@hasna/knowledge 0.2.26 → 0.2.28
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +61 -0
- package/bin/open-knowledge-mcp.js +85 -9
- package/bin/open-knowledge.js +86 -86
- package/dist/agent.d.ts +35 -0
- package/dist/artifact-store.d.ts +63 -0
- package/dist/auth.d.ts +35 -0
- package/dist/embeddings.d.ts +77 -0
- package/dist/index.d.ts +20 -0
- package/dist/index.js +5709 -0
- package/dist/knowledge-db.d.ts +27 -0
- package/dist/manifest-ingest.d.ts +35 -0
- package/dist/outbox-consume.d.ts +25 -0
- package/dist/provenance.d.ts +50 -0
- package/dist/providers.d.ts +89 -0
- package/dist/reindex.d.ts +37 -0
- package/dist/remote-client.d.ts +108 -0
- package/dist/retrieval.d.ts +71 -0
- package/dist/safety.d.ts +70 -0
- package/dist/sdk.d.ts +72 -0
- package/dist/search.d.ts +65 -0
- package/dist/service.d.ts +117 -0
- package/dist/source-ingest.d.ts +18 -0
- package/dist/source-ref.d.ts +30 -0
- package/dist/source-resolver.d.ts +92 -0
- package/dist/storage-contract.d.ts +106 -0
- package/dist/web-search.d.ts +40 -0
- package/dist/wiki-compiler.d.ts +67 -0
- package/dist/wiki-layout.d.ts +23 -0
- package/dist/workspace.d.ts +111 -0
- package/docs/architecture/ai-native-knowledge-base.md +24 -0
- package/docs/architecture/hosted-wrapper-responsibilities.md +8 -0
- package/docs/canonical-secrets-bootstrap-2026-06-08.md +127 -0
- package/package.json +15 -7
- package/src/agent.ts +0 -367
- package/src/artifact-store.ts +0 -184
- package/src/auth.ts +0 -123
- package/src/cli.ts +0 -1181
- package/src/embeddings.ts +0 -516
- package/src/knowledge-db.ts +0 -354
- package/src/manifest-ingest.ts +0 -515
- package/src/mcp-http.js +0 -110
- package/src/mcp.js +0 -1503
- package/src/outbox-consume.ts +0 -463
- package/src/provenance.ts +0 -93
- package/src/providers.ts +0 -308
- package/src/reindex.ts +0 -260
- package/src/remote-client.ts +0 -268
- package/src/retrieval.ts +0 -326
- package/src/safety.ts +0 -265
- package/src/schema.js +0 -25
- package/src/search.ts +0 -510
- package/src/service.ts +0 -432
- package/src/source-ingest.ts +0 -268
- package/src/source-ref.ts +0 -104
- package/src/source-resolver.ts +0 -436
- package/src/storage-contract.ts +0 -293
- package/src/store.ts +0 -113
- package/src/web-search.ts +0 -330
- package/src/wiki-compiler.ts +0 -711
- package/src/wiki-layout.ts +0 -251
- package/src/workspace.ts +0 -213
package/bin/open-knowledge.js
CHANGED
|
@@ -1,8 +1,8 @@
|
|
|
1
1
|
#!/usr/bin/env bun
|
|
2
2
|
// @bun
|
|
3
|
-
var
|
|
4
|
-
`);return t}function
|
|
5
|
-
`)}function
|
|
3
|
+
var A=import.meta.require;import{readFileSync as ke,writeFileSync as Ee,existsSync as ye,renameSync as cr,unlinkSync as Ot}from"fs";import{randomUUID as Nt}from"crypto";import{existsSync as rr,mkdirSync as Je,readFileSync as ir,writeFileSync as bt}from"fs";import{homedir as vt}from"os";import{dirname as sr,join as D,resolve as or}from"path";var ne=D(".hasna","apps","knowledge"),O={division:"xyz",app_type:"opensource",app:"knowledge",env:"prod",local_path:ne,s3:{bucket:"hasna-xyz-opensource-knowledge-prod",region:"us-east-1",profile:"hasna-xyz-infra",prefix:".hasna/apps/knowledge",server_side_encryption:"AES256"},secrets:{env:"hasna/xyz/opensource/knowledge/prod/env",aws:"hasna/xyz/opensource/knowledge/prod/aws",s3:"hasna/xyz/opensource/knowledge/prod/s3",rds:null,future_rds:"hasna/xyz/opensource/knowledge/prod/rds"},source_owner:"open-files",evidence_doc:"docs/canonical-secrets-bootstrap-2026-06-08.md"};function wt(){return{type:"s3",artifacts_root:"artifacts",s3:{bucket:O.s3.bucket,prefix:O.s3.prefix,region:O.s3.region,profile:O.s3.profile,server_side_encryption:O.s3.server_side_encryption}}}function Ge(){return D(vt(),".open-knowledge","db.json")}function ze(){return D(vt(),".hasna","apps","knowledge")}function ar(e=process.cwd()){return or(e,ne)}function ce(e){return{home:e,configPath:D(e,"config.json"),jsonStorePath:D(e,"db.json"),knowledgeDbPath:D(e,"knowledge.db"),artifactsDir:D(e,"artifacts"),cacheDir:D(e,"cache"),exportsDir:D(e,"exports"),indexesDir:D(e,"indexes"),logsDir:D(e,"logs"),runsDir:D(e,"runs"),schemasDir:D(e,"schemas"),wikiDir:D(e,"wiki")}}function ur(){return{version:1,mode:"local",hosted:{api_url:"https://knowledge.hasna.xyz"},storage:{type:"local",artifacts_root:"artifacts"},sources:{preferred_ref:"open-files",allowed_schemes:["open-files","s3","file","https","http"]},providers:{default_model:"openai:gpt-5.2",aliases:{fast:"openai:gpt-5-mini",reasoning:"anthropic:claude-opus-4-6",sonnet:"anthropic:claude-sonnet-4-6",deepseek:"deepseek:deepseek-chat","deepseek-reasoning":"deepseek:deepseek-reasoner"},openai:{api_key_env:"OPENAI_API_KEY",default_model:"gpt-5.2"},anthropic:{api_key_env:"ANTHROPIC_API_KEY",default_model:"claude-sonnet-4-6"},deepseek:{api_key_env:"DEEPSEEK_API_KEY",default_model:"deepseek-chat"}},embeddings:{default_model:"openai:text-embedding-3-small",dimensions:1536,batch_size:64,max_parallel_calls:4},safety:{network:{web_search_enabled:!1,s3_reads_enabled:!1,allowed_s3_buckets:[]},redaction:{enabled:!0},approvals:{generated_writes_require_approval:!0}}}}function Tt(e){let t=ce(e);Je(t.home,{recursive:!0});for(let n of[t.artifactsDir,t.cacheDir,t.exportsDir,t.indexesDir,t.logsDir,t.runsDir,t.schemasDir,t.wikiDir])Je(n,{recursive:!0});if(!rr(t.configPath))bt(t.configPath,`${JSON.stringify(ur(),null,2)}
|
|
4
|
+
`);return t}function Rt(e,t=process.cwd()){if(e==="project"||e==="local")return ce(ar(t));return ce(ze())}function de(e){Je(sr(e),{recursive:!0})}function St(e){let t=ir(e,"utf8");return JSON.parse(t)}function xt(e,t){de(e),bt(e,`${JSON.stringify(t,null,2)}
|
|
5
|
+
`)}function Ye(){return ce(ze()).jsonStorePath}function Ve(e){if(!ye(e))if(de(e),e===Ye()&&ye(Ge()))Ee(e,ke(Ge(),"utf8"));else Ee(e,JSON.stringify({items:[]},null,2))}function dr(e){return`${e}.lock`}function lr(e,t){let i=Date.now();while(Date.now()-i<5000){try{if(!ye(e)){Ee(e,JSON.stringify({owner:t,ts:Date.now()}));return}let c=JSON.parse(ke(e,"utf8"));if(Date.now()-c.ts>1e4)Ot(e)}catch{}let s=Date.now();while(Date.now()-s<50);}throw Error(`Could not acquire lock on ${e} after 5000ms`)}function _r(e,t){try{if(ye(e)){if(JSON.parse(ke(e,"utf8")).owner===t)Ot(e)}}catch{}}function U(e){Ve(e);let t=ke(e,"utf8"),n=JSON.parse(t);if(!n||!Array.isArray(n.items))return{items:[]};return n}function $(e,t){let n=`${e}.tmp.${Nt()}`;Ee(n,JSON.stringify(t,null,2)),cr(n,e)}function P(e,t){let n=Nt(),r=dr(e);lr(r,n);try{return t()}finally{_r(r,n)}}function Qe(){return`k_${Date.now().toString(36)}_${Math.random().toString(36).slice(2,8)}`}function At(e){return e.replace(/^k_/,"").slice(0,12)}import{Database as fr}from"bun:sqlite";var gr=`
|
|
6
6
|
PRAGMA journal_mode = WAL;
|
|
7
7
|
PRAGMA foreign_keys = ON;
|
|
8
8
|
|
|
@@ -169,7 +169,7 @@ CREATE VIRTUAL TABLE IF NOT EXISTS chunks_fts USING fts5(
|
|
|
169
169
|
|
|
170
170
|
INSERT OR IGNORE INTO schema_versions(version, applied_at)
|
|
171
171
|
VALUES (1, datetime('now'));
|
|
172
|
-
`,
|
|
172
|
+
`,mr=`
|
|
173
173
|
DROP TABLE IF EXISTS chunks_fts;
|
|
174
174
|
|
|
175
175
|
CREATE VIRTUAL TABLE IF NOT EXISTS chunks_fts USING fts5(
|
|
@@ -182,7 +182,7 @@ CREATE VIRTUAL TABLE IF NOT EXISTS chunks_fts USING fts5(
|
|
|
182
182
|
|
|
183
183
|
INSERT OR IGNORE INTO schema_versions(version, applied_at)
|
|
184
184
|
VALUES (2, datetime('now'));
|
|
185
|
-
`,
|
|
185
|
+
`,pr=`
|
|
186
186
|
CREATE TABLE IF NOT EXISTS audit_events (
|
|
187
187
|
id TEXT PRIMARY KEY,
|
|
188
188
|
event_type TEXT NOT NULL,
|
|
@@ -213,7 +213,7 @@ CREATE INDEX IF NOT EXISTS idx_approval_gates_status ON approval_gates(status);
|
|
|
213
213
|
|
|
214
214
|
INSERT OR IGNORE INTO schema_versions(version, applied_at)
|
|
215
215
|
VALUES (3, datetime('now'));
|
|
216
|
-
`,
|
|
216
|
+
`,hr=`
|
|
217
217
|
CREATE TABLE IF NOT EXISTS vector_index_entries (
|
|
218
218
|
id TEXT PRIMARY KEY,
|
|
219
219
|
chunk_id TEXT NOT NULL REFERENCES chunks(id) ON DELETE CASCADE,
|
|
@@ -244,7 +244,7 @@ CREATE INDEX IF NOT EXISTS idx_vector_index_status ON vector_index_entries(statu
|
|
|
244
244
|
|
|
245
245
|
INSERT OR IGNORE INTO schema_versions(version, applied_at)
|
|
246
246
|
VALUES (4, datetime('now'));
|
|
247
|
-
`,
|
|
247
|
+
`,Er=`
|
|
248
248
|
CREATE TABLE IF NOT EXISTS reindex_queue (
|
|
249
249
|
id TEXT PRIMARY KEY,
|
|
250
250
|
kind TEXT NOT NULL,
|
|
@@ -265,9 +265,9 @@ CREATE INDEX IF NOT EXISTS idx_reindex_queue_source_uri ON reindex_queue(source_
|
|
|
265
265
|
|
|
266
266
|
INSERT OR IGNORE INTO schema_versions(version, applied_at)
|
|
267
267
|
VALUES (5, datetime('now'));
|
|
268
|
-
`;function b(e){
|
|
269
|
-
`,{mode:384}),r}function
|
|
270
|
-
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[n,t.run_id??null,t.provider,t.model,t.input_tokens,t.output_tokens,t.cost_usd,JSON.stringify(t.metadata),t.created_at??new Date().toISOString()]),n}import{createHash as
|
|
268
|
+
`;function b(e){de(e);let t=new fr(e);return t.exec("PRAGMA foreign_keys = ON;"),t.exec("PRAGMA busy_timeout = 5000;"),t}function x(e){let t=b(e);try{if(t.exec(gr),re(t)<2)t.exec(mr);if(re(t)<3)t.exec(pr);if(re(t)<4)t.exec(hr);if(re(t)<5)t.exec(Er);return{path:e,schema_version:re(t)}}finally{t.close()}}function re(e){return e.query("SELECT MAX(version) AS version FROM schema_versions").get()?.version??0}function L(e,t){return e.query(`SELECT COUNT(*) AS n FROM ${t}`).get()?.n??0}function It(e){let t=b(e);try{return{schema_version:re(t),sources:L(t,"sources"),source_revisions:L(t,"source_revisions"),chunks:L(t,"chunks"),wiki_pages:L(t,"wiki_pages"),citations:L(t,"citations"),indexes:L(t,"knowledge_indexes"),runs:L(t,"runs"),run_events:L(t,"run_events"),redaction_findings:L(t,"redaction_findings"),audit_events:L(t,"audit_events"),approval_gates:L(t,"approval_gates"),storage_objects:L(t,"storage_objects"),embeddings:L(t,"chunk_embeddings"),vector_entries:L(t,"vector_index_entries"),reindex_queue:L(t,"reindex_queue")}}finally{t.close()}}import{existsSync as yr,mkdirSync as Lt,readFileSync as kr,writeFileSync as br}from"fs";import{dirname as vr,join as Ze,relative as wr,sep as Tr}from"path";function le(e){let t=e.replace(/\\/g,"/").trim();if(!t||t.startsWith("/"))throw Error(`Invalid artifact key: ${e}`);let n=t.split("/").filter(Boolean);if(n.length===0||n.some((r)=>r==="."||r===".."))throw Error(`Invalid artifact key: ${e}`);return n.join("/")}function et(e,t){let n=wr(e,t);if(n.startsWith("..")||n===".."||n.startsWith(`..${Tr}`))throw Error(`Artifact path escapes root: ${t}`)}class Ct{root;type="local";canRead=!0;canWrite=!0;constructor(e){this.root=e;Lt(e,{recursive:!0})}async put(e){let t=le(e.key),n=Ze(this.root,t);return et(this.root,n),Lt(vr(n),{recursive:!0}),br(n,e.body),{key:t,uri:`file://${n}`}}async getText(e){let t=le(e),n=Ze(this.root,t);return et(this.root,n),kr(n,"utf8")}async exists(e){let t=le(e),n=Ze(this.root,t);return et(this.root,n),yr(n)}}class Dt{options;type="s3";canRead=!0;canWrite=!0;client;constructor(e){this.options=e;this.client=e.client}async getClient(){if(this.client)return this.client;let[{S3Client:e},{fromIni:t}]=await Promise.all([import("@aws-sdk/client-s3"),import("@aws-sdk/credential-providers")]);return this.client=new e({region:this.options.region,credentials:this.options.profile?t({profile:this.options.profile}):void 0,maxAttempts:this.options.max_attempts}),this.client}objectKey(e){let t=le(e),n=this.options.prefix?le(this.options.prefix):"";return n?`${n}/${t}`:t}async put(e){let[{PutObjectCommand:t},n]=await Promise.all([import("@aws-sdk/client-s3"),this.getClient()]),r=this.objectKey(e.key);return await n.send(new t({Bucket:this.options.bucket,Key:r,Body:e.body,ContentType:e.content_type,Metadata:e.metadata,ServerSideEncryption:this.options.server_side_encryption,SSEKMSKeyId:this.options.kms_key_id})),{key:r,uri:`s3://${this.options.bucket}/${r}`}}async getText(e){let[{GetObjectCommand:t},n]=await Promise.all([import("@aws-sdk/client-s3"),this.getClient()]),r=this.objectKey(e),i=await n.send(new t({Bucket:this.options.bucket,Key:r}));if(!i.Body)return"";return await i.Body.transformToString()}async exists(e){let[{HeadObjectCommand:t},n]=await Promise.all([import("@aws-sdk/client-s3"),this.getClient()]),r=this.objectKey(e);try{return await n.send(new t({Bucket:this.options.bucket,Key:r})),!0}catch(i){let s=i instanceof Error?i.name:"";if(s==="NotFound"||s==="NoSuchKey"||s==="NotFoundError")return!1;throw i}}}function Ut(e,t){if(e.storage.type==="s3"){if(!e.storage.s3?.bucket)throw Error("S3 artifact storage requires storage.s3.bucket");return new Dt({bucket:e.storage.s3.bucket,prefix:e.storage.s3.prefix,region:e.storage.s3.region,profile:e.storage.s3.profile,max_attempts:e.storage.s3.max_attempts,server_side_encryption:e.storage.s3.server_side_encryption,kms_key_id:e.storage.s3.kms_key_id})}return new Ct(t.artifactsDir)}import{existsSync as Rr,mkdirSync as Sr,readFileSync as xr,unlinkSync as Or,writeFileSync as Nr}from"fs";import{homedir as Ar}from"os";import{dirname as Ir,join as Pt}from"path";var tt="https://knowledge.hasna.xyz";function X(e){let t=new URL(e);if(t.protocol!=="http:"&&t.protocol!=="https:")throw Error("Knowledge API URL must use http or https.");let n=t.pathname.replace(/\/+$/,"");if(n==="/api"||n==="/api/v1")t.pathname="/";else if(n.endsWith("/api/v1"))t.pathname=n.slice(0,-7)||"/";else if(n.endsWith("/api"))t.pathname=n.slice(0,-4)||"/";return t.toString().replace(/\/+$/,"")}function ve(e=process.env){if(e.HASNA_KNOWLEDGE_AUTH_PATH)return e.HASNA_KNOWLEDGE_AUTH_PATH;let t=e.HASNA_KNOWLEDGE_AUTH_DIR??Pt(Ar(),".hasna","knowledge");return Pt(t,"auth.json")}function be(e,t=process.env){return X(t.KNOWLEDGE_API_URL??e?.hosted?.api_url??tt)}function Ft(e=process.env){try{let t=ve(e);if(!Rr(t))return null;let n=JSON.parse(xr(t,"utf8"));return typeof n.api_key==="string"&&n.api_key.length>0?n:null}catch{return null}}function Mt(e,t=process.env){let n=ve(t),r={...e,api_url:e.api_url?X(e.api_url):void 0,created_at:e.created_at??new Date().toISOString()};return Sr(Ir(n),{recursive:!0,mode:448}),Nr(n,`${JSON.stringify(r,null,2)}
|
|
269
|
+
`,{mode:384}),r}function jt(e=process.env){try{return Or(ve(e)),!0}catch{return!1}}function nt(e=process.env){if(e.KNOWLEDGE_API_KEY)return{apiKey:e.KNOWLEDGE_API_KEY,source:"env"};if(e.HASNA_KNOWLEDGE_API_KEY)return{apiKey:e.HASNA_KNOWLEDGE_API_KEY,source:"env"};let t=Ft(e);return t?.api_key?{apiKey:t.api_key,source:"file"}:{apiKey:null,source:"none"}}function Kt(e,t=process.env){let n=Ft(t),r=nt(t),i=t.KNOWLEDGE_API_URL?be(e,t):n?.api_url?X(n.api_url):be(e,t);return{authenticated:Boolean(r.apiKey),source:r.source,api_url:i,auth_path:ve(t),email:r.source==="file"?n?.email??null:null,org_id:r.source==="file"?n?.org_id??null:null,org_slug:r.source==="file"?n?.org_slug??null:null,user_id:r.source==="file"?n?.user_id??null:null,api_key_present:Boolean(r.apiKey)}}import{randomUUID as ln}from"crypto";import{randomUUID as Lr}from"crypto";var rt={openai:{api_key_env:"OPENAI_API_KEY",default_model:"gpt-5.2"},anthropic:{api_key_env:"ANTHROPIC_API_KEY",default_model:"claude-sonnet-4-6"},deepseek:{api_key_env:"DEEPSEEK_API_KEY",default_model:"deepseek-chat"}},Cr={openai:{text_generation:!0,structured_output:!0,tool_usage:!0,tool_streaming:!0,image_input:!0,native_web_search:!0,reasoning:!0,embeddings:!0},anthropic:{text_generation:!0,structured_output:!0,tool_usage:!0,tool_streaming:!0,image_input:!0,native_web_search:!1,reasoning:!0,embeddings:!1},deepseek:{text_generation:!0,structured_output:!0,tool_usage:!0,tool_streaming:!0,image_input:!1,native_web_search:!1,reasoning:!0,embeddings:!1}},Dr={default:"openai:gpt-5.2",fast:"openai:gpt-5-mini",reasoning:"anthropic:claude-opus-4-6",sonnet:"anthropic:claude-sonnet-4-6",deepseek:"deepseek:deepseek-chat","deepseek-reasoning":"deepseek:deepseek-reasoner"};function Wt(e){return e?.providers??{}}function q(e,t){let n=Wt(e)[t]??{};return{...rt[t],...n}}function Xt(e){let t=Wt(e);return{...Dr,...t.default_model?{default:t.default_model}:{},...t.aliases??{}}}function C(e){let[t,...n]=e.split(":"),r=n.join(":");if(t!=="openai"&&t!=="anthropic"&&t!=="deepseek")throw Error(`Unsupported AI provider: ${t}`);if(!r)throw Error(`Invalid model ref: ${e}. Expected provider:model.`);return{provider:t,model:r}}function H(e,t){return Xt(t)[e]??e}function it(e){let t=Xt(e);return Object.entries(t).map(([n,r])=>{let i=C(r);return{alias:n,model_ref:r,provider:i.provider,model:i.model,default:n==="default",capabilities:Cr[i.provider]}})}function qt(e,t=process.env){return Object.keys(rt).map((n)=>{let r=q(e,n),i=Boolean(t[r.api_key_env]);return{provider:n,api_key_env:r.api_key_env,configured:i,source:i?"env":"missing",base_url:r.base_url??null,default_model:r.default_model}})}function Ht(e,t=process.env){return{default_model:H("default",e),providers:qt(e,t),models:it(e)}}function z(e,t,n=process.env){let r=qt(t,n).find((i)=>i.provider===e);if(!r)throw Error(`Unsupported AI provider: ${e}`);if(!r.configured)throw Error(`Missing ${r.api_key_env} for ${e}. Set the env var to use this provider.`);return r}async function Ur(e){if(e==="openai"){let{createOpenAI:n}=await import("@ai-sdk/openai");return n}if(e==="anthropic"){let{createAnthropic:n}=await import("@ai-sdk/anthropic");return n}let{createDeepSeek:t}=await import("@ai-sdk/deepseek");return t}async function Pr(e={}){let{createProviderRegistry:t}=await import("ai"),n=e.env??process.env,r={};for(let i of Object.keys(rt)){let s=q(e.config,i),c=n[s.api_key_env];if(!c)continue;let d=e.factories?.[i]??await Ur(i);r[i]=d({apiKey:c,baseURL:s.base_url})}return t(r)}async function Bt(e,t={}){let n=H(e,t.config),r=C(n);return z(r.provider,t.config,t.env),(await Pr(t)).languageModel(n)}function $t(e,t){for(let n of t){let r=e[n];if(typeof r==="number"&&Number.isFinite(r))return r}return 0}function we(e){let t=e.usage??{};return{provider:e.provider,model:e.model,input_tokens:$t(t,["inputTokens","promptTokens","input_tokens","prompt_tokens"]),output_tokens:$t(t,["outputTokens","completionTokens","output_tokens","completion_tokens"]),cost_usd:e.costUsd??0,metadata:{usage:t,provider_metadata:e.providerMetadata??{}}}}function Te(e,t){let n=`usage_${Lr()}`;return e.run(`INSERT INTO provider_usage (id, run_id, provider, model, input_tokens, output_tokens, cost_usd, metadata_json, created_at)
|
|
270
|
+
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[n,t.run_id??null,t.provider,t.model,t.input_tokens,t.output_tokens,t.cost_usd,JSON.stringify(t.metadata),t.created_at??new Date().toISOString()]),n}import{createHash as ri}from"crypto";function st(e){return["deleted","stale","invalidated","reindex_required"].includes((e??"").toLowerCase())}function B(e){let t=e.status??null;return{source_owner:"open-files",source_ref:e.source_ref??null,source_uri:e.source_uri??null,source_kind:e.source_kind??null,source_revision_id:e.source_revision_id??null,revision:e.revision??null,hash:e.hash??null,chunk_id:e.chunk_id??null,start_offset:e.start_offset??null,end_offset:e.end_offset??null,status:t,read_only:!0,citation_required:!0,resolver:e.resolver??null,stale:st(t)}}function Y(e){return{source_owner:"open-files",generated_from:e.generated_from,artifact_key:e.artifact_key,source_refs:e.source_refs??[],read_only_sources:!0,citation_required:e.citation_required??!0,raw_source_bytes_stored_in_open_knowledge:!1}}function Jt(e,t){return{...e,provenance:t}}import{createHash as Yt}from"crypto";var Fr="openai:text-embedding-3-small",Vt=1536;function Re(e){return e?.embeddings??{}}function Gt(e,t){return`${e}_${Yt("sha256").update(t).digest("hex").slice(0,20)}`}function at(e){if(!e)return{};try{let t=JSON.parse(e);return t&&typeof t==="object"&&!Array.isArray(t)?t:{}}catch{return{}}}function j(e,t){for(let n of t){let r=e[n];if(typeof r==="string"&&r.length>0)return r}return null}function zt(e,t){for(let n of t){let r=e[n];if(typeof r==="number"&&Number.isFinite(r))return r}return null}function ot(e){return Math.sqrt(e.reduce((t,n)=>t+n*n,0))}function Mr(e,t,n=ot(t)){let r=ot(e);if(r===0||n===0)return 0;let i=Math.min(e.length,t.length),s=0;for(let c=0;c<i;c+=1)s+=e[c]*t[c];return s/(r*n)}function jr(e,t){let n=Yt("sha256").update(e).digest();return Array.from({length:t},(r,i)=>{let s=n[i%n.length]/255;return Number((s*2-1).toFixed(6))})}async function Kr(e,t,n=process.env){z("openai",t,n);let r=q(t,"openai"),{createOpenAI:i}=await import("@ai-sdk/openai"),s=i({apiKey:n[r.api_key_env],baseURL:r.base_url});if(s.embeddingModel)return s.embeddingModel(e);if(s.textEmbedding)return s.textEmbedding(e);if(s.textEmbeddingModel)return s.textEmbeddingModel(e);throw Error("OpenAI provider does not expose an embedding model factory.")}function V(e,t){if(!e||e==="default"||e==="embedding")return Re(t).default_model??Fr;return e}async function Qt(e,t={}){let n=V(t.modelRef,t.config),r=C(n);if(r.provider!=="openai")throw Error(`Embedding provider ${r.provider} is not supported yet. Use openai:text-embedding-3-small.`);let i=t.dimensions??Re(t.config).dimensions??Vt;if(t.fake)return{provider:r.provider,model:r.model,dimensions:i,vectors:e.map((a)=>jr(a,i)),usage:{input_tokens:e.reduce((a,u)=>a+Math.max(1,Math.ceil(u.split(/\s+/).filter(Boolean).length*1.25)),0)}};let{embedMany:s}=await import("ai"),c=await Kr(r.model,t.config,t.env),d=await s({model:c,values:e,maxParallelCalls:t.maxParallelCalls??Re(t.config).max_parallel_calls,providerOptions:{openai:{dimensions:i}}}),g=d.embeddings;return{provider:r.provider,model:r.model,dimensions:g[0]?.length??i,vectors:g,usage:{input_tokens:d.usage?.tokens??0}}}function $r(e,t){if(t.sourceRevisionId)return e.query(`SELECT
|
|
271
271
|
c.id,
|
|
272
272
|
c.text,
|
|
273
273
|
c.token_count,
|
|
@@ -305,7 +305,7 @@ VALUES (5, datetime('now'));
|
|
|
305
305
|
ON v.chunk_id = c.id AND v.provider = ? AND v.model = ?
|
|
306
306
|
WHERE v.id IS NULL
|
|
307
307
|
ORDER BY c.created_at ASC, c.ordinal ASC
|
|
308
|
-
LIMIT ?`).all(t.provider,t.model,t.limit)}function
|
|
308
|
+
LIMIT ?`).all(t.provider,t.model,t.limit)}function Wr(e){let t=at(e.metadata_json),n=t.provenance;if(n&&typeof n==="object"&&!Array.isArray(n))return n;return B({source_ref:j(t,["source_ref"]),source_uri:e.source_uri??j(t,["source_uri"]),source_kind:e.source_kind??j(t,["source_kind"]),source_revision_id:e.source_revision_id,revision:e.revision??j(t,["revision"]),hash:e.hash??j(t,["hash"]),chunk_id:e.id,start_offset:e.start_offset??zt(t,["start_offset"]),end_offset:e.end_offset??zt(t,["end_offset"]),status:j(t,["status"]),resolver:"open-files-read-only"})}function Xr(e,t,n,r){let i=e.prepare(`
|
|
309
309
|
INSERT INTO chunk_embeddings (id, chunk_id, provider, model, dimensions, vector_json, created_at)
|
|
310
310
|
VALUES (?, ?, ?, ?, ?, ?, ?)
|
|
311
311
|
ON CONFLICT(chunk_id, provider, model) DO UPDATE SET
|
|
@@ -334,10 +334,10 @@ VALUES (5, datetime('now'));
|
|
|
334
334
|
status = excluded.status,
|
|
335
335
|
metadata_json = excluded.metadata_json,
|
|
336
336
|
updated_at = excluded.updated_at
|
|
337
|
-
`);return e.transaction(()=>{for(let d=0;d<t.length;d+=1){let g=t[d],a=n.vectors[d];if(!a)continue;let
|
|
337
|
+
`);return e.transaction(()=>{for(let d=0;d<t.length;d+=1){let g=t[d],a=n.vectors[d];if(!a)continue;let u=at(g.metadata_json),o=Wr(g),_=o.source_ref??j(u,["source_ref"]),l=o.source_uri??g.source_uri??j(u,["source_uri"]),f=o.revision??g.revision??j(u,["revision"]),m=o.hash??g.hash??j(u,["hash"]),E=o.status??j(u,["status"])??"active",h=JSON.stringify(a);i.run(Gt("emb",`${g.id}\x00${n.provider}\x00${n.model}`),g.id,n.provider,n.model,n.dimensions,h,r),s.run(Gt("vec",`${g.id}\x00${n.provider}\x00${n.model}`),g.id,g.source_revision_id,n.provider,n.model,n.dimensions,h,ot(a),l,_,f,m,o.start_offset,o.end_offset,g.token_count,E,JSON.stringify({...u,provenance:o,embedded_at:r}),r,r)}})(),t.length}async function Se(e){let t=V(e.modelRef,e.config),n=C(t);if(n.provider!=="openai")throw Error(`Embedding provider ${n.provider} is not supported yet.`);let r=(e.now??new Date).toISOString(),i=Math.max(1,Math.min(e.limit??100,1000));x(e.dbPath);let s=b(e.dbPath),c;try{c=$r(s,{provider:n.provider,model:n.model,limit:i,sourceRevisionId:e.sourceRevisionId})}finally{s.close()}if(c.length===0)return{provider:n.provider,model:n.model,dimensions:e.dimensions??Re(e.config).dimensions??Vt,chunks_seen:0,chunks_embedded:0,embeddings_upserted:0,vector_entries_upserted:0,usage:{input_tokens:0}};let d=await Qt(c.map((a)=>a.text),e),g=b(e.dbPath);try{let a=Xr(g,c,d,r);return{provider:d.provider,model:d.model,dimensions:d.dimensions,chunks_seen:c.length,chunks_embedded:c.length,embeddings_upserted:a,vector_entries_upserted:a,usage:d.usage}}finally{g.close()}}function Zt(e){x(e);let t=b(e);try{let n=t.query("SELECT COUNT(*) AS n FROM chunk_embeddings").get()?.n??0,r=t.query("SELECT COUNT(*) AS n FROM vector_index_entries").get()?.n??0,i=t.query(`SELECT provider, model, dimensions, COUNT(*) AS entries, MAX(updated_at) AS updated_at
|
|
338
338
|
FROM vector_index_entries
|
|
339
339
|
GROUP BY provider, model, dimensions
|
|
340
|
-
ORDER BY provider, model`).all();return{total_embeddings:n,total_vector_entries:r,indexes:i}}finally{t.close()}}async function
|
|
340
|
+
ORDER BY provider, model`).all();return{total_embeddings:n,total_vector_entries:r,indexes:i}}finally{t.close()}}async function xe(e){let t=V(e.modelRef,e.config),n=C(t),r=Math.max(1,Math.min(e.limit??10,100)),i=await Qt([e.query],e),s=i.vectors[0]??[];x(e.dbPath);let c=b(e.dbPath);try{let g=c.query(`SELECT
|
|
341
341
|
v.chunk_id,
|
|
342
342
|
c.text,
|
|
343
343
|
v.vector_json,
|
|
@@ -349,7 +349,7 @@ VALUES (5, datetime('now'));
|
|
|
349
349
|
v.metadata_json
|
|
350
350
|
FROM vector_index_entries v
|
|
351
351
|
JOIN chunks c ON c.id = v.chunk_id
|
|
352
|
-
WHERE v.provider = ? AND v.model = ? AND v.status = 'active'`).all(n.provider,n.model).map((a)=>{let
|
|
352
|
+
WHERE v.provider = ? AND v.model = ? AND v.status = 'active'`).all(n.provider,n.model).map((a)=>{let u=JSON.parse(a.vector_json),o=at(a.metadata_json),_=o.provenance&&typeof o.provenance==="object"&&!Array.isArray(o.provenance)?o.provenance:null;return{chunk_id:a.chunk_id,score:Mr(s,u,a.vector_norm),text:a.text,source_uri:a.source_uri,source_ref:a.source_ref,revision:a.revision,hash:a.hash,provenance:_}}).sort((a,u)=>u.score-a.score).slice(0,r);return{provider:n.provider,model:n.model,dimensions:i.dimensions,query:e.query,results:g}}finally{c.close()}}function Ne(e){if(!e)return{};try{let t=JSON.parse(e);return t&&typeof t==="object"&&!Array.isArray(t)?t:{}}catch{return{}}}function F(e,t){for(let n of t){let r=e[n];if(typeof r==="string"&&r.length>0)return r}return null}function en(e,t){for(let n of t){let r=e[n];if(typeof r==="number"&&Number.isFinite(r))return r}return null}function tn(e){return Array.from(new Set(e))}function qr(e){let t=e.normalize("NFKC").toLowerCase().match(/[\p{L}\p{N}_]+/gu)??[];return tn(t.filter((n)=>n.length>0)).slice(0,16)}function Hr(e){if(e.length===0)return null;return e.map((t)=>`${t}*`).join(" OR ")}function Br(e){return e.replace(/[\\%_]/g,(t)=>`\\${t}`)}function nn(e,t){return e.flatMap((n)=>Array.from({length:t},()=>`%${Br(n)}%`))}function Jr(e,t){let n=Number.isFinite(e)?1/(1+Math.abs(e)):0,r=1/(1+t);return Ae(Math.max(n,r))}function rn(e,t){if(t.length===0)return 0;let n=t.filter((r)=>e.includes(r)).length;if(n===0)return 0;return Ae(Math.min(0.85,0.35+n/t.length*0.5))}function Gr(e){return Ae(Math.max(0,Math.min(1,(e+1)/2)))}function Ae(e){return Number(e.toFixed(6))}function _e(e,t){let n=e.keyword??0,r=e.semantic??0,i=e.catalog??0,s=t?.chunk_id?0.05:0;return Ae(Math.min(1,n*0.55+r*0.4+i*0.35+s))}function ut(e){let t=e.provenance;return t&&typeof t==="object"&&!Array.isArray(t)?t:null}function zr(e){let t=Ne(e.chunk_metadata_json),n=ut(t);if(n)return n;if(!e.source_revision_id&&!e.source_uri)return null;return B({source_ref:F(t,["source_ref"]),source_uri:e.source_uri??F(t,["source_uri"]),source_kind:e.source_kind??F(t,["source_kind"]),source_revision_id:e.source_revision_id,revision:e.revision??F(t,["revision"]),hash:e.hash??F(t,["hash"]),chunk_id:e.chunk_id,start_offset:e.start_offset??en(t,["start_offset"]),end_offset:e.end_offset??en(t,["end_offset"]),status:F(t,["status"]),resolver:"open-files-read-only"})}function Yr(e,t,n){if(!t)return[];return e.query(`SELECT
|
|
353
353
|
chunks_fts.chunk_id,
|
|
354
354
|
c.kind AS chunk_kind,
|
|
355
355
|
c.wiki_page_id,
|
|
@@ -378,96 +378,96 @@ VALUES (5, datetime('now'));
|
|
|
378
378
|
LEFT JOIN wiki_pages wp ON wp.id = c.wiki_page_id
|
|
379
379
|
WHERE chunks_fts MATCH ?
|
|
380
380
|
ORDER BY rank ASC
|
|
381
|
-
LIMIT ?`).all(t,n)}function
|
|
381
|
+
LIMIT ?`).all(t,n)}function sn(e,t){if(t.length===0)return"1 = 0";return t.map(()=>`(${e.map((r)=>`lower(COALESCE(${r}, '')) LIKE ? ESCAPE '\\'`).join(" OR ")})`).join(" OR ")}function Vr(e,t,n){let r=["path","title","artifact_uri","metadata_json"];return e.query(`SELECT id, path, title, artifact_uri, content_hash, status, metadata_json
|
|
382
382
|
FROM wiki_pages
|
|
383
|
-
WHERE status = 'active' AND (${
|
|
383
|
+
WHERE status = 'active' AND (${sn(r,t)})
|
|
384
384
|
ORDER BY updated_at DESC
|
|
385
|
-
LIMIT ?`).all(...
|
|
385
|
+
LIMIT ?`).all(...nn(t,r.length),n)}function Qr(e,t,n){let r=["kind","name","shard_key","artifact_uri","metadata_json"];return e.query(`SELECT id, kind, name, artifact_uri, shard_key, metadata_json
|
|
386
386
|
FROM knowledge_indexes
|
|
387
|
-
WHERE ${
|
|
387
|
+
WHERE ${sn(r,t)}
|
|
388
388
|
ORDER BY updated_at DESC
|
|
389
|
-
LIMIT ?`).all(...
|
|
389
|
+
LIMIT ?`).all(...nn(t,r.length),n)}function Zr(e,t){let n=Ne(e.chunk_metadata_json),r=zr(e),i=F(n,["source_ref"]),s=e.source_uri??F(n,["source_uri"]),c=Boolean(e.wiki_page_id),d={kind:c?"wiki_chunk":"source_chunk",id:e.chunk_id,title:c?e.wiki_title:e.source_title,text:e.text,score:0,scores:{keyword:t},source:s||i?{uri:s,ref:i,kind:e.source_kind??F(n,["source_kind"]),revision:e.revision??F(n,["revision"]),hash:e.hash??F(n,["hash"])}:null,citation:{chunk_id:e.chunk_id,start_offset:e.start_offset,end_offset:e.end_offset},artifact:c?{uri:e.wiki_artifact_uri,path:e.wiki_path,hash:e.wiki_content_hash,shard_key:e.wiki_path}:null,provenance:r,reasons:["keyword_match"]};return d.score=_e(d.scores,d.citation),d}function ei(e,t){let n=Ne(e.metadata_json),r=rn(`${e.path} ${e.title} ${e.artifact_uri??""} ${e.metadata_json}`.toLowerCase(),t),i={kind:"wiki_page",id:e.id,title:e.title,text:null,score:0,scores:{catalog:r},source:null,citation:null,artifact:{uri:e.artifact_uri,path:e.path,hash:e.content_hash,shard_key:e.path},provenance:ut(n),reasons:["wiki_catalog_match"]};return i.score=_e(i.scores,i.citation),i}function ti(e,t){let n=Ne(e.metadata_json),r=rn(`${e.kind} ${e.name} ${e.shard_key??""} ${e.artifact_uri??""} ${e.metadata_json}`.toLowerCase(),t),i={kind:"knowledge_index",id:e.id,title:e.name,text:null,score:0,scores:{catalog:r},source:null,citation:null,artifact:{uri:e.artifact_uri,path:F(n,["artifact_key"]),hash:F(n,["content_hash"]),shard_key:e.shard_key},provenance:ut(n),reasons:["index_catalog_match"]};return i.score=_e(i.scores,i.citation),i}function Oe(e,t){let n=`${t.kind}:${t.id}`,r=e.get(n);if(!r){e.set(n,t);return}r.scores={keyword:Math.max(r.scores.keyword??0,t.scores.keyword??0)||void 0,semantic:Math.max(r.scores.semantic??0,t.scores.semantic??0)||void 0,catalog:Math.max(r.scores.catalog??0,t.scores.catalog??0)||void 0},r.reasons=tn([...r.reasons,...t.reasons]),r.text=r.text??t.text,r.title=r.title??t.title,r.source=r.source??t.source,r.citation=r.citation??t.citation,r.artifact=r.artifact??t.artifact,r.provenance=r.provenance??t.provenance,r.score=_e(r.scores,r.citation)}function ni(e){let t={source_chunk:0,wiki_chunk:1,wiki_page:2,knowledge_index:3};return e.sort((n,r)=>{if(r.score!==n.score)return r.score-n.score;return t[n.kind]-t[r.kind]||n.id.localeCompare(r.id)})}async function Ie(e){let t=e.query.trim();if(!t)throw Error("Search query is required.");let n=Math.max(1,Math.min(e.limit??10,100)),r=qr(t),i=Hr(r),s=e.semantic===!0||e.fake===!0||Boolean(e.modelRef),c=[],d=null,g=null,a=null,u=0,o=0,_=0,l=new Map;x(e.dbPath);let f=b(e.dbPath);try{let E=Yr(f,i,Math.max(n*3,20));u=E.length,E.forEach((k,p)=>Oe(l,Zr(k,Jr(k.rank,p))));let h=Vr(f,r,Math.max(n,10)),T=Qr(f,r,Math.max(n,10));o=h.length+T.length,h.forEach((k)=>Oe(l,ei(k,r))),T.forEach((k)=>Oe(l,ti(k,r)))}finally{f.close()}if(s)try{let E=await xe({dbPath:e.dbPath,query:t,limit:Math.max(n*3,20),config:e.config,env:e.env,modelRef:e.modelRef,dimensions:e.dimensions,fake:e.fake,batchSize:e.batchSize,maxParallelCalls:e.maxParallelCalls});d=E.provider,g=E.model,a=E.dimensions,_=E.results.length;for(let h of E.results){let T={kind:"source_chunk",id:h.chunk_id,title:null,text:h.text,score:0,scores:{semantic:Gr(h.score)},source:{uri:h.source_uri,ref:h.source_ref,kind:h.provenance?.source_kind??null,revision:h.revision,hash:h.hash},citation:{chunk_id:h.chunk_id,start_offset:h.provenance?.start_offset??null,end_offset:h.provenance?.end_offset??null},artifact:null,provenance:h.provenance,reasons:["semantic_match"]};T.score=_e(T.scores,T.citation),Oe(l,T)}}catch(E){c.push(`semantic_search_failed: ${E instanceof Error?E.message:String(E)}`)}let m=ni(Array.from(l.values())).slice(0,n);return{query:t,limit:n,mode:{keyword:!0,catalog:!0,semantic:s},semantic_provider:d,semantic_model:g,semantic_dimensions:a,counts:{keyword_results:u,catalog_results:o,semantic_results:_,merged_results:m.length},warnings:c,results:m}}function on(e,t){return`${e}_${ri("sha256").update(t).digest("hex").slice(0,20)}`}function an(e){return e.normalize("NFKC").trim().replace(/\s+/g," ").toLowerCase()}function ii(e){return Array.from(new Set(an(e).match(/[\p{L}\p{N}_]+/gu)??[])).slice(0,16)}function si(e){return[e.title,e.text].filter(Boolean).join(" ").toLowerCase()}function oi(e,t){if(t.length===0)return 0;let n=si(e),r=t.filter((i)=>n.includes(i)).length;return Number((r/t.length).toFixed(6))}function ai(e){if(!e)return!0;if("read_only"in e)return e.read_only===!0;if("read_only_sources"in e)return e.read_only_sources===!0;return!0}function un(e){if(!e)return!1;if("stale"in e&&e.stale)return!0;if("status"in e)return st(e.status);return!1}function ui(e){if(un(e.provenance))return 0;if(e.source?.hash||e.source?.revision)return 1;if(e.artifact?.hash)return 0.85;if(e.provenance&&"source_refs"in e.provenance&&e.provenance.source_refs.length>0)return 0.75;return 0.55}function ci(e){if(e.citation?.chunk_id&&(e.source?.uri||e.artifact?.uri))return 1;if(e.provenance&&"citation_required"in e.provenance&&e.provenance.citation_required)return 0.75;if(e.artifact?.uri)return 0.65;return 0.35}function di(e){if(e.kind==="wiki_chunk")return 0.85;if(e.kind==="source_chunk")return 0.8;if(e.kind==="wiki_page")return 0.65;return 0.55}function li(e,t){let n={base_score:e.score,exact_score:oi(e,t),citation_score:ci(e),freshness_score:ui(e),authority_score:di(e)},r=Math.min(1,n.base_score*0.65+n.exact_score*0.1+n.citation_score*0.1+n.freshness_score*0.1+n.authority_score*0.05),i=new Set(e.reasons);if(n.exact_score>0.5)i.add("exact_term");if(n.citation_score>=0.75)i.add("cited_source");if(n.freshness_score>=0.85)i.add("fresh_source");return{...e,score:Number(r.toFixed(6)),reasons:Array.from(i),rerank:{...n,final_score:Number(r.toFixed(6))}}}function cn(e,t){let n=e.text??e.title;if(!n)return null;let r=n.replace(/\s+/g," ").trim();return r.length<=t?r:`${r.slice(0,Math.max(0,t-1)).trim()}...`}function _i(e){return{id:on("cite",`${e.kind}\x00${e.id}\x00${e.source?.uri??""}\x00${e.artifact?.uri??""}`),result_id:e.id,kind:e.kind,source_uri:e.source?.uri??null,source_ref:e.source?.ref??null,artifact_uri:e.artifact?.uri??null,artifact_path:e.artifact?.path??null,revision:e.source?.revision??null,hash:e.source?.hash??e.artifact?.hash??null,chunk_id:e.citation?.chunk_id??null,start_offset:e.citation?.start_offset??null,end_offset:e.citation?.end_offset??null,quote:cn(e,500),provenance:e.provenance}}function fi(e,t,n){let r=cn(e,n);if(!r)return null;return{id:on("excerpt",`${e.kind}\x00${e.id}`),result_id:e.id,citation_id:t.id,kind:e.kind,text:r,score:e.score}}function Le(e){return e.map(()=>"?").join(", ")}function gi(e,t){let n=t.map((d)=>d.citation?.chunk_id).filter((d)=>Boolean(d)),r=t.filter((d)=>d.kind==="wiki_page").map((d)=>d.id),i=[],s=[];if(n.length===0&&r.length===0)return{citations:i,backlinks:s};let c=b(e);try{if(n.length>0)i.push(...c.query(`SELECT id, wiki_page_id, chunk_id, source_uri, quote, start_offset, end_offset
|
|
390
390
|
FROM citations
|
|
391
|
-
WHERE chunk_id IN (${
|
|
391
|
+
WHERE chunk_id IN (${Le(n)})
|
|
392
392
|
ORDER BY created_at DESC
|
|
393
|
-
LIMIT 50`).all(...n));if(r.length>0)i.push(...
|
|
393
|
+
LIMIT 50`).all(...n));if(r.length>0)i.push(...c.query(`SELECT id, wiki_page_id, chunk_id, source_uri, quote, start_offset, end_offset
|
|
394
394
|
FROM citations
|
|
395
|
-
WHERE wiki_page_id IN (${
|
|
395
|
+
WHERE wiki_page_id IN (${Le(r)})
|
|
396
396
|
ORDER BY created_at DESC
|
|
397
|
-
LIMIT 50`).all(...r)),s.push(...
|
|
397
|
+
LIMIT 50`).all(...r)),s.push(...c.query(`SELECT from_page_id, to_page_id, label
|
|
398
398
|
FROM wiki_backlinks
|
|
399
|
-
WHERE from_page_id IN (${
|
|
400
|
-
LIMIT 50`).all(...r,...r))}finally{
|
|
401
|
-
`)}function
|
|
399
|
+
WHERE from_page_id IN (${Le(r)}) OR to_page_id IN (${Le(r)})
|
|
400
|
+
LIMIT 50`).all(...r,...r))}finally{c.close()}return{citations:i,backlinks:s}}async function Ce(e){let t=Math.max(200,Math.min(e.contextChars??1200,4000)),n=await Ie(e),r=ii(n.query),i=[...n.warnings],s=new Set,c=new Set,g=n.results.filter((o)=>{if(!ai(o.provenance))return i.push(`permission_filtered: ${o.kind}:${o.id}`),s.add("Dropped a result because provenance was not read-only."),!1;if(un(o.provenance))return i.push(`stale_filtered: ${o.kind}:${o.id}`),c.add("Dropped a stale result whose source status requires reindexing."),!1;return!0}).map((o)=>li(o,r)).sort((o,_)=>_.score-o.score||o.id.localeCompare(_.id)).slice(0,n.limit),a=g.map(_i),u=g.map((o,_)=>fi(o,a[_],t)).filter((o)=>Boolean(o));for(let o of g){if(o.provenance&&"read_only"in o.provenance&&o.provenance.read_only)s.add("All source-backed excerpts are read-only and citation-required.");if(o.rerank.freshness_score>=0.85)c.add("Fresh source revision/hash or artifact hash is present for top context.")}return{query:n.query,normalized_query:an(n.query),created_at:new Date().toISOString(),mode:n.mode,warnings:i,search_counts:n.counts,results:g,citations:a,excerpts:u,graph:gi(e.dbPath,g),notes:{permissions:Array.from(s),freshness:Array.from(c)}}}function ct(e){let t=e.trim().split(/\s+/).filter(Boolean).length;return Math.max(1,Math.ceil(t*1.25))}function lt(e){return`C${e+1}`}function mi(e,t){if(t.excerpts.length===0)return`No indexed knowledge matched the prompt: ${e}`;return[`Found ${t.excerpts.length} relevant knowledge excerpt(s) for: ${e}`,"",...t.excerpts.slice(0,5).map((r,i)=>{let s=t.citations.find((d)=>d.id===r.citation_id),c=s?.source_ref??s?.source_uri??s?.artifact_path??s?.artifact_uri??"unknown source";return`[${lt(i)}] ${r.text} (${c})`})].join(`
|
|
401
|
+
`)}function pi(e,t){let n=t.citations.map((i,s)=>({id:lt(s),source_ref:i.source_ref,source_uri:i.source_uri,artifact_path:i.artifact_path,revision:i.revision,hash:i.hash,quote:i.quote})),r=t.excerpts.map((i,s)=>({id:lt(s),kind:i.kind,text:i.text,score:i.score}));return[`Prompt: ${e}`,"","Use only the provided context. Cite claims with citation ids like [C1]. If context is insufficient, say what is missing.","",`Context excerpts:
|
|
402
402
|
${JSON.stringify(r,null,2)}`,"",`Citations:
|
|
403
403
|
${JSON.stringify(n,null,2)}`].join(`
|
|
404
|
-
`)}function
|
|
405
|
-
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[t.runId,"knowledge-prompt",t.prompt,t.status,t.provider,t.model,JSON.stringify(t.metadata),t.now,t.now])}finally{n.close()}}function
|
|
406
|
-
VALUES (?, ?, ?, ?, ?, ?)`,[`evt_${
|
|
404
|
+
`)}function hi(e,t){if(t.citations.length===0)return[];return[{kind:"answer_note",title:e.length>80?`${e.slice(0,77)}...`:e,citations:t.citations.map((n)=>n.id),requires_approval:!0}]}function Ei(e,t){let n=b(e);try{n.run(`INSERT INTO runs (id, type, prompt, status, provider, model, metadata_json, created_at, updated_at)
|
|
405
|
+
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[t.runId,"knowledge-prompt",t.prompt,t.status,t.provider,t.model,JSON.stringify(t.metadata),t.now,t.now])}finally{n.close()}}function dt(e,t){let n=b(e);try{n.run(`INSERT INTO run_events (id, run_id, level, event, metadata_json, created_at)
|
|
406
|
+
VALUES (?, ?, ?, ?, ?, ?)`,[`evt_${ln()}`,t.runId,t.level,t.event,JSON.stringify(t.metadata),t.now])}finally{n.close()}}function dn(e,t){let n=b(e);try{n.run(`UPDATE runs
|
|
407
407
|
SET status = ?, provider = ?, model = ?, metadata_json = ?, updated_at = ?
|
|
408
|
-
WHERE id = ?`,[t.status,t.provider,t.model,JSON.stringify(t.metadata),t.now,t.runId])}finally{n.close()}}function
|
|
408
|
+
WHERE id = ?`,[t.status,t.provider,t.model,JSON.stringify(t.metadata),t.now,t.runId])}finally{n.close()}}function yi(e,t,n,r,i,s,c={}){let d=b(e);try{Te(d,{run_id:t,provider:r,model:i,input_tokens:n.input_tokens,output_tokens:n.output_tokens,cost_usd:n.cost_usd,metadata:c,created_at:s})}finally{d.close()}}async function _n(e){let t=e.prompt.trim();if(!t)throw Error("Knowledge prompt is required.");let n=(e.now??new Date).toISOString(),r=`run_${ln()}`,i=H(e.modelRef??"default",e.config),s=C(i);x(e.dbPath),Ei(e.dbPath,{runId:r,prompt:t,status:e.generate?"running":"dry_run",provider:e.generate?s.provider:"local",model:e.generate?s.model:"context-draft",metadata:{semantic:e.semantic===!0||e.fake===!0||Boolean(e.modelRef),approve_write:e.approveWrite===!0,generated:e.generate===!0},now:n});let{prompt:c,generate:d,approveWrite:g,now:a,...u}=e,o=await Ce({...u,query:t});dt(e.dbPath,{runId:r,level:"info",event:"context_retrieved",metadata:{results:o.results.length,citations:o.citations.length,warnings:o.warnings},now:n});let _=mi(t,o),l=!1,f="local",m="context-draft",E={input_tokens:ct(t)+o.excerpts.reduce((p,R)=>p+ct(R.text),0),output_tokens:ct(_),cost_usd:0},h=[...o.warnings];if(e.generate)try{if(e.fake)l=!0,f=s.provider,m=s.model,_=`Fake generated answer for: ${t}
|
|
409
409
|
|
|
410
|
-
${_}`;else{let{generateText:
|
|
411
|
-
VALUES (?, ?, ?, ?, ?, ?, ?)`,[r,t.event_type,t.action,t.target_uri??null,t.decision,JSON.stringify(t.metadata??{}),n]),r}function
|
|
412
|
-
VALUES (?, ?, ?, ?, ?, ?, ?)`,[`redact_${
|
|
413
|
-
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[r,t.action,t.target_uri??null,"approved",t.reason??null,t.approved_by??"local-cli",JSON.stringify(t.metadata??{}),n,n]),{id:r,status:"approved"}}function
|
|
410
|
+
${_}`;else{let{generateText:p}=await import("ai"),R=await Bt(i,{config:e.config,env:e.env}),v=await p({model:R,system:"You answer company knowledge-base prompts using only provided context and citation ids.",prompt:pi(t,o)});l=!0,f=s.provider,m=s.model,_=v.text;let w=we({provider:f,model:m,usage:v.usage,providerMetadata:v.providerMetadata});E={input_tokens:w.input_tokens,output_tokens:w.output_tokens,cost_usd:w.cost_usd}}}catch(p){throw dt(e.dbPath,{runId:r,level:"error",event:"answer_generation_failed",metadata:{message:p instanceof Error?p.message:String(p)},now:n}),dn(e.dbPath,{runId:r,status:"failed",provider:s.provider,model:s.model,metadata:{generated:!1,error:p instanceof Error?p.message:String(p)},now:n}),p}let T=hi(t,o),k={approved:e.approveWrite===!0,durable_writes_performed:!1,reason:e.approveWrite?"Approval flag recorded; durable wiki writing is deferred to the wiki compile task.":"Dry-run mode: proposed wiki updates require approval before durable writes."};return dt(e.dbPath,{runId:r,level:"info",event:l?"answer_generated":"answer_drafted",metadata:{provider:f,model:m,proposed_updates:T.length,durable_writes_performed:!1},now:n}),yi(e.dbPath,r,E,f,m,n,{generated:l,citations:o.citations.length}),dn(e.dbPath,{runId:r,status:l?"completed":"dry_run",provider:f,model:m,metadata:{generated:l,citations:o.citations.length,proposed_updates:T.length,approve_write:e.approveWrite===!0},now:n}),{run_id:r,prompt:t,generated:l,provider:f,model:m,answer:_,context:o,citations:o.citations,proposed_wiki_updates:T,write_policy:k,usage:E,warnings:h}}import{createHash as Ii,randomUUID as Li}from"crypto";import{existsSync as Ci,readFileSync as Di}from"fs";import{basename as Ui}from"path";function fn(e,t){if(!e)throw Error(t);return e}function ki(e){let n=e.slice(13).split("/").filter(Boolean),r=n[0];if(r!=="file"&&r!=="source")throw Error("Invalid open-files ref. Expected open-files://file/<id>, open-files://file/<id>/revision/<revision_id>, or open-files://source/<id>/path/<path>.");let i=fn(n[1],"Invalid open-files ref. Missing id.");if(r==="file"){if(n.length===2)return{kind:"open-files",uri:e,entity:r,id:i};if(n[2]==="revision"&&n[3]&&n.length===4)return{kind:"open-files",uri:e,entity:r,id:i,revision_id:decodeURIComponent(n[3])};throw Error("Invalid open-files file ref. Expected open-files://file/<id>/revision/<revision_id>.")}let s=n.indexOf("path"),c=s>=0?decodeURIComponent(n.slice(s+1).join("/")):void 0;return{kind:"open-files",uri:e,entity:r,id:i,path:c}}function bi(e){let t=new URL(e),n=fn(t.hostname,"Invalid s3 ref. Missing bucket."),r=decodeURIComponent(t.pathname.replace(/^\/+/,""));if(!r)throw Error("Invalid s3 ref. Missing object key.");return{kind:"s3",uri:e,bucket:n,key:r}}function vi(e){let t=new URL(e);return{kind:"file",uri:e,path:decodeURIComponent(t.pathname)}}function wi(e){let t=new URL(e);return{kind:"web",uri:e,url:t.toString()}}function M(e){if(e.startsWith("open-files://"))return ki(e);if(e.startsWith("s3://"))return bi(e);if(e.startsWith("file://"))return vi(e);if(e.startsWith("https://")||e.startsWith("http://"))return wi(e);throw Error(`Unsupported source ref scheme: ${e}`)}function gn(e,t=M(e)){if(t.kind==="open-files"&&t.entity==="file"&&t.revision_id)return e.replace(/\/revision\/[^/]+$/,"");return e}function mn(e){let t=M(e);return t.kind==="open-files"&&t.entity==="file"?t.revision_id??null:null}import{createHash as Ti,randomUUID as _t}from"crypto";import{relative as Ri,resolve as hn,sep as Si}from"path";function pn(e){let t=process.env[e];return t==="1"||t==="true"||t==="yes"}function En(e,t){let n=e,r=new Set(n.safety?.network?.allowed_s3_buckets??[]);if(e.storage.type==="s3"&&e.storage.s3?.bucket)r.add(e.storage.s3.bucket);if(process.env.HASNA_KNOWLEDGE_ALLOWED_S3_BUCKETS)for(let i of process.env.HASNA_KNOWLEDGE_ALLOWED_S3_BUCKETS.split(",").map((s)=>s.trim()).filter(Boolean))r.add(i);return{mode:e.mode,allowWriteRoots:[t.home,t.artifactsDir,t.cacheDir,t.exportsDir,t.indexesDir,t.logsDir,t.runsDir,t.schemasDir,t.wikiDir].map((i)=>hn(i)),readOnlySourceAccess:!0,network:{webSearchEnabled:n.safety?.network?.web_search_enabled??pn("HASNA_KNOWLEDGE_WEB_SEARCH"),s3ReadsEnabled:n.safety?.network?.s3_reads_enabled??pn("HASNA_KNOWLEDGE_ALLOW_S3_READS"),allowedS3Buckets:[...r].sort()},redaction:{enabled:n.safety?.redaction?.enabled??!0},approvals:{generatedWritesRequireApproval:n.safety?.approvals?.generated_writes_require_approval??!0}}}function xi(e,t){let n=Ri(e,t);return n===""||!n.startsWith("..")&&n!==".."&&!n.startsWith(`..${Si}`)}function Q(e,t){let n=hn(e);if(!t.allowWriteRoots.some((r)=>xi(r,n)))throw Error(`Safety policy denied write outside .hasna/apps/knowledge: ${e}`)}function J(e,t){let r=new URL(e).hostname;if(!t.network.s3ReadsEnabled)throw Error("Safety policy denied S3 read. Set safety.network.s3_reads_enabled=true or HASNA_KNOWLEDGE_ALLOW_S3_READS=1.");if(!t.network.allowedS3Buckets.includes(r))throw Error(`Safety policy denied S3 bucket "${r}". Add it to safety.network.allowed_s3_buckets or HASNA_KNOWLEDGE_ALLOWED_S3_BUCKETS.`)}function ie(e){if(!e.network.webSearchEnabled)throw Error("Safety policy denied web search. Set safety.network.web_search_enabled=true or HASNA_KNOWLEDGE_WEB_SEARCH=1.")}var Oi=[{type:"private_key_block",severity:"high",regex:/-----BEGIN [A-Z ]*PRIVATE KEY-----[\s\S]*?-----END [A-Z ]*PRIVATE KEY-----/g,replacement:"[REDACTED:private_key_block]"},{type:"secret_assignment",severity:"high",regex:/\b(?:api[_-]?key|secret|token|password)\s*[:=]\s*['"]?[^'"\s]{8,}/gi,replacement:"[REDACTED:secret_assignment]"},{type:"openai_api_key",severity:"high",regex:/\bsk-[A-Za-z0-9_-]{20,}\b/g,replacement:"[REDACTED:openai_api_key]"},{type:"anthropic_api_key",severity:"high",regex:/\bsk-ant-[A-Za-z0-9_-]{20,}\b/g,replacement:"[REDACTED:anthropic_api_key]"},{type:"aws_access_key_id",severity:"high",regex:/\bA(?:KIA|SIA)[A-Z0-9]{16}\b/g,replacement:"[REDACTED:aws_access_key_id]"}];function De(e,t){if(t&&!t.redaction.enabled)return{text:e,findings:[]};let n=e,r=[];for(let i of Oi)n=n.replace(i.regex,(s,...c)=>{let d=typeof c.at(-2)==="number"?c.at(-2):n.indexOf(s);return r.push({type:i.type,severity:i.severity,start:Math.max(0,d),end:Math.max(0,d+s.length)}),i.replacement});return{text:n,findings:r}}function Ni(e){return`audit_${Ti("sha256").update(`${e.event_type}\x00${e.action}\x00${e.target_uri??""}\x00${e.created_at??""}\x00${JSON.stringify(e.metadata??{})}\x00${_t()}`).digest("hex").slice(0,24)}`}function I(e,t){let n=t.created_at??new Date().toISOString(),r=Ni({...t,created_at:n});return e.run(`INSERT INTO audit_events (id, event_type, action, target_uri, decision, metadata_json, created_at)
|
|
411
|
+
VALUES (?, ?, ?, ?, ?, ?, ?)`,[r,t.event_type,t.action,t.target_uri??null,t.decision,JSON.stringify(t.metadata??{}),n]),r}function Ue(e,t){let n=t.created_at??new Date().toISOString();for(let r of t.findings)e.run(`INSERT INTO redaction_findings (id, source_uri, run_id, severity, finding_type, metadata_json, created_at)
|
|
412
|
+
VALUES (?, ?, ?, ?, ?, ?, ?)`,[`redact_${_t()}`,t.source_uri??null,t.run_id??null,r.severity,r.type,JSON.stringify({...t.metadata??{},start:r.start,end:r.end}),n]);return t.findings.length}function yn(e,t){let n=t.created_at??new Date().toISOString(),r=`approval_${_t()}`;return e.run(`INSERT INTO approval_gates (id, action, target_uri, status, reason, approved_by, metadata_json, created_at, updated_at)
|
|
413
|
+
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[r,t.action,t.target_uri??null,"approved",t.reason??null,t.approved_by??"local-cli",JSON.stringify(t.metadata??{}),n,n]),{id:r,status:"approved"}}function Ai(e,t,n){let r=e.query(`SELECT id FROM approval_gates
|
|
414
414
|
WHERE action = ? AND status = 'approved' AND (target_uri IS NULL OR target_uri = ? OR ? IS NULL)
|
|
415
|
-
ORDER BY updated_at DESC LIMIT 1`).get(t,n??null,n??null);return Boolean(r)}function
|
|
415
|
+
ORDER BY updated_at DESC LIMIT 1`).get(t,n??null,n??null);return Boolean(r)}function kn(e,t,n,r){let i=n==="generated_write"&&t.approvals.generatedWritesRequireApproval,s=!i||Ai(e,n,r);return{action:n,target_uri:r??null,approval_required:i,approved:s,decision:s?"allow":"requires_approval"}}function Pe(e,t){return`${e}_${Ii("sha256").update(t).digest("hex").slice(0,20)}`}function se(e){return e&&typeof e==="object"&&!Array.isArray(e)?e:void 0}function N(e){return typeof e==="string"&&e.length>0?e:void 0}function Pi(e){let t=N(e.source_ref)??N(e.source_uri)??N(e.uri);if(t)return t;let n=N(e.file_id);if(n){let s=N(e.revision_id)??N(e.revision),c=`open-files://file/${encodeURIComponent(n)}`;return s?`${c}/revision/${encodeURIComponent(s)}`:c}let r=N(e.source_id),i=N(e.path);if(r&&i)return`open-files://source/${encodeURIComponent(r)}/path/${encodeURIComponent(i)}`;throw Error("Outbox event is missing source_ref, file_id, or source_id/path.")}function Fi(e,t){if(t.kind==="open-files"&&t.entity==="file"&&t.revision_id)return e.replace(/\/revision\/[^/]+$/,"");return e}function Mi(e){return N(e.hash)??N(e.checksum)??N(e.sha256)??null}function ji(e,t,n){return N(e.revision_id)??N(e.revision)??N(e.version_id)??(t.kind==="open-files"?t.revision_id:void 0)??n??null}function Ki(e){return(N(e.event)??N(e.type)??N(e.action)??N(e.change_type)??"changed").toLowerCase()}function $i(e){let t=N(e.path);return N(e.title)??N(e.name)??(t?Ui(t):null)}function Wi(e,t){let n=Pi(e),r=M(n),i=Mi(e);return{raw:e,eventType:Ki(e),sourceRef:n,sourceUri:Fi(n,r),kind:r.kind,title:$i(e),revision:ji(e,r,i),hash:i,status:N(e.status)?.toLowerCase()??null,updatedAt:N(e.updated_at)??t,acl:e.permissions??e.acl??void 0}}function Xi(e){let t=e.trim();if(!t)return[];if(t.startsWith("[")){let n=JSON.parse(t);if(!Array.isArray(n))throw Error("Outbox array parse failed.");return n.map((r)=>{let i=se(r);if(!i)throw Error("Outbox array entries must be objects.");return i})}if(t.startsWith("{"))try{let n=JSON.parse(t),r=se(n);if(!r)throw Error("Outbox object parse failed.");if(Array.isArray(r.events))return r.events.map((i)=>{let s=se(i);if(!s)throw Error("Outbox events entries must be objects.");return s});if("source_ref"in r||"source_uri"in r||"file_id"in r)return[r]}catch(n){let r=t.split(/\r?\n/).filter((i)=>i.trim().length>0);if(r.length<=1)throw n;return r.map((i)=>{let s=se(JSON.parse(i));if(!s)throw Error("Outbox JSONL entries must be objects.");return s})}return t.split(/\r?\n/).filter((n)=>n.trim().length>0).map((n)=>{let r=se(JSON.parse(n));if(!r)throw Error("Outbox JSONL entries must be objects.");return r})}async function qi(e,t,n){let r=new URL(e),i=r.hostname,s=decodeURIComponent(r.pathname.replace(/^\/+/,""));if(!i||!s)throw Error(`Invalid S3 outbox URI: ${e}`);if(n)J(e,n);let[{S3Client:c,GetObjectCommand:d},{fromIni:g}]=await Promise.all([import("@aws-sdk/client-s3"),import("@aws-sdk/credential-providers")]),a=t?.storage.type==="s3"&&t.storage.s3?.bucket===i?t.storage.s3:void 0,o=await new c({region:a?.region,credentials:a?.profile?g({profile:a.profile}):void 0,maxAttempts:a?.max_attempts}).send(new d({Bucket:i,Key:s}));if(!o.Body)return"";return await o.Body.transformToString()}async function Hi(e,t,n){if(e.startsWith("s3://"))return qi(e,t,n);if(!Ci(e))throw Error(`Outbox not found: ${e}`);return Di(e,"utf8")}function bn(e,t){let n={};if(e)try{n=se(JSON.parse(e))??{}}catch{n={}}return JSON.stringify({...n,...t})}function Bi(e,t,n){let r=Pe("src",t.sourceUri);e.run(`INSERT INTO sources (id, uri, kind, title, metadata_json, acl_json, created_at, updated_at)
|
|
416
416
|
VALUES (?, ?, ?, ?, ?, ?, ?, ?)
|
|
417
417
|
ON CONFLICT(uri) DO UPDATE SET
|
|
418
418
|
kind = excluded.kind,
|
|
419
419
|
title = COALESCE(excluded.title, sources.title),
|
|
420
|
-
updated_at = excluded.updated_at`,[r,t.sourceUri,t.kind,t.title,JSON.stringify({source_ref:t.sourceRef,source_uri:t.sourceUri,status:t.status,last_outbox_event:t.eventType}),JSON.stringify(t.acl??{}),n,t.updatedAt]);let i=e.query("SELECT id, metadata_json, acl_json FROM sources WHERE uri = ?").get(t.sourceUri);if(!i)throw Error(`Failed to upsert source for outbox event: ${t.sourceUri}`);let s={source_ref:t.sourceRef,source_uri:t.sourceUri,last_outbox_event:t.eventType,last_outbox_at:t.updatedAt};if(t.status)s.status=t.status;if(
|
|
420
|
+
updated_at = excluded.updated_at`,[r,t.sourceUri,t.kind,t.title,JSON.stringify({source_ref:t.sourceRef,source_uri:t.sourceUri,status:t.status,last_outbox_event:t.eventType}),JSON.stringify(t.acl??{}),n,t.updatedAt]);let i=e.query("SELECT id, metadata_json, acl_json FROM sources WHERE uri = ?").get(t.sourceUri);if(!i)throw Error(`Failed to upsert source for outbox event: ${t.sourceUri}`);let s={source_ref:t.sourceRef,source_uri:t.sourceUri,last_outbox_event:t.eventType,last_outbox_at:t.updatedAt};if(t.status)s.status=t.status;if(N(t.raw.path))s.path=t.raw.path;return e.run("UPDATE sources SET metadata_json = ?, acl_json = CASE WHEN ? IS NULL THEN acl_json ELSE ? END, updated_at = ? WHERE id = ?",[bn(i.metadata_json,s),t.acl===void 0?null:JSON.stringify(t.acl),t.acl===void 0?null:JSON.stringify(t.acl),t.updatedAt,i.id]),i.id}function Ji(e,t,n,r){if(!n.revision)return null;let i=Pe("rev",`${t}\x00${n.revision}`),s={source_ref:n.sourceRef,source_uri:n.sourceUri,status:n.status,last_outbox_event:n.eventType,reindex_required:!0};return e.run(`INSERT INTO source_revisions (id, source_id, revision, hash, extracted_text_uri, metadata_json, created_at)
|
|
421
421
|
VALUES (?, ?, ?, ?, ?, ?, ?)
|
|
422
422
|
ON CONFLICT(source_id, revision) DO UPDATE SET
|
|
423
423
|
hash = COALESCE(excluded.hash, source_revisions.hash),
|
|
424
|
-
metadata_json = excluded.metadata_json`,[i,t,n.revision,n.hash,
|
|
425
|
-
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[s,"open-files-outbox",e.input,"completed","local","open-files-outbox",JSON.stringify({path:e.input,events:r.length}),t,t]);let
|
|
426
|
-
VALUES (?, ?, ?, ?, ?, ?)`,[
|
|
427
|
-
VALUES (?, ?, ?, ?, 0, 0, 0, ?, ?)`,[
|
|
428
|
-
`);if(!r.trim())return[];let i=[],s=0;while(s<r.length){let
|
|
424
|
+
metadata_json = excluded.metadata_json`,[i,t,n.revision,n.hash,N(n.raw.extracted_text_ref)??null,JSON.stringify(s),r]),e.query("SELECT id FROM source_revisions WHERE source_id = ? AND revision = ?").get(t,n.revision)?.id??null}function Gi(e,t,n){if(n.revision)return e.query("SELECT id FROM source_revisions WHERE source_id = ? AND revision = ?").all(t,n.revision).map((r)=>r.id);if(n.hash)return e.query("SELECT id FROM source_revisions WHERE source_id = ? AND hash = ?").all(t,n.hash).map((r)=>r.id);return e.query("SELECT id FROM source_revisions WHERE source_id = ?").all(t).map((r)=>r.id)}function zi(e,t){let n=e.query("SELECT id FROM chunks WHERE source_revision_id = ?").all(t),r=0,i=0;for(let c of n){let d=e.query("SELECT COUNT(*) AS n FROM chunk_embeddings WHERE chunk_id = ?").get(c.id);r+=d?.n??0;let g=e.query("SELECT COUNT(*) AS n FROM vector_index_entries WHERE chunk_id = ?").get(c.id);i+=g?.n??0,e.run("DELETE FROM vector_index_entries WHERE chunk_id = ?",[c.id]),e.run("DELETE FROM chunk_embeddings WHERE chunk_id = ?",[c.id]),e.run("DELETE FROM chunks_fts WHERE chunk_id = ?",[c.id])}e.run("DELETE FROM chunks WHERE source_revision_id = ?",[t]);let s=e.query("SELECT metadata_json FROM source_revisions WHERE id = ?").get(t);return e.run("UPDATE source_revisions SET metadata_json = ? WHERE id = ?",[bn(s?.metadata_json,{reindex_required:!0,invalidated_at:new Date().toISOString()}),t]),{chunksDeleted:n.length,embeddingsDeleted:r,vectorEntriesDeleted:i}}function Yi(e,t){return t==="deleted"||["delete","deleted","remove","removed"].includes(e)}function Vi(e){return["move","moved","rename","renamed","path_changed"].includes(e)}function Qi(e){return["permission","permissions","permission_changed","acl_changed"].includes(e)}async function vn(e){let t=(e.now??new Date).toISOString();if(e.safetyPolicy)Q(e.dbPath,e.safetyPolicy);x(e.dbPath);let n=await Hi(e.input,e.config,e.safetyPolicy),r=Xi(n),i=b(e.dbPath),s=`run_${Li()}`;try{return i.transaction(()=>{i.run(`INSERT INTO runs (id, type, prompt, status, provider, model, metadata_json, created_at, updated_at)
|
|
425
|
+
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[s,"open-files-outbox",e.input,"completed","local","open-files-outbox",JSON.stringify({path:e.input,events:r.length}),t,t]);let c=new Set,d=new Set,g=0,a=0,u=0,o=0,_=0,l=0,f=0;return I(i,{event_type:"source_read",action:e.input.startsWith("s3://")?"s3_outbox_read":"local_outbox_read",target_uri:e.input,decision:"allow",metadata:{events:r.length,read_only:!0},created_at:t}),r.forEach((m,E)=>{let h=Wi(m,t),T=Bi(i,h,t);c.add(T);let k=Ji(i,T,h,t);if(k)d.add(k);let p=Gi(i,T,h);for(let R of p){d.add(R);let v=zi(i,R);g+=v.chunksDeleted,a+=v.embeddingsDeleted,u+=v.vectorEntriesDeleted,o+=1}if(Yi(h.eventType,h.status))_+=1;if(Vi(h.eventType))l+=1;if(Qi(h.eventType)||h.acl!==void 0)f+=1;i.run(`INSERT INTO run_events (id, run_id, level, event, metadata_json, created_at)
|
|
426
|
+
VALUES (?, ?, ?, ?, ?, ?)`,[Pe("evt",`${s}\x00${E}\x00${h.sourceRef}\x00${h.eventType}`),s,"info",h.eventType,JSON.stringify({source_ref:h.sourceRef,source_uri:h.sourceUri,revision:h.revision,hash:h.hash,status:h.status,affected_revisions:p.length}),h.updatedAt])}),i.run(`INSERT INTO provider_usage (id, run_id, provider, model, input_tokens, output_tokens, cost_usd, metadata_json, created_at)
|
|
427
|
+
VALUES (?, ?, ?, ?, 0, 0, 0, ?, ?)`,[Pe("usage",s),s,"local","open-files-outbox",JSON.stringify({note:"No model provider used for outbox invalidation."}),t]),I(i,{event_type:"write",action:"knowledge_outbox_invalidation",target_uri:e.dbPath,decision:"allow",metadata:{run_id:s,events:r.length,sources:c.size,revisions:d.size,chunks_deleted:g,embeddings_deleted:a,vector_entries_deleted:u},created_at:t}),{path:e.input,db_path:e.dbPath,run_id:s,events_seen:r.length,sources_touched:c.size,revisions_touched:d.size,chunks_deleted:g,embeddings_deleted:a,vector_entries_deleted:u,stale_revisions:o,deleted_sources:_,moved_sources:l,permission_updates:f}})()}finally{i.close()}}import{createHash as Zi}from"crypto";import{existsSync as es,readFileSync as ts}from"fs";import{basename as ns}from"path";function ft(e,t){return`${e}_${Zi("sha256").update(t).digest("hex").slice(0,20)}`}function oe(e){return e&&typeof e==="object"&&!Array.isArray(e)?e:void 0}function S(e){return typeof e==="string"&&e.length>0?e:void 0}function rs(e){return typeof e==="number"&&Number.isFinite(e)?e:void 0}function is(e){let t=S(e.source_ref)??S(e.source_uri)??S(e.uri);if(t)return t;let n=S(e.file_id);if(n){let s=S(e.revision_id)??S(e.revision),c=`open-files://file/${encodeURIComponent(n)}`;return s?`${c}/revision/${encodeURIComponent(s)}`:c}let r=S(e.source_id),i=S(e.path);if(r&&i)return`open-files://source/${encodeURIComponent(r)}/path/${encodeURIComponent(i)}`;throw Error("Manifest item is missing source_ref, file_id, or source_id/path.")}function ss(e,t){if(t.kind==="open-files"&&t.entity==="file"&&t.revision_id)return e.replace(/\/revision\/[^/]+$/,"");return e}function os(e){let t=S(e.extracted_text)??S(e.text)??S(e.content_text)??S(e.markdown);if(t!==void 0)return t;let n=e.content;return typeof n==="string"?n:null}function as(e){let t=S(e.extracted_text_ref)??S(e.extracted_text_uri)??S(e.text_ref);if(t)return t;let n=oe(e.content);return S(n?.extracted_text_ref)??S(n?.extracted_text_uri)??null}function us(e){let t=S(e.path);return S(e.title)??S(e.name)??(t?ns(t):null)}function cs(e){return S(e.hash)??S(e.checksum)??S(e.sha256)??null}function ds(e,t,n){return S(e.revision_id)??S(e.revision)??S(e.version_id)??(t.kind==="open-files"?t.revision_id:void 0)??n??S(e.updated_at)??"current"}function ls(e,t){let n={};for(let[r,i]of Object.entries(e)){if(["text","content","content_text","extracted_text","markdown"].includes(r))continue;n[r]=i}return n.source_ref=t.sourceRef,n.source_uri=t.sourceUri,n.status=t.status,n}function _s(e,t){let n=is(e),r=M(n),i=ss(n,r),s=cs(e),c=S(e.status)??"active";return{raw:e,sourceRef:n,sourceUri:i,kind:r.kind,title:us(e),revision:ds(e,r,s),hash:s,extractedTextUri:as(e),text:os(e),metadata:ls(e,{sourceRef:n,sourceUri:i,status:c}),acl:e.permissions??e.acl??{},status:c,updatedAt:S(e.updated_at)??t}}function fs(e){let t=e.trim();if(!t)return[];if(t.startsWith("[")){let n=JSON.parse(t);if(!Array.isArray(n))throw Error("Manifest array parse failed.");return n.map((r)=>{let i=oe(r);if(!i)throw Error("Manifest array entries must be objects.");return i})}if(t.startsWith("{"))try{let n=JSON.parse(t),r=oe(n);if(!r)throw Error("Manifest object parse failed.");if(Array.isArray(r.items))return r.items.map((i)=>{let s=oe(i);if(!s)throw Error("Manifest items entries must be objects.");return s});if("source_ref"in r||"source_uri"in r||"file_id"in r)return[r]}catch(n){let r=t.split(/\r?\n/).filter((i)=>i.trim().length>0);if(r.length<=1)throw n;return r.map((i)=>{let s=oe(JSON.parse(i));if(!s)throw Error("Manifest JSONL entries must be objects.");return s})}return t.split(/\r?\n/).filter((n)=>n.trim().length>0).map((n)=>{let r=oe(JSON.parse(n));if(!r)throw Error("Manifest JSONL entries must be objects.");return r})}async function gs(e,t,n){let r=new URL(e),i=r.hostname,s=decodeURIComponent(r.pathname.replace(/^\/+/,""));if(!i||!s)throw Error(`Invalid S3 manifest URI: ${e}`);if(n)J(e,n);let[{S3Client:c,GetObjectCommand:d},{fromIni:g}]=await Promise.all([import("@aws-sdk/client-s3"),import("@aws-sdk/credential-providers")]),a=t?.storage.type==="s3"&&t.storage.s3?.bucket===i?t.storage.s3:void 0,o=await new c({region:a?.region,credentials:a?.profile?g({profile:a.profile}):void 0,maxAttempts:a?.max_attempts}).send(new d({Bucket:i,Key:s}));if(!o.Body)return"";return await o.Body.transformToString()}async function ms(e,t,n){if(e.startsWith("s3://"))return gs(e,t,n);if(!es(e))throw Error(`Manifest not found: ${e}`);return ts(e,"utf8")}function ps(e,t,n){let r=e.replace(/\r\n/g,`
|
|
428
|
+
`);if(!r.trim())return[];let i=[],s=0;while(s<r.length){let c=Math.min(r.length,s+t),d=c;if(c<r.length){let a=r.lastIndexOf(`
|
|
429
429
|
|
|
430
|
-
`,
|
|
430
|
+
`,c),u=r.lastIndexOf(". ",c),o=Math.max(a,u);if(o>s+Math.floor(t*0.5))d=o+(o===a?2:1)}let g=r.slice(s,d).trim();if(g)i.push({ordinal:i.length,text:g,startOffset:s,endOffset:d});if(d>=r.length)break;s=Math.max(0,d-n)}return i}function hs(e){let t=e.trim().split(/\s+/).filter(Boolean).length;return Math.max(1,Math.ceil(t*1.25))}function Es(e,t){let n=e.query("SELECT id FROM chunks WHERE source_revision_id = ?").all(t);for(let r of n)e.run("DELETE FROM chunks_fts WHERE chunk_id = ?",[r.id]);return e.run("DELETE FROM chunks WHERE source_revision_id = ?",[t]),n.length}function ys(e,t,n){let r=ft("src",t.sourceUri);e.run(`INSERT INTO sources (id, uri, kind, title, metadata_json, acl_json, created_at, updated_at)
|
|
431
431
|
VALUES (?, ?, ?, ?, ?, ?, ?, ?)
|
|
432
432
|
ON CONFLICT(uri) DO UPDATE SET
|
|
433
433
|
kind = excluded.kind,
|
|
434
434
|
title = excluded.title,
|
|
435
435
|
metadata_json = excluded.metadata_json,
|
|
436
436
|
acl_json = excluded.acl_json,
|
|
437
|
-
updated_at = excluded.updated_at`,[r,t.sourceUri,t.kind,t.title,JSON.stringify(t.metadata),JSON.stringify(t.acl??{}),n,t.updatedAt]);let i=e.query("SELECT id FROM sources WHERE uri = ?").get(t.sourceUri);if(!i)throw Error(`Failed to upsert source: ${t.sourceUri}`);return i.id}function
|
|
437
|
+
updated_at = excluded.updated_at`,[r,t.sourceUri,t.kind,t.title,JSON.stringify(t.metadata),JSON.stringify(t.acl??{}),n,t.updatedAt]);let i=e.query("SELECT id FROM sources WHERE uri = ?").get(t.sourceUri);if(!i)throw Error(`Failed to upsert source: ${t.sourceUri}`);return i.id}function ks(e,t,n,r){let i=ft("rev",`${t}\x00${n.revision}`);e.run(`INSERT INTO source_revisions (id, source_id, revision, hash, extracted_text_uri, metadata_json, created_at)
|
|
438
438
|
VALUES (?, ?, ?, ?, ?, ?, ?)
|
|
439
439
|
ON CONFLICT(source_id, revision) DO UPDATE SET
|
|
440
440
|
hash = excluded.hash,
|
|
441
441
|
extracted_text_uri = excluded.extracted_text_uri,
|
|
442
|
-
metadata_json = excluded.metadata_json`,[i,t,n.revision,n.hash,n.extractedTextUri,JSON.stringify(n.metadata),r]);let s=e.query("SELECT id FROM source_revisions WHERE source_id = ? AND revision = ?").get(t,n.revision);if(!s)throw Error(`Failed to upsert source revision: ${n.sourceRef}`);return s.id}function
|
|
443
|
-
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,[
|
|
442
|
+
metadata_json = excluded.metadata_json`,[i,t,n.revision,n.hash,n.extractedTextUri,JSON.stringify(n.metadata),r]);let s=e.query("SELECT id FROM source_revisions WHERE source_id = ? AND revision = ?").get(t,n.revision);if(!s)throw Error(`Failed to upsert source revision: ${n.sourceRef}`);return s.id}function bs(e,t,n,r,i,s,c){if(!n.text||n.status.toLowerCase()==="deleted")return{chunksInserted:0,redactions:0};let d=De(n.text,c);if(d.findings.length>0)Ue(e,{source_uri:n.sourceUri,findings:d.findings,metadata:{source_ref:n.sourceRef,revision:n.revision},created_at:r}),I(e,{event_type:"redaction",action:"source_text_redact",target_uri:n.sourceUri,decision:"redacted",metadata:{findings:d.findings.length,source_ref:n.sourceRef,revision:n.revision},created_at:r});let g=ps(d.text,i,s);for(let a of g){let u=ft("chk",`${t}\x00${a.ordinal}\x00${a.text}`),o=B({source_ref:n.sourceRef,source_uri:n.sourceUri,source_kind:n.kind,source_revision_id:t,revision:n.revision,hash:n.hash,chunk_id:u,start_offset:a.startOffset,end_offset:a.endOffset,status:n.status,resolver:"open-files-read-only"}),_=Jt({source_ref:n.sourceRef,source_uri:n.sourceUri,source_kind:n.kind,source_revision_id:t,revision:n.revision,hash:n.hash,status:n.status,path:S(n.raw.path)??null,mime:S(n.raw.mime)??S(n.raw.content_type)??null,size:rs(n.raw.size)??null},o);e.run(`INSERT INTO chunks (id, source_revision_id, kind, ordinal, text, token_count, start_offset, end_offset, metadata_json, created_at)
|
|
443
|
+
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,[u,t,"source",a.ordinal,a.text,hs(a.text),a.startOffset,a.endOffset,JSON.stringify(_),r]),e.run("INSERT INTO chunks_fts (chunk_id, text, title, source_uri) VALUES (?, ?, ?, ?)",[u,a.text,n.title??"",n.sourceUri])}return{chunksInserted:g.length,redactions:d.findings.length}}async function wn(e){let t=e.now??new Date;if(e.safetyPolicy)Q(e.dbPath,e.safetyPolicy);x(e.dbPath);let n=await ms(e.input,e.config,e.safetyPolicy),r=fs(n);return fe({dbPath:e.dbPath,items:r,sourceLabel:e.input,safetyPolicy:e.safetyPolicy,now:t,maxChunkChars:e.maxChunkChars,chunkOverlapChars:e.chunkOverlapChars})}async function fe(e){let t=(e.now??new Date).toISOString(),n=e.maxChunkChars??4000,r=e.chunkOverlapChars??200;if(n<500)throw Error("maxChunkChars must be at least 500.");if(r<0||r>=n)throw Error("chunkOverlapChars must be less than maxChunkChars.");if(e.safetyPolicy)Q(e.dbPath,e.safetyPolicy);x(e.dbPath);let i=b(e.dbPath);try{return i.transaction(()=>{let c=new Set,d=new Set,g=0,a=0,u=0,o=0;I(i,{event_type:"source_read",action:e.readAction??(e.sourceLabel.startsWith("s3://")?"s3_manifest_read":"local_manifest_read"),target_uri:e.sourceLabel,decision:"allow",metadata:{items:e.items.length,read_only:!0},created_at:t});for(let _ of e.items){let l=_s(_,t),f=ys(i,l,t),m=ks(i,f,l,t);if(c.add(f),d.add(m),l.text||l.status.toLowerCase()==="deleted")a+=Es(i,m);let E=bs(i,m,l,t,n,r,e.safetyPolicy);g+=E.chunksInserted,u+=E.redactions}return I(i,{event_type:"write",action:"knowledge_manifest_ingest",target_uri:e.dbPath,decision:"allow",metadata:{items:e.items.length,sources:c.size,revisions:d.size,chunks_inserted:g,redactions:u},created_at:t}),{path:e.sourceLabel,db_path:e.dbPath,items_seen:e.items.length,sources_upserted:c.size,revisions_upserted:d.size,chunks_inserted:g,chunks_deleted:a,redactions:u,skipped:o}})()}finally{i.close()}}import{createHash as Os}from"crypto";import{existsSync as Ns,readFileSync as As}from"fs";import{basename as je}from"path";function Fe(e){if(!e)return{};try{let t=JSON.parse(e);return t&&typeof t==="object"&&!Array.isArray(t)?t:{}}catch{return{}}}function Z(e,t){for(let n of t){let r=e[n];if(typeof r==="string"&&r.length>0)return r}return null}function Tn(e,t){for(let n of t){let r=e[n];if(typeof r==="number"&&Number.isFinite(r))return r}return null}function vs(e,t){let n=e.mode;if(typeof n==="string"&&n!=="read_only")throw Error(`Source resolver denied ${t}. Permission mode is ${n}, expected read_only.`);let r=e.denied_purposes;if(Array.isArray(r)&&r.includes(t))throw Error(`Source resolver denied ${t}. Purpose is explicitly denied.`);let i=e.allowed_purposes;if(Array.isArray(i)&&i.length>0&&!i.includes(t))throw Error(`Source resolver denied ${t}. Allowed purposes: ${i.join(", ")}`)}function ws(e,t,n){if(!t)return n;try{let r=M(e);if(r.kind==="open-files"&&r.entity==="file")return`${e}/revision/${encodeURIComponent(t.revision)}`}catch{return n}return n}function Ts(e,t,n){return e.query(`SELECT id, uri, kind, title, metadata_json, acl_json, updated_at
|
|
444
444
|
FROM sources
|
|
445
445
|
WHERE uri = ? OR uri = ?
|
|
446
446
|
ORDER BY CASE WHEN uri = ? THEN 0 ELSE 1 END
|
|
447
|
-
LIMIT 1`).get(t,n,t)??null}function
|
|
447
|
+
LIMIT 1`).get(t,n,t)??null}function Rs(e,t,n){if(n)return e.query(`SELECT id, revision, hash, extracted_text_uri, metadata_json, created_at
|
|
448
448
|
FROM source_revisions
|
|
449
449
|
WHERE source_id = ? AND revision = ?
|
|
450
450
|
LIMIT 1`).get(t,n)??null;return e.query(`SELECT id, revision, hash, extracted_text_uri, metadata_json, created_at
|
|
451
451
|
FROM source_revisions
|
|
452
452
|
WHERE source_id = ?
|
|
453
453
|
ORDER BY created_at DESC, revision DESC
|
|
454
|
-
LIMIT 1`).get(t)??null}function Ss(e,t){if(!t)return 0;return e.query("SELECT COUNT(*) AS n FROM chunks WHERE source_revision_id = ?").get(t)?.n??0}function
|
|
454
|
+
LIMIT 1`).get(t)??null}function Ss(e,t){if(!t)return 0;return e.query("SELECT COUNT(*) AS n FROM chunks WHERE source_revision_id = ?").get(t)?.n??0}function xs(e,t,n){if(!t||n<=0)return[];return e.query(`SELECT id, kind, ordinal, text, token_count, start_offset, end_offset, metadata_json
|
|
455
455
|
FROM chunks
|
|
456
456
|
WHERE source_revision_id = ?
|
|
457
457
|
ORDER BY ordinal ASC
|
|
458
|
-
LIMIT ?`).all(t,n)}async function
|
|
458
|
+
LIMIT ?`).all(t,n)}async function Me(e){let t=e.purpose??"knowledge_answer",n=Math.max(0,Math.min(e.limit??10,100)),r=(e.now??new Date).toISOString(),i=M(e.sourceRef),s=gn(e.sourceRef,i),c=mn(e.sourceRef);if(e.safetyPolicy){if(!e.safetyPolicy.readOnlySourceAccess)throw Error("Safety policy denied source resolution.");Q(e.dbPath,e.safetyPolicy)}x(e.dbPath);let d=b(e.dbPath);try{return d.transaction(()=>{let g=Ts(d,s,e.sourceRef);if(!g)return I(d,{event_type:"source_read",action:"open_files_resolve_missing",target_uri:e.sourceRef,decision:"allow",metadata:{purpose:t,read_only:!0,source_uri:s},created_at:r}),{source_ref:e.sourceRef,source_uri:s,purpose:t,read_only:!0,resolved:!1,resolver:{name:"open-files-read-only",mode:"local_catalog",contract:"open-files-knowledge-source-v1"},source:null,revision:null,content:{mime:null,size:null,hash:null,text_available:!1,chunks_total:0,chunks_returned:0,char_count_returned:0,extracted_text_ref:null,bytes_available:!1,bytes_exposed:!1},chunks:[],citations:[]};let a=Fe(g.metadata_json),u=Fe(g.acl_json);try{vs(u,t)}catch(p){throw I(d,{event_type:"source_read",action:"open_files_resolve",target_uri:e.sourceRef,decision:"deny",metadata:{purpose:t,read_only:!0,source_uri:g.uri,error:p instanceof Error?p.message:String(p)},created_at:r}),p}let o=Rs(d,g.id,c),_=Fe(o?.metadata_json),l=Ss(d,o?.id??null),f=xs(d,o?.id??null,n),m=ws(g.uri,o,e.sourceRef),E=f.map((p)=>{let R=Fe(p.metadata_json),v={resolver:"open-files-read-only",mode:"local_catalog",purpose:t,read_only:!0,source_ref:Z(R,["source_ref"])??m,source_uri:g.uri,source_revision_id:o?.id??null,revision:o?.revision??null,hash:o?.hash??Z(R,["hash"]),chunk_id:p.id,start_offset:p.start_offset,end_offset:p.end_offset,resolved_at:r},w=B({source_ref:v.source_ref,source_uri:v.source_uri,source_kind:g.kind,source_revision_id:v.source_revision_id,revision:v.revision,hash:v.hash,chunk_id:p.id,start_offset:p.start_offset,end_offset:p.end_offset,status:Z(R,["status"]),resolver:v.resolver});return{id:p.id,kind:p.kind,ordinal:p.ordinal,text:p.text,token_count:p.token_count,start_offset:p.start_offset,end_offset:p.end_offset,metadata:R,evidence:v,provenance:w}}),h=E.map((p)=>({source_ref:p.evidence.source_ref,source_uri:g.uri,chunk_id:p.id,quote:p.text.slice(0,500),start_offset:p.start_offset,end_offset:p.end_offset,evidence:p.evidence,provenance:p.provenance}));I(d,{event_type:"source_read",action:"open_files_resolve",target_uri:e.sourceRef,decision:"allow",metadata:{purpose:t,read_only:!0,source_uri:g.uri,revision:o?.revision??null,chunks_returned:E.length,chunks_total:l},created_at:r});let T=Z(a,["mime","content_type"])??Z(_,["mime","content_type"]),k=Tn(a,["size","size_bytes"])??Tn(_,["size","size_bytes"]);return{source_ref:m,source_uri:g.uri,purpose:t,read_only:!0,resolved:!0,resolver:{name:"open-files-read-only",mode:"local_catalog",contract:"open-files-knowledge-source-v1"},source:{id:g.id,uri:g.uri,kind:g.kind,title:g.title,metadata:a,permissions:u,updated_at:g.updated_at},revision:o?{id:o.id,revision:o.revision,hash:o.hash,extracted_text_uri:o.extracted_text_uri,metadata:_,created_at:o.created_at,reindex_required:_.reindex_required===!0}:null,content:{mime:T,size:k,hash:o?.hash??Z(a,["hash","checksum","sha256"]),text_available:l>0,chunks_total:l,chunks_returned:E.length,char_count_returned:E.reduce((p,R)=>p+R.text.length,0),extracted_text_ref:o?.extracted_text_uri??Z(_,["extracted_text_ref","extracted_text_uri"]),bytes_available:!1,bytes_exposed:!1},chunks:E,citations:h}})()}finally{d.close()}}function ae(e){return`sha256:${Os("sha256").update(e).digest("hex")}`}function Is(e){return e.replace(/<script[\s\S]*?<\/script>/gi," ").replace(/<style[\s\S]*?<\/style>/gi," ").replace(/<[^>]+>/g," ").replace(/ /g," ").replace(/&/g,"&").replace(/</g,"<").replace(/>/g,">").replace(/\s+\n/g,`
|
|
459
459
|
`).replace(/\n\s+/g,`
|
|
460
|
-
`).replace(/[ \t]{2,}/g," ").trim()}async function
|
|
460
|
+
`).replace(/[ \t]{2,}/g," ").trim()}async function Ls(e,t,n){let r=new URL(e),i=r.hostname,s=decodeURIComponent(r.pathname.replace(/^\/+/,""));if(!i||!s)throw Error(`Invalid S3 source URI: ${e}`);if(n)J(e,n);let[{S3Client:c,GetObjectCommand:d},{fromIni:g}]=await Promise.all([import("@aws-sdk/client-s3"),import("@aws-sdk/credential-providers")]),a=t?.storage.type==="s3"&&t.storage.s3?.bucket===i?t.storage.s3:void 0,o=await new c({region:a?.region,credentials:a?.profile?g({profile:a.profile}):void 0,maxAttempts:a?.max_attempts}).send(new d({Bucket:i,Key:s}));if(!o.Body)return"";return await o.Body.transformToString()}async function Cs(e,t){if(t)ie(t);let n=await fetch(e,{headers:{accept:"text/markdown,text/plain,text/html,application/json;q=0.8,*/*;q=0.5","user-agent":"@hasna/knowledge source-ingest"}});if(!n.ok)throw Error(`Web source read failed ${n.status}: ${e}`);let r=n.headers.get("content-type"),i=await n.text();return{text:r?.includes("html")?Is(i):i,mime:r}}function Ke(e){if(e.kind==="file")return je(e.path);if(e.kind==="s3")return je(e.key);if(e.kind==="web")return je(new URL(e.url).pathname)||e.url;return e.path?je(e.path):e.id}async function Rn(e,t,n){if(e.kind==="file"){if(!Ns(e.path))throw Error(`Source file not found: ${e.path}`);let r=As(e.path,"utf8");return{text:r,contentSource:"file",title:Ke(e),mime:"text/plain",size:r.length,hash:ae(r),revision:null,extractedTextRef:null,metadata:{path:e.path},permissions:{mode:"read_only"}}}if(e.kind==="s3"){let r=await Ls(e.uri,t,n);return{text:r,contentSource:"s3",title:Ke(e),mime:"text/plain",size:r.length,hash:ae(r),revision:null,extractedTextRef:null,metadata:{bucket:e.bucket,key:e.key},permissions:{mode:"read_only"}}}if(e.kind==="web"){let r=await Cs(e.url,n);return{text:r.text,contentSource:"web",title:Ke(e),mime:r.mime,size:r.text.length,hash:ae(r.text),revision:null,extractedTextRef:null,metadata:{url:e.url},permissions:{mode:"read_only"}}}throw Error(`Direct source reading is not available for ${e.uri}`)}async function Ds(e,t,n){if(e.startsWith("open-files://"))throw Error("Open-files extracted text refs require an open-files resolver API. Ingest an open-files manifest with extracted_text or an extracted_text_ref using file://, s3://, or https://.");let r=M(e);return{text:(await Rn(r,t,n)).text,contentSource:"extracted_text_ref"}}async function Us(e){let t=await Me({dbPath:e.dbPath,sourceRef:e.sourceRef,purpose:e.purpose??"knowledge_index",limit:100,safetyPolicy:e.safetyPolicy,now:e.now});if(!t.resolved)throw Error("Open-files source is not in the local knowledge catalog. Ingest an open-files manifest first or use the open-files resolver API.");if(t.revision?.extracted_text_uri&&!t.content.text_available){let r=await Ds(t.revision.extracted_text_uri,e.config,e.safetyPolicy);return{text:r.text,contentSource:r.contentSource,title:t.source?.title??null,mime:t.content.mime,size:r.text.length,hash:t.revision.hash??ae(r.text),revision:t.revision.revision,extractedTextRef:t.revision.extracted_text_uri,metadata:t.source?.metadata??{},permissions:t.source?.permissions??{mode:"read_only"}}}if(t.chunks.length===0)throw Error("Open-files source has no extracted text chunks yet. Ingest an open-files manifest with extracted_text or extracted_text_ref first.");let n=t.chunks.map((r)=>r.text).join(`
|
|
461
461
|
|
|
462
|
-
`);return{text:n,contentSource:"catalog_chunks",title:t.source?.title??null,mime:t.content.mime,size:n.length,hash:t.revision?.hash??
|
|
462
|
+
`);return{text:n,contentSource:"catalog_chunks",title:t.source?.title??null,mime:t.content.mime,size:n.length,hash:t.revision?.hash??ae(n),revision:t.revision?.revision??null,extractedTextRef:t.revision?.extracted_text_uri??null,metadata:t.source?.metadata??{},permissions:t.source?.permissions??{mode:"read_only"}}}function Ps(e,t,n,r){let i=n.hash??ae(n.text),s={...n.metadata,source_ref:e,content_source:n.contentSource,read_only:!0},c={source_ref:e,name:n.title??Ke(t),mime:n.mime??"text/plain",size:n.size??n.text.length,hash:i,revision:n.revision??i,status:"active",updated_at:new Date().toISOString(),permissions:{mode:"read_only",allowed_purposes:[r],...n.permissions},metadata:s,extracted_text_ref:n.extractedTextRef,extracted_text:n.text};if(t.kind==="open-files"){if(t.entity==="file")c.file_id=t.id;if(t.entity==="source")c.source_id=t.id,c.path=t.path}if(t.kind==="file")c.path=t.path;if(t.kind==="s3")c.path=t.key;if(t.kind==="web")c.url=t.url;return c}async function Sn(e){let t=e.purpose??"knowledge_index",n=M(e.sourceRef),r=n.kind==="open-files"?await Us(e):await Rn(n,e.config,e.safetyPolicy),i=Ps(e.sourceRef,n,r,t);return{...await fe({dbPath:e.dbPath,items:[i],sourceLabel:e.sourceRef,readAction:"source_ref_ingest_read",safetyPolicy:e.safetyPolicy,now:e.now}),source_ref:e.sourceRef,content_source:r.contentSource,read_only:!0,hash:String(i.hash)}}import{createHash as Fs,randomUUID as xn}from"crypto";function Ms(e,t){return`${e}_${Fs("sha256").update(t).digest("hex").slice(0,20)}`}function js(e){let t=b(e);try{let n=t.query("SELECT status, COUNT(*) AS n FROM reindex_queue GROUP BY status ORDER BY status").all();return Object.fromEntries(n.map((r)=>[r.status,r.n]))}finally{t.close()}}function On(e,t){let n=V(t.modelRef,t.config),r=C(n),i=b(e);try{return i.query(`SELECT c.id AS chunk_id, c.source_revision_id, s.uri AS source_uri
|
|
463
463
|
FROM chunks c
|
|
464
464
|
LEFT JOIN source_revisions sr ON sr.id = c.source_revision_id
|
|
465
465
|
LEFT JOIN sources s ON s.id = sr.source_id
|
|
466
466
|
LEFT JOIN vector_index_entries v ON v.chunk_id = c.id AND v.provider = ? AND v.model = ?
|
|
467
467
|
WHERE v.id IS NULL
|
|
468
|
-
ORDER BY c.created_at ASC, c.ordinal ASC`).all(r.provider,r.model)}finally{i.close()}}function
|
|
469
|
-
WHERE metadata_json LIKE '%"reindex_required":true%' OR metadata_json LIKE '%"status":"stale"%'`).get()?.n??0;return{schema_version:n,chunks:r,vector_entries:i,missing_embeddings:s,queued:
|
|
470
|
-
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[a,"embedding",g.chunk_id,g.source_uri,n,"pending",JSON.stringify({source_revision_id:g.source_revision_id}),t,t]),s+=1}})()}finally{i.close()}return{enqueued:s,already_queued:
|
|
468
|
+
ORDER BY c.created_at ASC, c.ordinal ASC`).all(r.provider,r.model)}finally{i.close()}}function Nn(e){x(e.dbPath);let t=b(e.dbPath);try{let n=t.query("SELECT MAX(version) AS version FROM schema_versions").get()?.version??0,r=t.query("SELECT COUNT(*) AS n FROM chunks").get()?.n??0,i=t.query("SELECT COUNT(*) AS n FROM vector_index_entries").get()?.n??0,s=On(e.dbPath,e).length,c=t.query(`SELECT COUNT(*) AS n FROM source_revisions
|
|
469
|
+
WHERE metadata_json LIKE '%"reindex_required":true%' OR metadata_json LIKE '%"status":"stale"%'`).get()?.n??0;return{schema_version:n,chunks:r,vector_entries:i,missing_embeddings:s,queued:js(e.dbPath),stale_revisions:c}}finally{t.close()}}function gt(e){x(e.dbPath);let t=(e.now??new Date).toISOString(),n=e.reason??"missing_embedding",r=On(e.dbPath,e),i=b(e.dbPath),s=0,c=0;try{i.transaction(()=>{for(let g of r){let a=Ms("rq",`embedding\x00${g.chunk_id}\x00${n}`);if(i.query("SELECT id FROM reindex_queue WHERE kind = ? AND target_id = ? AND reason = ?").get("embedding",g.chunk_id,n)){c+=1;continue}i.run(`INSERT INTO reindex_queue (id, kind, target_id, source_uri, reason, status, metadata_json, created_at, updated_at)
|
|
470
|
+
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[a,"embedding",g.chunk_id,g.source_uri,n,"pending",JSON.stringify({source_revision_id:g.source_revision_id}),t,t]),s+=1}})()}finally{i.close()}return{enqueued:s,already_queued:c,reason:n}}function Ks(e){let t=b(e);try{let n=t.query("SELECT COUNT(*) AS n FROM chunk_embeddings").get()?.n??0,r=t.query("SELECT COUNT(*) AS n FROM vector_index_entries").get()?.n??0;return t.run("DELETE FROM vector_index_entries"),t.run("DELETE FROM chunk_embeddings"),{embeddings:n,vectorEntries:r}}finally{t.close()}}function $s(e,t,n){let r=V(t.modelRef,t.config),i=C(r),s=b(e);try{return s.run(`UPDATE reindex_queue
|
|
471
471
|
SET status = ?, updated_at = ?
|
|
472
472
|
WHERE kind = ?
|
|
473
473
|
AND status = ?
|
|
@@ -476,12 +476,12 @@ ${_}`;else{let{generateText:m}=await import("ai"),T=await Xt(i,{config:e.config,
|
|
|
476
476
|
WHERE v.chunk_id = reindex_queue.target_id
|
|
477
477
|
AND v.provider = ?
|
|
478
478
|
AND v.model = ?
|
|
479
|
-
)`,["completed",n,"embedding","pending",i.provider,i.model]).changes}finally{s.close()}}async function
|
|
480
|
-
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[n,"embedding-refresh",e.full?"full":"incremental","running","local",
|
|
481
|
-
VALUES (?, ?, ?, ?, ?, ?)`,[`evt_${
|
|
482
|
-
`),d
|
|
483
|
-
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[o,"provider-web-search",t,"running",a,
|
|
484
|
-
VALUES (?, ?, ?, ?, ?, ?)`,[`evt_${
|
|
479
|
+
)`,["completed",n,"embedding","pending",i.provider,i.model]).changes}finally{s.close()}}async function An(e){x(e.dbPath);let t=(e.now??new Date).toISOString(),n=`run_${xn()}`,r=e.full?Ks(e.dbPath):{embeddings:0,vectorEntries:0},i=gt({...e,reason:e.full?"full_embedding_rebuild":"missing_embedding"}),s=b(e.dbPath);try{s.run(`INSERT INTO runs (id, type, prompt, status, provider, model, metadata_json, created_at, updated_at)
|
|
480
|
+
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[n,"embedding-refresh",e.full?"full":"incremental","running","local",V(e.modelRef,e.config),JSON.stringify({full:e.full===!0,queued:i}),t,t])}finally{s.close()}let c=await Se({dbPath:e.dbPath,config:e.config,env:e.env,modelRef:e.modelRef,dimensions:e.dimensions,fake:e.fake,limit:e.limit,now:e.now}),d=$s(e.dbPath,e,t),g=b(e.dbPath);try{g.run("UPDATE runs SET status = ?, metadata_json = ?, updated_at = ? WHERE id = ?",["completed",JSON.stringify({full:e.full===!0,queued:i,indexed:c,completed_queue_items:d}),t,n]),g.run(`INSERT INTO run_events (id, run_id, level, event, metadata_json, created_at)
|
|
481
|
+
VALUES (?, ?, ?, ?, ?, ?)`,[`evt_${xn()}`,n,"info","embedding_refresh_completed",JSON.stringify({queued:i,indexed:c,completed_queue_items:d}),t])}finally{g.close()}return{run_id:n,full:e.full===!0,deleted_embeddings:r.embeddings,deleted_vector_entries:r.vectorEntries,queued:i,indexed:c,completed_queue_items:d}}var $e=1;function In(e){return Boolean(e&&typeof e==="object"&&!Array.isArray(e))}function K(e,t){let n=e[t];return typeof n==="string"?n:void 0}function Ws(e,t){let n=e[t];return typeof n==="number"&&Number.isFinite(n)?n:void 0}function Ln(e,t){let n=e[t];return Array.isArray(n)?n:void 0}function ge(e,t){let n=In(e)?e:{};return{contract_version:$e,id:K(n,"id")??t?.id,type:K(n,"type")??t?.type,status:K(n,"status")??t?.status,query:K(n,"query")??t?.query,prompt:K(n,"prompt")??t?.prompt,output_preview:Object.prototype.hasOwnProperty.call(n,"output_preview")?n.output_preview:t?.output_preview,citations:Ln(n,"citations")??t?.citations,artifacts:Ln(n,"artifacts")??t?.artifacts,usage:In(n.usage)?n.usage:t?.usage,created_at:K(n,"created_at")??t?.created_at,started_at:K(n,"started_at")??t?.started_at,completed_at:K(n,"completed_at")??t?.completed_at,duration_ms:Ws(n,"duration_ms")??t?.duration_ms,error_code:K(n,"error_code")??t?.error_code,error_message:K(n,"error_message")??t?.error_message,error:K(n,"error")??t?.error,details:Object.prototype.hasOwnProperty.call(n,"details")?n.details:t?.details}}function Cn(e){return{contract_version:$e,service:"open-knowledge",mode:e.mode,capabilities:["registry","search","ask","build","sync","status","logs","artifacts","open-files-source-refs","s3-generated-artifacts"],endpoints:{registry:"/api/v1/knowledge/registry",search:"/api/v1/knowledge/search",ask:"/api/v1/knowledge/ask",build:"/api/v1/knowledge/build",sync:"/api/v1/knowledge/sync",run_status:"/api/v1/knowledge/runs/{run_id}",run_logs:"/api/v1/knowledge/runs/{run_id}/logs",run_artifacts:"/api/v1/knowledge/runs/{run_id}/artifacts"},source_contract:{owner:"open-files",preferred_ref:"open-files",allowed_schemes:e.sourceSchemes,raw_source_bytes_stored_in_open_knowledge:!1},artifact_contract:{storage_type:e.storageType,uri_prefix:e.artifactUriPrefix,generated_only:!0}}}class We{apiKey;apiUrl;constructor(e,t){this.apiKey=e;this.apiUrl=t}static fromConfig(e,t=process.env){let n=nt(t);if(!n.apiKey)return null;return new We(n.apiKey,be(e,t))}async request(e,t={}){return fetch(`${this.apiUrl}${e}`,{...t,headers:{Authorization:`Bearer ${this.apiKey}`,"Content-Type":"application/json",...t.headers}})}async registry(){return(await this.request("/api/v1/knowledge/registry")).json()}async search(e){let t=await this.request("/api/v1/knowledge/search",{method:"POST",body:JSON.stringify(e)});return ge(await t.json(),{type:"search",query:e.query})}async ask(e){let t=await this.request("/api/v1/knowledge/ask",{method:"POST",body:JSON.stringify(e)});return ge(await t.json(),{type:"ask",prompt:e.prompt})}async build(e){let t=await this.request("/api/v1/knowledge/build",{method:"POST",body:JSON.stringify(e)});return ge(await t.json(),{type:"build",prompt:e.prompt})}async sync(e={}){let t=await this.request("/api/v1/knowledge/sync",{method:"POST",body:JSON.stringify(e)});return ge(await t.json(),{type:"sync"})}async runStatus(e){let t=await this.request(`/api/v1/knowledge/runs/${encodeURIComponent(e)}`);if(!t.ok)return null;return ge(await t.json(),{id:e,type:"status"})}async runLogs(e){let t=await this.request(`/api/v1/knowledge/runs/${encodeURIComponent(e)}/logs`);if(!t.ok)return[];let n=await t.json();return Array.isArray(n)?n:[]}async runArtifacts(e){let t=await this.request(`/api/v1/knowledge/runs/${encodeURIComponent(e)}/artifacts`);if(!t.ok)return[];let n=await t.json();return Array.isArray(n)?n:[]}}import{createHash as Xs,randomUUID as Dn}from"crypto";function qs(e){return`sha256:${Xs("sha256").update(e).digest("hex")}`}function Un(e){let t=e.trim().split(/\s+/).filter(Boolean).length;return Math.max(1,Math.ceil(t*1.25))}function Pn(e){return e&&typeof e==="object"&&!Array.isArray(e)?e:{}}function G(e){return typeof e==="string"&&e.length>0?e:null}function Hs(e){let t=Pn(e),n=G(t.url)??G(t.uri)??G(t.sourceUrl);if(!n)return null;return{url:n,title:G(t.title)??G(t.name),snippet:G(t.snippet)??G(t.text)??G(t.description),provider_metadata:t}}function Xe(e,t){if(Array.isArray(e)){for(let i of e)Xe(i,t);return}let n=Hs(e);if(n)t.set(n.url,n);let r=Pn(e);for(let i of["sources","results","citations","annotations","output"])if(r[i])Xe(r[i],t)}function Bs(e,t){return Array.from({length:Math.min(t,3)},(n,r)=>({url:`https://example.com/knowledge-web-${r+1}`,title:`Fake web source ${r+1}`,snippet:`Deterministic web-search fixture for "${e}"`,provider_metadata:{fake:!0,rank:r+1}}))}async function Js(e){let{generateText:t}=await import("ai"),{createOpenAI:n}=await import("@ai-sdk/openai"),r=q(e.config,"openai"),i=n({apiKey:e.env[r.api_key_env],baseURL:r.base_url}),s=i.tools?.webSearch;if(!s)throw Error("OpenAI provider does not expose tools.webSearch.");return t({model:i(e.model),prompt:e.query,tools:{web_search:s({externalWebAccess:!0,searchContextSize:"medium",...e.domains.length>0?{allowedDomains:e.domains}:{}})},toolChoice:{type:"tool",toolName:"web_search"}})}async function Gs(e){let{generateText:t}=await import("ai"),{createAnthropic:n}=await import("@ai-sdk/anthropic"),r=q(e.config,"anthropic"),i=n({apiKey:e.env[r.api_key_env],baseURL:r.base_url}),s=i.tools?.webSearch_20250305??i.tools?.webSearch;if(!s)throw Error("Anthropic provider does not expose a web search tool.");return t({model:i(e.model),prompt:e.query,tools:{web_search:s({maxUses:e.maxUses,...e.domains.length>0?{allowedDomains:e.domains}:{}})}})}async function zs(e,t,n){if(!e.fileResults||t.length===0)return 0;let r=t.map((s)=>{let c=[s.title,s.snippet,s.url].filter(Boolean).join(`
|
|
482
|
+
`),d=qs(c);return{source_ref:s.url,name:s.title??s.url,url:s.url,mime:"text/plain",hash:d,revision:d,status:"active",updated_at:n,permissions:{mode:"read_only",allowed_purposes:["knowledge_answer","knowledge_index"]},metadata:{source_ref:s.url,content_source:"provider_web_search",provider_metadata:s.provider_metadata},extracted_text:c}});return(await fe({dbPath:e.dbPath,items:r,sourceLabel:`web-search:${e.query}`,readAction:"provider_web_search_file_results",safetyPolicy:e.safetyPolicy,now:new Date(n)})).sources_upserted}async function Fn(e){let t=e.query.trim();if(!t)throw Error("Web search query is required.");let n=e.env??process.env,r=(e.now??new Date).toISOString(),i=Math.max(1,Math.min(e.limit??5,20)),s=Math.max(1,Math.min(e.maxUses??3,10)),c=e.domains??[],d=H(e.modelRef??(e.provider?`${e.provider}:${q(e.config,e.provider).default_model}`:"default"),e.config),g=C(d),a=e.provider??g.provider,u=g.provider===a?g.model:q(e.config,a).default_model,o=`run_${Dn()}`;if(!e.fake&&e.safetyPolicy)ie(e.safetyPolicy);if(!e.fake&&a!=="openai"&&a!=="anthropic")throw Error(`Provider ${a} does not expose native web search yet.`);if(!e.fake)z(a,e.config,n);x(e.dbPath);let _=b(e.dbPath);try{_.run(`INSERT INTO runs (id, type, prompt, status, provider, model, metadata_json, created_at, updated_at)
|
|
483
|
+
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[o,"provider-web-search",t,"running",a,u,JSON.stringify({domains:c,max_uses:s,fake:e.fake===!0}),r,r]),I(_,{event_type:"source_read",action:e.fake?"fake_provider_web_search":"provider_web_search",target_uri:t,decision:"allow",metadata:{provider:a,model:u,domains:c,max_uses:s},created_at:r})}finally{_.close()}let l="",f=[],m={input_tokens:Un(t),output_tokens:0,cost_usd:0},E=[];if(e.fake)f=Bs(t,i),l=`Fake web search answer for: ${t}`,m.output_tokens=Un(l);else{let k=a==="openai"?await Js({query:t,model:u,config:e.config,env:n,maxUses:s,domains:c}):await Gs({query:t,model:u,config:e.config,env:n,maxUses:s,domains:c});l=k.text;let p=new Map;Xe(k.sources,p),Xe(k.toolResults,p),f=Array.from(p.values()).slice(0,i);let R=we({provider:a,model:u,usage:k.usage,providerMetadata:k.providerMetadata});m={input_tokens:R.input_tokens,output_tokens:R.output_tokens,cost_usd:R.cost_usd}}let h=await zs(e,f,r),T=b(e.dbPath);try{T.run("UPDATE runs SET status = ?, metadata_json = ?, updated_at = ? WHERE id = ?",["completed",JSON.stringify({domains:c,max_uses:s,sources:f.length,filed_sources:h,fake:e.fake===!0}),r,o]),T.run(`INSERT INTO run_events (id, run_id, level, event, metadata_json, created_at)
|
|
484
|
+
VALUES (?, ?, ?, ?, ?, ?)`,[`evt_${Dn()}`,o,"info","provider_web_search_completed",JSON.stringify({sources:f.length,filed_sources:h}),r]),Te(T,{run_id:o,provider:a,model:u,input_tokens:m.input_tokens,output_tokens:m.output_tokens,cost_usd:m.cost_usd,metadata:{web_search:!0,sources:f.length,filed_sources:h},created_at:r})}finally{T.close()}if(f.length===0)E.push("no_web_sources_returned");return{run_id:o,query:t,provider:a,model:u,answer:l,sources:f,filed_sources:h,usage:m,warnings:E}}import{createHash as Qs,randomUUID as Zs}from"crypto";import{createHash as Ys,randomUUID as Vs}from"crypto";var Mn=[{kind:"schema",prefix:"schemas/",description:"Machine-readable agent schemas and source rules."},{kind:"index",prefix:"indexes/",description:"Small orientation indexes and future shard manifests."},{kind:"log",prefix:"logs/",description:"Append-only JSONL run and wiki-maintenance log partitions."},{kind:"run",prefix:"runs/",description:"Prompt/tool/cost ledgers and generated output records."},{kind:"wiki_page",prefix:"wiki/",description:"Generated cited Markdown pages, not raw source files."},{kind:"export",prefix:"exports/",description:"Portable exports and snapshots of derived knowledge state."}];function qe(e){let t=typeof e==="string"?Buffer.from(e):Buffer.from(e);return{hash:`sha256:${Ys("sha256").update(t).digest("hex")}`,size_bytes:t.byteLength}}function jn(e){return Mn.find((n)=>e.startsWith(n.prefix))?.kind??"artifact"}function mt(e,t,n="global"){let r=pt(e,t),i=e.storage.s3??null,s=i?.prefix?.replace(/^\/+|\/+$/g,"")??"",c=i?`s3://${i.bucket}/${s?`${s}/`:""}`:"",d=O.s3.prefix.replace(/^\/+|\/+$/g,""),g=`s3://${O.s3.bucket}/${d}/`,a=e.storage.type==="s3"&&i?.bucket===O.s3.bucket&&(i.region??null)===O.s3.region;return{scope:n,mode:e.mode,storage_type:e.storage.type,workspace_home:t.home,local_layout:{app_path:ne,config_path:t.configPath,json_store_path:t.jsonStorePath,knowledge_db_path:t.knowledgeDbPath,directories:{artifacts:t.artifactsDir,cache:t.cacheDir,exports:t.exportsDir,indexes:t.indexesDir,logs:t.logsDir,runs:t.runsDir,schemas:t.schemasDir,wiki:t.wikiDir}},artifact_store:{type:e.storage.type,artifacts_root:e.storage.artifacts_root,uri_prefix:e.storage.type==="s3"?c:`file://${t.artifactsDir}/`,s3:i?{bucket:i.bucket,prefix:s,region:i.region??null,profile:i.profile??null,server_side_encryption:i.server_side_encryption??null,kms_key_configured:Boolean(i.kms_key_id)}:null},canonical_hasna_xyz:{division:O.division,app_type:O.app_type,app:O.app,env:O.env,active:a,local_path:O.local_path,s3:{bucket:O.s3.bucket,region:O.s3.region,profile:O.s3.profile,prefix:d,uri_prefix:g,server_side_encryption:O.s3.server_side_encryption},secrets:{env:O.secrets.env,aws:O.secrets.aws,s3:O.secrets.s3,rds:O.secrets.rds,future_rds:O.secrets.future_rds},evidence_doc:O.evidence_doc},hosted:{enabled:e.mode==="hosted",api_url:X(e.hosted?.api_url??tt),api_url_env:"KNOWLEDGE_API_URL",api_key_env:"KNOWLEDGE_API_KEY",auth_storage:"~/.hasna/knowledge/auth.json",remote_contract_version:$e,requires_hosted_account_for_local_use:!1},source_ownership:{owner:"open-files",preferred_ref:e.sources.preferred_ref,allowed_schemes:e.sources.allowed_schemes,raw_source_bytes_stored_in_open_knowledge:!1,stores:["source refs","source revisions and hashes","citation spans","redacted extracted chunks","embeddings","generated wiki artifacts","indexes","run ledgers"],does_not_store:["raw open-files bytes","S3 object credentials","connector secrets","hosted tenant ownership state"]},generated_artifacts:Mn,scalability:{catalog:"knowledge.db tracks sources, revisions, chunks, citations, indexes, runs, and storage_objects.",indexes:"Indexes are cataloged DB rows plus sharded artifacts, not one giant index.md.",logs:"Logs use dated JSONL partitions under logs/yyyy/mm/dd.jsonl.",markdown:"Markdown pages are the readable wiki layer over DB/object-store state."},warnings:r.warnings}}function pt(e,t){let n=[],r=[];if(!t.home.endsWith(ne))r.push(`Workspace home does not end with ${ne}: ${t.home}`);if(e.storage.type==="s3"){if(!e.storage.s3?.bucket)n.push("storage.s3.bucket is required when storage.type is s3.");if(!e.storage.s3?.prefix)r.push("storage.s3.prefix is empty; generated knowledge artifacts will be written at the bucket root.");if(e.mode==="local")r.push("storage.type is s3 while mode is local; this is valid for BYO S3, but hosted wrappers should set mode to hosted.")}if(e.storage.type==="local"&&e.storage.s3)r.push("storage.s3 is configured but ignored while storage.type is local.");if(e.sources.preferred_ref!=="open-files")r.push("sources.preferred_ref should stay open-files for durable company knowledge.");if(!e.sources.allowed_schemes.includes("open-files"))n.push("sources.allowed_schemes must include open-files.");if(e.mode==="hosted"&&e.hosted?.api_url)try{X(e.hosted.api_url)}catch{n.push("hosted.api_url must be an http(s) URL when mode is hosted.")}return{ok:n.length===0,errors:n,warnings:r}}function me(e,t,n=new Date){let r=n.toISOString(),i=e.prepare(`
|
|
485
485
|
INSERT INTO storage_objects (
|
|
486
486
|
id, artifact_uri, kind, content_type, hash, size_bytes, metadata_json, created_at, updated_at
|
|
487
487
|
)
|
|
@@ -493,7 +493,7 @@ ${_}`;else{let{generateText:m}=await import("ai"),T=await Xt(i,{config:e.config,
|
|
|
493
493
|
size_bytes = excluded.size_bytes,
|
|
494
494
|
metadata_json = excluded.metadata_json,
|
|
495
495
|
updated_at = excluded.updated_at
|
|
496
|
-
`);e.transaction((
|
|
496
|
+
`);e.transaction((c)=>{for(let d of c)i.run(Vs(),d.uri,d.kind,d.content_type??null,d.hash??null,d.size_bytes??null,JSON.stringify({key:d.key,...d.metadata??{}}),r,r)})(t)}function ue(e,t){return`${e}_${Qs("sha256").update(t).digest("hex").slice(0,20)}`}function ht(e){return e.normalize("NFKC").toLowerCase().replace(/[^a-z0-9]+/g,"-").replace(/^-+|-+$/g,"").slice(0,80)||"knowledge-page"}function eo(e){return{year:String(e.getUTCFullYear()),month:String(e.getUTCMonth()+1).padStart(2,"0"),day:String(e.getUTCDate()).padStart(2,"0")}}function to(e){let t=e.trim().split(/\s+/).filter(Boolean).length;return Math.max(1,Math.ceil(t*1.25))}function Kn(e){if(!e)return{};try{let t=JSON.parse(e);return t&&typeof t==="object"&&!Array.isArray(t)?t:{}}catch{return{}}}function no(e){return Array.from(new Set((e??"").toLowerCase().match(/[\p{L}\p{N}_]+/gu)??[])).slice(0,12)}function $n(e){return e.replace(/[\\%_]/g,(t)=>`\\${t}`)}function ro(e,t){let n=Math.max(1,Math.min(t.limit??10,50)),r=t.sourceRefs??[],i=no(t.query),s=["c.kind = 'source'"],c=[];if(r.length>0){s.push(`(${r.map(()=>"(s.uri = ? OR c.metadata_json LIKE ?)").join(" OR ")})`);for(let d of r)c.push(d,`%${$n(d)}%`)}if(i.length>0){s.push(`(${i.map(()=>"lower(c.text) LIKE ? ESCAPE '\\'").join(" OR ")})`);for(let d of i)c.push(`%${$n(d)}%`)}return c.push(n),e.query(`SELECT
|
|
497
497
|
c.id AS chunk_id,
|
|
498
498
|
c.text,
|
|
499
499
|
c.start_offset,
|
|
@@ -509,10 +509,10 @@ ${_}`;else{let{generateText:m}=await import("ai"),T=await Xt(i,{config:e.config,
|
|
|
509
509
|
JOIN sources s ON s.id = sr.source_id
|
|
510
510
|
WHERE ${s.join(" AND ")}
|
|
511
511
|
ORDER BY c.created_at ASC, c.ordinal ASC
|
|
512
|
-
LIMIT ?`).all(...
|
|
512
|
+
LIMIT ?`).all(...c)}function Wn(e,t=420){let n=e.replace(/\s+/g," ").trim();return n.length<=t?n:`${n.slice(0,t-1).trim()}...`}function io(e,t){if(e.title?.trim())return e.title.trim();if(e.query?.trim())return e.query.trim();return t[0]?.source_title??"Compiled Knowledge"}function so(e,t,n){let r=t.map((s,c)=>{return`- [${`S${c+1}`}] ${s.source_title??s.source_uri??"Source"} (${s.source_uri??"unknown"}, revision ${s.revision??"unknown"}, hash ${s.hash??"unknown"})`}),i=t.map((s,c)=>{let d=`S${c+1}`;return[`## ${s.source_title??`Source ${c+1}`}`,"",Wn(s.text),"",`Citation: [${d}]`].join(`
|
|
513
513
|
`)});return[`# ${e}`,"",`Generated at: ${n}`,"","## Sources","",...r,"",...i,""].join(`
|
|
514
|
-
`)}async function
|
|
515
|
-
`,content_type:"application/x-ndjson"})}function
|
|
514
|
+
`)}async function He(e,t){let n=await e.put(t);return{key:n.key,uri:n.uri,kind:t.key.startsWith("logs/")?"log":"wiki_page",content_type:t.content_type,...qe(t.body),metadata:{...t.metadata??{}}}}async function Xn(e,t,n){let{year:r,month:i,day:s}=eo(n),c=`logs/${r}/${i}/${s}.jsonl`,d="";try{d=await e.getText(c)}catch{d=""}return He(e,{key:c,body:`${d}${JSON.stringify(t)}
|
|
515
|
+
`,content_type:"application/x-ndjson"})}function Et(e,t){e.run(`INSERT INTO wiki_pages (id, path, title, artifact_uri, content_hash, status, metadata_json, created_at, updated_at)
|
|
516
516
|
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)
|
|
517
517
|
ON CONFLICT(path) DO UPDATE SET
|
|
518
518
|
title = excluded.title,
|
|
@@ -520,41 +520,41 @@ ${_}`;else{let{generateText:m}=await import("ai"),T=await Xt(i,{config:e.config,
|
|
|
520
520
|
content_hash = excluded.content_hash,
|
|
521
521
|
status = excluded.status,
|
|
522
522
|
metadata_json = excluded.metadata_json,
|
|
523
|
-
updated_at = excluded.updated_at`,[t.pageId,t.path,t.title,t.artifactUri,t.contentHash,"active",JSON.stringify({artifact_key:t.path,provenance:t.provenance}),t.now,t.now]);let n=e.query("SELECT id FROM chunks WHERE wiki_page_id = ?").all(t.pageId);for(let i of n)e.run("DELETE FROM chunks_fts WHERE chunk_id = ?",[i.id]);e.run("DELETE FROM chunks WHERE wiki_page_id = ?",[t.pageId]);let r=
|
|
524
|
-
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,[r,t.pageId,"wiki",0,t.body,
|
|
525
|
-
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[
|
|
523
|
+
updated_at = excluded.updated_at`,[t.pageId,t.path,t.title,t.artifactUri,t.contentHash,"active",JSON.stringify({artifact_key:t.path,provenance:t.provenance}),t.now,t.now]);let n=e.query("SELECT id FROM chunks WHERE wiki_page_id = ?").all(t.pageId);for(let i of n)e.run("DELETE FROM chunks_fts WHERE chunk_id = ?",[i.id]);e.run("DELETE FROM chunks WHERE wiki_page_id = ?",[t.pageId]);let r=ue("chk",`${t.pageId}\x00${t.contentHash}`);e.run(`INSERT INTO chunks (id, wiki_page_id, kind, ordinal, text, token_count, start_offset, end_offset, metadata_json, created_at)
|
|
524
|
+
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,[r,t.pageId,"wiki",0,t.body,to(t.body),0,t.body.length,JSON.stringify({artifact_key:t.path,artifact_uri:t.artifactUri,content_hash:t.contentHash,provenance:t.provenance}),t.now]),e.run("INSERT INTO chunks_fts (chunk_id, text, title, source_uri) VALUES (?, ?, ?, ?)",[r,t.body,t.title,t.artifactUri])}function qn(e,t,n,r){e.run("DELETE FROM citations WHERE wiki_page_id = ?",[t]);for(let i of n)e.run(`INSERT INTO citations (id, wiki_page_id, chunk_id, source_uri, quote, start_offset, end_offset, metadata_json, created_at)
|
|
525
|
+
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,[ue("cit",`${t}\x00${i.source_uri}\x00${i.chunk_id??Zs()}`),t,i.chunk_id,i.source_uri,i.quote,i.start_offset,i.end_offset,JSON.stringify(i.metadata),r]);return n.length}function Hn(e,t){return e.run(`INSERT INTO knowledge_indexes (id, kind, name, artifact_uri, shard_key, metadata_json, created_at, updated_at)
|
|
526
526
|
VALUES (?, ?, ?, ?, ?, ?, ?, ?)
|
|
527
527
|
ON CONFLICT(kind, name, shard_key) DO UPDATE SET
|
|
528
528
|
artifact_uri = excluded.artifact_uri,
|
|
529
529
|
metadata_json = excluded.metadata_json,
|
|
530
|
-
updated_at = excluded.updated_at`,[
|
|
531
|
-
`),h=
|
|
532
|
-
VALUES (?, ?, ?, ?)`,[_,
|
|
533
|
-
`),g=
|
|
530
|
+
updated_at = excluded.updated_at`,[ue("idx",`wiki-topic\x00${t.path}`),"wiki_topic",t.title,t.artifactUri,t.path,JSON.stringify({artifact_key:t.path,content_hash:t.contentHash}),t.now,t.now]),1}function oo(e){return e.toLowerCase().match(/[a-z0-9][a-z0-9-]{2,}/)?.[0]??"knowledge"}async function Bn(e){let t=e.now??new Date,n=t.toISOString();x(e.dbPath);let r=b(e.dbPath),i;try{i=ro(r,e)}finally{r.close()}if(i.length===0)throw Error("No source chunks matched wiki compile input.");let s=io(e,i),d=`wiki/generated/${ht(s)}.md`,g=so(s,i,n),a=i.map((v)=>{let w=Kn(v.metadata_json);return typeof w.source_ref==="string"?w.source_ref:v.source_uri}).filter((v)=>Boolean(v)),u=Y({generated_from:"wiki_compile",artifact_key:d,source_refs:a}),o=await He(e.store,{key:d,body:g,content_type:"text/markdown",metadata:{generated_from:"wiki_compile"}}),_=ue("wiki",d),l=i.map((v)=>({chunk_id:v.chunk_id,source_uri:v.source_uri??"unknown",quote:Wn(v.text,240),start_offset:v.start_offset,end_offset:v.end_offset,metadata:{source_revision_id:v.source_revision_id,revision:v.revision,hash:v.hash,source_ref:Kn(v.metadata_json).source_ref??v.source_uri}})),f=oo(s),m=`wiki/concepts/${ht(f)}.md`,E=[`# ${f}`,"",`Related page: [[${d}]]`,""].join(`
|
|
531
|
+
`),h=Y({generated_from:"wiki_compile_concept",artifact_key:m,source_refs:a}),T=await He(e.store,{key:m,body:E,content_type:"text/markdown",metadata:{generated_from:"wiki_compile_concept"}}),k=ue("wiki",m),p=await Xn(e.store,{ts:n,event:"wiki_compile_completed",page_key:d,source_refs:a,chunks_seen:i.length},t),R=b(e.dbPath);try{me(R,[o,T,p],t),Et(R,{pageId:_,path:d,title:s,artifactUri:o.uri,contentHash:o.hash??"",body:g,provenance:u,now:n}),Et(R,{pageId:k,path:m,title:f,artifactUri:T.uri,contentHash:T.hash??"",body:E,provenance:h,now:n}),R.run(`INSERT OR REPLACE INTO wiki_backlinks (from_page_id, to_page_id, label, created_at)
|
|
532
|
+
VALUES (?, ?, ?, ?)`,[_,k,"concept",n]);let v=qn(R,_,l,n),w=Hn(R,{title:s,path:d,artifactUri:o.uri,contentHash:o.hash??"",now:n});return{page_id:_,path:d,artifact_uri:o.uri,content_hash:o.hash??"",chunks_seen:i.length,citations_written:v,concept_page_id:k,indexes_updated:w,log_key:p.key,warnings:[]}}finally{R.close()}}async function Jn(e){if(!e.approveWrite)return{approved:!1,durable_writes_performed:!1,page_id:null,path:null,artifact_uri:null,citations_written:0,log_key:null,message:"Dry-run: answer filing requires --approve-write."};let t=e.now??new Date,n=t.toISOString(),r=e.prompt.length>80?`${e.prompt.slice(0,77)}...`:e.prompt,s=`wiki/answers/${ht(r)}.md`,c=e.context.citations,d=[`# ${r}`,"",e.answer,"","## Citations","",...c.map((f,m)=>`- [C${m+1}] ${f.source_ref??f.source_uri??f.artifact_path??f.artifact_uri??"unknown"} ${f.hash?`(hash ${f.hash})`:""}`),""].join(`
|
|
533
|
+
`),g=c.map((f)=>f.source_ref??f.source_uri).filter((f)=>Boolean(f)),a=Y({generated_from:"knowledge_answer",artifact_key:s,source_refs:g}),u=await He(e.store,{key:s,body:d,content_type:"text/markdown",metadata:{generated_from:"knowledge_answer"}}),o=await Xn(e.store,{ts:n,event:"wiki_answer_filed",page_key:s,prompt:e.prompt,citations:c.length},t),_=ue("wiki",s),l=b(e.dbPath);try{me(l,[u,o],t),Et(l,{pageId:_,path:s,title:r,artifactUri:u.uri,contentHash:u.hash??"",body:d,provenance:a,now:n});let f=qn(l,_,c.map((m)=>({chunk_id:m.chunk_id,source_uri:m.source_uri??m.artifact_uri??"unknown",quote:m.quote,start_offset:m.start_offset,end_offset:m.end_offset,metadata:{source_ref:m.source_ref,artifact_path:m.artifact_path,revision:m.revision,hash:m.hash}})),n);return Hn(l,{title:r,path:s,artifactUri:u.uri,contentHash:u.hash??"",now:n}),{approved:!0,durable_writes_performed:!0,page_id:_,path:s,artifact_uri:u.uri,citations_written:f,log_key:o.key,message:`Filed answer to ${s}`}}finally{l.close()}}function ee(e,t){e.push(t)}function Gn(e){x(e.dbPath);let t=b(e.dbPath),n=[];try{let r=t.query("SELECT COUNT(*) AS n FROM wiki_pages WHERE status = 'active'").get()?.n??0,i=t.query("SELECT COUNT(*) AS n FROM citations").get()?.n??0,s=t.query("SELECT COUNT(*) AS n FROM wiki_backlinks").get()?.n??0,c=t.query(`SELECT wp.id, wp.path
|
|
534
534
|
FROM wiki_pages wp
|
|
535
535
|
LEFT JOIN citations c ON c.wiki_page_id = wp.id
|
|
536
536
|
WHERE wp.status = 'active' AND wp.path LIKE 'wiki/generated/%'
|
|
537
537
|
GROUP BY wp.id
|
|
538
|
-
HAVING COUNT(c.id) = 0`).all();for(let l of
|
|
538
|
+
HAVING COUNT(c.id) = 0`).all();for(let l of c)ee(n,{type:"missing_citation",severity:"error",page_id:l.id,path:l.path,message:"Generated wiki page has no citations."});let d=t.query(`SELECT wp.id AS page_id, wp.path, c.source_uri, c.chunk_id
|
|
539
539
|
FROM citations c
|
|
540
540
|
JOIN wiki_pages wp ON wp.id = c.wiki_page_id
|
|
541
541
|
LEFT JOIN chunks ch ON ch.id = c.chunk_id
|
|
542
|
-
WHERE ch.metadata_json LIKE '%"stale":true%' OR ch.metadata_json LIKE '%"status":"stale"%' OR ch.metadata_json LIKE '%"status":"deleted"%'`).all();for(let l of d)
|
|
542
|
+
WHERE ch.metadata_json LIKE '%"stale":true%' OR ch.metadata_json LIKE '%"status":"stale"%' OR ch.metadata_json LIKE '%"status":"deleted"%'`).all();for(let l of d)ee(n,{type:"stale_citation",severity:"warn",page_id:l.page_id,path:l.path,source_uri:l.source_uri,chunk_id:l.chunk_id??void 0,message:"Page cites a stale or deleted source chunk."});let g=t.query(`SELECT lower(title) AS title, COUNT(*) AS n
|
|
543
543
|
FROM wiki_pages
|
|
544
544
|
WHERE status = 'active'
|
|
545
545
|
GROUP BY lower(title)
|
|
546
|
-
HAVING COUNT(*) > 1`).all();for(let l of g)
|
|
546
|
+
HAVING COUNT(*) > 1`).all();for(let l of g)ee(n,{type:"duplicate_page",severity:"warn",message:`Duplicate active wiki title: ${l.title} (${l.n} pages).`});let a=t.query(`SELECT wp.id, wp.path
|
|
547
547
|
FROM wiki_pages wp
|
|
548
548
|
LEFT JOIN wiki_backlinks wb1 ON wb1.from_page_id = wp.id
|
|
549
549
|
LEFT JOIN wiki_backlinks wb2 ON wb2.to_page_id = wp.id
|
|
550
550
|
WHERE wp.status = 'active'
|
|
551
551
|
AND wp.path NOT IN ('wiki/README.md')
|
|
552
552
|
GROUP BY wp.id
|
|
553
|
-
HAVING COUNT(wb1.to_page_id) = 0 AND COUNT(wb2.from_page_id) = 0`).all();for(let l of a)
|
|
553
|
+
HAVING COUNT(wb1.to_page_id) = 0 AND COUNT(wb2.from_page_id) = 0`).all();for(let l of a)ee(n,{type:"orphan_page",severity:"info",page_id:l.id,path:l.path,message:"Wiki page has no backlinks."});let u=t.query(`SELECT wp.id AS page_id, wp.path, c.source_uri
|
|
554
554
|
FROM citations c
|
|
555
555
|
JOIN wiki_pages wp ON wp.id = c.wiki_page_id
|
|
556
556
|
LEFT JOIN sources s ON s.uri = c.source_uri
|
|
557
|
-
WHERE s.id IS NULL AND c.source_uri NOT LIKE 'file://%' AND c.source_uri NOT LIKE 's3://%' AND c.source_uri NOT LIKE 'https://%' AND c.source_uri NOT LIKE 'open-files://%'`).all();for(let l of
|
|
557
|
+
WHERE s.id IS NULL AND c.source_uri NOT LIKE 'file://%' AND c.source_uri NOT LIKE 's3://%' AND c.source_uri NOT LIKE 'https://%' AND c.source_uri NOT LIKE 'open-files://%'`).all();for(let l of u)ee(n,{type:"unresolved_source_ref",severity:"error",page_id:l.page_id,path:l.path,source_uri:l.source_uri,message:"Citation source URI cannot be resolved to a known or allowed source ref."});let o=t.query("SELECT id, path FROM wiki_pages WHERE lower(metadata_json) LIKE '%contradiction%'").all();for(let l of o)ee(n,{type:"contradiction_marker",severity:"warn",page_id:l.id,path:l.path,message:"Page metadata contains a contradiction marker."});let _=t.query(`SELECT c.id AS chunk_id, s.uri AS source_uri
|
|
558
558
|
FROM chunks c
|
|
559
559
|
JOIN source_revisions sr ON sr.id = c.source_revision_id
|
|
560
560
|
JOIN sources s ON s.id = sr.source_id
|
|
@@ -562,7 +562,7 @@ ${_}`;else{let{generateText:m}=await import("ai"),T=await Xt(i,{config:e.config,
|
|
|
562
562
|
WHERE c.kind = 'source'
|
|
563
563
|
GROUP BY c.id
|
|
564
564
|
HAVING COUNT(cit.id) = 0
|
|
565
|
-
LIMIT 25`).all();for(let l of _)
|
|
565
|
+
LIMIT 25`).all();for(let l of _)ee(n,{type:"new_article_candidate",severity:"info",chunk_id:l.chunk_id,source_uri:l.source_uri??void 0,message:"Source chunk is indexed but not cited by any wiki page yet."});return{ok:n.every((l)=>l.severity!=="error"),issue_count:n.length,issues:n,counts:{active_pages:r,citations:i,backlinks:s,new_article_candidates:_.length}}}finally{t.close()}}import{createHash as ao}from"crypto";function uo(e){let t=String(e.getUTCFullYear()),n=String(e.getUTCMonth()+1).padStart(2,"0"),r=String(e.getUTCDate()).padStart(2,"0");return{year:t,month:n,day:r}}function yt(e,t){return`${e}_${ao("sha256").update(t).digest("hex").slice(0,20)}`}function co(e){let t=e.trim().split(/\s+/).filter(Boolean).length;return Math.max(1,Math.ceil(t*1.25))}function lo(){return`# Knowledge Agent Schema v1
|
|
566
566
|
|
|
567
567
|
## Source Rules
|
|
568
568
|
|
|
@@ -587,7 +587,7 @@ ${_}`;else{let{generateText:m}=await import("ai"),T=await Xt(i,{config:e.config,
|
|
|
587
587
|
## Lint Rules
|
|
588
588
|
|
|
589
589
|
- Flag stale pages, missing citations, contradictions, orphan pages, duplicate pages, and unresolved source refs.
|
|
590
|
-
`}function
|
|
590
|
+
`}function _o(){return`# Knowledge Index
|
|
591
591
|
|
|
592
592
|
This is a compact orientation index for agents. It is not the full search index.
|
|
593
593
|
|
|
@@ -602,19 +602,19 @@ This is a compact orientation index for agents. It is not the full search index.
|
|
|
602
602
|
|
|
603
603
|
Raw source files are resolved through open-files. This app stores source refs,
|
|
604
604
|
citations, chunks, generated wiki artifacts, indexes, and run records.
|
|
605
|
-
`}function
|
|
605
|
+
`}function zn(){return`# Wiki
|
|
606
606
|
|
|
607
607
|
Generated durable knowledge pages live here.
|
|
608
608
|
|
|
609
609
|
Pages should be concise, cited, and organized for both humans and agents.
|
|
610
|
-
`}async function
|
|
611
|
-
`,content_type:"application/x-ndjson"}],o=await Promise.all(
|
|
612
|
-
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,[d,t,"wiki",0,i,
|
|
610
|
+
`}async function Yn(e,t=new Date){let{year:n,month:r,day:i}=uo(t),s="schemas/v1.md",c="indexes/root.md",d="wiki/README.md",g=`logs/${n}/${r}/${i}.jsonl`,a={ts:t.toISOString(),event:"wiki_layout_initialized",schema_key:"schemas/v1.md",root_index_key:"indexes/root.md",wiki_readme_key:"wiki/README.md"},u=[{key:"schemas/v1.md",body:lo(),content_type:"text/markdown"},{key:"indexes/root.md",body:_o(),content_type:"text/markdown"},{key:"wiki/README.md",body:zn(),content_type:"text/markdown"},{key:g,body:`${JSON.stringify(a)}
|
|
611
|
+
`,content_type:"application/x-ndjson"}],o=await Promise.all(u.map(async(_)=>{let l=await e.put(_);return{key:l.key,uri:l.uri,kind:jn(_.key),content_type:_.content_type,metadata:{provenance:Y({generated_from:"wiki_layout_init",artifact_key:_.key,citation_required:_.key.startsWith("wiki/")||_.key.startsWith("indexes/")})},...qe(_.body)}}));return{schema_key:"schemas/v1.md",root_index_key:"indexes/root.md",wiki_readme_key:"wiki/README.md",log_key:g,artifacts:o,written:["schemas/v1.md","indexes/root.md","wiki/README.md",g]}}function kt(e){let t=e.metadata?.provenance;if(t&&typeof t==="object"&&!Array.isArray(t))return t;return Y({generated_from:"wiki_layout_init",artifact_key:e.key})}function fo(e,t,n,r,i,s){let c=kt(r),d=yt("chk",`${t}\x00${r.hash??r.uri}`),g=e.query("SELECT id FROM chunks WHERE wiki_page_id = ?").all(t);for(let a of g)e.run("DELETE FROM chunks_fts WHERE chunk_id = ?",[a.id]);e.run("DELETE FROM chunks WHERE wiki_page_id = ?",[t]),e.run(`INSERT INTO chunks (id, wiki_page_id, kind, ordinal, text, token_count, start_offset, end_offset, metadata_json, created_at)
|
|
612
|
+
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,[d,t,"wiki",0,i,co(i),0,i.length,JSON.stringify({artifact_key:r.key,artifact_uri:r.uri,content_hash:r.hash??null,provenance:c}),s]),e.run("INSERT INTO chunks_fts (chunk_id, text, title, source_uri) VALUES (?, ?, ?, ?)",[d,i,n,r.uri])}function Vn(e,t,n=new Date){let r=n.toISOString(),i=t.find((c)=>c.key.endsWith("indexes/root.md")),s=t.find((c)=>c.key.endsWith("wiki/README.md"));if(i)e.run(`INSERT INTO knowledge_indexes (id, kind, name, artifact_uri, shard_key, metadata_json, created_at, updated_at)
|
|
613
613
|
VALUES (?, ?, ?, ?, ?, ?, ?, ?)
|
|
614
614
|
ON CONFLICT(kind, name, shard_key) DO UPDATE SET
|
|
615
615
|
artifact_uri = excluded.artifact_uri,
|
|
616
616
|
metadata_json = excluded.metadata_json,
|
|
617
|
-
updated_at = excluded.updated_at`,[
|
|
617
|
+
updated_at = excluded.updated_at`,[yt("idx","root:indexes/root.md"),"root","root",i.uri,"root",JSON.stringify({artifact_key:i.key,content_hash:i.hash??null,provenance:kt(i)}),r,r]);if(s){let c=yt("wiki","wiki/README.md");e.run(`INSERT INTO wiki_pages (id, path, title, artifact_uri, content_hash, status, metadata_json, created_at, updated_at)
|
|
618
618
|
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)
|
|
619
619
|
ON CONFLICT(path) DO UPDATE SET
|
|
620
620
|
title = excluded.title,
|
|
@@ -622,7 +622,7 @@ Pages should be concise, cited, and organized for both humans and agents.
|
|
|
622
622
|
content_hash = excluded.content_hash,
|
|
623
623
|
status = excluded.status,
|
|
624
624
|
metadata_json = excluded.metadata_json,
|
|
625
|
-
updated_at = excluded.updated_at`,[u,"wiki/README.md","Wiki",s.uri,s.hash??null,"active",JSON.stringify({artifact_key:s.key,provenance:Et(s)}),r,r]),lo(e,u,"Wiki",s,Jn(),r)}}function _o(e){if(!e)return;let t=e.trim().toLowerCase();if(t==="local"||t==="offline")return"local";if(t==="hosted"||t==="remote"||t==="knowledge.hasna.xyz")return"hosted";throw Error("Invalid setup mode. Use hosted or local.")}class Yn{options;ensuredWorkspace;cachedConfig;constructor(e={}){this.options=e}get scope(){return this.options.scope??"global"}get workspace(){return this.ensuredWorkspace??wt(this.options.scope,this.options.cwd)}ensureWorkspace(){if(!this.ensuredWorkspace)this.ensuredWorkspace=bt(this.workspace.home);return this.ensuredWorkspace}jsonStorePath(){return this.ensureWorkspace().jsonStorePath}config(){if(!this.cachedConfig){let e=this.ensureWorkspace();this.cachedConfig=vt(e.configPath)}return this.cachedConfig}safetyPolicy(){return pn(this.config(),this.ensureWorkspace())}artifactStore(){return Lt(this.config(),this.ensureWorkspace())}storageContract(){return Kn(this.config(),this.ensureWorkspace(),this.scope)}validateStorage(){return gt(this.config(),this.ensureWorkspace())}setup(e={}){let t=this.ensureWorkspace(),n=this.config(),r=_o(e.mode)??n.mode,i=e.apiUrl?$(e.apiUrl):n.hosted?.api_url?$(n.hosted.api_url):null,s={...n,mode:r,hosted:{...n.hosted??{},...i?{api_url:i}:{}}};return St(t.configPath,s),this.cachedConfig=s,{ok:!0,mode:r,api_url:s.hosted?.api_url??null,config_path:t.configPath,next:r==="hosted"?["open-knowledge auth login --api-key <key>","open-knowledge remote contracts --json"]:["open-knowledge search <query>","knowledge <prompt>"],message:`Set knowledge mode to ${r}`}}authStatus(e=process.env){return Kt(this.config(),e)}saveAuth(e,t=process.env){let n=e.apiUrl??this.config().hosted?.api_url;return Dt({api_key:e.apiKey,email:e.email,org_id:e.orgId,org_slug:e.orgSlug,user_id:e.userId,api_url:n},t)}clearAuth(e=process.env){return Ut(e)}remoteContract(){let e=this.storageContract();return An({mode:this.config().mode,sourceSchemes:this.config().sources.allowed_schemes,storageType:e.artifact_store.type,artifactUriPrefix:e.artifact_store.uri_prefix})}remoteClient(e=process.env){return We.fromConfig(this.config(),e)}paths(){let e=this.ensureWorkspace();return{ok:!0,scope:this.scope,home:e.home,config_path:e.configPath,json_store_path:e.jsonStorePath,knowledge_db_path:e.knowledgeDbPath,artifacts_dir:e.artifactsDir,indexes_dir:e.indexesDir,logs_dir:e.logsDir,runs_dir:e.runsDir,schemas_dir:e.schemasDir,wiki_dir:e.wikiDir,config:this.config(),message:e.home}}initDb(){return x(this.ensureWorkspace().knowledgeDbPath)}dbStats(){let e=this.ensureWorkspace();return x(e.knowledgeDbPath),Ot(e.knowledgeDbPath)}async initWiki(){let e=this.ensureWorkspace();x(e.knowledgeDbPath);let t=await zn(this.artifactStore()),n=b(e.knowledgeDbPath);try{ge(n,t.artifacts),Gn(n,t.artifacts)}finally{n.close()}return t}async compileWiki(e={}){let t=this.ensureWorkspace();return qn({...e,dbPath:t.knowledgeDbPath,store:this.artifactStore()})}async fileAnswer(e){let t=this.ensureWorkspace(),n=await this.retrieveContext({query:e.prompt,limit:e.limit,semantic:e.semantic,modelRef:e.modelRef,dimensions:e.dimensions,fake:e.fake});return Hn({dbPath:t.knowledgeDbPath,store:this.artifactStore(),prompt:e.prompt,answer:e.answer,context:n,approveWrite:e.approveWrite})}lintWiki(){let e=this.ensureWorkspace();return Bn({dbPath:e.knowledgeDbPath})}async ingestManifest(e){let t=this.ensureWorkspace();return yn({dbPath:t.knowledgeDbPath,input:e,config:this.config(),safetyPolicy:this.safetyPolicy()})}async ingestSource(e,t){let n=this.ensureWorkspace();return vn({dbPath:n.knowledgeDbPath,sourceRef:e,purpose:t,config:this.config(),safetyPolicy:this.safetyPolicy()})}async resolveSource(e,t={}){let n=this.ensureWorkspace();return Ke({dbPath:n.knowledgeDbPath,sourceRef:e,purpose:t.purpose,limit:t.limit,safetyPolicy:this.safetyPolicy()})}async consumeOutbox(e){let t=this.ensureWorkspace();return kn({dbPath:t.knowledgeDbPath,input:e,config:this.config(),safetyPolicy:this.safetyPolicy()})}reindexHealth(e={}){let t=this.ensureWorkspace();return Rn({...e,dbPath:t.knowledgeDbPath,config:this.config()})}enqueueReindex(e={}){let t=this.ensureWorkspace();return ft({...e,dbPath:t.knowledgeDbPath,config:this.config()})}async refreshEmbeddings(e={}){let t=this.ensureWorkspace();return xn({...e,dbPath:t.knowledgeDbPath,config:this.config()})}providerStatus(e=process.env){return $t(this.config(),e)}modelRegistry(){return rt(this.config())}embeddingStatus(){let e=this.ensureWorkspace();return Yt(e.knowledgeDbPath)}async indexEmbeddings(e={}){let t=this.ensureWorkspace();return Te({...e,dbPath:t.knowledgeDbPath,config:this.config()})}async semanticSearch(e){let t=this.ensureWorkspace();return Re({...e,dbPath:t.knowledgeDbPath,config:this.config()})}async search(e){let t=this.ensureWorkspace();return Ae({...e,dbPath:t.knowledgeDbPath,config:this.config()})}async retrieveContext(e){let t=this.ensureWorkspace();return Le({...e,dbPath:t.knowledgeDbPath,config:this.config()})}async runPrompt(e){let t=this.ensureWorkspace();return un({...e,dbPath:t.knowledgeDbPath,config:this.config()})}async webSearch(e){let t=this.ensureWorkspace();return Pn({...e,dbPath:t.knowledgeDbPath,config:this.config(),safetyPolicy:this.safetyPolicy()})}}function Vn(e={}){return new Yn(e)}import{basename as go}from"path";var pe={name:"@hasna/knowledge",version:"0.2.26",description:"Agent-friendly local knowledge CLI with JSON output, pagination, and safe destructive actions",type:"module",bin:{knowledge:"bin/open-knowledge.js","open-knowledge":"bin/open-knowledge.js","open-knowledge-mcp":"bin/open-knowledge-mcp.js"},files:["bin","src","docs","LICENSE","README.md"],scripts:{test:"bun test","test:cli":"bun test tests/cli.test.ts",build:"bun build --target=bun --outfile=bin/open-knowledge.js --minify --external @aws-sdk/client-s3 --external @aws-sdk/credential-providers --external ai --external @ai-sdk/openai --external @ai-sdk/anthropic --external @ai-sdk/deepseek src/cli.ts && bun build --target=bun --outfile=bin/open-knowledge-mcp.js --external @modelcontextprotocol/sdk --external @aws-sdk/client-s3 --external @aws-sdk/credential-providers --external ai --external @ai-sdk/openai --external @ai-sdk/anthropic --external @ai-sdk/deepseek src/mcp.js",prepublishOnly:"bun run build",postinstall:"bun run build"},keywords:["knowledge","cli","agents","json","notes","local","store"],license:"Apache-2.0",publishConfig:{registry:"https://registry.npmjs.org",access:"public"},repository:{type:"git",url:"git+https://github.com/hasna/knowledge.git"},bugs:{url:"https://github.com/hasna/knowledge/issues"},author:"Hasna Inc. <hasna@example.com>",engines:{bun:">=1.0",node:">=18"},dependencies:{"@aws-sdk/client-s3":"^3.1063.0","@aws-sdk/credential-providers":"^3.1063.0","@ai-sdk/anthropic":"^3.0.81","@ai-sdk/deepseek":"^2.0.35","@ai-sdk/openai":"^3.0.68","@modelcontextprotocol/sdk":"^1.29.0",ai:"^6.0.197",zod:"^4.3.6"},devDependencies:{"@types/bun":"^1.3.14"}};var Qn={debug:0,info:1,warn:2,error:3},po=()=>{if(process.env.DEBUG)return"debug";if(process.env.LOG_LEVEL==="debug")return"debug";if(process.env.LOG_LEVEL==="warn")return"warn";if(process.env.LOG_LEVEL==="error")return"error";return"info"};function ee(e,t,n){if(Qn[e]<Qn[po()])return;let r={debug:"[DEBUG]",info:"[INFO]",warn:"[WARN]",error:"[ERROR]"}[e],i=n?`${r} ${t} ${JSON.stringify(n)}`:`${r} ${t}`;if(e==="error")console.error(i);else console.error(i)}var Zn=["add","list","get","delete","update","archive","restore","upsert","untag","export","prune","dedupe","stats","paths","setup","auth","remote","storage","db","wiki","source","ingest","reindex","search","web","ask","build","embeddings","providers","safety","help"],er={ls:"list",rm:"delete",edit:"update",unarchive:"restore",knowledge:"ask"};function mo(e){let t=[],n={};for(let r=0;r<e.length;r+=1){let i=e[r];if(!i.startsWith("-")){t.push(i);continue}switch(i){case"--json":n.json=!0;break;case"--yes":case"-y":n.yes=!0;break;case"--help":case"-h":n.help=!0;break;case"--version":case"-v":n.version=!0;break;case"--desc":n.desc=!0;break;case"--page":case"-p":n.page=Number(e[r+1]),r+=1;break;case"--limit":case"-l":n.limit=Number(e[r+1]),r+=1;break;case"--search":case"-s":n.search=e[r+1],r+=1;break;case"--sort":n.sort=e[r+1],r+=1;break;case"--id":n.id=e[r+1],r+=1;break;case"--store":n.store=e[r+1],r+=1;break;case"--title":n.title=e[r+1],r+=1;break;case"--content":n.content=e[r+1],r+=1;break;case"--url":n.url=e[r+1],r+=1;break;case"--tag":case"-t":n.tag=e[r+1],r+=1;break;case"--format":n.format=e[r+1],r+=1;break;case"--completions":n.completions=e[r+1],r+=1;break;case"--purpose":n.purpose=e[r+1],r+=1;break;case"--model":n.model=e[r+1],r+=1;break;case"--dimensions":n.dimensions=Number(e[r+1]),r+=1;break;case"--semantic":n.semantic=!0;break;case"--context":n.context=!0;break;case"--generate":n.generate=!0;break;case"--approve-write":n.approveWrite=!0;break;case"--provider":n.provider=e[r+1],r+=1;break;case"--mode":n.mode=e[r+1],r+=1;break;case"--api-url":n.apiUrl=e[r+1],r+=1;break;case"--api-key":n.apiKey=e[r+1],r+=1;break;case"--email":n.email=e[r+1],r+=1;break;case"--org":n.org=e[r+1],r+=1;break;case"--org-id":n.orgId=e[r+1],r+=1;break;case"--user-id":n.userId=e[r+1],r+=1;break;case"--domain":n.domain=[...n.domain??[],e[r+1]],r+=1;break;case"--file-results":n.fileResults=!0;break;case"--full":n.full=!0;break;case"--fake":n.fake=!0;break;case"--no-color":n.noColor=!0;break;case"--scope":n.scope=e[r+1],r+=1;break;case"--older-than":n.olderThan=Number(e[r+1]),r+=1;break;case"--empty":n.empty=!0;break;case"--archived":n.archived=!0;break;case"--include-archived":n.includeArchived=!0;break;default:throw Error(`Unknown flag: ${i}. Run 'open-knowledge --help' for valid options.`)}}return{positional:t,flags:n}}function ho(e){if(!e)return"";return er[e]??e}function Eo(e,t){let n=Array.from({length:e.length+1},()=>Array(t.length+1).fill(0));for(let r=0;r<=e.length;r+=1)n[r][0]=r;for(let r=0;r<=t.length;r+=1)n[0][r]=r;for(let r=1;r<=e.length;r+=1)for(let i=1;i<=t.length;i+=1){let s=e[r-1]===t[i-1]?0:1;n[r][i]=Math.min(n[r-1][i]+1,n[r][i-1]+1,n[r-1][i-1]+s)}return n[e.length][t.length]}function ko(e){if(!e)return"";let t=[...Zn,...Object.keys(er)],n="",r=Number.POSITIVE_INFINITY;for(let i of t){let s=Eo(e,i);if(s<r)r=s,n=i}return r<=3?n:""}function yo(){return go(process.argv[1]??"")==="knowledge"}function bo(){console.log(`open-knowledge - local agent knowledge store
|
|
625
|
+
updated_at = excluded.updated_at`,[c,"wiki/README.md","Wiki",s.uri,s.hash??null,"active",JSON.stringify({artifact_key:s.key,provenance:kt(s)}),r,r]),fo(e,c,"Wiki",s,zn(),r)}}function go(e){if(!e)return;let t=e.trim().toLowerCase();if(t==="local"||t==="offline")return"local";if(t==="hosted"||t==="remote"||t==="knowledge.hasna.xyz")return"hosted";throw Error("Invalid setup mode. Use hosted or local.")}class Qn{options;ensuredWorkspace;cachedConfig;constructor(e={}){this.options=e}get scope(){return this.options.scope??"global"}get workspace(){return this.ensuredWorkspace??Rt(this.options.scope,this.options.cwd)}ensureWorkspace(){if(!this.ensuredWorkspace)this.ensuredWorkspace=Tt(this.workspace.home);return this.ensuredWorkspace}jsonStorePath(){return this.ensureWorkspace().jsonStorePath}config(){if(!this.cachedConfig){let e=this.ensureWorkspace();this.cachedConfig=St(e.configPath)}return this.cachedConfig}safetyPolicy(){return En(this.config(),this.ensureWorkspace())}artifactStore(){return Ut(this.config(),this.ensureWorkspace())}storageContract(){return mt(this.config(),this.ensureWorkspace(),this.scope)}validateStorage(){return pt(this.config(),this.ensureWorkspace())}setup(e={}){let t=this.ensureWorkspace(),n=this.config(),r=go(e.mode)??n.mode,i=e.apiUrl?X(e.apiUrl):n.hosted?.api_url?X(n.hosted.api_url):null,s={...n,mode:r,hosted:{...n.hosted??{},...i?{api_url:i}:{}},storage:e.canonicalHasnaXyz?wt():n.storage};xt(t.configPath,s),this.cachedConfig=s;let c=mt(s,t,this.scope);return{ok:!0,mode:r,api_url:s.hosted?.api_url??null,storage_type:s.storage.type,artifact_uri_prefix:c.artifact_store.uri_prefix,canonical_hasna_xyz:c.canonical_hasna_xyz,config_path:t.configPath,next:r==="hosted"?["open-knowledge auth login --api-key <key>","open-knowledge storage status --json","open-knowledge remote contracts --json"]:["open-knowledge search <query>","knowledge <prompt>"],message:`Set knowledge mode to ${r}`}}authStatus(e=process.env){return Kt(this.config(),e)}saveAuth(e,t=process.env){let n=e.apiUrl??this.config().hosted?.api_url;return Mt({api_key:e.apiKey,email:e.email,org_id:e.orgId,org_slug:e.orgSlug,user_id:e.userId,api_url:n},t)}clearAuth(e=process.env){return jt(e)}remoteContract(){let e=this.storageContract();return Cn({mode:this.config().mode,sourceSchemes:this.config().sources.allowed_schemes,storageType:e.artifact_store.type,artifactUriPrefix:e.artifact_store.uri_prefix})}remoteClient(e=process.env){return We.fromConfig(this.config(),e)}paths(){let e=this.ensureWorkspace();return{ok:!0,scope:this.scope,home:e.home,config_path:e.configPath,json_store_path:e.jsonStorePath,knowledge_db_path:e.knowledgeDbPath,artifacts_dir:e.artifactsDir,indexes_dir:e.indexesDir,logs_dir:e.logsDir,runs_dir:e.runsDir,schemas_dir:e.schemasDir,wiki_dir:e.wikiDir,config:this.config(),message:e.home}}initDb(){return x(this.ensureWorkspace().knowledgeDbPath)}dbStats(){let e=this.ensureWorkspace();return x(e.knowledgeDbPath),It(e.knowledgeDbPath)}async initWiki(){let e=this.ensureWorkspace();x(e.knowledgeDbPath);let t=await Yn(this.artifactStore()),n=b(e.knowledgeDbPath);try{me(n,t.artifacts),Vn(n,t.artifacts)}finally{n.close()}return t}async compileWiki(e={}){let t=this.ensureWorkspace();return Bn({...e,dbPath:t.knowledgeDbPath,store:this.artifactStore()})}async fileAnswer(e){let t=this.ensureWorkspace(),n=await this.retrieveContext({query:e.prompt,limit:e.limit,semantic:e.semantic,modelRef:e.modelRef,dimensions:e.dimensions,fake:e.fake});return Jn({dbPath:t.knowledgeDbPath,store:this.artifactStore(),prompt:e.prompt,answer:e.answer,context:n,approveWrite:e.approveWrite})}lintWiki(){let e=this.ensureWorkspace();return Gn({dbPath:e.knowledgeDbPath})}async ingestManifest(e){let t=this.ensureWorkspace();return wn({dbPath:t.knowledgeDbPath,input:e,config:this.config(),safetyPolicy:this.safetyPolicy()})}async ingestSource(e,t){let n=this.ensureWorkspace();return Sn({dbPath:n.knowledgeDbPath,sourceRef:e,purpose:t,config:this.config(),safetyPolicy:this.safetyPolicy()})}async resolveSource(e,t={}){let n=this.ensureWorkspace();return Me({dbPath:n.knowledgeDbPath,sourceRef:e,purpose:t.purpose,limit:t.limit,safetyPolicy:this.safetyPolicy()})}async consumeOutbox(e){let t=this.ensureWorkspace();return vn({dbPath:t.knowledgeDbPath,input:e,config:this.config(),safetyPolicy:this.safetyPolicy()})}reindexHealth(e={}){let t=this.ensureWorkspace();return Nn({...e,dbPath:t.knowledgeDbPath,config:this.config()})}enqueueReindex(e={}){let t=this.ensureWorkspace();return gt({...e,dbPath:t.knowledgeDbPath,config:this.config()})}async refreshEmbeddings(e={}){let t=this.ensureWorkspace();return An({...e,dbPath:t.knowledgeDbPath,config:this.config()})}providerStatus(e=process.env){return Ht(this.config(),e)}modelRegistry(){return it(this.config())}embeddingStatus(){let e=this.ensureWorkspace();return Zt(e.knowledgeDbPath)}async indexEmbeddings(e={}){let t=this.ensureWorkspace();return Se({...e,dbPath:t.knowledgeDbPath,config:this.config()})}async semanticSearch(e){let t=this.ensureWorkspace();return xe({...e,dbPath:t.knowledgeDbPath,config:this.config()})}async search(e){let t=this.ensureWorkspace();return Ie({...e,dbPath:t.knowledgeDbPath,config:this.config()})}async retrieveContext(e){let t=this.ensureWorkspace();return Ce({...e,dbPath:t.knowledgeDbPath,config:this.config()})}async runPrompt(e){let t=this.ensureWorkspace();return _n({...e,dbPath:t.knowledgeDbPath,config:this.config()})}async webSearch(e){let t=this.ensureWorkspace();return Fn({...e,dbPath:t.knowledgeDbPath,config:this.config(),safetyPolicy:this.safetyPolicy()})}}function Zn(e={}){return new Qn(e)}import{basename as po}from"path";var pe={name:"@hasna/knowledge",version:"0.2.28",description:"Agent-friendly local knowledge CLI with JSON output, pagination, and safe destructive actions",type:"module",exports:{".":{import:"./dist/index.js",types:"./dist/index.d.ts"}},main:"./dist/index.js",types:"./dist/index.d.ts",bin:{knowledge:"bin/open-knowledge.js","open-knowledge":"bin/open-knowledge.js","open-knowledge-mcp":"bin/open-knowledge-mcp.js"},files:["bin","dist","docs","LICENSE","README.md"],scripts:{test:"bun test","test:cli":"bun test tests/cli.test.ts",build:"rm -rf dist && bun build --target=bun --outfile=bin/open-knowledge.js --minify --external @aws-sdk/client-s3 --external @aws-sdk/credential-providers --external ai --external @ai-sdk/openai --external @ai-sdk/anthropic --external @ai-sdk/deepseek src/cli.ts && bun build --target=bun --outfile=bin/open-knowledge-mcp.js --external @modelcontextprotocol/sdk --external @aws-sdk/client-s3 --external @aws-sdk/credential-providers --external ai --external @ai-sdk/openai --external @ai-sdk/anthropic --external @ai-sdk/deepseek src/mcp.js && bun build ./src/index.ts --outdir ./dist --target bun --external @aws-sdk/client-s3 --external @aws-sdk/credential-providers --external ai --external @ai-sdk/openai --external @ai-sdk/anthropic --external @ai-sdk/deepseek && bunx tsc -p tsconfig.build.json",prepublishOnly:"bun run build"},keywords:["knowledge","cli","agents","json","notes","local","store"],license:"Apache-2.0",publishConfig:{registry:"https://registry.npmjs.org",access:"public"},repository:{type:"git",url:"git+https://github.com/hasna/knowledge.git"},bugs:{url:"https://github.com/hasna/knowledge/issues"},author:"Hasna Inc. <hasna@example.com>",engines:{bun:">=1.0",node:">=18"},dependencies:{"@ai-sdk/anthropic":"^3.0.81","@ai-sdk/deepseek":"^2.0.35","@ai-sdk/openai":"^3.0.68","@aws-sdk/client-s3":"^3.1063.0","@aws-sdk/credential-providers":"^3.1063.0","@modelcontextprotocol/sdk":"^1.29.0","@types/json-schema":"^7.0.15",ai:"^6.0.197",zod:"^4.3.6"},devDependencies:{"@types/bun":"^1.3.14"}};var er={debug:0,info:1,warn:2,error:3},ho=()=>{if(process.env.DEBUG)return"debug";if(process.env.LOG_LEVEL==="debug")return"debug";if(process.env.LOG_LEVEL==="warn")return"warn";if(process.env.LOG_LEVEL==="error")return"error";return"info"};function te(e,t,n){if(er[e]<er[ho()])return;let r={debug:"[DEBUG]",info:"[INFO]",warn:"[WARN]",error:"[ERROR]"}[e],i=n?`${r} ${t} ${JSON.stringify(n)}`:`${r} ${t}`;if(e==="error")console.error(i);else console.error(i)}var tr=["add","list","get","delete","update","archive","restore","upsert","untag","export","prune","dedupe","stats","paths","setup","auth","remote","storage","db","wiki","source","ingest","reindex","search","web","ask","build","embeddings","providers","safety","help"],nr={ls:"list",rm:"delete",edit:"update",unarchive:"restore",knowledge:"ask"};function Eo(e){let t=[],n={};for(let r=0;r<e.length;r+=1){let i=e[r];if(!i.startsWith("-")){t.push(i);continue}switch(i){case"--json":n.json=!0;break;case"--yes":case"-y":n.yes=!0;break;case"--help":case"-h":n.help=!0;break;case"--version":case"-v":n.version=!0;break;case"--desc":n.desc=!0;break;case"--page":case"-p":n.page=Number(e[r+1]),r+=1;break;case"--limit":case"-l":n.limit=Number(e[r+1]),r+=1;break;case"--search":case"-s":n.search=e[r+1],r+=1;break;case"--sort":n.sort=e[r+1],r+=1;break;case"--id":n.id=e[r+1],r+=1;break;case"--store":n.store=e[r+1],r+=1;break;case"--title":n.title=e[r+1],r+=1;break;case"--content":n.content=e[r+1],r+=1;break;case"--url":n.url=e[r+1],r+=1;break;case"--tag":case"-t":n.tag=e[r+1],r+=1;break;case"--format":n.format=e[r+1],r+=1;break;case"--completions":n.completions=e[r+1],r+=1;break;case"--purpose":n.purpose=e[r+1],r+=1;break;case"--model":n.model=e[r+1],r+=1;break;case"--dimensions":n.dimensions=Number(e[r+1]),r+=1;break;case"--semantic":n.semantic=!0;break;case"--context":n.context=!0;break;case"--generate":n.generate=!0;break;case"--approve-write":n.approveWrite=!0;break;case"--provider":n.provider=e[r+1],r+=1;break;case"--mode":n.mode=e[r+1],r+=1;break;case"--api-url":n.apiUrl=e[r+1],r+=1;break;case"--canonical-hasna-xyz":n.canonicalHasnaXyz=!0;break;case"--api-key":n.apiKey=e[r+1],r+=1;break;case"--email":n.email=e[r+1],r+=1;break;case"--org":n.org=e[r+1],r+=1;break;case"--org-id":n.orgId=e[r+1],r+=1;break;case"--user-id":n.userId=e[r+1],r+=1;break;case"--domain":n.domain=[...n.domain??[],e[r+1]],r+=1;break;case"--file-results":n.fileResults=!0;break;case"--full":n.full=!0;break;case"--fake":n.fake=!0;break;case"--no-color":n.noColor=!0;break;case"--scope":n.scope=e[r+1],r+=1;break;case"--older-than":n.olderThan=Number(e[r+1]),r+=1;break;case"--empty":n.empty=!0;break;case"--archived":n.archived=!0;break;case"--include-archived":n.includeArchived=!0;break;default:throw Error(`Unknown flag: ${i}. Run 'open-knowledge --help' for valid options.`)}}return{positional:t,flags:n}}function yo(e){if(!e)return"";return nr[e]??e}function ko(e,t){let n=Array.from({length:e.length+1},()=>Array(t.length+1).fill(0));for(let r=0;r<=e.length;r+=1)n[r][0]=r;for(let r=0;r<=t.length;r+=1)n[0][r]=r;for(let r=1;r<=e.length;r+=1)for(let i=1;i<=t.length;i+=1){let s=e[r-1]===t[i-1]?0:1;n[r][i]=Math.min(n[r-1][i]+1,n[r][i-1]+1,n[r-1][i-1]+s)}return n[e.length][t.length]}function bo(e){if(!e)return"";let t=[...tr,...Object.keys(nr)],n="",r=Number.POSITIVE_INFINITY;for(let i of t){let s=ko(e,i);if(s<r)r=s,n=i}return r<=3?n:""}function vo(){return po(process.argv[1]??"")==="knowledge"}function wo(){console.log(`open-knowledge - local agent knowledge store
|
|
626
626
|
|
|
627
627
|
Usage:
|
|
628
628
|
open-knowledge <command> [options]
|
|
@@ -642,7 +642,7 @@ Commands:
|
|
|
642
642
|
dedupe Remove duplicate items by title+content (requires --yes)
|
|
643
643
|
stats Show knowledge base statistics
|
|
644
644
|
paths Show resolved workspace/store paths
|
|
645
|
-
setup Configure local or
|
|
645
|
+
setup Configure local, hosted, or canonical Hasna XYZ S3 mode
|
|
646
646
|
auth login|whoami|logout Manage hosted API credentials
|
|
647
647
|
remote contracts|status Inspect hosted client contracts/readiness
|
|
648
648
|
storage status|validate Inspect local/S3 artifact storage contract
|
|
@@ -719,5 +719,5 @@ Export Options:
|
|
|
719
719
|
|
|
720
720
|
Prune Options:
|
|
721
721
|
--older-than <days> Remove items older than N days
|
|
722
|
-
--empty Remove items with empty content`)}function
|
|
723
|
-
_open_knowledge() { _arguments -C "1: :(add list get update archive restore upsert untag delete export prune dedupe stats paths setup auth remote storage db wiki source ingest reindex search web ask build embeddings providers safety help ls rm edit unarchive knowledge)" "(--json)--json" "(--yes)-y" "(--help)--help" "(--version)--version" "(--desc)--desc" "(--archived)--archived" "(--include-archived)--include-archived" "(--semantic)--semantic" "(--context)--context" "(--generate)--generate" "(--approve-write)--approve-write" "(--file-results)--file-results" "(--full)--full" "(--fake)--fake" "(-p --page)"{-p,--page}"[page number]:number:" "(-l --limit)"{-l,--limit}"[items per page]:number:" "(-s --search)"{-s,--search}"[search text]:text:" "(--sort)--sort"{created,title}:" "(--id)--id[item id]:id:" "(--store)--store[store path]:path:" "(--title)--title[new title]:" "(--content)--content[new content]:" "(--url)--url[source url]:" "(-t --tag)"{-t,--tag}"[tag]:tag:" "(--format)--format[json|jsonl]:" "(--completions)--completions[output completions]:shell:(bash zsh fish):" "(--purpose)--purpose[purpose]:" "(--model)--model[model ref]:" "(--dimensions)--dimensions[embedding dimensions]:number:" "(--provider)--provider[provider]:" "(--mode)--mode"{local,hosted}:" "(--api-url)--api-url[hosted API URL]:" "(--api-key)--api-key[hosted API key]:" "(--email)--email[email]:" "(--org)--org[org slug]:" "(--org-id)--org-id[org id]:" "(--user-id)--user-id[user id]:" "(--domain)--domain[domain]:" "(--no-color)--no-color[disable color]" "(--scope)--scope"{local,global,project}:" }; _open_knowledge`);else if(a==="fish")console.log('complete -c open-knowledge -f; complete -c open-knowledge -a "add list get update archive restore upsert untag delete export prune dedupe stats paths setup auth remote storage db wiki source ingest reindex search web ask build embeddings providers safety help ls rm edit unarchive knowledge"; complete -c open-knowledge -l json; complete -c open-knowledge -l yes -s y; complete -c open-knowledge -l help -s h; complete -c open-knowledge -l version -s v; complete -c open-knowledge -l desc; complete -c open-knowledge -l archived; complete -c open-knowledge -l include-archived; complete -c open-knowledge -l semantic; complete -c open-knowledge -l context; complete -c open-knowledge -l generate; complete -c open-knowledge -l approve-write; complete -c open-knowledge -l provider; complete -c open-knowledge -l mode; complete -c open-knowledge -l api-url; complete -c open-knowledge -l api-key; complete -c open-knowledge -l email; complete -c open-knowledge -l org; complete -c open-knowledge -l org-id; complete -c open-knowledge -l user-id; complete -c open-knowledge -l domain; complete -c open-knowledge -l file-results; complete -c open-knowledge -l full; complete -c open-knowledge -l fake; complete -c open-knowledge -s p -l page; complete -c open-knowledge -s l -l limit; complete -c open-knowledge -s s -l search; complete -c open-knowledge -l sort; complete -c open-knowledge -l id; complete -c open-knowledge -l store; complete -c open-knowledge -l title; complete -c open-knowledge -l content; complete -c open-knowledge -l url; complete -c open-knowledge -s t -l tag; complete -c open-knowledge -l format; complete -c open-knowledge -l completions; complete -c open-knowledge -l purpose; complete -c open-knowledge -l model; complete -c open-knowledge -l dimensions; complete -c open-knowledge -l no-color; complete -c open-knowledge -l scope -a "local global project"');else throw Error("Invalid --completions value. Use 'bash', 'zsh', or 'fish'.");return}let r=ho(t[0]),i=1;if(yo()&&r&&!Zn.includes(r))r="ask",i=0;if(!r||n.help||r==="help"){wo(t[1]);return}let s=Vn({scope:n.scope}),u=n.store;if(!u)if(n.scope==="project"||n.scope==="local")u=s.jsonStorePath();else u=Ge();if(r==="paths"){k(s.paths(),n.json);return}if(r==="setup"){let a=s.setup({mode:n.mode,apiUrl:n.apiUrl});k(a,n.json);return}if(r==="auth"){let a=t[1]??"whoami";if(a==="whoami"||a==="status"){let c=s.authStatus(process.env);k({ok:!0,...c,message:c.authenticated?`Authenticated via ${c.source}`:"Not authenticated"},n.json);return}if(a==="login"){let c=n.apiKey??process.env.KNOWLEDGE_API_KEY??process.env.HASNA_KNOWLEDGE_API_KEY;if(!c)throw Error("Usage: open-knowledge auth login --api-key <key> [--email <email>]");let o=s.saveAuth({apiKey:c,email:n.email,orgSlug:n.org,orgId:n.orgId,userId:n.userId,apiUrl:n.apiUrl},process.env);k({ok:!0,authenticated:!0,email:o.email??null,org_slug:o.org_slug??null,api_url:o.api_url??s.authStatus(process.env).api_url,auth_path:s.authStatus(process.env).auth_path,message:`Saved hosted credentials for ${o.email??"API key"}`},n.json);return}if(a==="logout"){let c=s.clearAuth(process.env);k({ok:!0,removed:c,message:c?"Removed hosted credentials":"No hosted credentials found"},n.json);return}throw Error("Invalid auth action. Use 'login', 'whoami', or 'logout'.")}if(r==="remote"){let a=t[1]??"status";if(a==="contracts"||a==="contract"){let c=s.authStatus(process.env);k({ok:!0,authenticated:c.authenticated,api_url:c.api_url,contract:s.remoteContract(),message:`Remote contract v${s.remoteContract().contract_version}`},n.json);return}if(a==="status"){let c=s.authStatus(process.env),o=s.remoteContract();k({ok:!0,mode:s.config().mode,authenticated:c.authenticated,auth_source:c.source,api_url:c.api_url,client_ready:Boolean(s.remoteClient(process.env)),contract_version:o.contract_version,capabilities:o.capabilities,message:c.authenticated?`Remote client ready for ${c.api_url}`:"Remote client not authenticated"},n.json);return}throw Error("Invalid remote action. Use 'contracts' or 'status'.")}if(r==="storage"){let a=t[1]??"status";if(a==="status"){let c=s.storageContract(),o=s.validateStorage();k({ok:o.ok,...c,validation:o,message:`${c.storage_type} artifact storage at ${c.artifact_store.uri_prefix}`},n.json);return}if(a==="validate"){let c=s.validateStorage();k({ok:c.ok,validation:c,message:c.ok?"Storage contract valid":`Storage contract invalid: ${c.errors.join("; ")}`},n.json);return}throw Error("Invalid storage action. Use 'status' or 'validate'.")}if(r==="db"){let a=t[1]??"init";if(a!=="init"&&a!=="stats")throw Error("Invalid db action. Use 'init' or 'stats'.");if(a==="init"){let o=s.initDb();k({ok:!0,...o,message:`Initialized ${o.path}`},n.json);return}let c=s.dbStats();k({ok:!0,path:s.workspace.knowledgeDbPath,...c,message:`knowledge.db schema v${c.schema_version}`},n.json);return}if(r==="wiki"){let a=t[1]??"init";if(a==="init"){let c=await s.initWiki();k({ok:!0,...c,message:`Initialized wiki layout in ${s.workspace.home}`},n.json);return}if(a==="compile"){let c=t.slice(2),o=c.filter((f)=>/^(open-files|file|s3|https?):\/\//.test(f)),_=c.filter((f)=>!/^(open-files|file|s3|https?):\/\//.test(f)).join(" "),l=await s.compileWiki({title:n.title,query:_||n.search,sourceRefs:o.length>0?o:void 0,limit:n.limit});k({ok:!0,...l,message:`Compiled wiki page ${l.path}`},n.json);return}if(a==="file-answer"||a==="answer"){let c=t.slice(2).join(" ");if(!c)throw Error("Usage: open-knowledge wiki file-answer <prompt> --content <answer> --approve-write");if(!n.content)throw Error("Missing --content <answer> for wiki file-answer.");let o=await s.fileAnswer({prompt:c,answer:n.content,approveWrite:n.approveWrite,limit:n.limit,semantic:n.semantic,modelRef:n.model,dimensions:n.dimensions,fake:n.fake});k({ok:!0,...o},n.json);return}if(a==="lint"){let c=s.lintWiki();k({ok:c.ok,...c,message:c.ok?"Wiki lint passed":`Wiki lint found ${c.issue_count} issue(s)`},n.json);return}throw Error("Invalid wiki action. Use 'init', 'compile', 'file-answer', or 'lint'.")}if(r==="safety"){let a=t[1]??"status",c=s.ensureWorkspace(),o=s.safetyPolicy();s.initDb();let _=b(c.knowledgeDbPath);try{if(a==="status"){k({ok:!0,mode:o.mode,workspace:c.home,allow_write_roots:o.allowWriteRoots,read_only_source_access:o.readOnlySourceAccess,network:o.network,redaction:o.redaction,approvals:o.approvals,message:`Safety policy: ${o.mode}`},n.json);return}if(a==="check"){let l=t[2]??"generated_write",f=t[3]??null,p;try{if(l==="web_search")ne(o),p={action:l,target_uri:f,approval_required:!1,approved:!0,decision:"allow"};else if(l==="s3_read"){if(!f)throw Error("safety check s3_read requires an s3:// target.");B(f,o),p={action:l,target_uri:f,approval_required:!1,approved:!0,decision:"allow"}}else p=hn(_,o,l,f);A(_,{event_type:"safety_check",action:l,target_uri:f,decision:p.decision==="allow"?"allow":"requires_approval",metadata:p}),k({ok:!0,...p,message:`Safety check ${p.decision}`},n.json);return}catch(E){throw A(_,{event_type:"safety_check",action:l,target_uri:f,decision:"deny",metadata:{error:E instanceof Error?E.message:String(E)}}),E}}if(a==="approve"){let l=t[2]??"generated_write",f=t[3]??null,p=mn(_,{action:l,target_uri:f,reason:"local-cli approval",metadata:{scope:n.scope??"global"}});A(_,{event_type:"approval",action:l,target_uri:f,decision:"allow",metadata:{approval_id:p.id}}),k({ok:!0,...p,action:l,target_uri:f,message:`Approved ${l}`},n.json);return}if(a==="audit"){let l=_.query("SELECT id, event_type, action, target_uri, decision, metadata_json, created_at FROM audit_events ORDER BY created_at DESC LIMIT 50").all().map((f)=>({id:f.id,event_type:f.event_type,action:f.action,target_uri:f.target_uri,decision:f.decision,metadata:JSON.parse(f.metadata_json),created_at:f.created_at}));k({ok:!0,events:l,message:`${l.length} audit event(s)`},n.json);return}if(a==="redact"){let l=t.slice(2).join(" ");if(!l)throw Error("Usage: open-knowledge safety redact <text>");let f=Ce(l,o);if(f.findings.length>0)Pe(_,{source_uri:"safety://redact",findings:f.findings,metadata:{command:"safety redact"}});A(_,{event_type:"redaction",action:"safety_redact",target_uri:"safety://redact",decision:f.findings.length>0?"redacted":"allow",metadata:{findings:f.findings.length}}),k({ok:!0,text:f.text,findings:f.findings,message:`Redacted ${f.findings.length} finding(s)`},n.json);return}throw Error("Invalid safety action. Use 'status', 'check', 'approve', 'audit', or 'redact'.")}finally{_.close()}}if(r==="source"){if((t[1]??"")!=="resolve")throw Error("Invalid source action. Use 'resolve'.");let c=t[2];if(!c)throw Error("Usage: open-knowledge source resolve <source-ref>");let o=await s.resolveSource(c,{purpose:n.purpose,limit:n.limit});k({ok:!0,...o,message:o.resolved?`Resolved ${o.source_ref} (${o.content.chunks_returned}/${o.content.chunks_total} chunks)`:`Source not indexed: ${c}`},n.json);return}if(r==="ingest"){let a=t[1]??"";if(a==="manifest"){let c=t[2];if(!c)throw Error("Usage: open-knowledge ingest manifest <file|s3://bucket/key>");let o=await s.ingestManifest(c);k({ok:!0,...o,message:`Ingested ${o.items_seen} manifest item(s)`},n.json);return}if(a==="source"){let c=t[2];if(!c)throw Error("Usage: open-knowledge ingest source <source-ref>");let o=await s.ingestSource(c,n.purpose);k({ok:!0,...o,message:`Ingested source ${o.source_ref} (${o.chunks_inserted} chunks)`},n.json);return}throw Error("Invalid ingest action. Use 'manifest' or 'source'.")}if(r==="reindex"){let a=t[1]??"status";if(a==="status"){let c=s.reindexHealth({modelRef:n.model,dimensions:n.dimensions,fake:n.fake});k({ok:!0,...c,message:`${c.missing_embeddings} chunk(s) missing embeddings`},n.json);return}if(a==="enqueue"){let c=s.enqueueReindex({modelRef:n.model,dimensions:n.dimensions,fake:n.fake});k({ok:!0,...c,message:`Queued ${c.enqueued} embedding refresh item(s)`},n.json);return}if(a==="embeddings"){let c=await s.refreshEmbeddings({full:n.full,limit:n.limit,modelRef:n.model,dimensions:n.dimensions,fake:n.fake});k({ok:!0,...c,message:`Embedded ${c.indexed.chunks_embedded} chunk(s)`},n.json);return}if(a==="outbox"){let c=t[2];if(!c)throw Error("Usage: open-knowledge reindex outbox <file|s3://bucket/key>");let o=await s.consumeOutbox(c);k({ok:!0,...o,message:`Consumed ${o.events_seen} outbox event(s)`},n.json);return}throw Error("Invalid reindex action. Use 'status', 'enqueue', 'embeddings', or 'outbox'.")}if(r==="embeddings"){let a=t[1]??"status";if(a==="status"){let c=s.embeddingStatus();k({ok:!0,...c,message:`${c.total_vector_entries} vector index entries`},n.json);return}if(a==="index"){let c=await s.indexEmbeddings({limit:n.limit,modelRef:n.model,dimensions:n.dimensions,fake:n.fake});k({ok:!0,...c,message:`Embedded ${c.chunks_embedded} chunk(s)`},n.json);return}if(a==="search"){let c=t.slice(2).join(" ");if(!c)throw Error("Usage: open-knowledge embeddings search <query>");let o=await s.semanticSearch({query:c,limit:n.limit,modelRef:n.model,dimensions:n.dimensions,fake:n.fake});k({ok:!0,...o,message:`${o.results.length} semantic result(s)`},n.json);return}throw Error("Invalid embeddings action. Use 'status', 'index', or 'search'.")}if(r==="search"){let a=t.slice(1).join(" ");if(!a)throw Error("Usage: open-knowledge search <query>");if(n.context){let o=await s.retrieveContext({query:a,limit:n.limit,semantic:n.semantic,modelRef:n.model,dimensions:n.dimensions,fake:n.fake});k({ok:!0,...o,message:`${o.excerpts.length} context excerpt(s)`},n.json);return}let c=await s.search({query:a,limit:n.limit,semantic:n.semantic,modelRef:n.model,dimensions:n.dimensions,fake:n.fake});k({ok:!0,...c,message:`${c.results.length} search result(s)`},n.json);return}if(r==="web"){if((t[1]??"search")!=="search")throw Error("Invalid web action. Use 'search'.");let c=t.slice(2).join(" ");if(!c)throw Error("Usage: open-knowledge web search <query>");let o=await s.webSearch({query:c,limit:n.limit,modelRef:n.model,provider:n.provider,domains:n.domain,fake:n.fake,fileResults:n.fileResults});k({ok:!0,...o,message:`${o.sources.length} web source(s)`},n.json);return}if(r==="ask"||r==="build"){let a=t.slice(i).join(" ");if(!a)throw Error("Usage: open-knowledge ask <prompt>");let c=await s.runPrompt({prompt:a,limit:n.limit,semantic:n.semantic,modelRef:n.model,dimensions:n.dimensions,fake:n.fake,generate:n.generate,approveWrite:n.approveWrite});k({ok:!0,...c,message:c.generated?"Generated answer with citations":"Prepared citation context draft"},n.json);return}if(r==="providers"){let a=t[1]??"status";if(a==="status"){let c=s.providerStatus(),o=c.providers.filter((_)=>_.configured).length;k({ok:!0,...c,message:`${o}/${c.providers.length} provider credential(s) configured`},n.json);return}if(a==="models"){let c=s.modelRegistry();k({ok:!0,models:c,message:`${c.length} model alias(es)`},n.json);return}if(a==="check"){let c=t[2]??"default",o=q(c,s.config()),_=L(o),l=z(_.provider,s.config());k({ok:!0,target:c,model_ref:o,provider:_.provider,model:_.model,credential:l,message:`${_.provider} credentials configured`},n.json);return}throw Error("Invalid providers action. Use 'status', 'models', or 'check'.")}if(Ye(u),r==="add"){let a=t[1],c=t[2];if(!a||!c)throw Error("Usage: open-knowledge add <title> <content>");D(u,()=>{let o=P(u),_={id:Ve(),title:a,content:c,url:n.url??null,tags:n.tag?[n.tag]:[],created_at:new Date().toISOString(),updated_at:new Date().toISOString()};o.items.push(_),j(u,o),ee("info","Item added",{id:_.id,title:_.title}),k({ok:!0,item:_,message:`Added ${_.id}`},n.json)});return}if(r==="list"){if(n.format!==void 0&&n.format!=="table"&&n.format!=="json")throw Error("Invalid --format value for list. Use 'table' or 'json'.");D(u,()=>{let a=P(u),c=Number.isFinite(n.page)&&n.page>0?n.page:1,o=Number.isFinite(n.limit)&&n.limit>0?n.limit:20,_=n.search?String(n.search).toLowerCase():"",l=n.tag?String(n.tag).toLowerCase():"",f=n.format==="table"||!n.json&&!n.format&&vo(n),p=n.json||n.format==="json",E=a.items;if(n.archived)E=E.filter((v)=>v.archived===!0);else if(!n.includeArchived)E=E.filter((v)=>!v.archived);if(_)E=E.filter((v)=>v.title.toLowerCase().includes(_)||v.content.toLowerCase().includes(_));if(l)E=E.filter((v)=>v.tags&&v.tags.map((He)=>He.toLowerCase()).includes(l));let{sorted:h,sort:S,direction:y}=So(E,n),m=(c-1)*o,T=h.slice(m,m+o),w=Math.max(1,Math.ceil(h.length/o));if(p){k({ok:!0,page:c,limit:o,total:h.length,total_pages:w,sort:S,direction:y,items:T},!0);return}if(T.length===0){k(`No items found (search=${_||"none"}, tag=${l||"none"})`,!1);return}if(f){let v=(W)=>W,He=`${v("ID")} ${v("TITLE")} ${v("CREATED")} ${v("URL")} ${v("TAGS")}`;console.log(He);for(let W of T)console.log(`${W.id} ${v(W.title)} ${W.created_at} ${W.url?v(W.url):""} ${W.tags?.length?v(`[${W.tags.join(", ")}]`):""}`);console.log(`Page ${c}/${w} | showing ${T.length} of ${h.length} | sort=${S} ${y} | search=${_||"none"} | tag=${l||"none"}`)}else{for(let v of T)console.log(`${v.id} ${v.title} ${v.created_at}${v.url?` ${v.url}`:""}${v.tags?.length?` [${v.tags.join(", ")}]`:""}`);console.log(`Page ${c}/${w} | showing ${T.length} of ${h.length} | sort=${S} ${y} | search=${_||"none"} | tag=${l||"none"}`)}});return}if(r==="get"){me(n),D(u,()=>{let c=P(u).items.find((o)=>o.id===n.id||o.short_id===n.id);if(!c)throw Error(`Item not found: ${n.id}`);k({ok:!0,item:c,message:`${c.id}: ${c.title}`},n.json)});return}if(r==="update"){me(n),D(u,()=>{let a=P(u),c=a.items.findIndex((_)=>_.id===n.id||_.short_id===n.id);if(c===-1)throw Error(`Item not found: ${n.id}`);let o=a.items[c];if(n.title!==void 0)o.title=n.title;if(n.content!==void 0)o.content=n.content;if(n.url!==void 0)o.url=n.url;if(n.tag!==void 0){if(o.tags=o.tags||[],!o.tags.map((_)=>_.toLowerCase()).includes(n.tag.toLowerCase()))o.tags.push(n.tag)}o.updated_at=new Date().toISOString(),a.items[c]=o,j(u,a),k({ok:!0,item:o,message:`Updated ${o.id}`},n.json)});return}if(r==="archive"||r==="restore"){me(n),D(u,()=>{let a=P(u),c=a.items.findIndex((_)=>_.id===n.id||_.short_id===n.id);if(c===-1)throw Error(`Item not found: ${n.id}`);let o=a.items[c];o.archived=r==="archive",o.updated_at=new Date().toISOString(),a.items[c]=o,j(u,a),k({ok:!0,item:o,message:`${r==="archive"?"Archived":"Restored"} ${o.id}`},n.json)});return}if(r==="untag"){if(me(n),!n.tag)throw Error("Missing required --tag. Example: open-knowledge untag --id <id> -t <tag>");D(u,()=>{let a=P(u),c=a.items.findIndex((l)=>l.id===n.id||l.short_id===n.id);if(c===-1)throw Error(`Item not found: ${n.id}`);let o=a.items[c],_=o.tags?.length??0;o.tags=(o.tags??[]).filter((l)=>l.toLowerCase()!==n.tag.toLowerCase()),o.updated_at=new Date().toISOString(),a.items[c]=o,j(u,a),k({ok:!0,item:o,removed:_-o.tags.length,message:`Removed tag from ${o.id}`},n.json)});return}if(r==="upsert"){let a=n.title??t[1],c=n.content??t[2];D(u,()=>{let o=P(u),_=n.id?o.items.findIndex((p)=>p.id===n.id||p.short_id===n.id):-1,l=new Date().toISOString();if(_===-1){if(!a||!c)throw Error("New item requires title and content. Example: open-knowledge upsert <title> <content> [--id <id>]");let p=n.id??Ve(),E={id:p,short_id:xt(p),title:a,content:c,url:n.url??null,tags:n.tag?[n.tag]:[],metadata:{},archived:!1,created_at:l,updated_at:l};o.items.push(E),j(u,o),k({ok:!0,created:!0,item:E,message:`Upserted ${E.id}`},n.json);return}let f=o.items[_];if(a!==void 0)f.title=a;if(c!==void 0)f.content=c;if(n.url!==void 0)f.url=n.url;if(n.tag!==void 0){if(f.tags=f.tags||[],!f.tags.map((p)=>p.toLowerCase()).includes(n.tag.toLowerCase()))f.tags.push(n.tag)}f.updated_at=l,o.items[_]=f,j(u,o),k({ok:!0,created:!1,item:f,message:`Upserted ${f.id}`},n.json)});return}if(r==="delete"){if(me(n),!n.yes)throw Error("Refusing delete without --yes. Re-run with: open-knowledge delete --id <id> --yes");D(u,()=>{let a=P(u),c=a.items.length;a.items=a.items.filter((_)=>_.id!==n.id&&_.short_id!==n.id);let o=c!==a.items.length;if(j(u,a),!o)throw Error(`Item not found: ${n.id}`);ee("info","Item deleted",{id:n.id}),k({ok:!0,deleted_id:n.id,message:`Deleted ${n.id}`},n.json)});return}if(r==="export"){let a=n.format??"json";if(a!=="json"&&a!=="jsonl")throw Error("Invalid --format. Use 'json' or 'jsonl'.");D(u,()=>{let c=P(u);if(a==="jsonl")for(let o of c.items)console.log(JSON.stringify(o));else k({ok:!0,items:c.items},n.json)});return}if(r==="prune"){if(!n.yes)throw Error("Refusing prune without --yes. Re-run with: open-knowledge prune --yes [--older-than <days>] [--empty]");D(u,()=>{let a=P(u),c=a.items.length;if(n.olderThan!==void 0){let _=new Date;_.setDate(_.getDate()-n.olderThan),a.items=a.items.filter((l)=>new Date(l.created_at)>=_)}if(n.empty)a.items=a.items.filter((_)=>_.content.trim().length>0);let o=c-a.items.length;j(u,a),ee("info","Prune completed",{pruned:o,remaining:a.items.length}),k({ok:!0,pruned:o,remaining:a.items.length,message:`Pruned ${o} item(s)`},n.json)});return}if(r==="dedupe"){if(!n.yes)throw Error("Refusing dedupe without --yes. Re-run with: open-knowledge dedupe --yes [--json]");D(u,()=>{let a=P(u),c=new Set,o=a.items.length;a.items=a.items.filter((l)=>{let f=`${l.title}\x00${l.content}`;if(c.has(f))return!1;return c.add(f),!0});let _=o-a.items.length;j(u,a),ee("info","Dedupe completed",{removed:_,remaining:a.items.length}),k({ok:!0,removed:_,remaining:a.items.length,message:`Dedupe removed ${_} duplicate(s)`},n.json)});return}if(r==="stats"){D(u,()=>{let a=P(u),c=a.items.filter((y)=>!y.archived),o=c.length,_=a.items.length-o,l=c.filter((y)=>y.url).length,f=c.filter((y)=>y.tags&&y.tags.length>0).length,p=o>0?c.map((y)=>y.created_at).sort()[0]:null,E=o>0?c.map((y)=>y.created_at).sort()[o-1]:null,h={};for(let y of c)for(let m of y.tags||[])h[m]=(h[m]||0)+1;let S=Object.entries(h).sort((y,m)=>m[1]-y[1]).slice(0,5).map(([y,m])=>({tag:y,count:m}));k({ok:!0,total:o,archived:_,with_url:l,with_tags:f,oldest:p,newest:E,top_tags:S,message:`${o} items | ${l} with URL | ${f} with tags`},n.json)});return}let d=ko(t[0]),g=d?` Did you mean '${d}'?`:"";throw ee("warn","Unknown command",{input:t[0],suggestion:d}),Error(`Unknown command: ${t[0]}.${g} Run 'open-knowledge --help' for available commands.`)}if(import.meta.main)To(process.argv.slice(2)).catch((e)=>{let t=e instanceof Error?e.message:String(e);ee("error","CLI error",{message:t,stack:e instanceof Error?e.stack:void 0}),console.error(`Error: ${t}`),process.exitCode=1});export{ko as suggestCommand,So as sortItems,To as run,mo as parseArgs};
|
|
722
|
+
--empty Remove items with empty content`)}function To(e){if(e==="add"){console.log("Usage: open-knowledge add <title> <content> [--url <url>] [-t <tag>] [--json]");return}if(e==="list"||e==="ls"){console.log("Usage: open-knowledge list|ls [--format table|json] [-p <page>] [-l <limit>] [-s <search>] [-t <tag>] [--sort created|title] [--desc] [--json]");return}if(e==="get"){console.log("Usage: open-knowledge get --id <id> [--json]");return}if(e==="update"||e==="edit"){console.log("Usage: open-knowledge update|edit --id <id> [--title <title>] [--content <content>] [--url <url>] [-t <tag>] [--json]");return}if(e==="archive"){console.log("Usage: open-knowledge archive --id <id> [--json]");return}if(e==="restore"||e==="unarchive"){console.log("Usage: open-knowledge restore|unarchive --id <id> [--json]");return}if(e==="upsert"){console.log("Usage: open-knowledge upsert [title] [content] [--id <id>] [--title <title>] [--content <content>] [--url <url>] [-t <tag>] [--json]");return}if(e==="untag"){console.log("Usage: open-knowledge untag --id <id> -t <tag> [--json]");return}if(e==="delete"||e==="rm"){console.log("Usage: open-knowledge delete|rm --id <id> -y [--json]");return}if(e==="export"){console.log("Usage: open-knowledge export [--format jsonl] [--json]");return}if(e==="prune"){console.log("Usage: open-knowledge prune --yes [--older-than <days>] [--empty] [--json]");return}if(e==="dedupe"){console.log("Usage: open-knowledge dedupe --yes [--json]");return}if(e==="stats"){console.log("Usage: open-knowledge stats [--json]");return}if(e==="paths"){console.log("Usage: open-knowledge paths [--scope local|global|project] [--json]");return}if(e==="setup"){console.log("Usage: open-knowledge setup --mode local|hosted [--api-url https://...] [--canonical-hasna-xyz] [--scope local|global|project] [--json]");return}if(e==="auth"){console.log("Usage: open-knowledge auth login|whoami|logout [--api-key <key>] [--email <email>] [--org <slug>] [--api-url https://...] [--scope local|global|project] [--json]");return}if(e==="remote"){console.log("Usage: open-knowledge remote contracts|status [--scope local|global|project] [--json]");return}if(e==="storage"){console.log("Usage: open-knowledge storage status|validate [--scope local|global|project] [--json]");return}if(e==="db"){console.log("Usage: open-knowledge db init|stats [--scope local|global|project] [--json]");return}if(e==="wiki"){console.log("Usage: open-knowledge wiki init|compile|file-answer|lint [query|prompt] [--title <title>] [--content <answer>] [--approve-write] [--limit <n>] [--scope local|global|project] [--json]");return}if(e==="source"){console.log("Usage: open-knowledge source resolve <source-ref> [--purpose knowledge_answer|knowledge_index] [--limit <n>] [--scope local|global|project] [--json]");return}if(e==="ingest"){console.log("Usage: open-knowledge ingest manifest <file|s3://bucket/key> | source <source-ref> [--purpose knowledge_index] [--scope local|global|project] [--json]");return}if(e==="reindex"){console.log("Usage: open-knowledge reindex status|enqueue|embeddings|outbox [file|s3://bucket/key] [--full] [--fake] [--scope local|global|project] [--json]");return}if(e==="search"){console.log("Usage: open-knowledge search <query> [--context] [--semantic] [--model openai:text-embedding-3-small] [--limit <n>] [--dimensions <n>] [--fake] [--scope local|global|project] [--json]");return}if(e==="web"){console.log("Usage: open-knowledge web search <query> [--provider openai|anthropic] [--model provider:model] [--domain <domain>] [--file-results] [--fake] [--scope local|global|project] [--json]");return}if(e==="ask"||e==="build"||e==="knowledge"){console.log("Usage: open-knowledge ask|build <prompt> [--generate] [--semantic] [--model default|provider:model] [--approve-write] [--scope local|global|project] [--json]");return}if(e==="embeddings"){console.log("Usage: open-knowledge embeddings status|index|search [query] [--model openai:text-embedding-3-small] [--limit <n>] [--dimensions <n>] [--fake] [--scope local|global|project] [--json]");return}if(e==="providers"){console.log("Usage: open-knowledge providers status|models|check [provider|model-alias] [--scope local|global|project] [--json]");return}if(e==="safety"){console.log("Usage: open-knowledge safety status|check|approve|audit|redact [args] [--scope local|global|project] [--json]");return}wo()}function Ro(e){if(e.noColor||process.env.NO_COLOR)return!1;if(process.env.FORCE_COLOR)return!0;return process.stdout.isTTY===!0}function y(e,t,n){if(t){console.log(JSON.stringify(e,null,2));return}if(typeof e==="string"){console.log(e);return}console.log(e.message??JSON.stringify(e,null,2))}function he(e){if(!e.id)throw Error("Missing required --id. Example: open-knowledge get --id <id>")}function So(e,t){let n=t.sort??"created";if(n!=="created"&&n!=="title")throw Error("Invalid --sort value. Use 'created' or 'title'.");let r=[...e].sort((i,s)=>{if(n==="title")return i.title.localeCompare(s.title);return i.created_at.localeCompare(s.created_at)});if(t.desc)r.reverse();return{sorted:r,sort:n,direction:t.desc?"desc":"asc"}}async function xo(e){let{positional:t,flags:n}=Eo(e);if(te("debug","CLI invoked",{command:t[0],flags:{json:n.json,store:n.store}}),n.version){console.log(n.json?JSON.stringify({name:pe.name,version:pe.version},null,2):`${pe.name} ${pe.version}`);return}if(n.completions){let a=n.completions;if(a==="bash")console.log('_open_knowledge() { local cur; cur="${COMP_WORDS[COMP_CWORD]}"; COMPREPLY=($(compgen -W "add list get update archive restore upsert untag delete export prune dedupe stats paths setup auth remote storage db wiki source ingest reindex search web ask build embeddings providers safety help ls rm edit unarchive knowledge --json --yes --help --version --desc --page --limit --search --sort --id --store --title --content --url --tag --format --completions --purpose --model --dimensions --semantic --context --generate --approve-write --provider --mode --api-url --canonical-hasna-xyz --api-key --email --org --org-id --user-id --domain --file-results --full --fake --no-color --scope --archived --include-archived" -- "$cur")); }; complete -F _open_knowledge open-knowledge');else if(a==="zsh")console.log(`#compdef open-knowledge
|
|
723
|
+
_open_knowledge() { _arguments -C "1: :(add list get update archive restore upsert untag delete export prune dedupe stats paths setup auth remote storage db wiki source ingest reindex search web ask build embeddings providers safety help ls rm edit unarchive knowledge)" "(--json)--json" "(--yes)-y" "(--help)--help" "(--version)--version" "(--desc)--desc" "(--archived)--archived" "(--include-archived)--include-archived" "(--semantic)--semantic" "(--context)--context" "(--generate)--generate" "(--approve-write)--approve-write" "(--canonical-hasna-xyz)--canonical-hasna-xyz" "(--file-results)--file-results" "(--full)--full" "(--fake)--fake" "(-p --page)"{-p,--page}"[page number]:number:" "(-l --limit)"{-l,--limit}"[items per page]:number:" "(-s --search)"{-s,--search}"[search text]:text:" "(--sort)--sort"{created,title}:" "(--id)--id[item id]:id:" "(--store)--store[store path]:path:" "(--title)--title[new title]:" "(--content)--content[new content]:" "(--url)--url[source url]:" "(-t --tag)"{-t,--tag}"[tag]:tag:" "(--format)--format[json|jsonl]:" "(--completions)--completions[output completions]:shell:(bash zsh fish):" "(--purpose)--purpose[purpose]:" "(--model)--model[model ref]:" "(--dimensions)--dimensions[embedding dimensions]:number:" "(--provider)--provider[provider]:" "(--mode)--mode"{local,hosted}:" "(--api-url)--api-url[hosted API URL]:" "(--api-key)--api-key[hosted API key]:" "(--email)--email[email]:" "(--org)--org[org slug]:" "(--org-id)--org-id[org id]:" "(--user-id)--user-id[user id]:" "(--domain)--domain[domain]:" "(--no-color)--no-color[disable color]" "(--scope)--scope"{local,global,project}:" }; _open_knowledge`);else if(a==="fish")console.log('complete -c open-knowledge -f; complete -c open-knowledge -a "add list get update archive restore upsert untag delete export prune dedupe stats paths setup auth remote storage db wiki source ingest reindex search web ask build embeddings providers safety help ls rm edit unarchive knowledge"; complete -c open-knowledge -l json; complete -c open-knowledge -l yes -s y; complete -c open-knowledge -l help -s h; complete -c open-knowledge -l version -s v; complete -c open-knowledge -l desc; complete -c open-knowledge -l archived; complete -c open-knowledge -l include-archived; complete -c open-knowledge -l semantic; complete -c open-knowledge -l context; complete -c open-knowledge -l generate; complete -c open-knowledge -l approve-write; complete -c open-knowledge -l canonical-hasna-xyz; complete -c open-knowledge -l provider; complete -c open-knowledge -l mode; complete -c open-knowledge -l api-url; complete -c open-knowledge -l api-key; complete -c open-knowledge -l email; complete -c open-knowledge -l org; complete -c open-knowledge -l org-id; complete -c open-knowledge -l user-id; complete -c open-knowledge -l domain; complete -c open-knowledge -l file-results; complete -c open-knowledge -l full; complete -c open-knowledge -l fake; complete -c open-knowledge -s p -l page; complete -c open-knowledge -s l -l limit; complete -c open-knowledge -s s -l search; complete -c open-knowledge -l sort; complete -c open-knowledge -l id; complete -c open-knowledge -l store; complete -c open-knowledge -l title; complete -c open-knowledge -l content; complete -c open-knowledge -l url; complete -c open-knowledge -s t -l tag; complete -c open-knowledge -l format; complete -c open-knowledge -l completions; complete -c open-knowledge -l purpose; complete -c open-knowledge -l model; complete -c open-knowledge -l dimensions; complete -c open-knowledge -l no-color; complete -c open-knowledge -l scope -a "local global project"');else throw Error("Invalid --completions value. Use 'bash', 'zsh', or 'fish'.");return}let r=yo(t[0]),i=1;if(vo()&&r&&!tr.includes(r))r="ask",i=0;if(!r||n.help||r==="help"){To(t[1]);return}let s=Zn({scope:n.scope}),c=n.store;if(!c)if(n.scope==="project"||n.scope==="local")c=s.jsonStorePath();else c=Ye();if(r==="paths"){y(s.paths(),n.json);return}if(r==="setup"){let a=s.setup({mode:n.mode,apiUrl:n.apiUrl,canonicalHasnaXyz:n.canonicalHasnaXyz});y(a,n.json);return}if(r==="auth"){let a=t[1]??"whoami";if(a==="whoami"||a==="status"){let u=s.authStatus(process.env);y({ok:!0,...u,message:u.authenticated?`Authenticated via ${u.source}`:"Not authenticated"},n.json);return}if(a==="login"){let u=n.apiKey??process.env.KNOWLEDGE_API_KEY??process.env.HASNA_KNOWLEDGE_API_KEY;if(!u)throw Error("Usage: open-knowledge auth login --api-key <key> [--email <email>]");let o=s.saveAuth({apiKey:u,email:n.email,orgSlug:n.org,orgId:n.orgId,userId:n.userId,apiUrl:n.apiUrl},process.env);y({ok:!0,authenticated:!0,email:o.email??null,org_slug:o.org_slug??null,api_url:o.api_url??s.authStatus(process.env).api_url,auth_path:s.authStatus(process.env).auth_path,message:`Saved hosted credentials for ${o.email??"API key"}`},n.json);return}if(a==="logout"){let u=s.clearAuth(process.env);y({ok:!0,removed:u,message:u?"Removed hosted credentials":"No hosted credentials found"},n.json);return}throw Error("Invalid auth action. Use 'login', 'whoami', or 'logout'.")}if(r==="remote"){let a=t[1]??"status";if(a==="contracts"||a==="contract"){let u=s.authStatus(process.env);y({ok:!0,authenticated:u.authenticated,api_url:u.api_url,contract:s.remoteContract(),message:`Remote contract v${s.remoteContract().contract_version}`},n.json);return}if(a==="status"){let u=s.authStatus(process.env),o=s.remoteContract();y({ok:!0,mode:s.config().mode,authenticated:u.authenticated,auth_source:u.source,api_url:u.api_url,client_ready:Boolean(s.remoteClient(process.env)),contract_version:o.contract_version,capabilities:o.capabilities,message:u.authenticated?`Remote client ready for ${u.api_url}`:"Remote client not authenticated"},n.json);return}throw Error("Invalid remote action. Use 'contracts' or 'status'.")}if(r==="storage"){let a=t[1]??"status";if(a==="status"){let u=s.storageContract(),o=s.validateStorage();y({ok:o.ok,...u,validation:o,message:`${u.storage_type} artifact storage at ${u.artifact_store.uri_prefix}`},n.json);return}if(a==="validate"){let u=s.validateStorage();y({ok:u.ok,validation:u,message:u.ok?"Storage contract valid":`Storage contract invalid: ${u.errors.join("; ")}`},n.json);return}throw Error("Invalid storage action. Use 'status' or 'validate'.")}if(r==="db"){let a=t[1]??"init";if(a!=="init"&&a!=="stats")throw Error("Invalid db action. Use 'init' or 'stats'.");if(a==="init"){let o=s.initDb();y({ok:!0,...o,message:`Initialized ${o.path}`},n.json);return}let u=s.dbStats();y({ok:!0,path:s.workspace.knowledgeDbPath,...u,message:`knowledge.db schema v${u.schema_version}`},n.json);return}if(r==="wiki"){let a=t[1]??"init";if(a==="init"){let u=await s.initWiki();y({ok:!0,...u,message:`Initialized wiki layout in ${s.workspace.home}`},n.json);return}if(a==="compile"){let u=t.slice(2),o=u.filter((f)=>/^(open-files|file|s3|https?):\/\//.test(f)),_=u.filter((f)=>!/^(open-files|file|s3|https?):\/\//.test(f)).join(" "),l=await s.compileWiki({title:n.title,query:_||n.search,sourceRefs:o.length>0?o:void 0,limit:n.limit});y({ok:!0,...l,message:`Compiled wiki page ${l.path}`},n.json);return}if(a==="file-answer"||a==="answer"){let u=t.slice(2).join(" ");if(!u)throw Error("Usage: open-knowledge wiki file-answer <prompt> --content <answer> --approve-write");if(!n.content)throw Error("Missing --content <answer> for wiki file-answer.");let o=await s.fileAnswer({prompt:u,answer:n.content,approveWrite:n.approveWrite,limit:n.limit,semantic:n.semantic,modelRef:n.model,dimensions:n.dimensions,fake:n.fake});y({ok:!0,...o},n.json);return}if(a==="lint"){let u=s.lintWiki();y({ok:u.ok,...u,message:u.ok?"Wiki lint passed":`Wiki lint found ${u.issue_count} issue(s)`},n.json);return}throw Error("Invalid wiki action. Use 'init', 'compile', 'file-answer', or 'lint'.")}if(r==="safety"){let a=t[1]??"status",u=s.ensureWorkspace(),o=s.safetyPolicy();s.initDb();let _=b(u.knowledgeDbPath);try{if(a==="status"){y({ok:!0,mode:o.mode,workspace:u.home,allow_write_roots:o.allowWriteRoots,read_only_source_access:o.readOnlySourceAccess,network:o.network,redaction:o.redaction,approvals:o.approvals,message:`Safety policy: ${o.mode}`},n.json);return}if(a==="check"){let l=t[2]??"generated_write",f=t[3]??null,m;try{if(l==="web_search")ie(o),m={action:l,target_uri:f,approval_required:!1,approved:!0,decision:"allow"};else if(l==="s3_read"){if(!f)throw Error("safety check s3_read requires an s3:// target.");J(f,o),m={action:l,target_uri:f,approval_required:!1,approved:!0,decision:"allow"}}else m=kn(_,o,l,f);I(_,{event_type:"safety_check",action:l,target_uri:f,decision:m.decision==="allow"?"allow":"requires_approval",metadata:m}),y({ok:!0,...m,message:`Safety check ${m.decision}`},n.json);return}catch(E){throw I(_,{event_type:"safety_check",action:l,target_uri:f,decision:"deny",metadata:{error:E instanceof Error?E.message:String(E)}}),E}}if(a==="approve"){let l=t[2]??"generated_write",f=t[3]??null,m=yn(_,{action:l,target_uri:f,reason:"local-cli approval",metadata:{scope:n.scope??"global"}});I(_,{event_type:"approval",action:l,target_uri:f,decision:"allow",metadata:{approval_id:m.id}}),y({ok:!0,...m,action:l,target_uri:f,message:`Approved ${l}`},n.json);return}if(a==="audit"){let l=_.query("SELECT id, event_type, action, target_uri, decision, metadata_json, created_at FROM audit_events ORDER BY created_at DESC LIMIT 50").all().map((f)=>({id:f.id,event_type:f.event_type,action:f.action,target_uri:f.target_uri,decision:f.decision,metadata:JSON.parse(f.metadata_json),created_at:f.created_at}));y({ok:!0,events:l,message:`${l.length} audit event(s)`},n.json);return}if(a==="redact"){let l=t.slice(2).join(" ");if(!l)throw Error("Usage: open-knowledge safety redact <text>");let f=De(l,o);if(f.findings.length>0)Ue(_,{source_uri:"safety://redact",findings:f.findings,metadata:{command:"safety redact"}});I(_,{event_type:"redaction",action:"safety_redact",target_uri:"safety://redact",decision:f.findings.length>0?"redacted":"allow",metadata:{findings:f.findings.length}}),y({ok:!0,text:f.text,findings:f.findings,message:`Redacted ${f.findings.length} finding(s)`},n.json);return}throw Error("Invalid safety action. Use 'status', 'check', 'approve', 'audit', or 'redact'.")}finally{_.close()}}if(r==="source"){if((t[1]??"")!=="resolve")throw Error("Invalid source action. Use 'resolve'.");let u=t[2];if(!u)throw Error("Usage: open-knowledge source resolve <source-ref>");let o=await s.resolveSource(u,{purpose:n.purpose,limit:n.limit});y({ok:!0,...o,message:o.resolved?`Resolved ${o.source_ref} (${o.content.chunks_returned}/${o.content.chunks_total} chunks)`:`Source not indexed: ${u}`},n.json);return}if(r==="ingest"){let a=t[1]??"";if(a==="manifest"){let u=t[2];if(!u)throw Error("Usage: open-knowledge ingest manifest <file|s3://bucket/key>");let o=await s.ingestManifest(u);y({ok:!0,...o,message:`Ingested ${o.items_seen} manifest item(s)`},n.json);return}if(a==="source"){let u=t[2];if(!u)throw Error("Usage: open-knowledge ingest source <source-ref>");let o=await s.ingestSource(u,n.purpose);y({ok:!0,...o,message:`Ingested source ${o.source_ref} (${o.chunks_inserted} chunks)`},n.json);return}throw Error("Invalid ingest action. Use 'manifest' or 'source'.")}if(r==="reindex"){let a=t[1]??"status";if(a==="status"){let u=s.reindexHealth({modelRef:n.model,dimensions:n.dimensions,fake:n.fake});y({ok:!0,...u,message:`${u.missing_embeddings} chunk(s) missing embeddings`},n.json);return}if(a==="enqueue"){let u=s.enqueueReindex({modelRef:n.model,dimensions:n.dimensions,fake:n.fake});y({ok:!0,...u,message:`Queued ${u.enqueued} embedding refresh item(s)`},n.json);return}if(a==="embeddings"){let u=await s.refreshEmbeddings({full:n.full,limit:n.limit,modelRef:n.model,dimensions:n.dimensions,fake:n.fake});y({ok:!0,...u,message:`Embedded ${u.indexed.chunks_embedded} chunk(s)`},n.json);return}if(a==="outbox"){let u=t[2];if(!u)throw Error("Usage: open-knowledge reindex outbox <file|s3://bucket/key>");let o=await s.consumeOutbox(u);y({ok:!0,...o,message:`Consumed ${o.events_seen} outbox event(s)`},n.json);return}throw Error("Invalid reindex action. Use 'status', 'enqueue', 'embeddings', or 'outbox'.")}if(r==="embeddings"){let a=t[1]??"status";if(a==="status"){let u=s.embeddingStatus();y({ok:!0,...u,message:`${u.total_vector_entries} vector index entries`},n.json);return}if(a==="index"){let u=await s.indexEmbeddings({limit:n.limit,modelRef:n.model,dimensions:n.dimensions,fake:n.fake});y({ok:!0,...u,message:`Embedded ${u.chunks_embedded} chunk(s)`},n.json);return}if(a==="search"){let u=t.slice(2).join(" ");if(!u)throw Error("Usage: open-knowledge embeddings search <query>");let o=await s.semanticSearch({query:u,limit:n.limit,modelRef:n.model,dimensions:n.dimensions,fake:n.fake});y({ok:!0,...o,message:`${o.results.length} semantic result(s)`},n.json);return}throw Error("Invalid embeddings action. Use 'status', 'index', or 'search'.")}if(r==="search"){let a=t.slice(1).join(" ");if(!a)throw Error("Usage: open-knowledge search <query>");if(n.context){let o=await s.retrieveContext({query:a,limit:n.limit,semantic:n.semantic,modelRef:n.model,dimensions:n.dimensions,fake:n.fake});y({ok:!0,...o,message:`${o.excerpts.length} context excerpt(s)`},n.json);return}let u=await s.search({query:a,limit:n.limit,semantic:n.semantic,modelRef:n.model,dimensions:n.dimensions,fake:n.fake});y({ok:!0,...u,message:`${u.results.length} search result(s)`},n.json);return}if(r==="web"){if((t[1]??"search")!=="search")throw Error("Invalid web action. Use 'search'.");let u=t.slice(2).join(" ");if(!u)throw Error("Usage: open-knowledge web search <query>");let o=await s.webSearch({query:u,limit:n.limit,modelRef:n.model,provider:n.provider,domains:n.domain,fake:n.fake,fileResults:n.fileResults});y({ok:!0,...o,message:`${o.sources.length} web source(s)`},n.json);return}if(r==="ask"||r==="build"){let a=t.slice(i).join(" ");if(!a)throw Error("Usage: open-knowledge ask <prompt>");let u=await s.runPrompt({prompt:a,limit:n.limit,semantic:n.semantic,modelRef:n.model,dimensions:n.dimensions,fake:n.fake,generate:n.generate,approveWrite:n.approveWrite});y({ok:!0,...u,message:u.generated?"Generated answer with citations":"Prepared citation context draft"},n.json);return}if(r==="providers"){let a=t[1]??"status";if(a==="status"){let u=s.providerStatus(),o=u.providers.filter((_)=>_.configured).length;y({ok:!0,...u,message:`${o}/${u.providers.length} provider credential(s) configured`},n.json);return}if(a==="models"){let u=s.modelRegistry();y({ok:!0,models:u,message:`${u.length} model alias(es)`},n.json);return}if(a==="check"){let u=t[2]??"default",o=H(u,s.config()),_=C(o),l=z(_.provider,s.config());y({ok:!0,target:u,model_ref:o,provider:_.provider,model:_.model,credential:l,message:`${_.provider} credentials configured`},n.json);return}throw Error("Invalid providers action. Use 'status', 'models', or 'check'.")}if(Ve(c),r==="add"){let a=t[1],u=t[2];if(!a||!u)throw Error("Usage: open-knowledge add <title> <content>");P(c,()=>{let o=U(c),_={id:Qe(),title:a,content:u,url:n.url??null,tags:n.tag?[n.tag]:[],created_at:new Date().toISOString(),updated_at:new Date().toISOString()};o.items.push(_),$(c,o),te("info","Item added",{id:_.id,title:_.title}),y({ok:!0,item:_,message:`Added ${_.id}`},n.json)});return}if(r==="list"){if(n.format!==void 0&&n.format!=="table"&&n.format!=="json")throw Error("Invalid --format value for list. Use 'table' or 'json'.");P(c,()=>{let a=U(c),u=Number.isFinite(n.page)&&n.page>0?n.page:1,o=Number.isFinite(n.limit)&&n.limit>0?n.limit:20,_=n.search?String(n.search).toLowerCase():"",l=n.tag?String(n.tag).toLowerCase():"",f=n.format==="table"||!n.json&&!n.format&&Ro(n),m=n.json||n.format==="json",E=a.items;if(n.archived)E=E.filter((w)=>w.archived===!0);else if(!n.includeArchived)E=E.filter((w)=>!w.archived);if(_)E=E.filter((w)=>w.title.toLowerCase().includes(_)||w.content.toLowerCase().includes(_));if(l)E=E.filter((w)=>w.tags&&w.tags.map((Be)=>Be.toLowerCase()).includes(l));let{sorted:h,sort:T,direction:k}=So(E,n),p=(u-1)*o,R=h.slice(p,p+o),v=Math.max(1,Math.ceil(h.length/o));if(m){y({ok:!0,page:u,limit:o,total:h.length,total_pages:v,sort:T,direction:k,items:R},!0);return}if(R.length===0){y(`No items found (search=${_||"none"}, tag=${l||"none"})`,!1);return}if(f){let w=(W)=>W,Be=`${w("ID")} ${w("TITLE")} ${w("CREATED")} ${w("URL")} ${w("TAGS")}`;console.log(Be);for(let W of R)console.log(`${W.id} ${w(W.title)} ${W.created_at} ${W.url?w(W.url):""} ${W.tags?.length?w(`[${W.tags.join(", ")}]`):""}`);console.log(`Page ${u}/${v} | showing ${R.length} of ${h.length} | sort=${T} ${k} | search=${_||"none"} | tag=${l||"none"}`)}else{for(let w of R)console.log(`${w.id} ${w.title} ${w.created_at}${w.url?` ${w.url}`:""}${w.tags?.length?` [${w.tags.join(", ")}]`:""}`);console.log(`Page ${u}/${v} | showing ${R.length} of ${h.length} | sort=${T} ${k} | search=${_||"none"} | tag=${l||"none"}`)}});return}if(r==="get"){he(n),P(c,()=>{let u=U(c).items.find((o)=>o.id===n.id||o.short_id===n.id);if(!u)throw Error(`Item not found: ${n.id}`);y({ok:!0,item:u,message:`${u.id}: ${u.title}`},n.json)});return}if(r==="update"){he(n),P(c,()=>{let a=U(c),u=a.items.findIndex((_)=>_.id===n.id||_.short_id===n.id);if(u===-1)throw Error(`Item not found: ${n.id}`);let o=a.items[u];if(n.title!==void 0)o.title=n.title;if(n.content!==void 0)o.content=n.content;if(n.url!==void 0)o.url=n.url;if(n.tag!==void 0){if(o.tags=o.tags||[],!o.tags.map((_)=>_.toLowerCase()).includes(n.tag.toLowerCase()))o.tags.push(n.tag)}o.updated_at=new Date().toISOString(),a.items[u]=o,$(c,a),y({ok:!0,item:o,message:`Updated ${o.id}`},n.json)});return}if(r==="archive"||r==="restore"){he(n),P(c,()=>{let a=U(c),u=a.items.findIndex((_)=>_.id===n.id||_.short_id===n.id);if(u===-1)throw Error(`Item not found: ${n.id}`);let o=a.items[u];o.archived=r==="archive",o.updated_at=new Date().toISOString(),a.items[u]=o,$(c,a),y({ok:!0,item:o,message:`${r==="archive"?"Archived":"Restored"} ${o.id}`},n.json)});return}if(r==="untag"){if(he(n),!n.tag)throw Error("Missing required --tag. Example: open-knowledge untag --id <id> -t <tag>");P(c,()=>{let a=U(c),u=a.items.findIndex((l)=>l.id===n.id||l.short_id===n.id);if(u===-1)throw Error(`Item not found: ${n.id}`);let o=a.items[u],_=o.tags?.length??0;o.tags=(o.tags??[]).filter((l)=>l.toLowerCase()!==n.tag.toLowerCase()),o.updated_at=new Date().toISOString(),a.items[u]=o,$(c,a),y({ok:!0,item:o,removed:_-o.tags.length,message:`Removed tag from ${o.id}`},n.json)});return}if(r==="upsert"){let a=n.title??t[1],u=n.content??t[2];P(c,()=>{let o=U(c),_=n.id?o.items.findIndex((m)=>m.id===n.id||m.short_id===n.id):-1,l=new Date().toISOString();if(_===-1){if(!a||!u)throw Error("New item requires title and content. Example: open-knowledge upsert <title> <content> [--id <id>]");let m=n.id??Qe(),E={id:m,short_id:At(m),title:a,content:u,url:n.url??null,tags:n.tag?[n.tag]:[],metadata:{},archived:!1,created_at:l,updated_at:l};o.items.push(E),$(c,o),y({ok:!0,created:!0,item:E,message:`Upserted ${E.id}`},n.json);return}let f=o.items[_];if(a!==void 0)f.title=a;if(u!==void 0)f.content=u;if(n.url!==void 0)f.url=n.url;if(n.tag!==void 0){if(f.tags=f.tags||[],!f.tags.map((m)=>m.toLowerCase()).includes(n.tag.toLowerCase()))f.tags.push(n.tag)}f.updated_at=l,o.items[_]=f,$(c,o),y({ok:!0,created:!1,item:f,message:`Upserted ${f.id}`},n.json)});return}if(r==="delete"){if(he(n),!n.yes)throw Error("Refusing delete without --yes. Re-run with: open-knowledge delete --id <id> --yes");P(c,()=>{let a=U(c),u=a.items.length;a.items=a.items.filter((_)=>_.id!==n.id&&_.short_id!==n.id);let o=u!==a.items.length;if($(c,a),!o)throw Error(`Item not found: ${n.id}`);te("info","Item deleted",{id:n.id}),y({ok:!0,deleted_id:n.id,message:`Deleted ${n.id}`},n.json)});return}if(r==="export"){let a=n.format??"json";if(a!=="json"&&a!=="jsonl")throw Error("Invalid --format. Use 'json' or 'jsonl'.");P(c,()=>{let u=U(c);if(a==="jsonl")for(let o of u.items)console.log(JSON.stringify(o));else y({ok:!0,items:u.items},n.json)});return}if(r==="prune"){if(!n.yes)throw Error("Refusing prune without --yes. Re-run with: open-knowledge prune --yes [--older-than <days>] [--empty]");P(c,()=>{let a=U(c),u=a.items.length;if(n.olderThan!==void 0){let _=new Date;_.setDate(_.getDate()-n.olderThan),a.items=a.items.filter((l)=>new Date(l.created_at)>=_)}if(n.empty)a.items=a.items.filter((_)=>_.content.trim().length>0);let o=u-a.items.length;$(c,a),te("info","Prune completed",{pruned:o,remaining:a.items.length}),y({ok:!0,pruned:o,remaining:a.items.length,message:`Pruned ${o} item(s)`},n.json)});return}if(r==="dedupe"){if(!n.yes)throw Error("Refusing dedupe without --yes. Re-run with: open-knowledge dedupe --yes [--json]");P(c,()=>{let a=U(c),u=new Set,o=a.items.length;a.items=a.items.filter((l)=>{let f=`${l.title}\x00${l.content}`;if(u.has(f))return!1;return u.add(f),!0});let _=o-a.items.length;$(c,a),te("info","Dedupe completed",{removed:_,remaining:a.items.length}),y({ok:!0,removed:_,remaining:a.items.length,message:`Dedupe removed ${_} duplicate(s)`},n.json)});return}if(r==="stats"){P(c,()=>{let a=U(c),u=a.items.filter((k)=>!k.archived),o=u.length,_=a.items.length-o,l=u.filter((k)=>k.url).length,f=u.filter((k)=>k.tags&&k.tags.length>0).length,m=o>0?u.map((k)=>k.created_at).sort()[0]:null,E=o>0?u.map((k)=>k.created_at).sort()[o-1]:null,h={};for(let k of u)for(let p of k.tags||[])h[p]=(h[p]||0)+1;let T=Object.entries(h).sort((k,p)=>p[1]-k[1]).slice(0,5).map(([k,p])=>({tag:k,count:p}));y({ok:!0,total:o,archived:_,with_url:l,with_tags:f,oldest:m,newest:E,top_tags:T,message:`${o} items | ${l} with URL | ${f} with tags`},n.json)});return}let d=bo(t[0]),g=d?` Did you mean '${d}'?`:"";throw te("warn","Unknown command",{input:t[0],suggestion:d}),Error(`Unknown command: ${t[0]}.${g} Run 'open-knowledge --help' for available commands.`)}if(import.meta.main)xo(process.argv.slice(2)).catch((e)=>{let t=e instanceof Error?e.message:String(e);te("error","CLI error",{message:t,stack:e instanceof Error?e.stack:void 0}),console.error(`Error: ${t}`),process.exitCode=1});export{bo as suggestCommand,So as sortItems,xo as run,Eo as parseArgs};
|