veryfront 0.1.83 → 0.1.85

package/esm/cli/commands/knowledge/parser-source.js

@@ -14,6 +14,53 @@ def yaml_quote(value: Any) -> str:
 
 
 CODE_FENCE = chr(96) * 3
+TEXT_FILE_EXTENSIONS = {
+    ".c",
+    ".cc",
+    ".conf",
+    ".cpp",
+    ".cs",
+    ".css",
+    ".go",
+    ".h",
+    ".hpp",
+    ".ini",
+    ".java",
+    ".js",
+    ".jsonl",
+    ".jsx",
+    ".kt",
+    ".less",
+    ".lua",
+    ".mjs",
+    ".ndjson",
+    ".php",
+    ".pl",
+    ".py",
+    ".r",
+    ".rb",
+    ".rs",
+    ".sass",
+    ".scala",
+    ".scss",
+    ".sh",
+    ".sql",
+    ".swift",
+    ".toml",
+    ".ts",
+    ".tsx",
+    ".xml",
+    ".yaml",
+    ".yml",
+    ".zsh",
+}
+TEXT_FILE_NAMES = {
+    "dockerfile",
+    "makefile",
+    "readme",
+    "license",
+    "changelog",
+}
 
 
 def slugify(value: str) -> str:
@@ -406,6 +453,7 @@ def parse_json(path: str):
 
 def select_parser(path: Path):
     ext = path.suffix.lower()
+    name = path.name.lower()
     if ext == ".pdf":
         return "pdf", prefer_kreuzberg("pdf", parse_pdf)
     if ext in {".csv", ".tsv"}:
@@ -424,6 +472,10 @@ def select_parser(path: Path):
         return ext.lstrip("."), parse_text
     if ext == ".json":
         return "json", parse_json
+    if ext in TEXT_FILE_EXTENSIONS:
+        return ext.lstrip("."), parse_text
+    if not ext and name in TEXT_FILE_NAMES:
+        return "text", parse_text
     raise ValueError(f"Unsupported file type: {ext}")
 
 

package/esm/cli/commands/knowledge/result.d.ts

@@ -0,0 +1,54 @@
+export type KnowledgeIngestSkipReason = "hidden_path" | "ignored_directory" | "unsupported_file_type";
+export type KnowledgeIngestFailureReason = "parser_error" | "upload_error";
+export interface KnowledgeIngestFileResult {
+    source: string;
+    localSourcePath: string;
+    outputPath: string;
+    remotePath: string;
+    slug: string;
+    sourceType: string;
+    summary: string;
+    stats: Record<string, unknown>;
+    warnings: string[];
+}
+export interface KnowledgeIngestSkippedFileResult {
+    source: string;
+    localSourcePath: string | null;
+    reason: KnowledgeIngestSkipReason;
+    message: string;
+}
+export interface KnowledgeIngestFailedFileResult {
+    source: string;
+    localSourcePath: string;
+    reason: KnowledgeIngestFailureReason;
+    message: string;
+}
+export interface KnowledgeIngestSummary {
+    requested_count: number;
+    ingested_count: number;
+    skipped_count: number;
+    failed_count: number;
+}
+export interface KnowledgeIngestResultMetadata {
+    requested_count: number;
+    source_mode: "explicit_sources" | "path_prefix";
+    knowledge_path: string;
+}
+export interface KnowledgeIngestJobResult {
+    kind: "knowledge_ingest";
+    version: 1;
+    metadata: KnowledgeIngestResultMetadata;
+    summary: KnowledgeIngestSummary;
+    ingested: KnowledgeIngestFileResult[];
+    skipped: KnowledgeIngestSkippedFileResult[];
+    failed: KnowledgeIngestFailedFileResult[];
+}
+export declare function buildKnowledgeIngestJobResult(input: {
+    requestedCount: number;
+    sourceMode: KnowledgeIngestResultMetadata["source_mode"];
+    knowledgePath: string;
+    ingested: KnowledgeIngestFileResult[];
+    skipped?: KnowledgeIngestSkippedFileResult[];
+    failed?: KnowledgeIngestFailedFileResult[];
+}): KnowledgeIngestJobResult;
+//# sourceMappingURL=result.d.ts.map

package/esm/cli/commands/knowledge/result.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"result.d.ts","sourceRoot":"","sources":["../../../../src/cli/commands/knowledge/result.ts"],"names":[],"mappings":"AAAA,MAAM,MAAM,yBAAyB,GACjC,aAAa,GACb,mBAAmB,GACnB,uBAAuB,CAAC;AAE5B,MAAM,MAAM,4BAA4B,GACpC,cAAc,GACd,cAAc,CAAC;AAEnB,MAAM,WAAW,yBAAyB;IACxC,MAAM,EAAE,MAAM,CAAC;IACf,eAAe,EAAE,MAAM,CAAC;IACxB,UAAU,EAAE,MAAM,CAAC;IACnB,UAAU,EAAE,MAAM,CAAC;IACnB,IAAI,EAAE,MAAM,CAAC;IACb,UAAU,EAAE,MAAM,CAAC;IACnB,OAAO,EAAE,MAAM,CAAC;IAChB,KAAK,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IAC/B,QAAQ,EAAE,MAAM,EAAE,CAAC;CACpB;AAED,MAAM,WAAW,gCAAgC;IAC/C,MAAM,EAAE,MAAM,CAAC;IACf,eAAe,EAAE,MAAM,GAAG,IAAI,CAAC;IAC/B,MAAM,EAAE,yBAAyB,CAAC;IAClC,OAAO,EAAE,MAAM,CAAC;CACjB;AAED,MAAM,WAAW,+BAA+B;IAC9C,MAAM,EAAE,MAAM,CAAC;IACf,eAAe,EAAE,MAAM,CAAC;IACxB,MAAM,EAAE,4BAA4B,CAAC;IACrC,OAAO,EAAE,MAAM,CAAC;CACjB;AAED,MAAM,WAAW,sBAAsB;IACrC,eAAe,EAAE,MAAM,CAAC;IACxB,cAAc,EAAE,MAAM,CAAC;IACvB,aAAa,EAAE,MAAM,CAAC;IACtB,YAAY,EAAE,MAAM,CAAC;CACtB;AAED,MAAM,WAAW,6BAA6B;IAC5C,eAAe,EAAE,MAAM,CAAC;IACxB,WAAW,EAAE,kBAAkB,GAAG,aAAa,CAAC;IAChD,cAAc,EAAE,MAAM,CAAC;CACxB;AAED,MAAM,WAAW,wBAAwB;IACvC,IAAI,EAAE,kBAAkB,CAAC;IACzB,OAAO,EAAE,CAAC,CAAC;IACX,QAAQ,EAAE,6BAA6B,CAAC;IACxC,OAAO,EAAE,sBAAsB,CAAC;IAChC,QAAQ,EAAE,yBAAyB,EAAE,CAAC;IACtC,OAAO,EAAE,gCAAgC,EAAE,CAAC;IAC5C,MAAM,EAAE,+BAA+B,EAAE,CAAC;CAC3C;AAED,wBAAgB,6BAA6B,CAAC,KAAK,EAAE;IACnD,cAAc,EAAE,MAAM,CAAC;IACvB,UAAU,EAAE,6BAA6B,CAAC,aAAa,CAAC,CAAC;IACzD,aAAa,EAAE,MAAM,CAAC;IACtB,QAAQ,EAAE,yBAAyB,EAAE,CAAC;IACtC,OAAO,CAAC,EAAE,gCAAgC,EAAE,CAAC;IAC7C,MAAM,CAAC,EAAE,+BAA+B,EAAE,CAAC;CAC5C,GAAG,wBAAwB,CAsB3B"}

package/esm/cli/commands/knowledge/result.js

@@ -0,0 +1,22 @@
+export function buildKnowledgeIngestJobResult(input) {
+    const skipped = input.skipped ?? [];
+    const failed = input.failed ?? [];
+    return {
+        kind: "knowledge_ingest",
+        version: 1,
+        metadata: {
+            requested_count: input.requestedCount,
+            source_mode: input.sourceMode,
+            knowledge_path: input.knowledgePath,
+        },
+        summary: {
+            requested_count: input.requestedCount,
+            ingested_count: input.ingested.length,
+            skipped_count: skipped.length,
+            failed_count: failed.length,
+        },
+        ingested: input.ingested,
+        skipped,
+        failed,
+    };
+}

package/esm/cli/commands/knowledge/source-policy.d.ts

@@ -0,0 +1,11 @@
+import type { KnowledgeIngestSkipReason } from "./result.js";
+export type KnowledgeIngestSourceDecision = {
+    kind: "ingest";
+} | {
+    kind: "skip";
+    reason: KnowledgeIngestSkipReason;
+    message: string;
+};
+export declare function classifyKnowledgeDirectoryPath(path: string): KnowledgeIngestSourceDecision;
+export declare function classifyKnowledgeSourcePath(path: string): KnowledgeIngestSourceDecision;
+//# sourceMappingURL=source-policy.d.ts.map

package/esm/cli/commands/knowledge/source-policy.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"source-policy.d.ts","sourceRoot":"","sources":["../../../../src/cli/commands/knowledge/source-policy.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,yBAAyB,EAAE,MAAM,aAAa,CAAC;AAuE7D,MAAM,MAAM,6BAA6B,GACrC;IAAE,IAAI,EAAE,QAAQ,CAAA;CAAE,GAClB;IACA,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,EAAE,yBAAyB,CAAC;IAClC,OAAO,EAAE,MAAM,CAAC;CACjB,CAAC;AAiEJ,wBAAgB,8BAA8B,CAAC,IAAI,EAAE,MAAM,GAAG,6BAA6B,CAE1F;AAED,wBAAgB,2BAA2B,CAAC,IAAI,EAAE,MAAM,GAAG,6BAA6B,CAgBvF"}

package/esm/cli/commands/knowledge/source-policy.js

@@ -0,0 +1,135 @@
+import { basename, extname, normalize } from "../../../src/platform/compat/path/index.js";
+const RICH_FILE_EXTENSIONS = new Set([
+    ".pdf",
+    ".csv",
+    ".tsv",
+    ".docx",
+    ".xlsx",
+    ".xls",
+    ".pptx",
+    ".html",
+    ".htm",
+    ".txt",
+    ".json",
+    ".md",
+    ".mdx",
+]);
+const TEXT_FALLBACK_EXTENSIONS = new Set([
+    ".c",
+    ".cc",
+    ".conf",
+    ".cpp",
+    ".cs",
+    ".css",
+    ".go",
+    ".h",
+    ".hpp",
+    ".ini",
+    ".java",
+    ".js",
+    ".jsonl",
+    ".jsx",
+    ".kt",
+    ".less",
+    ".lua",
+    ".mjs",
+    ".ndjson",
+    ".php",
+    ".pl",
+    ".py",
+    ".r",
+    ".rb",
+    ".rs",
+    ".sass",
+    ".scala",
+    ".scss",
+    ".sh",
+    ".sql",
+    ".swift",
+    ".toml",
+    ".tsx",
+    ".ts",
+    ".xml",
+    ".yaml",
+    ".yml",
+    ".zsh",
+]);
+const TEXT_FALLBACK_FILENAMES = new Set([
+    "dockerfile",
+    "makefile",
+    "readme",
+    "license",
+    "changelog",
+]);
+const IGNORED_DIRECTORY_NAMES = new Set([
+    "node_modules",
+]);
+function normalizePathSegments(value) {
+    return normalize(value)
+        .replaceAll("\\", "/")
+        .split("/")
+        .map((segment) => segment.trim())
+        .filter((segment) => segment.length > 0 && segment !== "." && segment !== "..");
+}
+function findHiddenSegment(segments) {
+    for (const segment of segments) {
+        if (segment.startsWith(".")) {
+            return segment;
+        }
+    }
+    return null;
+}
+function findIgnoredDirectory(segments) {
+    for (const segment of segments) {
+        if (IGNORED_DIRECTORY_NAMES.has(segment.toLowerCase())) {
+            return segment;
+        }
+    }
+    return null;
+}
+function isSupportedKnowledgeFile(path) {
+    const extension = extname(path).toLowerCase();
+    if (extension && (RICH_FILE_EXTENSIONS.has(extension) || TEXT_FALLBACK_EXTENSIONS.has(extension))) {
+        return true;
+    }
+    const fileName = basename(path).toLowerCase();
+    return TEXT_FALLBACK_FILENAMES.has(fileName);
+}
+function classifyCommonKnowledgePath(path) {
+    const segments = normalizePathSegments(path);
+    const hiddenSegment = findHiddenSegment(segments);
+    if (hiddenSegment != null) {
+        return {
+            kind: "skip",
+            reason: "hidden_path",
+            message: `Hidden file or directory skipped: ${hiddenSegment}`,
+        };
+    }
+    const ignoredDirectory = findIgnoredDirectory(segments);
+    if (ignoredDirectory != null) {
+        return {
+            kind: "skip",
+            reason: "ignored_directory",
+            message: `Ignored directory skipped: ${ignoredDirectory}`,
+        };
+    }
+    return null;
+}
+export function classifyKnowledgeDirectoryPath(path) {
+    return classifyCommonKnowledgePath(path) ?? { kind: "ingest" };
+}
+export function classifyKnowledgeSourcePath(path) {
+    const commonDecision = classifyCommonKnowledgePath(path);
+    if (commonDecision != null) {
+        return commonDecision;
+    }
+    if (!isSupportedKnowledgeFile(path)) {
+        const extension = extname(path).toLowerCase();
+        return {
+            kind: "skip",
+            reason: "unsupported_file_type",
+            message: `Unsupported file type: ${extension || basename(path)}`,
+        };
+    }
+    return { kind: "ingest" };
+}

package/esm/deno.js

@@ -1,6 +1,6 @@
 export default {
     "name": "veryfront",
-    "version": "0.1.83",
+    "version": "0.1.85",
     "license": "Apache-2.0",
     "nodeModulesDir": "auto",
     "exclude": [

package/esm/src/jobs/index.d.ts

@@ -30,5 +30,5 @@
  */
 import "../../_dnt.polyfills.js";
 export { type CreateCronJobInput, type CreateJobInput, createJobsClient, type ListBatchJobsOptions, type ListCronJobsOptions, type ListJobEventsOptions, type ListJobsOptions, type ProjectScopedOptions, type UpdateCronJobInput, VeryfrontJobsClient, type VeryfrontJobsClientConfig, } from "./jobs-client.js";
-export { type CronJob, CronJobSchema, type CronJobStatus, CronJobStatusSchema, type Job, type JobBatch, type JobBatchResult, JobBatchResultSchema, JobBatchSchema, type JobBatchStatusCounts, JobBatchStatusCountsSchema, type JobEvent, JobEventSchema, type JobEventsResponse, JobEventsResponseSchema, type JobLogsResponse, JobLogsResponseSchema, type JobResult, JobResultSchema, JobSchema, type JobStatus, JobStatusSchema, type JobTargetDefinition, JobTargetDefinitionSchema, type JobTargetDefinitionsResponse, JobTargetDefinitionsResponseSchema, type KnowledgeIngestBatchSource, KnowledgeIngestBatchSourceSchema, type KnowledgeIngestFileResult, KnowledgeIngestFileResultSchema, PageInfoSchema, type PaginatedCronJobsResponse, PaginatedCronJobsResponseSchema, type PaginatedJobsResponse, PaginatedJobsResponseSchema, type ReservedJobTargetFamily, ReservedJobTargetFamilySchema, } from "./schemas.js";
+export { type CronJob, CronJobSchema, type CronJobStatus, CronJobStatusSchema, type Job, type JobBatch, type JobBatchResult, JobBatchResultSchema, JobBatchSchema, type JobBatchStatusCounts, JobBatchStatusCountsSchema, type JobEvent, JobEventSchema, type JobEventsResponse, JobEventsResponseSchema, type JobKind, JobKindSchema, type JobListItem, JobListItemSchema, type JobLogsResponse, JobLogsResponseSchema, type JobResult, JobResultSchema, type JobResultSummary, JobResultSummarySchema, JobSchema, type JobStatus, JobStatusSchema, type JobTargetDefinition, JobTargetDefinitionSchema, type JobTargetDefinitionsResponse, JobTargetDefinitionsResponseSchema, type KnowledgeIngestBatchSource, KnowledgeIngestBatchSourceSchema, type KnowledgeIngestBatchSourceWithMessage, KnowledgeIngestBatchSourceWithMessageSchema, type KnowledgeIngestFailedFileResult, KnowledgeIngestFailedFileResultSchema, type KnowledgeIngestFileResult, KnowledgeIngestFileResultSchema, type KnowledgeIngestJobResult, KnowledgeIngestJobResultCountsSchema, KnowledgeIngestJobResultMetadataSchema, KnowledgeIngestJobResultSchema, type KnowledgeIngestSkippedFileResult, KnowledgeIngestSkippedFileResultSchema, PageInfoSchema, type PaginatedCronJobsResponse, PaginatedCronJobsResponseSchema, type PaginatedJobsResponse, PaginatedJobsResponseSchema, type ReservedJobTargetFamily, ReservedJobTargetFamilySchema, } from "./schemas.js";
 //# sourceMappingURL=index.d.ts.map

package/esm/src/jobs/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../src/src/jobs/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;AACH,OAAO,yBAAyB,CAAC;AAGjC,OAAO,EACL,KAAK,kBAAkB,EACvB,KAAK,cAAc,EACnB,gBAAgB,EAChB,KAAK,oBAAoB,EACzB,KAAK,mBAAmB,EACxB,KAAK,oBAAoB,EACzB,KAAK,eAAe,EACpB,KAAK,oBAAoB,EACzB,KAAK,kBAAkB,EACvB,mBAAmB,EACnB,KAAK,yBAAyB,GAC/B,MAAM,kBAAkB,CAAC;AAC1B,OAAO,EACL,KAAK,OAAO,EACZ,aAAa,EACb,KAAK,aAAa,EAClB,mBAAmB,EACnB,KAAK,GAAG,EACR,KAAK,QAAQ,EACb,KAAK,cAAc,EACnB,oBAAoB,EACpB,cAAc,EACd,KAAK,oBAAoB,EACzB,0BAA0B,EAC1B,KAAK,QAAQ,EACb,cAAc,EACd,KAAK,iBAAiB,EACtB,uBAAuB,EACvB,KAAK,eAAe,EACpB,qBAAqB,EACrB,KAAK,SAAS,EACd,eAAe,EACf,SAAS,EACT,KAAK,SAAS,EACd,eAAe,EACf,KAAK,mBAAmB,EACxB,yBAAyB,EACzB,KAAK,4BAA4B,EACjC,kCAAkC,EAClC,KAAK,0BAA0B,EAC/B,gCAAgC,EAChC,KAAK,yBAAyB,EAC9B,+BAA+B,EAC/B,cAAc,EACd,KAAK,yBAAyB,EAC9B,+BAA+B,EAC/B,KAAK,qBAAqB,EAC1B,2BAA2B,EAC3B,KAAK,uBAAuB,EAC5B,6BAA6B,GAC9B,MAAM,cAAc,CAAC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../src/src/jobs/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;AACH,OAAO,yBAAyB,CAAC;AAGjC,OAAO,EACL,KAAK,kBAAkB,EACvB,KAAK,cAAc,EACnB,gBAAgB,EAChB,KAAK,oBAAoB,EACzB,KAAK,mBAAmB,EACxB,KAAK,oBAAoB,EACzB,KAAK,eAAe,EACpB,KAAK,oBAAoB,EACzB,KAAK,kBAAkB,EACvB,mBAAmB,EACnB,KAAK,yBAAyB,GAC/B,MAAM,kBAAkB,CAAC;AAC1B,OAAO,EACL,KAAK,OAAO,EACZ,aAAa,EACb,KAAK,aAAa,EAClB,mBAAmB,EACnB,KAAK,GAAG,EACR,KAAK,QAAQ,EACb,KAAK,cAAc,EACnB,oBAAoB,EACpB,cAAc,EACd,KAAK,oBAAoB,EACzB,0BAA0B,EAC1B,KAAK,QAAQ,EACb,cAAc,EACd,KAAK,iBAAiB,EACtB,uBAAuB,EACvB,KAAK,OAAO,EACZ,aAAa,EACb,KAAK,WAAW,EAChB,iBAAiB,EACjB,KAAK,eAAe,EACpB,qBAAqB,EACrB,KAAK,SAAS,EACd,eAAe,EACf,KAAK,gBAAgB,EACrB,sBAAsB,EACtB,SAAS,EACT,KAAK,SAAS,EACd,eAAe,EACf,KAAK,mBAAmB,EACxB,yBAAyB,EACzB,KAAK,4BAA4B,EACjC,kCAAkC,EAClC,KAAK,0BAA0B,EAC/B,gCAAgC,EAChC,KAAK,qCAAqC,EAC1C,2CAA2C,EAC3C,KAAK,+BAA+B,EACpC,qCAAqC,EACrC,KAAK,yBAAyB,EAC9B,+BAA+B,EAC/B,KAAK,wBAAwB,EAC7B,oCAAoC,EACpC,sCAAsC,EACtC,8BAA8B,EAC9B,KAAK,gCAAgC,EACrC,sCAAsC,EACtC,cAAc,EACd,KAAK,yBAAyB,EAC9B,+BAA+B,EAC/B,KAAK,qBAAqB,EAC1B,2BAA2B,EAC3B,KAAK,uBAAuB,EAC5B,6BAA6B,GAC9B,MAAM,cAAc,CAAC"}

package/esm/src/jobs/index.js

@@ -30,4 +30,4 @@
  */
 import "../../_dnt.polyfills.js";
 export { createJobsClient, VeryfrontJobsClient, } from "./jobs-client.js";
-export { CronJobSchema, CronJobStatusSchema, JobBatchResultSchema, JobBatchSchema, JobBatchStatusCountsSchema, JobEventSchema, JobEventsResponseSchema, JobLogsResponseSchema, JobResultSchema, JobSchema, JobStatusSchema, JobTargetDefinitionSchema, JobTargetDefinitionsResponseSchema, KnowledgeIngestBatchSourceSchema, KnowledgeIngestFileResultSchema, PageInfoSchema, PaginatedCronJobsResponseSchema, PaginatedJobsResponseSchema, ReservedJobTargetFamilySchema, } from "./schemas.js";
+export { CronJobSchema, CronJobStatusSchema, JobBatchResultSchema, JobBatchSchema, JobBatchStatusCountsSchema, JobEventSchema, JobEventsResponseSchema, JobKindSchema, JobListItemSchema, JobLogsResponseSchema, JobResultSchema, JobResultSummarySchema, JobSchema, JobStatusSchema, JobTargetDefinitionSchema, JobTargetDefinitionsResponseSchema, KnowledgeIngestBatchSourceSchema, KnowledgeIngestBatchSourceWithMessageSchema, KnowledgeIngestFailedFileResultSchema, KnowledgeIngestFileResultSchema, KnowledgeIngestJobResultCountsSchema, KnowledgeIngestJobResultMetadataSchema, KnowledgeIngestJobResultSchema, KnowledgeIngestSkippedFileResultSchema, PageInfoSchema, PaginatedCronJobsResponseSchema, PaginatedJobsResponseSchema, ReservedJobTargetFamilySchema, } from "./schemas.js";