sphica 0.6.35 → 0.6.36

package/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://anthropic.com/claude-code/plugin.schema.json",
   "name": "sphica",
-  "version": "0.6.35",
+  "version": "0.6.36",
   "description": "Records Claude Code and Codex sessions on your machine and keeps past implementation and decisions, with their sources, for your agent to find.",
   "author": {
     "name": "iroha924",

package/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "sphica",
-  "version": "0.6.35",
+  "version": "0.6.36",
   "description": "Records Claude Code and Codex sessions on your machine and keeps past implementation and decisions, with their sources, for your agent to find.",
   "author": {
     "name": "iroha924",

package/README.md

@@ -74,7 +74,7 @@ In Codex, open `/hooks` and mark Sphica's hooks as trusted. Automatic recording
 **3. Set up in your repository**
 
 ```bash
-cd ~/Projects/your-repo
+cd path/to/your-repo
 sphica init
 ```
 

package/db/migrations/0011.sql

@@ -0,0 +1,26 @@
+-- Revision 10 → 11 of generation 2: a harvest run keeps the sources it may cite, chosen when it begins (harvest_run_source). A running
+-- harvest has none, so it is removed and its session gets "Begin again"; nothing cites a running run (a save also finishes it).
+-- Every statement that stays is the same as in the schema at revision 11.
+
+create temp table sphica_migration_note (rule text, item text, action text);
+
+create table harvest_run_source (
+  run_id integer not null references extraction_run (id) on delete cascade,
+  source_id integer not null references source (id) on delete cascade,
+  primary key (run_id, source_id)
+) strict;
+create index harvest_run_source_source on harvest_run_source (source_id);
+
+create trigger harvest_run_source_run before insert on harvest_run_source begin
+  select raise(abort, 'only a harvest run keeps sources, from its own project')
+  where (select origin from extraction_run where id = new.run_id) is not 'harvest'
+    or (select project_id from source where id = new.source_id) is not (select project_id from extraction_run where id = new.run_id);
+end;
+
+insert into sphica_migration_note
+select 'a harvest run still running, begun before its sources were kept', 'run ' || id || ' ' || target || ' begun ' || started_at,
+  'removed; begin the harvest again'
+from extraction_run where origin = 'harvest' and status = 'running' order by id;
+delete from extraction_run where origin = 'harvest' and status = 'running';
+
+pragma user_version = 11;

package/db/schema.sql

@@ -6,7 +6,7 @@
 --   captured sources   session, source, artifact_link, edit_observation: what was said or written, never rewritten
 --                      (the owner can forget chosen sources: forget_batch and source_forgotten keep what was removed, without its text)
 --   extracted units    unit and its option, evidence, adoption, link, state, anchor, alias tables: what was decided or implemented
---   processing         extraction_run, source_processing: what has been looked at and saved, so gaps are counted
+--   processing         extraction_run, source_processing, harvest_run_source: what has been looked at and saved, so gaps are counted
 --   work and delivery  work, delivery, delivery_unit: the current work status and what the hooks injected
 -- Every table is STRICT and every primary key is not null. Times are ISO 8601 UTC (`Date#toISOString()`); `strftime(...) is column` rejects others.
 -- Paths (a source's, an edit observation's, an anchor's) are repository-relative with forward slashes, in one form, so one place
@@ -293,6 +293,15 @@ create table source_processing (
 ) strict;
 create index source_processing_run on source_processing (run_id);
 
+-- The sources a harvest run may show and cite, chosen when it begins, so another harvest of the same pull request cannot change them.
+-- Ids only; a source the owner forgets leaves with its row
+create table harvest_run_source (
+  run_id integer not null references extraction_run (id) on delete cascade,
+  source_id integer not null references source (id) on delete cascade,
+  primary key (run_id, source_id)
+) strict;
+create index harvest_run_source_source on harvest_run_source (source_id);
+
 -- An extracted unit. Its text is never rewritten: corrections are successors, withdrawals, retractions, and anchor replacements.
 -- extraction: supported (every evidence span was found in retained text) or quarantined (with reason).
 -- lifecycle changes only through unit_state (its trigger sets this column). revision rises with every change to the unit's relations,
@@ -981,6 +990,11 @@ create trigger source_processing_project before insert on source_processing begi
   select raise(abort, 'source and run belong to different projects')
   where (select project_id from source where id = new.source_id) is not (select project_id from extraction_run where id = new.run_id);
 end;
+create trigger harvest_run_source_run before insert on harvest_run_source begin
+  select raise(abort, 'only a harvest run keeps sources, from its own project')
+  where (select origin from extraction_run where id = new.run_id) is not 'harvest'
+    or (select project_id from source where id = new.source_id) is not (select project_id from extraction_run where id = new.run_id);
+end;
 
 -- Every change to a unit's relations raises its revision (stale drafts are refused against it)
 create trigger unit_rev_evidence_i after insert on unit_evidence begin update unit set revision = revision + 1 where id = new.unit_id; end;
@@ -1211,4 +1225,4 @@ create trigger capture_delivery_prune_insert instead of insert on capture_delive
     order by d.at, d.id limit 200);
 end;
 
-pragma user_version = 10;
+pragma user_version = 11;

package/dist/capture.js

@@ -11396,7 +11396,7 @@ var HOST_MARKS = new Set([".orphaned_at", ".in_use"]);
 
 // server/src/sqlite.ts
 var SCHEMA_GENERATION = 2;
-var SCHEMA_REVISION = 10;
+var SCHEMA_REVISION = 11;
 var sphicaHome = () => process.env.SPHICA_HOME || path2.join(os.homedir(), ".sphica");
 var dbFile = () => process.env.SPHICA_DB || path2.join(sphicaHome(), "sphica.db");
 function requireRuntime() {
@@ -11597,6 +11597,7 @@ var INGEST_INSERTS = new Set([
   "ingest_source",
   "extraction_run",
   "source_processing",
+  "harvest_run_source",
   "unit",
   "unit_option",
   "unit_evidence",
@@ -11692,6 +11693,7 @@ var FORGET_WRITES = {
     "unit_evidence",
     "unit_adoption",
     "source_processing",
+    "harvest_run_source",
     "source_fts",
     "field_def",
     "unit_field",

package/dist/cli.js

@@ -93,7 +93,6 @@ var require_src = __commonJS(function(exports, module) {
 
 // server/src/cli.ts
 import fs10 from "node:fs";
-import os4 from "node:os";
 import path10 from "node:path";
 
 // server/node_modules/@clack/core/dist/index.mjs
@@ -4912,7 +4911,7 @@ ${indent}${caution(note)}` : ""}`);
 
 // server/src/sqlite.ts
 var SCHEMA_GENERATION = 2;
-var SCHEMA_REVISION = 10;
+var SCHEMA_REVISION = 11;
 var sphicaHome = () => process.env.SPHICA_HOME || path3.join(os2.homedir(), ".sphica");
 var dbFile = () => process.env.SPHICA_DB || path3.join(sphicaHome(), "sphica.db");
 function requireRuntime() {
@@ -15957,6 +15956,7 @@ var INGEST_INSERTS = new Set([
   "ingest_source",
   "extraction_run",
   "source_processing",
+  "harvest_run_source",
   "unit",
   "unit_option",
   "unit_evidence",
@@ -16052,6 +16052,7 @@ var FORGET_WRITES = {
     "unit_evidence",
     "unit_adoption",
     "source_processing",
+    "harvest_run_source",
     "source_fts",
     "field_def",
     "unit_field",
@@ -16120,7 +16121,6 @@ var sessionId = (projectId, host, externalId) => uuidFrom(String(projectId), hos
 // server/src/project.ts
 import { execFileSync as execFileSync2 } from "node:child_process";
 import fs6 from "node:fs";
-import os3 from "node:os";
 import path6 from "node:path";
 
 // server/src/anchors.ts
@@ -16348,48 +16348,6 @@ function nameLocal(dir, name) {
   });
   return { key: `local:${name}`, root, name };
 }
-function underHome(p, homes, api = path6) {
-  for (const h2 of homes) {
-    const rel = api.relative(h2, p);
-    if (rel === "")
-      return "~";
-    if (rel !== ".." && !rel.startsWith(`..${api.sep}`) && !api.isAbsolute(rel))
-      return `~${api.sep}${rel}`;
-  }
-  return p;
-}
-var projectsDir = () => path6.join(os3.homedir(), "Projects");
-function localRoots(roots = [projectsDir()]) {
-  const seen = new Map;
-  const add = (p) => {
-    if (p)
-      seen.set(p.key, [...new Set([...seen.get(p.key) ?? [], p.root])]);
-  };
-  for (const r2 of roots) {
-    let entries = [];
-    try {
-      entries = fs6.readdirSync(r2, { withFileTypes: true });
-    } catch {
-      continue;
-    }
-    for (const e of entries)
-      if (e.isDirectory() && !e.name.startsWith("."))
-        add(identify(path6.join(r2, e.name)));
-  }
-  for (const [root, name] of Object.entries(localMap())) {
-    if (LOCAL_KEY.test(name) && fs6.existsSync(root))
-      add({ key: `local:${name}`, root, name });
-  }
-  const found = new Map;
-  const ambiguous = new Map;
-  for (const [key, dirs] of seen) {
-    if (dirs.length === 1 && dirs[0])
-      found.set(key, dirs[0]);
-    else
-      ambiguous.set(key, dirs);
-  }
-  return { found, ambiguous };
-}
 function relativeTo(root, file, cwd = root) {
   const abs = path6.resolve(cwd, file);
   const rel = path6.relative(root, abs);
@@ -18769,14 +18727,7 @@ async function doctor(cwd) {
         if (twice.length)
           say2("warn", "Records", `${plural(twice.length, "set")} of live records hold the same words. Read them with Sphica's read and withdraw the extra ones with /sphica:glean`);
         await adoptionStops(db, q.waiting, say2);
-        const { found, ambiguous } = localRoots();
-        const homes = [os4.homedir()];
-        try {
-          homes.push(fs10.realpathSync.native(os4.homedir()));
-        } catch {}
-        const tilde = (p) => underHome(p, homes);
         const rows = await db.selectFrom("project as p").select((eb) => [
-          "p.key",
           "p.name",
           eb.selectFrom("unit as u").select((u3) => u3.fn.countAll().as("n")).whereRef("u.project_id", "=", "p.id").as("records"),
           eb.selectFrom("extraction_run as r").select((r2) => r2.fn.max("r.finished_at").as("at")).whereRef("r.project_id", "=", "p.id").where("r.status", "=", "saved").where("r.origin", "<>", "migration").as("extracted")
@@ -18784,11 +18735,8 @@ async function doctor(cwd) {
         if (rows.length)
           console.log(section("Projects", true));
         const column = Math.max(...rows.map((x) => width(inline(x.name)))) + 2;
-        for (const x of rows) {
-          const copies = ambiguous.get(x.key);
-          const where = copies ? ` (${copies.length} copies: ${copies.map((c) => inline(tilde(c))).sort().join(", ")})` : found.get(x.key) ? "" : ` (not found in ${inline(tilde(projectsDir()))} or the named projects)`;
-          console.log(indent(`  ${mark("none")} ${pad(inline(x.name), column)}${plural(Number(x.records ?? 0), "record")}${x.extracted ? ` / last extraction ${new Date(x.extracted).toLocaleString("sv-SE")}` : ""}${where}`));
-        }
+        for (const x of rows)
+          console.log(indent(`  ${mark("none")} ${pad(inline(x.name), column)}${plural(Number(x.records ?? 0), "record")}${x.extracted ? ` / last extraction ${new Date(x.extracted).toLocaleString("sv-SE")}` : ""}`));
       });
     } catch (e) {
       say2("fail", "DB", `cannot read: ${plain(reason(e))}`);

package/dist/deliver.js

@@ -11427,7 +11427,7 @@ var HOST_MARKS = new Set([".orphaned_at", ".in_use"]);
 
 // server/src/sqlite.ts
 var SCHEMA_GENERATION = 2;
-var SCHEMA_REVISION = 10;
+var SCHEMA_REVISION = 11;
 var sphicaHome = () => process.env.SPHICA_HOME || path3.join(os.homedir(), ".sphica");
 var dbFile = () => process.env.SPHICA_DB || path3.join(sphicaHome(), "sphica.db");
 function requireRuntime() {
@@ -11652,6 +11652,7 @@ var INGEST_INSERTS = new Set([
   "ingest_source",
   "extraction_run",
   "source_processing",
+  "harvest_run_source",
   "unit",
   "unit_option",
   "unit_evidence",
@@ -11747,6 +11748,7 @@ var FORGET_WRITES = {
     "unit_evidence",
     "unit_adoption",
     "source_processing",
+    "harvest_run_source",
     "source_fts",
     "field_def",
     "unit_field",

package/dist/mcp-record.js

@@ -46183,7 +46183,7 @@ var HOST_MARKS = new Set([".orphaned_at", ".in_use"]);
 
 // server/src/sqlite.ts
 var SCHEMA_GENERATION = 2;
-var SCHEMA_REVISION = 10;
+var SCHEMA_REVISION = 11;
 var sphicaHome = () => process.env.SPHICA_HOME || path2.join(os.homedir(), ".sphica");
 var dbFile = () => process.env.SPHICA_DB || path2.join(sphicaHome(), "sphica.db");
 function requireRuntime() {
@@ -46384,6 +46384,7 @@ var INGEST_INSERTS = new Set([
   "ingest_source",
   "extraction_run",
   "source_processing",
+  "harvest_run_source",
   "unit",
   "unit_option",
   "unit_evidence",
@@ -46479,6 +46480,7 @@ var FORGET_WRITES = {
     "unit_evidence",
     "unit_adoption",
     "source_processing",
+    "harvest_run_source",
     "source_fts",
     "field_def",
     "unit_field",
@@ -47787,6 +47789,7 @@ var item = (v) => ({
   commit: null,
   ...v
 });
+var said = (body) => body?.trim() ? body : "";
 var sha = (s) => s && /^[0-9a-f]{40}$/.test(s) ? s : null;
 var cleanPath = (p) => p && !p.startsWith("/") && !p.includes("\\") && !/\p{Cc}/u.test(p) && !p.split("/").some((x) => x === "" || x === "." || x === "..") ? p : null;
 async function readPull(get, number5, repo = null) {
@@ -47807,52 +47810,49 @@ async function readPull(get, number5, repo = null) {
     association: p.author_association ?? null,
     url: p.html_url,
     createdAt: p.created_at,
-    text: p.body?.trim() ? p.body : ""
+    text: said(p.body)
   }));
   for (const c of comments)
-    if (c.body?.trim())
-      items.push(item({
-        kind: "pr_comment",
-        artifact,
-        externalId: `comment:${c.id}`,
-        author: c.user ?? null,
-        association: c.author_association ?? null,
-        url: c.html_url,
-        createdAt: c.created_at,
-        text: c.body
-      }));
+    items.push(item({
+      kind: "pr_comment",
+      artifact,
+      externalId: `comment:${c.id}`,
+      author: c.user ?? null,
+      association: c.author_association ?? null,
+      url: c.html_url,
+      createdAt: c.created_at,
+      text: said(c.body)
+    }));
   for (const r of reviews)
-    if (r.body?.trim())
-      items.push(item({
-        kind: "review",
-        artifact,
-        externalId: `review:${r.id}`,
-        author: r.user ?? null,
-        association: r.author_association ?? null,
-        url: r.html_url,
-        createdAt: r.submitted_at ?? p.created_at,
-        text: r.body
-      }));
-  for (const c of reviewComments)
-    if (c.body?.trim()) {
-      const end = c.line ?? null;
-      const start = c.start_line && c.start_line <= (end ?? 0) && (c.start_side ?? c.side) === c.side ? c.start_line : end;
-      items.push(item({
-        kind: "review_comment",
-        artifact,
-        externalId: `review_comment:${c.id}`,
-        author: c.user ?? null,
-        association: c.author_association ?? null,
-        parent: c.in_reply_to_id ? `review_comment:${c.in_reply_to_id}` : null,
-        url: c.html_url,
-        createdAt: c.created_at,
-        text: c.body,
-        path: cleanPath(c.path),
-        lines: cleanPath(c.path) && end && start ? [start, end] : null,
-        hunk: c.diff_hunk ?? null,
-        commit: sha(c.commit_id)
-      }));
-    }
+    items.push(item({
+      kind: "review",
+      artifact,
+      externalId: `review:${r.id}`,
+      author: r.user ?? null,
+      association: r.author_association ?? null,
+      url: r.html_url,
+      createdAt: r.submitted_at ?? p.created_at,
+      text: said(r.body)
+    }));
+  for (const c of reviewComments) {
+    const end = c.line ?? null;
+    const start = c.start_line && c.start_line <= (end ?? 0) && (c.start_side ?? c.side) === c.side ? c.start_line : end;
+    items.push(item({
+      kind: "review_comment",
+      artifact,
+      externalId: `review_comment:${c.id}`,
+      author: c.user ?? null,
+      association: c.author_association ?? null,
+      parent: c.in_reply_to_id ? `review_comment:${c.in_reply_to_id}` : null,
+      url: c.html_url,
+      createdAt: c.created_at,
+      text: said(c.body),
+      path: cleanPath(c.path),
+      lines: cleanPath(c.path) && end && start ? [start, end] : null,
+      hunk: c.diff_hunk ?? null,
+      commit: sha(c.commit_id)
+    }));
+  }
   for (const c of commits)
     items.push(item({
       kind: "commit_message",
@@ -47885,29 +47885,27 @@ async function readIssue(get, n) {
   const issue2 = await get(`issues/${n}`);
   if (issue2.pull_request)
     return items;
-  if (issue2.body?.trim())
+  items.push(item({
+    kind: "issue_body",
+    artifact: `issue:${n}`,
+    externalId: `issue:${n}`,
+    author: issue2.user ?? null,
+    association: issue2.author_association ?? null,
+    url: issue2.html_url,
+    createdAt: issue2.created_at,
+    text: said(issue2.body)
+  }));
+  for (const c of await get(`issues/${n}/comments?per_page=100`, true))
     items.push(item({
-      kind: "issue_body",
+      kind: "issue_comment",
       artifact: `issue:${n}`,
-      externalId: `issue:${n}`,
-      author: issue2.user ?? null,
-      association: issue2.author_association ?? null,
-      url: issue2.html_url,
-      createdAt: issue2.created_at,
-      text: issue2.body
+      externalId: `comment:${c.id}`,
+      author: c.user ?? null,
+      association: c.author_association ?? null,
+      url: c.html_url,
+      createdAt: c.created_at,
+      text: said(c.body)
     }));
-  for (const c of await get(`issues/${n}/comments?per_page=100`, true))
-    if (c.body?.trim())
-      items.push(item({
-        kind: "issue_comment",
-        artifact: `issue:${n}`,
-        externalId: `comment:${c.id}`,
-        author: c.user ?? null,
-        association: c.author_association ?? null,
-        url: c.html_url,
-        createdAt: c.created_at,
-        text: c.body
-      }));
   return items;
 }
 function githubTarget(repo, url2) {
@@ -47996,12 +47994,16 @@ async function linkIssues(db, projectId2, number5, closes) {
       kind: "closes"
     }).onConflict((oc) => oc.doNothing()).execute();
 }
-async function pullSources(db, projectId2, number5) {
+async function pullSourceIds(db, projectId2, number5) {
   const artifacts = [
     `pr:${number5}`,
     ...(await db.selectFrom("artifact_link").select("to_artifact").where("project_id", "=", projectId2).where("from_artifact", "=", `pr:${number5}`).execute()).map((l) => l.to_artifact)
   ];
-  return db.selectFrom("source as s").where("s.project_id", "=", projectId2).where("s.artifact", "in", artifacts).where(({ not, exists, selectFrom }) => not(exists(selectFrom("source as n").select("n.id").whereRef("n.project_id", "=", "s.project_id").whereRef("n.kind", "=", "s.kind").whereRef("n.external_id", "=", "s.external_id").whereRef("n.revision", ">", "s.revision")))).where(({ not, exists, selectFrom }) => not(exists(selectFrom("source_forgotten as f").select("f.source_id").whereRef("f.project_id", "=", "s.project_id").whereRef("f.kind", "=", "s.kind").whereRef("f.external_id", "=", "s.external_id").whereRef("f.revision", ">", "s.revision")))).select([
+  const rows = await db.selectFrom("source as s").where("s.project_id", "=", projectId2).where("s.artifact", "in", artifacts).where(({ not, exists, selectFrom }) => not(exists(selectFrom("source as n").select("n.id").whereRef("n.project_id", "=", "s.project_id").whereRef("n.kind", "=", "s.kind").whereRef("n.external_id", "=", "s.external_id").whereRef("n.revision", ">", "s.revision")))).where(({ not, exists, selectFrom }) => not(exists(selectFrom("source_forgotten as f").select("f.source_id").whereRef("f.project_id", "=", "s.project_id").whereRef("f.kind", "=", "s.kind").whereRef("f.external_id", "=", "s.external_id").whereRef("f.revision", ">", "s.revision")))).select("s.id").execute();
+  return rows.map((r) => r.id);
+}
+async function runSources(db, runId) {
+  return db.selectFrom("harvest_run_source as h").innerJoin("source as s", "s.id", "h.source_id").where("h.run_id", "=", runId).where(({ not, exists, selectFrom }) => not(exists(selectFrom("source_forgotten as f").select("f.source_id").whereRef("f.project_id", "=", "s.project_id").whereRef("f.kind", "=", "s.kind").whereRef("f.external_id", "=", "s.external_id").whereRef("f.revision", ">", "s.revision")))).select([
     "s.id",
     "s.kind",
     "s.artifact",
@@ -48010,7 +48012,6 @@ async function pullSources(db, projectId2, number5) {
     "s.author_login",
     "s.author_association",
     "s.created_at",
-    "s.captured_at",
     "s.path",
     "s.line_start",
     "s.text"
@@ -49495,8 +49496,8 @@ async function checkGlean(db, target, raw, facts = prepareGlean(target.root, raw
     if (op.op === "replace_anchor") {
       if (!repoPath(op.to.path))
         errors3.push(`${what}: the path is not inside the repository`);
-      const said = await span(op.source, op.quote, what);
-      if (said && said.s.author_kind !== "owner")
+      const said2 = await span(op.source, op.quote, what);
+      if (said2 && said2.s.author_kind !== "owner")
         errors3.push(`${what}: only the owner's words can move an anchor`);
       const from = repoPath(op.from.path);
       if (!from)
@@ -49974,7 +49975,7 @@ async function beginHarvest(db, projectId2, number5, get, beginCall) {
     const kept = (await storeItems(trx, projectId2, pull.items)).filter((id) => id !== null).length;
     await linkIssues(trx, projectId2, number5, pull.closes);
     const run = newRunId();
-    await openRun(trx, {
+    const runId = await openRun(trx, {
       projectId: projectId2,
       origin: "harvest",
       target: `pr:${number5}`,
@@ -49982,6 +49983,8 @@ async function beginHarvest(db, projectId2, number5, get, beginCall) {
       draftId: run,
       beginCall
     });
+    for (const sourceId of await pullSourceIds(trx, projectId2, number5))
+      await trx.insertInto("harvest_run_source").values({ run_id: runId, source_id: sourceId }).execute();
     return { run, sources: kept };
   });
 }
@@ -50036,7 +50039,7 @@ async function agentRun(db, run, call) {
 async function scopeOf(db, run, root) {
   if (run.origin === "harvest") {
     const number5 = Number(run.target.slice("pr:".length));
-    const sources2 = (await pullSources(db, run.project_id, number5)).filter((x) => x.captured_at <= run.started_at);
+    const sources2 = await runSources(db, run.id);
     return {
       target: {
         projectId: run.project_id,

package/dist/mcp.js

@@ -46004,7 +46004,7 @@ var HOST_MARKS = new Set([".orphaned_at", ".in_use"]);
 
 // server/src/sqlite.ts
 var SCHEMA_GENERATION = 2;
-var SCHEMA_REVISION = 10;
+var SCHEMA_REVISION = 11;
 var sphicaHome = () => process.env.SPHICA_HOME || path2.join(os.homedir(), ".sphica");
 var dbFile = () => process.env.SPHICA_DB || path2.join(sphicaHome(), "sphica.db");
 function requireRuntime() {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "sphica",
-  "version": "0.6.35",
+  "version": "0.6.36",
   "description": "Records Claude Code and Codex sessions on your machine and keeps past implementation and decisions, with their sources, for your agent to find.",
   "license": "MIT",
   "type": "module",