scai 0.1.117 → 0.1.118

package/dist/scripts/dbcheck.js

@@ -1,269 +1,184 @@
 import Database from "better-sqlite3";
 import path from "path";
 import os from "os";
-import { Config } from "../config.js";
 import fs from "fs";
+import { execSync } from "child_process";
+import { Config } from "../config.js";
+/* ───────────────────────── bootstrap ───────────────────────── */
 const cfg = Config.getRaw();
 const repoKey = cfg.activeRepo;
 if (!repoKey) {
-    console.error("❌ No active repo found. Use `scai set-index` to set one.");
+    console.error("❌ No active repo found. Use `scai set-index`.");
     process.exit(1);
 }
 const repoName = path.basename(repoKey);
 const scaiRepoRoot = path.join(os.homedir(), ".scai", "repos", repoName);
 const dbPath = path.join(scaiRepoRoot, "db.sqlite");
 if (!fs.existsSync(dbPath)) {
-    console.error(`❌ No database found at ${dbPath}`);
+    console.error(`❌ DB not found: ${dbPath}`);
     process.exit(1);
 }
 const db = new Database(dbPath);
-// Utility to log simple table stats
-function tableStats(table, column) {
-    const total = db.prepare(`SELECT COUNT(*) AS count FROM ${table}`).get().count;
-    let extra = "";
-    if (column) {
-        const nonNull = db.prepare(`SELECT COUNT(*) AS count FROM ${table} WHERE ${column} IS NOT NULL AND ${column} != ''`).get().count;
-        extra = ` | Non-null ${column}: ${nonNull}`;
-    }
-    console.log(`📊 ${table}: ${total}${extra}`);
+/* ───────────────────────── helpers ───────────────────────── */
+function tableCount(table) {
+    return db.prepare(`SELECT COUNT(*) AS c FROM ${table}`).get().c;
 }
-// === Files Table ===
-console.log("\n🗂 Table: files");
-tableStats("files", "summary");
-tableStats("files", "embedding");
-const processingStatuses = ["extracted", "skipped", "failed", "unprocessed"];
-for (const status of processingStatuses) {
-    const count = db.prepare(`SELECT COUNT(*) AS count FROM files WHERE processing_status = ?`).get(status).count;
-    console.log(`   Status '${status}': ${count}`);
+function nonEmptyCount(table, col) {
+    return db.prepare(`SELECT COUNT(*) AS c FROM ${table} WHERE ${col} IS NOT NULL AND ${col} != ''`).get().c;
 }
-// === Example Files ===
-const exampleFiles = db.prepare(`
-  SELECT id, filename, substr(summary,1,50) AS summary, substr(embedding,1,50) AS embedding
+function header(title) {
+    console.log(`\n${title}`);
+}
+/* ───────────────────────── files ───────────────────────── */
+header("🗂 files");
+const totalFiles = tableCount("files");
+const filesWithContent = nonEmptyCount("files", "content_text");
+console.log(`📊 total files: ${totalFiles}`);
+console.log(`📄 files with content: ${filesWithContent}`);
+header("⚙️ processing_status");
+const statuses = db.prepare(`
+  SELECT processing_status, COUNT(*) AS count
   FROM files
-  LIMIT 5
-`).all();
-console.log("\n🧾 Example Files:");
-exampleFiles.forEach(f => {
-    console.log(`   [${f.id}] ${f.filename} | summary: "${f.summary}" | embedding: "${f.embedding}"`);
-});
-// === FTS5 Check ===
-const ftsExists = db.prepare(`SELECT name FROM sqlite_master WHERE type='table' AND name='files_fts'`).get();
-console.log("\n🔍 FTS5 Table:", ftsExists ? "✅ exists" : "❌ missing");
-// === Functions Table ===
-console.log("\n🧑‍💻 Table: functions");
-tableStats("functions");
-const sampleFuncs = db.prepare(`
-  SELECT id, name, file_id, substr(content,1,50) AS preview
-  FROM functions
-  ORDER BY RANDOM()
-  LIMIT 5
+  GROUP BY processing_status
+  ORDER BY count DESC
 `).all();
-sampleFuncs.forEach(f => {
-    console.log(`   [${f.id}] ${f.name} (file ${f.file_id}): ${f.preview}`);
-});
-// === Graph Classes Table ===
-console.log("\n🏷 Table: graph_classes");
-tableStats("graph_classes");
-const sampleClasses = db.prepare(`
-  SELECT id, name, file_id, substr(content,1,50) AS preview
-  FROM graph_classes
-  ORDER BY RANDOM()
-  LIMIT 5
-`).all();
-sampleClasses.forEach(c => {
-    console.log(`   [${c.id}] ${c.name} (file ${c.file_id}): ${c.preview}`);
-});
-// === Graph Edges Table ===
-console.log("\n🔗 Table: graph_edges");
-tableStats("graph_edges", "relation");
-// Dynamically get column names
-const columns = db.prepare(`PRAGMA table_info(graph_edges)`).all();
-const colNames = columns.map(c => c.name);
-// Determine source/target ID columns
-const sourceCol = colNames.includes("source_id") ? "source_id" : "source_unique_id";
-const targetCol = colNames.includes("target_id") ? "target_id" : "target_unique_id";
-const sampleEdges = db.prepare(`
-  SELECT id, source_type, ${sourceCol} AS source_id,
-         target_type, ${targetCol} AS target_id,
-         relation
-  FROM graph_edges
-  ORDER BY RANDOM()
-  LIMIT 5
-`).all();
-sampleEdges.forEach(e => {
-    console.log(`   [${e.id}] ${e.source_type}(${e.source_id}) -[${e.relation}]-> ${e.target_type}(${e.target_id})`);
-});
-// === Tags Tables ===
-console.log("\n🏷 Table: graph_tags_master");
-tableStats("graph_tags_master");
-const sampleTags = db.prepare(`
-  SELECT id, name
-  FROM graph_tags_master
+statuses.forEach(s => console.log(`   ${s.processing_status ?? "NULL"}: ${s.count}`));
+/* ───────────────────────── FTS ───────────────────────── */
+header("🔍 files_fts");
+const ftsExists = db
+    .prepare(`SELECT 1 FROM sqlite_master WHERE type='table' AND name='files_fts'`)
+    .get();
+if (!ftsExists) {
+    console.log("❌ files_fts missing — search is broken");
+}
+else {
+    const cols = db.prepare(`PRAGMA table_info(files_fts)`).all();
+    const hasContentText = cols.some(c => c.name === "content_text");
+    if (!hasContentText) {
+        console.log("❌ files_fts.content_text missing");
+    }
+    else {
+        const indexed = nonEmptyCount("files_fts", "content_text");
+        console.log(`📦 indexed rows with content_text: ${indexed}`);
+        const sample = db.prepare(`
+      SELECT filename, substr(content_text,1,80) AS preview
+      FROM files_fts
+      WHERE files_fts MATCH 'function'
+      LIMIT 3
+    `).all();
+        if (sample.length === 0) {
+            console.log("⚠️ no FTS hits for 'function'");
+        }
+        else {
+            console.log("✅ FTS sample:");
+            sample.forEach(r => console.log(`   ${r.filename} | "${r.preview}"`));
+        }
+    }
+}
+/* ───────────────────────── folder capsules ───────────────────────── */
+header("📁 folder_capsules");
+console.log(`📊 total capsules: ${tableCount("folder_capsules")}`);
+const emptyCapsules = db.prepare(`
+  SELECT COUNT(*) AS c
+  FROM folder_capsules
+  WHERE capsule_json IS NULL OR capsule_json = ''
+`).get();
+if (emptyCapsules.c > 0) {
+    console.log(`⚠️ ${emptyCapsules.c} capsules have empty JSON`);
+}
+const sampleCaps = db.prepare(`
+  SELECT path, depth, confidence, source_file_count, capsule_json
+  FROM folder_capsules
   ORDER BY RANDOM()
   LIMIT 5
 `).all();
-sampleTags.forEach(t => {
-    console.log(`   [${t.id}] ${t.name}`);
+sampleCaps.forEach(c => {
+    let parsed = null;
+    try {
+        parsed = JSON.parse(c.capsule_json);
+    }
+    catch { }
+    console.log(`   ${c.path}`);
+    console.log(`      depth=${c.depth} confidence=${c.confidence} files=${c.source_file_count}`);
+    if (parsed) {
+        console.log(`      roles=${parsed.roles?.length ?? 0} ` +
+            `concerns=${parsed.concerns?.length ?? 0} ` +
+            `keyFiles=${parsed.keyFiles?.length ?? 0}`);
+    }
 });
-console.log("\n🔖 Table: graph_entity_tags");
-tableStats("graph_entity_tags");
-const sampleEntityTags = db.prepare(`
-  SELECT id, entity_type, entity_unique_id AS entity_id, tag_id
-  FROM graph_entity_tags
+/* ───────────────────────── functions ───────────────────────── */
+header("🧑‍💻 functions");
+console.log(`📊 total functions: ${tableCount("functions")}`);
+const funcSamples = db.prepare(`
+  SELECT id, name, file_id, substr(content,1,60) AS preview
+  FROM functions
   ORDER BY RANDOM()
   LIMIT 5
 `).all();
-sampleEntityTags.forEach(et => {
-    console.log(`   [${et.id}] ${et.entity_type}(${et.entity_id}) -> tag ${et.tag_id}`);
-});
-// === Function → Edges Check ===
-console.log("\n🕸 Function Calls / CalledBy consistency");
-// 1. Check for edges pointing to missing functions
-const missingFuncs = db.prepare(`
-  SELECT e.id, e.source_unique_id, e.target_unique_id
+funcSamples.forEach(f => console.log(`   [${f.id}] ${f.name} (file ${f.file_id}) "${f.preview}"`));
+/* ───────────────────────── graph (KG) ───────────────────────── */
+header("🏷 graph_classes");
+console.log(`📊 total classes: ${tableCount("graph_classes")}`);
+header("🔗 graph_edges");
+console.log(`📊 total edges: ${tableCount("graph_edges")}`);
+const danglingFuncs = db.prepare(`
+  SELECT e.id
   FROM graph_edges e
-  WHERE e.source_type = 'function'
+  WHERE e.source_type='function'
     AND NOT EXISTS (SELECT 1 FROM functions f WHERE f.unique_id = e.source_unique_id)
   UNION
-  SELECT e.id, e.source_unique_id, e.target_unique_id
+  SELECT e.id
   FROM graph_edges e
-  WHERE e.target_type = 'function'
+  WHERE e.target_type='function'
     AND NOT EXISTS (SELECT 1 FROM functions f WHERE f.unique_id = e.target_unique_id)
   LIMIT 10
 `).all();
-if (missingFuncs.length === 0) {
-    console.log("✅ All edges reference valid functions.");
+if (danglingFuncs.length === 0) {
+    console.log("✅ no dangling function edges");
 }
 else {
-    console.log("❌ Found edges pointing to missing functions:");
-    missingFuncs.forEach(e => {
-        console.log(`   Edge ${e.id}: ${e.source_unique_id} -> ${e.target_unique_id}`);
-    });
+    console.log(`❌ dangling function edges: ${danglingFuncs.length}`);
 }
-// 2. Functions with outgoing calls
-const funcWithCalls = db.prepare(`
-  SELECT f.id, f.name, COUNT(e.id) AS callCount
-  FROM functions f
-  JOIN graph_edges e
-    ON e.source_type = 'function'
-   AND e.source_unique_id = f.unique_id
-  GROUP BY f.id
-  HAVING callCount > 0
-  ORDER BY callCount DESC
-  LIMIT 5
-`).all();
-funcWithCalls.forEach(f => {
-    console.log(`   🔹 Function [${f.id}] ${f.name} has ${f.callCount} outgoing calls`);
-});
-// 3. Functions with incoming calls
-const funcWithCalledBy = db.prepare(`
-  SELECT f.id, f.name, COUNT(e.id) AS calledByCount
-  FROM functions f
-  JOIN graph_edges e
-    ON e.target_type = 'function'
-   AND e.target_unique_id = f.unique_id
-  GROUP BY f.id
-  HAVING calledByCount > 0
-  ORDER BY calledByCount DESC
-  LIMIT 5
-`).all();
-funcWithCalledBy.forEach(f => {
-    console.log(`   🔸 Function [${f.id}] ${f.name} is called by ${f.calledByCount} functions`);
-});
-// 4. Check for duplicate edges (same source→target)
-const duplicateEdges = db.prepare(`
-  SELECT source_unique_id, target_unique_id, COUNT(*) as dupCount
-  FROM graph_edges
-  WHERE source_type = 'function' AND target_type = 'function'
-  GROUP BY source_unique_id, target_unique_id
-  HAVING dupCount > 1
-  LIMIT 5
-`).all();
-if (duplicateEdges.length > 0) {
-    console.log("⚠️ Duplicate function call edges found:");
-    duplicateEdges.forEach(d => {
-        console.log(`   ${d.source_unique_id} -> ${d.target_unique_id} (x${d.dupCount})`);
-    });
-}
-else {
-    console.log("✅ No duplicate function call edges.");
-}
-// === File-specific check (AskCmd.ts) ===
-const targetPath = "/Users/rzs/dev/repos/scai/cli/src/commands/AskCmd.ts";
-// --- MUST KEEP: find the file row ---
-const targetFile = db.prepare(`
-  SELECT id, path, filename, processing_status
-  FROM files
-  WHERE path = ?
-`).get(targetPath);
-if (!targetFile) {
-    console.log(`❌ File not found in DB: ${targetPath}`);
-}
-else {
-    console.log(`\n📂 File found: [${targetFile.id}] ${targetFile.filename} (${targetFile.processing_status})`);
-    // --- MUST KEEP: list functions in this file ---
-    const funcs = db.prepare(`
-    SELECT id, name, unique_id, start_line, end_line
-    FROM functions
-    WHERE file_id = ?
-    ORDER BY start_line
-  `).all(targetFile.id);
-    console.log(`   Functions (${funcs.length}):`);
-    funcs.forEach((f) => {
-        console.log(`     [${f.id}] ${f.name} (${f.unique_id}) lines ${f.start_line}-${f.end_line}`);
-    });
-    // --- OPTIONAL: outgoing calls from this file’s functions ---
-    const outgoing = db.prepare(`
-    SELECT f.name AS source_name, e.target_unique_id, e.relation
-    FROM functions f
-    JOIN graph_edges e
-      ON e.source_unique_id = f.unique_id
-     AND e.source_type = 'function'
-    WHERE f.file_id = ?
-  `).all(targetFile.id);
-    console.log(`   Outgoing calls (${outgoing.length}):`);
-    outgoing.forEach((o) => {
-        console.log(`     ${o.source_name} -[${o.relation}]-> ${o.target_unique_id}`);
-    });
-    // --- OPTIONAL: incoming calls to this file’s functions ---
-    const incoming = db.prepare(`
-    SELECT f.name AS target_name, e.source_unique_id, e.relation
-    FROM functions f
-    JOIN graph_edges e
-      ON e.target_unique_id = f.unique_id
-     AND e.target_type = 'function'
-    WHERE f.file_id = ?
-  `).all(targetFile.id);
-    console.log(`   Incoming calls (${incoming.length}):`);
-    incoming.forEach((i) => {
-        console.log(`     ${i.source_unique_id} -[${i.relation}]-> ${i.target_name}`);
-    });
-    // --- NICE TO HAVE: dangling edges check ---
-    const dangling = db.prepare(`
-    SELECT e.id, e.source_unique_id, e.target_unique_id, e.relation
-    FROM graph_edges e
-    WHERE e.source_unique_id IN (SELECT unique_id FROM functions WHERE file_id = ?)
-       OR e.target_unique_id IN (SELECT unique_id FROM functions WHERE file_id = ?)
-  `).all(targetFile.id, targetFile.id);
-    console.log(`   Edges referencing this file's functions (dangling check): ${dangling.length}`);
-    dangling.forEach((d) => {
-        console.log(`     Edge ${d.id}: ${d.source_unique_id} -[${d.relation}]-> ${d.target_unique_id}`);
-    });
-    // --- OPTIONAL: consistency check summary ---
-    if (funcs.length > 0 && outgoing.length === 0 && incoming.length === 0 && dangling.length === 0) {
-        console.log("⚠️  This file has functions but no graph edges. Possible causes:");
-        console.log("     - Edges were never extracted for this file, OR");
-        console.log("     - unique_id mismatch between functions and edges.");
+/* ───────────────────────── Graphviz ───────────────────────── */
+header("🧩 graphviz");
+try {
+    const edges = db.prepare(`
+    SELECT source_type, source_unique_id, target_type, target_unique_id, relation
+    FROM graph_edges
+    LIMIT 300
+  `).all();
+    if (edges.length === 0) {
+        console.log("⚠️ no edges — skipping export");
     }
     else {
-        console.log("✅ Edges and functions are consistent for this file.");
+        const lines = [
+            "digraph G {",
+            "  rankdir=LR;",
+            '  node [shape=box, style=filled, color="#eaeaea"];'
+        ];
+        for (const e of edges) {
+            const s = `${e.source_type}_${e.source_unique_id}`.replace(/[^a-zA-Z0-9_]/g, "_");
+            const t = `${e.target_type}_${e.target_unique_id}`.replace(/[^a-zA-Z0-9_]/g, "_");
+            const label = (e.relation ?? "").replace(/"/g, "'");
+            lines.push(`  ${s} -> ${t} [label="${label}"];`);
+        }
+        lines.push("}");
+        const outDir = path.join(scaiRepoRoot, "graphs");
+        fs.mkdirSync(outDir, { recursive: true });
+        const dot = path.join(outDir, "graph-overview.dot");
+        const png = path.join(outDir, "graph-overview.png");
+        fs.writeFileSync(dot, lines.join("\n"));
+        console.log(`✅ dot written: ${dot}`);
+        try {
+            execSync(`dot -Tpng "${dot}" -o "${png}"`);
+            console.log(`✅ png written: ${png}`);
+        }
+        catch {
+            console.log("⚠️ graphviz not installed — png skipped");
+        }
     }
-    const funcsDebug = db.prepare(`
-  SELECT id, name, unique_id
-  FROM functions
-  WHERE file_id = ?
-`).all(targetFile.id);
-    console.log("Funcsdebug:\n");
-    console.log(funcsDebug);
 }
-console.log("\n✅ DB check completed.\n");
+catch (err) {
+    console.error("❌ graphviz export failed", err);
+}
+console.log("\n✅ DB check complete\n");