scai 0.1.116 → 0.1.118

package/dist/db/sqlTemplates.js

@@ -1,8 +1,8 @@
 // --- Files ---
 // Upsert file metadata into `files`
 export const upsertFileTemplate = `
-  INSERT INTO files (path, filename, summary, type, last_modified, indexed_at, embedding)
-  VALUES (:path, :filename, :summary, :type, :lastModified, :indexedAt, :embedding)
+  INSERT INTO files (path, filename, summary, type, last_modified, indexed_at)
+  VALUES (:path, :filename, :summary, :type, :lastModified, :indexedAt)
   ON CONFLICT(path) DO UPDATE SET
     filename = excluded.filename,
     summary = CASE
@@ -12,17 +12,12 @@ export const upsertFileTemplate = `
     END,
     type = excluded.type,
     last_modified = excluded.last_modified,
-    indexed_at = excluded.indexed_at,
-    embedding = CASE
-      WHEN excluded.embedding IS NOT NULL AND excluded.embedding != files.embedding
-      THEN excluded.embedding
-      ELSE files.embedding
-    END
+    indexed_at = excluded.indexed_at
 `;
 // Insert or replace into FTS
 export const upsertFileFtsTemplate = `
-  INSERT OR REPLACE INTO files_fts (rowid, filename, summary, path)
-  VALUES ((SELECT id FROM files WHERE path = :path), :filename, :summary, :path)
+  INSERT OR REPLACE INTO files_fts (rowid, filename, summary, path, content_text)
+  VALUES ((SELECT id FROM files WHERE path = :path), :filename, :summary, :path, :contentText)
 `;
 // Simple file query (no scoring)
 export const queryFilesTemplate = `
@@ -35,7 +30,7 @@ export const queryFilesTemplate = `
 // FTS search with bm25 scoring
 export const searchFilesTemplate = `
   SELECT fts.rowid AS id, f.path, f.filename, f.summary, f.type,
-         bm25(files_fts) AS bm25Score, f.embedding
+         bm25(files_fts) AS bm25Score
   FROM files f
   JOIN files_fts fts ON f.id = fts.rowid
   WHERE fts.files_fts MATCH ?
@@ -44,13 +39,13 @@ export const searchFilesTemplate = `
 `;
 // --- Functions ---
 export const insertFunctionTemplate = `
-  INSERT INTO functions (file_id, name, start_line, end_line, content, embedding, lang)
-  VALUES (:file_id, :name, :start_line, :end_line, :content, :embedding, :lang)
+  INSERT INTO functions (file_id, name, start_line, end_line, content, lang, unique_id)
+  VALUES (:file_id, :name, :start_line, :end_line, :content, :lang, :unique_id)
 `;
 // --- Graph ---
 export const insertGraphClassTemplate = `
-  INSERT INTO graph_classes (file_id, name, start_line, end_line, content, embedding, lang)
-  VALUES (:file_id, :name, :start_line, :end_line, :content, :embedding, :lang)
+  INSERT INTO graph_classes (file_id, name, start_line, end_line, content, lang)
+  VALUES (:file_id, :name, :start_line, :end_line, :content, :lang)
 `;
 export const insertEdgeTemplate = `
   INSERT INTO graph_edges (
@@ -89,6 +84,36 @@ export const insertGraphEntityTagTemplate = `
   ON CONFLICT(entity_type, entity_unique_id, tag_id) DO NOTHING
 `;
 // --- File Processing Status ---
+/* processing_status values:
+   "unprocessed" → newly discovered, not yet indexed
+   "indexed"     → content read and inserted into FTS
+   "capsuled"    → folder capsule created
+   "extracted"   → functions/classes extracted for knowledge graph
+   "kg_done"     → knowledge graph updated
+   "skipped"     → ignored (invalid, missing, or binary)
+   "failed"      → could not be processed due to error
+*/
+export const markFileAsProcessedByPath = `
+  UPDATE files
+  SET processing_status = 'processed'
+  WHERE path = @path
+`;
+export const markFileAsIndexed = `
+  UPDATE files
+  SET processing_status = 'indexed',
+      indexed_at = datetime('now')
+  WHERE path = @path
+`;
+export const markFileAsCapsuled = `
+  UPDATE files
+SET processing_status = 'capsulized'
+WHERE path LIKE ?
+`;
+export const markFileAsKgDone = `
+  UPDATE files
+  SET processing_status = 'kg_done'
+  WHERE path = @path
+`;
 export const markFileAsExtractedTemplate = `
   UPDATE files
   SET processing_status = 'extracted',
@@ -108,9 +133,9 @@ export const markFileAsFailedTemplate = `
   WHERE id = :id
 `;
 export const selectUnprocessedFiles = `
-  SELECT id, path, type, summary, indexed_at, last_modified, processing_status
+  SELECT id, path, type, content_text, indexed_at, last_modified, processing_status
   FROM files
-  WHERE processing_status = 'unprocessed' OR summary IS NULL OR summary = ''
+  WHERE processing_status = 'unprocessed'
   ORDER BY last_modified DESC
   LIMIT ?
 `;
@@ -120,10 +145,16 @@ export const markFileAsSkippedByPath = `
       functions_extracted_at = NULL
   WHERE path = @path
 `;
-export const updateFileWithSummaryAndEmbedding = `
+export const updateFileWithSummary = `
   UPDATE files
   SET summary = @summary,
-      embedding = @embedding,
+      content_text = @contentText,
       indexed_at = datetime('now')
   WHERE path = @path
 `;
+// --- File Processing Status Checks ---
+export const countUnprocessedFiles = `
+  SELECT COUNT(*) as count
+  FROM files
+  WHERE processing_status = 'unprocessed'
+`;

package/dist/fileRules/builtins.js

@@ -0,0 +1,31 @@
+export const BUILTINS = {
+    ts: new Set([
+        // String methods
+        "trim", "toLowerCase", "toUpperCase", "includes", "startsWith", "endsWith", "replace", "match", "split",
+        // Array methods
+        "map", "forEach", "filter", "reduce", "some", "every", "find", "findIndex", "push", "pop", "shift", "unshift", "slice", "splice",
+        // Object methods
+        "keys", "values", "entries", "assign",
+        // Number/Math
+        "parseInt", "parseFloat", "isNaN", "isFinite", "floor", "ceil", "round", "abs", "max", "min", "random",
+        // JSON / Promise / RegExp
+        "stringify", "parse", "then", "catch", "finally", "test",
+        // Console
+        "log", "error", "warn", "info",
+    ]),
+    java: new Set([
+        // Object methods
+        "equals", "hashCode", "toString", "clone", "finalize", "notify", "notifyAll", "wait",
+        // String methods
+        "charAt", "compareTo", "contains", "endsWith", "equalsIgnoreCase", "indexOf", "isEmpty",
+        "length", "replace", "split", "startsWith", "substring", "toLowerCase", "toUpperCase", "trim",
+        // Math methods
+        "abs", "max", "min", "floor", "ceil", "round", "sqrt", "pow", "random",
+        // Collections (common)
+        "add", "addAll", "remove", "removeAll", "contains", "containsAll", "size", "clear", "get", "put",
+        // System
+        "exit", "currentTimeMillis", "arraycopy",
+        // Misc
+        "format", "parseInt", "parseDouble",
+    ]),
+};

package/dist/fileRules/codeAllowedExtensions.js

@@ -0,0 +1,4 @@
+// src/fileRules/codeAllowedExtensions.ts
+export const CODE_ALLOWED_EXTENSIONS = [
+    'js', 'javascript', 'ts', 'typescript', 'java',
+];

package/dist/fileRules/fileExceptions.js

@@ -1,9 +1,6 @@
 export const specificFileExceptions = [
     // 🧑‍💻 Project Configuration Files
     'package.json', // Keep package.json for NPM/Yarn dependency management
-    'package-lock.json', // Keep package-lock.json for npm lockfile
-    'yarn.lock', // Keep yarn.lock for Yarn dependency lockfile
-    'pnpm-lock.yaml', // Keep pnpm-lock.yaml for pnpm lockfile
     'tsconfig.json', // Keep TypeScript configuration file
     'tsconfig.build.json', // Keep build-specific tsconfig file
     'tsconfig.prod.json', // Keep production-specific tsconfig file
@@ -41,10 +38,6 @@ export const specificFileExceptions = [
     'build.gradle', // Keep Gradle build file
     'pom.xml', // Keep Maven Project Object Model (POM) file
     'settings.gradle', // Keep Gradle settings file
-    'build.sh', // Keep shell script for building the project
-    'build.bash', // Keep bash build script
-    'deploy.sh', // Keep shell script for deployment
-    'ci.sh', // Keep shell script for CI
     // 🔧 Other Project Files
     'Makefile.am', // Keep Automake Makefile
     'config.yaml', // Keep general config file in YAML format
@@ -68,12 +61,6 @@ export const specificFileExceptions = [
     'README.txt', // Keep documentation in text format
     'data.json', // Keep data JSON file
     'data.yml', // Keep data YAML file
-    'env.json', // Keep environment JSON file
-    'env.yml', // Keep environment YAML file
-    '.env', // Keep environment variable files
-    '.env.local', // Keep local environment variables
-    '.env.production', // Keep production environment variables
-    '.env.development', // Keep development environment variables
     // 🚧 Test-related files
     'test.config.js', // Keep test config for testing frameworks
     'test-utils.js', // Keep test utility files

package/dist/fileRules/ignoredExtensions.js

@@ -63,3 +63,13 @@ export const IGNORED_EXTENSIONS = [
     // ─── Misc Hashes & Metadata ────────────────────────────────
     '.md5', '.sha1', '.sha256', '.xlsb', '.xar',
 ];
+// ─── Knowledge Graph Ignored Extensions ─────────────────────
+export const KG_IGNORED_EXTENSIONS = [
+    // ─── Frequently ignored / useless for KG ─────────────────
+    '.json',
+    // ─── Executables / Binaries ───────────────────────────────
+    '.html', '.htm', '.xml', '.exe', '.dll', '.bin', '.iso', '.img', '.o', '.obj', '.so', '.dylib', '.a', '.lib',
+    '.cmd', '.bat', '.ps1', '.sh', '.bash', '.run', '.vbs', '.jscript',
+    // ─── Other meaningless files for KG ───────────────────────
+    '.msm', '.lock', '.lockfile', '.lck', '.db-lock', '.pid', '.sock', '.socket',
+];