Fix repository indexing accuracy and model-context relevance

backend/services/analyzer.py

@@ -14,7 +14,12 @@ def analyze(root):
     paths={e["path"].lower() for e in entries}; names={Path(e["path"]).name.lower() for e in entries}
     signals={
       "readme":any(n.startswith("readme") for n in names),
-      "tests":any("/test" in p or p.startswith("test") or "/tests/" in p or p.startswith("tests/") for p in paths),
+      "tests":any(
+          any(part in {"test", "tests", "__tests__"} for part in p.split("/")[:-1])
+          or Path(p).name.startswith(("test_", "test."))
+          or Path(p).name.endswith(("_test.py", "_test.go", ".test.js", ".spec.js", ".test.ts", ".spec.ts", ".test.tsx", ".spec.tsx"))
+          for p in paths
+      ),
       "gitignore":".gitignore" in names,
       "manifest":any(n in names for n in ["package.json","requirements.txt","pyproject.toml","pom.xml","cargo.toml","go.mod","composer.json","gemfile"]),
       "ci":any(p.startswith(".github/workflows/") for p in paths),
@@ -70,10 +75,10 @@ def build_context(root,q,max_files=8,max_chars=18000):
     tokens={x for x in re.findall(r"[\w-]+", q.casefold()) if len(x)>2}; ranked=[]
     for p in iter_files(root):
         text=read_text(p); rel=str(p.relative_to(root)).replace("\\","/")
-        low=(rel+"\n"+text[:30000]).casefold()
+        low=(rel+"\n"+text[:4000]).casefold()
         matched=sum(t in low for t in tokens)
         bonus=sum(k in rel.casefold() for k in ["readme","app.","main.","index.","route","api"])
-        if matched or bonus: ranked.append((matched,bonus,rel,text))
+        if matched or (not q.strip() and bonus): ranked.append((matched,bonus,rel,text))
     ranked.sort(key=lambda x: (-x[0], -x[1], x[2].casefold(), x[2])); out=""; used=0; selected=0
     for _,_,rel,text in ranked:
         block=f"\n--- FILE: {rel} ---\n{text[:4000]}\n"

backend/services/github.py

@@ -427,7 +427,7 @@ def iter_files(root: Path):
         try: size = p.stat().st_size
         except OSError: continue
         if size > 300_000: continue
-        if p.suffix.lower() in TEXT_EXTENSIONS or p.name.lower() in {"dockerfile","makefile",".gitignore"}:
+        if p.suffix.lower() in TEXT_EXTENSIONS or p.name.lower().endswith(".env.example") or p.name.lower() in {"dockerfile","makefile",".gitignore"}:
             yield p
 
 def read_text(path: Path): return path.read_text(encoding="utf-8", errors="replace")

backend/tests/test_analyzer_edge_cases.py

@@ -103,5 +103,22 @@ def test_context_budget_counts_included_not_considered_files(self):
         self.assertNotIn("--- FILE: a_large.py", context)
 
 
+    def test_context_selected_file_exposes_the_query_match(self):
+        self.write("a_hidden.py", "x" * 5000 + "needle_for_context")
+        self.write("z_visible.py", "needle_for_context = 42")
+        context = build_context(self.root, "needle_for_context", max_files=1)
+        self.assertIn("z_visible.py", context)
+        self.assertIn("needle_for_context = 42", context)
+        self.assertNotIn("a_hidden.py", context)
+
+
+    def test_unmatched_question_does_not_return_unrelated_readme(self):
+        self.write("README.md", "setup notes without the requested identifier")
+        self.write("app.py", "def start(): pass")
+        context = build_context(self.root, "nonexistent_unique_identifier")
+        self.assertIn("No strongly matching files", context)
+        self.assertNotIn("setup notes", context)
+
+
 if __name__ == "__main__":
     unittest.main()

backend/tests/test_repository_safety.py

@@ -192,6 +192,30 @@ def test_zip_slip_is_refused_without_writing_outside(self):
                     github.extract_zip(z)
                 self.assertFalse((self.root / "escape.py").exists())
 
+    def test_environment_examples_are_indexed_without_loading_env_secrets(self):
+        imported = self.repos / ("a" * 12)
+        imported.mkdir()
+        (imported / ".env.example").write_text("SAMPLE_OPTION=placeholder")
+        (imported / "service.env.example").write_text("SERVICE_OPTION=example")
+        (imported / ".env").write_text("REAL_PRIVATE_SETTING=secret")
+        listed = {p.name for p in github.iter_files(imported)}
+        self.assertIn(".env.example", listed)
+        self.assertIn("service.env.example", listed)
+        self.assertNotIn(".env", listed)
+        self.assertIn("SAMPLE_OPTION", build_context(imported, "SAMPLE_OPTION"))
+        self.assertNotIn("REAL_PRIVATE_SETTING", str(analyze(imported)))
+
+
+    def test_test_signal_ignores_unrelated_source_names(self):
+        imported = self.repos / ("b" * 12)
+        imported.mkdir()
+        (imported / "src").mkdir()
+        (imported / "src" / "testimonials.py").write_text("def describe(): pass")
+        (imported / "src" / "contest.py").write_text("def compete(): pass")
+        self.assertFalse(analyze(imported)["signals"]["tests"])
+        (imported / "src" / "feature.test.ts").write_text("it('works', () => {})")
+        self.assertTrue(analyze(imported)["signals"]["tests"])
+
 
 if __name__ == "__main__":
     unittest.main()