fix(filesystem): require real grep source lines

Do not emit source-file grep fallback candidates unless an actual source line matches the query.

pageindex/filesystem/commands.py

@@ -1466,6 +1466,8 @@ class PIFSCommandExecutor:
             if direct_only and self._folder_path_for_source_path(file_row["source_path"]) != folder_path:
                 continue
             line_number, text = self._first_matching_source_line(path, query)
+            if line_number is None:
+                continue
             hits.append(
                 {
                     "file_ref": file_row["file_ref"],
@@ -1560,15 +1562,15 @@ class PIFSCommandExecutor:
                     break
         return filtered
 
-    def _first_matching_source_line(self, path: Path, query: str) -> tuple[int, str]:
+    def _first_matching_source_line(self, path: Path, query: str) -> tuple[int | None, str]:
         try:
             lines = path.read_text(encoding="utf-8", errors="ignore").splitlines()
         except OSError:
-            return 1, ""
+            return None, ""
         for line_number, line in enumerate(lines, 1):
             if self._line_matches(line, query):
                 return line_number, self._compact_text(line, max_chars=220)
-        return 1, self._compact_text(lines[0], max_chars=220) if lines else ""
+        return None, ""
 
     def _source_root(self) -> Path | None:
         with self.filesystem.store.connect() as conn:

tests/test_pageindex_filesystem_scope.py

@@ -87,6 +87,40 @@ def test_semantic_search_scope_filters_explicit_source_type_facets():
     ) == {}
 
 
+def test_grep_source_file_requires_terms_on_same_line(tmp_path):
+    from pageindex.filesystem import PIFSCommandExecutor, PageIndexFileSystem
+
+    source_dir = tmp_path / "source" / "documents"
+    source_dir.mkdir(parents=True)
+    source = source_dir / "split.json"
+    source.write_text(
+        '{\n  "first": "alpha evidence lives here",\n'
+        '  "second": "omega evidence lives there"\n}\n',
+        encoding="utf-8",
+    )
+    filesystem = PageIndexFileSystem(workspace=tmp_path / "workspace")
+    filesystem.register_file(
+        storage_uri=str(source),
+        source_path="documents/split.json",
+        folder_path="/documents",
+        external_id="doc_split_terms",
+        title="Split source terms",
+        content="registered artifact without the searched tokens",
+    )
+    executor = PIFSCommandExecutor(filesystem, json_output=True)
+
+    result = json.loads(executor.execute('grep -R "alpha omega" /documents'))
+
+    assert result["data"]["mode"] == "files"
+    assert result["data"]["data"] == []
+
+    matched = json.loads(executor.execute('grep -R "alpha evidence" /documents'))
+
+    assert matched["data"]["data"][0]["external_id"] == "doc_split_terms"
+    assert matched["data"]["data"][0]["line"] == 2
+    assert "alpha evidence" in matched["data"]["data"][0]["text"]
+
+
 def test_existing_summary_projection_index_configures_retrieval_backend(tmp_path, monkeypatch):
     from pageindex.filesystem import PageIndexFileSystem
     from pageindex.filesystem.semantic_index import SemanticIndexRecord, SQLiteVecSemanticIndex