codegraphcontext 0.6.6__tar.gz → 0.6.8__tar.gz

{codegraphcontext-0.6.6/src/codegraphcontext.egg-info → codegraphcontext-0.6.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: codegraphcontext
-Version: 0.6.6
+Version: 0.6.8
 Summary: An MCP server that indexes local code into a graph database to provide context to AI assistants.
 Author-email: Shashank Shekhar Singh <shashankshekharsingh1205@gmail.com>
 License: MIT License

{codegraphcontext-0.6.6 → codegraphcontext-0.6.8}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "codegraphcontext"
-version = "0.6.6"
+version = "0.6.8"
 description = "An MCP server that indexes local code into a graph database to provide context to AI assistants."
 authors = [{ name = "Shashank Shekhar Singh", email = "shashankshekharsingh1205@gmail.com" }]
 readme = "README.md"

{codegraphcontext-0.6.6 → codegraphcontext-0.6.8}/src/codegraphcontext/cli/cli_helpers.py

@@ -351,7 +351,7 @@ def index_helper(path: str, context: Optional[str] = None, no_progress: bool = F
                 file_count = record["file_count"] if record else 0
                 
                 if file_count > 0:
-                    expected = graph_builder.estimate_processing_time(path_obj) if path_obj.is_dir() else None
+                    expected = graph_builder.estimate_processing_time(path_obj, cgcignore_path=ctx.cgcignore_path) if path_obj.is_dir() else None
                     expected_file_count = expected[0] if expected else None
                     if expected_file_count is None or file_count >= expected_file_count:
                         console.print(f"[yellow]Repository '{path}' is already indexed with {file_count} files. Skipping.[/yellow]")
@@ -361,6 +361,32 @@ def index_helper(path: str, context: Optional[str] = None, no_progress: bool = F
                     console.print(
                         f"[yellow]Repository '{path}' has only {file_count} of {expected_file_count} files indexed. Continuing.[/yellow]"
                     )
+                    # Name the gap: without this, a pipeline that can never
+                    # produce nodes for some files loops on the message above
+                    # forever with nothing actionable (#1673).
+                    try:
+                        from codegraphcontext.tools.indexing.discovery import discover_files_to_index
+                        discovered, _ = discover_files_to_index(
+                            path_obj,
+                            cgcignore_path=ctx.cgcignore_path,
+                            supported_extensions=set(graph_builder.parsers.keys()),
+                        )
+                        with db_manager.get_driver().session() as s2:
+                            res2 = s2.run(
+                                "MATCH (r:Repository {path: $path})-[:CONTAINS*]->(f:File) RETURN DISTINCT f.path AS p",
+                                path=repo_path_str,
+                            )
+                            have = {row["p"] for row in res2 if row.get("p")}
+                        missing = [
+                            str(f) for f in discovered
+                            if str(f.resolve()) not in have and f.resolve().as_posix() not in have
+                        ]
+                        if missing:
+                            shown = ", ".join(Path(m).name for m in missing[:5])
+                            more = f" (+{len(missing) - 5} more)" if len(missing) > 5 else ""
+                            console.print(f"[dim]Missing: {shown}{more}[/dim]")
+                    except Exception:
+                        pass
                 else:
                     console.print(f"[yellow]Repository '{path}' exists but has no files (likely interrupted). Re-indexing...[/yellow]")
         except Exception as e:

{codegraphcontext-0.6.6 → codegraphcontext-0.6.8}/src/codegraphcontext/tools/graph_builder.py

@@ -534,7 +534,7 @@ class GraphBuilder:
             # still reports "Successfully finished indexing".
             return {"path": str(path), "error": str(e), "parse_failed": True}
 
-    def estimate_processing_time(self, path: Path) -> Optional[Tuple[int, float]]:
+    def estimate_processing_time(self, path: Path, cgcignore_path: Optional[str] = None) -> Optional[Tuple[int, float]]:
         """Estimate the time required to index a repository.
 
         Parameters
@@ -550,8 +550,13 @@ class GraphBuilder:
         try:
             from codegraphcontext.tools.indexing.discovery import discover_files_to_index
             supported_extensions = set(self.parsers.keys())
+            # cgcignore_path must match what the indexing pipelines use:
+            # counting files a context-specific ignore excludes made the
+            # resume check expect more files than indexing will ever produce
+            # ("only N of M files indexed" on every run, #1673).
             files, _ = discover_files_to_index(
                 path,
+                cgcignore_path=cgcignore_path,
                 supported_extensions=supported_extensions,
             )
             total_files = len(files)
@@ -566,6 +571,9 @@ class GraphBuilder:
     ):
         from . import scip_indexer
 
+        # Reset here too: without it a SCIP run left (and printed) the
+        # PREVIOUS Tree-sitter run's summary from the same process (#1673).
+        self.last_index_summary = {}
         await run_scip_index_async(
             path,
             is_dependency,
@@ -577,6 +585,7 @@ class GraphBuilder:
             self.get_parser,
             scip_indexer,
             cgcignore_path,
+            index_summary=self.last_index_summary,
         )
 
     def _name_from_symbol(self, symbol: str) -> str:

{codegraphcontext-0.6.6 → codegraphcontext-0.6.8}/src/codegraphcontext/tools/indexing/schema.py

@@ -166,6 +166,22 @@ def create_graph_schema(driver: Any, db_manager: Any) -> None:
                         f"CREATE INDEX {_label.lower()}_name_path IF NOT EXISTS "
                         f"FOR ({_var}:{_label}) ON ({_var}.name, {_var}.path)"
                     )
+                # EnumMember, Object, Mixin, Extension and Parameter had index
+                # statements only in the is_falkordb branch, so on Neo4j the
+                # writer's `MATCH (m:EnumMember {name: …, path: …})` (and the
+                # per-label CONTAINS linking generally) was an unavoidable
+                # label scan — found while auditing #1653's lookup sites.
+                for _extra_label, _extra_var in (
+                    ("EnumMember", "em"),
+                    ("Object", "o"),
+                    ("Mixin", "mx"),
+                    ("Extension", "ex"),
+                    ("Parameter", "p"),
+                ):
+                    ddl.run(
+                        f"CREATE INDEX {_extra_label.lower()}_name_path IF NOT EXISTS "
+                        f"FOR ({_extra_var}:{_extra_label}) ON ({_extra_var}.name, {_extra_var}.path)"
+                    )
 
             ddl.run("CREATE INDEX function_lang IF NOT EXISTS FOR (f:Function) ON (f.lang)")
             ddl.run("CREATE INDEX class_lang IF NOT EXISTS FOR (c:Class) ON (c.lang)")

{codegraphcontext-0.6.6 → codegraphcontext-0.6.8}/src/codegraphcontext/tools/indexing/scip_pipeline.py

@@ -42,6 +42,7 @@ async def run_scip_index_async(
     get_parser: Callable[[str], Any],
     scip_indexer_mod: Any,
     cgcignore_path: Optional[str] = None,
+    index_summary: Optional[dict] = None,
 ) -> None:
     """Run SCIP CLI, write graph, supplement with Tree-sitter, write SCIP CALLS edges."""
     ScipIndexer = scip_indexer_mod.ScipIndexer
@@ -171,7 +172,18 @@ async def run_scip_index_async(
                 except Exception as e:
                     debug_log(f"Tree-sitter supplement failed for {abs_path_str}: {e}")
 
-            writer.add_file_to_graph(file_data, repo_name, imports_map)
+            try:
+                writer.add_file_to_graph(file_data, repo_name, imports_map)
+            except Exception as e:
+                # One unwritable file must not abort the run, and it must
+                # still be accounted for — otherwise the graph stays short of
+                # the discovery census and `cgc index` reports "only N of M
+                # files indexed. Continuing." on every run, forever (#1673).
+                warning_logger(f"SCIP file write failed for {abs_path_str}: {e}; recording minimal node")
+                try:
+                    writer.add_minimal_file_node(Path(abs_path_str), index_root, is_dependency)
+                except Exception as e2:
+                    debug_log(f"Minimal node fallback also failed for {abs_path_str}: {e2}")
 
             processed += 1
             if job_id:
@@ -236,6 +248,14 @@ async def run_scip_index_async(
             try:
                 ts_data = ts_parser.parse(repo_file, is_dependency, index_source=True)
                 if "error" in ts_data:
+                    # Same accounting rule as the Tree-sitter pipeline: a file
+                    # that failed to parse still gets a minimal File node so
+                    # the graph count converges with discovery (#1673).
+                    await asyncio.to_thread(
+                        writer.add_minimal_file_node, repo_file, index_root, is_dependency
+                    )
+                    minimal_nodes += 1
+                    processed += 1
                     continue
                 ts_data["repo_path"] = str(index_root)
                 ts_data.setdefault("function_calls_scip", [])
@@ -251,6 +271,14 @@ async def run_scip_index_async(
                     )
             except Exception as e:
                 debug_log(f"Tree-sitter supplement (non-SCIP file) failed for {abs_str}: {e}")
+                try:
+                    await asyncio.to_thread(
+                        writer.add_minimal_file_node, repo_file, index_root, is_dependency
+                    )
+                    minimal_nodes += 1
+                    processed += 1
+                except Exception as e2:
+                    debug_log(f"Minimal node fallback also failed for {abs_str}: {e2}")
         if supplemented:
             # Re-run pre_scan with the expanded file list so imports_map is complete
             all_paths = [Path(p) for p in files_data.keys() if Path(p).exists()]
@@ -284,6 +312,30 @@ async def run_scip_index_async(
 
         writer.write_scip_call_edges(files_data, name_from_symbol)
 
+        # CLI-facing summary — the SCIP path never populated one, so
+        # `cgc index` finished silently with no file/function/class report
+        # at all (#1673).
+        if index_summary is not None:
+            from collections import Counter
+            extension_counts = Counter()
+            for fp in files_data.keys():
+                suffix = Path(fp).suffix or Path(fp).name
+                extension_counts[suffix] += 1
+            index_summary.clear()
+            index_summary.update({
+                "total_scanned_files": len(files_data) + minimal_nodes,
+                "files_by_extension": dict(extension_counts),
+                "function_nodes": sum(
+                    1 for fd in all_file_data for fn in fd.get("functions", [])
+                    if fn.get("name") != "<module>"
+                ),
+                "class_nodes": sum(len(fd.get("classes", [])) for fd in all_file_data),
+                "call_edges": sum(len(g) for g in resolved_calls),
+                "serialization_seconds": 0.0,
+                "failed_files": 0,
+                "minimal_file_nodes": minimal_nodes,
+            })
+
         if job_id:
             job_manager.update_job(job_id, status=JobStatus.COMPLETED, end_time=datetime.now())
 

{codegraphcontext-0.6.6 → codegraphcontext-0.6.8}/src/codegraphcontext/tools/languages/kotlin.py

@@ -169,7 +169,10 @@ class KotlinTreeSitterParser:
                 return (
                     self._get_node_text(name_node) if name_node else None,
                     curr.type,
-                    curr.start_point[0] + 1,
+                    # The name node's line, not the node's: an annotated
+                    # declaration starts at its annotation, and consumers join
+                    # this against the function's line_number (#1660).
+                    (name_node.start_point[0] + 1) if name_node else (curr.start_point[0] + 1),
                 )
             if curr.type in ("class_declaration", "interface_declaration", "object_declaration"):
                 for child in curr.children:
@@ -177,7 +180,7 @@ class KotlinTreeSitterParser:
                          return (
                             self._get_node_text(child),
                             curr.type,
-                            curr.start_point[0] + 1,
+                            child.start_point[0] + 1,
                         )
                 # Check for secondary constructors
                 if curr.type == "secondary_constructor":
@@ -1330,6 +1333,12 @@ class KotlinTreeSitterParser:
                             
                     if name_node:
                         func_name = self._get_node_text(name_node)
+                        # Annotations/modifiers are children of the declaration
+                        # node, so node.start_point is the ANNOTATION line for
+                        # any annotated function. The name node sits on the
+                        # declaration proper — the established convention
+                        # (python decorated_definition, csharp #1659, #1660).
+                        start_line = name_node.start_point[0] + 1
                         
                         params_node = None
                         for child in node.children:
@@ -1457,6 +1466,10 @@ class KotlinTreeSitterParser:
                         for child in node.children:
                             if child.type in ("type_identifier", "simple_identifier"):
                                 class_name = self._get_node_text(child)
+                                # Same convention as functions: an annotated
+                                # class node starts at its annotation line;
+                                # report the declaration line (#1660).
+                                start_line = child.start_point[0] + 1
                                 break
                             
                     source_text = self._get_node_text(node)

{codegraphcontext-0.6.6 → codegraphcontext-0.6.8}/src/codegraphcontext/tools/languages/php.py

@@ -162,7 +162,10 @@ class PhpTreeSitterParser:
                 return (
                     self._get_node_text(name_node) if name_node else None,
                     curr.type,
-                    curr.start_point[0] + 1,
+                    # Name-node line: attributed declarations start at their
+                    # #[Attribute], and consumers join this against the
+                    # declaration's line_number (#1660).
+                    (name_node.start_point[0] + 1) if name_node else (curr.start_point[0] + 1),
                 )
             curr = curr.parent
         return None, None, None
@@ -189,7 +192,7 @@ class PhpTreeSitterParser:
             if curr.type in ("class_declaration", "interface_declaration", "trait_declaration", "enum_declaration"):
                 name_node = curr.child_by_field_name("name")
                 name = self._get_node_text(name_node) if name_node else None
-                return (name, curr.start_point[0] + 1) if name else None
+                return (name, name_node.start_point[0] + 1) if name else None
             curr = curr.parent
         return None
 
@@ -245,6 +248,11 @@ class PhpTreeSitterParser:
                     name_node = node.child_by_field_name("name")
                     if name_node:
                         func_name = self._get_node_text(name_node)
+                        # #[Attribute] lines are part of the declaration node,
+                        # so node.start_point is the attribute line; report the
+                        # declaration proper (python/csharp/kotlin convention,
+                        # #1660).
+                        start_line = name_node.start_point[0] + 1
                         
                         params_node = node.child_by_field_name("parameters")
                         parameters = []
@@ -321,6 +329,9 @@ class PhpTreeSitterParser:
                     name_node = node.child_by_field_name("name")
                     if name_node:
                         type_name = self._get_node_text(name_node)
+                        # Same convention as functions: attributes are children
+                        # of the declaration node (#1660).
+                        start_line = name_node.start_point[0] + 1
                         source_text = self._get_node_text(node)
                         
                         bases = []

{codegraphcontext-0.6.6 → codegraphcontext-0.6.8}/src/codegraphcontext/tools/languages/ruby.py

@@ -49,6 +49,14 @@ RUBY_QUERIES = {
             left: (instance_variable) @name
             right: (_) @value
         )
+        (assignment
+            left: (class_variable) @name
+            right: (_) @value
+        )
+        (assignment
+            left: (global_variable) @name
+            right: (_) @value
+        )
     """,
     "comments": """
         (comment) @comment
@@ -537,7 +545,12 @@ class RubyTreeSitterParser:
                 while current and current.type != 'assignment':
                     current = current.parent
                 if current:
-                    assignment_id = id(current)
+                    # id() of a tree-sitter node is a fresh wrapper each
+                    # access (n1 == n2 while id(n1) != id(n2)), which split
+                    # the @name/@value captures of ONE assignment into two
+                    # buckets — value was always None (#1660). Key on the
+                    # node's span instead.
+                    assignment_id = (current.start_byte, current.end_byte)
                     if assignment_id not in captures_by_assignment:
                         captures_by_assignment[assignment_id] = {'node': current, 'name': None, 'value': None}
                     captures_by_assignment[assignment_id]['name'] = self._get_node_text(node)
@@ -547,7 +560,7 @@ class RubyTreeSitterParser:
                 while current and current.type != 'assignment':
                     current = current.parent
                 if current:
-                    assignment_id = id(current)
+                    assignment_id = (current.start_byte, current.end_byte)
                     if assignment_id not in captures_by_assignment:
                         captures_by_assignment[assignment_id] = {'node': current, 'name': None, 'value': None}
                     captures_by_assignment[assignment_id]['value'] = self._get_node_text(node)
@@ -560,10 +573,12 @@ class RubyTreeSitterParser:
             if name:
                 # Determine variable type based on name prefix
                 var_type = "local"
-                if name.startswith("@"):
-                    var_type = "instance"
-                elif name.startswith("@@"):
+                # @@ must be tested before @: startswith("@") matches
+                # "@@count" too, which left the class branch dead (#1660).
+                if name.startswith("@@"):
                     var_type = "class"
+                elif name.startswith("@"):
+                    var_type = "instance"
                 elif name.startswith("$"):
                     var_type = "global"
 

{codegraphcontext-0.6.6 → codegraphcontext-0.6.8/src/codegraphcontext.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: codegraphcontext
-Version: 0.6.6
+Version: 0.6.8
 Summary: An MCP server that indexes local code into a graph database to provide context to AI assistants.
 Author-email: Shashank Shekhar Singh <shashankshekharsingh1205@gmail.com>
 License: MIT License