Method _parse_notebook_cells

code_review_graph/parser.py:1372–1501

Parse notebook cells grouped by language. Args: path: Notebook file path. cells: List of CellInfo with index, language, and source. default_language: Default language for the File node.

(
        self,
        path: Path,
        cells: list[CellInfo],
        default_language: str,
    )

        return self._parse_notebook_cells(path, cells, kernel_lang)

    def _parse_notebook_cells(
        self,
        path: Path,
        cells: list[CellInfo],
        default_language: str,
    ) -> tuple[list[NodeInfo], list[EdgeInfo]]:
        """Parse notebook cells grouped by language.

        Args:
            path: Notebook file path.
            cells: List of CellInfo with index, language, and source.
            default_language: Default language for the File node.
        """
        file_path_str = str(path)
        test_file = _is_test_file(file_path_str)

        # Group cells by language
        lang_cells: dict[str, list[CellInfo]] = {}
        for cell in cells:
            lang_cells.setdefault(cell.language, []).append(cell)

        all_nodes: list[NodeInfo] = []
        all_edges: list[EdgeInfo] = []

        # Track offsets per language for cell_index tagging.
        # Each language group is parsed independently by Tree-sitter,
        # so line numbers restart at 1 for each group.
        all_cell_offsets: list[tuple[int, int, int]] = []
        max_line = 1

        for lang, lang_group in lang_cells.items():
            if lang == "sql":
                # SQL: regex-based table extraction
                for cell in lang_group:
                    for match in _SQL_TABLE_RE.finditer(cell.source):
                        table_name = match.group(1).replace("`", "")
                        all_edges.append(EdgeInfo(
                            kind="IMPORTS_FROM",
                            source=file_path_str,
                            target=table_name,
                            file_path=file_path_str,
                            line=1,
                        ))
                continue

            if lang not in ("python", "r"):
                continue

            ts_parser = self._get_parser(lang)
            if not ts_parser:
                continue

            # Concatenate cells of this language.
            # Line numbers start at 1 for each language group because
            # Tree-sitter parses each concatenation independently.
            code_chunks: list[str] = []
            cell_offsets: list[tuple[int, int, int]] = []
            current_line = 1