Class PipelineController

onecite/core.py:124–183

Runs the 4-stage pipeline: Parse -> Identify -> Enrich -> Format.

class PipelineController:
    """Runs the 4-stage pipeline: Parse -> Identify -> Enrich -> Format."""
    
    def __init__(self, use_google_scholar: bool = False):
        """Initialize the pipeline controller.

        Args:
            use_google_scholar: If ``True``, enable Google Scholar as an
                additional data source for identification and enrichment.
                Requires the optional ``scholarly`` package.  Defaults to
                ``False``.
        """
        self.logger = logging.getLogger(__name__)
        from .pipeline import ParserModule, IdentifierModule, EnricherModule, FormatterModule
        
        self.template_loader = TemplateLoader()
        self.parser = ParserModule()
        self.identifier = IdentifierModule(use_google_scholar=use_google_scholar)
        self.enricher = EnricherModule(use_google_scholar=use_google_scholar)
        self.formatter = FormatterModule()
    
    def process(self, input_content: str, input_type: str, template_name: str,
                output_format: str, interactive_callback: Callable[[List[Dict]], int]) -> Dict[str, Any]:
        """Run all four pipeline stages and return results with a report.

        Args:
            input_content: The raw text or BibTeX content to process.
            input_type: Format of *input_content* — ``"txt"`` for
                plain-text references or ``"bib"`` for BibTeX.
            template_name: Name of the YAML template that controls
                which fields are collected (e.g.
                ``"journal_article_full"``).
            output_format: Output format. Only ``"bibtex"`` is
                supported; any other value raises ``FormatError``.
            interactive_callback: A callable that receives a list of
                candidate dictionaries and returns the index of the
                selected candidate.

        Returns:
            A dictionary with two keys:

            * ``results`` — a list of formatted citation strings.
            * ``report`` — a dict containing ``total``, ``succeeded``,
              and ``failed_entries``.
        """
        self.logger.info("Starting OneCite processing pipeline")
        
        try:
            template = self.template_loader.load_template(template_name)
            raw_entries = self.parser.parse(input_content, input_type)
            identified_entries = self.identifier.identify(raw_entries, interactive_callback)
            completed_entries = self.enricher.enrich(identified_entries, template, raw_entries)
            result = self.formatter.format(completed_entries, output_format)
            
            self.logger.info("OneCite processing pipeline completed")
            return result
            
        except Exception as e: