app/controllers/repository_controller.py

from app.core.config import REPOSITORIES_DIR

from app.services.github_service import GithubService
from app.services.metadata_service import MetadataService
from app.services.parser_service import ParserService
from app.services.index_service import IndexService
from app.services.graph_service import GraphService
from app.services.resolver_service import ResolverService
from app.services.embedding_service import EmbeddingService

from app.utils.file_utils import get_python_files

from app.core.constants import (
    METADATA_FILE,
    CODE_INDEX_FILE,
    RESOLVED_CODE_FILE,
)


class RepositoryController:

    def __init__(self):

        self.github = GithubService()

        self.metadata = MetadataService()

        self.parser = ParserService()

        self.index = IndexService()

        self.resolver = ResolverService()

        self.graph_service = GraphService()

        self.embedding_service = EmbeddingService()

    def clone(
        self,
        url: str,
    ):

        clone_result = self.github.clone(
            url,
            REPOSITORIES_DIR,
        )

        repository_id = clone_result["repository_id"]

        repo_path = clone_result["repo_path"]

        metadata = self.metadata.extract(
            repo_path,
        )

        self.index.save(
            repo_path,
            METADATA_FILE,
            metadata,
        )

        parsed_files = []

        for file in get_python_files(repo_path):

            parsed_files.append(
                self.parser.parse_file(
                    file,
                )
            )

        self.index.save(
            repo_path,
            CODE_INDEX_FILE,
            parsed_files,
        )

        code_index = [
            item.model_dump()
            for item in parsed_files
        ]

        resolved = self.resolver.resolve(
            code_index,
        )

        self.index.save(
            repo_path,
            RESOLVED_CODE_FILE,
            resolved,
        )

        print("Building graph...")

        self.graph_service.build(
            repository_id=repository_id,
            resolved=resolved,
        )

        print("Building embeddings...")

        self.embedding_service.build(
            repository_id=repository_id,
            resolved=resolved,
        )

        print("Repository indexed successfully.")

        metadata.repository_id = repository_id

        return metadata