Commit 154c3cd7 authored by Jan Reimes's avatar Jan Reimes
Browse files

♻️ refactor(config): rename TDocCrawlerConfig to ThreeGPPConfig, deduplicate path defaults 

Remove backward-compatible TDocCrawlerConfig alias now that all
consumers have been migrated.  Deduplicate _DEFAULT_* path constants
by sourcing them from CacheManager instead of duplicating in settings.
parent a575d180

src/tdoc_crawler/cli/crawl.py

+4 −4
Original line number Diff line number Diff line
@@ -46,7 +46,7 @@ from tdoc_crawler.cli.args import ( 
)

from tdoc_crawler.cli.formatting import format_output

from tdoc_crawler.cli.printing import print_spec_crawl_table, spec_crawl_to_dict

from tdoc_crawler.config import TDocCrawlerConfig

from tdoc_crawler.config import ThreeGPPConfig

from tdoc_crawler.credentials import set_credentials

from tdoc_crawler.database import MeetingDatabase, TDocDatabase

from tdoc_crawler.database.specs import SpecCrawlResult, SpecDatabase
@@ -147,7 +147,7 @@ def crawl_tdocs( 
    """

    set_verbosity(verbosity)



    crawler_config = TDocCrawlerConfig.from_settings()

    crawler_config = ThreeGPPConfig.from_settings()

    # Override cache_dir if provided (deprecated but still supported)

    if cache_dir is not None:

        crawler_config.path.cache_dir = cache_dir
@@ -297,7 +297,7 @@ def crawl_meetings( 
    set_verbosity(verbosity)

    set_credentials(eol_username, eol_password, prompt=prompt_credentials)



    crawler_config = TDocCrawlerConfig.from_settings()

    crawler_config = ThreeGPPConfig.from_settings()

    if cache_dir is not None:

        crawler_config.path.cache_dir = cache_dir

    crawler_config.ensure_paths()
@@ -401,7 +401,7 @@ def crawl_specs( 
) -> None:

    """Crawl spec metadata from configured sources."""

    set_verbosity(verbosity)

    crawler_config = TDocCrawlerConfig.from_settings()

    crawler_config = ThreeGPPConfig.from_settings()

    if cache_dir is not None:

        crawler_config.path.cache_dir = cache_dir

    crawler_config.ensure_paths()

src/tdoc_crawler/config/__init__.py

+0 −2
Original line number Diff line number Diff line
@@ -16,7 +16,6 @@ from tdoc_crawler.config.settings import ( 
    CredentialsConfig,

    HttpConfig,

    PathConfig,

    TDocCrawlerConfig,

    ThreeGPPConfig,

)

from tdoc_crawler.config.sources import (
@@ -42,7 +41,6 @@ __all__ = [ 
    "CredentialsConfig",

    "HttpConfig",

    "PathConfig",

    "TDocCrawlerConfig",

    "ThreeGPPConfig",

    "discover_config_files",

    "load_config_file",

src/tdoc_crawler/config/cache_manager.py

+5 −0
Original line number Diff line number Diff line
@@ -108,6 +108,11 @@ class CacheManager: 
        """Path to workspace registry JSON file."""

        return self._cache_dir / WORKSPACE_REGISTRY_FILENAME



    @classmethod

    def is_registered(cls) -> bool:

        """Check if a CacheManager instance has been registered."""

        return cls._instance is not None





def resolve_cache_manager() -> CacheManager:

    """Resolve the registered CacheManager instance.

src/tdoc_crawler/config/export.py

+2 −2
Original line number Diff line number Diff line
@@ -87,7 +87,7 @@ class ConfigExporter: 
        """Export to TOML with comments from field descriptions."""

        lines = ["# 3GPP Crawler Configuration", ""]

        lines.append("# This file was generated by `tdoc-crawler config init`")

        lines.append("# Default values are defined in TDocCrawlerConfig using Field(default=...)")

        lines.append("# Default values are defined in ThreeGPPConfig using Field(default=...)")

        lines.append("")



        current_section = None
@@ -125,7 +125,7 @@ class ConfigExporter: 
        lines = [

            "# 3GPP Crawler Configuration",

            "# This file was generated by `tdoc-crawler config init`",

            "# Default values are defined in TDocCrawlerConfig using Field(default=...)",

            "# Default values are defined in ThreeGPPConfig using Field(default=...)",

            "",

        ]

src/tdoc_crawler/config/settings.py

+13 −21
Original line number Diff line number Diff line
@@ -17,16 +17,16 @@ from pathlib import Path 
from pydantic import AliasChoices, Field, field_validator

from pydantic_settings import BaseSettings, SettingsConfigDict



from tdoc_crawler.config.cache_manager import (

    DEFAULT_CHECKOUT_DIRNAME,

    DEFAULT_DATABASE_FILENAME,

    DEFAULT_HTTP_CACHE_FILENAME,

    DEFAULT_WORKSPACES_DIRNAME,

)

from tdoc_crawler.config.env_vars import ConfigEnvVar

from tdoc_crawler.config.sources import discover_config_files, load_config_file, merge_configs



# Default values (not constants - just module-level defaults)

# Actual paths are resolved by CacheManager at runtime

_DEFAULT_CACHE_DIR_STR = "~/.3gpp-crawler"

_DEFAULT_DATABASE_FILENAME = "3gpp_crawler.db"

_DEFAULT_HTTP_CACHE_FILENAME = "http-cache.sqlite3"

_DEFAULT_CHECKOUT_DIRNAME = "checkout"

_DEFAULT_WORKSPACES_DIRNAME = "workspaces"





class PathConfig(BaseSettings):
@@ -44,37 +44,33 @@ class PathConfig(BaseSettings): 
        description="Root cache directory for storing downloaded files and metadata",

    )

    db_filename: str = Field(

        default=_DEFAULT_DATABASE_FILENAME,

        default=DEFAULT_DATABASE_FILENAME,

        description="SQLite database filename for storing crawl metadata",

    )

    checkout_dirname: str = Field(

        default=_DEFAULT_CHECKOUT_DIRNAME,

        default=DEFAULT_CHECKOUT_DIRNAME,

        description="Subdirectory name for checked-out documents",

    )

    workspaces_dirname: str = Field(

        default=_DEFAULT_WORKSPACES_DIRNAME,

        default=DEFAULT_WORKSPACES_DIRNAME,

        description="Subdirectory name for workspace data (sources, wiki)",

    )



    @property

    def db_file(self) -> Path:

        """Path to the SQLite database file."""

        return self.cache_dir / _DEFAULT_DATABASE_FILENAME

        return self.cache_dir / DEFAULT_DATABASE_FILENAME



    @property

    def http_cache_file(self) -> Path:

        """Path to the HTTP cache database file."""

        return self.cache_dir / _DEFAULT_HTTP_CACHE_FILENAME

        return self.cache_dir / DEFAULT_HTTP_CACHE_FILENAME



    @property

    def checkout_dir(self) -> Path:

        """Path to the checkout directory for documents."""

        return self.cache_dir / _DEFAULT_CHECKOUT_DIRNAME

        return self.cache_dir / DEFAULT_CHECKOUT_DIRNAME



    @property

    def workspaces_dir(self) -> Path:

        """Path to the workspaces directory (separate from document checkouts)."""

        return self.cache_dir / _DEFAULT_WORKSPACES_DIRNAME

        return self.cache_dir / DEFAULT_WORKSPACES_DIRNAME



    @field_validator("cache_dir", mode="before")

    @classmethod
@@ -327,14 +323,10 @@ class ThreeGPPConfig(BaseSettings): 
        return cls(**merged)





# Backward-compatible alias — remove after all consumers are updated

TDocCrawlerConfig = ThreeGPPConfig



__all__ = [

    "CrawlConfig",

    "CredentialsConfig",

    "HttpConfig",

    "PathConfig",

    "TDocCrawlerConfig",

    "ThreeGPPConfig",

]