Commit f69b9c33 authored by Jan Reimes's avatar Jan Reimes
Browse files

refactor(02-05): consolidate normalize_release_label function 

- Moved normalize_release_label from metadata.py to core normalization.py
- Updated metadata.py to import from core normalization module
- Single source of truth for release label normalization
- No breaking changes - function behavior remains identical
parent 50f0f30f

packages/3gpp-ai/threegpp_ai/lightrag/metadata.py

+4 −21
Original line number Diff line number Diff line
@@ -7,27 +7,10 @@ structured 3GPP metadata to document text before LightRAG insertion. 
from __future__ import annotations



from pydantic import BaseModel, Field

from tdoc_crawler.utils.normalization import normalize_release, normalize_tdoc_id





def normalize_release_label(release: str | None) -> str | None:

    """Normalize release strings to a stable label format for metadata display."""

    normalized_label: str | None = None

    if release is not None:

        cleaned = release.strip()

        if cleaned:

            normalized_label = cleaned

            try:

                release_type, normalized_value, _ = normalize_release(cleaned)

                if release_type == "latest":

                    normalized_label = "latest"

                elif release_type == "all":

                    normalized_label = "all"

                elif normalized_value is not None:

                    normalized_label = f"Rel-{normalized_value.lstrip('-')}"

            except ValueError:

                pass

    return normalized_label

from tdoc_crawler.utils.normalization import (

    normalize_release_label,

    normalize_tdoc_id,

)





class RAGMetadata(BaseModel):

src/tdoc_crawler/utils/normalization.py

+27 −0
Original line number Diff line number Diff line
@@ -10,6 +10,33 @@ _RANGE_SPLIT_PATTERN = re.compile(r"\s*([-:])\s*") 
_RELEASE_PREFIX_PATTERN = re.compile(r"^(?:v|rel|rel[-])", re.IGNORECASE)





def normalize_release_label(release: str | None) -> str | None:

    """Normalize release strings to a stable label format for metadata display.



    Args:

        release: Raw release string (e.g., "18", "Rel-18", "latest")



    Returns:

        Normalized label (e.g., "Rel-18", "latest", "all") or None if input is None/empty

    """

    normalized_label: str | None = None

    if release is not None:

        cleaned = release.strip()

        if cleaned:

            normalized_label = cleaned

            try:

                release_type, normalized_value, _ = normalize_release(cleaned)

                if release_type == "latest":

                    normalized_label = "latest"

                elif release_type == "all":

                    normalized_label = "all"

                elif normalized_value is not None:

                    normalized_label = f"Rel-{normalized_value.lstrip('-')}"

            except ValueError:

                pass

    return normalized_label





class NormalizationError(ValueError):

    """Raised when spec identifier normalization or expansion fails."""