Function is_draft_content

scripts/fetch_nmpa_fulltext.py:644–664

Detect if extracted text is a draft for comments (征求意见稿). Only official/final versions should be included in the knowledge base. Draft documents are identified by explicit markers in the first ~3000 chars.

(text: str)

def is_draft_content(text: str) -> bool:
    """Detect if extracted text is a draft for comments (征求意见稿).

    Only official/final versions should be included in the knowledge base.
    Draft documents are identified by explicit markers in the first ~3000 chars.
    """
    if not text:
        return False

    head = text[:3000]

    draft_patterns = [
        r'征求意见稿',
        r'征求\s*意见\s*稿',
    ]

    for pat in draft_patterns:
        if re.search(pat, head):
            return True

    return False


def validate_extracted_content(