INNER CODE UNIT · Python

_structural_nonheading

Laurent00TT/PharosRAG · src/chunker/core.py:136

def _structural_nonheading(txt):
    """Text-shape rejects shared by the leveling pass and the reset-aware pre-pass."""
    return bool(TOC_RE.search(txt) or TOC_TAIL_RE.match(txt)
                or len(txt) > MAX_HEADING_LEN or BULLET_RE.match(txt))


def _leading_number(txt):
    """Parse a leading section/list number -> ('dotted', depth) | ('bare', value) | None.
    Year-leading (2026 ...) is treated as not-a-number (review F1/F4)."""
    m = NUM_RE.match(txt)
    if not m:
        return None
    tok = m.group(1)
    if YEAR_RE.match(tok.split(".")[0]):
        return None
    if "." in tok:
        return ("dotted", tok.count(".") + 1)
    if tok.isdigit():

View source record →