Demo: form answers to an AI-written PDF report

_NUMBER = re.compile(r"\d+(?:\.\d+)?")
# Figures written as words slip past the digit check, so they are not allowed at all.
# "one" is left out: it is everyday English ("one cause", "one at a time"), not a figure.
_NUMBER_WORD = re.compile(
    r"\b(two|three|four|five|six|seven|eight|nine|ten|eleven|twelve|thirteen|fourteen|fifteen|"
    r"sixteen|seventeen|eighteen|nineteen|twenty|thirty|forty|fifty|sixty|seventy|eighty|ninety|"
    r"hundred|thousand|million|billion|half|halve|halves|quarter|third|double|doubles|twice|triple|"
    r"thrice|dozen)\b", re.IGNORECASE)


def allowed_numbers(sub: dict, blocks: list[dict]) -> set[str]:
    """Every number the AI may use: the answers, plus the selected blocks' values and text."""
    allowed = set()
    sources = [str(v) for v in sub["values"].values()]
    for b in blocks:
        sources += [fmt_num(b["value"]), b["title"], b["text"]]
    for text in sources:
        allowed.update(fmt_num(n) for n in _NUMBER.findall(text))
    return allowed


def ai_text_fields(reply: dict):
    yield "summary", reply.get("summary", "")
    yield "links_between_issues", reply.get("links_between_issues", "")
    for i, p in enumerate(reply.get("priorities", []), start=1):
        yield f"priorities[{i}].headline", p.get("headline", "")
        yield f"priorities[{i}].why", p.get("why", "")


def validate_ai(reply: dict, sub: dict, blocks: list[dict]) -> list[str]:
    problems = []
    for key, typ in (("summary", str), ("priorities", list),
                     ("links_between_issues", str), ("chosen_block_ids", list)):
        if not isinstance(reply.get(key), typ):
            problems.append(f"'{key}' is missing or not a {typ.__name__}")
    if problems:
        return problems

    selected = {b["id"] for b in blocks}
    for bid in reply["chosen_block_ids"]:
        if bid not in selected:
            problems.append(f"chosen_block_ids: '{bid}' was not selected by the rules")
    if len(set(reply["chosen_block_ids"])) != len(reply["chosen_block_ids"]):
        problems.append("chosen_block_ids: contains duplicates")
    if not 1 <= len(reply["priorities"]) <= 3:
        problems.append(f"priorities: expected 1-3, got {len(reply['priorities'])}")
    for i, p in enumerate(reply["priorities"], start=1):
        if p.get("block_id") not in reply["chosen_block_ids"]:
            problems.append(f"priorities[{i}]: block '{p.get('block_id')}' is not in chosen_block_ids")

    allowed = allowed_numbers(sub, blocks)
    for field, text in ai_text_fields(reply):
        if not str(text).strip():
            problems.append(f"{field}: empty")
        for num in _NUMBER.findall(str(text)):
            if fmt_num(num) not in allowed:
                problems.append(f"{field}: number {num} is not in the answers or the content sheet")
        for word in _NUMBER_WORD.findall(str(text)):
            problems.append(f"{field}: '{word}' is a figure written as a word; use the digits "
                            "from the answers or the content sheet, or leave it out")
    return problems