| from __future__ import annotations |
|
|
| from io import BytesIO |
| from typing import Any |
|
|
| import fitz |
|
|
| from src.domain import MAX_PAGES |
| from src.normalize import fold_digits |
| from src.schemas import ValidationReport |
|
|
|
|
| def build_pdf(text: str, metadata: dict[str, str] | None = None) -> bytes: |
| doc = fitz.open() |
| page = doc.new_page() |
| |
| page.insert_text((72, 72), text or " ", fontsize=11) |
| if metadata: |
| doc.set_metadata({key: value for key, value in metadata.items() if value}) |
| payload = doc.tobytes() |
| doc.close() |
| return payload |
|
|
|
|
| def inspect_pdf(data: bytes, report: ValidationReport | None = None) -> dict[str, Any]: |
| issues = report or ValidationReport(ok=True) |
| try: |
| doc = fitz.open(stream=data, filetype="pdf") |
| except Exception: |
| issues.add("pdf_unreadable", "The PDF could not be opened.", "pdf") |
| return {"text": "", "pages": 0, "metadata": {}, "scanned": False, "report": issues} |
| pages = doc.page_count |
| if pages > MAX_PAGES: |
| issues.add("too_many_pages", f"PDFs are limited to {MAX_PAGES} text pages.", "pdf") |
| texts = [] |
| for page in doc: |
| texts.append(page.get_text("text") or "") |
| body = "\n".join(texts).strip() |
| metadata = {key: str(value or "") for key, value in (doc.metadata or {}).items()} |
| scanned = pages > 0 and len(body) < max(12, pages * 8) |
| if scanned: |
| issues.add( |
| "scanned_pdf", |
| "Image-only / scanned PDFs are out of scope. This MVP reads a text layer only.", |
| "pdf", |
| severity="warning", |
| ) |
| result = { |
| "text": body, |
| "pages": pages, |
| "metadata": metadata, |
| "scanned": scanned, |
| "report": issues, |
| } |
| doc.close() |
| return result |
|
|
|
|
| def redact_pdf(data: bytes, values: list[str], extra_metadata_clear: bool = True) -> tuple[bytes, dict[str, Any]]: |
| doc = fitz.open(stream=data, filetype="pdf") |
| hits = 0 |
| folded_needles = [fold_digits(value) for value in values if value and len(value) >= 4] |
| for page in doc: |
| for value in values: |
| if not value or len(value) < 3: |
| continue |
| for rect in page.search_for(value): |
| page.add_redact_annot(rect, fill=(0, 0, 0)) |
| hits += 1 |
| page.apply_redactions() |
| if extra_metadata_clear: |
| empty = {key: "" for key in (doc.metadata or {})} |
| doc.set_metadata(empty) |
| try: |
| doc.del_xml_metadata() |
| except Exception: |
| pass |
| payload = doc.tobytes() |
| verify = inspect_pdf(payload) |
| leftover = [] |
| extracted = fold_digits(verify["text"]) |
| meta_blob = fold_digits(" ".join(verify["metadata"].values())) |
| for needle, original in zip(folded_needles, [v for v in values if v and len(v) >= 4]): |
| if needle in extracted or needle in meta_blob: |
| leftover.append(original) |
| doc.close() |
| return payload, {"hits": hits, "leftover": leftover, "pages": verify["pages"], "metadata": verify["metadata"], "text": verify["text"]} |
|
|
|
|
| def write_pdf(path, data: bytes) -> None: |
| path.write_bytes(data) |
|
|
|
|
| def pdf_from_path(path) -> bytes: |
| return path.read_bytes() |
|
|
|
|
| def buffer(data: bytes) -> BytesIO: |
| return BytesIO(data) |
|
|
