diff options
| author | J08nY | 2022-07-26 10:05:46 +0200 |
|---|---|---|
| committer | J08nY | 2022-07-26 10:06:22 +0200 |
| commit | 63e986b66e8578836d587bd91c17ed1aa70e15fe (patch) | |
| tree | 49cb1d3e7aff39347a432bf674a5a63e182a9f4b | |
| parent | 5bb98186b9126f2cadc5f8996e27ba437086dd9a (diff) | |
| download | sec-certs-63e986b66e8578836d587bd91c17ed1aa70e15fe.tar.gz sec-certs-63e986b66e8578836d587bd91c17ed1aa70e15fe.tar.zst sec-certs-63e986b66e8578836d587bd91c17ed1aa70e15fe.zip | |
Ignore errors in hyperlink collection from PDFs.
| -rw-r--r-- | sec_certs/utils/pdf.py | 17 |
1 files changed, 10 insertions, 7 deletions
diff --git a/sec_certs/utils/pdf.py b/sec_certs/utils/pdf.py index 0f272d1b..9b55d370 100644 --- a/sec_certs/utils/pdf.py +++ b/sec_certs/utils/pdf.py @@ -195,13 +195,16 @@ def extract_pdf_metadata(filepath: Path) -> Tuple[str, Optional[Dict[str, Any]]] annots = reduce(lambda x, y: x + y, annots) links = set() for a in annots: - if isinstance(a, IndirectObject): - note = a.getObject() - else: - note = a - link = note.get("/A", {}).get("/URI") - if link: - links.add(map_metadata_value(link)) + try: + if isinstance(a, IndirectObject): + note = a.getObject() + else: + note = a + link = note.get("/A", {}).get("/URI") + if link: + links.add(map_metadata_value(link)) + except Exception: + pass metadata["pdf_hyperlinks"] = links except Exception as e: |