diff options
Diffstat (limited to 'python/sandcrawler/html.py')
| -rw-r--r-- | python/sandcrawler/html.py | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/python/sandcrawler/html.py b/python/sandcrawler/html.py index 7e1e10d..2117eb0 100644 --- a/python/sandcrawler/html.py +++ b/python/sandcrawler/html.py @@ -91,4 +91,6 @@ def extract_fulltext_url(html_url, html_body): if iframe and '.pdf' in iframe['src']: return dict(pdf_url=iframe['src']) + # TODO: hrmars.com. anchor with .pdf href, and anchor text is "PDF" + return dict() |