diff options
Diffstat (limited to 'python')
| -rw-r--r-- | python/sandcrawler/html.py | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/python/sandcrawler/html.py b/python/sandcrawler/html.py index 091162d..34da876 100644 --- a/python/sandcrawler/html.py +++ b/python/sandcrawler/html.py @@ -223,7 +223,7 @@ def extract_fulltext_url(html_url, html_body): # american archivist (OA) # https://americanarchivist.org/doi/abs/10.17723/aarc.62.2.j475270470145630 - if "://americanarchivist.org/doi/abs/" in html_url: + if "://americanarchivist.org/doi/" in html_url and not "/doi/pdf" in html_url: # <a href="/doi/pdf/10.17723/aarc.62.2.j475270470145630" target="_blank"> hrefs = soup.find_all('a', attrs={"target":"_blank"}) for href in hrefs: |