diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2022-07-12 15:52:51 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2022-07-12 15:52:51 -0700 |
| commit | 0d5f6468960a7bbf7a499654feaaa5ecab7d9130 (patch) | |
| tree | 800195bfaf9cdfc0d37d5299231416ebda3cf29a /python | |
| parent | 8f85ab294eae50e31efa9e31bb0bca1bca76cf8b (diff) | |
| download | sandcrawler-0d5f6468960a7bbf7a499654feaaa5ecab7d9130.tar.gz sandcrawler-0d5f6468960a7bbf7a499654feaaa5ecab7d9130.zip | |
ingest: doaj.org article landing page access links
Diffstat (limited to 'python')
| -rw-r--r-- | python/sandcrawler/html_metadata.py | 12 | ||||
| -rw-r--r-- | python/sandcrawler/ingest_file.py | 1 |
2 files changed, 12 insertions, 1 deletions
diff --git a/python/sandcrawler/html_metadata.py b/python/sandcrawler/html_metadata.py index 2fb500c..edaf89f 100644 --- a/python/sandcrawler/html_metadata.py +++ b/python/sandcrawler/html_metadata.py @@ -255,6 +255,12 @@ HTML_FULLTEXT_PATTERNS: List[Dict[str, str]] = [ "attr": "href", "technique": "dovepress fulltext link", }, + { + "in_doc_url": "://doaj.org/article/", + "selector": "section.col-md-8 a[target='_blank'].button--primary", + "attr": "href", + "technique": "doaj.org access link", + }, ] COMPONENT_FULLTEXT_PATTERNS: List[Dict[str, str]] = [ @@ -627,6 +633,12 @@ PDF_FULLTEXT_PATTERNS: List[Dict[str, str]] = [ "attr": "href", "technique": "PDF link (scitemed.com)", }, + { + "in_doc_url": "://doaj.org/article/", + "selector": "section.col-md-8 a[target='_blank'].button--primary", + "attr": "href", + "technique": "doaj.org access link", + }, ] FULLTEXT_URL_PATTERNS_SKIP: List[str] = [ diff --git a/python/sandcrawler/ingest_file.py b/python/sandcrawler/ingest_file.py index c79973f..25ae7d2 100644 --- a/python/sandcrawler/ingest_file.py +++ b/python/sandcrawler/ingest_file.py @@ -128,7 +128,6 @@ class IngestFileWorker(SandcrawlerWorker): "://www.openlibrary.org/", "://fatcat.wiki/", "://orcid.org/", - "://doaj.org/", # Domain squats "://bartandjones.com", "://ijretm.com", |