From 8f8ce8cc7585ededd2fb6d66abc13134d3182f0d Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Fri, 26 Mar 2021 12:10:13 -0700
Subject: ingest: handle current degruyter PDF link pattern

---
 python/sandcrawler/html_metadata.py | 8 ++++++++
 1 file changed, 8 insertions(+)

(limited to 'python')

diff --git a/python/sandcrawler/html_metadata.py b/python/sandcrawler/html_metadata.py
index f9f48a6..1c054fa 100644
--- a/python/sandcrawler/html_metadata.py
+++ b/python/sandcrawler/html_metadata.py
@@ -375,6 +375,14 @@ PDF_FULLTEXT_PATTERNS: List[dict] = [
         "technique": "PDF URL link",
         "example_page": "http://www.bezmialemscience.org/archives/archive-detail/article-preview/editorial/20439",
     },
+    {
+        "in_doc_url": "degruyter.com/document/",
+        "in_fulltext_url": "/pdf",
+        "selector": "a.downloadPdf",
+        "attr": "href",
+        "technique": "PDF URL link",
+        "example_page": "https://www.degruyter.com/document/doi/10.1515/zaw-2021-0001/html",
+    },
 ]
 
 FULLTEXT_URL_PATTERNS_SKIP = [
-- 
cgit v1.2.3