From 9516fd423d0a0545f8206e0ef42439d0578fa20c Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Tue, 12 Jul 2022 21:01:20 -0700
Subject: ingest: random site PDF link pattern

---
 python/sandcrawler/html_metadata.py | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/python/sandcrawler/html_metadata.py b/python/sandcrawler/html_metadata.py
index edaf89f..7b44bfe 100644
--- a/python/sandcrawler/html_metadata.py
+++ b/python/sandcrawler/html_metadata.py
@@ -639,6 +639,13 @@ PDF_FULLTEXT_PATTERNS: List[Dict[str, str]] = [
         "attr": "href",
         "technique": "doaj.org access link",
     },
+    {
+        "in_doc_url": "/jvi.aspx",
+        "in_fulltext_url": "download_fulltext",
+        "selector": "div.siteMainWrapper div.siteArticleShare a[target='_blank'].list-group-item",
+        "attr": "href",
+        "technique": "erciyesmedj.com publication system PDF download link",
+    },
 ]
 
 FULLTEXT_URL_PATTERNS_SKIP: List[str] = [
-- 
cgit v1.2.3