Add Crossref search engine and DOAI rewrite plugin

searx/plugins/__init__.py

@@ -19,7 +19,8 @@ from searx import logger
 
 logger = logger.getChild('plugins')
 
-from searx.plugins import (https_rewrite,
+from searx.plugins import (doai_rewrite,
+                           https_rewrite,
                            open_results_on_new_tab,
                            self_info,
                            search_on_category_select,
@@ -73,6 +74,7 @@ class PluginStore():
 
 
 plugins = PluginStore()
+plugins.register(doai_rewrite)
 plugins.register(https_rewrite)
 plugins.register(open_results_on_new_tab)
 plugins.register(self_info)

searx/plugins/doai_rewrite.py

@@ -0,0 +1,31 @@
+from flask_babel import gettext
+import re
+from urlparse import urlparse, parse_qsl
+
+regex = re.compile(r'10\.\d{4,9}/[^\s]+')
+
+name = gettext('DOAI rewrite')
+description = gettext('Avoid paywalls by redirecting to open-access versions of publications when available')
+default_on = False
+
+
+def extract_doi(url):
+    match = regex.search(url.path)
+    if match:
+        return match.group(0)
+    for _, v in parse_qsl(url.query):
+        match = regex.search(v)
+        if match:
+            return match.group(0)
+    return None
+
+
+def on_result(request, ctx):
+    doi = extract_doi(ctx['result']['parsed_url'])
+    if doi and len(doi) < 50:
+        for suffix in ('/', '.pdf', '/full', '/meta', '/abstract'):
+            if doi.endswith(suffix):
+                doi = doi[:-len(suffix)]
+        ctx['result']['url'] = 'http://doai.io/' + doi
+        ctx['result']['parsed_url'] = urlparse(ctx['result']['url'])
+    return True