- fix(nexus): Fix DOI detection in messages

GitOrigin-RevId: e4d4519221319c26134d8cbbd011eca2def2b6b3

nexus/meta_api/README.md

@@ -1,5 +1,5 @@
 # Nexus Search: Meta API
 
 ```
-NEXUS_META_API_summa.url=http://summa bazel run -c opt binary
+NEXUS_META_API_summa.url=http://summa bazel run binary
 ```

nexus/meta_api/query_extensionner/checks.py

@@ -5,6 +5,7 @@ from nexus.nlptools.regex import (
     DOI_REGEX,
     ISBN_REGEX,
     NID_REGEX,
+    ONLY_DOI_REGEX,
     URL_REGEX,
 )
 
@@ -20,10 +21,10 @@ class QueryClass(Enum):
 
 
 def check_doi(query) -> (QueryClass, str):
-    # ToDo: rewrite normally, just hotfixed
-    if query.startswith('references:'):
-        return
-    if r := re.search(DOI_REGEX, query):
+    if (
+        ((r := re.search(DOI_REGEX, query)) and re.search(URL_REGEX, query))
+        or re.search(ONLY_DOI_REGEX, query)
+    ):
         doi = (r[1] + '/' + r[2]).lower()
         return {
             'doi': doi,

nexus/nlptools/regex.py

@@ -29,4 +29,5 @@ DOI_REGEX = re.compile(r'(10.\d{4,9})\s?/\s?([-._;()<>/:A-Za-z0-9]+[^.?\s])')
 ISBN_REGEX = re.compile(r'^(?:[iI][sS][bB][nN]\:?\s*)?((97(8|9))?\-?\d{9}(\d|X))$')
 MD5_REGEX = re.compile(r'([A-Fa-f0-9]{32})')
 NID_REGEX = re.compile(r'(?:[Nn][Ii][Dd]\s?:?\s*)([0-9]+)')
+ONLY_DOI_REGEX = re.compile(r'^(10.\d{4,9})\s?/\s?([-._;()<>/:A-Za-z0-9]+[^.?\s])$')
 PUBMED_ID_REGEX = re.compile(r'(?:(?:https?://)?(?:www.)?ncbi.nlm.nih.gov/pubmed/|[Pp][Mm][Ii][Dd]\s?:?\s*)([0-9]+)')