Fix 404 warning on s2 DOI not found

bibtexautocomplete/APIs/semantic_scholar.py

@@ -59,6 +59,13 @@ class SemanticScholarLookup(JSON_DT_Lookup):
     # so an average of 1 query every 3 seconds
     query_delay: float = 3.0  # = (5 * 60) / 100
 
+    def get_no_warning_codes(self) -> List[int]:
+        """Ignore 404 returned on invalid DOIs"""
+        codes = super().get_no_warning_codes()
+        if self.doi is not None:
+            codes.append(404)
+        return codes
+
     def get_base_path(self) -> str:
         if self.doi is not None:
             return self.path + "/DOI:" + self.doi

bibtexautocomplete/bibtex/matching.py

@@ -30,7 +30,8 @@
 
 
 def author_set(entry: BibtexEntry) -> Set[str]:
-    """Returns the set of normalized author lastname from entry"""
+    """Returns the set of normalized author lastname from entry
+    I.E. lowercase loast names, with spaces removed"""
     authors = entry.author
     last_names: Set[str] = set()
     for author in authors:
@@ -42,7 +43,9 @@ def author_set(entry: BibtexEntry) -> Set[str]:
 
 def common_authors(a: BibtexEntry, b: BibtexEntry) -> Tuple[int, int, int]:
     """Returns:
-    number of common authors, number of a authors, number of authors of b"""
+    - number of common authors;
+    - number of authors of a only,
+    - number of authors of b only"""
     authors_a = author_set(a)
     authors_b = author_set(b)
     common = authors_a.intersection(authors_b)

bibtexautocomplete/lookups/search_mixin.py

@@ -36,8 +36,20 @@ class SearchResultMixin(Generic[result]):
       defaults to [200]
     """
 
+    # HTTP codes that indicate we (likely) got valid data
     ok_codes: List[int] = [200]
 
+    # HTTP codes that shouldn't raise a warning, but indicate we got no data
+    # Typically 404 for websites that return 404 on unknown DOIs.
+    no_warning_codes: List[int] = []
+
+    def get_no_warning_codes(self) -> List[int]:
+        """return a list of HTTP codes that shouldn't raise a warning, but
+        indicate we got no data. Typically 404 for websites that return 404 on
+        unknown DOIs. Override this for dynamic setting, else just change
+        the no_warning_codes attribute"""
+        return self.no_warning_codes
+
     def get_results(self, data: bytes) -> Optional[Iterable[result]]:
         """Parse the data into a list of results to check
         Return None if no results/invalid data"""
@@ -59,12 +71,13 @@ def match_score(self, entry: BibtexEntry, res: result) -> int:
     def process_data(self, data: Data) -> Optional[BibtexEntry]:
         """Iterate through results until one matches"""
         if data.code not in self.ok_codes:
-            logger.warn(
-                "response: {FgYellow}{status}{reason}{Reset} in {delay}s",
-                status=data.code,
-                reason=" " + data.reason if data.reason else "",
-                delay=data.delay,
-            )
+            if data.code not in self.get_no_warning_codes():
+                logger.warn(
+                    "response: {FgYellow}{status}{reason}{Reset} in {delay}s",
+                    status=data.code,
+                    reason=" " + data.reason if data.reason else "",
+                    delay=data.delay,
+                )
             return None
         results = self.get_results(data.data)
         if results is None: