EXT-7: Handle LangDetectException gracefully, return correct HTTP sta…

…tus codes (#8)

* EXT-7: Handle LangDetectException gracefully

* EXT-7: Return correct status code in HTTP exception handler

* Linting

* EXT-7: Return status 500 on unexpected internal error

* EXT-7: Re-raise original exception

main.py

@@ -79,5 +79,6 @@ def display_error_message_in_template(request: Request, exc: StarletteHTTPExcept
                 'error': str(exc.detail)
             },
             "root_path": Utils.get_environment_prefix()
-        }
+        },
+        status_code=exc.status_code
     )

metadata_extract/finder.py

@@ -8,6 +8,7 @@
 from dateutil.parser import parse
 from dateparser.search import search_dates
 import langdetect
+from langdetect.lang_detect_exception import LangDetectException
 from . import text, author_name
 from .candidate import Candidate, Origin
 from .infopage import InfoPage
@@ -218,8 +219,11 @@ def get_year_from_info(self) -> None:
 
     def get_language(self) -> None:
         """Detects language of concatenated text, and adds it as a candidate."""
-        lang = langdetect.detect(' '.join(self.doc.pages.values()))
-        self.metadata.add_candidate('language', Candidate(lang, Origin.LANGUAGE_MODEL))
+        try:
+            lang = langdetect.detect(' '.join(self.doc.pages.values()))
+            self.metadata.add_candidate('language', Candidate(lang, Origin.LANGUAGE_MODEL))
+        except LangDetectException:
+            return
 
     def read_info_page(self) -> None:
         """Finds the infopage and searches for candidate values for title, publisher and authors."""

metadata_extract/metadata.py

@@ -78,6 +78,11 @@ def rank_years(self) -> Optional[CandidateType]:
                               reverse=True)
         return sorted_years[0].to_dict()
 
+    def choose_language(self) -> Optional[CandidateType]:
+        if 'language' not in self.candidates:
+            return None
+        return self.candidates['language'][0].to_dict()
+
     def choose_isxn(self, identifier: str) -> Optional[CandidateType]:
         if identifier not in self.candidates:
             return None
@@ -143,7 +148,7 @@ def choose_authors(self) -> list[CandidateType]:
 
     def choose_best(self) -> None:
         self.results['year'] = self.rank_years()
-        self.results['language'] = self.candidates['language'][0].to_dict()
+        self.results['language'] = self.choose_language()
         self.results['title'] = self.choose_title()
         self.results['publisher'] = self.choose_publishers()
         self.results['publicationType'] = self.choose_doc_type()

src/util.py

@@ -93,9 +93,10 @@ def process_and_remove(
         try:
             results = self.meteor.run(filepath)
             return results
-        except Exception:
+        except Exception as exc:
             print(traceback.format_exc())
-            return {'error': f'Error while processing file {filename if filename else ""}'}
+            raise HTTPException(detail=f'Error while processing file {filename}',
+                                status_code=500) from exc
         finally:
             if delete_immediately:
                 os.remove(filepath)