Harden scripts.extract_detected_languages against flat directory structures

scripts/extract_detected_languages.py

@@ -27,11 +27,15 @@ THRESHOLD = float(sys.argv[5]) / 100.0
 
 
 def get_languages_worker(filename):
-    basename = str(INPUT_OCR_ROOT / filename.parent / filename.stem)
+    basename = INPUT_OCR_ROOT / filename.stem
     try:
-        languages = read_page_languages(basename, DETECTED_LANGUAGES, algorithm='OLDA')
+        languages = read_page_languages(str(basename), DETECTED_LANGUAGES, algorithm='OLDA')
     except IOError:
-        return 'not-exists'
+        basename = INPUT_OCR_ROOT / filename.parent / filename.stem
+        try:
+            languages = read_page_languages(str(basename), DETECTED_LANGUAGES, algorithm='OLDA')
+        except IOError:
+            return 'not-exists'
     return (filename, languages)