GVersion 0.2.2

Author: JorjMcKie

pymupdf4llm/pymupdf4llm/helpers/check_ocr.py

@@ -184,9 +184,6 @@ def should_ocr_page(
 
     if decision["has_text"]:
         # early exit if any text exists
-        print(
-            f"{decision['has_text']=}, {decision['readable_text']=}, {decision['should_ocr']=}"
-        )
         return decision
 
     # Check for image coverage

pymupdf4llm/pymupdf4llm/helpers/document_layout.py

@@ -130,7 +130,7 @@ def is_superscripted(line):
         return True
     if len(spans) < 2:  # single span line: skip
         return False
-    if spans0["origin"][1] < spans[1]["origin"][1] and span0["size"] < spans[1]["size"]:
+    if span0["origin"][1] < spans[1]["origin"][1] and span0["size"] < spans[1]["size"]:
         return True
     return False