Hotfixes

Author: JorjMcKie

pdf4llm/setup.py

@@ -6,7 +6,7 @@
 with open(os.path.join(setup_py_cwd, "README.md"), encoding="utf-8") as f:
     readme = f.read()
 
-version = "0.2.1"
+version = "0.2.2"
 
 classifiers = [
     "Development Status :: 5 - Production/Stable",

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
 

pymupdf4llm/pymupdf4llm/versions_file.py

@@ -1,3 +1,3 @@
 # Generated file - do not edit.
 MINIMUM_PYMUPDF_VERSION = (1, 26, 6)
-VERSION = '0.2.1'
+VERSION = '0.2.2'

pymupdf4llm/setup.py

@@ -14,7 +14,7 @@
     "Topic :: Utilities",
 ]
 
-version = "0.2.1"
+version = "0.2.2"
 requires = ["pymupdf>=1.26.6", "tabulate"]
 
 text = requires[0].split("=")[1]