kreuzberg 3.19.0__py3-none-any.whl → 3.19.1__py3-none-any.whl

kreuzberg/_ocr/_tesseract.py

@@ -301,8 +301,15 @@ class TesseractBackend(OCRBackend[TesseractConfig]):
             "OFF",
         ]
 
-        if run_config["tesseract_format"] != "text":
-            command.append(run_config["tesseract_format"])
+        # Handle output format - use config option for HOCR to ensure Windows compatibility
+        # Windows Tesseract 5.5.0 doesn't respect 'hocr' configfile, needs explicit config
+        tesseract_format = run_config["tesseract_format"]
+        if tesseract_format == "hocr":
+            command.extend(["-c", "tessedit_create_hocr=1"])
+        elif tesseract_format == "tsv":
+            command.append("tsv")
+        elif tesseract_format != "text":
+            command.append(tesseract_format)
 
         for kwarg, value in run_config["remaining_kwargs"].items():
             if kwarg.startswith("table_"):
@@ -1162,7 +1169,13 @@ class TesseractBackend(OCRBackend[TesseractConfig]):
             "OFF",
         ]
 
-        if output_format != "text":
+        # Handle output format - use config option for HOCR to ensure Windows compatibility
+        # Windows Tesseract 5.5.0 doesn't respect 'hocr' configfile, needs explicit config
+        if output_format == "hocr":
+            command.extend(["-c", "tessedit_create_hocr=1"])
+        elif output_format == "tsv":
+            command.append("tsv")
+        elif output_format != "text":
             command.append(output_format)
 
         for kwarg, value in kwargs.items():

{kreuzberg-3.19.0.dist-info → kreuzberg-3.19.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: kreuzberg
-Version: 3.19.0
+Version: 3.19.1
 Summary: Document intelligence framework for Python - Extract text, metadata, and structured data from diverse file formats
 Project-URL: documentation, https://kreuzberg.dev
 Project-URL: homepage, https://github.com/Goldziher/kreuzberg

{kreuzberg-3.19.0.dist-info → kreuzberg-3.19.1.dist-info}/RECORD

@@ -36,7 +36,7 @@ kreuzberg/_ocr/_base.py,sha256=ZvOJvW8DtylQJZdCPk9vlVNZiBFK-dC4Oj7Kb6-mWkY,1419
 kreuzberg/_ocr/_easyocr.py,sha256=bHz2S_8nNHaPHPemcJK-U0al9_qP-vUmWE4ECVlf7AA,15485
 kreuzberg/_ocr/_paddleocr.py,sha256=CV9cCjkRe-3cNJ5tRu_sBXd_HNghEwfPIgWwxAZTeRY,15026
 kreuzberg/_ocr/_table_extractor.py,sha256=LhBiCX8R_xR-uK1FH3ONA_vqOmqUWANZJ2HMCBLsmNY,5513
-kreuzberg/_ocr/_tesseract.py,sha256=Uu6H1LMh1WSC1OmKhPx-miG98r9KEfc0GF7b8isS33E,52420
+kreuzberg/_ocr/_tesseract.py,sha256=6NSKSvXpH66j_ACKwz2rs_b9Exg0xOWRHelhZAcuw80,53162
 kreuzberg/_token_reduction/__init__.py,sha256=y_2WgPxJes8_PD-VMfx7vQT0hGjFIixzS8PjaIseAGg,311
 kreuzberg/_token_reduction/_reducer.py,sha256=shAfMPznP69sTSzwX_bE1LpcBmoia9cpd7r6bSc4R5Q,13609
 kreuzberg/_token_reduction/_stopwords.py,sha256=mu-5CapG0RCP7LYzjhdTM6WWLtmt3cjZ08OOsyQkJVg,3608
@@ -122,8 +122,8 @@ kreuzberg/_utils/_string.py,sha256=wVyvEHByHBeu_6evmqJGv9Ml-NAwkyz60n8l-7L5Cw0,4
 kreuzberg/_utils/_sync.py,sha256=gb828WYfVtkB4wKslJrPMmrdeI1h3htWceq-gywHtO4,3184
 kreuzberg/_utils/_table.py,sha256=OVg6T2QnerMhVNb1juLTBSIjyjFiE5-OrUWr5NSCgnQ,6493
 kreuzberg/_utils/_tmp.py,sha256=mwZ0BFzhGPfYa2tt8qSjUjfcHnSYvbQT4VlPRCRc_q8,2038
-kreuzberg-3.19.0.dist-info/METADATA,sha256=fV1j2iWA2-rcZodFFV3kmSsuBJhoDsW6OuyIu9Myf4A,12492
-kreuzberg-3.19.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-kreuzberg-3.19.0.dist-info/entry_points.txt,sha256=GplGhFryCP7kyAG_k-Mdahznvo2fwi73qLFg5yQfH_A,91
-kreuzberg-3.19.0.dist-info/licenses/LICENSE,sha256=-8caMvpCK8SgZ5LlRKhGCMtYDEXqTKH9X8pFEhl91_4,1066
-kreuzberg-3.19.0.dist-info/RECORD,,
+kreuzberg-3.19.1.dist-info/METADATA,sha256=CpVI7_0uQVLHP1NnQBB0mXqgjwfAW-BCO3kxTiKdvnY,12492
+kreuzberg-3.19.1.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+kreuzberg-3.19.1.dist-info/entry_points.txt,sha256=GplGhFryCP7kyAG_k-Mdahznvo2fwi73qLFg5yQfH_A,91
+kreuzberg-3.19.1.dist-info/licenses/LICENSE,sha256=-8caMvpCK8SgZ5LlRKhGCMtYDEXqTKH9X8pFEhl91_4,1066
+kreuzberg-3.19.1.dist-info/RECORD,,