pembot 0.1.10__tar.gz → 0.1.11__tar.gz

{pembot-0.1.10 → pembot-0.1.11}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pembot
-Version: 0.1.10
+Version: 0.1.11
 Summary: A Python Package to convert PEM blog content to usseful information by leveraging LLMs
 Author-email: cyto <aryan_sidhwani@protonmail.com>
 License-Expression: MIT

pembot-0.1.11/pembot/.git/COMMIT_EDITMSG

@@ -0,0 +1 @@
+raised error if gpu limit exceeded instead of returning empty

pembot-0.1.11/pembot/.git/ORIG_HEAD

@@ -0,0 +1 @@
+5d2259f832675fd5a5031ee436a1bc8909acd6c9

{pembot-0.1.10 → pembot-0.1.11}/pembot/.git/logs/HEAD

@@ -22,3 +22,6 @@ e89cb4f5af158d26dcff5eed03dba6671a818739 e3c62c141fc65ef2be0095c49b23e06263f0b73
 e3c62c141fc65ef2be0095c49b23e06263f0b734 d440b20aae1265dabbd3ddaafb24c35e40e3ab3c cyto <silverstone965@gmail.com> 1759127664 +0530	commit: cyto/fixed the excel file input bug, in the conversion method; added exceptions where there is invalid input
 d440b20aae1265dabbd3ddaafb24c35e40e3ab3c 8d58ec13be39949ecfe7211b42c56acd2a83dc72 cyto <silverstone965@gmail.com> 1761392079 +0530	commit: added DeepSeek-OCR as a model option to use spaces
 8d58ec13be39949ecfe7211b42c56acd2a83dc72 e2dc7f666268cf207253a15d55a7fc73093c6d3b cyto <silverstone965@gmail.com> 1761539934 +0530	commit: added logs; fixed typo
+e2dc7f666268cf207253a15d55a7fc73093c6d3b 5d2259f832675fd5a5031ee436a1bc8909acd6c9 cyto <silverstone965@gmail.com> 1761794401 +0530	commit: raised error if gpu limit exceeded instead of returning empty
+5d2259f832675fd5a5031ee436a1bc8909acd6c9 e2dc7f666268cf207253a15d55a7fc73093c6d3b cyto <silverstone965@gmail.com> 1761794509 +0530	reset: moving to HEAD~
+e2dc7f666268cf207253a15d55a7fc73093c6d3b ddd2e5c09eea11798956f2de7df33a9b1eee7e62 cyto <silverstone965@gmail.com> 1761794413 +0530	commit: raised error if gpu limit exceeded instead of returning empty

{pembot-0.1.10 → pembot-0.1.11}/pembot/.git/logs/refs/heads/main

@@ -22,3 +22,6 @@ e89cb4f5af158d26dcff5eed03dba6671a818739 e3c62c141fc65ef2be0095c49b23e06263f0b73
 e3c62c141fc65ef2be0095c49b23e06263f0b734 d440b20aae1265dabbd3ddaafb24c35e40e3ab3c cyto <silverstone965@gmail.com> 1759127664 +0530	commit: cyto/fixed the excel file input bug, in the conversion method; added exceptions where there is invalid input
 d440b20aae1265dabbd3ddaafb24c35e40e3ab3c 8d58ec13be39949ecfe7211b42c56acd2a83dc72 cyto <silverstone965@gmail.com> 1761392079 +0530	commit: added DeepSeek-OCR as a model option to use spaces
 8d58ec13be39949ecfe7211b42c56acd2a83dc72 e2dc7f666268cf207253a15d55a7fc73093c6d3b cyto <silverstone965@gmail.com> 1761539934 +0530	commit: added logs; fixed typo
+e2dc7f666268cf207253a15d55a7fc73093c6d3b 5d2259f832675fd5a5031ee436a1bc8909acd6c9 cyto <silverstone965@gmail.com> 1761794401 +0530	commit: raised error if gpu limit exceeded instead of returning empty
+5d2259f832675fd5a5031ee436a1bc8909acd6c9 e2dc7f666268cf207253a15d55a7fc73093c6d3b cyto <silverstone965@gmail.com> 1761794509 +0530	reset: moving to HEAD~
+e2dc7f666268cf207253a15d55a7fc73093c6d3b ddd2e5c09eea11798956f2de7df33a9b1eee7e62 cyto <silverstone965@gmail.com> 1761794413 +0530	commit: raised error if gpu limit exceeded instead of returning empty

{pembot-0.1.10 → pembot-0.1.11}/pembot/.git/logs/refs/remotes/origin/main

@@ -21,3 +21,4 @@ e89cb4f5af158d26dcff5eed03dba6671a818739 e3c62c141fc65ef2be0095c49b23e06263f0b73
 e3c62c141fc65ef2be0095c49b23e06263f0b734 d440b20aae1265dabbd3ddaafb24c35e40e3ab3c cyto <silverstone965@gmail.com> 1759127680 +0530	update by push
 d440b20aae1265dabbd3ddaafb24c35e40e3ab3c 8d58ec13be39949ecfe7211b42c56acd2a83dc72 cyto <silverstone965@gmail.com> 1761392127 +0530	update by push
 8d58ec13be39949ecfe7211b42c56acd2a83dc72 e2dc7f666268cf207253a15d55a7fc73093c6d3b cyto <silverstone965@gmail.com> 1761539997 +0530	update by push
+e2dc7f666268cf207253a15d55a7fc73093c6d3b ddd2e5c09eea11798956f2de7df33a9b1eee7e62 cyto <silverstone965@gmail.com> 1761794465 +0530	update by push

pembot-0.1.11/pembot/.git/refs/heads/main

@@ -0,0 +1 @@
+ddd2e5c09eea11798956f2de7df33a9b1eee7e62

pembot-0.1.11/pembot/.git/refs/remotes/origin/main

@@ -0,0 +1 @@
+ddd2e5c09eea11798956f2de7df33a9b1eee7e62

{pembot-0.1.10 → pembot-0.1.11}/pembot/__init__.py

@@ -1,6 +1,6 @@
 """
 A Python Package to convert PEM blog content to usseful information by leveraging LLMs
 """
-__version__ = '0.1.10'
+__version__ = '0.1.11'
 from .main import save_to_json_file, make_query
 __all__ = ["save_to_json_file", "make_query"]

{pembot-0.1.10 → pembot-0.1.11}/pembot/config/config.yaml

@@ -2,4 +2,4 @@ OUTPUT_DIR: /home/cyto/dev/pembotdir
 PAGE_DELIMITER: ___________________________ NEXT PAGE ___________________________
 app:
   name: pembot
-  version: 0.1.10
+  version: 0.1.11

{pembot-0.1.10 → pembot-0.1.11}/pembot/pdf2markdown/extract.py

@@ -3,8 +3,7 @@ import pdfplumber
 import re
 import yaml
 import pytesseract
-import numpy as np
-from typing import Literal, final
+from typing import Literal
 from PIL import Image
 import os
 import logging
@@ -18,7 +17,6 @@ from google import genai
 from google.genai import types
 import mimetypes
 from gradio_client import Client, handle_file
-import gradio as gr
 import tempfile
 
 
@@ -187,11 +185,11 @@ class MarkdownPDFExtractor(PDFExtractor):
                             prompt_type= 'markdown',
 
                             # spaces zerogpu
-                            api_name="/predict"
+                            # api_name="/predict"
                         )
                     print("ocr'd: ", result[:100] + "...")
             except Exception as e:
-                print("Error during nanonet inference", e)
+                print("Error during HF OCR inference", e)
                 error_message = str(e)
                 if "You have exceeded your Pro GPU quota" in error_message:
                     # print("\n\n\nFALLING BACK TO TESS\n\n\n")
@@ -426,7 +424,7 @@ class MarkdownPDFExtractor(PDFExtractor):
         except Exception as e:
             self.logger.error(f"Error performing OCR: {e}")
             self.logger.exception(traceback.format_exc())
-            return ""
+            raise e
 
     def caption_image(self, image, image_bytes):
         """Generate a caption for the given image."""

{pembot-0.1.10 → pembot-0.1.11}/pembot/requirements.txt

@@ -45,7 +45,7 @@ pandas==2.3.0
 pathlib==1.0.1
 pdfminer.six==20250506
 pdfplumber==0.11.7
-pembot==0.1.10
+pembot==0.1.11
 pillow==11.3.0
 primp==0.15.0
 pyasn1==0.6.1

pembot-0.1.10/pembot/.git/COMMIT_EDITMSG

@@ -1 +0,0 @@
-added logs; fixed typo

pembot-0.1.10/pembot/.git/refs/heads/main

@@ -1 +0,0 @@
-e2dc7f666268cf207253a15d55a7fc73093c6d3b

pembot-0.1.10/pembot/.git/refs/remotes/origin/main

@@ -1 +0,0 @@
-e2dc7f666268cf207253a15d55a7fc73093c6d3b