pembot 0.0.5__tar.gz → 0.0.7__tar.gz

{pembot-0.0.5 → pembot-0.0.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pembot
-Version: 0.0.5
+Version: 0.0.7
 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-0.0.7/pembot/.git/COMMIT_EDITMSG

@@ -0,0 +1 @@
+added model name to convertor

{pembot-0.0.5 → pembot-0.0.7}/pembot/.git/logs/HEAD

@@ -5,3 +5,6 @@ ac9c9018c62fa30dc142665c1b5a375f4e056880 72f047cda92abcd1ddc857f6461de605f866833
 e91172752e9a421ae463112d2b0506b37498c98d 0c8d9b2690545bf1906b05cd9f18b783b3eb74f1 cyto <silverstone965@gmail.com> 1749716350 +0530	commit: added a pem blog chunking module for updating from local, and, an embedding loop to embed all the blogs, with document id as the filter in the search, and the first line title as the filter in updation
 0c8d9b2690545bf1906b05cd9f18b783b3eb74f1 eb75e1c49f1e5b79dca17ccdbec8067756523238 cyto <silverstone965@gmail.com> 1750856653 +0530	commit: made arrangements for the cases when custom file bytes are to be processed to text output; handled a ollama running / crashing error
 eb75e1c49f1e5b79dca17ccdbec8067756523238 0bdb4169fc0f312b8698f1df17a258fff163aeaa cyto <silverstone965@gmail.com> 1750937276 +0530	commit: fixed the output_dir bug; fixed the excel to json function; ran some tests on convertor; incremented the version on the package; removed dependency on schema / structure, and shifted required fields to a pickle file path in the cli args;
+0bdb4169fc0f312b8698f1df17a258fff163aeaa 9528bbccd167e3f4ad583a1ae9fac98a52620e27 cyto <silverstone965@gmail.com> 1750947488 +0530	commit: handled local llm nonexistent error properly for choice of just passing None as llm_client;
+9528bbccd167e3f4ad583a1ae9fac98a52620e27 ef0503a60244391590b16042019032e91d7cc30d cyto <silverstone965@gmail.com> 1751872559 +0530	commit: added a model_name_parameter to change models quicky
+ef0503a60244391590b16042019032e91d7cc30d af80ddb5890f062e364ea8ade2d602df4e12de8c cyto <silverstone965@gmail.com> 1751896700 +0530	commit: added model name to convertor

{pembot-0.0.5 → pembot-0.0.7}/pembot/.git/logs/refs/heads/main

@@ -5,3 +5,6 @@ ac9c9018c62fa30dc142665c1b5a375f4e056880 72f047cda92abcd1ddc857f6461de605f866833
 e91172752e9a421ae463112d2b0506b37498c98d 0c8d9b2690545bf1906b05cd9f18b783b3eb74f1 cyto <silverstone965@gmail.com> 1749716350 +0530	commit: added a pem blog chunking module for updating from local, and, an embedding loop to embed all the blogs, with document id as the filter in the search, and the first line title as the filter in updation
 0c8d9b2690545bf1906b05cd9f18b783b3eb74f1 eb75e1c49f1e5b79dca17ccdbec8067756523238 cyto <silverstone965@gmail.com> 1750856653 +0530	commit: made arrangements for the cases when custom file bytes are to be processed to text output; handled a ollama running / crashing error
 eb75e1c49f1e5b79dca17ccdbec8067756523238 0bdb4169fc0f312b8698f1df17a258fff163aeaa cyto <silverstone965@gmail.com> 1750937276 +0530	commit: fixed the output_dir bug; fixed the excel to json function; ran some tests on convertor; incremented the version on the package; removed dependency on schema / structure, and shifted required fields to a pickle file path in the cli args;
+0bdb4169fc0f312b8698f1df17a258fff163aeaa 9528bbccd167e3f4ad583a1ae9fac98a52620e27 cyto <silverstone965@gmail.com> 1750947488 +0530	commit: handled local llm nonexistent error properly for choice of just passing None as llm_client;
+9528bbccd167e3f4ad583a1ae9fac98a52620e27 ef0503a60244391590b16042019032e91d7cc30d cyto <silverstone965@gmail.com> 1751872559 +0530	commit: added a model_name_parameter to change models quicky
+ef0503a60244391590b16042019032e91d7cc30d af80ddb5890f062e364ea8ade2d602df4e12de8c cyto <silverstone965@gmail.com> 1751896700 +0530	commit: added model name to convertor

{pembot-0.0.5 → pembot-0.0.7}/pembot/.git/logs/refs/remotes/origin/main

@@ -4,3 +4,6 @@ ac9c9018c62fa30dc142665c1b5a375f4e056880 72f047cda92abcd1ddc857f6461de605f866833
 e91172752e9a421ae463112d2b0506b37498c98d 0c8d9b2690545bf1906b05cd9f18b783b3eb74f1 cyto <silverstone965@gmail.com> 1749716371 +0530	update by push
 0c8d9b2690545bf1906b05cd9f18b783b3eb74f1 eb75e1c49f1e5b79dca17ccdbec8067756523238 cyto <silverstone965@gmail.com> 1750856672 +0530	update by push
 eb75e1c49f1e5b79dca17ccdbec8067756523238 0bdb4169fc0f312b8698f1df17a258fff163aeaa cyto <silverstone965@gmail.com> 1750937389 +0530	update by push
+0bdb4169fc0f312b8698f1df17a258fff163aeaa 9528bbccd167e3f4ad583a1ae9fac98a52620e27 cyto <silverstone965@gmail.com> 1750947502 +0530	update by push
+9528bbccd167e3f4ad583a1ae9fac98a52620e27 ef0503a60244391590b16042019032e91d7cc30d cyto <silverstone965@gmail.com> 1751872581 +0530	update by push
+ef0503a60244391590b16042019032e91d7cc30d af80ddb5890f062e364ea8ade2d602df4e12de8c cyto <silverstone965@gmail.com> 1751896713 +0530	update by push

pembot-0.0.7/pembot/.git/objects/ef/0503a60244391590b16042019032e91d7cc30d

@@ -0,0 +1,3 @@
+x
��Kj1D��)z� ��ĴZ-{�h���
�>2�AVU��Qom��9��j�5�d��,��#��tQF&T�J|��ۀ�t̙�(�T��E�
+9U��i�%� �>�}z�Dz��~��q����pYߩ�OP���s	ޤ3R�v
+�*�.��z���a��Ԟ�t��¯����B���k�U�

pembot-0.0.7/pembot/.git/refs/heads/main

@@ -0,0 +1 @@
+af80ddb5890f062e364ea8ade2d602df4e12de8c

pembot-0.0.7/pembot/.git/refs/remotes/origin/main

@@ -0,0 +1 @@
+af80ddb5890f062e364ea8ade2d602df4e12de8c

{pembot-0.0.5 → pembot-0.0.7}/pembot/AnyToText/convertor.py

@@ -31,10 +31,14 @@ EXCEL_FILE_TYPES= [
 class Convertor():
 
 
-    def __init__(self, myfile: Path | None= None, output_dir: Path | None= None, file_bytes: bytes | None= None, suffix: str | None= None, file_type: str | None= None):
+    def __init__(self, myfile: Path | None= None, output_dir: Path | None= None, file_bytes: bytes | None= None, suffix: str | None= None, file_type: str | None= None, model_name: str | None = None):
 
         self.output= ""
 
+        if model_name is None:
+            # model_name=  "gemini-2.5-flash"
+            model_name=  "Nanonets-OCR-s"
+
         # file_type can be pdf, excel, etc.
         if output_dir is None and myfile is None and file_bytes is not None and suffix is not None:
             with tempfile.TemporaryDirectory() as dp:
@@ -43,7 +47,7 @@ class Convertor():
                     myfile= Path(fp.name)
                     output_dir= Path(dp)
                     if file_type == 'pdf':
-                        extractor= MarkdownPDFExtractor(str(myfile), output_path= str(output_dir), page_delimiter= "-- NEXT PAGE --")
+                        extractor= MarkdownPDFExtractor(str(myfile), output_path= str(output_dir), page_delimiter= "-- NEXT PAGE --", model_name= model_name)
                         extractor.extract()
                         with open(output_dir / (myfile.stem + '.md')) as output_file:
                             self.output= output_file.read()
@@ -67,7 +71,7 @@ class Convertor():
                 print("the file was json")
             elif mt == 'application/pdf':
                 print("the file was pdf, outputting in: ", output_dir)
-                extractor= MarkdownPDFExtractor(str(myfile), output_path= str(self.output_dir), page_delimiter= "-- NEXT PAGE --")
+                extractor= MarkdownPDFExtractor(str(myfile), output_path= str(self.output_dir), page_delimiter= "-- NEXT PAGE --", model_name= model_name)
                 extractor.extract()
 
             elif mt in EXCEL_FILE_TYPES:
@@ -333,10 +337,10 @@ def chunk_text(text, chunk_size=500, overlap_size=50):
 if __name__ == '__main__':
     print("Test Run Start:")
     try:
-        # print("Test 1: scaned pdf page, bytes")
-        # with open("/home/cyto/Documents/scanned.pdf", "rb") as imgpdf:
-        #     conv= Convertor(file_bytes= imgpdf.read(), suffix= ".pdf", file_type= "pdf")
-        #     print(conv.output)
+        print("Test 1: scaned pdf page, bytes")
+        with open("/home/cyto/Documents/scanned.pdf", "rb") as imgpdf:
+            conv= Convertor(file_bytes= imgpdf.read(), suffix= ".pdf", file_type= "pdf")
+            print(conv.output)
 
         # print("Test 2: JD pdf, bytes")
         # with open("/home/cyto/dev/pembotdir/jds/PM Trainee.pdf", "rb") as imgpdf:

{pembot-0.0.5 → pembot-0.0.7}/pembot/__init__.py

@@ -1,6 +1,6 @@
 """
 A Python Package to convert PEM blog content to usseful information by leveraging LLMs
 """
-__version__ = '0.0.5'
+__version__ = '0.0.7'
 from .main import save_to_json_file, make_query
 __all__ = ["save_to_json_file", "make_query"]

{pembot-0.0.5 → pembot-0.0.7}/pembot/config/config.yaml

@@ -2,4 +2,4 @@ OUTPUT_DIR: /home/cyto/dev/pembotdir
 PAGE_DELIMITER: ___________________________ NEXT PAGE ___________________________
 app:
   name: pembot
-  version: 0.0.5
+  version: 0.0.7

pembot-0.0.7/pembot/pdf2markdown/.git/COMMIT_EDITMSG

@@ -0,0 +1 @@
+handled the gpu errors non-gracefully so that it stops

{pembot-0.0.5 → pembot-0.0.7}/pembot/pdf2markdown/.git/config

@@ -9,3 +9,6 @@
 [branch "main"]
 	remote = origin
 	merge = refs/heads/main
+[remote "myorigin"]
+	url = https://github.com/silverstone-git/pdf-to-markdown.git
+	fetch = +refs/heads/*:refs/remotes/myorigin/*

pembot-0.0.7/pembot/pdf2markdown/.git/logs/HEAD

@@ -0,0 +1,5 @@
+0000000000000000000000000000000000000000 ffb759ee4605b232366a9ee58134532913c3f9e0 cyto <cyto@callisto.localdomain> 1747745478 +0530	clone: from https://github.com/iamarunbrahma/pdf-to-markdown
+ffb759ee4605b232366a9ee58134532913c3f9e0 b8702320e56074e9680181d8b7897d6a0a552e2d cyto <silverstone965@gmail.com> 1750947962 +0530	commit: handled config loading errors gracefully; added gemini support, as an option; added huggingface nanonets transformers support (as an option); redesigned the extract markdown for captioning and image ocr (block image and full-page image);
+b8702320e56074e9680181d8b7897d6a0a552e2d 14251b198e0bac39a3dc3b42f9e57b20c01465fb cyto <silverstone965@gmail.com> 1751604763 +0530	commit: removed deps on torch and transformers; used gradio client for ocr through public spaces;
+14251b198e0bac39a3dc3b42f9e57b20c01465fb b48d697aa9fd97151eb2a84a1af5d408b7630232 cyto <silverstone965@gmail.com> 1751871887 +0530	commit: cyto/argument-list-bug-fix;authentication-used-in-gradio-client
+b48d697aa9fd97151eb2a84a1af5d408b7630232 f3b2d76c75bbd50e04fc4c2ad17fc94ca6daed32 cyto <silverstone965@gmail.com> 1751896628 +0530	commit: handled the gpu errors non-gracefully so that it stops

pembot-0.0.7/pembot/pdf2markdown/.git/logs/refs/heads/main

@@ -0,0 +1,5 @@
+0000000000000000000000000000000000000000 ffb759ee4605b232366a9ee58134532913c3f9e0 cyto <cyto@callisto.localdomain> 1747745478 +0530	clone: from https://github.com/iamarunbrahma/pdf-to-markdown
+ffb759ee4605b232366a9ee58134532913c3f9e0 b8702320e56074e9680181d8b7897d6a0a552e2d cyto <silverstone965@gmail.com> 1750947962 +0530	commit: handled config loading errors gracefully; added gemini support, as an option; added huggingface nanonets transformers support (as an option); redesigned the extract markdown for captioning and image ocr (block image and full-page image);
+b8702320e56074e9680181d8b7897d6a0a552e2d 14251b198e0bac39a3dc3b42f9e57b20c01465fb cyto <silverstone965@gmail.com> 1751604763 +0530	commit: removed deps on torch and transformers; used gradio client for ocr through public spaces;
+14251b198e0bac39a3dc3b42f9e57b20c01465fb b48d697aa9fd97151eb2a84a1af5d408b7630232 cyto <silverstone965@gmail.com> 1751871887 +0530	commit: cyto/argument-list-bug-fix;authentication-used-in-gradio-client
+b48d697aa9fd97151eb2a84a1af5d408b7630232 f3b2d76c75bbd50e04fc4c2ad17fc94ca6daed32 cyto <silverstone965@gmail.com> 1751896628 +0530	commit: handled the gpu errors non-gracefully so that it stops

pembot-0.0.7/pembot/pdf2markdown/.git/logs/refs/remotes/myorigin/main

@@ -0,0 +1,4 @@
+0000000000000000000000000000000000000000 b8702320e56074e9680181d8b7897d6a0a552e2d cyto <silverstone965@gmail.com> 1750948073 +0530	update by push
+b8702320e56074e9680181d8b7897d6a0a552e2d 14251b198e0bac39a3dc3b42f9e57b20c01465fb cyto <silverstone965@gmail.com> 1751604904 +0530	update by push
+14251b198e0bac39a3dc3b42f9e57b20c01465fb b48d697aa9fd97151eb2a84a1af5d408b7630232 cyto <silverstone965@gmail.com> 1751872077 +0530	update by push
+b48d697aa9fd97151eb2a84a1af5d408b7630232 f3b2d76c75bbd50e04fc4c2ad17fc94ca6daed32 cyto <silverstone965@gmail.com> 1751896663 +0530	update by push

pembot-0.0.7/pembot/pdf2markdown/.git/objects/f3/b2d76c75bbd50e04fc4c2ad17fc94ca6daed32

@@ -0,0 +1 @@
+x
��Kj!3vw� �l媷?`k�v��>�!�C'��:'Hk�f't:�lȺ�6g�u2j�߈G�TV��ةN��gb�rp���F��ɚ���RI��<Z���

pembot-0.0.7/pembot/pdf2markdown/.git/refs/heads/main

@@ -0,0 +1 @@
+f3b2d76c75bbd50e04fc4c2ad17fc94ca6daed32

pembot-0.0.7/pembot/pdf2markdown/.git/refs/remotes/myorigin/main

@@ -0,0 +1 @@
+f3b2d76c75bbd50e04fc4c2ad17fc94ca6daed32

{pembot-0.0.5 → pembot-0.0.7}/pembot/pdf2markdown/extract.py

@@ -2,11 +2,9 @@ import fitz
 import pdfplumber
 import re
 import yaml
-# import pytesseract
+import pytesseract
 import numpy as np
-from transformers import AutoTokenizer, AutoProcessor, AutoModelForImageTextToText, VisionEncoderDecoderModel, ViTImageProcessor
 from typing import Literal, final
-import torch
 from PIL import Image
 import os
 import logging
@@ -19,6 +17,9 @@ import io
 from google import genai
 from google.genai import types
 import mimetypes
+from gradio_client import Client, handle_file
+import gradio as gr
+import tempfile
 
 
 
@@ -75,25 +76,18 @@ class MarkdownPDFExtractor(PDFExtractor):
         super().__init__(pdf_path)
 
         if model_name is None:
-            self.MODEL_NAME= "gemini-2.5-flash"
+            # self.MODEL_NAME= "gemini-2.5-flash"
+            self.MODEL_NAME= "Nanonets-OCR-s"
         else:
             self.MODEL_NAME= model_name
 
         if  "gemini" in self.MODEL_NAME:
             self.gclient = genai.Client(api_key= os.getenv("GEMINI_API_KEY", ''))
-        else:
-            model_path = "nanonets/Nanonets-OCR-s"
-            self.model = AutoModelForImageTextToText.from_pretrained(
-                model_path,
-                torch_dtype="auto",
-                device_map="auto",
-                attn_implementation="flash_attention_2"
-            )
-            self.model.eval()
-            self.tokenizer = AutoTokenizer.from_pretrained(model_path)
-            self.processor = AutoProcessor.from_pretrained(model_path)
-            self.setup_image_captioning()
+        elif "anonet" in self.MODEL_NAME:
+            # self.nclient= Client("prithivMLmods/Multimodal-OCR2")
 
+            # zerogpu public
+            self.nclient= Client("deepak-mehta/ocr-simplify", hf_token= os.getenv('HF_TOKEN', ''))
 
 
         self.markdown_content= ""
@@ -108,25 +102,6 @@ class MarkdownPDFExtractor(PDFExtractor):
 
 
 
-    def setup_image_captioning(self):
-        """Set up the image captioning model."""
-        try:
-            self.model = VisionEncoderDecoderModel.from_pretrained(
-                "nlpconnect/vit-gpt2-image-captioning"
-            )
-            self.feature_extractor = ViTImageProcessor.from_pretrained(
-                "nlpconnect/vit-gpt2-image-captioning"
-            )
-            self.tokenizer = AutoTokenizer.from_pretrained(
-                "nlpconnect/vit-gpt2-image-captioning"
-            )
-            self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-            self.model.to(self.device)
-            self.logger.info("Image captioning model set up successfully.")
-        except Exception as e:
-            self.logger.error(f"Error setting up image captioning model: {e}")
-            self.logger.exception(traceback.format_exc())
-
     def extract(self):
         try:
             markdown_content, markdown_pages = self.extract_markdown()
@@ -140,15 +115,25 @@ class MarkdownPDFExtractor(PDFExtractor):
         except Exception as e:
             self.logger.error(f"Error processing PDF: {e}")
             self.logger.exception(traceback.format_exc())
+
+            error_message= str(e).lower()
+            if "GPU" in error_message and "quota" in error_message:
+                return "GPU quota error", []
             return "", []
 
 
-    def ocr_page_with_nanonets_s(self, pil_image, img_bytes, max_new_tokens: int | None = None):
-        prompt = """Extract the text from the above document as if you were reading it naturally. Return the tables in html format. Return the equations in LaTeX representation. If there is an image in the document and image caption is not present, add a small description of the image inside the <img></img> tag; otherwise, add the image caption inside <img></img>. Watermarks should be wrapped in brackets. Ex: <watermark>OFFICIAL COPY</watermark>. Page numbers should be wrapped in brackets. Ex: <page_number>14</page_number> or <page_number>9/22</page_number>. Prefer using ☐ and ☑ for check boxes."""
+    def image_ocr(self, pil_image, img_bytes, max_new_tokens: int | None = None, prompt: str | None= None):
+        if prompt is None:
+            prompt = """Extract the text from the above document as if you were reading it naturally. Return the tables in html format. Return the equations in LaTeX representation. If there is an image in the document and image caption is not present, add a small description of the image inside the <img></img> tag; otherwise, add the image caption inside <img></img>. Watermarks should be wrapped in brackets. Ex: <watermark>OFFICIAL COPY</watermark>. Page numbers should be wrapped in brackets. Ex: <page_number>14</page_number> or <page_number>9/22</page_number>. Prefer using ☐ and ☑ for check boxes."""
         if max_new_tokens is None:
             max_new_tokens= 4096
 
-        if 'gemini' in self.MODEL_NAME:
+        w, h= pil_image.size
+        if w < 200 or h < 50:
+            return "<img> A small image </img>"
+
+        model_name= self.MODEL_NAME.lower()
+        if 'gemini' in model_name:
 
             image_format = pil_image.format
             dummy_filename = f"dummy.{image_format.lower()}"
@@ -165,24 +150,46 @@ class MarkdownPDFExtractor(PDFExtractor):
             )
             # print("response :", response)
             return response.text
-        else:
-            image = pil_image
-            messages = [
-                {"role": "system", "content": "You are a helpful assistant."},
-                {"role": "user", "content": [
-                    {"type": "image", "image": image},
-                    {"type": "text", "text": prompt},
-                ]},
-            ]
-            text = self.processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
-            inputs = self.processor(text=[text], images=[image], padding=True, return_tensors="pt")
-            inputs = inputs.to(self.model.device)
-
-            output_ids = self.model.generate(**inputs, max_new_tokens=max_new_tokens, do_sample=False)
-            generated_ids = [output_ids[len(input_ids):] for input_ids, output_ids in zip(inputs.input_ids, output_ids)]
+        elif 'nanonet' in model_name:
 
-            output_text = self.processor.batch_decode(generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=True)
-            return output_text[0]
+            result= ""
+            try:
+                with tempfile.NamedTemporaryFile(suffix=f'.{pil_image.format.lower()}', mode= 'w') as temp_file:
+                    pil_image.save(temp_file.name)
+                    print("file name: ", temp_file.name)
+                    gr_image= handle_file(temp_file.name)
+                    print("gr image : ", gr_image)
+                    result = self.nclient.predict(
+                        # model_name="Nanonets-OCR-s",
+                        # text= prompt,
+                        gr_image,
+                        # max_new_tokens=max_new_tokens,
+                        # temperature=0.6,
+                        # top_p=0.9,
+                        # top_k=50,
+                        # repetition_penalty=1.2,
+
+                        # prithiv model
+                        # api_name="/generate_image"
+
+                        max_new_tokens,
+
+                        # spaces zerogpu
+                        api_name="/predict"
+                    )
+                    print("ocr'd: ", result[:100] + "...")
+            except Exception as e:
+                print("Error during nanonet 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")
+                    # return pytesseract.image_to_string(pil_image)
+                    raise e
+
+            
+            return result
+        else:
+            return pytesseract.image_to_string(pil_image)
 
 
 
@@ -219,7 +226,7 @@ class MarkdownPDFExtractor(PDFExtractor):
                 for page_num, page in enumerate(doc):
                     current_page_markdown_blocks = [] # Collect markdown blocks for the current page
                     page_has_searchable_text = False
-                    page_has_embedded_images = False
+                    # page_has_embedded_images = False
 
                     self.logger.info(f"\nProcessing page {page_num + 1}...")
 
@@ -252,7 +259,7 @@ class MarkdownPDFExtractor(PDFExtractor):
                             try:
                                 image_bytes= io.BytesIO(img_data)
                                 pil_image = Image.open(image_bytes)
-                                ocr_text_from_block_image = self.ocr_page_with_nanonets_s(
+                                ocr_text_from_block_image = self.image_ocr(
                                     pil_image, image_bytes, max_new_tokens=15000
                                 )
 
@@ -265,6 +272,9 @@ class MarkdownPDFExtractor(PDFExtractor):
                             except Exception as e:
                                 self.logger.error(f"    Error processing embedded image block for OCR: {e}")
                                 current_page_markdown_blocks.append("\n\n![Image Processing Error](error_on_page_{page_num+1}_block_{block_num+1}.png)\n\n")
+                                error_message= str(e).lower()
+                                if "GPU" in error_message and "quota" in error_message:
+                                    raise e
 
 
                     # Insert tables at their approximate positions (after blocks are processed for the page)
@@ -292,7 +302,7 @@ class MarkdownPDFExtractor(PDFExtractor):
                             image_bytestream= io.BytesIO(img_bytes)
                             pil_image = Image.open(image_bytestream)
 
-                            ocr_text_from_page = self.ocr_page_with_nanonets_s(
+                            ocr_text_from_page = self.image_ocr(
                                 pil_image, image_bytestream, max_new_tokens=15000
                             )
 
@@ -309,6 +319,9 @@ class MarkdownPDFExtractor(PDFExtractor):
                                 self.logger.info(f"  Full-page OCR yielded no text for page {page_num+1}.")
                         except Exception as e:
                             self.logger.error(f"  Error during full-page OCR on page {page_num+1}: {e}")
+                            error_message= str(e).lower()
+                            if "GPU" in error_message and "quota" in error_message:
+                                raise e
                     else:
                         self.logger.info(f"  Page {page_num + 1} has sufficient searchable text or embedded image OCR; skipping full-page OCR.")
 
@@ -332,7 +345,12 @@ class MarkdownPDFExtractor(PDFExtractor):
             except Exception as e:
                 self.logger.critical(f"An unexpected error occurred during markdown extraction: {e}")
                 self.logger.exception(traceback.format_exc())
-                return "", []
+
+                error_message= str(e).lower()
+                if "GPU" in error_message and "quota" in error_message:
+                    return "GPU quota error", []
+                else:
+                    return "", []
 
     def extract_tables(self):
         """Extract tables from PDF using pdfplumber."""
@@ -389,7 +407,7 @@ class MarkdownPDFExtractor(PDFExtractor):
             # ocr_result = pytesseract.image_to_string(
             #     image
             # )
-            ocr_result= self.ocr_page_with_nanonets_s(image, image_bytes, max_new_tokens=15000)
+            ocr_result= self.image_ocr(image, image_bytes, max_new_tokens=15000)
 
 
             return ocr_result.strip()
@@ -409,41 +427,15 @@ class MarkdownPDFExtractor(PDFExtractor):
             if image.mode != "RGB":
                 image = image.convert("RGB")
 
-            image_format = image.format
-            dummy_filename = f"dummy.{image_format.lower()}"
-            mime_type, _ = mimetypes.guess_type(dummy_filename)
-
-            if "gemini" in self.MODEL_NAME:
-                response=  self.gclient.models.generate_content(
-                    model= self.MODEL_NAME,
-                    contents=[
-                        types.Part.from_bytes(
-                            data=image_bytes.getvalue(),
-                            mime_type= mime_type
-                        ),
-                        "Write a caption for this image"
-                    ]
-                )
-                return response.text
-            else:
-                # Ensure the image is in the correct shape
-                image = np.array(image).transpose(2, 0, 1)  # Convert to (C, H, W) format
+            caption= self.image_ocr(image, image_bytes, max_new_tokens=15000, prompt= "Write a caption for this image")
+            return caption
 
-                inputs = self.feature_extractor(images=image, return_tensors="pt").to(
-                    self.device
-                )
-                pixel_values = inputs.pixel_values
-
-                generated_ids = self.model.generate(pixel_values, max_length=30)
-
-                generated_ids = self.model.generate(pixel_values, max_length=30)
-                generated_caption = self.tokenizer.batch_decode(
-                    generated_ids, skip_special_tokens=True
-                )[0]
-                return generated_caption.strip()
         except Exception as e:
             self.logger.error(f"Error captioning image: {e}")
             self.logger.exception(traceback.format_exc())
+            error_message= str(e)
+            if "GPU" in error_message and "quota" in error_message:
+                raise e
             return ""
 
     def clean_text(self, text):
@@ -758,6 +750,7 @@ class MarkdownPDFExtractor(PDFExtractor):
             self.logger.exception(traceback.format_exc())
             return ""
 
+
     def get_header_level(self, font_size):
         """Determine header level based on font size."""
         if font_size > 24:

pembot-0.0.7/pembot/pdf2markdown/pyrightconfig.json

@@ -0,0 +1,4 @@
+{
+  "venvPath": "../..",
+  "venv": "venvpem"
+}

pembot-0.0.7/pembot/requirements.txt

@@ -0,0 +1,80 @@
+aiofiles==24.1.0
+annotated-types==0.7.0
+anyio==4.9.0
+audioop-lts==0.2.1
+cachetools==5.5.2
+certifi==2025.6.15
+cffi==1.17.1
+charset-normalizer==3.4.2
+click==8.2.1
+cryptography==45.0.5
+dnspython==2.7.0
+et_xmlfile==2.0.0
+fastapi==0.115.14
+ffmpy==0.6.0
+filelock==3.18.0
+fsspec==2025.5.1
+google-auth==2.40.3
+google-genai==1.24.0
+gradio==5.35.0
+gradio_client==1.10.4
+greenlet==3.2.3
+groovy==0.1.2
+h11==0.16.0
+hf-xet==1.1.5
+httpcore==1.0.9
+httpx==0.28.1
+huggingface-hub==0.33.2
+idna==3.10
+Jinja2==3.1.6
+markdown-it-py==3.0.0
+MarkupSafe==3.0.2
+mdurl==0.1.2
+msgpack==1.1.1
+numpy==2.3.1
+ollama==0.5.1
+openpyxl==3.1.5
+orjson==3.10.18
+packaging==25.0
+pandas==2.3.0
+pathlib==1.0.1
+pdfminer.six==20250506
+pdfplumber==0.11.7
+pembot==0.0.6
+pillow==11.3.0
+pyasn1==0.6.1
+pyasn1_modules==0.4.2
+pycparser==2.22
+pydantic==2.11.7
+pydantic_core==2.33.2
+pydub==0.25.1
+Pygments==2.19.2
+pymongo==4.13.2
+PyMuPDF==1.26.3
+pynvim==0.5.2
+pypdfium2==4.30.1
+pytesseract==0.3.13
+python-dateutil==2.9.0.post0
+python-multipart==0.0.20
+pytz==2025.2
+PyYAML==6.0.2
+requests==2.32.4
+rich==14.0.0
+rsa==4.9.1
+ruff==0.12.1
+safehttpx==0.1.6
+semantic-version==2.10.0
+shellingham==1.5.4
+six==1.17.0
+sniffio==1.3.1
+starlette==0.46.2
+tenacity==8.5.0
+tomlkit==0.13.3
+tqdm==4.67.1
+typer==0.16.0
+typing-inspection==0.4.1
+typing_extensions==4.14.0
+tzdata==2025.2
+urllib3==2.5.0
+uvicorn==0.35.0
+websockets==15.0.1

pembot-0.0.5/pembot/.git/COMMIT_EDITMSG

@@ -1 +0,0 @@
-fixed the output_dir bug; fixed the excel to json function; ran some tests on convertor; incremented the version on the package; removed dependency on schema / structure, and shifted required fields to a pickle file path in the cli args;

pembot-0.0.5/pembot/.git/refs/heads/main

@@ -1 +0,0 @@
-0bdb4169fc0f312b8698f1df17a258fff163aeaa

pembot-0.0.5/pembot/.git/refs/remotes/origin/main

@@ -1 +0,0 @@
-0bdb4169fc0f312b8698f1df17a258fff163aeaa