pembot 0.1.7__py2.py3-none-any.whl → 0.1.9__py2.py3-none-any.whl

pembot/.git/COMMIT_EDITMSG

@@ -1 +1 @@
-silly willy mistake
+added DeepSeek-OCR as a model option to use spaces

pembot/.git/logs/HEAD

@@ -19,3 +19,5 @@ c3cc0da3d955ecec0f865c46c030a0c073697495 e6adbc3c373070269f97ef82d4f63027d7878f6
 e6adbc3c373070269f97ef82d4f63027d7878f67 81d01e1c63d48b096c77aae83471d42272ca9fce cyto <silverstone965@gmail.com> 1758890262 +0530	commit: cyto/fixed a file already open error and wrote to file if myfile is excel type and output_dir is given
 81d01e1c63d48b096c77aae83471d42272ca9fce e89cb4f5af158d26dcff5eed03dba6671a818739 cyto <silverstone965@gmail.com> 1758891589 +0530	commit: cyto/made the excel convertor compatible with suffix input from both system file and with bytes + manual suffix input
 e89cb4f5af158d26dcff5eed03dba6671a818739 e3c62c141fc65ef2be0095c49b23e06263f0b734 cyto <silverstone965@gmail.com> 1758892053 +0530	commit: silly willy mistake
+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

pembot/.git/logs/refs/heads/main

@@ -19,3 +19,5 @@ c3cc0da3d955ecec0f865c46c030a0c073697495 e6adbc3c373070269f97ef82d4f63027d7878f6
 e6adbc3c373070269f97ef82d4f63027d7878f67 81d01e1c63d48b096c77aae83471d42272ca9fce cyto <silverstone965@gmail.com> 1758890262 +0530	commit: cyto/fixed a file already open error and wrote to file if myfile is excel type and output_dir is given
 81d01e1c63d48b096c77aae83471d42272ca9fce e89cb4f5af158d26dcff5eed03dba6671a818739 cyto <silverstone965@gmail.com> 1758891589 +0530	commit: cyto/made the excel convertor compatible with suffix input from both system file and with bytes + manual suffix input
 e89cb4f5af158d26dcff5eed03dba6671a818739 e3c62c141fc65ef2be0095c49b23e06263f0b734 cyto <silverstone965@gmail.com> 1758892053 +0530	commit: silly willy mistake
+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

pembot/.git/logs/refs/remotes/origin/main

@@ -18,3 +18,5 @@ c3cc0da3d955ecec0f865c46c030a0c073697495 e6adbc3c373070269f97ef82d4f63027d7878f6
 e6adbc3c373070269f97ef82d4f63027d7878f67 81d01e1c63d48b096c77aae83471d42272ca9fce cyto <silverstone965@gmail.com> 1758890281 +0530	update by push
 81d01e1c63d48b096c77aae83471d42272ca9fce e89cb4f5af158d26dcff5eed03dba6671a818739 cyto <silverstone965@gmail.com> 1758891600 +0530	update by push
 e89cb4f5af158d26dcff5eed03dba6671a818739 e3c62c141fc65ef2be0095c49b23e06263f0b734 cyto <silverstone965@gmail.com> 1758892065 +0530	update by push
+e3c62c141fc65ef2be0095c49b23e06263f0b734 d440b20aae1265dabbd3ddaafb24c35e40e3ab3c cyto <silverstone965@gmail.com> 1759127680 +0530	update by push
+d440b20aae1265dabbd3ddaafb24c35e40e3ab3c 8d58ec13be39949ecfe7211b42c56acd2a83dc72 cyto <silverstone965@gmail.com> 1761392127 +0530	update by push

pembot/.git/objects/8d/58ec13be39949ecfe7211b42c56acd2a83dc72

@@ -0,0 +1 @@
+x
��Mj�0@�u�ٗ�ѯc(����	F3��Բ��z���~��q-e���}�*�L�<%o��!���O�40KD<{�f�]�f�Dj]�B9�!���TOٳ�[����Wxm��{�u�1���B��µ������a�g�͡�`�RC"*�}����>.@

pembot/.git/refs/heads/main

@@ -1 +1 @@
-e3c62c141fc65ef2be0095c49b23e06263f0b734
+8d58ec13be39949ecfe7211b42c56acd2a83dc72

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

@@ -1 +1 @@
-e3c62c141fc65ef2be0095c49b23e06263f0b734
+8d58ec13be39949ecfe7211b42c56acd2a83dc72

pembot/AnyToText/convertor.py

@@ -5,7 +5,6 @@ from pembot.pdf2markdown.extract import MarkdownPDFExtractor
 import os
 import pandas as pd
 from typing import Literal, Union
-import tempfile
 from datetime import datetime, date
 from tabulate import tabulate
 
@@ -36,6 +35,10 @@ class Convertor():
         self.input_filepath= None
         self.output= ""
         self.suffix= suffix
+        self.file_bytes= file_bytes
+
+        if (file_bytes and not suffix) or (file_bytes and not file_type):
+            raise Exception("wrong use of convertor library, have to give suffix and file_type along with the file_bytes")
 
 
         if model_name is None:
@@ -44,7 +47,7 @@ class Convertor():
 
         # file_type can be pdf, excel, etc.
         if file_bytes and suffix:
-            with tempfile.TemporaryDirectory() as dp:
+            with TemporaryDirectory() as dp:
                 output_dir = Path(dp)
                 myfile = output_dir / f"input{suffix}"
                 myfile.write_bytes(file_bytes)
@@ -102,42 +105,51 @@ class Convertor():
         markdown_output = []
 
         file_suffix= ''
-        try:
-            if not input_filepath.exists():
-                file_suffix= self.suffix
-            else:
-                file_suffix = input_filepath.suffix.lower()
-
-            current_engine: PandasReadEngineType = excel_ods_engine
 
-            if file_suffix in ['.xls', '.xlsx', '.ods']:
-                if file_suffix == '.ods':
-                    if current_engine is None:
-                        current_engine = 'odf'
-                    elif current_engine != 'odf':
-                        print(f"Warning: Specified engine '{current_engine}' may not be optimal for ODS. Forcing 'odf'.")
-                        current_engine = 'odf'
-
-                excel_file = pd.ExcelFile(input_filepath, engine=current_engine)
-                if not excel_file.sheet_names:
-                    return f"Warning: File '{input_filepath.name}' contains no sheets."
-
-                for sheet_name in excel_file.sheet_names:
-                    df = excel_file.parse(sheet_name)
-                    markdown_output.append(f"## {sheet_name}\n")
+        try:
+            with TemporaryDirectory() as dp:
+                dp_path= Path(dp)
+
+                if input_filepath is None:
+                    file_suffix= self.suffix
+                    if file_suffix and self.file_bytes:
+                        input_filepath= dp_path / ("my_excel_file" + file_suffix)
+                        input_filepath.write_bytes(self.file_bytes)
+                    else:
+                        raise Exception("no input file name, and no file bytes either")
+                elif input_filepath.exists():
+                    file_suffix = input_filepath.suffix.lower()
+
+                current_engine: PandasReadEngineType = excel_ods_engine
+
+                if file_suffix in ['.xls', '.xlsx', '.ods']:
+                    if file_suffix == '.ods':
+                        if current_engine is None:
+                            current_engine = 'odf'
+                        elif current_engine != 'odf':
+                            print(f"Warning: Specified engine '{current_engine}' may not be optimal for ODS. Forcing 'odf'.")
+                            current_engine = 'odf'
+
+                    excel_file = pd.ExcelFile(input_filepath, engine=current_engine)
+                    if not excel_file.sheet_names:
+                        return f"Warning: File '{input_filepath.name}' contains no sheets."
+
+                    for sheet_name in excel_file.sheet_names:
+                        df = excel_file.parse(sheet_name)
+                        markdown_output.append(f"## {sheet_name}\n")
+                        markdown_table = tabulate(df, headers='keys', tablefmt='pipe')
+                        markdown_output.append(markdown_table)
+                        markdown_output.append("\n")
+
+                    return "\n".join(markdown_output)
+
+                elif file_suffix == '.csv':
+                    df = pd.read_csv(input_filepath)
                     markdown_table = tabulate(df, headers='keys', tablefmt='pipe')
-                    markdown_output.append(markdown_table)
-                    markdown_output.append("\n")
+                    return markdown_table
 
-                return "\n".join(markdown_output)
-
-            elif file_suffix == '.csv':
-                df = pd.read_csv(input_filepath)
-                markdown_table = tabulate(df, headers='keys', tablefmt='pipe')
-                return markdown_table
-
-            else:
-                return f"Error: Unsupported file type: '{file_suffix}'. Please provide a CSV, XLS, XLSX, or ODS file."
+                else:
+                    return f"Error: Unsupported file type: '{file_suffix}'. Please provide a CSV, XLS, XLSX, or ODS file."
 
         except ImportError as ie:
             if 'odfpy' in str(ie).lower() and file_suffix == '.ods':
@@ -183,10 +195,10 @@ if __name__ == '__main__':
         #     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:
-        #     conv= Convertor(file_bytes= imgpdf.read(), suffix= ".pdf", file_type= "pdf")
-        #     print(conv.output)
+        print("Test 2: balance sheet, bytes")
+        with open("/home/cyto/Downloads/balance_sheet_2023-24_final.xlsx", "rb") as xl:
+            conv= Convertor(file_bytes= xl.read(), suffix= ".xlsx", file_type= "excel")
+            print(conv.output)
 
         print("Test 3: excel schedule, bytes")
         with open("/home/cyto/Downloads/Assignment schedule.xlsx", "rb") as imgpdf:
@@ -194,13 +206,13 @@ if __name__ == '__main__':
             print(conv.output)
 
         # without bytes example:
-        print("Test 4: scanned pdf, path")
-        conv= Convertor(myfile= Path('/home/cyto/Documents/scanned.pdf'), output_dir= Path('/home/cyto/Documents'))
-        print(conv.output)
-
-        print("Test 5: schedule excel, path")
-        conv= Convertor(myfile= Path('/home/cyto/Downloads/Assignment schedule.xlsx'), output_dir= Path('/home/cyto/Downloads'))
-        print(conv.output)
+        # print("Test 4: scanned pdf, path")
+        # conv= Convertor(myfile= Path('/home/cyto/Documents/scanned.pdf'), output_dir= Path('/home/cyto/Documents'))
+        # print(conv.output)
+        #
+        # print("Test 5: schedule excel, path")
+        # conv= Convertor(myfile= Path('/home/cyto/Downloads/Assignment schedule.xlsx'), output_dir= Path('/home/cyto/Downloads'))
+        # print(conv.output)
     except FileNotFoundError as fe:
         print("file not found, modify the driver code to get sample files to test:\n\n", fe)
     except Exception as e:

pembot/__init__.py

@@ -1,6 +1,6 @@
 """
 A Python Package to convert PEM blog content to usseful information by leveraging LLMs
 """
-__version__ = '0.1.7'
+__version__ = '0.1.9'
 from .main import save_to_json_file, make_query
 __all__ = ["save_to_json_file", "make_query"]

pembot/config/config.yaml

@@ -2,4 +2,4 @@ OUTPUT_DIR: /home/cyto/dev/pembotdir
 PAGE_DELIMITER: ___________________________ NEXT PAGE ___________________________
 app:
   name: pembot
-  version: 0.1.7
+  version: 0.1.9

pembot/pdf2markdown/extract.py

@@ -88,6 +88,9 @@ class MarkdownPDFExtractor(PDFExtractor):
 
             # zerogpu public
             self.nclient= Client("deepak-mehta/ocr-simplify", hf_token= os.getenv('HF_TOKEN', ''))
+        elif "DeepSeek-OCR" in self.MODEL_NAME:
+            # zerogpu private
+            self.dclient= Client("deepak-mehta/deepseek-ocr", hf_token= os.getenv('HF_TOKEN', ''))
 
 
         self.markdown_content= ""
@@ -150,7 +153,7 @@ class MarkdownPDFExtractor(PDFExtractor):
             )
             # print("response :", response)
             return response.text
-        elif 'nanonet' in model_name:
+        elif 'nanonet' in model_name or 'DeepSeek-OCR' in model_name:
 
             result= ""
             try:
@@ -159,24 +162,33 @@ class MarkdownPDFExtractor(PDFExtractor):
                     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"
-                    )
+                    if 'nanonet' in model_name:
+                        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"
+                        )
+                    else:
+                        result = self.dclient.predict(
+                            file_input=gr_image,
+                            prompt_type= 'markdown',
+
+                            # spaces zerogpu
+                            api_name="/predict"
+                        )
                     print("ocr'd: ", result[:100] + "...")
             except Exception as e:
                 print("Error during nanonet inference", e)

pembot/requirements.txt

@@ -45,7 +45,7 @@ pandas==2.3.0
 pathlib==1.0.1
 pdfminer.six==20250506
 pdfplumber==0.11.7
-pembot==0.1.7
+pembot==0.1.9
 pillow==11.3.0
 primp==0.15.0
 pyasn1==0.6.1

{pembot-0.1.7.dist-info → pembot-0.1.9.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pembot
-Version: 0.1.7
+Version: 0.1.9
 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-0.1.7.dist-info → pembot-0.1.9.dist-info}/RECORD

@@ -1,19 +1,19 @@
 pembot/.gitignore,sha256=yyDEUmeqZekG4AOrU9Zvu2ZQhJvEzEg_lQp2CDfBhXM,92
 pembot/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
-pembot/__init__.py,sha256=EX9fGAqPH699o3Da6RqEsOR2Wr3MKdTPah0RJwfK7mw,211
+pembot/__init__.py,sha256=smnlaO-5RkabqARB-idge6rJg8yodM3L9mzVPS6qHUs,211
 pembot/gartner.py,sha256=3ALknQ5mSXIimmwCa3JFDzB_EW2hHEcQO1T2odyBquk,5408
 pembot/main.py,sha256=lZLIV8XPonvNoY4LVS-5fct1y9URMXWoSGJUKMw3Yg8,9667
 pembot/output_structure_local.py,sha256=YfpHzfTNeLMSsB_CjAamha9D6Iz7E1IC-tW9xPCMWFc,3000
 pembot/pem.py,sha256=mv6iGcN1peSY7z2dtCQ_BKj31EFBNfczBhps_d-0XDo,6377
 pembot/pyrightconfig.json,sha256=j2O2tc8Z-Zu7hEnhN9neoKk6-iLkAlp4qOmAxFyHB7Y,368
 pembot/query.py,sha256=zgfIJsSMDatFPl0Fw3MhK7fO8uBB0Yj4rxEAExqGyGA,18054
-pembot/requirements.txt,sha256=VdlT1GjMuEBd0aaXBvYu8jVGNEHqS-HYP0PxFkjy1mE,1548
+pembot/requirements.txt,sha256=ds-Oa7buELYeuYmmk_m7Ue2AXFfODHkL7oD-U_uiACc,1548
 pembot/search.py,sha256=IW0F8QjE-HSYP47v5P9EqfnzKgFEf5CGxeICtHDDrkE,9137
-pembot/.git/COMMIT_EDITMSG,sha256=8Aj0tny5TMtJMf8GXlLD0iWIa3014XUbB4kdQ_DpywE,20
+pembot/.git/COMMIT_EDITMSG,sha256=-sRwEF42WGCK3nwVw43YEXAN7ezI3j0LfpddK1HWff8,51
 pembot/.git/HEAD,sha256=KNJb-Cr0wOK3L1CVmyvrhZ4-YLljCl6MYD2tTdsrboA,21
 pembot/.git/config,sha256=ZFl9d2GyxirgRXRsv8iULIieKxwGC9P6SAjB_AmTkmQ,271
 pembot/.git/description,sha256=hatsFj1DoX6pz3eIMIvKFGbxsKjRzJLibpv2PaQGKu4,73
-pembot/.git/index,sha256=D-X98FZtV9oFdGMYpyz21lyOIKkGYQT3H1cAkggVsCo,2054
+pembot/.git/index,sha256=DVip8TexqW-hFAlcFVfCUupKN3pTLAsb1zJ1tBLUbO4,2054
 pembot/.git/packed-refs,sha256=7DECsr7q7vJ6Gw6a2gS3dE4v-YzbxGiWYoSWM43DgsQ,112
 pembot/.git/hooks/applypatch-msg.sample,sha256=AiNJeguLAzqlijpSG4YphpOGz3qw4vEBlj0yiqYhk_c,478
 pembot/.git/hooks/commit-msg.sample,sha256=H3TV6SkpebVz69WXQdRsuT_zkazdCD00C5Q3B1PZJDc,896
@@ -30,15 +30,17 @@ pembot/.git/hooks/push-to-checkout.sample,sha256=pT0HQXmLKHxt16-mSu5HPzBeZdP0lGO
 pembot/.git/hooks/sendemail-validate.sample,sha256=ROv8kj3FRmvACWAvDs8Ge5xlRZq_6IaN3Em3jmztepI,2308
 pembot/.git/hooks/update.sample,sha256=jV8vqD4QPPCLV-qmdSHfkZT0XL28s32lKtWGCXoU0QY,3650
 pembot/.git/info/exclude,sha256=ZnH-g7egfIky7okWTR8nk7IxgFjri5jcXAbuClo7DsE,240
-pembot/.git/logs/HEAD,sha256=jUmGH5OkLoVPnJR660a56FOJ7CUtST1Ca8iwVwJ1XP8,5018
-pembot/.git/logs/refs/heads/main,sha256=jUmGH5OkLoVPnJR660a56FOJ7CUtST1Ca8iwVwJ1XP8,5018
+pembot/.git/logs/HEAD,sha256=1bmHS5n4N0g5a9FoGTyCcjLjOLjUokarBu41xIuX2VA,5456
+pembot/.git/logs/refs/heads/main,sha256=1bmHS5n4N0g5a9FoGTyCcjLjOLjUokarBu41xIuX2VA,5456
 pembot/.git/logs/refs/remotes/origin/HEAD,sha256=OrkNquczPPh6fEGtutFKva_-_JhAdwnvXpCCPC4N6jk,194
-pembot/.git/logs/refs/remotes/origin/main,sha256=cvNjhHmt2b8MNQIyud3Qv2nz80GBqiKaAnNSL6cz3bk,2920
+pembot/.git/logs/refs/remotes/origin/main,sha256=YloiqLDHJGSAiRvtvpfH5YuPME6B2BTuTzqOgG6EKtU,3212
 pembot/.git/objects/00/3ba85af0ed7b9f6ab099ca298c3d0c18fb002b,sha256=pnk_IbjhUJWavx5BKSlXX8CEvWEMoSm8Dv1tQrUmzn4,169
 pembot/.git/objects/01/5f71967c525963c827d7fe5415ae2c040c4a64,sha256=-qlT-5utWcwFnO3ADkH2SA2LBsdcph6wE2iePxJxkHs,170
 pembot/.git/objects/05/5e82e69847a636258cb994bb920c03a93b5ff4,sha256=eNZTNvT7qgsLCfJvRfTETWZIkk_vBEEroLNbPC8RRa4,90
+pembot/.git/objects/06/ef9ad559094e5b48fe2a1b437dca5cea07c06b,sha256=Muyv2SPX9C670CzFqB_ss1KWQMRJD0EP03aClyo_Yhg,905
 pembot/.git/objects/06/f7563094fe405dfd9c69f05e357f4e20fc5979,sha256=1BPcr84XzGPl2ULpwFPcVi1zGLpgdd_zmvxICXD34Hk,90
 pembot/.git/objects/09/ee34d94dffd4c286df1d6d528b2e98e2a6cce2,sha256=FSXPGn6UBhR7s1Ug-afzCYLfGy8dE3Umn8dBKaahkDM,203
+pembot/.git/objects/0a/2121ea3115562cc205df572ea26532aaac5244,sha256=D0g2Dq2cUDMPiH6czmZePougntu2TNoVVEshqMCzJzo,905
 pembot/.git/objects/0a/6dd69f9258bba08a669efec17ff170fdd1509e,sha256=k6_BM-BMD6mZXEmMuqCuEJSdHF_j3kmxe2QlF6-Nc4I,527
 pembot/.git/objects/0a/fb3a98cdc55b1434b44534ec2bf22c56cfa26c,sha256=Xxw20vI57zuhERWopDAZpQw6rAOhFtUr05lzpGyCTTE,120
 pembot/.git/objects/0b/db4169fc0f312b8698f1df17a258fff163aeaa,sha256=hsOHhX0Yajg27Y7B9lo-WjDXzW1KNMg2CBr93G116EY,387
@@ -65,6 +67,7 @@ pembot/.git/objects/2f/f2a190e96dded527d8dd1ab00b706f95348d99,sha256=1td2g361orB
 pembot/.git/objects/35/97e518a8658280be9f377f78edf1dfa1f23814,sha256=gfc5bFLVZpwNQb1Ox2VosDYAjw0Lc5ZLjmvNA8gWcmg,2546
 pembot/.git/objects/37/175696b3ca7a5d17379f03fb61a1023d50aeba,sha256=XaF3EsJ1wSIWtgBtgKsZkwiMK0NM8acFy9nnqE9_d0s,3085
 pembot/.git/objects/3a/54acc088992fa8e890b93e83115ec6dc019835,sha256=EZlP672_7dB3SZ_ZwyBsFmVTJpLoblFqDdfoW-2v990,2879
+pembot/.git/objects/3c/80f6d984dff52ee250436fcf3da2e809967ae4,sha256=Aa0Oja-UqBgdV9q1d2uSQJNdj-voW-wpVGfPQmj9Xkg,170
 pembot/.git/objects/3d/07d3b29ff53d95de3898fb786d61732f210515,sha256=A9MNZO3QZ6ghGd1MyfmJ6H3dBTpF4HZcRosVxWytx8E,4077
 pembot/.git/objects/3d/e536f9c1fd05a23c2dec66423ed610afb0cf5f,sha256=omF4gmE9IQFZR8t6ybAKfnW02tdn9ZaVWKRhv_o1V4c,2083
 pembot/.git/objects/3e/23850624fcf5f111d6ea88ddd64adf924cf82f,sha256=ygVUpaLo7cxUdIgjFlaBh2BkllV6BIYYkzLIxsPKjWE,4111
@@ -73,6 +76,7 @@ pembot/.git/objects/3f/78215d7e17da726fb352fd92b3c117db9b63ba,sha256=J8r5hqTEgAw
 pembot/.git/objects/3f/e072cf3cb6a9f30c3e9936e3ddf622e80270d0,sha256=Z-UoKi2MYe0qGTtBxAr5cnIOHKkhoEXMgalevFUz9lA,2992
 pembot/.git/objects/41/ae8fa8f8baa2daee5ec0aa21ae17922ae051a0,sha256=TLuVmtSH9K33qB-WHMxKDUihHCrwdTtCKtjBs-rAnJ4,56
 pembot/.git/objects/41/cbeb6bcb4c6fa9ef9be571082d95ecb4ea0ee3,sha256=waMrzjG_o5D4JgHkjjqcDQCwuS17w60JRkVr25ZFlcI,117
+pembot/.git/objects/41/cca8bf63122d1044d2fb36f63467ccd500832e,sha256=EdQFpFKwRO06A_EiJGZGmMGzYp4IO68KHmOXNSFezUA,526
 pembot/.git/objects/42/ef76e19df247993cf5b64aba5dccaf8587a375,sha256=aDo41HaWG05-IuCeY0RPuYfi31yFhtB4y0D36A92Dkw,115
 pembot/.git/objects/42/f03e1b66aa56bbb36a1c3a8dea9e1e727faffa,sha256=n4W2gcagesjI1rStKNxQ98q5UOHlfwFJGUADFeYldoE,418
 pembot/.git/objects/44/86da0f89c566c3bae8abf435d37aeca87f1632,sha256=S2hY860Ep-0c7gQcbgrH6ioG7-Hw9a3BwYHcCkwy1Hg,3884
@@ -82,6 +86,7 @@ pembot/.git/objects/4b/c4370a037feed828cca0915ebb0bb94b24a9d4,sha256=jt9lsSz8c3d
 pembot/.git/objects/4d/a03134f70896f72053fbdc0cd4f4c76d4ac1d8,sha256=GBhAvxM1omIt-PN6mNXYlIJMN5nx2AUE0ZOf68El5pc,117
 pembot/.git/objects/50/39b29fda67743a044993436df6a4a1db7b8888,sha256=NYNmYtOq8IMmH32GaQSOBpTRTTm6jEJfY3vytVpzfKM,115
 pembot/.git/objects/51/9e780574933d7627a083222bd10dd74f430904,sha256=3e3Iu2-waVySghbLYXmwhDPpfhV4PF82suvjcYkSVog,3604
+pembot/.git/objects/55/a26fb846654d84aacea136307a35fb0c46c9c8,sha256=ufE1kt8BOEfq3iqwX6k5H86bt775jQtCBfAyHCgfmhM,115
 pembot/.git/objects/59/69ac8b9d6b44a601385c3ed8c710a69d05216e,sha256=3IOcUn5myiozgeId1iWJZX-r7cS65xXnzQCEjrc-1ZA,168
 pembot/.git/objects/5b/efa3b2f18d2b5d332c6de503a7054f4af0569f,sha256=g84QcQu-1NZ4-MfLHRosIUOnlK0VItVBqqFW5ffGDNI,882
 pembot/.git/objects/5c/4f01d3ce9e243bbb8a693f97e5c7d13a857cb7,sha256=BnHoA5JBo5NY2ReemhwmZ-dOdx6CwXWY1TQsc-FSM5o,242
@@ -104,11 +109,13 @@ pembot/.git/objects/79/4431c1d34c60b4f3fb963823f77f33bd947cc7,sha256=4jqvkfzW3aD
 pembot/.git/objects/7a/7d28b0313a3d9d509823faaae31949af8610ef,sha256=X59k-p9VNLBpmJlL53qIz8mntLeCSpnjw-rq9u9z_6I,90
 pembot/.git/objects/7e/0907822f7d316ebe0be07e1f6918bef412c80b,sha256=lFc55Bu-vEXF8In553gHxlEsB47Vg2qFXHiJqepWEqg,5167
 pembot/.git/objects/81/d01e1c63d48b096c77aae83471d42272ca9fce,sha256=dPITERwChdrVyBsZnhkOBblW83G0w3Zm7Y6e1CnodtA,216
+pembot/.git/objects/82/f733fe4edc22fe2f4caa889d2533b24a7bf9df,sha256=1fhgN9PebFoGr-0k49nbV0ztRH5OrSu1K43arcakiI0,3046
 pembot/.git/objects/86/cdaec229f1fbebf43042266b03878944669f25,sha256=eTvQhUeYXP8E181oTOcBydcgmImr62IizaH_Jbcbg8g,4077
 pembot/.git/objects/87/d6df5217a4a374f8c1211a05f9bd657f72c9a7,sha256=OGq5-x1lFa94vTX7WYO6o4TGvCZwAvZ6LXm6N3dpiKM,3881
 pembot/.git/objects/88/0c3d45ac59940344dfb6c45005f7e908173138,sha256=7VMQzB6baLdC2Uj5f84w-X6XLM3GinXGBQjewhXupAc,914
 pembot/.git/objects/89/d2439385c82b98104f27edf39bcf28a631233f,sha256=7jIYfDS2dTNKkTxZKOBmgsSZeoaE6e48JEPzIhI1ySQ,91
 pembot/.git/objects/8b/5be2af9b16f290549193859c214cd9072212e8,sha256=DhGeGisCdFZ0TcRKp5angRpaseI87TQDt5FtGZInstk,117
+pembot/.git/objects/8d/58ec13be39949ecfe7211b42c56acd2a83dc72,sha256=sJLULXCzo0CAnARNynNiXwWposMLEdmT2ICZnxRWJZQ,189
 pembot/.git/objects/8d/adc1d7891c79de24ba2c7c38b4c830bf61870a,sha256=QJaAleJXlBhybaUcSeKB7nC9OJg9gjP_xc071Wyq8BM,115
 pembot/.git/objects/8f/c00bf69f4ad3e50c13acc4a0988b6c0fe72b5a,sha256=uJVaujaQWN_NwzK9P0SM7cYp3I6GQFXdlYBPrnqVhcg,159
 pembot/.git/objects/90/f067b86364ab243a7e3bc75f936319ba9eac88,sha256=FLAmmgvYuEAx1-ZBU30rvDzP0ppXWRSVrzPWVnArIb0,203
@@ -127,6 +134,7 @@ pembot/.git/objects/ab/c6b15265171457b41e2cfdaf3b8c3994a59eb7,sha256=ivRCkHzUZHX
 pembot/.git/objects/ab/f77db148e3fb3b26913af14ae43130396f3269,sha256=rJJenBYvGWdXvwNgHWqIJCF3q2oxeA3eT19eRxlNmJA,3346
 pembot/.git/objects/ac/9c9018c62fa30dc142665c1b5a375f4e056880,sha256=P_8LPBV0v4D17Akj4f5Cr2dhgNFUsh4o7DLK78CfNPo,349
 pembot/.git/objects/ac/e2b51971b1bbade48afffc8d117c74d18f123c,sha256=U0sHygSqCMG4GOXK-fga5jBweWkzFfmR-iEphGInoac,905
+pembot/.git/objects/af/66fa89b4e1d1a8ac32dca38126f5510faea126,sha256=sg2s7hhVck8NLGWJmoze5Yl0CeimyXzAE1dFhvygSEc,56
 pembot/.git/objects/af/80ddb5890f062e364ea8ade2d602df4e12de8c,sha256=QELzH3NdMCFohFEcf5oAAu_e54VFr-LhTyPbXY7GjSk,169
 pembot/.git/objects/b1/1173d9b68db117437ccb9551461152e1e8a77d,sha256=6cl8NMNQ9b5fBh97GPEQNssOVrh-EQLJfhqSBbNb_vU,205
 pembot/.git/objects/b1/ddf2869bc7d213b35dabd6fa5bfae44cd6b7a7,sha256=zC9EjJo4qRxy8d1zuRafZ36QVP9KnxehomIfIC0ZZEo,915
@@ -139,15 +147,20 @@ pembot/.git/objects/bf/068a0714e2145de83a5c004f4213b091439d0e,sha256=MpiiCqAk6GQ
 pembot/.git/objects/bf/32a7e6872e5dc4025ee3df3c921ec7ade0855f,sha256=lwL9ickzIFtMJgNKaPp6nTGDlMhPs6fkZTWevQWK_Lc,56
 pembot/.git/objects/bf/518686b06069d2a8abd3689908b7e1a6e16b05,sha256=w-HgdJdX2_ZdiIptJv8BcWdeDEyhl42WEk8P72X8YKU,421
 pembot/.git/objects/c0/793458db6e1bee7f79f1a504fb8ff4963f8ed3,sha256=b8lo_OrMeGgirc9yY_OFjv5xVpG6FBpZnBf7jbtlmyw,421
+pembot/.git/objects/c0/e6cf6ef4d7f0100113213d8bab75b966cd79ef,sha256=CiFonN9bS6HaZOKU8-wi72dUkS4kP1kiII3QeMfyoBo,170
 pembot/.git/objects/c0/f948ab4636a125bc202368e6c9cbe80d76169a,sha256=GPQso_R_RWWLx_pF3g58MiM4HyeSnpXTeLeKDfhkyPc,526
 pembot/.git/objects/c2/443060c07101948487cfa93cc39e082e9e0f5f,sha256=d9rjB8sgBOUQ-HQ8yu5I-c5Dqr_q2z0OOCXSufjDAak,3998
 pembot/.git/objects/c2/926f040b089a52edfb8351480f63619ab7e0ab,sha256=HAXSsWokz2tuk9Y952ogIEzSBlbUC4lZ1CjvWBc22Cg,56
 pembot/.git/objects/c2/ce19d575a8cccf6886862c4fa6afefba142511,sha256=kxbbFUJ1TpEVIrqgiLzepP5Z1k_kF3FjCHvJ04yCBvs,3370
 pembot/.git/objects/c3/cc0da3d955ecec0f865c46c030a0c073697495,sha256=7ZXWsXqapYhbZZJwaaeAwqGcgX8JwoS5DazqOGaRHeQ,179
 pembot/.git/objects/c6/b72ea9f8856d3bde28cb75775ebea9840535b8,sha256=dNygXpD49awTkQZZUIQQ9uQDEN3aKHKrQEPuiyLBT6Q,527
+pembot/.git/objects/c9/d9d9b3a93b142e8b7266fc4e3e2417128a0b32,sha256=v3U1X176tqD2nTW1z3UQyCTradKh9nEb8KH-ZUXQZjE,115
 pembot/.git/objects/c9/efe79dee4c91d4bb8c3d3c6e01ff70ff79a722,sha256=mnJvaTyytYLB1XPeJS_MAY-q0X3karH0k7EeBpUM16k,115
+pembot/.git/objects/cc/348ab3677f744f8d7cd8b2ac7eb775528cfb1c,sha256=2GQxoBSPEaKEZoUWa_3qEockmzUQnuWN9ohEie28VW8,90
 pembot/.git/objects/ce/a4ffc1cf5eab61a2a0abd8f6dc941b580b69fd,sha256=yKUe_ZHD0UynTIrDRhuVqjDjKYDfZkWplqXjeSOD_bk,3894
 pembot/.git/objects/d0/937f7d832266337289d5ec09459f931a46fcf7,sha256=_RZ7Z2EZp1OOF_XZhY6e1tzWwhI8Fa5R9aaF_W8APBA,56
+pembot/.git/objects/d3/508f1537e9bd48bc784da569e14c342bc9c05e,sha256=0VpYRyf_U_go58Y-wncv2bgDw3o9gXBzBBuhft1eBQQ,55
+pembot/.git/objects/d4/40b20aae1265dabbd3ddaafb24c35e40e3ab3c,sha256=gHVL-l-koZnke8dJEK06U6Vz9HjqoP762qB4QUU9wE0,220
 pembot/.git/objects/d9/ec420cb55a82e7efbc8564e30ec7f4c0f6021e,sha256=NfSA8eUrylFWTcpPSd2FoA8_TcifJ_pk3jBYmbfkNKQ,90
 pembot/.git/objects/dd/82bd16a51b9bad8241d9fb46619b1c6755cafe,sha256=36R9xSCSTvnCmK5IhuTacIC8FLQYWkVYh2QwxXJ2e7s,56
 pembot/.git/objects/e0/9162dbd64d85bb5ed740aa99faefa73f293d78,sha256=I5fpz3BQ2maFPTSu43T1uvYMuLiep1C3K6CsX8UMNPI,196
@@ -163,6 +176,7 @@ pembot/.git/objects/ea/0af89e61a882c5afc2a8c281b2d96f174bfe58,sha256=lXbMvL_xl8P
 pembot/.git/objects/eb/75e1c49f1e5b79dca17ccdbec8067756523238,sha256=ltEINFUpQP86CkE4nAT1Afegz3ytY3Nlx1P6ibTFEbo,305
 pembot/.git/objects/ee/a73c7f24094ed83b014f7cfce46e10f817bec8,sha256=fFYq_ODekFhF9SwBL9GP_fGDsNavXVVOuI6kmnHlkiY,5140
 pembot/.git/objects/ef/0503a60244391590b16042019032e91d7cc30d,sha256=mrF9jZHY2oJm8tkd8nQdMgUPbrZfENOFaR3mvbwi1dg,187
+pembot/.git/objects/ef/141ba9482c729796968be2e562b1488a1e5552,sha256=p6CoPdq20oILGmCOolB6HqVsb22FRGz1LnVIi1-hzRk,526
 pembot/.git/objects/ef/3488a3c636d73d82ad138e70a92453249b7f37,sha256=09VHFwSnSxOExWa3sKBzWAfSMyx5INI9KK2mstW8-xs,203
 pembot/.git/objects/ef/c899c7f910cfa7a383692eee851cf5af36da8c,sha256=9o2hAtTleVJrHkDxERUrwEaze8hzc-E2n1Jq2t_Oo68,2853
 pembot/.git/objects/f1/2d2ef8948cbe4b24279bee282f934cf5a1b834,sha256=HvfHm5CnbujMp-uu0aucVzpaWCy2aKBzaXknvlYCnjg,3937
@@ -184,21 +198,21 @@ pembot/.git/objects/fe/cc5d8154b1e77e4c6beb23ce9cbe8fea55d34d,sha256=0it_Z3Lk5Mj
 pembot/.git/objects/pack/pack-d5469edc8c36e3bb1de5e0070e4d5b1eae935dd4.idx,sha256=CNzx_lz6v4PulPxRW2t9nz-ifvplpSFPhMA2M9WNUrA,3424
 pembot/.git/objects/pack/pack-d5469edc8c36e3bb1de5e0070e4d5b1eae935dd4.pack,sha256=dk3Sqrd0L-tNVLRy3uJdTYJNkw8v59mE1hV8zrCFNzc,41355
 pembot/.git/objects/pack/pack-d5469edc8c36e3bb1de5e0070e4d5b1eae935dd4.rev,sha256=7U3tpTWQ3dn5dwQo_KWMWxF31cKaDnCk2AzTO7Cx4Bg,388
-pembot/.git/refs/heads/main,sha256=6zKAFxl4876S5gjd1QK1hg0_Hv1g_W0E92fBZQYFzSQ,41
+pembot/.git/refs/heads/main,sha256=uEuah700WPimZUV8DPQJkB7P2y-1mRSsAFuF4_YiQKY,41
 pembot/.git/refs/remotes/origin/HEAD,sha256=K7aiSqD8bEhBAPXVGim7rYQc0sdV9dk_qiBOXbtOsrQ,30
-pembot/.git/refs/remotes/origin/main,sha256=6zKAFxl4876S5gjd1QK1hg0_Hv1g_W0E92fBZQYFzSQ,41
+pembot/.git/refs/remotes/origin/main,sha256=uEuah700WPimZUV8DPQJkB7P2y-1mRSsAFuF4_YiQKY,41
 pembot/AnyToText/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-pembot/AnyToText/convertor.py,sha256=OT6d2Dyz443cHqTIYagYWpB4T7adfTlkiLTQZs1Q3Tg,8845
+pembot/AnyToText/convertor.py,sha256=1wMzqcBvt6hgjvvdd2evxpFInOy-bmrr8dkOE-fUC4I,9611
 pembot/TextEmbedder/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 pembot/TextEmbedder/gemini_embedder.py,sha256=P679-2mmQESlYKML1vcrwx_-CSgWJgIQk7NL4F7BLQE,677
 pembot/TextEmbedder/mongodb_embedder.py,sha256=-xIr-zrAGzCmgNeojuX6qYj2t019EVO1I6g-Hwq0FL8,10799
 pembot/TextEmbedder/mongodb_index_creator.py,sha256=kopqdVYJii_wExVrXGZjMfqWZ2dD42b3PeNWo71weHI,5354
 pembot/TextEmbedder/vector_query.py,sha256=Kh1uhx9CatB-oQlQtnW-1I2Qz7MGHI20n2h_8peAChM,1986
-pembot/config/config.yaml,sha256=g3KOq-TnkamchzrNcHuEYay0P4oMxjoyiGQY4LSaSPI,156
+pembot/config/config.yaml,sha256=5sjUtvZ3xDMa4qRI1hyLKuBVGdyRiufGMLKfht9OGmw,156
 pembot/pdf2markdown/LICENSE,sha256=1JTJhQjUYDqJzFJhNtitm7mHyE71PRHgetIqRRWg6Pk,1068
 pembot/pdf2markdown/README.md,sha256=jitM1pwI69oa0N4mXv5-SY1ka9Sz3jsRNCDdpW-50kY,4545
 pembot/pdf2markdown/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-pembot/pdf2markdown/extract.py,sha256=0fnZnUqNy7shef6dijqEAMgBSf0YAiEx_mbRSU_pmQg,34418
+pembot/pdf2markdown/extract.py,sha256=oomC2qxisdVuYOtYV4fhF92BBjsfixOViWwBv0khjKM,35038
 pembot/pdf2markdown/pyrightconfig.json,sha256=Vt_k4N2LtZhth0lQOQAOnRKDOQkYYVzmdtb-bP3gu7M,47
 pembot/pdf2markdown/requirements.txt,sha256=0vZQzkSZKLNVUttd4euoDyYEy0nc2W3CIVxhepHW5Ho,76
 pembot/pdf2markdown/.git/COMMIT_EDITMSG,sha256=K6REOtE5mjRmxGSeQdpaFAr1luu1BmaZnzXkTjKINzY,55
@@ -250,7 +264,7 @@ pembot/pdf2markdown/config/config.yaml,sha256=w75W2Eg4-tu8rRk_23PqxWDh0010kRKLmP
 pembot/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 pembot/utils/inference_client.py,sha256=jeURmY2P5heVlH1dCV0XSgiX3U2qYGEmrnUv0KFpdww,5380
 pembot/utils/string_tools.py,sha256=gtRa5rBR0Q7GspTu2WtCnvhJQLFjPfWLvhmyiPkyStU,1883
-pembot-0.1.7.dist-info/licenses/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
-pembot-0.1.7.dist-info/WHEEL,sha256=Dyt6SBfaasWElUrURkknVFAZDHSTwxg3PaTza7RSbkY,100
-pembot-0.1.7.dist-info/METADATA,sha256=m0FazfiKAQQ6sx-4htib7ukF8BKeKPVJa9yi8ObwsgI,313
-pembot-0.1.7.dist-info/RECORD,,
+pembot-0.1.9.dist-info/licenses/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
+pembot-0.1.9.dist-info/WHEEL,sha256=Dyt6SBfaasWElUrURkknVFAZDHSTwxg3PaTza7RSbkY,100
+pembot-0.1.9.dist-info/METADATA,sha256=P4CgwerFdlz7tBVa0Aakloj6F26pkyaTIBsH01ffjak,313
+pembot-0.1.9.dist-info/RECORD,,