Duplicate from Raghav001/PDF

Co-authored-by: Raghavan <Raghav001@users.noreply.huggingface.co>

.gitattributes

@@ -0,0 +1,34 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,14 @@
+---
+title: ChatPDF
+emoji: 💻
+colorFrom: gray
+colorTo: yellow
+sdk: gradio
+sdk_version: 3.20.0
+app_file: app.py
+pinned: false
+license: apache-2.0
+duplicated_from: Raghav001/PDF
+---
+
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -0,0 +1,312 @@
+import requests
+import json
+import gradio as gr
+# from concurrent.futures import ThreadPoolExecutor
+import pdfplumber
+import pandas as pd
+import langchain
+import time
+from cnocr import CnOcr
+import pinecone
+import openai
+from langchain.vectorstores import Pinecone
+from langchain.embeddings.openai import OpenAIEmbeddings
+from langchain.text_splitter import CharacterTextSplitter
+
+# from langchain.document_loaders import PyPDFLoader
+from langchain.document_loaders import UnstructuredWordDocumentLoader
+from langchain.document_loaders import UnstructuredPowerPointLoader
+# from langchain.document_loaders.image import UnstructuredImageLoader
+
+
+from langchain.chains.question_answering import load_qa_chain
+from langchain import OpenAI
+
+from sentence_transformers import SentenceTransformer, models, util
+word_embedding_model = models.Transformer('sentence-transformers/all-MiniLM-L6-v2', do_lower_case=True)
+pooling_model = models.Pooling(word_embedding_model.get_word_embedding_dimension(), pooling_mode='cls')
+embedder = SentenceTransformer(modules=[word_embedding_model, pooling_model])
+ocr = CnOcr()
+# chat_url = 'https://Raghav001-API.hf.space/sale'
+chat_url = 'https://Raghav001-API.hf.space/chatpdf'
+chat_emd = 'https://Raghav001-API.hf.space/embedd'
+headers = {
+    'Content-Type': 'application/json',
+}
+# thread_pool_executor = ThreadPoolExecutor(max_workers=4)
+history_max_len = 500
+all_max_len = 3000
+
+
+
+# Initialize Pinecone client and create an index
+pinecone.init(api_key="ffb1f594-0915-4ebf-835f-c1eaa62fdcdc",environment = "us-west4-gcp-free")
+index = pinecone.Index(index_name="test")    
+
+
+def get_emb(text):
+    emb_url = 'https://Raghav001-API.hf.space/embeddings'
+    data = {"content": text}
+    try:
+        result = requests.post(url=emb_url,
+                               data=json.dumps(data),
+                               headers=headers
+                               )
+        print("--------------------------------Embeddings-----------------------------------")
+        print(result.json()['data'][0]['embedding'])
+        return result.json()['data'][0]['embedding']
+    except Exception as e:
+        print('data', data, 'result json', result.json())
+
+
+def doc_emb(doc: str):
+    texts = doc.split('\n')
+    # futures = []
+    emb_list = embedder.encode(texts)
+    print('emb_list',emb_list)
+    # for text in texts:
+    #     futures.append(thread_pool_executor.submit(get_emb, text))
+    # for f in futures:
+    #     emb_list.append(f.result())
+    print('\n'.join(texts))
+    pine(doc)
+    gr.Textbox.update(value="")
+    return texts, emb_list, gr.Textbox.update(visible=True), gr.Button.update(visible=True), gr.Markdown.update(
+        value="""success ! Let's talk"""), gr.Chatbot.update(visible=True)
+
+
+def get_response(msg, bot, doc_text_list, doc_embeddings):
+    # future = thread_pool_executor.submit(get_emb, msg)
+    gr.Textbox.update(value="")
+    now_len = len(msg)
+    req_json = {'question': msg}
+    his_bg = -1
+    for i in range(len(bot) - 1, -1, -1):
+        if now_len + len(bot[i][0]) + len(bot[i][1]) > history_max_len:
+            break
+        now_len += len(bot[i][0]) + len(bot[i][1])
+        his_bg = i
+    req_json['history'] = [] if his_bg == -1 else bot[his_bg:]
+    # query_embedding = future.result()
+    query_embedding = embedder.encode([msg])
+    cos_scores = util.cos_sim(query_embedding, doc_embeddings)[0]
+    score_index = [[score, index] for score, index in zip(cos_scores, [i for i in range(len(cos_scores))])]
+    score_index.sort(key=lambda x: x[0], reverse=True)
+    print('score_index:\n', score_index)
+    print('doc_emb_state', doc_emb_state)
+    index_set, sub_doc_list = set(), []
+    for s_i in score_index:
+        doc = doc_text_list[s_i[1]]
+        if now_len + len(doc) > all_max_len:
+            break
+        index_set.add(s_i[1])
+        now_len += len(doc)
+       # Maybe the paragraph is truncated wrong, so add the upper and lower paragraphs
+        if s_i[1] > 0 and s_i[1] -1 not in index_set:
+            doc = doc_text_list[s_i[1]-1]
+            if now_len + len(doc) > all_max_len:
+                break
+            index_set.add(s_i[1]-1)
+            now_len += len(doc)
+        if s_i[1] + 1 < len(doc_text_list) and s_i[1] + 1 not in index_set:
+            doc = doc_text_list[s_i[1]+1]
+            if now_len + len(doc) > all_max_len:
+                break
+            index_set.add(s_i[1]+1)
+            now_len += len(doc)
+
+    index_list = list(index_set)
+    index_list.sort()
+    for i in index_list:
+        sub_doc_list.append(doc_text_list[i])
+    req_json['doc'] = '' if len(sub_doc_list) == 0 else '\n'.join(sub_doc_list)
+    data = {"content": json.dumps(req_json)}
+    print('data:\n', req_json)
+    result = requests.post(url=chat_url,
+                           data=json.dumps(data),
+                           headers=headers
+                           )
+    res = result.json()['content']
+    bot.append([msg, res])
+    return bot[max(0, len(bot) - 3):]
+
+
+def up_file(fls):
+    doc_text_list = []
+
+    
+    names = []
+    print(names)
+    for i in fls:
+        names.append(str(i.name))
+
+    
+    pdf = []
+    docs = []
+    pptx = []
+
+    for i in names:
+        
+        if i[-3:] == "pdf":
+            pdf.append(i)
+        elif i[-4:] == "docx":
+            docs.append(i)
+        else:
+            pptx.append(i)
+
+
+    #Pdf Extracting
+    for idx, file in enumerate(pdf):
+        print("11111")
+        #print(file.name)
+        with pdfplumber.open(file) as pdf:
+            for i in range(len(pdf.pages)):
+                # Read page i+1 of a PDF document
+                page = pdf.pages[i]
+                res_list = page.extract_text().split('\n')[:-1]
+
+                for j in range(len(page.images)):
+                   # Get the binary stream of the image
+                    img = page.images[j]
+                    file_name = '{}-{}-{}.png'.format(str(time.time()), str(i), str(j))
+                    with open(file_name, mode='wb') as f:
+                        f.write(img['stream'].get_data())
+                    try:
+                        res = ocr.ocr(file_name)
+                        # res = PyPDFLoader(file_name)
+                    except Exception as e:
+                        res = []
+                    if len(res) > 0:
+                        res_list.append(' '.join([re['text'] for re in res]))
+
+                tables = page.extract_tables()
+                for table in tables:
+                    # The first column is used as the header
+                    df = pd.DataFrame(table[1:], columns=table[0])
+                    try:
+                        records = json.loads(df.to_json(orient="records", force_ascii=False))
+                        for rec in records:
+                            res_list.append(json.dumps(rec, ensure_ascii=False))
+                    except Exception as e:
+                        res_list.append(str(df))
+
+                doc_text_list += res_list
+
+        #pptx Extracting
+    for i in pptx:
+        loader = UnstructuredPowerPointLoader(i)
+        data = loader.load()
+        # content = str(data).split("'")
+        # cnt = content[1]
+        # # c = cnt.split('\\n\\n')
+        # # final = "".join(c)
+        # c = cnt.replace('\\n\\n',"").replace("<PAGE BREAK>","").replace("\t","")
+        doc_text_list.append(data)
+
+    
+
+    #Doc Extracting
+    for i in docs:
+        loader = UnstructuredWordDocumentLoader(i)
+        data = loader.load()
+        # content = str(data).split("'")
+        # cnt = content[1]
+        # # c = cnt.split('\\n\\n')
+        # # final = "".join(c)
+        # c = cnt.replace('\\n\\n',"").replace("<PAGE BREAK>","").replace("\t","")
+        doc_text_list.append(data)
+
+    # #Image Extraction
+    # for i in jpg:
+    #     loader = UnstructuredImageLoader(i)
+    #     data = loader.load()
+    #     # content = str(data).split("'")
+    #     # cnt = content[1]
+    #     # # c = cnt.split('\\n\\n')
+    #     # # final = "".join(c)
+    #     # c = cnt.replace('\\n\\n',"").replace("<PAGE BREAK>","").replace("\t","")
+    #     doc_text_list.append(data)
+                
+    doc_text_list = [str(text).strip() for text in doc_text_list if len(str(text).strip()) > 0]
+    # print(doc_text_list)
+    return gr.Textbox.update(value='\n'.join(doc_text_list), visible=True), gr.Button.update(
+        visible=True), gr.Markdown.update(
+        value="Processing")
+
+
+def pine(data):
+    char_text_spliter = CharacterTextSplitter(chunk_size = 1000, chunk_overlap=0)
+    # doc_text = char_text_spliter.split_documents(data)
+    doc_spilt = []
+    data = data.split(" ")
+    # print(len(data))
+    
+    c = 0
+    check = 0
+    for i in data:
+      # print(i)
+      if c == 350:
+        text = " ".join(data[check: check + c])
+        print(text)
+        print(check)
+        doc_spilt.append(text)
+        check = check + c
+        c = 0
+      else:
+        c = c+1
+
+
+    Embedding_model = "text-embedding-ada-002"
+    embeddings = OpenAIEmbeddings(openai_api_key="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")
+
+    print(requests.post(url = chat_emd))
+
+    # embeddings = requests.post(url=chat_emd,
+    #                        data=json.dumps(data),
+    #                        headers=headers
+    #                        )
+
+    pinecone.init(api_key = "ffb1f594-0915-4ebf-835f-c1eaa62fdcdc",
+              environment = "us-west4-gcp-free"
+              )
+
+    index_name = "test"
+    docstore = Pinecone.from_texts([d for d in doc_spilt],embeddings,index_name = index_name,namespace='a1')
+
+    return ''
+
+def get_answer(query_live):
+    
+    llm = OpenAI(temperature=0, openai='aaa')
+    qa_chain = load_qa_chain(llm,chain_type='stuff')
+    query = query_live
+    docs = docstore.similarity_search(query)
+    qa_chain.run(input_documents = docs, question = query)
+
+with gr.Blocks() as demo:
+    with gr.Row():
+        with gr.Column():
+            file = gr.File(file_types=['.pdf'], label='Click to upload Document', file_count='multiple')
+            doc_bu = gr.Button(value='Submit', visible=False)
+
+            
+            txt = gr.Textbox(label='result', visible=False)
+            
+            
+            doc_text_state = gr.State([])
+            doc_emb_state = gr.State([])
+            
+        with gr.Column():
+            md = gr.Markdown("Please Upload the PDF")
+            chat_bot = gr.Chatbot(visible=False)
+            msg_txt = gr.Textbox(visible = False)
+            chat_bu = gr.Button(value='Clear', visible=False)
+
+    file.change(up_file, [file], [txt, doc_bu, md]) #hiding the text
+    doc_bu.click(doc_emb, [txt], [doc_text_state, doc_emb_state, msg_txt, chat_bu, md, chat_bot])
+    msg_txt.submit(get_response, [msg_txt, chat_bot,doc_text_state, doc_emb_state], [chat_bot],queue=False)
+    chat_bu.click(lambda: None, None, chat_bot, queue=False)
+
+if __name__ == "__main__":
+    demo.queue().launch(show_api=False)
+    # demo.queue().launch(share=False, server_name='172.22.2.54', server_port=9191)

requirements.txt

@@ -0,0 +1,8 @@
+pdfplumber
+sentence_transformers
+cnocr
+langchain
+unstructured
+pinecone-client
+openai
+tiktoken