Update app.py

app.py

@@ -1,25 +1,65 @@
+import urllib.request
 import fitz
-import uuid
 import re
 import numpy as np
 import tensorflow_hub as hub
 import openai
 import gradio as gr
-import shutil
 import os
 from sklearn.neighbors import NearestNeighbors
-from tempfile import NamedTemporaryFile
-from PyPDF2 import PdfReader
 
-openAI_key = os.environ['OpenAPI']
+def download_pdf(url, output_path):
+    urllib.request.urlretrieve(url, output_path)
+
+
+def preprocess(text):
+    text = text.replace('\n', ' ')
+    text = re.sub('\s+', ' ', text)
+    return text
+
+
+def pdf_to_text(path, start_page=1, end_page=None):
+    doc = fitz.open(path)
+    total_pages = doc.page_count
+
+    if end_page is None:
+        end_page = total_pages
+
+    text_list = []
+
+    for i in range(start_page-1, end_page):
+        text = doc.load_page(i).get_text("text")
+        text = preprocess(text)
+        text_list.append(text)
+
+    doc.close()
+    return text_list
+
+
+def text_to_chunks(texts, word_length=150, start_page=1):
+    text_toks = [t.split(' ') for t in texts]
+    page_nums = []
+    chunks = []
+
+    for idx, words in enumerate(text_toks):
+        for i in range(0, len(words), word_length):
+            chunk = words[i:i+word_length]
+            if (i+word_length) > len(words) and (len(chunk) < word_length) and (
+                len(text_toks) != (idx+1)):
+                text_toks[idx+1] = chunk + text_toks[idx+1]
+                continue
+            chunk = ' '.join(chunk).strip()
+            chunk = f'[Page no. {idx+start_page}]' + ' ' + '"' + chunk + '"'
+            chunks.append(chunk)
+    return chunks
+
 
 class SemanticSearch:
-    
+
     def __init__(self):
         self.use = hub.load('https://tfhub.dev/google/universal-sentence-encoder/4')
         self.fitted = False
-    
-    
+
     def fit(self, data, batch=1000, n_neighbors=5):
         self.data = data
         self.embeddings = self.get_text_embedding(data, batch=batch)
@@ -27,87 +67,128 @@ class SemanticSearch:
         self.nn = NearestNeighbors(n_neighbors=n_neighbors)
         self.nn.fit(self.embeddings)
         self.fitted = True
-    
-    
+
     def __call__(self, text, return_data=True):
         inp_emb = self.use([text])
         neighbors = self.nn.kneighbors(inp_emb, return_distance=False)[0]
-        
+
         if return_data:
             return [self.data[i] for i in neighbors]
         else:
             return neighbors
-    
-    
+
     def get_text_embedding(self, texts, batch=1000):
         embeddings = []
         for i in range(0, len(texts), batch):
-            text_batch = texts[i:(i+batch)]
-            print(f"Processing batch {i//batch + 1} of {len(texts)//batch + 1}")
-            print(f"Text batch: {text_batch}")
-            emb_batch = self.use(text_batch)
-            print(f"Embedding batch: {emb_batch}")
-            embeddings.append(emb_batch)
-        embeddings = np.vstack(embeddings)
-        print(f"Final embeddings: {embeddings}")
-        return embeddings
+            text_batch = texts[i:(i+batch
+import urllib.request
+import fitz
+import re
+import numpy as np
+import tensorflow_hub as hub
+import openai
+import gradio as gr
+import os
+from sklearn.neighbors import NearestNeighbors
+
+def download_pdf(url, output_path):
+    urllib.request.urlretrieve(url, output_path)
 
 
-def pdf_to_text(pdf_path, start_page=1):
-    pdf = PdfReader(pdf_path)
-    text = ''
-    for i in range(start_page, len(pdf.pages)):
-        text += pdf.pages[i].extract_text()
+def preprocess(text):
+    text = text.replace('\n', ' ')
+    text = re.sub('\s+', ' ', text)
     return text
 
-def text_to_chunks(text, start_page=1, chunk_size=512):
-    chunks = [text[i:i+chunk_size] for i in range(0, len(text), chunk_size)]
+
+def pdf_to_text(path, start_page=1, end_page=None):
+    doc = fitz.open(path)
+    total_pages = doc.page_count
+
+    if end_page is None:
+        end_page = total_pages
+
+    text_list = []
+
+    for i in range(start_page-1, end_page):
+        text = doc.load_page(i).get_text("text")
+        text = preprocess(text)
+        text_list.append(text)
+
+    doc.close()
+    return text_list
+
+
+def text_to_chunks(texts, word_length=150, start_page=1):
+    text_toks = [t.split(' ') for t in texts]
+    page_nums = []
+    chunks = []
+
+    for idx, words in enumerate(text_toks):
+        for i in range(0, len(words), word_length):
+            chunk = words[i:i+word_length]
+            if (i+word_length) > len(words) and (len(chunk) < word_length) and (
+                len(text_toks) != (idx+1)):
+                text_toks[idx+1] = chunk + text_toks[idx+1]
+                continue
+            chunk = ' '.join(chunk).strip()
+            chunk = f'[Page no. {idx+start_page}]' + ' ' + '"' + chunk + '"'
+            chunks.append(chunk)
     return chunks
 
 
+class SemanticSearch:
 
-def unique_filename(basename):
-    # Append a unique ID to the end of the filename, before the extension
-    base, ext = os.path.splitext(basename)
-    return base + "_" + uuid.uuid4().hex + ext
+    def __init__(self):
+        self.use = hub.load('https://tfhub.dev/google/universal-sentence-encoder/4')
+        self.fitted = False
+
+    def fit(self, data, batch=1000, n_neighbors=5):
+        self.data = data
+        self.embeddings = self.get_text_embedding(data, batch=batch)
+        n_neighbors = min(n_neighbors, len(self.embeddings))
+        self.nn = NearestNeighbors(n_neighbors=n_neighbors)
+        self.nn.fit(self.embeddings)
+        self.fitted = True
+
+    def __call__(self, text, return_data=True):
+        inp_emb = self.use([text])
+        neighbors = self.nn.kneighbors(inp_emb, return_distance=False)[0]
+
+        if return_data:
+            return [self.data[i] for i in neighbors]
+        else:
+            return neighbors
+
+    def get_text_embedding(self, texts, batch=1000):
+        embeddings = []
+        for i in range(0, len(texts), batch):
+            text_batch = texts[i:(i+batch
+            emb_batch = self.use(text_batch)
+            embeddings.append(emb_batch)
+        embeddings = np.vstack(embeddings)
+        return embeddings
 
 
-def load_recommender(paths, start_page=1):
+def load_recommender(path, start_page=1):
     global recommender
-    chunks = []
-    for path in paths:
-        pdf_file = os.path.basename(path)
-        embeddings_file = f"{pdf_file}_{start_page}.npy"
-        
-        if os.path.isfile(embeddings_file):
-            embeddings = np.load(embeddings_file)
-            recommender.embeddings = embeddings
-            recommender.fitted = True
-            print("Embeddings loaded from file")
-            continue
-        
-        texts = pdf_to_text(path, start_page=start_page)
-        chunks.extend(text_to_chunks(texts, start_page=start_page))
-    
+    texts = pdf_to_text(path, start_page=start_page)
+    chunks = text_to_chunks(texts, start_page=start_page)
     recommender.fit(chunks)
-    np.save(embeddings_file, recommender.embeddings)
     return 'Corpus Loaded.'
 
 
-def generate_text(openAI_key, prompt, engine="gpt-3.5-turbo"):
+def generate_text(openAI_key, prompt, engine="text-davinci-003"):
     openai.api_key = openAI_key
-    messages = [{'role': 'system', 'content': 'You are a helpful assistant.'},
-                {'role': 'user', 'content': prompt}]
-    
-    completions = openai.ChatCompletion.create(
-        model=engine,
-        messages=messages,
+    completions = openai.Completion.create(
+        engine=engine,
+        prompt=prompt,
         max_tokens=512,
         n=1,
         stop=None,
         temperature=0.7,
     )
-    message = completions.choices[0].message['content']
+    message = completions.choices[0].text
     return message
 
 
@@ -117,80 +198,55 @@ def generate_answer(question, openAI_key):
     prompt += 'search results:\n\n'
     for c in topn_chunks:
         prompt += c + '\n\n'
-        
+
     prompt += "Instructions: Compose a comprehensive reply to the query using the search results given. "\
-              "Make sure the answer is correct and don't output false content. "\
-              "If you do not know the answer - answer 'information not provided' "\
-              "Answer should be short and concise. Answer step-by-step. \n\nQuery: {question}\nAnswer: "
-    
+              "Cite each reference using [ Page Number] notation (every result has this number at the beginning). "\
+              "Citation should be done at the end of each sentence. If the search results mention multiple subjects "\
+              "with the same name, create separate answers for each. Only include information found in the results and "\
+              "don't add any additional information. Make sure the answer is correct and don't output false content. "\
+              "If the text does not relate to the query, simply state 'Text Not Found in PDF'. Ignore outlier "\
+              "search results which has nothing to do with the question. Only answer what is asked. The "\
+              "answer should be short and concise. Answer step-by-step. \n\nQuery: {question}\nAnswer: "
+
     prompt += f"Query: {question}\nAnswer:"
-    answer = generate_text(openAI_key, prompt, "gpt-3.5-turbo")
+    answer = generate_text(openAI_key, prompt, "text-davinci-003")
     return answer
 
 
-def main_loop(url: str, files: list, question:
-str, openAI_key):
-    paths = []
+def question_answer(url, file, question, openAI_key):
+    if openAI_key.strip() == '':
+        return '[ERROR]: Please enter your Open AI Key. Get your key here: https://platform.openai.com/account/api-keys'
+    if url.strip() == '' and file is None:
+        return '[ERROR]: Both URL and PDF are empty. Provide at least one.'
+
+    if url.strip() != '' and file is not None:
+        return '[ERROR]: Both URL and PDF are provided. Please provide only one (either URL or PDF).'
 
     if url.strip() != '':
         glob_url = url
         download_pdf(glob_url, 'corpus.pdf')
-        paths.append('corpus.pdf')
-    if files is not None and len(files) > 0:
-        for file in files:
-            old_file_name = file.name
-            file_name = old_file_name[:-12] + old_file_name[-4:]
-            file_name = unique_filename(file_name)  # Ensure the new file name is unique
+        load_recommender('corpus.pdf')
 
-            # Copy the content of the old file to the new file and delete the old file
-            with open(old_file_name, 'rb') as src, open(file_name, 'wb') as dst:
-                shutil.copyfileobj(src, dst)
-            os.remove(old_file_name)
+    else:
+        old_file_name = file.name
+        file_name = file.name
+        file_name = file_name[:-12] + file_name[-4:]
+        os.rename(old_file_name, file_name)
+        load_recommender(file_name)
 
-            paths.append(file_name)
+    if question.strip() == '':
+        return '[ERROR]: Question field is empty'
 
-    load_recommender(paths)
-
-    if question.strip().lower() == 'exit':
-        return '', False
-    
-    answer = generate_answer(question, openAI_key)
-    return answer, True  # Assuming the function returns an answer in all other cases
-
-
-def on_click(*args):
-    answer.value = main_loop(url.value, files.value, question.value)
+    return generate_answer(question, openAI_key)
 
 
 recommender = SemanticSearch()
 
-title = 'Cognitive pdfGPT'
-description = """ Why use Cognitive Ask an Expert?
-This is Cognitive Chat. Here you can upload multiple PDF files and query them as a single corpus of knowledge. 🛑DO NOT USE CONFIDENTIAL INFORMATION """
-
-
-with gr.Blocks() as demo:
-    gr.Markdown(f'<center><h1>{title}</h1></center>')
-    gr.Markdown(description)
-
-    with gr.Row():
-
-        with gr.Group():
-            files = gr.Files(label='➡️ Upload your PDFs ⬅️  NO CONFIDENTIAL FILES ', file_types=['.pdf'])
-            url = gr.Textbox(label=' ')
-            question = gr.Textbox(label='🔤 Enter your question here 🔤')
-            btn = gr.Button(value='Submit')
-            btn.style(full_width=False)
-
-        with gr.Group():
-            gr.Image("logo.jpg") 
-            answer = gr.Textbox(label='The answer to your question is :')
-            
-        btn.click(main_loop, inputs=[url, files, question, openAI_key], outputs=[answer])
-
-
-     
+title = 'PDF GPT'
+description = """ PDF GPT allows you to chat with your PDF file using Universal Sentence Encoder and Open AI. It gives hallucination free response than other tools as the embeddings are better than OpenAI. The returned response can even cite the page number in square brackets([]) where the information is located, adding credibility to the responses and helping to locate pertinent information quickly."""
 
-demo.launch(share=False, debug=True, auth=None, auth_message=None)
+with gr.Interface(fn=question_answer, inputs
+    =[url, file, question, openAI_key], outputs=[answer], title=title, description=description) as iface:
+        iface.launch()