ai: Support scanning documents.

app.py

@@ -20,6 +20,15 @@ import os
 import threading
 import random
 import time
+import pytesseract
+import pdfplumber
+import docx
+import pandas as pd
+import pptx
+import fitz
+import io
+from pathlib import Path
+from PIL import Image
 
 LINUX_SERVER_HOSTS = [host for host in json.loads(os.getenv("LINUX_SERVER_HOST", "[]")) if host]
 LINUX_SERVER_PROVIDER_KEYS = [key for key in json.loads(os.getenv("LINUX_SERVER_PROVIDER_KEY", "[]")) if key]
@@ -40,6 +49,84 @@ session = requests.Session()
 def get_model_key(display_name):
     return next((k for k, v in MODEL_MAPPING.items() if v == display_name), MODEL_CHOICES[0])
 
+def extract_text(file_path):
+    ext = Path(file_path).suffix.lower()
+
+    if ext == ".txt":
+        try:
+            with open(file_path, "r", encoding="utf-8") as file:
+                return file.read()
+        except:
+            return ""
+
+    elif ext == ".pdf":
+        text = []
+        try:
+            with pdfplumber.open(file_path) as pdf:
+                for page in pdf.pages:
+                    text.append(page.extract_text() or "")
+            if not "".join(text).strip():
+                text = extract_text_from_pdf_images(file_path)
+        except:
+            return ""
+        return "\n".join(text)
+
+    elif ext in [".doc", ".docx"]:
+        try:
+            doc = docx.Document(file_path)
+            text = "\n".join([para.text for para in doc.paragraphs])
+            if not text.strip():
+                text = extract_text_from_doc_images(file_path)
+            return text
+        except:
+            return ""
+
+    elif ext in [".xls", ".xlsx"]:
+        try:
+            df = pd.read_excel(file_path)
+            return df.to_string()
+        except:
+            return ""
+
+    elif ext in [".ppt", ".pptx"]:
+        try:
+            prs = pptx.Presentation(file_path)
+            text = []
+            for slide in prs.slides:
+                for shape in slide.shapes:
+                    if hasattr(shape, "text"):
+                        text.append(shape.text)
+            return "\n".join(text)
+        except:
+            return ""
+
+    return ""
+
+def extract_text_from_pdf_images(pdf_path):
+    text = []
+    try:
+        doc = fitz.open(pdf_path)
+        for page_num in range(len(doc)):
+            pix = doc[page_num].get_pixmap()
+            img = Image.frombytes("RGB", [pix.width, pix.height], pix.samples)
+            text.append(pytesseract.image_to_string(img))
+    except:
+        return []
+    return text
+
+def extract_text_from_doc_images(doc_path):
+    text = []
+    try:
+        doc = docx.Document(doc_path)
+        for rel in doc.part.rels:
+            if "image" in doc.part.rels[rel].target_ref:
+                img_data = doc.part.rels[rel].target_part.blob
+                img = Image.open(io.BytesIO(img_data))
+                text.append(pytesseract.image_to_string(img))
+    except:
+        return []
+    return "\n".join(text)
+
 def simulate_streaming_response(text):
     for line in text.splitlines():
         if stop_event.is_set():
@@ -64,7 +151,7 @@ def chat_with_model(history, user_input, selected_model_display):
     messages.append({"role": "user", "content": user_input})
 
     data = {"model": selected_model, "messages": messages, **model_config}
-    
+
     random.shuffle(LINUX_SERVER_PROVIDER_KEYS)
     random.shuffle(LINUX_SERVER_HOSTS)
 
@@ -87,18 +174,21 @@ def chat_with_model(history, user_input, selected_model_display):
 
     yield RESPONSES["RESPONSE_3"]
 
-def respond(user_input, history, selected_model_display):
-    if not user_input.strip():
+def respond(user_input, file_path, history, selected_model_display):
+    file_text = extract_text(file_path) if file_path else ""
+    combined_input = f"{user_input}\n\n{file_text}".strip()
+
+    if not combined_input:
         yield history, gr.update(value=""), gr.update(visible=False, interactive=False), gr.update(visible=True)
         return
 
     stop_event.clear()
-    history.append([user_input, RESPONSES["RESPONSE_8"]])
+    history.append([combined_input, RESPONSES["RESPONSE_8"]])
 
     yield history, gr.update(value=""), gr.update(visible=False), gr.update(visible=True)
 
     ai_response = ""
-    for chunk in chat_with_model(history, user_input, selected_model_display):
+    for chunk in chat_with_model(history, combined_input, selected_model_display):
         if stop_event.is_set():
             history[-1][1] = RESPONSES["RESPONSE_1"]
             yield history, gr.update(value=""), gr.update(visible=True), gr.update(visible=False)
@@ -116,8 +206,8 @@ def stop_response():
 def change_model(new_model_display):
     return [], new_model_display
 
-def check_send_button_enabled(msg):
-    return gr.update(visible=bool(msg.strip()), interactive=bool(msg.strip()))
+def check_send_button_enabled(msg, file):
+    return gr.update(visible=bool(msg.strip()) or bool(file), interactive=bool(msg.strip()) or bool(file))
 
 with gr.Blocks(fill_height=True, fill_width=True, title=AI_TYPES["AI_TYPE_4"], head=META_TAGS) as demo:
     user_history = gr.State([])
@@ -131,9 +221,13 @@ with gr.Blocks(fill_height=True, fill_width=True, title=AI_TYPES["AI_TYPE_4"], h
         send_btn = gr.Button(RESPONSES["RESPONSE_6"], visible=True, interactive=False)
         stop_btn = gr.Button(RESPONSES["RESPONSE_7"], variant=RESPONSES["RESPONSE_9"], visible=False)
 
+    with gr.Accordion("See more...", open=False):
+        file_upload = gr.File(label=AI_TYPES["AI_TYPE_5"], file_count="single", type="filepath")
+
     model_dropdown.change(fn=change_model, inputs=[model_dropdown], outputs=[user_history, selected_model])
-    send_btn.click(respond, inputs=[msg, user_history, selected_model], outputs=[chatbot, msg, send_btn, stop_btn])
-    msg.change(fn=check_send_button_enabled, inputs=[msg], outputs=[send_btn])
+    send_btn.click(respond, inputs=[msg, file_upload, user_history, selected_model], outputs=[chatbot, msg, send_btn, stop_btn])
+    msg.change(fn=check_send_button_enabled, inputs=[msg, file_upload], outputs=[send_btn])
     stop_btn.click(fn=stop_response, outputs=[send_btn, stop_btn])
+    file_upload.change(fn=check_send_button_enabled, inputs=[msg, file_upload], outputs=[send_btn])
 
-demo.launch(show_api=False)
+demo.launch(show_api=False, max_file_size="1mb")

requirements.txt

@@ -1 +1,8 @@
 huggingface_hub
+pytesseract
+pdfplumber
+python-docx
+pandas
+python-pptx
+PyMuPDF
+Pillow