PDF to images

Dockerfile

@@ -4,6 +4,7 @@ WORKDIR /code
 
 COPY ./requirements.txt /code/requirements.txt
 
+RUN apt-get update && apt-get install -y poppler-utils
 RUN pip install --no-cache-dir --upgrade -r /code/requirements.txt
 
 COPY . .

app.py

@@ -1,4 +1,7 @@
 import base64
+from pdf2image import convert_from_path
+from io import BytesIO
+from PIL import Image
 
 from llama_index.core.schema import BaseNode
 from shiny import ui, render, App
@@ -37,16 +40,24 @@ data_dir = Path(__file__).parent / "data"
 docstore = SimpleDocumentStore.from_persist_path(str(data_dir / "storage_metadata" / "processed_docstore_storage.json"))
 
 
-def get_pdf_data_url(file_path):
-    """Convert PDF file to data URL for embedding"""
+def get_pdf_as_images(pdf_path):
+    """Convert PDF file to a list of base64 encoded images"""
     try:
-        with open(file_path, "rb") as f:
-            pdf_data = f.read()
-            b64_data = base64.b64encode(pdf_data).decode()
-            return f"data:application/pdf;base64,{b64_data}"
+        # Convert PDF to list of images
+        images = convert_from_path(pdf_path)
+        image_urls = []
+        
+        for img in images:
+            # Convert PIL image to base64
+            buffered = BytesIO()
+            img.save(buffered, format="PNG")
+            img_base64 = base64.b64encode(buffered.getvalue()).decode()
+            image_urls.append(f"data:image/png;base64,{img_base64}")
+        
+        return image_urls
     except Exception as e:
-        logger.error(f"Error reading PDF file: {e}")
-        return None
+        logger.error(f"Error converting PDF to images: {e}")
+        return []
 
 
 def get_str_structure(doc: BaseNode) -> str:
@@ -127,7 +138,15 @@ def server(input, output, session):
                 logger.error(f"Document does not have a filename: {input.selected_doc_id()}")
                 return ui.p("Error: Document does not have a filename")
             pdf_path = data_dir / "pdfs" / filename
-            pdf_data_url = get_pdf_data_url(pdf_path) if pdf_path else None
+            # pdf_data_url = get_pdf_data_url(pdf_path) if pdf_path else None
+            image_urls = get_pdf_as_images(pdf_path) if pdf_path.exists() else []
+            image_elements = [
+                ui.tags.img(
+                    src=img_url,
+                    style="width: 100%; margin-bottom: 10px; border: 1px solid #ddd;"
+                )
+                for img_url in image_urls
+            ]
 
             return ui.div(
                 ui.h3(f"Display panel for: {doc.metadata.get('filename', 'Unknown')}"),
@@ -135,13 +154,10 @@ def server(input, output, session):
                     ui.column(
                         6,  # Left column (PDF)
                         ui.h4("PDF View"),
-                        ui.tags.iframe(
-                            src=pdf_data_url,
-                            style="width: 100%; height: 800px; border: 1px solid #ddd;",
-                            type="application/pdf",
-                        )
-                        if pdf_data_url
-                        else ui.p("No PDF available"),
+                        ui.div(
+                            ui.div(image_elements) if image_elements else ui.p("No PDF available"),
+                            style="height: 800px; overflow-y: auto;"
+                        ),
                     ),
                     ui.column(
                         6,  # Right column (Markdown)

requirements.txt

@@ -1,2 +1,4 @@
 shiny
-llama-index-core
+llama-index-core
+pdf2image
+Pillow