Merge in master

chunk_convert.py

@@ -1,4 +1,4 @@
-from marker.scripts import chunk_convert_cli
+from marker.scripts.chunk_convert import chunk_convert_cli
 
 if __name__ == "__main__":
     chunk_convert_cli()

convert.py

@@ -1,4 +1,4 @@
-from marker.scripts import convert_cli
+from marker.scripts.convert import convert_cli
 
 if __name__ == "__main__":
     convert_cli()

convert_single.py

@@ -1,4 +1,4 @@
-from marker.scripts import convert_single_cli
+from marker.scripts.convert_single import convert_single_cli
 
 if __name__ == "__main__":
     convert_single_cli()

marker/scripts/__init__.py

@@ -1,5 +0,0 @@
-from marker.scripts.convert_single import convert_single_cli
-from marker.scripts.convert import convert_cli
-from marker.scripts.server import server_cli
-from marker.scripts.run_streamlit_app import streamlit_app_cli
-from marker.scripts.chunk_convert import chunk_convert_cli

marker/scripts/server.py

@@ -3,7 +3,6 @@ import traceback
 import click
 import os
 
-import uvicorn
 from pydantic import BaseModel, Field
 from starlette.responses import HTMLResponse
 
@@ -163,6 +162,7 @@ async def convert_pdf_upload(
 @click.option("--port", type=int, default=8000, help="Port to run the server on")
 @click.option("--host", type=str, default="127.0.0.1", help="Host to run the server on")
 def server_cli(port: int, host: str):
+    import uvicorn
     # Run the server
     uvicorn.run(
         app,

marker/scripts/streamlit_app.py

@@ -68,15 +68,12 @@ def markdown_insert_images(markdown, images):
 def get_page_image(pdf_file, page_num, dpi=96):
     if "pdf" in pdf_file.type:
         doc = open_pdf(pdf_file)
-        renderer = doc.render(
-            pypdfium2.PdfBitmap.to_pil,
-            page_indices=[page_num],
+        page = doc[page_num]
+        png_image = page.render(
             scale=dpi / 72,
-        )
-        png = list(renderer)[0]
-        png_image = png.convert("RGB")
+        ).to_pil().convert("RGB")
     else:
-        png_image = Image.open(in_file).convert("RGB")
+        png_image = Image.open(pdf_file).convert("RGB")
     return png_image
 
 

marker_app.py

@@ -1,4 +1,4 @@
-from marker.scripts import streamlit_app_cli
+from marker.scripts.run_streamlit_app import streamlit_app_cli
 
 if __name__ == "__main__":
     streamlit_app_cli()

marker_server.py

@@ -1,4 +1,4 @@
-from marker.scripts import server_cli
+from marker.scripts.server import server_cli
 
 if __name__ == "__main__":
     server_cli()

pyproject.toml

@@ -23,13 +23,12 @@ transformers = "^4.45.2"
 python-dotenv = "^1.0.0"
 torch = "^2.5.1"
 tqdm = "^4.66.1"
-tabulate = "^0.9.0"
 ftfy = "^6.1.1"
 texify = "^0.2.1"
 rapidfuzz = "^3.8.1"
-surya-ocr = "~0.8.3"
+surya-ocr = "~0.9.0"
 regex = "^2024.4.28"
-pdftext = "~0.4.1"
+pdftext = "~0.5.0"
 markdownify = "^0.13.1"
 click = "^8.1.7"
 google-generativeai = "^0.8.3"
@@ -49,6 +48,7 @@ pytest-mock = "^3.14.0"
 apted = "1.0.3"
 distance = "0.1.3"
 lxml = "5.3.0"
+tabulate = "^0.9.0"
 
 [tool.poetry.scripts]
 marker = "marker.scripts.convert:convert_cli"

signatures/version1/cla.json

@@ -111,6 +111,38 @@
       "created_at": "2024-12-05T13:13:34Z",
       "repoId": 712111618,
       "pullRequestNo": 416
+    },
+    {
+      "name": "tarun-menta",
+      "id": 66506307,
+      "comment_id": 2543907406,
+      "created_at": "2024-12-15T15:06:32Z",
+      "repoId": 712111618,
+      "pullRequestNo": 427
+    },
+    {
+      "name": "ZeyuTeng96",
+      "id": 96521059,
+      "comment_id": 2567236036,
+      "created_at": "2025-01-02T02:36:02Z",
+      "repoId": 712111618,
+      "pullRequestNo": 452
+    },
+    {
+      "name": "xiaoyao9184",
+      "id": 6614349,
+      "comment_id": 2571623521,
+      "created_at": "2025-01-05T13:15:34Z",
+      "repoId": 712111618,
+      "pullRequestNo": 463
+    },
+    {
+      "name": "yasyf",
+      "id": 709645,
+      "comment_id": 2571679069,
+      "created_at": "2025-01-05T16:23:12Z",
+      "repoId": 712111618,
+      "pullRequestNo": 464
     }
   ]
 }