add: pdf2images

app-ocr.py

@@ -0,0 +1,149 @@
+import base64
+import io
+import os
+import shutil
+import time
+import uuid
+from pathlib import Path
+
+# import numpy as np
+# import tempfile
+# from PIL import Image
+import gradio as gr
+from modelscope import AutoModel, AutoTokenizer
+
+UPLOAD_FOLDER = "./uploads"
+RESULTS_FOLDER = "./results"
+
+
+tokenizer = AutoTokenizer.from_pretrained("stepfun-ai/GOT-OCR2_0", trust_remote_code=True)
+model = AutoModel.from_pretrained("stepfun-ai/GOT-OCR2_0", trust_remote_code=True, low_cpu_mem_usage=True, device_map="cuda", use_safetensors=True)
+model = model.eval().cuda()
+
+for folder in [UPLOAD_FOLDER, RESULTS_FOLDER]:
+    if not os.path.exists(folder):
+        os.makedirs(folder)
+
+
+def image_to_base64(image):
+    buffered = io.BytesIO()
+    image.save(buffered, format="PNG")
+    return base64.b64encode(buffered.getvalue()).decode()
+
+
+def run_GOT(image, got_mode, fine_grained_mode="", ocr_color="", ocr_box=""):
+    unique_id = str(uuid.uuid4())
+    image_path = os.path.join(UPLOAD_FOLDER, f"{unique_id}.png")
+    result_path = os.path.join(RESULTS_FOLDER, f"{unique_id}.html")
+
+    shutil.copy(image, image_path)
+
+    try:
+        if got_mode == "plain texts OCR":
+            res = model.chat(tokenizer, image_path, ocr_type="ocr")
+            return res, None
+        elif got_mode == "format texts OCR":
+            res = model.chat(tokenizer, image_path, ocr_type="format", render=True, save_render_file=result_path)
+        elif got_mode == "plain multi-crop OCR":
+            res = model.chat_crop(tokenizer, image_path, ocr_type="ocr")
+            return res, None
+        elif got_mode == "format multi-crop OCR":
+            res = model.chat_crop(tokenizer, image_path, ocr_type="format", render=True, save_render_file=result_path)
+        elif got_mode == "plain fine-grained OCR":
+            res = model.chat(tokenizer, image_path, ocr_type="ocr", ocr_box=ocr_box, ocr_color=ocr_color)
+            return res, None
+        elif got_mode == "format fine-grained OCR":
+            res = model.chat(tokenizer, image_path, ocr_type="format", ocr_box=ocr_box, ocr_color=ocr_color, render=True, save_render_file=result_path)
+
+        # res_markdown = f"$$ {res} $$"
+        res_markdown = res
+
+        if "format" in got_mode and os.path.exists(result_path):
+            with open(result_path, "r") as f:
+                html_content = f.read()
+            encoded_html = base64.b64encode(html_content.encode("utf-8")).decode("utf-8")
+            iframe_src = f"data:text/html;base64,{encoded_html}"
+            iframe = f'<iframe src="{iframe_src}" width="100%" height="600px"></iframe>'
+            download_link = f'<a href="data:text/html;base64,{encoded_html}" download="result_{unique_id}.html">Download Full Result</a>'
+            return res_markdown, f"{download_link}<br>{iframe}"
+        else:
+            return res_markdown, None
+    except Exception as e:
+        return f"Error: {str(e)}", None
+    finally:
+        if os.path.exists(image_path):
+            os.remove(image_path)
+
+
+def task_update(task):
+    if "fine-grained" in task:
+        return [
+            gr.update(visible=True),
+            gr.update(visible=False),
+            gr.update(visible=False),
+        ]
+    else:
+        return [
+            gr.update(visible=False),
+            gr.update(visible=False),
+            gr.update(visible=False),
+        ]
+
+
+def fine_grained_update(task):
+    if task == "box":
+        return [
+            gr.update(visible=False, value=""),
+            gr.update(visible=True),
+        ]
+    elif task == "color":
+        return [
+            gr.update(visible=True),
+            gr.update(visible=False, value=""),
+        ]
+
+
+def cleanup_old_files():
+    current_time = time.time()
+    for folder in [UPLOAD_FOLDER, RESULTS_FOLDER]:
+        for file_path in Path(folder).glob("*"):
+            if current_time - file_path.stat().st_mtime > 3600:  # 1 hour
+                file_path.unlink()
+
+
+with gr.Blocks() as demo:
+    with gr.Row():
+        with gr.Column():
+            image_input = gr.Image(type="filepath", label="上传图片")
+            task_dropdown = gr.Dropdown(
+                choices=[
+                    "plain texts OCR",
+                    "format texts OCR",
+                    "plain multi-crop OCR",
+                    "format multi-crop OCR",
+                    "plain fine-grained OCR",
+                    "format fine-grained OCR",
+                ],
+                label="选择GOT模式",
+                value="plain texts OCR",
+            )
+            fine_grained_dropdown = gr.Dropdown(choices=["box", "color"], label="fine-grained type", visible=False)
+            color_dropdown = gr.Dropdown(choices=["red", "green", "blue"], label="color list", visible=False)
+            box_input = gr.Textbox(label="input box: [x1,y1,x2,y2]", placeholder="e.g., [0,0,100,100]", visible=False)
+            submit_button = gr.Button("Submit")
+
+        with gr.Column():
+            ocr_result = gr.Textbox(label="GOT output")
+
+    with gr.Column():
+        gr.Markdown("**如果选择带格式的模式，mathpix结果将自动呈现如下:**")
+        html_result = gr.HTML(label="rendered html", show_label=True)
+
+    task_dropdown.change(task_update, inputs=[task_dropdown], outputs=[fine_grained_dropdown, color_dropdown, box_input])
+    fine_grained_dropdown.change(fine_grained_update, inputs=[fine_grained_dropdown], outputs=[color_dropdown, box_input])
+
+    submit_button.click(run_GOT, inputs=[image_input, task_dropdown, fine_grained_dropdown, color_dropdown, box_input], outputs=[ocr_result, html_result])
+
+if __name__ == "__main__":
+    cleanup_old_files()
+    demo.launch()

app.py

@@ -1,150 +1,49 @@
-import base64
-import io
 import os
-import shutil
-import time
 import uuid
-from pathlib import Path
 
+import fitz  # PyMuPDF
 import gradio as gr
-from modelscope import AutoModel, AutoTokenizer
-
-# import numpy as np
-# import tempfile
-# from PIL import Image
-
-
-tokenizer = AutoTokenizer.from_pretrained("stepfun-ai/GOT-OCR2_0", trust_remote_code=True)
-model = AutoModel.from_pretrained("stepfun-ai/GOT-OCR2_0", trust_remote_code=True, low_cpu_mem_usage=True, device_map="cuda", use_safetensors=True)
-model = model.eval().cuda()
+from PIL import Image
 
 UPLOAD_FOLDER = "./uploads"
 RESULTS_FOLDER = "./results"
 
-for folder in [UPLOAD_FOLDER, RESULTS_FOLDER]:
-    if not os.path.exists(folder):
-        os.makedirs(folder)
-
-
-def image_to_base64(image):
-    buffered = io.BytesIO()
-    image.save(buffered, format="PNG")
-    return base64.b64encode(buffered.getvalue()).decode()
-
-
-def run_GOT(image, got_mode, fine_grained_mode="", ocr_color="", ocr_box=""):
-    unique_id = str(uuid.uuid4())
-    image_path = os.path.join(UPLOAD_FOLDER, f"{unique_id}.png")
-    result_path = os.path.join(RESULTS_FOLDER, f"{unique_id}.html")
 
-    shutil.copy(image, image_path)
+def pdf_to_images(pdf_path):
+    images = []
+    pdf_document = fitz.open(pdf_path)
+    for page_num in range(len(pdf_document)):
+        page = pdf_document.load_page(page_num)
+        pix = page.get_pixmap()
+        img = Image.frombytes("RGB", [pix.width, pix.height], pix.samples)
+        images.append(img)
+    pdf_document.close()
+    return images
 
-    try:
-        if got_mode == "plain texts OCR":
-            res = model.chat(tokenizer, image_path, ocr_type="ocr")
-            return res, None
-        elif got_mode == "format texts OCR":
-            res = model.chat(tokenizer, image_path, ocr_type="format", render=True, save_render_file=result_path)
-        elif got_mode == "plain multi-crop OCR":
-            res = model.chat_crop(tokenizer, image_path, ocr_type="ocr")
-            return res, None
-        elif got_mode == "format multi-crop OCR":
-            res = model.chat_crop(tokenizer, image_path, ocr_type="format", render=True, save_render_file=result_path)
-        elif got_mode == "plain fine-grained OCR":
-            res = model.chat(tokenizer, image_path, ocr_type="ocr", ocr_box=ocr_box, ocr_color=ocr_color)
-            return res, None
-        elif got_mode == "format fine-grained OCR":
-            res = model.chat(tokenizer, image_path, ocr_type="format", ocr_box=ocr_box, ocr_color=ocr_color, render=True, save_render_file=result_path)
 
-        # res_markdown = f"$$ {res} $$"
-        res_markdown = res
+def process_pdf(pdf_file):
+    temp_pdf_path = os.path.join(UPLOAD_FOLDER, f"{uuid.uuid4()}.pdf")
+    pdf_file.save(temp_pdf_path)
+    images = pdf_to_images(temp_pdf_path)
+    os.remove(temp_pdf_path)
+    return images
 
-        if "format" in got_mode and os.path.exists(result_path):
-            with open(result_path, "r") as f:
-                html_content = f.read()
-            encoded_html = base64.b64encode(html_content.encode("utf-8")).decode("utf-8")
-            iframe_src = f"data:text/html;base64,{encoded_html}"
-            iframe = f'<iframe src="{iframe_src}" width="100%" height="600px"></iframe>'
-            download_link = f'<a href="data:text/html;base64,{encoded_html}" download="result_{unique_id}.html">Download Full Result</a>'
-            return res_markdown, f"{download_link}<br>{iframe}"
-        else:
-            return res_markdown, None
-    except Exception as e:
-        return f"Error: {str(e)}", None
-    finally:
-        if os.path.exists(image_path):
-            os.remove(image_path)
 
+def display_images(images):
+    image_elements = [gr.Image(value=img, type="pil") for img in images]
+    return gr.Gallery(value=image_elements)
 
-def task_update(task):
-    if "fine-grained" in task:
-        return [
-            gr.update(visible=True),
-            gr.update(visible=False),
-            gr.update(visible=False),
-        ]
-    else:
-        return [
-            gr.update(visible=False),
-            gr.update(visible=False),
-            gr.update(visible=False),
-        ]
 
-
-def fine_grained_update(task):
-    if task == "box":
-        return [
-            gr.update(visible=False, value=""),
-            gr.update(visible=True),
-        ]
-    elif task == "color":
-        return [
-            gr.update(visible=True),
-            gr.update(visible=False, value=""),
-        ]
-
-
-def cleanup_old_files():
-    current_time = time.time()
-    for folder in [UPLOAD_FOLDER, RESULTS_FOLDER]:
-        for file_path in Path(folder).glob("*"):
-            if current_time - file_path.stat().st_mtime > 3600:  # 1 hour
-                file_path.unlink()
+def on_image_select(image):
+    return image
 
 
 with gr.Blocks() as demo:
-    with gr.Row():
-        with gr.Column():
-            image_input = gr.Image(type="filepath", label="上传图片")
-            task_dropdown = gr.Dropdown(
-                choices=[
-                    "plain texts OCR",
-                    "format texts OCR",
-                    "plain multi-crop OCR",
-                    "format multi-crop OCR",
-                    "plain fine-grained OCR",
-                    "format fine-grained OCR",
-                ],
-                label="选择GOT模式",
-                value="plain texts OCR",
-            )
-            fine_grained_dropdown = gr.Dropdown(choices=["box", "color"], label="fine-grained type", visible=False)
-            color_dropdown = gr.Dropdown(choices=["red", "green", "blue"], label="color list", visible=False)
-            box_input = gr.Textbox(label="input box: [x1,y1,x2,y2]", placeholder="e.g., [0,0,100,100]", visible=False)
-            submit_button = gr.Button("Submit")
-
-        with gr.Column():
-            ocr_result = gr.Textbox(label="GOT output")
-
-    with gr.Column():
-        gr.Markdown("**如果选择带格式的模式，mathpix结果将自动呈现如下:**")
-        html_result = gr.HTML(label="rendered html", show_label=True)
-
-    task_dropdown.change(task_update, inputs=[task_dropdown], outputs=[fine_grained_dropdown, color_dropdown, box_input])
-    fine_grained_dropdown.change(fine_grained_update, inputs=[fine_grained_dropdown], outputs=[color_dropdown, box_input])
+    pdf_input = gr.File(label="上传PDF文件")
+    image_gallery = gr.Gallery(label="PDF页面预览", columns=3, height="auto")
+    selected_image = gr.Image(label="选中的图片", type="pil")
 
-    submit_button.click(run_GOT, inputs=[image_input, task_dropdown, fine_grained_dropdown, color_dropdown, box_input], outputs=[ocr_result, html_result])
+    pdf_input.upload(fn=process_pdf, inputs=pdf_input, outputs=image_gallery)
+    image_gallery.select(fn=on_image_select, inputs=image_gallery, outputs=selected_image)
 
-if __name__ == "__main__":
-    cleanup_old_files()
-    demo.launch()
+    # 这里可以添加OCR转换功能的相关组件和逻辑

requirements.txt

@@ -1,3 +1,4 @@
+PyMuPDF
 verovio
 gradio
 numpy