Initial commit

olmocr/bench/sample_data/olmocr_pipeline/openstax_caculus_pg_273_pg1_repeat3.md

+3.4 EXERCISES
+
+For the following exercises, the given functions represent the position of a particle traveling along a horizontal line.
+
+a. Find the velocity and acceleration functions.
+
+b. Determine the time intervals when the object is slowing down or speeding up.
+
+150. \[ s(t) = 2t^3 - 3t^2 - 12t + 8 \]
+
+151. \[ s(t) = 2t^3 - 15t^2 + 36t - 10 \]
+
+152. \[ s(t) = \frac{t}{1 + t^2} \]
+
+153. A rocket is fired vertically upward from the ground. The distance \( s \) in feet that the rocket travels from the ground after \( t \) seconds is given by \( s(t) = -16t^2 + 560t. \)
+
+a. Find the velocity of the rocket 3 seconds after being fired.
+
+b. Find the acceleration of the rocket 3 seconds after being fired.
+
+154. A ball is thrown downward with a speed of 8 ft/s from the top of a 64-foot-tall building. After \( t \) seconds, its height above the ground is given by \( s(t) = -16t^2 - 8t + 64. \)
+
+a. Determine how long it takes for the ball to hit the ground.
+
+b. Determine the velocity of the ball when it hits the ground.
+
+155. The position function \( s(t) = t^2 - 3t - 4 \) represents the position of the back of a car backing out of a driveway and then driving in a straight line, where \( s \) is in feet and \( t \) is in seconds. In this case, \( s(t) = 0 \) represents the time at which the back of the car is at the garage door, so \( s(0) = -4 \) is the starting position of the car, 4 feet inside the garage.
+
+a. Determine the velocity of the car when \( s(t) = 0. \)
+
+b. Determine the velocity of the car when \( s(t) = 14. \)
+
+156. The position of a hummingbird flying along a straight line in \( t \) seconds is given by \( s(t) = 3t^3 - 7t \) meters.
+
+a. Determine the velocity of the bird at \( t = 1 \) sec.
+
+b. Determine the acceleration of the bird at \( t = 1 \) sec.
+
+c. Determine the acceleration of the bird when the velocity equals 0.
+
+157. A potato is launched vertically upward with an initial velocity of 100 ft/s from a potato gun at the top of an 85-foot-tall building. The distance in feet that the potato travels from the ground after \( t \) seconds is given by \( s(t) = -16t^2 + 100t + 85. \)
+
+a. Find the velocity of the potato after 0.5 s and 5.75 s.
+
+b. Find the speed of the potato at 0.5 s and 5.75 s.
+
+c. Determine when the potato reaches its maximum height.
+
+d. Find the acceleration of the potato at 0.5 s and 1.5 s.
+
+e. Determine how long the potato is in the air.
+
+f. Determine the velocity of the potato upon hitting the ground.
+
+158. The position function \( s(t) = t^3 - 8t \) gives the position in miles of a freight train where east is the positive direction and \( t \) is measured in hours.
+
+a. Determine the direction the train is traveling when \( s(t) = 0. \)
+
+b. Determine the direction the train is traveling when \( s(t) = 0. \)
+
+c. Determine the time intervals when the train is slowing down or speeding up.
+
+159. The following graph shows the position \( y = s(t) \) of an object moving along a straight line.
+
+![Graph of the position function](image)
+
+a. Use the graph of the position function to determine the time intervals when the velocity is positive, negative, or zero.
+
+b. Sketch the graph of the velocity function.
+
+c. Use the graph of the velocity function to determine the time intervals when the acceleration is positive, negative, or zero.
+
+d. Determine the time intervals when the object is speeding up or slowing down.
\ No newline at end of file

olmocr/bench/sample_data/olmocr_pipeline/openstax_caculus_pg_273_pg1_repeat4.md

+3.4 EXERCISES
+
+For the following exercises, the given functions represent the position of a particle traveling along a horizontal line.
+
+a. Find the velocity and acceleration functions.
+
+b. Determine the time intervals when the object is slowing down or speeding up.
+
+150. \( s(t) = 2t^3 - 3t^2 - 12t + 8 \)
+
+151. \( s(t) = 2t^3 - 15t^2 + 36t - 10 \)
+
+152. \( s(t) = \frac{t}{1 + t^2} \)
+
+153. A rocket is fired vertically upward from the ground. The distance \( s \) in feet that the rocket travels from the ground after \( t \) seconds is given by \( s(t) = -16t^2 + 560t + 155 \).
+
+a. Find the velocity of the rocket 3 seconds after being fired.
+
+b. Find the acceleration of the rocket 3 seconds after being fired.
+
+154. A ball is thrown downward with a speed of 8 \( \text{ft/s} \) from the top of a 64-foot-tall building. After \( t \) seconds, its height above the ground is given by \( s(t) = -16t^2 - 8t + 64 \).
+
+a. Determine how long it takes for the ball to hit the ground.
+
+b. Determine the velocity of the ball when it hits the ground.
+
+155. The position function \( s(t) = t^2 - 3t - 4 \) represents the position of the back of a car backing out of a driveway and then driving in a straight line, where \( s \) is in feet and \( t \) is in seconds. In this case, \( s(t) = 0 \) represents the time at which the back of the car is at the garage door, so \( s(0) = -4 \) is the starting position of the car, 4 feet inside the garage.
+
+a. Determine the velocity of the car when \( s(t) = 0 \).
+
+b. Determine the velocity of the car when \( s(t) = 14 \).
+
+156. The position of a hummingbird flying along a straight line in \( t \) seconds is given by \( s(t) = 3t^3 - 7t \) meters.
+
+a. Determine the velocity of the bird at \( t = 1 \) sec.
+
+b. Determine the acceleration of the bird at \( t = 1 \) sec.
+
+c. Determine the acceleration of the bird when the velocity equals 0.
+
+157. A potato is launched vertically upward with an initial velocity of 100 \( \text{ft/s} \) from a potato gun at the top of an 85-foot-tall building. The distance in feet that the potato travels from the ground after \( t \) seconds is given by \( s(t) = -16t^2 + 100t + 85 \).
+
+a. Find the velocity of the potato after 0.5 s and 5.75 s.
+
+b. Find the speed of the potato at 0.5 s and 5.75 s.
+
+c. Determine when the potato reaches its maximum height.
+
+d. Find the acceleration of the potato at 0.5 s and 1.5 s.
+
+e. Determine how long the potato is in the air.
+
+f. Determine the velocity of the potato upon hitting the ground.
+
+158. The position function \( s(t) = t^3 - 8t \) gives the position in miles of a freight train where east is the positive direction and \( t \) is measured in hours.
+
+a. Determine the direction the train is traveling when \( s(t) = 0 \).
+
+b. Determine the direction the train is traveling when \( s(t) = 0 \).
+
+c. Determine the time intervals when the train is slowing down or speeding up.
+
+159. The following graph shows the position \( y = s(t) \) of an object moving along a straight line.
+
+![Graph](image)
+
+a. Use the graph of the position function to determine the time intervals when the velocity is positive, negative, or zero.
+
+b. Sketch the graph of the velocity function.
+
+c. Use the graph of the velocity function to determine the time intervals when the acceleration is positive, negative, or zero.
+
+d. Determine the time intervals when the object is speeding up or slowing down.
\ No newline at end of file

olmocr/bench/sample_data/olmocr_pipeline/openstax_caculus_pg_273_pg1_repeat5.md

+3.4 EXERCISES
+
+For the following exercises, the given functions represent the position of a particle traveling along a horizontal line.
+
+a. Find the velocity and acceleration functions.
+
+b. Determine the time intervals when the object is slowing down or speeding up.
+
+150. \( s(t) = 2t^3 - 3t^2 - 12t + 8 \)
+
+151. \( s(t) = 2t^3 - 15t^2 + 36t - 10 \)
+
+152. \( s(t) = \frac{t}{1 + t^2} \)
+
+153. A rocket is fired vertically upward from the ground. The distance \( s \) in feet that the rocket travels from the ground after \( t \) seconds is given by \( s(t) = -16t^2 + 560t \).
+
+a. Find the velocity of the rocket 3 seconds after being fired.
+
+b. Find the acceleration of the rocket 3 seconds after being fired.
+
+154. A ball is thrown downward with a speed of 8 ft/ s from the top of a 64-foot-tall building. After \( t \) seconds, its height above the ground is given by \( s(t) = -16t^2 - 8t + 64 \).
+
+a. Determine how long it takes for the ball to hit the ground.
+
+b. Determine the velocity of the ball when it hits the ground.
+
+155. The position function \( s(t) = t^2 - 3t - 4 \) represents the position of the back of a car backing out of a driveway and then driving in a straight line, where \( s \) is in feet and \( t \) is in seconds. In this case, \( s(t) = 0 \) represents the time at which the back of the car is at the garage door, so \( s(0) = -4 \) is the starting position of the car, 4 feet inside the garage.
+
+a. Determine the velocity of the car when \( s(t) = 0 \).
+
+b. Determine the velocity of the car when \( s(t) = 14 \).
+
+156. The position of a hummingbird flying along a straight line in \( t \) seconds is given by \( s(t) = 3t^3 - 7t \) meters.
+
+a. Determine the velocity of the bird at \( t = 1 \) sec.
+
+b. Determine the acceleration of the bird at \( t = 1 \) sec.
+
+c. Determine the acceleration of the bird when the velocity equals 0.
+
+157. A potato is launched vertically upward with an initial velocity of 100 ft/s from a potato gun at the top of an 85-foot-tall building. The distance in feet that the potato travels from the ground after \( t \) seconds is given by \( s(t) = -16t^2 + 100t + 85 \).
+
+a. Find the velocity of the potato after 0.5 s and 5.75 s.
+
+b. Find the speed of the potato at 0.5 s and 5.75 s.
+
+c. Determine when the potato reaches its maximum height.
+
+d. Find the acceleration of the potato at 0.5 s and 1.5 s.
+
+e. Determine how long the potato is in the air.
+
+f. Determine the velocity of the potato upon hitting the ground.
+
+158. The position function \( s(t) = t^3 - 8t \) gives the position in miles of a freight train where east is the positive direction and \( t \) is measured in hours.
+
+a. Determine the direction the train is traveling when \( s(t) = 0 \).
+
+b. Determine the direction the train is traveling when \( s(t) = 0 \).
+
+c. Determine the time intervals when the train is slowing down or speeding up.
+
+159. The following graph shows the position \( y = s(t) \) of an object moving along a straight line.
+
+![Graph of position function](image)
+
+a. Use the graph of the position function to determine the time intervals when the velocity is positive, negative, or zero.
+
+b. Sketch the graph of the velocity function.
+
+Use the graph of the velocity function to determine the time intervals when the acceleration is positive, negative, or zero.
+
+d. Determine the time intervals when the object is speeding up or slowing down.
\ No newline at end of file

olmocr/bench/scripts/convert_all.sh

+#!/bin/bash
+
+# Exit on error but allow the trap to execute
+set -e
+
+# Global variable to track server PID
+SERVER_PID=""
+
+# Trap function to handle Ctrl+C (SIGINT)
+cleanup() {
+    echo -e "\n[INFO] Received interrupt signal. Cleaning up..."
+    
+    # Find and kill any Python processes started by this script
+    echo "[INFO] Stopping any running Python processes"
+    pkill -P $$ python || true
+    
+    # Stop server if running
+    if [ -n "$SERVER_PID" ] && kill -0 "$SERVER_PID" 2>/dev/null; then
+        echo "[INFO] Stopping server (PID: $SERVER_PID)"
+        kill -TERM "$SERVER_PID" 2>/dev/null || true
+        wait "$SERVER_PID" 2>/dev/null || true
+    fi
+
+    pkill vllm
+    
+    echo "[INFO] Cleanup complete. Exiting."
+    exit 1
+}
+
+# Set the trap for SIGINT (Ctrl+C)
+trap cleanup SIGINT
+
+# Function to check if port 30000 is in use
+check_port() {
+    port=30000
+    echo "[INFO] Checking if port $port is available..."
+    
+    if command -v lsof >/dev/null 2>&1; then
+        # Linux/macOS
+        if lsof -i :$port >/dev/null 2>&1; then
+            echo "[ERROR] Port $port is already in use. Process details:"
+            lsof -i :$port
+            echo "[ERROR] Please stop the process using this port and try again."
+            echo "        You can use: kill -9 <PID>"
+            return 1
+        fi
+    elif command -v netstat >/dev/null 2>&1; then
+        # Windows/other systems with netstat
+        if netstat -an | grep -q ":$port "; then
+            echo "[ERROR] Port $port is already in use. Process details:"
+            if command -v findstr >/dev/null 2>&1; then
+                # Windows
+                netstat -ano | findstr ":$port"
+                echo "[ERROR] Please stop the process using this port and try again."
+                echo "        You can use: taskkill /F /PID <PID>"
+            else
+                netstat -an | grep ":$port "
+                echo "[ERROR] Please stop the process using this port and try again."
+            fi
+            return 1
+        fi
+    else
+        # Fallback method using nc if available
+        if command -v nc >/dev/null 2>&1; then
+            nc -z localhost $port >/dev/null 2>&1
+            if [ $? -eq 0 ]; then
+                echo "[ERROR] Port $port is already in use, but cannot determine which process."
+                echo "[ERROR] Please ensure port $port is available before continuing."
+                return 1
+            fi
+        else
+            echo "[WARNING] Cannot check if port $port is in use (neither lsof, netstat, nor nc available)."
+            echo "[WARNING] Continuing anyway, but this might fail if the port is already in use."
+            return 0
+        fi
+    fi
+    
+    echo "[INFO] Port $port is available."
+    return 0
+}
+
+# Function to create conda environment if it doesn't exist
+create_conda_env() {
+    env_name=$1
+    python_version=$2
+    
+    # Check if environment exists
+    if conda info --envs | grep -q "^$env_name "; then
+        echo "Environment $env_name already exists, using it."
+    else
+        echo "Creating conda environment: $env_name"
+        conda create -y -n $env_name python=$python_version
+    fi
+}
+
+# Generic function to start a server (either vllm or sglang)
+start_server() {
+    server_type=$1
+    model_name=$2
+    shift 2  # Remove server type and model name from the argument list
+    
+    echo "Starting $server_type server for model: $model_name"
+    echo "Additional arguments: $@"
+    
+    if [ "$server_type" = "sglang" ]; then
+        python -m sglang.launch_server --port 30000 --model "$model_name" "$@" &
+    elif [ "$server_type" = "vllm" ]; then
+        vllm serve $model_name --port 30000 "$@" &
+    else
+        echo "Unsupported server type: $server_type"
+        exit 1
+    fi
+    SERVER_PID=$!
+    
+    # Check if the server process is running
+    if ! kill -0 $SERVER_PID 2>/dev/null; then
+        echo "Failed to start server process. Exiting."
+        exit 1
+    fi
+    
+    # Wait for the server to be ready by checking the models endpoint
+    echo "Waiting for server to be ready..."
+    max_attempts=300
+    attempt=0
+    while [ $attempt -lt $max_attempts ]; do
+        if curl -s "http://localhost:30000/v1/models" -o /dev/null -w "%{http_code}" | grep -q "200"; then
+            echo "Server is ready!"
+            return 0
+        fi
+        attempt=$((attempt + 1))
+        echo "Waiting for server... attempt $attempt/$max_attempts"
+        sleep 2
+    done
+    
+    echo "Server failed to become ready after multiple attempts. Exiting."
+    kill $SERVER_PID
+    SERVER_PID=""
+    exit 1
+}
+
+# Function to stop the server
+stop_server() {
+    echo "Stopping server with PID: $SERVER_PID"
+    if [ -n "$SERVER_PID" ] && kill -0 "$SERVER_PID" 2>/dev/null; then
+        kill $SERVER_PID
+        wait $SERVER_PID 2>/dev/null || true
+        echo "Server stopped."
+    else
+        echo "No server to stop."
+    fi
+    SERVER_PID=""
+}
+
+# Create and activate olmocr environment
+create_conda_env "olmocr" "3.11"
+source $(conda info --base)/etc/profile.d/conda.sh
+source activate olmocr
+
+# Run olmocr benchmarks, exactly as the pipeline.py does it
+echo "Running olmocr benchmarks..."
+python -m olmocr.bench.convert olmocr_pipeline --repeats 5
+
+# Install marker-pdf and run benchmarks
+echo "Installing marker-pdf and running benchmarks..."
+pip install marker-pdf==1.6.1
+python -m olmocr.bench.convert marker
+
+# Install verovio and run benchmarks
+# echo "Installing verovio and running benchmarks..."
+# pip install verovio
+# python -m olmocr.bench.convert gotocr
+
+# Run chatgpt benchmarks
+echo "Running chatgpt benchmarks..."
+python -m olmocr.bench.convert chatgpt
+#python -m olmocr.bench.convert chatgpt:name=chatgpt45:model=gpt-4.5-preview-2025-02-27
+
+# Run gemini benchmarks
+echo "Running gemini benchmarks..."
+python -m olmocr.bench.convert gemini:name=gemini_flash2:model=gemini-2.0-flash --parallel 4
+
+echo "Running mistral..."
+pip install mistralai
+python -m olmocr.bench.convert mistral
+
+# Run raw server benchmarks with generic server function
+# For each model, start server, run benchmark, then stop server
+
+# Check port availability at script start
+check_port || exit 1
+
+# olmocr_base_temp0_1 using sglang server
+start_server sglang "allenai/olmOCR-7B-0225-preview" --chat-template qwen2-vl --mem-fraction-static 0.7
+python -m olmocr.bench.convert server:name=olmocr_base_temp0_1:model=allenai/olmOCR-7B-0225-preview:temperature=0.1:prompt_template=fine_tune:response_template=json --repeats 5 --parallel 20
+python -m olmocr.bench.convert server:name=olmocr_base_temp0_8:model=allenai/olmOCR-7B-0225-preview:temperature=0.8:prompt_template=fine_tune:response_template=json --repeats 5 --parallel 20
+stop_server
+
+start_server vllm "allenai/olmOCR-7B-0225-preview"
+python -m olmocr.bench.convert server:name=olmocr_base_vllm_temp0_1:model=allenai/olmOCR-7B-0225-preview:temperature=0.1:prompt_template=fine_tune:response_template=json --repeats 5 --parallel 20
+python -m olmocr.bench.convert server:name=olmocr_base_vllm_temp0_8:model=allenai/olmOCR-7B-0225-preview:temperature=0.8:prompt_template=fine_tune:response_template=json --repeats 5 --parallel 20
+stop_server
+
+# Feel free to enable if you want
+# qwen2_vl_7b using sglang server
+# start_server sglang "Qwen/Qwen2-VL-7B-Instruct" --chat-template qwen2-vl --mem-fraction-static 0.7
+# python -m olmocr.bench.convert server:name=qwen2_vl_7b:model=Qwen/Qwen2-VL-7B-Instruct:temperature=0.1:prompt_template=full:response_template=plain --repeats 5 --parallel 20
+# stop_server
+
+# TODO: qwen2.5 Not working right now in sglang
+# qwen25_vl_7b
+# create_conda_env "qwen25" "3.11"
+# source activate qwen25
+# pip install olmocr
+# pip install "sglang[all]>=0.4.3.post2" --find-links https://flashinfer.ai/whl/cu124/torch2.5/flashinfer-python transformers==4.48.3
+# start_server sglang "Qwen/Qwen2.5-VL-7B-Instruct" --chat-template qwen2-vl --mem-fraction-static 0.7
+# python -m olmocr.bench.convert server:name=qwen25_vl_7b:model=Qwen/Qwen2.5-VL-7B-Instruct:temperature=0.1:prompt_template=full:response_template=plain --repeats 5 --parallel 20
+# stop_server
+
+# TODO: Fix this, I was not able to get it to all install successfully
+# Create and activate mineru environment
+# create_conda_env "mineru" "3.11"
+# source activate mineru
+
+# Install magic-pdf and run benchmarks
+# echo "Installing magic-pdf and running mineru benchmarks..."
+# pip install -U "magic-pdf[full]==1.2.2" --extra-index-url https://wheels.myhloli.com
+# python -m pip install paddlepaddle==3.0.0rc1 -i https://www.paddlepaddle.org.cn/packages/stable/cpu/
+# pip install huggingface_hub Pillow paddleocr ultralytics doclayout-yolo pycocotools 
+# wget https://github.com/opendatalab/MinerU/raw/master/scripts/download_models_hf.py -O download_models_hf.py
+# python download_models_hf.py
+# python -m olmocr.bench.convert mineru
+
+# Final cleanup
+if [ -n "$SERVER_PID" ] && kill -0 $SERVER_PID 2>/dev/null; then
+    stop_server
+fi
+
+echo "All benchmarks completed successfully."

olmocr/bench/scripts/difference_viewer.py

+#!/usr/bin/env python3
+import argparse
+import glob
+import json
+import os
+import sys
+from collections import defaultdict
+
+from runners.run_chatgpt import run_chatgpt
+from runners.run_gemini import run_gemini
+
+from olmocr.data.renderpdf import render_pdf_to_base64png
+
+
+def parse_rules_file(file_path):
+    """Parse the rules file and organize rules by PDF."""
+    pdf_rules = defaultdict(list)
+
+    with open(file_path, "r") as f:
+        for line in f:
+            line = line.strip()
+            if not line:
+                continue
+            try:
+                rule = json.loads(line)
+                if "pdf" in rule:
+                    pdf_rules[rule["pdf"]].append(rule)
+            except json.JSONDecodeError:
+                print(f"Warning: Could not parse line as JSON: {line}")
+
+    return pdf_rules
+
+
+def get_model_outputs(pdf_path):
+    """Get outputs from both models for a given PDF."""
+    try:
+        print(f"Attempting to process PDF: {pdf_path}")
+        print(f"File exists: {os.path.exists(pdf_path)}")
+
+        chatgpt_output = run_chatgpt(pdf_path)
+        gemini_output = run_gemini(pdf_path)
+        return chatgpt_output, gemini_output
+    except Exception as e:
+        print(f"Error getting model outputs for {pdf_path}: {str(e)}")
+        return f"Error: {str(e)}", f"Error: {str(e)}"
+
+
+def find_pdfs_in_directory(directory):
+    """Find all PDF files in the given directory."""
+    if not os.path.isdir(directory):
+        print(f"Warning: {directory} is not a directory.")
+        return []
+    pdf_files = []
+    for ext in ["pdf", "PDF"]:
+        pattern = os.path.join(directory, f"*.{ext}")
+        pdf_files.extend(glob.glob(pattern))
+
+    print(f"Found {len(pdf_files)} PDF files in {directory}")
+    for pdf in pdf_files:
+        print(f"  - {pdf}")
+    return pdf_files
+
+
+def generate_html(pdf_rules, rules_file_path, pdfs_to_process=None):
+    """Generate the HTML page with PDF renderings and model outputs."""
+    pdf_paths = []
+
+    if pdfs_to_process:
+        for pdf_item in pdfs_to_process:
+            if os.path.isdir(pdf_item):
+                pdf_paths.extend(find_pdfs_in_directory(pdf_item))
+            elif os.path.isfile(pdf_item):
+                pdf_paths.append(pdf_item)
+            else:
+                print(f"Warning: {pdf_item} is neither a valid file nor directory")
+    else:
+        pdf_base_dir = os.path.join(os.path.dirname(rules_file_path), "pdfs")
+        pdf_paths = [os.path.join(pdf_base_dir, pdf_name) for pdf_name in list(pdf_rules.keys())[:10]]
+    pdf_paths = list(set(pdf_paths))
+    print("Processing the following PDFs:")
+    for path in pdf_paths:
+        print(f"  - {path} (exists: {os.path.exists(path)})")
+
+    html = """
+    <!DOCTYPE html>
+    <html lang="en">
+    <head>
+        <meta charset="UTF-8">
+        <meta name="viewport" content="width=device-width, initial-scale=1.0">
+        <title>PDF Model Comparison</title>
+        <style>
+            body {
+                font-family: Arial, sans-serif;
+                margin: 0;
+                padding: 20px;
+                background-color: #f5f5f5;
+            }
+            
+            .container {
+                max-width: 1800px;
+                margin: 0 auto;
+            }
+            
+            h1 {
+                color: #333;
+                text-align: center;
+                margin-bottom: 30px;
+            }
+            
+            .pdf-container {
+                background-color: white;
+                border-radius: 8px;
+                box-shadow: 0 2px 10px rgba(0, 0, 0, 0.1);
+                margin-bottom: 30px;
+                overflow: hidden;
+            }
+            
+            .pdf-header {
+                background-color: #4a6fa5;
+                color: white;
+                padding: 15px;
+                font-size: 18px;
+                font-weight: bold;
+                display: flex;
+                justify-content: space-between;
+                align-items: center;
+            }
+            
+            .pdf-content {
+                display: flex;
+                flex-direction: row;
+                padding: 20px;
+            }
+            
+            @media (max-width: 1200px) {
+                .pdf-content {
+                    flex-direction: column;
+                }
+            }
+            
+            .pdf-image {
+                flex: 0 0 30%;
+                max-width: 500px;
+                text-align: center;
+                padding-right: 20px;
+            }
+            
+            .pdf-image img {
+                max-width: 100%;
+                height: auto;
+                border: 1px solid #ddd;
+            }
+            
+            .models-container {
+                flex: 1;
+                display: flex;
+                flex-direction: column;
+            }
+            
+            .model-outputs {
+                display: flex;
+                flex-direction: row;
+                margin-bottom: 20px;
+            }
+            
+            .model-output {
+                flex: 1;
+                margin: 0 10px;
+                border: 1px solid #ddd;
+                border-radius: 5px;
+                overflow: hidden;
+            }
+            
+            .model-header {
+                background-color: #4a6fa5;
+                color: white;
+                padding: 10px;
+                font-weight: bold;
+                text-align: center;
+            }
+            
+            .model-content {
+                padding: 15px;
+                height: 400px;
+                overflow-y: auto;
+                white-space: pre-wrap;
+                font-family: monospace;
+                font-size: 14px;
+                background-color: #f8f9fa;
+            }
+            
+            .difference-content {
+                padding: 15px;
+                height: 200px;
+                overflow-y: auto;
+                white-space: pre-wrap;
+                font-family: monospace;
+                font-size: 14px;
+                background-color: #f8f9fa;
+                display: none;
+                margin-top: 20px;
+                border: 1px solid #ddd;
+                border-radius: 5px;
+            }
+            
+            .controls {
+                display: flex;
+                justify-content: space-between;
+                align-items: center;
+                margin-bottom: 20px;
+                padding: 10px;
+                background-color: #e9ecef;
+                border-radius: 5px;
+            }
+            
+            .rating-controls {
+                display: flex;
+                flex-wrap: wrap;
+                gap: 10px;
+            }
+            
+            button {
+                padding: 8px 15px;
+                border: none;
+                border-radius: 4px;
+                cursor: pointer;
+                font-weight: bold;
+                transition: background-color 0.2s;
+            }
+            
+            button:hover {
+                opacity: 0.9;
+            }
+            
+            .rating-btn {
+                flex: 1;
+                min-width: 120px;
+            }
+            
+            .chatgpt-better {
+                background-color: #28a745;
+                color: white;
+            }
+            
+            .gemini-better {
+                background-color: #dc3545;
+                color: white;
+            }
+            
+            .both-good {
+                background-color: #17a2b8;
+                color: white;
+            }
+            
+            .both-bad {
+                background-color: #6c757d;
+                color: white;
+            }
+            
+            .invalid-pdf {
+                background-color: #343a40;
+                color: white;
+            }
+            
+            .show-diff-btn {
+                background-color: #fd7e14;
+                color: white;
+            }
+            
+            .highlight {
+                background-color: #ffff99;
+            }
+            
+            .rating-indicator {
+                display: inline-block;
+                padding: 5px 10px;
+                border-radius: 4px;
+                color: white;
+                font-weight: bold;
+                margin-left: 10px;
+            }
+            
+            .error {
+                color: #dc3545;
+                padding: 20px;
+                text-align: center;
+                border: 1px solid #dc3545;
+                border-radius: 5px;
+                margin: 20px;
+            }
+        </style>
+    </head>
+    <body>
+        <div class="container">
+            <h1>PDF Model Comparison</h1>
+    """
+
+    for pdf_path in pdf_paths:
+        if not os.path.exists(pdf_path) or not os.path.isfile(pdf_path):
+            print(f"Skipping non-existent or non-file path: {pdf_path}")
+            continue
+        pdf_name = os.path.basename(pdf_path)
+        pdf_id = pdf_name.replace(".", "-")
+        try:
+            print(f"Rendering PDF: {pdf_path}")
+            base64_img = render_pdf_to_base64png(pdf_path, 0)
+            img_html = f'<img src="data:image/png;base64,{base64_img}" alt="{pdf_name}">'
+        except Exception as e:
+            print(f"Error rendering PDF: {str(e)}")
+            img_html = f'<div class="error">Error rendering PDF: {str(e)}</div>'
+        print(f"Getting model outputs for: {pdf_path}")
+        chatgpt_output, gemini_output = get_model_outputs(pdf_path)
+
+        html += f"""
+        <div class="pdf-container" id="pdf-{pdf_id}">
+            <div class="pdf-header">
+                <span>{pdf_name}</span>
+                <span class="rating-indicator" id="rating-{pdf_id}"></span>
+            </div>
+            <div class="pdf-content">
+                <div class="pdf-image">
+                    {img_html}
+                </div>
+                <div class="models-container">
+                    <div class="controls">
+                        <div class="rating-controls">
+                            <button class="rating-btn chatgpt-better" onclick="rateModel('{pdf_id}', 'chatgpt')">ChatGPT Better</button>
+                            <button class="rating-btn gemini-better" onclick="rateModel('{pdf_id}', 'gemini')">Gemini Better</button>
+                            <button class="rating-btn both-good" onclick="rateModel('{pdf_id}', 'both-good')">Both Good</button>
+                            <button class="rating-btn both-bad" onclick="rateModel('{pdf_id}', 'both-bad')">Both Bad</button>
+                            <button class="rating-btn invalid-pdf" onclick="rateModel('{pdf_id}', 'invalid')">Invalid PDF</button>
+                        </div>
+                        <button class="show-diff-btn" onclick="toggleDifference('{pdf_id}')">Show Differences</button>
+                    </div>
+                    <div class="model-outputs">
+                        <div class="model-output">
+                            <div class="model-header">ChatGPT Output</div>
+                            <div class="model-content" id="chatgpt-{pdf_id}">{chatgpt_output}</div>
+                        </div>
+                        <div class="model-output">
+                            <div class="model-header">Gemini Output</div>
+                            <div class="model-content" id="gemini-{pdf_id}">{gemini_output}</div>
+                        </div>
+                    </div>
+                    <div class="difference-content" id="diff-{pdf_id}">
+                        <h3>Differences</h3>
+                        <p>Loading differences...</p>
+                    </div>
+                </div>
+            </div>
+        </div>
+        """
+
+    html += """
+        </div>
+        
+        <script>
+            // Store ratings
+            const ratings = {};
+            
+            function rateModel(pdfId, rating) {
+                // Save rating
+                ratings[pdfId] = rating;
+                
+                // Update visual indicator
+                const indicator = document.getElementById(`rating-${pdfId}`);
+                indicator.textContent = rating.replace('-', ' ').toUpperCase();
+                
+                // Apply class matching the rating
+                indicator.className = 'rating-indicator ' + rating;
+                
+                // Save ratings to localStorage
+                localStorage.setItem('pdf-model-ratings', JSON.stringify(ratings));
+            }
+            
+            function toggleDifference(pdfId) {
+                const diffElement = document.getElementById(`diff-${pdfId}`);
+                const chatgptContent = document.getElementById(`chatgpt-${pdfId}`).textContent;
+                const geminiContent = document.getElementById(`gemini-${pdfId}`).textContent;
+                
+                if (diffElement.style.display === 'none' || !diffElement.style.display) {
+                    diffElement.style.display = 'block';
+                    
+                    // If content is "Loading differences...", calculate differences
+                    if (diffElement.textContent.includes("Loading differences...")) {
+                        // Simple difference highlighting
+                        const diffResult = findDifferences(chatgptContent, geminiContent);
+                        diffElement.innerHTML = diffResult;
+                    }
+                } else {
+                    diffElement.style.display = 'none';
+                }
+            }
+            
+            function findDifferences(text1, text2) {
+                // Split into sentences for comparison
+                const sentences1 = text1.split(/(?<=[.!?])\\s+/);
+                const sentences2 = text2.split(/(?<=[.!?])\\s+/);
+                
+                let result = "<h4>ChatGPT unique content:</h4><ul>";
+                
+                // Find sentences in text1 that aren't in text2
+                sentences1.forEach(sentence => {
+                    if (sentence.trim().length > 10 && !text2.includes(sentence)) {
+                        result += `<li><span class="highlight">${sentence}</span></li>`;
+                    }
+                });
+                
+                result += "</ul><h4>Gemini unique content:</h4><ul>";
+                
+                // Find sentences in text2 that aren't in text1
+                sentences2.forEach(sentence => {
+                    if (sentence.trim().length > 10 && !text1.includes(sentence)) {
+                        result += `<li><span class="highlight">${sentence}</span></li>`;
+                    }
+                });
+                
+                result += "</ul>";
+                return result;
+            }
+            
+            // Load saved ratings on page load
+            window.onload = function() {
+                const savedRatings = localStorage.getItem('pdf-model-ratings');
+                if (savedRatings) {
+                    const parsedRatings = JSON.parse(savedRatings);
+                    for (const pdfId in parsedRatings) {
+                        rateModel(pdfId, parsedRatings[pdfId]);
+                    }
+                }
+            };
+        </script>
+    </body>
+    </html>
+    """
+
+    return html
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Generate an HTML visualization for comparing AI model outputs on PDFs.")
+    parser.add_argument("-r", "--rules_file", help="Rules file path", default="./sample_data/dataset.jsonl")
+    parser.add_argument("-o", "--output", help="Output HTML file path", default="pdf_model_comparison.html")
+    parser.add_argument("-p", "--pdfs", nargs="+", help="Specific PDF files or directories to process")
+    parser.add_argument("-l", "--limit", type=int, help="Limit the number of PDFs to process", default=10)
+
+    args = parser.parse_args()
+
+    if not os.path.exists(args.rules_file):
+        print(f"Error: Rules file not found: {args.rules_file}")
+        sys.exit(1)
+    if args.pdfs:
+        for pdf_path in args.pdfs:
+            if not os.path.exists(pdf_path):
+                print(f"WARNING: Path not found: {pdf_path}")
+
+    pdf_rules = parse_rules_file(args.rules_file)
+    html = generate_html(pdf_rules, args.rules_file, args.pdfs)
+
+    with open(args.output, "w") as f:
+        f.write(html)
+
+    print(f"HTML visualization created: {args.output}")
+    print("Open this file in a web browser to view and rate model outputs.")
+
+
+if __name__ == "__main__":
+    main()

olmocr/bench/scripts/run_difference.py

+import os
+
+from openai import OpenAI
+from prompts import build_find_difference_prompt
+from runners.run_chatgpt import run_chatgpt
+from runners.run_gemini import run_gemini
+
+from olmocr.data.renderpdf import render_pdf_to_base64png
+
+
+def combined_output(pdf_path: str) -> str:
+    chatgpt_output = run_chatgpt(pdf_path)
+    gemini_output = run_gemini(pdf_path)
+    return f"ChatGPT OUTPUT: \n" f"{chatgpt_output}\n\n" f"Gemini OUTPUT: \n" f"{gemini_output}"
+
+
+def run_difference(pdf_path: str, page_num: int = 1, model: str = "gpt-4o-2024-08-06", temperature: float = 0.1) -> str:
+    """
+    Convert page of a PDF file to markdown using GPT.
+
+    This function renders the first page of the PDF to an image, runs OCR on that image,
+    and returns the OCR result as a markdown-formatted string.
+
+    Args:
+        pdf_path (str): The local path to the PDF file.
+        page_num (int): Which page from document to pass.
+        model (str): Model used to process.
+        Temperature (float): Temperature used while utilizing the model.
+
+    Returns:
+        str: The result in markdown format.
+    """
+    # Convert the first page of the PDF to a base64-encoded PNG image.
+    image_base64 = render_pdf_to_base64png(pdf_path, page_num=page_num, target_longest_image_dim=2048)
+    anchor_text = combined_output(pdf_path)
+    client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
+
+    response = client.chat.completions.create(
+        model=model,
+        messages=[
+            {
+                "role": "user",
+                "content": [
+                    {"type": "text", "text": build_find_difference_prompt(anchor_text)},
+                    {"type": "image_url", "image_url": {"url": f"data:image/png;base64,{image_base64}"}},
+                ],
+            }
+        ],
+        temperature=temperature,
+        max_tokens=3000,
+    )
+
+    raw_response = response.choices[0].message.content
+
+    return raw_response

olmocr/bench/synth/render.py

+#!/usr/bin/env python3
+import os
+import asyncio
+from pathlib import Path
+from playwright.async_api import async_playwright
+
+# Simple configuration
+CONFIG = {
+    "input_file": os.path.join(os.path.dirname(__file__), "templates", "listpage.js"),  # React component file
+    "output_pdf": "book-page.pdf",     # Output PDF filename
+    "temp_html": "temp-render.html",   # Temporary HTML file
+    "wait_time": 1500,                 # Time to wait for rendering (ms)
+    "device_scale": 2,                 # Resolution multiplier
+    "debug": True                      # Keep temp files for debugging
+}
+
+async def create_html_file():
+    """Create a temporary HTML file that loads the React component from a file."""
+    try:
+        # Check if input file exists
+        input_path = Path(CONFIG["input_file"])
+        if not input_path.exists():
+            print(f"Error: Input file '{input_path}' not found")
+            return False
+        
+        # Read the component file
+        with open(input_path, 'r', encoding='utf-8') as f:
+            component_code = f.read()
+        
+        # Create HTML that will load our component
+        html_content = """
+<!DOCTYPE html>
+<html>
+<head>
+  <meta charset="UTF-8">
+  <meta name="viewport" content="width=device-width, initial-scale=1.0">
+  <title>Book Page Template</title>
+  <script src="https://unpkg.com/react@17/umd/react.development.js"></script>
+  <script src="https://unpkg.com/react-dom@17/umd/react-dom.development.js"></script>
+  <script src="https://unpkg.com/@babel/standalone/babel.min.js"></script>
+  <style>
+    * {
+      box-sizing: border-box;
+    }
+    
+    html, body {
+      margin: 0;
+      padding: 0;
+      width: 8.5in;
+      height: 11in;
+      overflow: hidden;
+    }
+    
+    #root {
+      width: 100%;
+      height: 100%;
+      padding: 0.25in;
+      overflow: hidden;
+    }
+    
+    @media print {
+      body {
+        -webkit-print-color-adjust: exact;
+        print-color-adjust: exact;
+      }
+    }
+  </style>
+</head>
+<body>
+  <div id="root"></div>
+
+  <script type="text/babel">
+    // The React component code loaded from external file
+    """ + component_code + """
+    
+    // Render only the book page part, not the controls
+    ReactDOM.render(
+      <BookPageTemplate />,
+      document.getElementById('root')
+    );
+  </script>
+</body>
+</html>
+        """
+        
+        with open(CONFIG["temp_html"], 'w', encoding='utf-8') as f:
+            f.write(html_content)
+            
+        print(f"Created HTML file: {CONFIG['temp_html']}")
+        print(f"Using React component from: {CONFIG['input_file']}")
+        return True
+    except Exception as e:
+        print(f"Error creating HTML file: {e}")
+        print(f"Exception details: {str(e)}")
+        import traceback
+        traceback.print_exc()
+        return False
+
+async def render_to_pdf():
+    """Render the React component to PDF using Playwright."""
+    try:
+        # Create the HTML file first
+        html_created = await create_html_file()
+        if not html_created:
+            print("Failed to create HTML file")
+            return
+        
+        print("Launching browser...")
+        async with async_playwright() as p:
+            # Launch the browser with more debugging options
+            browser = await p.chromium.launch(
+                headless=True,  # True for production, False for debugging
+            )
+            
+            # Create a new page for letter size paper
+            page = await browser.new_page(
+                viewport={"width": 816, "height": 1056},  # 8.5in x 11in at 96dpi
+                device_scale_factor=CONFIG["device_scale"]
+            )
+            
+            # Get absolute path to HTML file
+            html_path = Path(CONFIG["temp_html"]).absolute()
+            html_uri = f"file://{html_path}"
+            
+            print(f"Navigating to: {html_uri}")
+            
+            # Add event listeners for console messages and errors
+            page.on("console", lambda msg: print(f"Browser console: {msg.text}"))
+            page.on("pageerror", lambda err: print(f"Browser page error: {err}"))
+            
+            # Navigate with longer timeout and wait for network idle
+            await page.goto(html_uri, wait_until="networkidle", timeout=30000)
+            
+            # Wait for React to render
+            await page.wait_for_timeout(CONFIG["wait_time"])
+            
+            # Add a check to ensure the component rendered
+            element_count = await page.evaluate("""() => {
+                const root = document.getElementById('root');
+                return root.childElementCount;
+            }""")
+            
+            if element_count == 0:
+                print("Warning: No elements found in root. Component may not have rendered.")
+            else:
+                print(f"Found {element_count} elements in root. Component rendered successfully.")
+            
+            # Save debug screenshot
+            if CONFIG["debug"]:
+                await page.screenshot(path="debug-screenshot.png")
+                print("Debug screenshot saved")
+            
+            # Generate PDF
+            print("Generating PDF...")
+            await page.pdf(
+                path=CONFIG["output_pdf"],
+                format="Letter",
+                print_background=True,
+                margin={"top": "0", "right": "0", "bottom": "0", "left": "0"}
+            )
+            
+            print(f"PDF generated successfully: {CONFIG['output_pdf']}")
+            
+            # Close the browser
+            await browser.close()
+        
+        # Cleanup temp files if not in debug mode
+        if not CONFIG["debug"] and Path(CONFIG["temp_html"]).exists():
+            Path(CONFIG["temp_html"]).unlink()
+            print("Temporary HTML file removed")
+            
+    except Exception as e:
+        print(f"Error generating PDF: {e}")
+
+if __name__ == "__main__":
+    # Run the async function
+    try:
+        asyncio.run(render_to_pdf())
+    except Exception as e:
+        print(f"Fatal error: {e}")
+        import traceback
+        traceback.print_exc()
\ No newline at end of file

olmocr/bench/synth/templates/bookpage.js

+//import React from 'react';
+
+const BookPageTemplate = () => {
+  // Only three state variables as requested
+  const [title, setTitle] = React.useState("ADVENTURES OF DON QUIXOTE");
+  const [pageNumber, setPageNumber] = React.useState("289");
+  const [text, setText] = React.useState(
+    "deed,\" said Don Quixote, \"thou hast hit the point, Sancho, which can alone shake my resolution; I neither can, nor ought to, draw my sword, as I have often told thee, against those who are not dubbed knights. To thee which I had premeditated, thy share of the booty would have been at least the emperor's crown of gold and Cupid's painted wings; for I would have plucked them off perforce, and delivered them into thy hands.\" \"The"
+  );
+
+  // Styles for heavily degraded scan effect
+  const heavilyDegradedStyles = {
+    filter: 'grayscale(30%) contrast(120%) brightness(85%) sepia(20%)',
+    position: 'relative',
+    backgroundColor: '#e6ddc6', // More yellowed aged paper
+    backgroundImage: 'url("data:image/svg+xml,%3Csvg viewBox=\'0 0 200 200\' xmlns=\'http://www.w3.org/2000/svg\'%3E%3Cfilter id=\'noiseFilter\'%3E%3CfeTurbulence type=\'fractalNoise\' baseFrequency=\'0.85\' numOctaves=\'3\' stitchTiles=\'stitch\'/%3E%3C/filter%3E%3Crect width=\'100%25\' height=\'100%25\' filter=\'url(%23noiseFilter)\' opacity=\'0.25\'/%3E%3C/svg%3E")',
+    boxShadow: 'inset 0 0 70px rgba(0, 0, 0, 0.3), 0 0 5px rgba(0,0,0,0.1)',
+    padding: '32px',
+    borderRadius: '2px',
+    overflow: 'hidden',
+    transform: 'rotate(0.3deg)', // Slightly askew scan
+  };
+
+  // Heavily degraded text
+  const badScanTextStyle = {
+    fontFamily: '"Times New Roman", serif',
+    letterSpacing: '-0.01em',
+    wordSpacing: '0.02em',
+    fontWeight: '500',
+    color: '#222222',
+    textShadow: '0 0 1px rgba(0, 0, 0, 0.5)',
+    transform: 'scale(1.01, 0.99) rotate(-0.4deg)', // Distorted proportions
+  };
+
+  // Random coffee stain effect
+  const coffeeStain = {
+    position: 'absolute',
+    width: '100px',
+    height: '80px',
+    top: '25%',
+    right: '15%',
+    borderRadius: '50%',
+    background: 'radial-gradient(ellipse at center, rgba(139,69,19,0.15) 0%, rgba(139,69,19,0.1) 50%, rgba(139,69,19,0.05) 70%, rgba(139,69,19,0) 100%)',
+    transform: 'rotate(30deg) scale(1.5, 1)',
+    pointerEvents: 'none',
+    zIndex: 1,
+  };
+
+  // Water damage effect
+  const waterDamage = {
+    position: 'absolute',
+    width: '70%',
+    height: '40%',
+    bottom: '10%',
+    left: '5%',
+    opacity: 0.07,
+    background: 'radial-gradient(ellipse at center, rgba(0,0,0,0.2) 0%, rgba(0,0,0,0.1) 40%, rgba(0,0,0,0) 70%)',
+    borderRadius: '40% 60% 70% 30% / 40% 50% 60% 50%',
+    pointerEvents: 'none',
+    zIndex: 1,
+  };
+
+  // Add fold lines
+  const foldLine = {
+    position: 'absolute',
+    width: '100%',
+    height: '3px',
+    top: '30%',
+    left: 0,
+    background: 'linear-gradient(to right, rgba(0,0,0,0) 0%, rgba(0,0,0,0.03) 20%, rgba(0,0,0,0.08) 50%, rgba(0,0,0,0.03) 80%, rgba(0,0,0,0) 100%)',
+    boxShadow: '0 1px 3px rgba(255,255,255,0.2)',
+    pointerEvents: 'none',
+    zIndex: 2,
+  };
+
+  // Torn edge effect
+  const tornEdge = {
+    position: 'absolute',
+    top: 0,
+    right: 0,
+    width: '100%',
+    height: '100%',
+    background: 'linear-gradient(135deg, transparent 97%, #e6ddc6 97%, #e6ddc6 100%)',
+    pointerEvents: 'none',
+  };
+
+  return (
+    <div style={{
+      maxWidth: '800px',
+      margin: '0 auto',
+      padding: '16px',
+    }}>
+      {/* Heavily degraded scan container */}
+      <div style={heavilyDegradedStyles}>
+        {/* Noise overlay */}
+        <div style={{
+          position: 'absolute',
+          top: 0,
+          left: 0,
+          right: 0,
+          bottom: 0,
+          background: 'url("data:image/png;base64,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") repeat',
+          opacity: 0.15,
+          pointerEvents: 'none',
+        }}></div>
+        
+        {/* Scan lines effect */}
+        <div style={{
+          position: 'absolute',
+          top: 0,
+          left: 0,
+          right: 0,
+          bottom: 0,
+          background: 'linear-gradient(to bottom, rgba(0,0,0,0.03) 1px, transparent 1px)',
+          backgroundSize: '100% 2px',
+          opacity: 0.5,
+          pointerEvents: 'none',
+        }}></div>
+        
+        {/* Add coffee stain */}
+        <div style={coffeeStain}></div>
+        
+        {/* Add water damage */}
+        <div style={waterDamage}></div>
+        
+        {/* Add fold line */}
+        <div style={foldLine}></div>
+        
+        {/* Add torn edge */}
+        <div style={tornEdge}></div>
+        
+        {/* Header with skewed alignment */}
+        <div style={{
+          display: 'flex',
+          justifyContent: 'space-between',
+          alignItems: 'center',
+          borderBottom: '2px solid #000',
+          paddingBottom: '4px',
+          marginBottom: '24px',
+          position: 'relative',
+          opacity: 0.8,
+          transform: 'skew(-0.5deg, 0.3deg)',
+        }}>
+          <div style={{width: '48px'}}></div>
+          <h1 style={{
+            ...badScanTextStyle,
+            fontSize: '20px',
+            fontWeight: 'bold',
+            textAlign: 'center',
+            textTransform: 'uppercase',
+            letterSpacing: '1px',
+            opacity: 0.8,
+          }}>{title}</h1>
+          <div style={{
+            ...badScanTextStyle,
+            fontSize: '20px', 
+            fontWeight: 'bold',
+            opacity: 0.85,
+          }}>{pageNumber}</div>
+        </div>
+        
+        {/* Horizontal divider with uneven quality */}
+        <div style={{
+          borderBottom: '1px solid #444',
+          marginBottom: '24px',
+          opacity: 0.6,
+          filter: 'blur(0.3px)',
+          transform: 'scaleY(1.5) skew(0.7deg)',
+        }}></div>
+        
+        {/* Text content with severely degraded appearance */}
+        <div style={{
+          columnCount: 2,
+          columnGap: '20px',
+          columnRule: '1px solid rgba(0,0,0,0.1)',
+          textAlign: 'justify',
+          ...badScanTextStyle,
+          fontSize: '16px',
+          lineHeight: '1.5',
+          opacity: 0.78,
+          // Very uneven ink distribution with blurry and faded parts
+          WebkitMaskImage: 'linear-gradient(to bottom, rgba(0,0,0,0.9), rgba(0,0,0,0.75) 50%, rgba(0,0,0,0.85))',
+          // Text distortion
+          filter: 'blur(0.2px)',
+        }}>
+          {/* Bad scan text with random character fading */}
+          <p>{text.split('').map((char, index) => {
+            const opacity = Math.random() > 0.8 ? 0.4 + Math.random() * 0.5 : 0.9 + Math.random() * 0.1;
+            const blur = Math.random() > 0.95 ? 1 : 0;
+            return <span key={index} style={{opacity, filter: `blur(${blur}px)`}}>{char}</span>;
+          })}</p>
+        </div>
+        
+        {/* Extra random ink spill */}
+        <div style={{
+          position: 'absolute',
+          width: '10px',
+          height: '20px',
+          top: '60%',
+          left: '25%',
+          background: 'rgba(0,0,0,0.3)',
+          borderRadius: '50%',
+          transform: 'rotate(30deg)',
+          filter: 'blur(1px)',
+          zIndex: 3,
+        }}></div>
+      </div>
+      
+    </div>
+  );
+};
+
+//export default BookPageTemplate;
+window.BookPageTemplate = BookPageTemplate;
\ No newline at end of file