| import argparse
|
| import os
|
| import json
|
| import datetime
|
| from pypdf import PdfReader
|
| import glob
|
| import re
|
|
|
| def extract_text_from_pdf(pdf_path):
|
| try:
|
| reader = PdfReader(pdf_path)
|
| text = ""
|
| for page in reader.pages:
|
| text += page.extract_text() or ""
|
|
|
| return text.replace("\n", "").replace(" ", "")
|
| except Exception as e:
|
| return f"[ERROR] {str(e)}"
|
|
|
| def evaluate(split_dir, truth_dir, result_path):
|
| log = {
|
| "Process": True,
|
| "Result": False,
|
| "TimePoint": datetime.datetime.now().isoformat(),
|
| "comments": ""
|
| }
|
|
|
| try:
|
|
|
| split_files = glob.glob(os.path.join(split_dir, "**", "*.pdf"), recursive=True)
|
| truth_files = glob.glob(os.path.join(truth_dir, "**", "*.pdf"), recursive=True)
|
|
|
|
|
| split_map = {}
|
| for f in split_files:
|
| m = re.search(r"(\d+)", os.path.basename(f))
|
| if m:
|
| split_map[int(m.group(1))] = f
|
| truth_map = {}
|
| for f in truth_files:
|
| m = re.search(r"(\d+)", os.path.basename(f))
|
| if m:
|
| truth_map[int(m.group(1))] = f
|
|
|
|
|
| if not truth_map:
|
| log["Process"] = False
|
| log["comments"] += "❌ No valid .pdf files in ground truth directory"
|
| if not split_map:
|
| log["Process"] = False
|
| log["comments"] += "❌ No valid .pdf files in split directory"
|
| if not log["Process"]:
|
| write_result(result_path, log)
|
| return
|
|
|
| total_pages = len(truth_map)
|
| passed_pages = 0
|
| page_logs = []
|
|
|
|
|
| for page_num in sorted(truth_map.keys()):
|
| truth_file = truth_map[page_num]
|
| split_file = split_map.get(page_num)
|
| if not split_file:
|
| page_logs.append(f"❌ Page {page_num} split file not found")
|
| continue
|
|
|
| split_text = extract_text_from_pdf(split_file)
|
| truth_text = extract_text_from_pdf(truth_file)
|
|
|
|
|
| if isinstance(split_text, str) and split_text.startswith("[ERROR]"):
|
| page_logs.append(f"❌ Error reading {split_file}: {split_text}")
|
| continue
|
| if isinstance(truth_text, str) and truth_text.startswith("[ERROR]"):
|
| page_logs.append(f"❌ Error reading {truth_file}: {truth_text}")
|
| continue
|
|
|
| if not truth_text:
|
| page_logs.append(f"⚠️ Page {page_num} ground truth is empty, skipping")
|
| passed_pages += 1
|
| continue
|
|
|
|
|
| correct_chars = sum(1 for a, b in zip(split_text, truth_text) if a == b)
|
| accuracy = (correct_chars / len(truth_text)) * 100 if truth_text else 0
|
|
|
| if accuracy >= 95:
|
| passed_pages += 1
|
| else:
|
| page_logs.append(f"❌ Page {page_num} accuracy {accuracy:.2f}% < 95%")
|
|
|
| if passed_pages == total_pages:
|
| log["Result"] = True
|
| log["comments"] = f"✅ All {total_pages} pages passed accuracy >=95% check"
|
| else:
|
| log["comments"] = f"❌ {total_pages - passed_pages} pages failed accuracy requirement:\n" + "\n".join(page_logs)
|
|
|
| except Exception as e:
|
| log["Process"] = False
|
| log["comments"] += f"[Exception error] {str(e)}"
|
|
|
| print(log["comments"])
|
| write_result(result_path, log)
|
|
|
|
|
| def write_result(result_path, log):
|
| os.makedirs(os.path.dirname(result_path) or '.', exist_ok=True)
|
| with open(result_path, "a", encoding="utf-8") as f:
|
| f.write(json.dumps(log, ensure_ascii=False, default=str) + "\n")
|
|
|
| if __name__ == "__main__":
|
| parser = argparse.ArgumentParser()
|
| parser.add_argument("--output", type=str, required=True, help="Path to split pages directory")
|
| parser.add_argument("--groundtruth", type=str, required=True, help="Path to ground truth pages directory")
|
| parser.add_argument("--result", type=str, required=True, help="Path to output JSONL results file")
|
| args = parser.parse_args()
|
|
|
| evaluate(args.output, args.groundtruth, args.result) |
