Upload scrpt27.py with huggingface_hub
Browse files- scrpt27.py +93 -97
scrpt27.py
CHANGED
@@ -20,6 +20,7 @@ from magic_pdf.filter.pdf_classify_by_type import classify
|
|
20 |
import fitz # PyMuPDF
|
21 |
import time
|
22 |
import signal
|
|
|
23 |
|
24 |
# Set up logging
|
25 |
logging.basicConfig(level=logging.DEBUG, format='%(asctime)s - %(levelname)s - %(message)s')
|
@@ -28,6 +29,7 @@ logger = logging.getLogger(__name__)
|
|
28 |
# Minimum batch size
|
29 |
MIN_BATCH_SIZE = 1
|
30 |
|
|
|
31 |
def parse_arguments():
|
32 |
parser = argparse.ArgumentParser(description="Process multiple PDFs using Magic PDF")
|
33 |
parser.add_argument("--input", default="input", help="Input folder containing PDF files")
|
@@ -39,31 +41,35 @@ def parse_arguments():
|
|
39 |
parser.add_argument("--initial-batch-size", type=int, default=1, help="Initial batch size for processing")
|
40 |
return parser.parse_args()
|
41 |
|
|
|
42 |
def load_config(config_path):
|
43 |
with open(config_path, 'r') as f:
|
44 |
return json.load(f)
|
45 |
|
|
|
46 |
def get_available_memory(gpu_id):
|
47 |
return torch.cuda.get_device_properties(gpu_id).total_memory - torch.cuda.memory_allocated(gpu_id)
|
48 |
|
|
|
49 |
def extract_images(pdf_path, output_folder):
|
50 |
doc = fitz.open(pdf_path)
|
51 |
pdf_name = os.path.splitext(os.path.basename(pdf_path))[0]
|
52 |
images_folder = os.path.join(output_folder, 'images')
|
53 |
os.makedirs(images_folder, exist_ok=True)
|
54 |
-
|
55 |
for page_num, page in enumerate(doc):
|
56 |
for img_index, img in enumerate(page.get_images(full=True)):
|
57 |
xref = img[0]
|
58 |
base_image = doc.extract_image(xref)
|
59 |
image_bytes = base_image["image"]
|
60 |
image_ext = base_image["ext"]
|
61 |
-
image_filename = f'{pdf_name}_{page_num+1:03d}_{img_index+1:03d}.{image_ext}'
|
62 |
image_path = os.path.join(images_folder, image_filename)
|
63 |
with open(image_path, "wb") as image_file:
|
64 |
image_file.write(image_bytes)
|
65 |
doc.close()
|
66 |
|
|
|
67 |
class MagicModel:
|
68 |
def __init__(self, config):
|
69 |
self.config = config
|
@@ -73,7 +79,8 @@ class MagicModel:
|
|
73 |
with open(log_file_path, 'a') as log_file:
|
74 |
log_file.write(f"Entering process_pdf\n")
|
75 |
log_file.write(f" parse_type: {parse_type}, (expected: str)\n")
|
76 |
-
log_file.write(
|
|
|
77 |
for page_index, page_info in enumerate(layout_info):
|
78 |
try:
|
79 |
with open(log_file_path, 'a') as log_file:
|
@@ -106,6 +113,7 @@ class MagicModel:
|
|
106 |
log_file.write(f"Exiting process_page\n")
|
107 |
return result
|
108 |
|
|
|
109 |
def process_single_pdf(input_file, output_folder, gpu_id, config, timeout, use_bf16, model, log_file_path):
|
110 |
start_time = time.time()
|
111 |
pdf_name = os.path.splitext(os.path.basename(input_file))[0]
|
@@ -134,17 +142,18 @@ def process_single_pdf(input_file, output_folder, gpu_id, config, timeout, use_b
|
|
134 |
torch.set_default_device('cpu')
|
135 |
|
136 |
pdf_data = read_file(input_file, 'rb')
|
137 |
-
|
138 |
# Perform PDF metadata scan
|
139 |
metadata = pdf_meta_scan(pdf_data)
|
140 |
with open(log_file_path, 'a') as log_file:
|
141 |
log_file.write(f"Processing PDF: {input_file}\n")
|
142 |
log_file.write(f"Metadata (expected: dict): {json.dumps(metadata, indent=2)}\n")
|
143 |
-
|
144 |
# Check if metadata indicates the PDF should be dropped
|
145 |
if metadata.get("_need_drop", False):
|
146 |
with open(log_file_path, 'a') as log_file:
|
147 |
-
log_file.write(
|
|
|
148 |
return input_file, "Dropped", None
|
149 |
|
150 |
# Check if all required fields are present in metadata
|
@@ -153,7 +162,7 @@ def process_single_pdf(input_file, output_folder, gpu_id, config, timeout, use_b
|
|
153 |
for field in required_fields:
|
154 |
if field not in metadata:
|
155 |
raise ValueError(f"Required field '{field}' not found in metadata for {input_file}")
|
156 |
-
|
157 |
# Extract required fields for classify function
|
158 |
total_page = metadata['total_page']
|
159 |
page_width = metadata['page_width_pts']
|
@@ -172,26 +181,22 @@ def process_single_pdf(input_file, output_folder, gpu_id, config, timeout, use_b
|
|
172 |
log_file.write(f" img_sz_list (expected: list of lists): {img_sz_list[:5]}...\n")
|
173 |
log_file.write(f" text_len_list (expected: list of ints): {text_len_list[:5]}...\n")
|
174 |
log_file.write(f" img_num_list (expected: list of ints): {img_num_list[:5]}...\n")
|
175 |
-
log_file.write(
|
|
|
176 |
log_file.write(f" invalid_chars (expected: bool): {invalid_chars}\n")
|
177 |
|
178 |
# Classify PDF
|
179 |
-
|
180 |
-
|
181 |
-
|
182 |
-
|
183 |
-
|
184 |
-
|
185 |
-
|
186 |
-
|
187 |
-
|
188 |
-
|
189 |
|
190 |
-
except Exception as e:
|
191 |
-
with open(log_file_path, 'a') as log_file:
|
192 |
-
log_file.write(f"Error in classify function for {input_file}: {str(e)}\n")
|
193 |
-
return input_file, f"Classification Error: {str(e)}", None
|
194 |
-
|
195 |
image_writer = DiskReaderWriter(output_subfolder)
|
196 |
with open(log_file_path, 'a') as log_file:
|
197 |
log_file.write(f"Image writer initialized: {image_writer}\n")
|
@@ -203,7 +208,7 @@ def process_single_pdf(input_file, output_folder, gpu_id, config, timeout, use_b
|
|
203 |
unipipe = UNIPipe(pdf_data, jso_useful_key, image_writer)
|
204 |
with open(log_file_path, 'a') as log_file:
|
205 |
log_file.write(f"UNIPipe initialized: {unipipe}\n")
|
206 |
-
|
207 |
parse_type = unipipe.pipe_classify()
|
208 |
with open(log_file_path, 'a') as log_file:
|
209 |
log_file.write(f"pipe_classify result (expected: str): {parse_type}\n")
|
@@ -212,81 +217,58 @@ def process_single_pdf(input_file, output_folder, gpu_id, config, timeout, use_b
|
|
212 |
with open(log_file_path, 'a') as log_file:
|
213 |
log_file.write(f"Detailed pipe_analyze Inputs for {input_file}:\n")
|
214 |
log_file.write(f" parse_type (expected: str): {parse_type}\n")
|
215 |
-
|
216 |
-
|
217 |
-
|
218 |
-
|
219 |
-
except Exception as e:
|
220 |
-
with open(log_file_path, 'a') as log_file:
|
221 |
-
log_file.write(f"Error in pipe_analyze for {input_file}: {str(e)}\n")
|
222 |
-
return input_file, f"pipe_analyze Error: {str(e)}", None
|
223 |
|
224 |
# Use OCR if it's not classified as a text PDF
|
225 |
if not is_text_pdf:
|
226 |
parse_type = 'ocr'
|
227 |
with open(log_file_path, 'a') as log_file:
|
228 |
-
log_file.write(
|
229 |
-
|
|
|
230 |
# Process the PDF using the model
|
231 |
-
|
232 |
-
|
233 |
-
|
234 |
-
log_file.write(f"Model process_pdf result (expected: dict): {parse_result}\n")
|
235 |
-
except Exception as e:
|
236 |
-
with open(log_file_path, 'a') as log_file:
|
237 |
-
log_file.write(f"Error in model processing for {input_file}: {str(e)}\n")
|
238 |
-
return input_file, f"Model Processing Error: {str(e)}", None
|
239 |
|
240 |
-
|
241 |
-
|
242 |
-
|
243 |
-
|
244 |
-
|
245 |
-
|
246 |
-
|
247 |
-
|
248 |
-
return input_file, f"pipe_mk_markdown Error: {str(e)}", None
|
249 |
|
250 |
-
try:
|
251 |
-
uni_format = unipipe.pipe_mk_uni_format(parse_result)
|
252 |
-
with open(log_file_path, 'a') as log_file:
|
253 |
-
log_file.write(f"pipe_mk_uni_format result (expected: dict): {uni_format}\n")
|
254 |
-
except Exception as e:
|
255 |
-
with open(log_file_path, 'a') as log_file:
|
256 |
-
log_file.write(f"Error in pipe_mk_uni_format for {input_file}: {str(e)}\n")
|
257 |
-
log_file.write(f" parse_result (expected: dict): {parse_result}\n")
|
258 |
-
return input_file, f"pipe_mk_uni_format Error: {str(e)}", None
|
259 |
-
|
260 |
# Write markdown content
|
261 |
with open(os.path.join(output_subfolder, f'{pdf_name}.md'), 'w', encoding='utf-8') as f:
|
262 |
f.write(markdown_content)
|
263 |
-
|
264 |
# Write middle.json
|
265 |
with open(os.path.join(output_subfolder, 'middle.json'), 'w', encoding='utf-8') as f:
|
266 |
json.dump(parse_result, f, ensure_ascii=False, indent=2)
|
267 |
-
|
268 |
# Write model.json
|
269 |
with open(os.path.join(output_subfolder, 'model.json'), 'w', encoding='utf-8') as f:
|
270 |
json.dump(uni_format, f, ensure_ascii=False, indent=2)
|
271 |
-
|
272 |
# Copy original PDF
|
273 |
shutil.copy(input_file, os.path.join(output_subfolder, f'{pdf_name}.pdf'))
|
274 |
-
|
275 |
# Generate layout.pdf and spans.pdf
|
276 |
-
|
277 |
-
|
278 |
-
except Exception as e:
|
279 |
-
with open(log_file_path, 'a') as log_file:
|
280 |
-
log_file.write(f"Error in do_parse for {input_file}: {str(e)}\n")
|
281 |
-
return input_file, f"do_parse Error: {str(e)}", None
|
282 |
-
|
283 |
# Extract images
|
284 |
extract_images(input_file, output_subfolder)
|
285 |
-
|
286 |
processing_time = time.time() - start_time
|
287 |
with open(log_file_path, 'a') as log_file:
|
288 |
-
log_file.write(
|
289 |
-
|
|
|
290 |
# Prepare result for JSONL output
|
291 |
result = {
|
292 |
"file_name": pdf_name,
|
@@ -296,7 +278,7 @@ def process_single_pdf(input_file, output_folder, gpu_id, config, timeout, use_b
|
|
296 |
"classification": classification_results,
|
297 |
"is_text_pdf": is_text_pdf
|
298 |
}
|
299 |
-
|
300 |
return input_file, "Success", result
|
301 |
|
302 |
except ValueError as ve:
|
@@ -310,15 +292,22 @@ def process_single_pdf(input_file, output_folder, gpu_id, config, timeout, use_b
|
|
310 |
return input_file, "Timeout", None
|
311 |
|
312 |
except Exception as e:
|
313 |
-
|
314 |
-
|
315 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
316 |
|
317 |
finally:
|
318 |
signal.alarm(0) # Cancel the alarm
|
319 |
if gpu_id >= 0:
|
320 |
torch.cuda.empty_cache()
|
321 |
|
|
|
322 |
def process_pdf_batch(batch, output_folder, gpu_id, config, timeout, use_bf16, model, log_file_path):
|
323 |
results = []
|
324 |
for pdf_file in batch:
|
@@ -326,6 +315,7 @@ def process_pdf_batch(batch, output_folder, gpu_id, config, timeout, use_bf16, m
|
|
326 |
results.append(result)
|
327 |
return results
|
328 |
|
|
|
329 |
def write_to_jsonl(results, output_file):
|
330 |
with open(output_file, 'a') as f:
|
331 |
for result in results:
|
@@ -333,6 +323,7 @@ def write_to_jsonl(results, output_file):
|
|
333 |
json.dump(result[2], f)
|
334 |
f.write('\n')
|
335 |
|
|
|
336 |
def get_gpu_memory_usage(gpu_id):
|
337 |
if gpu_id < 0:
|
338 |
return 0, 0 # CPU mode
|
@@ -340,18 +331,19 @@ def get_gpu_memory_usage(gpu_id):
|
|
340 |
allocated_memory = torch.cuda.memory_allocated(gpu_id)
|
341 |
return allocated_memory, total_memory
|
342 |
|
|
|
343 |
def main():
|
344 |
mp.set_start_method('spawn', force=True)
|
345 |
-
|
346 |
args = parse_arguments()
|
347 |
config = load_config(args.config)
|
348 |
-
|
349 |
input_folder = args.input
|
350 |
output_folder = args.output
|
351 |
os.makedirs(output_folder, exist_ok=True)
|
352 |
-
|
353 |
pdf_files = [os.path.join(input_folder, f) for f in os.listdir(input_folder) if f.endswith('.pdf')]
|
354 |
-
|
355 |
num_gpus = torch.cuda.device_count()
|
356 |
if num_gpus == 0:
|
357 |
print("No GPUs available. Using CPU.")
|
@@ -359,20 +351,20 @@ def main():
|
|
359 |
gpu_ids = [-1]
|
360 |
else:
|
361 |
gpu_ids = list(range(num_gpus))
|
362 |
-
|
363 |
num_workers = args.max_workers or min(num_gpus, os.cpu_count())
|
364 |
-
|
365 |
main_jsonl = os.path.join(output_folder, 'processing_results.jsonl')
|
366 |
temp_jsonl = os.path.join(output_folder, 'temp_results.jsonl')
|
367 |
log_file_path = os.path.join(output_folder, 'processing_log.txt')
|
368 |
-
|
369 |
# Enable deterministic mode
|
370 |
torch.backends.cudnn.deterministic = True
|
371 |
torch.backends.cudnn.benchmark = False
|
372 |
-
|
373 |
# Load the model
|
374 |
model = MagicModel(config)
|
375 |
-
|
376 |
results = []
|
377 |
with ProcessPoolExecutor(max_workers=num_workers) as executor:
|
378 |
for gpu_id in gpu_ids:
|
@@ -382,16 +374,18 @@ def main():
|
|
382 |
while pdf_index < len(pdf_files):
|
383 |
batch = pdf_files[pdf_index:pdf_index + batch_size]
|
384 |
try:
|
385 |
-
future = executor.submit(process_pdf_batch, batch, output_folder, gpu_id, config, args.timeout,
|
|
|
386 |
batch_results = future.result()
|
387 |
results.extend(batch_results)
|
388 |
for result in batch_results:
|
389 |
write_to_jsonl([result], temp_jsonl)
|
390 |
-
|
391 |
# Print VRAM usage
|
392 |
allocated, total = get_gpu_memory_usage(gpu_id)
|
393 |
with open(log_file_path, 'a') as log_file:
|
394 |
-
log_file.write(
|
|
|
395 |
# If successful and OOM hasn't occurred yet, increase batch size
|
396 |
if not oom_occurred:
|
397 |
batch_size += 1
|
@@ -404,24 +398,25 @@ def main():
|
|
404 |
log_file.write(f"OOM error occurred. Reducing batch size to {batch_size}\n")
|
405 |
torch.cuda.empty_cache()
|
406 |
continue
|
407 |
-
|
408 |
# After processing each batch, move temp JSONL to main JSONL
|
409 |
if os.path.exists(temp_jsonl):
|
410 |
with open(temp_jsonl, 'r') as temp, open(main_jsonl, 'a') as main:
|
411 |
shutil.copyfileobj(temp, main)
|
412 |
os.remove(temp_jsonl)
|
413 |
-
|
414 |
# Clear GPU cache after each batch
|
415 |
if gpu_id >= 0:
|
416 |
torch.cuda.empty_cache()
|
417 |
-
|
418 |
success_count = sum(1 for _, status, _ in results if status == "Success")
|
419 |
timeout_count = sum(1 for _, status, _ in results if status == "Timeout")
|
420 |
error_count = len(results) - success_count - timeout_count
|
421 |
-
|
422 |
with open(log_file_path, 'a') as log_file:
|
423 |
-
log_file.write(
|
424 |
-
|
|
|
425 |
with open(os.path.join(output_folder, 'processing_summary.txt'), 'w') as summary:
|
426 |
summary.write(f"Total PDFs processed: {len(results)}\n")
|
427 |
summary.write(f"Successful: {success_count}\n")
|
@@ -431,5 +426,6 @@ def main():
|
|
431 |
for pdf, status, _ in [result for result in results if result[1] != "Success"]:
|
432 |
summary.write(f" - {pdf}: {status}\n")
|
433 |
|
|
|
434 |
if __name__ == '__main__':
|
435 |
main()
|
|
|
20 |
import fitz # PyMuPDF
|
21 |
import time
|
22 |
import signal
|
23 |
+
import traceback
|
24 |
|
25 |
# Set up logging
|
26 |
logging.basicConfig(level=logging.DEBUG, format='%(asctime)s - %(levelname)s - %(message)s')
|
|
|
29 |
# Minimum batch size
|
30 |
MIN_BATCH_SIZE = 1
|
31 |
|
32 |
+
|
33 |
def parse_arguments():
|
34 |
parser = argparse.ArgumentParser(description="Process multiple PDFs using Magic PDF")
|
35 |
parser.add_argument("--input", default="input", help="Input folder containing PDF files")
|
|
|
41 |
parser.add_argument("--initial-batch-size", type=int, default=1, help="Initial batch size for processing")
|
42 |
return parser.parse_args()
|
43 |
|
44 |
+
|
45 |
def load_config(config_path):
|
46 |
with open(config_path, 'r') as f:
|
47 |
return json.load(f)
|
48 |
|
49 |
+
|
50 |
def get_available_memory(gpu_id):
|
51 |
return torch.cuda.get_device_properties(gpu_id).total_memory - torch.cuda.memory_allocated(gpu_id)
|
52 |
|
53 |
+
|
54 |
def extract_images(pdf_path, output_folder):
|
55 |
doc = fitz.open(pdf_path)
|
56 |
pdf_name = os.path.splitext(os.path.basename(pdf_path))[0]
|
57 |
images_folder = os.path.join(output_folder, 'images')
|
58 |
os.makedirs(images_folder, exist_ok=True)
|
59 |
+
|
60 |
for page_num, page in enumerate(doc):
|
61 |
for img_index, img in enumerate(page.get_images(full=True)):
|
62 |
xref = img[0]
|
63 |
base_image = doc.extract_image(xref)
|
64 |
image_bytes = base_image["image"]
|
65 |
image_ext = base_image["ext"]
|
66 |
+
image_filename = f'{pdf_name}_{page_num + 1:03d}_{img_index + 1:03d}.{image_ext}'
|
67 |
image_path = os.path.join(images_folder, image_filename)
|
68 |
with open(image_path, "wb") as image_file:
|
69 |
image_file.write(image_bytes)
|
70 |
doc.close()
|
71 |
|
72 |
+
|
73 |
class MagicModel:
|
74 |
def __init__(self, config):
|
75 |
self.config = config
|
|
|
79 |
with open(log_file_path, 'a') as log_file:
|
80 |
log_file.write(f"Entering process_pdf\n")
|
81 |
log_file.write(f" parse_type: {parse_type}, (expected: str)\n")
|
82 |
+
log_file.write(
|
83 |
+
f" layout_info (length: {len(layout_info)}), (expected: list of dicts): {layout_info}\n")
|
84 |
for page_index, page_info in enumerate(layout_info):
|
85 |
try:
|
86 |
with open(log_file_path, 'a') as log_file:
|
|
|
113 |
log_file.write(f"Exiting process_page\n")
|
114 |
return result
|
115 |
|
116 |
+
|
117 |
def process_single_pdf(input_file, output_folder, gpu_id, config, timeout, use_bf16, model, log_file_path):
|
118 |
start_time = time.time()
|
119 |
pdf_name = os.path.splitext(os.path.basename(input_file))[0]
|
|
|
142 |
torch.set_default_device('cpu')
|
143 |
|
144 |
pdf_data = read_file(input_file, 'rb')
|
145 |
+
|
146 |
# Perform PDF metadata scan
|
147 |
metadata = pdf_meta_scan(pdf_data)
|
148 |
with open(log_file_path, 'a') as log_file:
|
149 |
log_file.write(f"Processing PDF: {input_file}\n")
|
150 |
log_file.write(f"Metadata (expected: dict): {json.dumps(metadata, indent=2)}\n")
|
151 |
+
|
152 |
# Check if metadata indicates the PDF should be dropped
|
153 |
if metadata.get("_need_drop", False):
|
154 |
with open(log_file_path, 'a') as log_file:
|
155 |
+
log_file.write(
|
156 |
+
f"Dropping PDF {input_file}: {metadata.get('_drop_reason', 'Unknown reason')}\n")
|
157 |
return input_file, "Dropped", None
|
158 |
|
159 |
# Check if all required fields are present in metadata
|
|
|
162 |
for field in required_fields:
|
163 |
if field not in metadata:
|
164 |
raise ValueError(f"Required field '{field}' not found in metadata for {input_file}")
|
165 |
+
|
166 |
# Extract required fields for classify function
|
167 |
total_page = metadata['total_page']
|
168 |
page_width = metadata['page_width_pts']
|
|
|
181 |
log_file.write(f" img_sz_list (expected: list of lists): {img_sz_list[:5]}...\n")
|
182 |
log_file.write(f" text_len_list (expected: list of ints): {text_len_list[:5]}...\n")
|
183 |
log_file.write(f" img_num_list (expected: list of ints): {img_num_list[:5]}...\n")
|
184 |
+
log_file.write(
|
185 |
+
f" text_layout_list (expected: list of strs): {text_layout_list[:5]}...\n")
|
186 |
log_file.write(f" invalid_chars (expected: bool): {invalid_chars}\n")
|
187 |
|
188 |
# Classify PDF
|
189 |
+
is_text_pdf, classification_results = classify(
|
190 |
+
total_page, page_width, page_height, img_sz_list[:total_page],
|
191 |
+
text_len_list[:total_page], img_num_list[:total_page],
|
192 |
+
text_layout_list[:len(text_layout_list)], invalid_chars
|
193 |
+
)
|
194 |
+
with open(log_file_path, 'a') as log_file:
|
195 |
+
log_file.write(f"Classification Results:\n")
|
196 |
+
log_file.write(f" is_text_pdf (expected: bool): {is_text_pdf}\n")
|
197 |
+
log_file.write(
|
198 |
+
f" classification_results (expected: dict): {classification_results}\n")
|
199 |
|
|
|
|
|
|
|
|
|
|
|
200 |
image_writer = DiskReaderWriter(output_subfolder)
|
201 |
with open(log_file_path, 'a') as log_file:
|
202 |
log_file.write(f"Image writer initialized: {image_writer}\n")
|
|
|
208 |
unipipe = UNIPipe(pdf_data, jso_useful_key, image_writer)
|
209 |
with open(log_file_path, 'a') as log_file:
|
210 |
log_file.write(f"UNIPipe initialized: {unipipe}\n")
|
211 |
+
|
212 |
parse_type = unipipe.pipe_classify()
|
213 |
with open(log_file_path, 'a') as log_file:
|
214 |
log_file.write(f"pipe_classify result (expected: str): {parse_type}\n")
|
|
|
217 |
with open(log_file_path, 'a') as log_file:
|
218 |
log_file.write(f"Detailed pipe_analyze Inputs for {input_file}:\n")
|
219 |
log_file.write(f" parse_type (expected: str): {parse_type}\n")
|
220 |
+
layout_info = unipipe.pipe_analyze()
|
221 |
+
with open(log_file_path, 'a') as log_file:
|
222 |
+
log_file.write(
|
223 |
+
f"pipe_analyze Results (expected: list of dicts, length: {len(layout_info)}): {layout_info}\n")
|
|
|
|
|
|
|
|
|
224 |
|
225 |
# Use OCR if it's not classified as a text PDF
|
226 |
if not is_text_pdf:
|
227 |
parse_type = 'ocr'
|
228 |
with open(log_file_path, 'a') as log_file:
|
229 |
+
log_file.write(
|
230 |
+
f"parse_type after OCR check (expected: str): {parse_type}\n")
|
231 |
+
|
232 |
# Process the PDF using the model
|
233 |
+
parse_result = model.process_pdf(pdf_data, parse_type, layout_info, log_file_path)
|
234 |
+
with open(log_file_path, 'a') as log_file:
|
235 |
+
log_file.write(f"Model process_pdf result (expected: dict): {parse_result}\n")
|
|
|
|
|
|
|
|
|
|
|
236 |
|
237 |
+
markdown_content = unipipe.pipe_mk_markdown(parse_result)
|
238 |
+
with open(log_file_path, 'a') as log_file:
|
239 |
+
log_file.write(
|
240 |
+
f"pipe_mk_markdown result (expected: str, length: {len(markdown_content)}): {markdown_content}\n")
|
241 |
+
|
242 |
+
uni_format = unipipe.pipe_mk_uni_format(parse_result)
|
243 |
+
with open(log_file_path, 'a') as log_file:
|
244 |
+
log_file.write(f"pipe_mk_uni_format result (expected: dict): {uni_format}\n")
|
|
|
245 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
246 |
# Write markdown content
|
247 |
with open(os.path.join(output_subfolder, f'{pdf_name}.md'), 'w', encoding='utf-8') as f:
|
248 |
f.write(markdown_content)
|
249 |
+
|
250 |
# Write middle.json
|
251 |
with open(os.path.join(output_subfolder, 'middle.json'), 'w', encoding='utf-8') as f:
|
252 |
json.dump(parse_result, f, ensure_ascii=False, indent=2)
|
253 |
+
|
254 |
# Write model.json
|
255 |
with open(os.path.join(output_subfolder, 'model.json'), 'w', encoding='utf-8') as f:
|
256 |
json.dump(uni_format, f, ensure_ascii=False, indent=2)
|
257 |
+
|
258 |
# Copy original PDF
|
259 |
shutil.copy(input_file, os.path.join(output_subfolder, f'{pdf_name}.pdf'))
|
260 |
+
|
261 |
# Generate layout.pdf and spans.pdf
|
262 |
+
do_parse(input_file, parse_type, output_subfolder, draw_bbox=True)
|
263 |
+
|
|
|
|
|
|
|
|
|
|
|
264 |
# Extract images
|
265 |
extract_images(input_file, output_subfolder)
|
266 |
+
|
267 |
processing_time = time.time() - start_time
|
268 |
with open(log_file_path, 'a') as log_file:
|
269 |
+
log_file.write(
|
270 |
+
f"Successfully processed {input_file} on GPU {gpu_id} in {processing_time:.2f} seconds\n")
|
271 |
+
|
272 |
# Prepare result for JSONL output
|
273 |
result = {
|
274 |
"file_name": pdf_name,
|
|
|
278 |
"classification": classification_results,
|
279 |
"is_text_pdf": is_text_pdf
|
280 |
}
|
281 |
+
|
282 |
return input_file, "Success", result
|
283 |
|
284 |
except ValueError as ve:
|
|
|
292 |
return input_file, "Timeout", None
|
293 |
|
294 |
except Exception as e:
|
295 |
+
# Save full traceback to a file
|
296 |
+
traceback_file = os.path.join(output_folder, 'traceback.txt')
|
297 |
+
with open(traceback_file, 'w') as f:
|
298 |
+
f.write(traceback.format_exc())
|
299 |
+
|
300 |
+
# Print error message and traceback location to CLI
|
301 |
+
print(f"Error occurred: {e}")
|
302 |
+
print(f"Full traceback saved to: {traceback_file}")
|
303 |
+
exit(1) # Terminate the script
|
304 |
|
305 |
finally:
|
306 |
signal.alarm(0) # Cancel the alarm
|
307 |
if gpu_id >= 0:
|
308 |
torch.cuda.empty_cache()
|
309 |
|
310 |
+
|
311 |
def process_pdf_batch(batch, output_folder, gpu_id, config, timeout, use_bf16, model, log_file_path):
|
312 |
results = []
|
313 |
for pdf_file in batch:
|
|
|
315 |
results.append(result)
|
316 |
return results
|
317 |
|
318 |
+
|
319 |
def write_to_jsonl(results, output_file):
|
320 |
with open(output_file, 'a') as f:
|
321 |
for result in results:
|
|
|
323 |
json.dump(result[2], f)
|
324 |
f.write('\n')
|
325 |
|
326 |
+
|
327 |
def get_gpu_memory_usage(gpu_id):
|
328 |
if gpu_id < 0:
|
329 |
return 0, 0 # CPU mode
|
|
|
331 |
allocated_memory = torch.cuda.memory_allocated(gpu_id)
|
332 |
return allocated_memory, total_memory
|
333 |
|
334 |
+
|
335 |
def main():
|
336 |
mp.set_start_method('spawn', force=True)
|
337 |
+
|
338 |
args = parse_arguments()
|
339 |
config = load_config(args.config)
|
340 |
+
|
341 |
input_folder = args.input
|
342 |
output_folder = args.output
|
343 |
os.makedirs(output_folder, exist_ok=True)
|
344 |
+
|
345 |
pdf_files = [os.path.join(input_folder, f) for f in os.listdir(input_folder) if f.endswith('.pdf')]
|
346 |
+
|
347 |
num_gpus = torch.cuda.device_count()
|
348 |
if num_gpus == 0:
|
349 |
print("No GPUs available. Using CPU.")
|
|
|
351 |
gpu_ids = [-1]
|
352 |
else:
|
353 |
gpu_ids = list(range(num_gpus))
|
354 |
+
|
355 |
num_workers = args.max_workers or min(num_gpus, os.cpu_count())
|
356 |
+
|
357 |
main_jsonl = os.path.join(output_folder, 'processing_results.jsonl')
|
358 |
temp_jsonl = os.path.join(output_folder, 'temp_results.jsonl')
|
359 |
log_file_path = os.path.join(output_folder, 'processing_log.txt')
|
360 |
+
|
361 |
# Enable deterministic mode
|
362 |
torch.backends.cudnn.deterministic = True
|
363 |
torch.backends.cudnn.benchmark = False
|
364 |
+
|
365 |
# Load the model
|
366 |
model = MagicModel(config)
|
367 |
+
|
368 |
results = []
|
369 |
with ProcessPoolExecutor(max_workers=num_workers) as executor:
|
370 |
for gpu_id in gpu_ids:
|
|
|
374 |
while pdf_index < len(pdf_files):
|
375 |
batch = pdf_files[pdf_index:pdf_index + batch_size]
|
376 |
try:
|
377 |
+
future = executor.submit(process_pdf_batch, batch, output_folder, gpu_id, config, args.timeout,
|
378 |
+
args.use_bf16, model, log_file_path)
|
379 |
batch_results = future.result()
|
380 |
results.extend(batch_results)
|
381 |
for result in batch_results:
|
382 |
write_to_jsonl([result], temp_jsonl)
|
383 |
+
|
384 |
# Print VRAM usage
|
385 |
allocated, total = get_gpu_memory_usage(gpu_id)
|
386 |
with open(log_file_path, 'a') as log_file:
|
387 |
+
log_file.write(
|
388 |
+
f"GPU {gpu_id} - Batch size: {batch_size}, VRAM usage: {allocated / 1024 ** 3:.2f}GB / {total / 1024 ** 3:.2f}GB\n")
|
389 |
# If successful and OOM hasn't occurred yet, increase batch size
|
390 |
if not oom_occurred:
|
391 |
batch_size += 1
|
|
|
398 |
log_file.write(f"OOM error occurred. Reducing batch size to {batch_size}\n")
|
399 |
torch.cuda.empty_cache()
|
400 |
continue
|
401 |
+
|
402 |
# After processing each batch, move temp JSONL to main JSONL
|
403 |
if os.path.exists(temp_jsonl):
|
404 |
with open(temp_jsonl, 'r') as temp, open(main_jsonl, 'a') as main:
|
405 |
shutil.copyfileobj(temp, main)
|
406 |
os.remove(temp_jsonl)
|
407 |
+
|
408 |
# Clear GPU cache after each batch
|
409 |
if gpu_id >= 0:
|
410 |
torch.cuda.empty_cache()
|
411 |
+
|
412 |
success_count = sum(1 for _, status, _ in results if status == "Success")
|
413 |
timeout_count = sum(1 for _, status, _ in results if status == "Timeout")
|
414 |
error_count = len(results) - success_count - timeout_count
|
415 |
+
|
416 |
with open(log_file_path, 'a') as log_file:
|
417 |
+
log_file.write(
|
418 |
+
f"Processed {len(results)} PDFs. {success_count} succeeded, {timeout_count} timed out, {error_count} failed.\n")
|
419 |
+
|
420 |
with open(os.path.join(output_folder, 'processing_summary.txt'), 'w') as summary:
|
421 |
summary.write(f"Total PDFs processed: {len(results)}\n")
|
422 |
summary.write(f"Successful: {success_count}\n")
|
|
|
426 |
for pdf, status, _ in [result for result in results if result[1] != "Success"]:
|
427 |
summary.write(f" - {pdf}: {status}\n")
|
428 |
|
429 |
+
|
430 |
if __name__ == '__main__':
|
431 |
main()
|