Kessinen/notes-cleaner
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Implement parallel processing for optimization steps using ThreadPoolExecutor.

Browse Source
This commit is contained in:
Esa Kataja
2024-12-03 23:45:35 +02:00
parent 825338a2a3
commit 6e083f39e8
1 changed files with 37 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files
pdf_cleaner.py
+37 -23
View File
@@ -15,6 +15,7 @@ from pdf2image import convert_from_path
import settings import settings
from settings import OptimizationLevel from settings import OptimizationLevel
from PIL import Image from PIL import Image
import concurrent.futures
class Settings: class Settings:
TRIM_PADDING_PIXELS = 20 TRIM_PADDING_PIXELS = 20
@@ -326,37 +327,50 @@ def optimize(level):
# Step 1: Convert to monochrome if level >= 2 # Step 1: Convert to monochrome if level >= 2
if int(opt_level) >= int(OptimizationLevel.MONOCHROME): if int(opt_level) >= int(OptimizationLevel.MONOCHROME):
print("\nConverting to monochrome...") print("\nConverting to monochrome...")
for i, file_path in enumerate(temp_files, 1): with concurrent.futures.ThreadPoolExecutor() as executor:
print(f"[{i}/{total_files}] Converting {os.path.basename(file_path)}...", end='', flush=True) futures = {executor.submit(convert_to_monochrome, file_path, (opt_level == OptimizationLevel.MONOCHROME)): file_path for file_path in temp_files}
# Only optimize if this is the final step (level 2) for future in concurrent.futures.as_completed(futures):
if convert_to_monochrome(file_path, is_final=(opt_level == OptimizationLevel.MONOCHROME)): file_path = futures[future]
successful['monochrome'] += 1 try:
print(" ") if future.result():
else: successful['monochrome'] += 1
print(" ") print(f"Converted {os.path.basename(file_path)}")
else:
print(f"Failed to convert {os.path.basename(file_path)}")
except Exception as exc:
print(f"{os.path.basename(file_path)} generated an exception: {exc}")
# Step 2: Always trim whitespace # Step 2: Always trim whitespace
print("\nTrimming whitespace from images...") print("\nTrimming whitespace from images...")
for i, file_path in enumerate(temp_files, 1): with concurrent.futures.ThreadPoolExecutor() as executor:
print(f"[{i}/{total_files}] Processing {os.path.basename(file_path)}...", end='', flush=True) futures = {executor.submit(trim_whitespace, file_path, (opt_level == OptimizationLevel.TRIM)): file_path for file_path in temp_files}
# Only optimize if this is the final step (level 1) for future in concurrent.futures.as_completed(futures):
if trim_whitespace(file_path, is_final=(opt_level == OptimizationLevel.TRIM)): file_path = futures[future]
successful['trim'] += 1 try:
print(" ") if future.result():
else: successful['trim'] += 1
print(" ") print(f"Trimmed {os.path.basename(file_path)}")
else:
print(f"Failed to trim {os.path.basename(file_path)}")
except Exception as exc:
print(f"{os.path.basename(file_path)} generated an exception: {exc}")
# Step 3: Run optipng if level = 3 # Step 3: Run optipng if level = 3
if opt_level == OptimizationLevel.FULL: if opt_level == OptimizationLevel.FULL:
if check_optipng_installed(): if check_optipng_installed():
print("\nOptimizing PNG files with optipng...") print("\nOptimizing PNG files with optipng...")
for i, file_path in enumerate(temp_files, 1): with concurrent.futures.ThreadPoolExecutor() as executor:
print(f"[{i}/{total_files}] Optimizing {os.path.basename(file_path)}...", end='', flush=True) futures = {executor.submit(run_optipng, file_path): file_path for file_path in temp_files}
if run_optipng(file_path): for future in concurrent.futures.as_completed(futures):
successful['optipng'] += 1 file_path = futures[future]
print(" ") try:
else: if future.result():
print(" ") successful['optipng'] += 1
print(f"Optimized {os.path.basename(file_path)}")
else:
print(f"Failed to optimize {os.path.basename(file_path)}")
except Exception as exc:
print(f"{os.path.basename(file_path)} generated an exception: {exc}")
else: else:
print("\nNote: optipng not found. Skipping PNG optimization.") print("\nNote: optipng not found. Skipping PNG optimization.")