Something went wrong. Try again.
[READ-ONLY] Mirror of https://github.com/mrgnw/scripts.
Something went wrong. Try again.
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403#!/usr/bin/env -S uv run --script# /// script# requires-python = ">=3.10"# dependencies = [# "pillow",# "pillow-heif",# "pymupdf",# "tqdm",# "Send2Trash",# ]# ///"""Extract images from PDFs using PyMuPDF (no external poppler dependency), convert to HEIC by default, and optimize outputs.
Usage: pdfimgs [directory] - If directory is not specified, uses current working directory - Phase 1: For each PDF found: - If PDF has 1 image: saves it directly as {pdf_name}.{format} - If PDF has multiple images: saves them in {pdf_name}/ subdirectory - Phase 2: Optimize all extracted images using optimizt (external CLI, optional)"""
import sysimport subprocessfrom concurrent.futures import ThreadPoolExecutor, as_completedfrom pathlib import Pathfrom typing import Optional
import fitz # PyMuPDFfrom PIL import Imageimport pillow_heiffrom tqdm import tqdmfrom send2trash import send2trash# Route all prints through tqdm.write to avoid breaking progress barsimport builtins
def _tqdm_print(*args, sep=" ", end="\n", file=None, flush=False): s = sep.join(str(a) for a in args) tqdm.write(s, file=file or sys.stdout, end=end) if flush: try: (file or sys.stdout).flush() except Exception: pass
builtins.print = _tqdm_print
# Register HEIF handler with Pillowpillow_heif.register_heif_opener()
def convert_to_heic(image_path: Path, quality: int = 60, overwrite: bool = True, conversion_bar: Optional[tqdm] = None) -> Optional[Path]: """ Convert a single image file to HEIC using Pillow + pillow_heif. Returns the output .heic path on success, or None on failure.
- quality: 0..100 (higher = larger file, better visual quality) - overwrite: if True, remove the original file after successful conversion """ try: with Image.open(image_path) as im: if im.mode in ("P", "1"): im = im.convert("RGBA") # Compute final HEIC name: # If this is a temp file like ".__tmp__{pdf_stem}__img-1-1.png", strip the temp prefix name = image_path.name # Our temp pattern is something like ".__tmp__{pdf_stem}__img-1-1.png" # To get the final base name, take the segment after the last "__" base = name.split("__")[-1] if "__" in name else name final_name = Path(base).with_suffix(".heic").name out_path = image_path.parent / final_name im.save(out_path, format="HEIF", quality=quality) if conversion_bar is not None: conversion_bar.update(1) except Exception as e: print(f"Error converting {image_path.name} to HEIC: {e}") return None
if overwrite: try: image_path.unlink(missing_ok=True) except Exception: pass
return out_path
def extract_images_from_pdf( pdf_path: Path, output_dir: Path, heic_enabled: bool = False, heic_quality: int = 60, heic_overwrite: bool = True, convert_executor: Optional[ThreadPoolExecutor] = None, conversion_futures: Optional[list] = None, image_extract_bar: Optional[tqdm] = None, conversion_bar: Optional[tqdm] = None,) -> Path | None: """ Extract images from a PDF using PyMuPDF.
Returns the path to the extracted content (file or folder) or None if no images found or on error. """ if not pdf_path.exists(): print(f"Error: file not found: {pdf_path}") return None
try: doc = fitz.open(pdf_path) except Exception as e: print(f"Error opening {pdf_path.name}: {e}") return None
extracted_files: list[Path] = []
try: for page_index in range(len(doc)): page = doc[page_index] # full=True includes images used in transparency/softmasks; default False is fine, but we'll keep full for parity for img in page.get_images(full=True): xref = img[0] try: img_info = doc.extract_image(xref) except Exception as e: print(f"Error extracting image xref {xref} from page {page_index+1} in {pdf_path.name}: {e}") continue
image_bytes: Optional[bytes] = img_info.get("image") ext: str = img_info.get("ext", "png") # default to png if extension is missing if not image_bytes: # Sometimes PyMuPDF may return None if image can't be extracted print(f"Warning: no bytes for image xref {xref} in {pdf_path.name}") continue
# Build a deterministic name similar to pdfimages: img-{pageindex}-{seq}.{ext} # We use the count of images extracted so far to create a simple sequence. seq = len(extracted_files) + 1 name = f"img-{page_index+1}-{seq}.{ext}"
# Write into output; if HEIC is enabled, stream into a per-PDF subdirectory and submit conversion if heic_enabled: subdir = output_dir / pdf_path.stem subdir.mkdir(exist_ok=True) temp_path = subdir / f".__tmp__{pdf_path.stem}__{name}" else: temp_path = output_dir / f".__tmp__{pdf_path.stem}__{name}" try: temp_path.write_bytes(image_bytes) extracted_files.append(temp_path) if image_extract_bar is not None: image_extract_bar.update(1) # Stream conversion immediately if enabled if heic_enabled and convert_executor is not None: if conversion_bar is not None: conversion_bar.total += 1 conversion_bar.refresh() fut = convert_executor.submit( convert_to_heic, temp_path, quality=heic_quality, overwrite=heic_overwrite, conversion_bar=conversion_bar, ) if conversion_futures is not None: conversion_futures.append(fut) except Exception as e: print(f"Error writing image {name} from {pdf_path.name}: {e}") continue finally: doc.close()
# No images found if not extracted_files: print(f"No images found in {pdf_path.name}") return None
# Determine output location based on number of images pdf_name_no_ext = pdf_path.stem
# If HEIC is enabled, we already wrote into a per-PDF subdirectory and streamed conversions if heic_enabled: subdir = output_dir / pdf_name_no_ext # Do not rename or delete temp files here; conversion tasks read them and write final HEICs. return subdir
# HEIC disabled: keep original single/multi behavior if len(extracted_files) == 1: # Single image: save directly with PDF name preserving image extension image_path = extracted_files[0] ext = image_path.suffix output_path = output_dir / f"{pdf_name_no_ext}{ext}" try: image_path.replace(output_path) except Exception: # Fallback to copy then remove temp data = image_path.read_bytes() output_path.write_bytes(data) try: image_path.unlink(missing_ok=True) except Exception: pass # log suppressed to keep tqdm clean return output_path else: # Multiple images: create subdirectory subdir = output_dir / pdf_name_no_ext subdir.mkdir(exist_ok=True) for image_path in extracted_files: output_path = subdir / image_path.name.replace(f".__tmp__{pdf_name_no_ext}__", "") try: image_path.replace(output_path) except Exception: data = image_path.read_bytes() output_path.write_bytes(data) try: image_path.unlink(missing_ok=True) except Exception: pass return subdir
def optimize_images(path: Path) -> None: """ Optimize images using optimizt (external CLI).
If path is a directory, optimizes all contents recursively. If path is a file, optimizes the file.
Note: This step depends on 'optimizt' being installed in your environment (e.g., via npm). If not available, this step will be skipped with a message. """ try: subprocess.run( ["optimizt", str(path)], check=True, capture_output=True, ) if path.is_dir(): print(f"Optimized images in {path.name}/") else: print(f"Optimized {path.name}") except FileNotFoundError: print("Info: 'optimizt' not found. Skipping optimization step. Install it if you want optimization.") except subprocess.CalledProcessError as e: print(f"Error optimizing {path.name}: {e.stderr.decode()}")
def main(): import argparse
parser = argparse.ArgumentParser(description="Extract images from PDFs and convert to HEIC by default, then optimize.") parser.add_argument("directory", nargs="?", help="Directory to search for PDFs (default: cwd)") parser.add_argument("--no-heic", action="store_true", help="Disable HEIC conversion (enabled by default)") parser.add_argument("--heic-quality", type=int, default=60, help="HEIC quality (0..100, default: 60)") parser.add_argument("--keep-originals", action="store_true", help="Keep original extracted raster images alongside HEIC outputs (default is to replace originals)") parser.add_argument("--optimize-originals", action="store_true", help="When HEIC is enabled and originals are kept, also optimize the original non-HEIC images") parser.add_argument("--keep-pdfs", action="store_true", help="Keep source PDFs; default is to move them to Trash after processing") args = parser.parse_args()
# Determine directory to search search_dir = Path(args.directory) if args.directory else Path.cwd()
if not search_dir.is_dir(): print(f"Error: {search_dir} is not a valid directory") sys.exit(1)
# Find all PDFs in the directory (non-recursive) pdf_files = sorted(search_dir.glob("*.pdf"))
if not pdf_files: print(f"No PDF files found in {search_dir}") return
print(f"Found {len(pdf_files)} PDF file(s) in {search_dir}") print()
# Phase 1: Extract images from PDFs in parallel print("Phase 1: Extracting images...") output_paths = [] conversion_futures: list = [] if not args.no_heic: convert_executor_ctx = ThreadPoolExecutor(max_workers=None) else: convert_executor_ctx = None try: conversion_bar = tqdm(total=0, desc="Converting to HEIC", unit="img") if not args.no_heic else None with ThreadPoolExecutor(max_workers=None) as executor, tqdm(total=0, desc="Extracting images", unit="img") as image_extract_bar: futures = [] future_to_pdf = {} for pdf_path in pdf_files: f = executor.submit( extract_images_from_pdf, pdf_path, search_dir, heic_enabled=not args.no_heic, heic_quality=args.heic_quality, heic_overwrite=not args.keep_originals, convert_executor=convert_executor_ctx, conversion_futures=conversion_futures, image_extract_bar=image_extract_bar, conversion_bar=conversion_bar, ) futures.append(f) future_to_pdf[f] = pdf_path with tqdm(total=len(futures), desc="Extracting", unit="pdf") as pbar: for future in as_completed(futures): path = future.result() _ = future_to_pdf.get(future) if path is not None: output_paths.append(path) pbar.update(1) finally: if convert_executor_ctx is not None: # Do not shut down yet; conversions may still be running pass
if not output_paths: print("No images were extracted.") return
# Phase 2: HEIC conversion (enabled by default unless --no-heic) if not args.no_heic: heic_outputs: list[Path] = [] for future in as_completed(conversion_futures): out = future.result() if out: heic_outputs.append(out) if conversion_bar is not None: conversion_bar.close()
# If we overwrote originals (deleted them), keep HEIC outputs as the new set. # If not overwriting, optimize only HEICs by default unless --optimize-originals is set. if not args.keep_originals: output_paths = heic_outputs else: if args.optimize_originals: output_paths.extend(heic_outputs) else: output_paths = heic_outputs
# Phase 3: Optimize all extracted images in parallel (optional external tool) # Cleanup leftover temp files from extraction for p in output_paths: if p.is_dir(): for f in p.iterdir(): if f.is_file() and f.name.startswith(".__tmp__"): try: f.unlink(missing_ok=True) except Exception: pass else: if p.name.startswith(".__tmp__"): try: p.unlink(missing_ok=True) except Exception: pass # Denest single-image directories by grouping converted files by parent by_parent: dict[Path, list[Path]] = {} for f in output_paths: by_parent.setdefault(f.parent, []).append(f)
denested: list[Path] = [] for parent, files in by_parent.items(): if len(files) == 1: sole = files[0] dest = parent.parent / f"{parent.name}{sole.suffix}" try: sole.replace(dest) except Exception: dest.write_bytes(sole.read_bytes()) try: sole.unlink(missing_ok=True) except Exception: pass try: parent.rmdir() except Exception: pass denested.append(dest) else: denested.extend(files) output_paths = denested
print() # Phase 3: Optimizing images... print("Phase 3: Optimizing images...") with ThreadPoolExecutor(max_workers=None) as executor: futures = [executor.submit(optimize_images, path) for path in output_paths] for future in futures: future.result()
if not args.keep_pdfs: for pdf in pdf_files: try: send2trash(str(pdf)) except Exception as e: print(f"Warning: failed to move to Trash: {pdf.name}: {e}") print() print("Done!")
if __name__ == "__main__": main()