xenofem/dlibrary
1
0
Fork 0
Code Issues 3 Pull requests Packages Projects Releases Wiki Activity

prompt for how to handle PDF weird pages, instead of immediately bailing out

Browse source
This commit is contained in:
xenofem 2024-03-12 03:23:57 -04:00
parent c7f95d50f9
commit 3a9199b847
1 changed files with 26 additions and 28 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

44
dlibrary/dlibrary.py
View file

@ -533,37 +533,35 @@ def pdf_images(pdf, force=False):
return (extract_image(pdf, images[0][0]) for (images, _) in images_by_page) return (extract_image(pdf, images[0][0]) for (images, _) in images_by_page)
print("Checking PDF images the quick way failed, trying the slow way") print("Checking PDF images the quick way failed, trying the slow way")
def xref_or_image_generator(): print(f'0/{pdf.page_count} pages processed...', end='')
xref_mode = not force image_extractors = []
for (idx, page) in enumerate(pdf): for (idx, page) in enumerate(pdf):
page_images = page.get_image_info(xrefs=True) page_images = page.get_image_info(xrefs=True)
if len(page_images) == 1 and page_images[0]['xref'] != 0 and is_single_image(page): if len(page_images) == 1 and page_images[0]['xref'] != 0:
xref = page_images[0]['xref'] xref = page_images[0]['xref']
if xref_mode:
yield xref
else: else:
yield extract_image(pdf, xref) xref = None
if xref is not None and is_single_image(page):
image_extractors.append(lambda p=pdf, x=xref: extract_image(p, x))
else: else:
print(f'\nPage {idx+1}: {len(page_images)} images, {len([img for img in page_images if img["xref"] == 0])} non-xref images, {len(relevant_blocks(page))} total relevant objects') print(f'\nPage {idx+1}: {len(page_images)} images, {len([img for img in page_images if img["xref"] == 0])} non-xref images, {len(relevant_blocks(page))} total relevant objects')
if xref_mode:
raise ValueError
else:
print(f'Generating pixmap for page {idx+1}')
pix = page.get_pixmap(dpi=PDF_FALLBACK_DPI)
yield { 'ext': 'png', 'image': pix.tobytes('png') }
print(f'\x1b[2K\r{idx+1}/{pdf.page_count} pages processed...', end='')
print('')
if force: if force:
return xref_or_image_generator() print(f'Converting page {idx+1}')
choice = 'c'
try: else:
xrefs = list(xref_or_image_generator()) if xref is None:
except ValueError: choice = input('[N]ope out of this PDF or [c]onvert the page lossily to an image? [N/c] ')
print('\nFailed') else:
choice = input('[N]ope out of this PDF, e[x]tract the image without the text, or [c]onvert the entire page lossily to an image? [N/x/c] ')
if xref is not None and choice != '' and choice[0].lower() == 'x':
image_extractors.append(lambda p=pdf, x=xref: extract_image(p, x))
elif choice != '' and choice[0].lower() == 'c':
image_extractors.append(lambda p=page: { 'ext': 'png', 'image': p.get_pixmap(dpi=PDF_FALLBACK_DPI).tobytes('png') })
else:
return None return None
print('Success') print(f'\x1b[2K\r{idx+1}/{pdf.page_count} pages processed...', end=('' if idx+1 < pdf.page_count else '\n'))
return (extract_image(pdf, xref) for xref in xrefs)
return (extractor() for extractor in image_extractors)
def nfc(s): def nfc(s):
return unicodedata.normalize('NFC', s) return unicodedata.normalize('NFC', s)