add option to display weird PDF pages as sixel graphics before deciding how to handle them

This commit is contained in:
xenofem 2024-03-12 04:35:09 -04:00
parent 3a9199b847
commit ee9eba3287
3 changed files with 54 additions and 6 deletions

View file

@ -3,6 +3,7 @@
import argparse
import asyncio
import importlib_resources as resources
from io import BytesIO
from pathlib import Path
import os
from os.path import relpath, splitext
@ -17,6 +18,7 @@ import zipfile
from dlsite_async import DlsiteAPI
import fitz
from libsixel import *
from PIL import Image
from jinja2 import Environment, PackageLoader, select_autoescape
import requests
@ -77,7 +79,8 @@ IMAGE_FILE_EXTENSIONS = ['.png', '.jpg', '.jpeg', '.gif', '.tiff', '.bmp']
IGNOREABLE_FILES = ['Thumbs.db', '__MACOSX', '.DS_Store']
IGNOREABLE_EXTENSIONS = ['.txt', '.html', '.htm', '.psd', '.mp4']
PDF_FALLBACK_DPI = 300
PDF_CONVERSION_DPI = 300
PDF_PREVIEW_DPI = 72
IRRELEVANT_PDF_BLOCK_REGEX = re.compile(r'\bTCPDF\b', re.I)
@ -527,6 +530,45 @@ def extract_image(pdf, xref):
pix = fitz.Pixmap(pdf, xref)
return { 'ext': 'png', 'image': pix.tobytes('png') }
def display_sixel_page(page):
s = BytesIO()
image = Image.open(BytesIO(page.get_pixmap(dpi=PDF_PREVIEW_DPI).tobytes('png')))
width, height = image.size
try:
data = image.tobytes()
except NotImplementedError:
data = image.tostring()
output = sixel_output_new(lambda data, s: s.write(data), s)
try:
if image.mode == 'RGBA':
dither = sixel_dither_new(256)
sixel_dither_initialize(dither, data, width, height, SIXEL_PIXELFORMAT_RGBA8888)
elif image.mode == 'RGB':
dither = sixel_dither_new(256)
sixel_dither_initialize(dither, data, width, height, SIXEL_PIXELFORMAT_RGB888)
elif image.mode == 'P':
palette = image.getpalette()
dither = sixel_dither_new(256)
sixel_dither_set_palette(dither, palette)
sixel_dither_set_pixelformat(dither, SIXEL_PIXELFORMAT_PAL8)
elif image.mode == 'L':
dither = sixel_dither_get(SIXEL_BUILTIN_G8)
sixel_dither_set_pixelformat(dither, SIXEL_PIXELFORMAT_G8)
elif image.mode == '1':
dither = sixel_dither_get(SIXEL_BUILTIN_G1)
sixel_dither_set_pixelformat(dither, SIXEL_PIXELFORMAT_G1)
else:
raise RuntimeError('unexpected image mode')
try:
sixel_encode(data, width, height, 1, dither, output)
print(s.getvalue().decode('ascii'))
finally:
sixel_dither_unref(dither)
finally:
sixel_output_unref(output)
def pdf_images(pdf, force=False):
images_by_page = [(page.get_images(), is_single_image(page)) for page in pdf]
if all(len(images) == 1 and single for (images, single) in images_by_page):
@ -549,14 +591,18 @@ def pdf_images(pdf, force=False):
print(f'Converting page {idx+1}')
choice = 'c'
else:
if xref is None:
choice = input('[N]ope out of this PDF or [c]onvert the page lossily to an image? [N/c] ')
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] ')
shown = False
while True:
choice = input(f'[N]ope out of this PDF / [c]onvert page to image{"" if xref is None else " / e[x]tract embedded image without text"}{"" if shown else " / [s]how page before deciding"}? [N/c{"" if xref is None else "/x"}{"" if shown else "/s"}] ')
if not shown and choice != '' and choice[0].lower() == 's':
display_sixel_page(page)
shown = True
else:
break
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') })
image_extractors.append(lambda p=page: { 'ext': 'png', 'image': p.get_pixmap(dpi=PDF_CONVERSION_DPI).tobytes('png') })
else:
return None
print(f'\x1b[2K\r{idx+1}/{pdf.page_count} pages processed...', end=('' if idx+1 < pdf.page_count else '\n'))

View file

@ -57,6 +57,7 @@
jinja2
importlib-resources
setuptools
libsixel
];
src = ./.;
};

View file

@ -11,6 +11,7 @@ dependencies = [
"dlsite-async",
"jinja2",
"importlib_resources",
"libsixel",
]
[project.scripts]