Webcomic cutter.py

import os, re
from PIL import Image


def moreBreaks(pic, low_limit, high_limit):
##    print('== More breaks ==')
##    print(pic, low_limit, high_limit)
    new_breaks = []
    pic = pic.convert('L')
    lines = {}
    for y in range(low_limit, high_limit+1):
        lines[y] = pic.crop((0,y,pic.size[0],y+1)).getdata()
    for y in range(low_limit, high_limit):
        if max(lines[y])-min(lines[y]) <= 5:
            new_breaks = new_breaks + [y]
        if sum([abs(lines[y][x]-lines[y+1][x]) for x in range(pic.size[0])])/pic.size[0]>96:
            new_breaks = new_breaks + [y]
    accepted_breaks = []
    if len(new_breaks)>0:
        y = min(new_breaks)
        while y < high_limit:
            d = 0
            while y+d in new_breaks: d += 1
            accepted_breaks = accepted_breaks + [y + d//2]
            y = min([high_limit]+[line for line in new_breaks if line > y+d])
    else:
        print('No new breaks found')
##    print('New breaks count:', len(new_breaks))
##    print('Accepted breaks count:', len(accepted_breaks))
##    print('Accepted breaks:', accepted_breaks)
##    import pdb; pdb.set_trace()
    return accepted_breaks


def cutPic(pic_path, rename=True):
    exceedingSizeWarning = '-ExceedingSize'
    prepaginizedWarning = '-Prepaginized'
    page_ratio = 1.6
    with Image.open(pic_path) as pic:
        print()
        print('File:', pic_path)
        column = list(pic.convert('L').resize(
            (1,pic.size[1]),
            Image.ANTIALIAS).getdata())
        max_height = page_ratio * pic.size[0]
        print('Picture size:', pic.size)
        if pic.size[1]<=max_height or (prepaginizedWarning in pic_path) or (exceedingSizeWarning in pic_path):
            print('Procession not needed.')
            return None
        print('Max page height:', max_height)
##        blacks = [i for i, l in enumerate(column) if l == 0]
##        whites = [i for i, l in enumerate(column) if l == 255]
        breaks = []
        i = 0
        while i<len(column):
            if column[i] in [0,255]:
                j = 0
                while i+j<len(column) and (column[i] == column[i+j]):
                    j += 1
                breaks = breaks + [i+j//2]
                i += j
            else:
                i += 1
        breaks = [0] + breaks + [len(column)]
        additional_breaks = []
        print('Breaks:',breaks)
        for i,b in enumerate(breaks[:-1]):
            if breaks[i+1]-breaks[i] > max_height*1.5:
                additional_breaks = additional_breaks + moreBreaks(pic,breaks[i],breaks[i+1])
        if len(breaks)==2:
            additional_breaks = additional_breaks + moreBreaks(pic,*breaks)
##        import pdb; pdb.set_trace()
        breaks = sorted(breaks + additional_breaks)
        print('Breaks:', breaks)
        page_breaks = [0]
        while page_breaks[-1]<len(column):
##            import pdb; pdb.set_trace()
            new_page_break = max([page_breaks[-1]]+[b for b in breaks if b < page_breaks[-1]+max_height])
            if new_page_break == page_breaks[-1]:
                new_page_break = min([b for b in breaks if b >= page_breaks[-1]+max_height])
            page_breaks = page_breaks + [new_page_break]
##        print('Page breaks:', page_breaks)
        for i in range(len(page_breaks)-1):
            page_pic = pic.crop([
                    0,
                    page_breaks[i],
                    pic.size[0],
                    page_breaks[i+1]
                ])
            if page_pic.size[1]>max_height and rename:
                exceeding_marker = exceedingSizeWarning
            else:
                exceeding_marker = ''
            page_path = re.sub(r'(.*)\.',r'\1-{:02d}{}.'.format(i, exceeding_marker),pic_path)
            print('Page path:', page_path)
            page_pic.save(page_path)
        if rename: os.rename(
                pic_path,
                re.sub(r'(.*)\.',r'\1{}.'.format(prepaginizedWarning),pic_path)
            )
##    import pdb; pdb.set_trace()


def cutPicFolder(folder_path):
    filelist = []
    for a,b,c in os.walk(folder_path):
        for file in c:
            if file.lower().endswith(('.png','.jpg','.jpeg')):
                filelist = filelist + [os.path.join(a, file)]
    filelist.sort()
    for file in filelist: cutPic(file)


if __name__=="__main__":
##    cutPic('/home/hellerick/Documents/Akiba/Manga/Ensemble (mangatraders.org)/Selected/0000.jpg', rename=False)
    cutPicFolder('/home/hellerick/Documents/Akiba/Manga/Ensemble (mangatraders.org)/')