def pdf_to_jpg(pdf_path):
with open(pdf_path, 'rb') as fp:
# 画像を格納するディレクトリを作成
out_dir = utils.make_outdir(pdf_to_jpg_dir, basename(pdf_path))
# 既に実行結果のようなものがある場合はスキップ
if len(os.listdir(out_dir)) > 10:
print('dir exists: skip')
return
with utils.timer('extract: ' + pdf_path):
parser, document = prev_pdf_parser(fp)
# Check if the document allows text extraction. If not, abort.
if not document.is_extractable:
raise PDFException
pages = PDFPage.create_pages(document)
# Process each page contained in the document.
for idx, page in enumerate(pages):
page_num = str(idx).zfill(4)
layouts = prev_pdf_layout(page)
for layout in layouts:
for thing in layout:
if isinstance(thing, LTImage):
# save_image(thing, out_dir, 'image-' + page_num + '.png')
save_image(thing, out_dir,
'image-' + page_num + '.jpg')
def pdf_to_png(root_path):
for dirpath, _, filenames in os.walk(root_path):
for filename in filenames:
if not filename.endswith('.pdf'):
continue
# if 'kurumi' not in filename:
# continue
# 画像を格納するディレクトリを作成
output_path = utils.make_outdir(root_path, filename[:-4])
# 既に実行結果のようなものがある場合はスキップ
if len(os.listdir(output_path)) > 3:
continue
pdf_path = os.path.join(dirpath, filename)
png_path = os.path.join(output_path, 'image-%03d.jpg')
# png_path = os.path.join(output_path, 'image-%03d.png')
argv = ['convert',
'-define', 'jpeg:size={0}x{0}'.format(config.MAX_HEIGHT), # 大きい画像を変換するときに速くなるかも
# '-deskew', '40%', # 傾きを補正
'-density', str(config.DENSITY_LOW), # 解像度を指定。未指定は72dpiになってしまう
# '-geometry', '{0}x{0}'.format(MAX_HEIGHT), # 最大の画像大きさを指定。縦横比は守られる
pdf_path, png_path]
print('exec_code: {}'.format(argv))
with utils.timer('convert: {} -> {}'.format(pdf_path, png_path)):
# print('convert: {} -> {}'.format(pdf_path, png_path))
# コマンドを実行
if subprocess.call(argv) != 0:
print('failed: {}'.format(pdf_path))
def devide_jpgs(jpg_dir, norm_img_size=True):
'''カラーならc.jpg、グレーでよければグレスケ化してg.jpgで保存'''
out_dir = utils.make_outdir(jpg_dir, NEW_JPG_DIR)
for jpg_path in utils.get_path_list(jpg_dir, 'jpg'):
img = cv2.imread(jpg_path)
if img is None:
continue
# norm_img_sizeが指定されていればリサイズ
if norm_img_size:
img = utils.img_resize(img, max_height=config.MAX_HEIGHT)
if is_color_img(img):
cv2.imwrite(
join(out_dir,
basename(jpg_path).replace('.jpg', 'c.jpg')), img)
else:
img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
cv2.imwrite(
join(out_dir,
basename(jpg_path).replace('.jpg', 'g.jpg')), img)
box = [(v.x, v.y) for v in text.bounding_poly.vertices]
draw.line(box + [box[0]], width=2, fill=color)
return img_p
pickle_path = Path(utils.check_argv_path(sys.argv))
img_dir = pickle_path / '..' / pickle_path.name
path_serif_list = []
ta_list = utils.pickle_load(pickle_path)['values']
for idx, val in enumerate(ta_list[:]):
img_path = val['image_path']
tas_orig = val['text_annotation']
bal = Balloon(img_path, tas_orig)
draw = bal.highlight_texts()
serif = bal.make_serif()
path_serif_list.append([img_path, serif])
csv_dir = utils.make_outdir((pickle_path / '..' / '..').resolve(), 'csv')
with open(Path(csv_dir) / f'{pickle_path.name}.csv', 'w') as csv_file:
fieldnames = ['img_path', 'serif']
writer = csv.DictWriter(csv_file, fieldnames=fieldnames)
writer.writeheader()
[
writer.writerow({
fieldnames[0]: p,
fieldnames[1]: s
}) for p, s in path_serif_list
]
type=int,
help=('set padding size(px) for y'))
parser.add_argument('--ext', default='jpg', help=('target ext'))
args = parser.parse_args()
print(args)
image_dir = utils.check_argv_path(args)
padding_x = args.pad_x
padding_y = args.pad_y
good_cons_num_y = 4
if args.with_subtitle or args.shave_subtitle:
good_cons_num_y += 1
# 出力ディレクトリ・パスを準備
outdir_name = '2_paint_out'
output_path = utils.make_outdir(image_dir, outdir_name)
output_koma_path = utils.make_outdir(output_path, '0_koma')
if len(os.listdir(output_koma_path)) >= 3:
shutil.rmtree(output_path)
output_path = utils.make_outdir(image_dir, outdir_name)
output_koma_path = utils.make_outdir(output_path, '0_koma')
output_shaved_path = utils.make_outdir(output_koma_path, '0_padding_shave')
# paint_out処理: 1st
img_path_list = utils.get_path_list(image_dir, args.ext)
print('pages:', len(img_path_list) - (args.start + args.end))
with utils.timer('paint_out処理: 1st 切り抜き位置が求められた画像を切り抜き'):
odd_cp_list = [] # 奇数idxページのカットポイントを格納
even_cp_list = [] # 偶数idxページのカットポイントを格納
not_cut_img_path_dict = {}
# 1ページの画像から各コマを切り出す処理
import os
import sys
import cv2
from cut import AverageDiffCut
import utils
if __name__ == '__main__':
image_dir = utils.check_argv_path(sys.argv)
outdir_name = 'koma'
output_path = utils.make_outdir(image_dir, outdir_name)
img_path_list = utils.get_path_list(image_dir, 'jpg')
adc = AverageDiffCut()
# 切り出し座標＝カットポイント(cp)を探すためのループ
print('切り出し座標を検出しています')
odd_cp_list = [] # 奇数idxページのカットポイントを格納
even_cp_list = [] # 偶数idxページのカットポイントを格納
not_cut_img_path_dict = {}
for idx, img_path in enumerate(img_path_list):
img = cv2.imread(img_path)
img_gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# 切り出し座標の探索
cp_dict = adc.search_cut_point(img_gray)
print(idx, cp_dict, len(cp_dict['x']), len(cp_dict['y']))
# 切り出し座標が規定の数だけ帰ってきたらリストに追加
def pdf_to_page1(pdf_path):
with open(pdf_path, 'rb') as fp:
parser, document = prev_pdf_parser(fp)
# Check if the document allows text extraction. If not, abort.
if not document.is_extractable:
raise PDFException
# 最初のページを取得
page1 = PDFPage.create_pages(document).__next__()
layouts = prev_pdf_layout(page1)
for layout in layouts:
for thing in layout:
save_image(thing,
dirname(pdf_path),
TMP_FILENAME_PAGE1,
max_width=MAX_WIDTH_JPG1)
return join(dirname(pdf_path), TMP_FILENAME_PAGE1)
if __name__ == '__main__':
pdf_dir = utils.check_argv_path(sys.argv)
pdf_path_list = utils.get_path_list(pdf_dir, 'pdf')
pdf_path_list = [path for path in pdf_path_list if not os.path.isdir(path)]
pdf_to_jpg_dir = utils.make_outdir(pdf_dir, PDF_TO_JPG_DIR)
print('pdf files:', len(pdf_path_list))
for pdf_path in pdf_path_list:
pdf_to_jpg(pdf_path)
box = [(v.get('x', 0.0), v.get('y', 0.0))
for v in text['boundingPoly']['vertices']]
draw.line(box + [box[0]], width=2, fill=color)
return img
if __name__ == '__main__':
image_dir = Path(utils.check_argv_path(sys.argv))
key_file_path = '/Users/esuji/Dropbox/program/cvtest/yonkoma2data/src/MyFirstProject-b8427fd28b8d.json'
os.environ['GOOGLE_APPLICATION_CREDENTIALS'] = key_file_path
# Instantiates a client
client = vision.ImageAnnotatorClient()
pickle_dir = utils.make_outdir(image_dir / '..', 'pickles')
image_path_list = sorted(utils.get_path_list(image_dir, '.jpg'))
ta_list = []
for idx_koma, image_path in enumerate(image_path_list[100:110]):
print(idx_koma, image_path)
try:
image = get_image(image_path)
ta = get_response_text_annotations(image)
ta_list.append({
'image_path': image_path,
'text_annotation': list(ta)
})
print(ta[0].description)
except Exception as e:
import sys
import shutil
from collections import defaultdict
import utils
if __name__ == '__main__':
pdf_dir = utils.check_argv_path(sys.argv)
pdf_path_list = utils.get_path_list(pdf_dir, 'pdf')
# 作者名：pdfパスの辞書を作る
author_dict = defaultdict(list)
for pdf_path in pdf_path_list:
if len(pdf_path.split('[')) != 3:
continue
author = '[{}]'.format(pdf_path.split('[')[2].split(']')[0]).replace(' ', '')
author_dict[author].append(pdf_path)
# 作者名毎にmkdirしてpdfファイルを移動
for author, writing_path_list in author_dict.items():
author_dir = utils.make_outdir(pdf_dir, author)
for writing_path in writing_path_list:
shutil.move(writing_path, author_dir)
def jpg_to_pdf(jpg_dir_path, pdf_path):
# convert -quality 20 *.jpg 元pdf_clean.pdf
argv = [
'convert',
'-quality',
str(config.PDF_QUALITY), # pdfの圧縮率
join(jpg_dir_path, '*.jpg'), # すべてのjpgを対象
pdf_path.replace('.pdf', '{}.pdf'.format(CLEAN_PDF_SUFFIX)) # 出力するpdf
]
run_process(argv)
if __name__ == '__main__':
pdf_dir = utils.check_argv_path(sys.argv)
pdf_path_list = utils.get_path_list(pdf_dir, '.pdf')
clean_pdf_dir = utils.make_outdir(pdf_dir, CLEAN_PDF_DIR)
pdf_path_list = [path for path in pdf_path_list if '[' not in path]
pdf_path_list = [path for path in pdf_path_list if '31' in path]
print('pdf files:', len(pdf_path_list))
for pdf_path in pdf_path_list:
jpg_dir_path = join(os.path.dirname(pdf_path), PDF_TO_JPG_DIR,
basename(pdf_path))
if not os.path.isdir(jpg_dir_path):
continue
with utils.timer('jpgファイルをグレスケとカラーで分別'):
devide_jpgs(jpg_dir_path, norm_img_size=False)