You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
bazarr/libs/subzero/modification/mods/ocr_fixes.py

56 lines
2.4 KiB

# coding=utf-8
import logging
import re
from subzero.modification.mods import SubtitleTextModification
from subzero.modification.processors.string_processor import MultipleLineProcessor, WholeLineProcessor
from subzero.modification.processors.re_processor import MultipleWordReProcessor, NReProcessor
from subzero.modification import registry
from subzero.modification.dictionaries.data import data as OCR_fix_data
logger = logging.getLogger(__name__)
class FixOCR(SubtitleTextModification):
identifier = "OCR_fixes"
description = "Fix common OCR issues"
exclusive = True
order = 10
data_dict = None
long_description = "Fix issues that happen when a subtitle gets converted from bitmap to text through OCR"
def __init__(self, parent):
super(FixOCR, self).__init__(parent)
data_dict = OCR_fix_data.get(parent.language.alpha3t)
if not data_dict:
logger.debug("No SnR-data available for language %s", parent.language)
return
self.data_dict = data_dict
self.processors = self.get_processors()
def get_processors(self):
if not self.data_dict:
return []
return [
# remove broken HI tag colons (ANNOUNCER'., ". instead of :) after at least 3 uppercase chars
# don't modify stuff inside quotes
NReProcessor(re.compile(ur'(?u)(^[^"\'’ʼ❜‘‛”“‟„]*(?<=[A-ZÀ-Ž]{3})[A-ZÀ-Ž-_\s0-9]+)'
ur'(["\'’ʼ❜‘‛”“‟„]*[.,‚،⹁、;]+)(\s*)(?!["\'’ʼ❜‘‛”“‟„])'),
r"\1:\3", name="OCR_fix_HI_colons", supported=lambda p: not p.only_uppercase),
# fix F'bla
NReProcessor(re.compile(ur'(?u)(\bF)(\')([A-zÀ-ž]*\b)'), r"\1\3", name="OCR_fix_F"),
WholeLineProcessor(self.data_dict["WholeLines"], name="OCR_replace_line"),
MultipleWordReProcessor(self.data_dict["WholeWords"], name="OCR_replace_word"),
MultipleWordReProcessor(self.data_dict["BeginLines"], name="OCR_replace_beginline"),
MultipleWordReProcessor(self.data_dict["EndLines"], name="OCR_replace_endline"),
MultipleWordReProcessor(self.data_dict["PartialLines"], name="OCR_replace_partialline"),
MultipleLineProcessor(self.data_dict["PartialWordsAlways"], name="OCR_replace_partialwordsalways")
]
registry.register(FixOCR)