ag/pixelocr
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Use last known glyph for guessing bold/italic attributes of totally unknown words.

Browse source
This commit is contained in:
Andrey Golovizin 2014-09-15 23:32:24 +02:00
parent afcab12b0c
commit 4e4848e1bb
2 changed files with 9 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

4
pixelocr/document.py
View file

@ -73,8 +73,10 @@ class Document(object):
try: try:
glyph_data = self.glyphdb[glyph] glyph_data = self.glyphdb[glyph]
except KeyError: except KeyError:
guessed_bold = glyph.word.guess_bold(default=self.last_style.bold)
guessed_italic = glyph.word.guess_italic(default=self.last_style.italic)
self.switch_layout(glyph.word.guess_language()) self.switch_layout(glyph.word.guess_language())
text, bold, italic = self.ui.ask_for_help(glyph, glyph.word.guess_bold(), glyph.word.guess_italic()) text, bold, italic = self.ui.ask_for_help(glyph, guessed_bold, guessed_italic)
glyph_data = self.glyphdb.add_glyph(glyph, text, bold, italic) glyph_data = self.glyphdb.add_glyph(glyph, text, bold, italic)
self.last_style = glyph_data.style self.last_style = glyph_data.style
return glyph_data return glyph_data

12
pixelocr/page.py
View file

@ -280,16 +280,16 @@ class Word(PageObject):
language, count = sorted(counts.items(), reverse=True)[0] language, count = sorted(counts.items(), reverse=True)[0]
return language return language
def guess_bold(self): def guess_bold(self, *args, **kwargs):
return self._guess('bold') return self._guess('bold', *args, **kwargs)
def guess_italic(self): def guess_italic(self, *args, **kwargs):
return self._guess('italic') return self._guess('italic', *args, **kwargs)
def _guess(self, attr): def _guess(self, attr, default=None):
known_glyph_info = self._known_glyph_info() known_glyph_info = self._known_glyph_info()
if not known_glyph_info: if not known_glyph_info:
return None return default
total = sum(getattr(glyph_info.style, attr) for glyph_info in self._known_glyph_info()) total = sum(getattr(glyph_info.style, attr) for glyph_info in self._known_glyph_info())
avg = total / len(known_glyph_info) avg = total / len(known_glyph_info)
return avg >= 0.5 return avg >= 0.5