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

Use last the language of the last known glyph for totally unknown words.

Browse source
This commit is contained in:
Andrey Golovizin 2014-09-15 23:37:06 +02:00
parent 4e4848e1bb
commit 807ab6291c
2 changed files with 5 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

4
pixelocr/document.py
View file

@ -39,6 +39,7 @@ class Document(object):
self.config = Configuration.load_file(path.join(self.dirname, 'config.yaml')) self.config = Configuration.load_file(path.join(self.dirname, 'config.yaml'))
self.output_format = output_format self.output_format = output_format
self.last_style = Style(bold=False, italic=False, color=(255, 255, 255)) # FIXME get rid of hardcoded value self.last_style = Style(bold=False, italic=False, color=(255, 255, 255)) # FIXME get rid of hardcoded value
self.last_language = None
def save_glyphdb(self): def save_glyphdb(self):
self.glyphdb.save() self.glyphdb.save()
@ -75,10 +76,11 @@ class Document(object):
except KeyError: except KeyError:
guessed_bold = glyph.word.guess_bold(default=self.last_style.bold) guessed_bold = glyph.word.guess_bold(default=self.last_style.bold)
guessed_italic = glyph.word.guess_italic(default=self.last_style.italic) 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(self.last_language))
text, bold, italic = self.ui.ask_for_help(glyph, guessed_bold, guessed_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
self.last_language = self.language_map.get(glyph_data.text)
return glyph_data return glyph_data
@cached_property @cached_property

4
pixelocr/page.py
View file

@ -269,12 +269,12 @@ class Word(PageObject):
image = line.image[:, beginning.left:end.right] image = line.image[:, beginning.left:end.right]
super().__init__(line.document, image) super().__init__(line.document, image)
def guess_language(self): def guess_language(self, default=None):
counts = defaultdict(int) counts = defaultdict(int)
language_map = self.document.language_map language_map = self.document.language_map
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
for glyph_info in known_glyph_info: for glyph_info in known_glyph_info:
counts[language_map.get(glyph_info.text)] += 1 counts[language_map.get(glyph_info.text)] += 1
language, count = sorted(counts.items(), reverse=True)[0] language, count = sorted(counts.items(), reverse=True)[0]