diff --git a/pixelocr/config.py b/pixelocr/config.py index 0a61b74..0388195 100644 --- a/pixelocr/config.py +++ b/pixelocr/config.py @@ -22,6 +22,7 @@ class Configuration(BaseConfiguration): min_body_height = 10 min_word_distance = 15 + space_width = 15 diacritic_box_left = -3 diacritic_box_right = +3 diff --git a/pixelocr/document.py b/pixelocr/document.py index 0a445c9..c816ee8 100644 --- a/pixelocr/document.py +++ b/pixelocr/document.py @@ -28,8 +28,6 @@ from .utils import cached_property class Document(object): - SPACE_WIDTH = 15 - def __init__(self, dirname, ui, skip=0, limit=None, output_format='text'): super().__init__() self.dirname = dirname @@ -61,7 +59,7 @@ class Document(object): return ''.join(self.output_format.format(glyph_data_seq)) def recognize_line(self, line): - yield from [SPACE] * int(line.indent / self.SPACE_WIDTH) + yield from [SPACE] * int(line.indent / self.config.space_width) for glyph in line.glyphs: yield self.recognize_glyph(glyph) yield NEWLINE