Use entry points for output formats, add --output-format option.

pixelocr/gui/__init__.py

@@ -19,6 +19,8 @@ import signal
 from threading import Thread
 from argparse import ArgumentParser
 
+from pkg_resources import iter_entry_points
+
 import sip
 sip.setapi('QString', 2)
 
@@ -38,9 +40,17 @@ parser = ArgumentParser(description='PixelOCR')
 parser.add_argument('--skip', metavar='N', type=int, default=0)
 parser.add_argument('--limit', metavar='N', type=int, default=None)
 parser.add_argument('--quit', action='store_true')
+parser.add_argument('-f', '--output-format', type=str, default='text')
 parser.add_argument('filename')
 
 
+def load_entry_point(group, name):
+    try:
+        return next(iter_entry_points(group, name))
+    except StopIteration:
+        raise ValueError('Entry point {} in group {} not found'.format(name, group))
+
+
 def main():
     app = QApplication(sys.argv)
 
@@ -48,7 +58,13 @@ def main():
     QApplication.setApplicationName("PixelOCR");
 
     args = parser.parse_args()
-    ocr = OCREngine(args.filename, skip=args.skip, limit=args.limit, quit=args.quit)
+    ocr = OCREngine(
+        args.filename,
+        skip=args.skip,
+        limit=args.limit,
+        quit=args.quit,
+        output_format=load_entry_point('pixelocr.formatting', args.output_format).load()(),
+    )
     app.aboutToQuit.connect(ocr.save_glyphdb)
 
     win = MainWindow(ocr)

pixelocr/gui/ocrengine.py

@@ -39,13 +39,14 @@ class OCREngine(QThread):
     unknownGlyph = signal([Glyph])
     pageChanged = signal([Page])
 
-    def __init__(self, dirname, skip=0, limit=None, quit=False):
+    def __init__(self, dirname, skip=0, limit=None, quit=False, output_format='text'):
         super().__init__()
         self.dirname = dirname
         self.filenames = sorted(glob(path.join(dirname, '*.png')))[skip:skip + limit if limit else None]
         self.glyphdb = GlyphDB(path.join(self.dirname, 'glyphdb.pickle'))
         self.help_queue = Queue()
         self.quit = quit
+        self.output_format = output_format
 
     def save_glyphdb(self):
         self.glyphdb.save()
@@ -67,8 +68,7 @@ class OCREngine(QThread):
 
     def recognize_page(self, page):
         glyph_data_seq = itertools.chain(*(self.recognize_line(line) for line in page.lines))
-        output_format = formatting.HTMLFormat()
+        return ''.join(self.output_format.format(glyph_data_seq))
-        return ''.join(output_format.format(glyph_data_seq))
 
     def recognize_line(self, line):
         yield from [SPACE] * int(line.indent / self.SPACE_WIDTH)

setup.py

@@ -18,7 +18,11 @@ setup(
         'Operating System :: OS Independent',
     ],
     entry_points={
-        'console_scripts': ['pixelocr = pixelocr.gui:main']
+        'console_scripts': ['pixelocr = pixelocr.gui:main'],
+        'pixelocr.formatting': [
+            'text = pixelocr.formatting:TextFormat',
+            'html = pixelocr.formatting:HTMLFormat',
+        ],
     },
     packages=find_packages(),
 )