Replace hardcoded values in the diacritic detection code with config variables.

pixelocr/config.py

@@ -20,6 +20,23 @@ from confire import Configuration as BaseConfiguration
 class Configuration(BaseConfiguration):
     min_body_height = 10
 
+    diacritic_box_left = -3
+    diacritic_box_right = +3
+    diacritic_box_top = -10
+    diacritic_box_bottom = +10
+
+    apostrophe_box_left = -5
+    apostrophe_box_right = +7
+    apostrophe_box_top = -5
+    apostrophe_box_bottom = +10
+    apostrophe_min_height = +3
+    apostrophe_max_distance = +4
+
+    dot_box_left = -3
+    dot_box_right = +3
+    dot_box_top = +1
+    dot_box_bottom = +10
+
     @classmethod
     def load_file(cls, filename):
         class MyConfiguration(cls):

pixelocr/page.py

@@ -246,8 +246,6 @@ class Line(PageObject):
 
 
 class Glyph(PageObject):
-    MIN_BODY_HEIGHT = 10
-
     def __init__(self, line, image, elevation):
         super().__init__(line.document, image)
         self.elevation = elevation
@@ -255,7 +253,7 @@ class Glyph(PageObject):
 
     def is_body(self):
         """Return True if the glyph is definitely not diacritic."""
-        return self.height >= self.MIN_BODY_HEIGHT
+        return self.height >= self.config.min_body_height
 
     def optical_distance(self, other):
         distance = other.left - self.right
@@ -267,32 +265,34 @@ class Glyph(PageObject):
         if not self.is_body():
             return False
 
-        #TODO remove hardcoded sizes
-
         # diacritic above the letter
         if glyph.fits(
-            self.left - 3,
-            self.top - 10,
-            self.right + 3,
-            self.top + 10,
+            self.left + self.config.diacritic_box_left,
+            self.top + self.config.diacritic_box_top,
+            self.right + self.config.diacritic_box_right,
+            self.top + self.config.diacritic_box_bottom,
         ):
             return True
 
         # apostrophe, like in ť
-        if glyph.fits(
-            self.right - 5,
-            self.top - 5,
-            self.right + 7,
-            self.top + 10,
-        ) and glyph.height > 3 and self.optical_distance(glyph) < 4:
+        if (
+            glyph.fits(
+                self.right + self.config.apostrophe_box_left,
+                self.top + self.config.apostrophe_box_top,
+                self.right + self.config.apostrophe_box_right,
+                self.top + self.config.apostrophe_box_bottom,
+            )
+            and glyph.height > self.config.apostrophe_min_height
+            and self.optical_distance(glyph) < self.config.apostrophe_max_distance
+        ):
             return True
 
         # dot in ? and !
         if glyph.fits(
-            self.left - 3,
-            self.bottom + 1,
-            self.right + 3,
-            self.bottom + 10,
+            self.left + self.config.dot_box_left,
+            self.bottom + self.config.dot_box_top,
+            self.right + self.config.dot_box_right,
+            self.bottom + self.config.dot_box_bottom,
         ):
             return True