From dd847b4e98f7a64ec2b7a26d7df849fcc1e03721 Mon Sep 17 00:00:00 2001
From: Andrey Golovizin <ag@sologoc.com>
Date: Mon, 25 Aug 2014 15:30:28 +0200
Subject: [PATCH] Tweak diacritic detection to include ? and ! characters.

---
 pixelocr/page.py | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/pixelocr/page.py b/pixelocr/page.py
index fbf2cb5..6ad9482 100644
--- a/pixelocr/page.py
+++ b/pixelocr/page.py
@@ -183,6 +183,7 @@ class Glyph(PageObject):
     DIACRITIC_WINDOW_LEFT = 3
     DIACRITIC_WINDOW_RIGHT = 3
     DIACRITIC_MIN_ELEVATION = 5
+    MIN_BODY_HEIGHT = 10
 
     def __init__(self, image, elevation):
         super().__init__(image)
@@ -195,7 +196,7 @@ class Glyph(PageObject):
 
     def is_body(self):
         """Return True if the glyph is definitely not diacritic."""
-        return self.elevation <= 0
+        return self.height >= self.MIN_BODY_HEIGHT
 
     def detect_diacritic(self, glyph):
         """Check if the given glyph can be our diacritic and return a numeric score.
@@ -203,9 +204,9 @@ class Glyph(PageObject):
         Higher score means higher probability. Zero means "absolutely not".
         """
 
-        if glyph.elevation < self.DIACRITIC_MIN_ELEVATION:
+        if glyph.elevation > 0 and glyph.elevation < self.DIACRITIC_MIN_ELEVATION:
             return 0
-        if glyph.top >= self.top:
+        if glyph.top >= self.top and glyph.top < self.bottom:
             return 0
         if (
             glyph.left < self.left - self.DIACRITIC_WINDOW_LEFT