use reasonable timeout for tesseract callsw

pagenum/image.py

@@ -1,10 +1,11 @@
 import pytesseract
 import subprocess
+import logging
 
 from .system import randomtempname
 from PIL import Image
 
-
+_logger = logging.getLogger(__name__)
 ZONES = ('topleft', 'topright',
          'bottomleft', 'bottomright',
          'topcenter', 'bottomcenter')
@@ -54,8 +55,7 @@ class Zone:
             self.margin_top,
             self.margin_right,
             self.margin_bottom,
-            self.margin_left
-            )
+            self.margin_left)
 
     @staticmethod
     def from_string(s):
@@ -92,7 +92,6 @@ def zonecrop(png_path, z):
     elif z.isright():
         x1 = iw-z.width
     elif z.iscenter():
-        # not tested
         x1 = int(iw/2-z.width/2)
 
     if z.istop():
@@ -116,9 +115,15 @@ def zonecrop(png_path, z):
 
 
 def img2pagenum(img_file, maxlen):
-    s = pytesseract.image_to_string(img_file,
-                                    lang='eng',
-                                    config='--psm 13 -c tessedit_char_whitelist=0123456789')
+    try:
+        s = pytesseract.image_to_string(img_file,
+                                        timeout=10,
+                                        lang='eng',
+                                        config='--psm 13 -c tessedit_char_whitelist=0123456789')
+    except RuntimeError as timeout_error:
+        _logger.exception(timeout_error)
+        return None
+
     for r in '_-.,—':
         s = s.replace(r, '')
     s = s.strip()