remove no print glyphs

upstream-src/src/reportlab/lib/utils.py

@@ -1189,4 +1189,17 @@ def escapeOnce(data):
     data = data.replace(">", ">")
     data = data.replace("<", "<")
     return data
+
+import re
+
+# 'Other, Format' http://www.fileformat.info/info/unicode/category/Cf/list.htm
+noprint_chars = [unichr(int(c)) for c in '173 1536 1537 1538 1539 1757 1807 6068 6069 8203 8204 8205 8206 8207 8234 8235 8236 8237 8238 8288 8289 8290 8291 8292 8298 8299 8300 8301 8302 8303 65279 65529 65530 65531'.split()]
+
+repl_regex = re.compile('|'.join(noprint_chars))
+
+def remove_noprint(s):
+    if isinstance(s, str):
+        return repl_regex.sub('', s.decode('utf-8')).encode('utf-8')
+    else:
+        return repl_regex.sub('', s)
 

upstream-src/src/reportlab/pdfbase/pdfmetrics.py

@@ -22,7 +22,7 @@
 from types import StringType, ListType, TupleType
 from reportlab.pdfbase import _fontdata
 from reportlab.lib.logger import warnOnce
-from reportlab.lib.utils import rl_isfile, rl_glob, rl_isdir, open_and_read, open_and_readlines, findInPaths
+from reportlab.lib.utils import rl_isfile, rl_glob, rl_isdir, open_and_read, open_and_readlines, findInPaths, remove_noprint
 from reportlab import rl_config
 defaultEncoding = rl_config.defaultEncoding
 T1SearchPath = rl_config.T1SearchPath
@@ -732,9 +732,7 @@ def stringWidth(text, fontName, fontSize, encoding='utf8'):
     """Compute width of string in points;
     not accelerated as fast enough because of _instanceStringWidthU"""
     text = log2vis(text, DIR_RTL if rl_config.rtl else DIR_LTR)
-    rpl = unichr(65279)
-    rpl = rpl.encode('utf-8') if isinstance(text, str) else rpl
-    text = text.replace(rpl, '') #ignore zero width no-break space
+    text = remove_noprint(text)
     return getFont(fontName).stringWidth(text, fontSize, encoding=encoding)
 
 try:

upstream-src/src/reportlab/pdfgen/textobject.py

@@ -12,7 +12,7 @@
 import string
 from types import *
 from reportlab.lib.colors import Color, CMYKColor, CMYKColorSep, toColor, black, white, _CMYK_black, _CMYK_white
-from reportlab.lib.utils import fp_str
+from reportlab.lib.utils import fp_str, remove_noprint
 from reportlab.pdfbase import pdfmetrics
 
 # try to import pyfribidi
@@ -363,9 +363,7 @@ def _formatText(self, text):
         # Use pyfribidi to write the text in the correct visual order.
         directions = { 'LTR': DIR_LTR, 'RTL': DIR_RTL }
         text = log2vis(text, directions.get(self.direction, DIR_ON), reordernsm=False)
-        rpl = unichr(65279)
-        rpl = rpl.encode('utf-8') if isinstance(text, str) else rpl
-        text = text.replace(rpl, '') #ignore zero width no-break space
+        text = remove_noprint(text)
         canv = self._canvas
         font = pdfmetrics.getFont(self._fontname)
         R = []