[Docutils-checkins] SF.net SVN: docutils:[8195] trunk/docutils

[<mi...@us...>]

Revision: 8195
          http://sourceforge.net/p/docutils/code/8195
Author:   milde
Date:     2017-11-04 10:06:45 +0000 (Sat, 04 Nov 2017)
Log Message:
-----------
New function `utils.unescape_rawsource`.

Removes backslashes and backslash-escaped whitespace from
`rawsource` arguments.
Required for fixes in definition list terms, author docinfo extraction,
and smartquote escaping.

Modified Paths:
--------------
    trunk/docutils/HISTORY.txt
    trunk/docutils/docutils/utils/__init__.py
    trunk/docutils/test/test_utils.py

Modified: trunk/docutils/HISTORY.txt
===================================================================
--- trunk/docutils/HISTORY.txt	2017-11-04 09:40:27 UTC (rev 8194)
+++ trunk/docutils/HISTORY.txt	2017-11-04 10:06:45 UTC (rev 8195)
@@ -69,6 +69,7 @@
 * docutils/utils/__init__.py:
 
   - Deprecate `unique_combinations` (obsoleted by `itertools.combinations`).
+  - New function `unescape_rawsource`.
 
 
 Release 0.14 (2017-08-03)

Modified: trunk/docutils/docutils/utils/__init__.py
===================================================================
--- trunk/docutils/docutils/utils/__init__.py	2017-11-04 09:40:27 UTC (rev 8194)
+++ trunk/docutils/docutils/utils/__init__.py	2017-11-04 10:06:45 UTC (rev 8195)
@@ -588,6 +588,13 @@
             text = ''.join(text.split(sep))
         return text
 
+def unescape_rawsource(text):
+    """Remove escape-backslashes and escaped whitespace."""
+    # remove escaped whitespace or backslash at end of text
+    text = re.sub(r'(?<!\\)\\([ \n]|$)', r'', text)
+    # remove backslash-escapes
+    return re.sub(r'\\(.)', r'\1', text)
+
 def split_escaped_whitespace(text):
     """
     Split `text` on escaped whitespace (null+space or null+newline).

Modified: trunk/docutils/test/test_utils.py
===================================================================
--- trunk/docutils/test/test_utils.py	2017-11-04 09:40:27 UTC (rev 8194)
+++ trunk/docutils/test/test_utils.py	2017-11-04 10:06:45 UTC (rev 8195)
@@ -321,6 +321,22 @@
         self.assertEqual(utils.find_file_in_dirs('gibts/nicht.txt', dirs),
                          'gibts/nicht.txt')
 
+    # samples for the (un)escaping tests:
+    raw = r'spa\m\ and \\ham' + '\\'
+    nulled = 'spa\x00m\x00 and \x00\ham\x00'
+    unescaped = 'spamand \\ham'
 
+    def test_escape2null(self):
+        self.assertEqual(utils.escape2null(self.raw), self.nulled)
+
+    def test_unescape(self):
+        self.assertEqual(utils.unescape(self.nulled), self.unescaped)
+        self.assertEqual(utils.unescape(self.nulled, restore_backslashes=True),
+                         self.raw)
+
+    def test_unescape_rawsource(self):
+        self.assertEqual(utils.unescape_rawsource(self.raw), self.unescaped)
+
+
 if __name__ == '__main__':
     unittest.main()

This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.