[Assorted-commits] SF.net SVN: assorted:[1017] python-commons/trunk/src/commons/strs.py

[<yan...@us...>]

Revision: 1017
          http://assorted.svn.sourceforge.net/assorted/?rev=1017&view=rev
Author:   yangzhang
Date:     2008-10-14 03:34:36 +0000 (Tue, 14 Oct 2008)

Log Message:
-----------
added remove_empty_lines, nat_lang_join, or_join, and_join, unit tests for nat_lang_join et al.

Modified Paths:
--------------
    python-commons/trunk/src/commons/strs.py

Modified: python-commons/trunk/src/commons/strs.py
===================================================================
--- python-commons/trunk/src/commons/strs.py	2008-10-14 03:32:58 UTC (rev 1016)
+++ python-commons/trunk/src/commons/strs.py	2008-10-14 03:34:36 UTC (rev 1017)
@@ -6,18 +6,22 @@
 """
 
 __all__ = '''
+and_join
+cp1252_to_unicode
+cp1252_to_unicode_translations
+dos2unix
 format
+indent
+nat_lang_join
+or_join
+remove_empty_lines
 safe_ascii
-cp1252_to_unicode_translations
-cp1252_to_unicode
-unwrap
-indent
 underline
-dos2unix
 unicode2html
+unwrap
 '''.split()
 
-import itertools, cgi, re
+import itertools, cgi, re, unittest
 
 def format( *args ):
     """Formats the args as they would be by the C{print} built-in."""
@@ -86,6 +90,27 @@
     if isinstance(s, str): s = s.split('\n')
     return '\n'.join( ind + line for line in s )
 
+def unindent(text, amt = None):
+    """
+    If L{amt} is 0, removes all leading whitespace from each line in L{text}.
+    If L{amt} is L{None}, finds the smallest amount of leading whitespace on
+    any non-empty line and removes that many chars from each line.  If L{amt}
+    is positive, removes L{amt} chars from each line.
+    """
+    lines = text.split('\n')
+    if amt == 0:
+        return '\n'.join( line.lstrip() for line in lines )
+    def count_indent(line):
+        for i,c in enumerate(line):
+            if not c.isspace(): return i
+    if amt is None:
+        amt = min( count_indent(line) for line in lines if line.strip() != '' )
+    return '\n'.join( line[amt:] for line in lines )
+
+def remove_empty_lines(s):
+    "Removes all empty lines (or lines of just whitespace)."
+    return '\n'.join( line for line in s.split('\n') if line.strip() != '' )
+
 def underline(s, sep):
     """
     Appends to L{s} a newline and a number of repetitions of L{sep}; the number
@@ -101,3 +126,41 @@
 def unicode2html(s):
     "Extends cgi.escape() with escapes for all unicode characters."
     return pat.sub(lambda m: '&#%d;' % ord(m.group()), cgi.escape(s))
+
+def nat_lang_join(xs, last_glue, two_glue = None, glue = ', '):
+    """
+    Natural-language join.  Join a sequence of strings together into a
+    comma-separated list, but where the last pair is joined with the given
+    special glue.  (You may also override the non-last glue, which defaults to
+    a ', '.)
+
+    @param xs: The sequence of strings.  This must be a list-like sequence, not
+    a generated one.
+
+    @param last_glue: The string used to join the final pair of elements, when
+    there are more than two elements.
+
+    @param two_glue: The string used to join both elements in a 2-element
+    sequence.  Defaults to None, which means to use last_glue.
+
+    @param glue: The string used to join all the other elements.
+    """
+    if len(xs) == 0: return ''
+    elif len(xs) == 1: return xs[0]
+    elif len(xs) == 2: return xs[0] + two_glue + xs[1]
+    else: return glue.join(xs[:-1]) + last_glue + xs[-1]
+
+def or_join(xs): return nat_lang_join(xs, ', or ', ' or ')
+def and_join(xs): return nat_lang_join(xs, ', and ', ' and ')
+
+class str_test( unittest.TestCase ):
+    def test_nat_lang_join( self ):
+        self.assertEqual( nat_lang_join( 'alpha beta gamma'.split(), ' | ' ),
+                          'alpha, beta | gamma' )
+        self.assertEqual( and_join( 'alpha beta gamma'.split() ),
+                          'alpha, beta, and gamma' )
+        self.assertEqual( or_join( 'alpha beta'.split() ),
+                          'alpha or beta' )
+
+if __name__ == '__main__':
+    unittest.main()


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