re-pcre: findall(): Rework for compliant implementation.

pfalcon · pfalcon · commit 3ec82997d58d · 2015-05-29T02:04:27.000+03:00
diff --git a/re-pcre/re.py b/re-pcre/re.py
@@ -124,8 +124,9 @@ def split(self, s, maxsplit=0):
 
     def findall(self, s):
         res = []
+        start = 0
         while True:
-            m = self.search(s)
+            m = self.search(s, start)
             if not m:
                 return res
             if m.num == 1:
@@ -135,7 +136,7 @@ def findall(self, s):
             else:
                 res.append(m.groups())
             beg, end = m.span(0)
-            s = s[end:]
+            start = end
 
 
 def compile(pattern, flags=0):
diff --git a/re-pcre/test_re.py b/re-pcre/test_re.py
@@ -42,3 +42,12 @@
 
 text = "He was carefully disguised but captured quickly by police."
 assert re.findall(r"(\w+)ly", text) == ['careful', 'quick']
+
+_leading_whitespace_re = re.compile('(^[ \t]*)(?:[^ \t\n])', re.MULTILINE)
+text = "\tfoo\n\tbar"
+indents = _leading_whitespace_re.findall(text)
+assert indents == ['\t', '\t']
+
+text = "  \thello there\n  \t  how are you?"
+indents = _leading_whitespace_re.findall(text)
+assert indents == ['  \t', '  \t  ']
