Added strip raw option to wordpress xml importer

tests/support.py

@@ -6,6 +6,11 @@ __all__ = [
 
 import os
 import subprocess
+import re
+import sys
+import cStringIO
+
+from functools import wraps
 from contextlib import contextmanager
 from tempfile import mkdtemp
 from shutil import rmtree
@@ -28,8 +33,87 @@ def temporary_folder():
             # do whatever you want
     """
     tempdir = mkdtemp()
-    yield tempdir
-    rmtree(tempdir)
+    try:
+        yield tempdir
+    finally:
+        rmtree(tempdir)
+
+
+def isplit(s, sep=None):
+    """
+        Behave like str.split but returns a generator instead of a list.
+
+        >>> list(isplit('\tUse the force\n')) == '\tUse the force\n'.split()
+        True
+        >>> list(isplit('\tUse the force\n')) == ['Use', 'the', 'force']
+        True
+        >>> list(isplit('\tUse the force\n', "e")) == '\tUse the force\n'.split("e")
+        True
+        >>> list(isplit('Use the force', "e")) == 'Use the force'.split("e")
+        True
+        >>> list(isplit('Use the force', "e")) == ['Us', ' th', ' forc', '']
+        True
+
+    """
+    sep, hardsep = r'\s+' if sep is None else re.escape(sep), sep is not None
+    exp, pos, l = re.compile(sep), 0, len(s)
+    while True:
+        m = exp.search(s, pos)
+        if not m:
+            if pos < l or hardsep:
+                #      ^ mimic "split()": ''.split() returns []
+                yield s[pos:]
+            break
+        start = m.start()
+        if pos < start or hardsep:
+            #           ^ mimic "split()": includes trailing empty string
+            yield s[pos:start]
+        pos = m.end()
+
+
+def mute(returns_output=False):
+    """
+        Decorate a function that prints to stdout, intercepting the output.
+        If "returns_output" is True, the function will return a generator
+        yielding the printed lines instead of the return values.
+
+        The decorator litterally hijack sys.stdout during each function
+        execution, so be careful with what you apply it to.
+
+        >>> def numbers():
+            print "42"
+            print "1984"
+        ...
+        >>> numbers()
+        42
+        1984
+        >>> mute()(numbers)()
+        >>> list(mute(True)(numbers)())
+        ['42', '1984']
+
+    """
+
+    def decorator(func):
+
+        @wraps(func)
+        def wrapper(*args, **kwargs):
+
+            saved_stdout = sys.stdout
+            sys.stdout = cStringIO.StringIO()
+
+            try:
+                out = func(*args, **kwargs)
+                if returns_output:
+                    out = isplit(sys.stdout.getvalue().strip())
+            finally:
+                sys.stdout = saved_stdout
+
+            return out
+
+        return wrapper
+
+    return decorator
+
 
 
 def get_article(title, slug, content, lang, extra_metadata=None):