support pi.get() and pi.attrib by parsing the text context for pseudo-attributes

3 files changed, 57 insertions, 27 deletions

diff --git a/src/lxml/lxml.etree.pyx b/src/lxml/lxml.etree.pyx
index de97facf..8f5a1a51 100644
--- a/src/lxml/lxml.etree.pyx
+++ b/src/lxml/lxml.etree.pyx
@@ -1579,6 +1579,32 @@ cdef class _ProcessingInstruction(__ContentOnlyElement):
         else:
             return u"<?%s?>" % self.target
 
+    def get(self, key, default=None):
+        u"""get(self, key, default=None)
+
+        Try to parse pseudo-attributes from the text content of the
+        processing instruction, search for one with the given key as
+        name and return its associated value.
+
+        Note that this is only a convenience method for the most
+        common case that all text content is structured in
+        attribute-like name-value pairs with properly quoted values.
+        It is not guaranteed to work for all possible text content.
+        """
+        return self.attrib.get(key, default)
+
+    property attrib:
+        u"""Returns a dict containing all pseudo-attributes that can be
+        parsed from the text content of this processing instruction.
+        Note that modifying the dict currently has no effect on the
+        XML node, although this is not guaranteed to stay this way.
+        """
+        def __get__(self):
+            return { attr : (value1 or value2)
+                     for attr, value1, value2 in _FIND_PI_ATTRIBUTES(u' ' + self.text) }
+
+cdef object _FIND_PI_ATTRIBUTES = re.compile(ur'\s+(\w+)\s*=\s*(?:\'([^\']*)\'|"([^"]*)")', re.U).findall
+
 cdef class _Entity(__ContentOnlyElement):
     property tag:
         def __get__(self):
diff --git a/src/lxml/tests/test_etree.py b/src/lxml/tests/test_etree.py
index 9a7dd63c..093c0f11 100644
--- a/src/lxml/tests/test_etree.py
+++ b/src/lxml/tests/test_etree.py
@@ -449,6 +449,28 @@ class ETreeOnlyTestCase(HelperTestCase):
         self.assertEquals(root[0].target, "mypi")
         self.assertEquals(root[0].text, "my test ")
 
+    def test_pi_pseudo_attributes_get(self):
+        XML = self.etree.XML
+        root = XML(_bytes("<test><?mypi my='1' test=\" abc \" quotes=\"' '\" only names ?></test>"))
+        self.assertEquals(root[0].target, "mypi")
+        self.assertEquals(root[0].get('my'), "1")
+        self.assertEquals(root[0].get('test'), " abc ")
+        self.assertEquals(root[0].get('quotes'), "' '")
+        self.assertEquals(root[0].get('only'), None)
+        self.assertEquals(root[0].get('names'), None)
+        self.assertEquals(root[0].get('nope'), None)
+
+    def test_pi_pseudo_attributes_attrib(self):
+        XML = self.etree.XML
+        root = XML(_bytes("<test><?mypi my='1' test=\" abc \" quotes=\"' '\" only names ?></test>"))
+        self.assertEquals(root[0].target, "mypi")
+        self.assertEquals(root[0].attrib['my'], "1")
+        self.assertEquals(root[0].attrib['test'], " abc ")
+        self.assertEquals(root[0].attrib['quotes'], "' '")
+        self.assertRaises(KeyError, root[0].attrib.__getitem__, 'only')
+        self.assertRaises(KeyError, root[0].attrib.__getitem__, 'names')
+        self.assertRaises(KeyError, root[0].attrib.__getitem__, 'nope')
+
     def test_deepcopy_pi(self):
         # previously caused a crash
         ProcessingInstruction = self.etree.ProcessingInstruction
diff --git a/src/lxml/xslt.pxi b/src/lxml/xslt.pxi
index 6cb2b349..1773c552 100644
--- a/src/lxml/xslt.pxi
+++ b/src/lxml/xslt.pxi
@@ -801,20 +801,10 @@ xslt.exsltRegisterAll()
 ################################################################################
 # XSLT PI support
 
-cdef object _FIND_PI_ATTRIBUTES
-_FIND_PI_ATTRIBUTES = re.compile(ur'\s+(\w+)\s*=\s*["\']([^"\']+)["\']', re.U).findall
-
-cdef object _RE_PI_HREF
-_RE_PI_HREF = re.compile(ur'\s+href\s*=\s*["\']([^"\']+)["\']')
-
-cdef object _FIND_PI_HREF
-_FIND_PI_HREF = _RE_PI_HREF.findall
-
-cdef object _REPLACE_PI_HREF
-_REPLACE_PI_HREF = _RE_PI_HREF.sub
-
-cdef XPath __findStylesheetByID
-__findStylesheetByID = None
+cdef object _RE_PI_HREF = re.compile(ur'\s+href\s*=\s*(?:\'([^\']*)\'|"([^"]*)")')
+cdef object _FIND_PI_HREF = _RE_PI_HREF.findall
+cdef object _REPLACE_PI_HREF = _RE_PI_HREF.sub
+cdef XPath __findStylesheetByID = None
 
 cdef _findStylesheetByID(_Document doc, id):
     global __findStylesheetByID
@@ -843,10 +833,11 @@ cdef class _XSLTProcessingInstruction(PIBase):
         _assertValidNode(self)
         if self._c_node.content is NULL:
             raise ValueError, u"PI lacks content"
-        hrefs = _FIND_PI_HREF(u' ' + funicode(self._c_node.content))
+        hrefs = _FIND_PI_HREF(u' ' + self._c_node.content.decode('UTF-8'))
         if len(hrefs) != 1:
             raise ValueError, u"malformed PI attributes"
-        href_utf = utf8(hrefs[0])
+        hrefs = hrefs[0]
+        href_utf = utf8(hrefs[0] or hrefs[1])
         c_href = _cstr(href_utf)
 
         if c_href[0] != c'#':
@@ -881,7 +872,8 @@ cdef class _XSLTProcessingInstruction(PIBase):
     def set(self, key, value):
         u"""set(self, key, value)
 
-        Sets a pseudo attribute in the text of the processing instruction.
+        Supports setting the 'href' pseudo-attribute in the text of
+        the processing instruction.
         """
         if key != u"href":
             raise AttributeError, \
@@ -897,13 +889,3 @@ cdef class _XSLTProcessingInstruction(PIBase):
             self.text = _REPLACE_PI_HREF(attrib, text)
         else:
             self.text = text + attrib
-
-    def get(self, key, default=None):
-        u"""get(self, key, default=None)
-
-        Parses a pseudo attribute from the text of the processing instruction.
-        """
-        for attr, value in _FIND_PI_ATTRIBUTES(u' ' + self.text):
-            if attr == key:
-                return value
-        return default