more lxml form fixes and test cases. #111

* Do not treat "coord" attribute specially, just pass "NN,NN" as clickdata value
* Raise explicit ValueError if not clickable is found
* Fix bug looking for clickeables trough xpath when there is more than one form
* Test from_response with multiple clickdata

scrapy/http/request/form.py

@@ -118,17 +118,14 @@ def _get_clickable(clickdata, clickables, form):
 
     # We didn't find it, so now we build an XPath expression out of the other
     # arguments, because they can be used as such
-    xpath_pred = []
-    for k, v in clickdata.items():
-        if k == 'coord':
-            v = ','.join(str(c) for c in v)
-        xpath_pred.append('[@%s="%s"]' % (k, v))
-
-    xpath_expr = '//*%s' % ''.join(xpath_pred)
-    el = form.xpath(xpath_expr)
-    if len(el) > 1:
+    xpath = u'.//*' + \
+            u''.join(u'[@%s="%s"]' % tuple(c) for c in clickdata.iteritems())
+    el = form.xpath(xpath)
+    if len(el) == 1:
+        return (el[0].name, el[0].value)
+    elif len(el) > 1:
         raise MultipleElementsFound("Multiple elements found (%r) "
                                     "matching the criteria in clickdata: %r"
                                     % (el, clickdata))
     else:
-        return (el[0].name, el[0].value)
+        raise ValueError('No clickeable element matching clickdata')

scrapy/tests/test_http_request.py

@@ -306,6 +306,59 @@ class FormRequestTest(RequestTest):
         self.assertEqual(urlargs['one'], ['1'])
         self.assertEqual(urlargs['two'], ['2'])
 
+    def test_from_response_multiple_clickdata(self):
+        respbody = """
+<form action="get.php" method="GET">
+<input type="submit" name="clickeable" value="clicked1">
+<input type="submit" name="clickeable" value="clicked2">
+<input type="hidden" name="one" value="clicked1">
+<input type="hidden" name="two" value="clicked2">
+</form>
+        """
+        response = HtmlResponse("http://www.example.com/this/list.html", body=respbody)
+        r1 = self.request_class.from_response(response, \
+                clickdata={'name': 'clickeable', 'value': 'clicked2'})
+        urlargs = cgi.parse_qs(urlparse(r1.url).query)
+        self.assertEqual(urlargs['clickeable'], ['clicked2'])
+        self.assertEqual(urlargs['one'], ['clicked1'])
+        self.assertEqual(urlargs['two'], ['clicked2'])
+
+    def test_from_response_unicode_clickdata(self):
+        body = u"""
+<form action="get.php" method="GET">
+<input type="submit" name="price in \u00a3" value="\u00a3 1000">
+<input type="submit" name="price in \u20ac" value="\u20ac 2000">
+<input type="hidden" name="poundsign" value="\u00a3">
+<input type="hidden" name="eurosign" value="\u20ac">
+</form>
+        """
+        response = HtmlResponse("http://www.example.com", body=body, \
+                                encoding='utf-8')
+        r1 = self.request_class.from_response(response, \
+                clickdata={'name': u'price in \u00a3'})
+        urlargs = cgi.parse_qs(urlparse(r1.url).query)
+        self.assertTrue(urlargs[u'price in \u00a3'.encode('utf-8')])
+
+    def test_from_response_multiple_forms_clickdata(self):
+        body = u"""
+        <form name="form1">
+          <input type="submit" name="clickeable" value="clicked">
+          <input type="hidden" name="field1" value="value1">
+        </form>
+        <form name="form2">
+          <input type="submit" name="clickeable" value="clicked">
+          <input type="hidden" name="field2" value="value2">
+        </form>
+        """
+        res = HtmlResponse("http://example.com", body=body, encoding='utf-8')
+        req = self.request_class.from_response(res, \
+                                              formname='form2', \
+                                              clickdata={'name': 'clickeable'})
+        urlargs = cgi.parse_qs(urlparse(req.url).query)
+        self.assertEqual(urlargs['clickeable'], ['clicked'])
+        self.assertEqual(urlargs['field2'], ['value2'])
+        self.assertFalse('field1' in urlargs, urlargs)
+
     def test_from_response_dont_click(self):
         respbody = """
 <form action="get.php" method="GET">
@@ -336,6 +389,17 @@ class FormRequestTest(RequestTest):
                           response,
                           clickdata={'type': 'submit'})
 
+    def test_from_response_non_matching_clickdata(self):
+        body = """
+        <form>
+          <input type="submit" name="clickeable" value="clicked">
+        </form>
+        """
+        res = HtmlResponse("http://example.com", body=body)
+        self.assertRaises(ValueError,
+                          self.request_class.from_response, res,
+                                clickdata={'nonexistent': 'notme'})
+
     def test_from_response_errors_noform(self):
         respbody = """<html></html>"""
         response = HtmlResponse("http://www.example.com/lala.html", body=respbody)