Run tests on both DOMlite and etree trees. At the moment further tree imps have to be added by hand

jgraham · jgraham · commit 2566f09dffb5 · 2007-01-02T14:04:47.000Z
--HG--
extra : convert_revision : svn%3Aacbfec75-9323-0410-a652-858a13e371e0/trunk%40400
diff --git a/parse.py b/parse.py
@@ -31,7 +31,7 @@ def parse():
     except IndexError:
         print "No filename provided. Use -h for help"
         sys.exit(1)
-    if hasattr(opts, "treebuilder"):
+    if opts.treebuilder is not None:
         try:
             #This isn't a great way to do this
             exec("import treebuilders.%s")%opts.treebuilder.split(".")[0]
@@ -42,7 +42,9 @@ def parse():
         except:
             treebuilder = treebuilders.DOMlite.TreeBuilder
     else:
-        treebuilder = treebuilders.DOMlite.TreeBuilder
+        import treebuilders.DOMlite
+        treebuilder = treebuilders.DOMlite
+
     p = parser.HTMLParser(tree=treebuilder)
 
     if opts.profile:
@@ -62,12 +64,16 @@ def parse():
         document = p.parse(f)
         t1 = time.time()
         print p.tree.testSerializer(document)
+        if opts.error:
+            print "\nParse errors:\n" + "\n".join(p.errors)
         t2 = time.time()
         print "\n\nRun took: %fs (plus %fs to print the output)"%(t1-t0, t2-t1)
     else:
         document = p.parse(f)
+        print document
         print p.tree.testSerializer(document)
-        print "\nParse errors:\n" + "\n".join(p.errors)
+        if opts.error:
+            print "\nParse errors:\n" + "\n".join(p.errors)
 
 def getOptParser():
     parser = OptionParser(usage=__doc__)
@@ -83,6 +89,9 @@ def getOptParser():
     parser.add_option("-b", "--treebuilder", action="store", type="string",
                       dest="treebuilder")
 
+    parser.add_option("-e", "--error", action="store_true", default=False,
+                      dest="error", help="Print a list of parse errors")
+
     return parser
 
 if __name__ == "__main__":
diff --git a/src/treebuilders/DOMlite.py b/src/treebuilders/DOMlite.py
@@ -1,4 +1,4 @@
-import base
+import _base
 
 # Really crappy basic implementation of a DOM-core like thing
 class Node(object):
@@ -117,7 +117,7 @@ def __init__(self, data):
     def __str__(self):
         return "<!-- %s -->" % self.data
 
-class TreeBuilder(base.TreeBuilder):
+class TreeBuilder(_base.TreeBuilder):
     documentClass = Document
     doctypeClass = DocumentType
     elementClass = Element
diff --git a/src/treebuilders/_base.py b/src/treebuilders/_base.py
diff --git a/src/treebuilders/etree.py b/src/treebuilders/etree.py
@@ -3,7 +3,7 @@
 except ImportError:
     from elementtree import ElementTree
 
-import base
+import _base
 
 class Element(object):
     def __init__(self, name):
@@ -146,7 +146,7 @@ def serializeElement(element, indent=0):
 
     return "\n".join(rv)
 
-class TreeBuilder(base.TreeBuilder):
+class TreeBuilder(_base.TreeBuilder):
     documentClass = Document
     doctypeClass = DocumentType
     elementClass = Element
@@ -164,4 +164,4 @@ def reparentChildren(self, oldParent, newParent):
         else:
             newParent._element.text += oldParent._element.text
         oldParent._element.text = ""
-        base.TreeBuilder.reparentChildren(self, oldParent, newParent)
+        _base.TreeBuilder.reparentChildren(self, oldParent, newParent)
diff --git a/tests/test_parser.py b/tests/test_parser.py
@@ -10,7 +10,12 @@
 sys.path.insert(0, os.path.abspath(os.path.join(os.pardir, "src")))
 
 import parser
-import treebuilders
+#Run tests over all treebuilders
+#XXX - it would be nice to automate finding all treebuilders or to allow running just one
+from treebuilders import DOMlite, etree
+
+treetypes = {"DOMlite":DOMlite.TreeBuilder, 
+             "ElementTree":etree.TreeBuilder}
 
 def parseTestcase(testString):
     testString = testString.split("\n")
@@ -52,39 +57,40 @@ def convertTreeDump(treedump):
     return "\n".join(rv)
 
 class TestCase(unittest.TestCase):
-    def runParserTest(self, input, output, errors):
+    def runParserTest(self, input, output, errors, treeClass):
         #XXX - move this out into the setup function
         #concatenate all consecutive character tokens into a single token
-        p = parser.HTMLParser()
+        p = parser.HTMLParser(tree = treeClass)
         document = p.parse(StringIO.StringIO(input))
         errorMsg = "\n".join(["\n\nExpected:", output, "\nRecieved:",
-          convertTreeDump(p.tree.testSerializer(document))])
-        self.assertEquals(output, 
+                              convertTreeDump(p.tree.testSerializer(document))])
+        self.assertEquals(output,
                           convertTreeDump(p.tree.testSerializer(document)),
                           errorMsg)
 #         errorMsg2 = "\n".join(["\n\nInput errors:\n" + "\n".join(errors),
 #           "Actual errors:\n" + "\n".join(p.errors)])
 #         self.assertEquals(len(p.errors), len(errors), errorMsg2)
 
 def test_parser():
-    for filename in glob.glob('tree-construction/*.dat'):
-        f = open(filename)
-        tests = f.read().split("#data\n")
-        for test in tests:
-            if test == "":
-                continue
-            test = "#data\n" + test
-            input, output, errors = parseTestcase(test)
-            yield TestCase.runParserTest, input, output, errors
+    for name, cls in treetypes.iteritems():
+        for filename in glob.glob('tree-construction/*.dat'):
+            f = open(filename)
+            tests = f.read().split("#data\n")
+            for test in tests:
+                if test == "":
+                    continue
+                test = "#data\n" + test
+                input, output, errors = parseTestcase(test)
+                yield TestCase.runParserTest, input, output, errors, name, cls
 
 def buildTestSuite():
     tests = 0
-    for func, input, output, errors in test_parser():
+    for func, input, output, errors, treeName, treeCls in test_parser():
         tests += 1
         testName = 'test%d' % tests
         testFunc = lambda self, method=func, input=input, output=output, \
-            errors=errors: method(self, input, output, errors)
-        testFunc.__doc__ = 'Parser %s: %s' % (testName, input)
+            errors=errors, treeCls=treeCls: method(self, input, output, errors, treeCls)
+        testFunc.__doc__ = 'Parser %s Tree %s Input: %s'%(testName, treeName, input)
         instanceMethod = new.instancemethod(testFunc, None, TestCase)
         setattr(TestCase, testName, instanceMethod)
     return unittest.TestLoader().loadTestsFromTestCase(TestCase)
@@ -94,7 +100,4 @@ def main():
     unittest.main()
 
 if __name__ == "__main__":
-    # XXX Allow us to import the sibling module
-    os.chdir(os.path.split(os.path.abspath(__file__))[0])
-    sys.path.insert(0, os.path.abspath(os.path.join(os.pardir, "src")))
     main()
