Don't try to dereference paths or URIs in inputstream.py

jgraham · jgraham · commit 0335f27143d2 · 2007-01-10T21:18:52.000Z
--HG--
extra : convert_revision : svn%3Aacbfec75-9323-0410-a652-858a13e371e0/trunk%40446
diff --git a/parse.py b/parse.py
@@ -28,9 +28,21 @@ def parse():
 
     try:
         f = args[-1]
+        # Try opening from the internet
+        if f.startswith('http://'):
+            try:
+                import urllib
+                f = urllib.urlopen(f).read()
+            except: pass
+        else:
+            try:
+                # Try opening from file system
+                f = open(f)
+            except IOError: pass
     except IndexError:
         print "No filename provided. Use -h for help"
         sys.exit(1)
+
     if opts.treebuilder is not None:
         try:
             treebuilder = __import__("treebuilders." + opts.treebuilder,
diff --git a/src/inputstream.py b/src/inputstream.py
@@ -74,18 +74,6 @@ def openStream(self, source):
         if hasattr(source, 'read'):
             stream = source
         else:
-            # Try opening from file system
-            try:
-                return open(source)
-            except: pass
-
-            # Try opening from the internet
-            try:
-                if source.startswith('http://'):
-                    import urllib
-                    source = urllib.urlopen(source).read()
-            except: pass
-
             # Otherwise treat source as a string and convert to a file object
             import cStringIO
             stream = cStringIO.StringIO(str(source))
diff --git a/tests/test_parser.py b/tests/test_parser.py
@@ -12,9 +12,10 @@
 import html5parser
 #Run tests over all treebuilders
 #XXX - it would be nice to automate finding all treebuilders or to allow running just one
+
 from treebuilders import simpletree, etree, dom
 
-treetypes = {"simpletree":simpletree.TreeBuilder, 
+treetypes = {"simpletree":simpletree.TreeBuilder,
              "ElementTree":etree.TreeBuilder,
              "DOM":dom.TreeBuilder}
 
