Improve type guessing for the default WebDAV PUT factory (#1004)

* - Improve type guessing for the default WebDAV PUT factory

* - defer to guess_content_type

* - cover case with no content-type header

CHANGES.rst

@@ -10,6 +10,9 @@ https://zope.readthedocs.io/en/2.13/CHANGES.html
 4.6.4 (unreleased)
 ------------------
 
+- Improve type guessing for the default WebDAV PUT factory
+  (`#997 <https://github.com/zopefoundation/Zope/issues/997>`_)
+
 - Enable WebDAV PUT factories to change a newly created object's ID
   (`#997 <https://github.com/zopefoundation/Zope/issues/997>`_)
 

src/Products/PageTemplates/PageTemplateFile.py

@@ -47,7 +47,7 @@ def guess_type(filename, body):
     # detect text/xml  if 'filename' won't end with .xml
     # XXX: fix this in zope.contenttype
 
-    if body.startswith(b'<?xml'):
+    if body.startswith(b'<?xml') or filename.lower().endswith('.xml'):
         return 'text/xml'
 
     content_type, ignored_encoding = guess_content_type(filename, body)

src/Products/PageTemplates/ZopePageTemplate.py

@@ -303,7 +303,7 @@ def PUT(self, REQUEST, RESPONSE):
         self.dav__init(REQUEST, RESPONSE)
         self.dav__simpleifhandler(REQUEST, RESPONSE, refresh=1)
         text = REQUEST.get('BODY', '')
-        content_type = guess_type('', text)
+        content_type = guess_type(self.getId(), text)
         self.pt_edit(text, content_type)
         RESPONSE.setStatus(204)
         return RESPONSE

src/webdav/NullResource.py

@@ -100,16 +100,22 @@ def HEAD(self, REQUEST, RESPONSE):
     def _default_PUT_factory(self, name, typ, body):
         # See if the name contains a file extension
         shortname, ext = os.path.splitext(name)
+        ext = ext.lower()
 
         # Make sure the body is bytes
         if not isinstance(body, bytes):
             body = body.encode('UTF-8')
 
+        # Guess the type of file if the passed content-type is
+        # just the generic application/octet-stream
+        if not typ or typ == 'application/octet-stream':
+            typ, encoding = guess_content_type(name, body)
+
         if ext == '.dtml':
             ob = DTMLDocument('', __name__=name)
         elif typ in ('text/html', 'text/xml'):
             ob = ZopePageTemplate(name, body, content_type=typ)
-        elif typ[:6] == 'image/':
+        elif typ.startswith('image/'):
             ob = Image(name, '', body, content_type=typ)
         else:
             ob = File(name, '', body, content_type=typ)

src/webdav/tests/testPUT_factory.py

@@ -112,3 +112,80 @@ def custom_put_factory(name, typ, body):
         put = request.traverse('/folder/doc')
         put(request, request.RESPONSE)
         self.assertTrue('newname' in self.folder.objectIds())
+
+    def test_default_PUT_factory_type_guessing(self):
+        # Check how the default PUT factory guesses the type of object to
+        # create. It is based on either the content-type request header or the
+        # file name.
+        from OFS.DTMLDocument import DTMLDocument
+        from OFS.Image import File
+        from OFS.Image import Image
+        from Products.PageTemplates.ZopePageTemplate import ZopePageTemplate
+        request = self.app.REQUEST
+
+        # DTML documents
+        put = request.traverse('/folder/doc.dtml')
+        put(request, request.RESPONSE)
+        self.assertIsInstance(self.folder['doc.dtml'], DTMLDocument)
+
+        # Page Templates
+        # PUT and content type guessing is messed up for ZopePageTemplates
+        request.environ['CONTENT_TYPE'] = 'text/html'
+        request['BODY'] = b'bar'
+        put = request.traverse('/folder/zpt1')
+        put(request, request.RESPONSE)
+        self.assertIsInstance(self.folder.zpt1, ZopePageTemplate)
+        self.assertEqual(self.folder.zpt1.content_type, 'text/html')
+
+        request['BODY'] = b'<?xml version="1.0" encoding="UTF-8"?>bar'
+        put = request.traverse('/folder/zpt2')
+        put(request, request.RESPONSE)
+        self.assertIsInstance(self.folder.zpt2, ZopePageTemplate)
+        self.assertEqual(self.folder.zpt2.content_type, 'text/xml')
+
+        for html_extension in ('.pt', '.zpt', '.html', '.htm'):
+            ob_id = 'zpt%s' % html_extension
+            request.environ['CONTENT_TYPE'] = 'application/octet-stream'
+            request['BODY'] = b'<html></html>'
+            put = request.traverse('/folder/%s' % ob_id)
+            put(request, request.RESPONSE)
+            zope_ob = getattr(self.folder, ob_id)
+            self.assertIsInstance(zope_ob, ZopePageTemplate)
+            self.assertEqual(zope_ob.content_type, 'text/html')
+
+        request.environ['CONTENT_TYPE'] = 'application/octet-stream'
+        put = request.traverse('/folder/zpt.xml')
+        put(request, request.RESPONSE)
+        zope_ob = self.folder['zpt.xml']
+        self.assertIsInstance(zope_ob, ZopePageTemplate)
+        self.assertEqual(zope_ob.content_type, 'text/xml')
+
+        # Images
+        for content_type in ('image/jpg', 'image/gif', 'image/png'):
+            request.environ['CONTENT_TYPE'] = content_type
+            ob_id = 'img_%s' % content_type.replace('/', '_')
+            put = request.traverse('/folder/%s' % ob_id)
+            put(request, request.RESPONSE)
+            zope_ob = getattr(self.folder, ob_id)
+            self.assertIsInstance(zope_ob, Image)
+            self.assertEqual(zope_ob.content_type, content_type)
+
+        for extension in ('.jpg', '.jpeg', '.gif', '.png', '.tiff'):
+            ob_id = 'img%s' % extension
+            request.environ['CONTENT_TYPE'] = 'application/octet-stream'
+            put = request.traverse('/folder/%s' % ob_id)
+            put(request, request.RESPONSE)
+            zope_ob = getattr(self.folder, ob_id)
+            self.assertIsInstance(zope_ob, Image)
+
+        # File, the last fallback
+        for content_type in ('text/plain', 'application/pdf',
+                             'application/octet-stream'):
+            request.environ['CONTENT_TYPE'] = content_type
+            request['BODY'] = b'foobar'
+            ob_id = 'file_%s' % content_type.replace('/', '_')
+            put = request.traverse('/folder/%s' % ob_id)
+            put(request, request.RESPONSE)
+            zope_ob = getattr(self.folder, ob_id)
+            self.assertIsInstance(zope_ob, File)
+            self.assertEqual(zope_ob.content_type, content_type)