Trunk fix for #577 (GzipFilter doesn't force an update of the Content-Length header). All code which could change the length of response.body should delete the Content-Length header (if already set).

4 files changed, 49 insertions, 12 deletions

diff --git a/cherrypy/lib/cptools.py b/cherrypy/lib/cptools.py
index 9ae76086..2aced1b6 100644
--- a/cherrypy/lib/cptools.py
+++ b/cherrypy/lib/cptools.py
@@ -210,6 +210,9 @@ Message: %(error_msg)s
         if error_msg:
             body = self.login_screen(from_page, username, error_msg)
             cherrypy.response.body = body
+            if cherrypy.response.headers.has_key("Content-Length"):
+                # Delete Content-Length header so finalize() recalcs it.
+                del cherrypy.response.headers["Content-Length"]
             return True
         else:
             cherrypy.session[self.session_key] = username
@@ -235,6 +238,9 @@ Message: %(error_msg)s
             sess[self.session_key] = username = self.anonymous()
         if not username:
             cherrypy.response.body = self.login_screen(cherrypy.url(qs=request.query_string))
+            if cherrypy.response.headers.has_key("Content-Length"):
+                # Delete Content-Length header so finalize() recalcs it.
+                del cherrypy.response.headers["Content-Length"]
             return True
         
         self.on_check(username)
diff --git a/cherrypy/lib/encoding.py b/cherrypy/lib/encoding.py
index 82a1d081..14391880 100644
--- a/cherrypy/lib/encoding.py
+++ b/cherrypy/lib/encoding.py
@@ -100,6 +100,18 @@ def find_acceptable_charset(encoding=None, default_encoding='utf-8', errors='str
     else:
         response.collapse_body()
         encoder = encode_string
+        if response.headers.has_key("Content-Length"):
+            # Delete Content-Length header so finalize() recalcs it.
+            # Encoded strings may be of different lengths from their
+            # unicode equivalents, and even from each other. For example:
+            # >>> t = u"\u7007\u3040"
+            # >>> len(t)
+            # 2
+            # >>> len(t.encode("UTF-8"))
+            # 6
+            # >>> len(t.encode("utf7"))
+            # 8
+            del response.headers["Content-Length"]
     
     # Parse the Accept-Charset request header, and try to provide one
     # of the requested charsets (in order of user preference).
@@ -217,5 +229,8 @@ def gzip(compress_level=9, mime_types=['text/html', 'text/plain']):
                 
                 response.headers['Content-Encoding'] = 'gzip'
                 response.body = compress(response.body, compress_level)
+                if response.headers.has_key("Content-Length"):
+                    # Delete Content-Length header so finalize() recalcs it.
+                    del response.headers["Content-Length"]
             return
     cherrypy.HTTPError(406, "identity, gzip").set_response()
diff --git a/cherrypy/lib/static.py b/cherrypy/lib/static.py
index c28cf006..dee635e7 100644
--- a/cherrypy/lib/static.py
+++ b/cherrypy/lib/static.py
@@ -17,7 +17,7 @@ from cherrypy.lib import cptools, http
 def serve_file(path, content_type=None, disposition=None, name=None):
     """Set status, headers, and body in order to serve the given file.
     
-    The Content-Type header will be set to the content_ype arg, if provided.
+    The Content-Type header will be set to the content_type arg, if provided.
     If not provided, the Content-Type will be guessed by its extension.
     
     If disposition is not None, the Content-Disposition header will be set
@@ -96,7 +96,9 @@ def serve_file(path, content_type=None, disposition=None, name=None):
                 boundary = mimetools.choose_boundary()
                 ct = "multipart/byteranges; boundary=%s" % boundary
                 response.headers['Content-Type'] = ct
-##                del response.headers['Content-Length']
+                if response.headers.has_key("Content-Length"):
+                    # Delete Content-Length header so finalize() recalcs it.
+                    del response.headers["Content-Length"]
                 
                 def file_ranges():
                     # Apache compatibility:
diff --git a/cherrypy/lib/tidy.py b/cherrypy/lib/tidy.py
index cac5f9f8..25e34984 100644
--- a/cherrypy/lib/tidy.py
+++ b/cherrypy/lib/tidy.py
@@ -18,11 +18,13 @@ def tidy(temp_dir, tidy_path, strict_xml=False, errors_to_ignore=None,
     stable and it crashes on some HTML pages (which means that the
     server would also crash)
     """
+    response = cherrypy.response
+    
     # the tidy tool, by its very nature it's not generator friendly, 
     # so we just collapse the body and work with it.
-    orig_body = cherrypy.response.collapse_body()
+    orig_body = response.collapse_body()
     
-    fct = cherrypy.response.headers.get('Content-Type', '')
+    fct = response.headers.get('Content-Type', '')
     ct = fct.split(';')[0]
     encoding = ''
     i = fct.find('charset=')
@@ -73,8 +75,10 @@ def tidy(temp_dir, tidy_path, strict_xml=False, errors_to_ignore=None,
                     new_errs.append(err)
         
         if new_errs:
-            cherrypy.response.body = wrong_content('<br />'.join(new_errs),
-                                                   orig_body)
+            response.body = wrong_content('<br />'.join(new_errs), orig_body)
+            if response.headers.has_key("Content-Length"):
+                # Delete Content-Length header so finalize() recalcs it.
+                del response.headers["Content-Length"]
             return
         elif strict_xml:
             # The HTML is OK, but is it valid XML?
@@ -96,11 +100,17 @@ def tidy(temp_dir, tidy_path, strict_xml=False, errors_to_ignore=None,
                 body_file = StringIO.StringIO()
                 traceback.print_exc(file = body_file)
                 body_file = '<br />'.join(body_file.getvalue())
-                cherrypy.response.body = wrong_content(body_file, orig_body, "XML")
+                response.body = wrong_content(body_file, orig_body, "XML")
+                if response.headers.has_key("Content-Length"):
+                    # Delete Content-Length header so finalize() recalcs it.
+                    del response.headers["Content-Length"]
                 return
         
         if use_output:
-            cherrypy.response.body = [output]
+            response.body = [output]
+            if response.headers.has_key("Content-Length"):
+                # Delete Content-Length header so finalize() recalcs it.
+                del response.headers["Content-Length"]
 
 def html_space(text):
     """Escape text, replacing space with nbsp and tab with 4 nbsp's."""
@@ -118,11 +128,13 @@ def wrong_content(header, body, content_type="HTML"):
 
 
 def nsgmls(temp_dir, nsgmls_path, catalog_path, errors_to_ignore=None):
+    response = cherrypy.response
+    
     # the tidy tool, by its very nature it's not generator friendly, 
     # so we just collect the body and work with it.
-    orig_body = cherrypy.response.collapse_body()
+    orig_body = response.collapse_body()
     
-    fct = cherrypy.response.headers.get('Content-Type', '')
+    fct = response.headers.get('Content-Type', '')
     ct = fct.split(';')[0]
     encoding = ''
     i = fct.find('charset=')
@@ -162,6 +174,8 @@ def nsgmls(temp_dir, nsgmls_path, catalog_path, errors_to_ignore=None):
                 new_errs.append(err)
         
         if new_errs:
-            cherrypy.response.body = wrong_content('<br />'.join(new_errs),
-                                                   orig_body)
+            response.body = wrong_content('<br />'.join(new_errs), orig_body)
+            if response.headers.has_key("Content-Length"):
+                # Delete Content-Length header so finalize() recalcs it.
+                del response.headers["Content-Length"]