Fixed various issues related to unicode and CGI variables (assuming that CGI variables are utf-8 encoded, and sometime percent encoded).

Delayed computation of `value' and `name' from WSF_STRING.
Fixed computation of REQUEST_URI when the server does not provide it (this is rare, but possible).
   compute it as SERVER_NAME + encoded-PATH_INFO + {? + QUERY_STRING}
This commit is contained in:
2014-07-01 19:57:14 +02:00
parent 0b1697f20d
commit 7dfc6ea67a
3 changed files with 140 additions and 19 deletions

View File

@@ -70,6 +70,7 @@ feature -- Request processing
e: EXECUTION_ENVIRONMENT
enc: URL_ENCODER
utf: UTF_CONVERTER
do
l_request_uri := a_handler.uri
l_headers_map := a_handler.request_header_map
@@ -80,7 +81,7 @@ feature -- Request processing
across
vars as c
loop
env.force (enc.encoded_string (c.item), enc.encoded_string (c.key))
env.force (utf.utf_32_string_to_utf_8_string_8 (c.item), utf.utf_32_string_to_utf_8_string_8 (c.key))
end
else
create env.make (0)

View File

@@ -306,8 +306,11 @@ feature {NONE} -- Element change: CGI meta parameter related to PATH_INFO
s: like meta_string_variable
table: STRING_TABLE [READABLE_STRING_8]
l_query_string: like query_string
l_request_uri: detachable STRING_32
l_request_uri: detachable READABLE_STRING_8
s8: STRING_8
l_empty_string: like empty_string
enc: PERCENT_ENCODER
utf: UTF_CONVERTER
do
create {STRING_8} l_empty_string.make_empty
empty_string := l_empty_string
@@ -319,8 +322,11 @@ feature {NONE} -- Element change: CGI meta parameter related to PATH_INFO
until
a_vars.after
loop
-- Warning: truncated value to ascii !!!
if attached {READABLE_STRING_32} a_vars.item_for_iteration as s32 then
table.force (utf.utf_32_string_to_utf_8_string_8 (s32), a_vars.key_for_iteration)
else
table.force (a_vars.item_for_iteration.to_string_8, a_vars.key_for_iteration)
end
a_vars.forth
end
@@ -370,14 +376,22 @@ feature {NONE} -- Element change: CGI meta parameter related to PATH_INFO
if s /= Void then
l_request_uri := s
else
--| It might occur that REQUEST_URI is not available, so let's compute it from SCRIPT_NAME
create l_request_uri.make_from_string (script_name)
--| REQUEST_URI is not always available, in this case,
--| compute it from SCRIPT_NAME, PATH_INFO and QUERY_STRING which are required by CGI.
create s8.make_from_string (script_name)
create enc
enc.append_partial_percent_encoded_string_to (utf.utf_8_string_8_to_string_32 (path_info), s8, <<'/', '!', '$', '&', '%'', '(', ')', '*', '+', ',', ';', '='>>)
if not l_query_string.is_empty then
l_request_uri.append_character ('?')
l_request_uri.append (l_query_string)
s8.append_character ('?')
s8.append (l_query_string)
end
l_request_uri := s8
end
request_uri := single_slash_starting_string (l_request_uri)
--| FIXME: should it strip "////abc/def" to "/abc/def" ?
--| Not sure why this was done before.
-- l_request_uri := single_slash_starting_string (l_request_uri)
request_uri := l_request_uri
set_meta_string_variable ({WGI_META_NAMES}.request_uri, l_request_uri)
end
set_orig_path_info (s: READABLE_STRING_8)
@@ -482,7 +496,7 @@ invariant
empty_string_unchanged: empty_string.is_empty
note
copyright: "2011-2013, Jocelyn Fiat, Javier Velilla, Eiffel Software and others"
copyright: "2011-2014, Jocelyn Fiat, Javier Velilla, Eiffel Software and others"
license: "Eiffel Forum License v2 (see http://www.eiffel.com/licensing/forum.txt)"
source: "[
Eiffel Software

View File

@@ -25,11 +25,8 @@ feature {NONE} -- Initialization
make (a_name: READABLE_STRING_8; a_string: READABLE_STRING_8)
do
name := url_decoded_string (a_name)
value := url_decoded_string (a_string)
url_encoded_name := a_name
url_encoded_value := a_string
url_encoded_name := utf_8_percent_encoded_string (a_name)
url_encoded_value := utf_8_percent_encoded_string (a_string)
end
feature -- Access
@@ -38,11 +35,31 @@ feature -- Access
-- <Precursor>
--| Note that the value might be html encoded as well
--| this is the application responsibility to html decode it
local
v: like internal_name
do
v := internal_name
if v = Void then
v := url_decoded_string (url_encoded_name)
internal_name := v
end
Result := v
end
value: READABLE_STRING_32
-- <Precursor>
--| Note that the value might be html encoded as well
--| this is the application responsibility to html decode it
local
v: like internal_value
do
v := internal_value
if v = Void then
v := url_decoded_string (url_encoded_value)
internal_value := v
end
Result := v
end
url_encoded_name: READABLE_STRING_8
-- URL encoded string of `name'.
@@ -50,6 +67,14 @@ feature -- Access
url_encoded_value: READABLE_STRING_8
-- URL encoded string of `value'.
feature {NONE} -- Access: internals
internal_name: detachable like name
-- Cached value of `name'.
internal_value: detachable like value
-- Cached value of `value'.
feature -- Conversion
integer_value: INTEGER
@@ -81,7 +106,7 @@ feature -- Element change
change_name (a_name: like name)
do
name := url_decoded_string (a_name)
internal_name := Void
url_encoded_name := a_name
ensure then
a_name.same_string (url_encoded_name)
@@ -122,8 +147,89 @@ feature -- Visitor
vis.process_string (Current)
end
feature {NONE} -- Implementation
utf_8_percent_encoded_string (s: READABLE_STRING_8): READABLE_STRING_8
-- Percent-encode the UTF-8 sequence characters from UTF-8 encoded `s' and
-- return the Result.
local
s8: STRING_8
i, n, nb: INTEGER
do
-- First check if there are such UTF-8 character
-- If it has, convert them and return a new object as Result
-- otherwise return `s' directly to avoid creating a new object
from
i := 1
n := s.count
nb := 0
until
i > n
loop
if s.code (i) > 0x7F then -- >= 128
nb := nb + 1
end
i := i + 1
end
if nb > 0 then
create s8.make (s.count + nb * 3)
utf_8_string_8_into_percent_encoded_string_8 (s, s8)
Result := s8
else
Result := s
end
end
utf_8_string_8_into_percent_encoded_string_8 (s: READABLE_STRING_8; a_result: STRING_8)
-- Copy STRING_32 corresponding to UTF-8 sequence `s' appended into `a_result'.
local
i: INTEGER
n: INTEGER
c: NATURAL_32
do
from
n := s.count
a_result.grow (a_result.count + n)
until
i >= n
loop
i := i + 1
c := s.code (i)
if c <= 0x7F then
-- 0xxxxxxx
a_result.append_code (c)
elseif c <= 0xDF then
-- 110xxxxx 10xxxxxx
url_encoder.append_percent_encoded_character_code_to (c, a_result)
i := i + 1
if i <= n then
url_encoder.append_percent_encoded_character_code_to (s.code (i), a_result)
end
elseif c <= 0xEF then
-- 1110xxxx 10xxxxxx 10xxxxxx
url_encoder.append_percent_encoded_character_code_to (s.code (i), a_result)
i := i + 2
if i <= n then
url_encoder.append_percent_encoded_character_code_to (s.code (i - 1), a_result)
url_encoder.append_percent_encoded_character_code_to (s.code (i), a_result)
end
elseif c <= 0xF7 then
-- 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
url_encoder.append_percent_encoded_character_code_to (s.code (i), a_result)
i := i + 3
if i <= n then
url_encoder.append_percent_encoded_character_code_to (s.code (i - 2), a_result)
url_encoder.append_percent_encoded_character_code_to (s.code (i - 1), a_result)
url_encoder.append_percent_encoded_character_code_to (s.code (i), a_result)
end
else
a_result.append_code (c)
end
end
end
note
copyright: "2011-2013, Jocelyn Fiat, Javier Velilla, Olivier Ligot, Eiffel Software and others"
copyright: "2011-2014, Jocelyn Fiat, Javier Velilla, Olivier Ligot, Colin Adams, Eiffel Software and others"
license: "Eiffel Forum License v2 (see http://www.eiffel.com/licensing/forum.txt)"
source: "[
Eiffel Software