Better support for unicode path and values.

Added WSF_REQUEST.percent_encoded_path_info: READABLE_STRING_8 to keep url encoded path info, as it is useful for specific component The router is now using WSF_REQUEST.percent_encoded_path_info since URI_TEMPLATE are handling URI (and not IRI) this fixes an issue with unicode path parameters. This should not break existing code, and this fixes various unicode related issues related to PATH parameter and path info but also any component using file names. (required EiffelStudio >= 7.2)
2013-06-12 18:00:55 +02:00
parent a982286dd4
commit cc4ef1a575
28 changed files with 1056 additions and 449 deletions
--- a/library/text/encoder/encoder-safe.ecf
+++ b/library/text/encoder/encoder-safe.ecf
@@ -11,22 +11,10 @@
 			<assertions precondition="true"/>
 		</option>
 		<library name="base" location="$ISE_LIBRARY\library\base\base-safe.ecf"/>
-		<library name="encoding" location="$ISE_LIBRARY\library\encoding\encoding-safe.ecf"/>
 		<cluster name="src" location="src\" recursive="true">
 			<file_rule>
 				<exclude>/tests$</exclude>
-				<exclude>/spec$</exclude>
 			</file_rule>
-			<cluster name="src_before_70" location="$|spec\before_70\" recursive="true">
-				<condition>
-					<version type="compiler" max="7.0.8.7585"/>
-				</condition>
-			</cluster>
-			<cluster name="src_70" location="$|spec\70\" recursive="true">
-				<condition>
-					<version type="compiler" min="7.0.8.7586"/>
-				</condition>
-			</cluster>
 		</cluster>
 	</target>
 </system>
--- a/library/text/encoder/encoder.ecf
+++ b/library/text/encoder/encoder.ecf
@@ -11,22 +11,10 @@
 			<assertions precondition="true"/>
 		</option>
 		<library name="base" location="$ISE_LIBRARY\library\base\base.ecf"/>
-		<library name="encoding" location="$ISE_LIBRARY\library\encoding\encoding.ecf"/>
 		<cluster name="src" location="src\" recursive="true">
 			<file_rule>
 				<exclude>/tests$</exclude>
-				<exclude>/spec$</exclude>
 			</file_rule>
-			<cluster name="src_before_70" location="$\spec\before_70" recursive="true">
-				<condition>
-					<version type="compiler" max="7.0.8.7585"/>
-				</condition>
-			</cluster>
-			<cluster name="src_70" location="$\spec\70" recursive="true">
-				<condition>
-					<version type="compiler" min="7.0.8.7586"/>
-				</condition>
-			</cluster>
 		</cluster>
 	</target>
 </system>
--- a/library/text/encoder/src/spec/70/utf8_encoder_helper.e
+++ b/library/text/encoder/src/spec/70/utf8_encoder_helper.e
@@ -1,26 +1,21 @@
 note
-	description : "Objects that ..."
-	author      : "$Author$"
-	date        : "$Date$"
-	revision    : "$Revision$"
+	description: "Objects to access the shared once UTF8_URL_ENCODER ..."
+	date: "$Date$"
+	revision: "$Revision$"

-deferred class
-	UTF8_ENCODER_HELPER
+class
+	SHARED_UTF8_URL_ENCODER

-inherit
-	ANY
+feature -- Encoder

-	UNICODE_CONVERSION
-		export
-			{NONE} all
-			{ANY} is_valid_utf8
-		undefine
-			is_little_endian
+	url_encoder: UTF8_URL_ENCODER
+			-- Shared UTF8 URL encoder.
+		once
+			create Result
 		end

-
 note
-	copyright: "2011-2011, Eiffel Software and others"
+	copyright: "2011-2012, Eiffel Software and others"
 	license: "Eiffel Forum License v2 (see http://www.eiffel.com/licensing/forum.txt)"
 	source: "[
 			Eiffel Software
@@ -29,4 +24,5 @@ note
 			Website http://www.eiffel.com
 			Customer support http://support.eiffel.com
 		]"
+
 end
--- a/library/text/encoder/src/spec/before_70/utf8_encoder_helper.e
+++ b/library/text/encoder/src/spec/before_70/utf8_encoder_helper.e
@@ -1,73 +0,0 @@
-note
-	description : "Objects that ..."
-	author      : "$Author$"
-	date        : "$Date$"
-	revision    : "$Revision$"
-
-deferred class
-	UTF8_ENCODER_HELPER
-
-inherit
-	ANY
-
-	UNICODE_CONVERSION
-		export
-			{NONE} all
-		undefine
-			is_little_endian
-		end
-
-feature -- Status report
-
-	is_valid_utf8 (a_string: STRING): BOOLEAN
-			-- Is `a_string' valid UTF-8 string?
-		require
-			a_string_not_void: a_string /= Void
-		local
-			l_nat8: NATURAL_8
-			l_code: NATURAL_32
-			i, nb: INTEGER
-		do
-			from
-				i := 1
-				nb := a_string.count
-				Result := True
-			until
-				i > nb or not Result
-			loop
-				l_nat8 := a_string.code (i).to_natural_8
-				if l_nat8 <= 127 then
-						-- Form 0xxxxxxx.
-				elseif (l_nat8 & 0xE0) = 0xC0 then
-						-- Form 110xxxxx 10xxxxxx.
-					l_code := (l_nat8 & 0x1F).to_natural_32 |<< 6
-					i := i + 1
-				elseif (l_nat8 & 0xF0) = 0xE0 then
-					-- Form 1110xxxx 10xxxxxx 10xxxxxx.
-					i := i + 2
-				elseif (l_nat8 & 0xF8) = 0xF0 then
-					-- Form 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx.
-					i := i + 3
-				elseif (l_nat8 & 0xFC) = 0xF8 then
-					-- Starts with 111110xx
-					Result := False
-				else
-					-- Starts with 1111110x
-					Result := False
-				end
-				i := i + 1
-			end
-		end
-
-
-note
-	copyright: "2011-2011, Eiffel Software and others"
-	license: "Eiffel Forum License v2 (see http://www.eiffel.com/licensing/forum.txt)"
-	source: "[
-			Eiffel Software
-			5949 Hollister Ave., Goleta, CA 93117 USA
-			Telephone 805-685-1006, Fax 805-685-6869
-			Website http://www.eiffel.com
-			Customer support http://support.eiffel.com
-		]"
-end
--- a/library/text/encoder/src/utf8_encoder.e
+++ b/library/text/encoder/src/utf8_encoder.e
@@ -15,8 +15,6 @@ class
 inherit
 	ENCODER [READABLE_STRING_32, READABLE_STRING_8]

-	UTF8_ENCODER_HELPER
-
 	PLATFORM
 		export
 			{NONE} all
@@ -37,9 +35,13 @@ feature -- Encoder

 	encoded_string (s: READABLE_STRING_32): STRING_8
 			-- UTF8-encoded value of `s'.
+		do
+			Result := general_encoded_string (s)
+		end
+
+	general_encoded_string (s: READABLE_STRING_GENERAL): STRING_8
 		do
 			Result := utf32_to_utf8 (s)
-			has_error := not last_conversion_successful
 		end

 feature -- Decoder
@@ -48,11 +50,34 @@ feature -- Decoder
 			-- The UTF8-encoded equivalent of the given string
 		do
 			Result := utf8_to_utf32 (v)
-			has_error := not last_conversion_successful
+			has_error := not is_valid_utf8 (v)
+		end
+
+feature {NONE} -- UTF implementation
+
+	utf32_to_utf8 (s: READABLE_STRING_GENERAL): STRING_8
+		local
+			utf: UTF_CONVERTER
+		do
+			Result := utf.utf_32_string_to_utf_8_string_8 (s)
+		end
+
+	utf8_to_utf32 (s: READABLE_STRING_8): STRING_32
+		local
+			utf: UTF_CONVERTER
+		do
+			Result := utf.utf_8_string_8_to_string_32 (s)
+		end
+
+	is_valid_utf8 (s: READABLE_STRING_8): BOOLEAN
+		local
+			utf: UTF_CONVERTER
+		do
+			Result := utf.is_valid_utf_8_string_8 (s)
 		end

 note
-	copyright: "2011-2012, Eiffel Software and others"
+	copyright: "2011-2013, Eiffel Software and others"
 	license: "Eiffel Forum License v2 (see http://www.eiffel.com/licensing/forum.txt)"
 	source: "[
 			Eiffel Software
--- a/library/text/encoder/src/utf8_url_encoder.e
+++ b/library/text/encoder/src/utf8_url_encoder.e
@@ -15,23 +15,24 @@ class
 inherit
 	URL_ENCODER
 		redefine
-			default_create,
 			name,
 			general_encoded_string,
 			encoded_string, partial_encoded_string,
 			decoded_string
+		select
+			encoded_string,
+			decoded_string,
+			has_error
 		end

-	UTF8_ENCODER_HELPER
+	UTF8_ENCODER
+		rename
+			general_encoded_string as utf8_general_encoded_string,
+			encoded_string as utf8_encoded_string,
+			decoded_string as utf8_decoded_string,
+			has_error as utf8_has_error
 		redefine
-			default_create
-		end
-
-feature {NONE} -- Initialization
-
-	default_create
-		do
-			Precursor {UTF8_ENCODER_HELPER}
+			name
 		end

 feature -- Access
@@ -46,27 +47,22 @@ feature -- Encoder
 	encoded_string (s: READABLE_STRING_32): STRING_8
 			-- URL-encoded value of `s'.
 		do
-			Result := utf32_to_utf8 (s)
-			Result := Precursor (Result)
+			Result := general_encoded_string (s)
 		end

 	general_encoded_string (s: READABLE_STRING_GENERAL): STRING_8
 		do
-			if attached {READABLE_STRING_32} s as s32 then
-				Result := utf32_to_utf8 (s32)
-			else
-				Result := s.as_string_8
-			end
-			Result := Precursor (Result)
+			Result := utf8_general_encoded_string (s)
+			Result := Precursor {URL_ENCODER} (Result)
+			has_error := has_error or utf8_has_error
 		end

 	partial_encoded_string (s: READABLE_STRING_GENERAL; a_ignore: ARRAY [CHARACTER]): STRING_8
 			-- URL-encoded value of `s'.
 		do
-			Result := Precursor (s, a_ignore)
-			if not has_error then
-				Result := utf32_to_utf8 (Result)
-			end
+			Result := utf8_general_encoded_string (s)
+			Result := Precursor {URL_ENCODER} (Result, a_ignore)
+			has_error := has_error or utf8_has_error
 		end

 feature -- Decoder
@@ -74,17 +70,15 @@ feature -- Decoder
 	decoded_string (v: READABLE_STRING_8): STRING_32
 			-- The URL-encoded equivalent of the given string
 		do
-			Result := Precursor (v)
+			Result := Precursor {URL_ENCODER} (v)
 			if not has_error then
-				if is_valid_utf8 (Result) then
-					Result := utf8_to_utf32 (Result)
-					has_error := not last_conversion_successful
-				end
+				Result := utf8_decoded_string (Result)
+				has_error := utf8_has_error
 			end
 		end

 note
-	copyright: "2011-2012, Eiffel Software and others"
+	copyright: "2011-2013, Eiffel Software and others"
 	license: "Eiffel Forum License v2 (see http://www.eiffel.com/licensing/forum.txt)"
 	source: "[
 			Eiffel Software