http.rb 1.3.4 → 1.3.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a6fd7a7020d8a692cd3852ed9aa2820322ad24b9e61a368d9e70dd4a2a9b6682
-  data.tar.gz: 4b535cc818bef4bfd17ea1e3b2ef177706fede6a5772c48f7184d57e252314ee
+  metadata.gz: b19f251bfdf7beb262c7f6fe8c00950fb580937b1679cef33059346e4695d8a4
+  data.tar.gz: 6dd0dfa38e25d6805552778398c36df2bf8e42d6f507f23fe8bde2aeb229945c
 SHA512:
-  metadata.gz: a0c53543092fbf97a7fd94418b3c6360320dc4d4eb8711bc092c7a95d1801965e7c62e51b67988e6d28e637531f7bdcc74c4ed1633fabc4ae7be8bfb70b29bdf
-  data.tar.gz: becf134eada7e1ba1d3af4433c17599e2aaecce3b100b538a9818b8f0c4ea269086642f6d8b8acce59c41872898b0ee333bf37702f2841e1fb710623b8498a75
+  metadata.gz: 923a61906a76fd676e1f47808ccc3d295cc9d7c6b5b977d5ec1ff9a496daf8488dfd5f49f1169b650343d66aba9d9994cc166d9caf9d2a64a9e6440c5ac3fb64
+  data.tar.gz: 672784fee166a077e19e8b79866c24b84793c9aa82c7e4bbe217cbca60565fa0ad241c3d739bb3808108a4caa9ac2f036428943c7d1337327f35b8083ddc737b

data/CHANGELOG

@@ -1,5 +1,18 @@
 # CHANGELOG
 
+## 20261011
+
+1.3.5: Non-ASCII is URL-encoded in full, and without a warning.
+
+A query argument holding a multibyte character was encoded one byte per character, so only the first byte of each went out: 'Älaskår' became '%C3lask%C3r' where it should have been '%C3%84lask%C3%A5r', and a server decoding it got invalid UTF-8. The warning this was first noticed by, "historical binary regexp match /.../n against UTF-8 string", came from the same line and is only a warning, but dropping the /n alone would have left the bytes truncated, which is what kept the warning from seeming to matter.
+
+1. ~ Thoran::String::Urlencode#urlencode: /$1.size/$1.bytesize/, so that each byte of a character is encoded rather than the first only.
+2. ~ Thoran::String::Urlencode#urlencode: - /n flag. Matched against a UTF-8 string, a binary regexp warns on current Ruby, and the match is the same without it: a run of non-ASCII characters is still matched whole and then unpacked byte by byte, and a binary string is still matched byte-wise.
+3. ~ Thoran/String/Urlencode/urlencode.rb: /0.4.0/0.4.1/ in the file's own header.
+4. + test/Thoran/String/Urlencode/urlencode_test.rb: unreserved and reserved characters, a space, two and three byte characters, a binary string, the absence of a warning, and the url_encode alias. The multibyte, warning, and alias examples fail against 1.3.4.
+5. ~ HTTP::VERSION: /1.3.4/1.3.5/
+
+
 ## 20261011
 
 1.3.4: ~ LICENSE: /2026/2009-2026/

data/lib/HTTP/VERSION.rb

@@ -2,5 +2,5 @@
 # HTTP::VERSION
 
 module HTTP
-  VERSION = '1.3.4'
+  VERSION = '1.3.5'
 end

data/lib/Thoran/String/Urlencode/urlencode.rb

@@ -1,8 +1,8 @@
 # Thoran/String/Urlencode/urlencode.rb
 # Thoran::String::Urlencode#urlencode
 
-# 20260713
-# 0.4.0
+# 20261011
+# 0.4.1
 
 # Acknowledgements: I've simply ripped off and refashioned the code from the CGI module!...
 
@@ -10,14 +10,17 @@
 # -/0: More consistent naming.
 # 1. /UrlEncode/Urlencode/
 # 2. /url_encode/urlencode/
+# 0/1: Non-ASCII is encoded in full, and without a warning.
+# 3. /$1.size/$1.bytesize/: unpack('H2' * $1.size) took one byte per character, so only the first byte of a multibyte character was encoded and 'Älaskår' became '%C3lask%C3r'.
+# 4. - /n flag: a binary regexp matched against a UTF-8 string warns on current Ruby (seen on 4.0.5), and the match is the same without it.
 
 module Thoran
   module String
     module Urlencode
 
       def urlencode
-        self.gsub(/([^ a-zA-Z0-9_.-]+)/n) do
-          '%' + $1.unpack('H2' * $1.size).join('%').upcase
+        self.gsub(/([^ a-zA-Z0-9_.-]+)/) do
+          '%' + $1.unpack('H2' * $1.bytesize).join('%').upcase
         end.tr(' ', '+')
       end
       alias_method :url_encode, :urlencode

data/test/Thoran/String/Urlencode/urlencode_test.rb

@@ -0,0 +1,41 @@
+# test/Thoran/String/Urlencode/urlencode_test.rb
+
+require_relative '../../../helper'
+
+describe Thoran::String::Urlencode do
+  describe "#urlencode" do
+    it "leaves unreserved characters alone" do
+      _('aZ09_.-'.urlencode).must_equal('aZ09_.-')
+    end
+
+    it "encodes a space as a plus" do
+      _('a b'.urlencode).must_equal('a+b')
+    end
+
+    it "encodes reserved characters" do
+      _('a/b?c=d&e'.urlencode).must_equal('a%2Fb%3Fc%3Dd%26e')
+    end
+
+    it "encodes every byte of a two byte character" do
+      _('Älaskår'.urlencode).must_equal('%C3%84lask%C3%A5r')
+    end
+
+    it "encodes every byte of a three byte character" do
+      _('✓'.urlencode).must_equal('%E2%9C%93')
+    end
+
+    it "encodes a binary string byte by byte" do
+      _('Ä'.b.urlencode).must_equal('%C3%84')
+    end
+
+    it "does not warn on a UTF-8 string" do
+      _{'Älaskår'.urlencode}.must_be_silent
+    end
+  end
+
+  describe "#url_encode" do
+    it "is an alias of #urlencode" do
+      _('Ashër'.url_encode).must_equal('Ash%C3%ABr')
+    end
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: http.rb
 version: !ruby/object:Gem::Version
-  version: 1.3.4
+  version: 1.3.5
 platform: ruby
 authors:
 - thoran
@@ -122,6 +122,7 @@ files:
 - test/HTTP/request_without_body_test.rb
 - test/HTTP/trace_test.rb
 - test/Net/HTTPResponse/StatusPredicates_test.rb
+- test/Thoran/String/Urlencode/urlencode_test.rb
 - test/helper.rb
 homepage: https://github.com/thoran/http
 licenses: