summaryrefslogtreecommitdiff
path: root/spec/ruby/core/string/shared
diff options
context:
space:
mode:
Diffstat (limited to 'spec/ruby/core/string/shared')
-rw-r--r--spec/ruby/core/string/shared/chars.rb12
-rw-r--r--spec/ruby/core/string/shared/codepoints.rb8
-rw-r--r--spec/ruby/core/string/shared/concat.rb1
-rw-r--r--spec/ruby/core/string/shared/dedup.rb6
-rw-r--r--spec/ruby/core/string/shared/each_codepoint_without_block.rb6
-rw-r--r--spec/ruby/core/string/shared/each_line.rb2
-rw-r--r--spec/ruby/core/string/shared/encode.rb185
-rw-r--r--spec/ruby/core/string/shared/eql.rb10
-rw-r--r--spec/ruby/core/string/shared/length.rb10
-rw-r--r--spec/ruby/core/string/shared/replace.rb1
-rw-r--r--spec/ruby/core/string/shared/slice.rb21
-rw-r--r--spec/ruby/core/string/shared/succ.rb3
-rw-r--r--spec/ruby/core/string/shared/to_sym.rb4
13 files changed, 235 insertions, 34 deletions
diff --git a/spec/ruby/core/string/shared/chars.rb b/spec/ruby/core/string/shared/chars.rb
index e9fdf89fd6..c730643cf4 100644
--- a/spec/ruby/core/string/shared/chars.rb
+++ b/spec/ruby/core/string/shared/chars.rb
@@ -21,12 +21,12 @@ describe :string_chars, shared: true do
end
it "returns characters in the same encoding as self" do
- "&%".force_encoding('Shift_JIS').send(@method).to_a.all? {|c| c.encoding.name.should == 'Shift_JIS'}
+ "&%".dup.force_encoding('Shift_JIS').send(@method).to_a.all? {|c| c.encoding.name.should == 'Shift_JIS'}
"&%".encode('BINARY').send(@method).to_a.all? {|c| c.encoding.should == Encoding::BINARY }
end
it "works with multibyte characters" do
- s = "\u{8987}".force_encoding("UTF-8")
+ s = "\u{8987}".dup.force_encoding("UTF-8")
s.bytesize.should == 3
s.send(@method).to_a.should == [s]
end
@@ -39,14 +39,14 @@ describe :string_chars, shared: true do
end
it "returns a different character if the String is transcoded" do
- s = "\u{20AC}".force_encoding('UTF-8')
- s.encode('UTF-8').send(@method).to_a.should == ["\u{20AC}".force_encoding('UTF-8')]
+ s = "\u{20AC}".dup.force_encoding('UTF-8')
+ s.encode('UTF-8').send(@method).to_a.should == ["\u{20AC}".dup.force_encoding('UTF-8')]
s.encode('iso-8859-15').send(@method).to_a.should == [[0xA4].pack('C').force_encoding('iso-8859-15')]
- s.encode('iso-8859-15').encode('UTF-8').send(@method).to_a.should == ["\u{20AC}".force_encoding('UTF-8')]
+ s.encode('iso-8859-15').encode('UTF-8').send(@method).to_a.should == ["\u{20AC}".dup.force_encoding('UTF-8')]
end
it "uses the String's encoding to determine what characters it contains" do
- s = "\u{24B62}"
+ s = +"\u{24B62}"
s.force_encoding('UTF-8').send(@method).to_a.should == [
s.force_encoding('UTF-8')
diff --git a/spec/ruby/core/string/shared/codepoints.rb b/spec/ruby/core/string/shared/codepoints.rb
index 0b2e078e0a..1c28ba3d5e 100644
--- a/spec/ruby/core/string/shared/codepoints.rb
+++ b/spec/ruby/core/string/shared/codepoints.rb
@@ -1,4 +1,4 @@
-# -*- encoding: binary -*-
+# encoding: binary
describe :string_codepoints, shared: true do
it "returns self" do
s = "foo"
@@ -7,7 +7,7 @@ describe :string_codepoints, shared: true do
end
it "raises an ArgumentError when self has an invalid encoding and a method is called on the returned Enumerator" do
- s = "\xDF".force_encoding(Encoding::UTF_8)
+ s = "\xDF".dup.force_encoding(Encoding::UTF_8)
s.valid_encoding?.should be_false
-> { s.send(@method).to_a }.should raise_error(ArgumentError)
end
@@ -21,7 +21,7 @@ describe :string_codepoints, shared: true do
end
it "raises an ArgumentError if self's encoding is invalid and a block is given" do
- s = "\xDF".force_encoding(Encoding::UTF_8)
+ s = "\xDF".dup.force_encoding(Encoding::UTF_8)
s.valid_encoding?.should be_false
-> { s.send(@method) { } }.should raise_error(ArgumentError)
end
@@ -49,7 +49,7 @@ describe :string_codepoints, shared: true do
it "round-trips to the original String using Integer#chr" do
s = "\u{13}\u{7711}\u{1010}"
- s2 = ""
+ s2 = +""
s.send(@method) {|n| s2 << n.chr(Encoding::UTF_8)}
s.should == s2
end
diff --git a/spec/ruby/core/string/shared/concat.rb b/spec/ruby/core/string/shared/concat.rb
index ee5ef2a98f..dded9a69e7 100644
--- a/spec/ruby/core/string/shared/concat.rb
+++ b/spec/ruby/core/string/shared/concat.rb
@@ -1,3 +1,4 @@
+# frozen_string_literal: false
describe :string_concat, shared: true do
it "concatenates the given argument to self and returns self" do
str = 'hello '
diff --git a/spec/ruby/core/string/shared/dedup.rb b/spec/ruby/core/string/shared/dedup.rb
index 893fd1e360..1ffd6aa0fd 100644
--- a/spec/ruby/core/string/shared/dedup.rb
+++ b/spec/ruby/core/string/shared/dedup.rb
@@ -1,3 +1,4 @@
+# frozen_string_literal: false
describe :string_dedup, shared: true do
it 'returns self if the String is frozen' do
input = 'foo'.freeze
@@ -47,9 +48,4 @@ describe :string_dedup, shared: true do
dynamic.send(@method).should_not equal("this string is frozen".send(@method).freeze)
dynamic.send(@method).should equal(dynamic)
end
-
- it "interns the provided string if it is frozen" do
- dynamic = "this string is unique and frozen #{rand}".freeze
- dynamic.send(@method).should equal(dynamic)
- end
end
diff --git a/spec/ruby/core/string/shared/each_codepoint_without_block.rb b/spec/ruby/core/string/shared/each_codepoint_without_block.rb
index 92b7f76032..c88e5c54c7 100644
--- a/spec/ruby/core/string/shared/each_codepoint_without_block.rb
+++ b/spec/ruby/core/string/shared/each_codepoint_without_block.rb
@@ -1,4 +1,4 @@
-# -*- encoding: binary -*-
+# encoding: binary
describe :string_each_codepoint_without_block, shared: true do
describe "when no block is given" do
it "returns an Enumerator" do
@@ -6,7 +6,7 @@ describe :string_each_codepoint_without_block, shared: true do
end
it "returns an Enumerator even when self has an invalid encoding" do
- s = "\xDF".force_encoding(Encoding::UTF_8)
+ s = "\xDF".dup.force_encoding(Encoding::UTF_8)
s.valid_encoding?.should be_false
s.send(@method).should be_an_instance_of(Enumerator)
end
@@ -23,7 +23,7 @@ describe :string_each_codepoint_without_block, shared: true do
end
it "should return the size of the string even when the string has an invalid encoding" do
- s = "\xDF".force_encoding(Encoding::UTF_8)
+ s = "\xDF".dup.force_encoding(Encoding::UTF_8)
s.valid_encoding?.should be_false
s.send(@method).size.should == 1
end
diff --git a/spec/ruby/core/string/shared/each_line.rb b/spec/ruby/core/string/shared/each_line.rb
index a14b4d7779..231a6d9d4f 100644
--- a/spec/ruby/core/string/shared/each_line.rb
+++ b/spec/ruby/core/string/shared/each_line.rb
@@ -106,7 +106,7 @@ describe :string_each_line, shared: true do
end
it "does not care if the string is modified while substituting" do
- str = "hello\nworld."
+ str = +"hello\nworld."
out = []
str.send(@method){|x| out << x; str[-1] = '!' }.should == "hello\nworld!"
out.should == ["hello\n", "world."]
diff --git a/spec/ruby/core/string/shared/encode.rb b/spec/ruby/core/string/shared/encode.rb
index a73de5b943..9466308886 100644
--- a/spec/ruby/core/string/shared/encode.rb
+++ b/spec/ruby/core/string/shared/encode.rb
@@ -1,4 +1,5 @@
# -*- encoding: utf-8 -*-
+# frozen_string_literal: false
describe :string_encode, shared: true do
describe "when passed no options" do
it "transcodes to Encoding.default_internal when set" do
@@ -193,6 +194,190 @@ describe :string_encode, shared: true do
end
end
+ describe "given the fallback option" do
+ context "given a hash" do
+ it "looks up the replacement value from the hash" do
+ encoded = "B\ufffd".encode(Encoding::US_ASCII, fallback: { "\ufffd" => "bar" })
+ encoded.should == "Bbar"
+ end
+
+ it "calls to_str on the returned value" do
+ obj = Object.new
+ obj.should_receive(:to_str).and_return("bar")
+ encoded = "B\ufffd".encode(Encoding::US_ASCII, fallback: { "\ufffd" => obj })
+ encoded.should == "Bbar"
+ end
+
+ it "does not call to_s on the returned value" do
+ obj = Object.new
+ obj.should_not_receive(:to_s)
+ -> {
+ "B\ufffd".encode(Encoding::US_ASCII, fallback: { "\ufffd" => obj })
+ }.should raise_error(TypeError, "no implicit conversion of Object into String")
+ end
+
+ it "raises an error if the key is not present in the hash" do
+ -> {
+ "B\ufffd".encode(Encoding::US_ASCII, fallback: { "foo" => "bar" })
+ }.should raise_error(Encoding::UndefinedConversionError, "U+FFFD from UTF-8 to US-ASCII")
+ end
+
+ it "raises an error if the value is itself invalid" do
+ -> {
+ "B\ufffd".encode(Encoding::US_ASCII, fallback: { "\ufffd" => "\uffee" })
+ }.should raise_error(ArgumentError, "too big fallback string")
+ end
+
+ it "uses the hash's default value if set" do
+ hash = {}
+ hash.default = "bar"
+ encoded = "B\ufffd".encode(Encoding::US_ASCII, fallback: hash)
+ encoded.should == "Bbar"
+ end
+
+ it "uses the result of calling default_proc if set" do
+ hash = {}
+ hash.default_proc = -> _, _ { "bar" }
+ encoded = "B\ufffd".encode(Encoding::US_ASCII, fallback: hash)
+ encoded.should == "Bbar"
+ end
+ end
+
+ context "given an object inheriting from Hash" do
+ before do
+ klass = Class.new(Hash)
+ @hash_like = klass.new
+ @hash_like["\ufffd"] = "bar"
+ end
+
+ it "looks up the replacement value from the object" do
+ encoded = "B\ufffd".encode(Encoding::US_ASCII, fallback: @hash_like)
+ encoded.should == "Bbar"
+ end
+ end
+
+ context "given an object responding to []" do
+ before do
+ klass = Class.new do
+ def [](c) = c.bytes.inspect
+ end
+ @hash_like = klass.new
+ end
+
+ it "calls [] on the object, passing the invalid character" do
+ encoded = "B\ufffd".encode(Encoding::US_ASCII, fallback: @hash_like)
+ encoded.should == "B[239, 191, 189]"
+ end
+ end
+
+ context "given an object not responding to []" do
+ before do
+ @non_hash_like = Object.new
+ end
+
+ it "raises an error" do
+ -> {
+ "B\ufffd".encode(Encoding::US_ASCII, fallback: @non_hash_like)
+ }.should raise_error(Encoding::UndefinedConversionError, "U+FFFD from UTF-8 to US-ASCII")
+ end
+ end
+
+ context "given a proc" do
+ it "calls the proc to get the replacement value, passing in the invalid character" do
+ encoded = "B\ufffd".encode(Encoding::US_ASCII, fallback: proc { |c| c.bytes.inspect })
+ encoded.should == "B[239, 191, 189]"
+ end
+
+ it "calls to_str on the returned value" do
+ obj = Object.new
+ obj.should_receive(:to_str).and_return("bar")
+ encoded = "B\ufffd".encode(Encoding::US_ASCII, fallback: proc { |c| obj })
+ encoded.should == "Bbar"
+ end
+
+ it "does not call to_s on the returned value" do
+ obj = Object.new
+ obj.should_not_receive(:to_s)
+ -> {
+ "B\ufffd".encode(Encoding::US_ASCII, fallback: proc { |c| obj })
+ }.should raise_error(TypeError, "no implicit conversion of Object into String")
+ end
+
+ it "raises an error if the returned value is itself invalid" do
+ -> {
+ "B\ufffd".encode(Encoding::US_ASCII, fallback: -> c { "\uffee" })
+ }.should raise_error(ArgumentError, "too big fallback string")
+ end
+ end
+
+ context "given a lambda" do
+ it "calls the lambda to get the replacement value, passing in the invalid character" do
+ encoded = "B\ufffd".encode(Encoding::US_ASCII, fallback: -> c { c.bytes.inspect })
+ encoded.should == "B[239, 191, 189]"
+ end
+
+ it "calls to_str on the returned value" do
+ obj = Object.new
+ obj.should_receive(:to_str).and_return("bar")
+ encoded = "B\ufffd".encode(Encoding::US_ASCII, fallback: -> c { obj })
+ encoded.should == "Bbar"
+ end
+
+ it "does not call to_s on the returned value" do
+ obj = Object.new
+ obj.should_not_receive(:to_s)
+ -> {
+ "B\ufffd".encode(Encoding::US_ASCII, fallback: -> c { obj })
+ }.should raise_error(TypeError, "no implicit conversion of Object into String")
+ end
+
+ it "raises an error if the returned value is itself invalid" do
+ -> {
+ "B\ufffd".encode(Encoding::US_ASCII, fallback: -> c { "\uffee" })
+ }.should raise_error(ArgumentError, "too big fallback string")
+ end
+ end
+
+ context "given a method" do
+ def replace(c) = c.bytes.inspect
+ def replace_bad(c) = "\uffee"
+
+ def replace_to_str(c)
+ obj = Object.new
+ obj.should_receive(:to_str).and_return("bar")
+ obj
+ end
+
+ def replace_to_s(c)
+ obj = Object.new
+ obj.should_not_receive(:to_s)
+ obj
+ end
+
+ it "calls the method to get the replacement value, passing in the invalid character" do
+ encoded = "B\ufffd".encode(Encoding::US_ASCII, fallback: method(:replace))
+ encoded.should == "B[239, 191, 189]"
+ end
+
+ it "calls to_str on the returned value" do
+ encoded = "B\ufffd".encode(Encoding::US_ASCII, fallback: method(:replace_to_str))
+ encoded.should == "Bbar"
+ end
+
+ it "does not call to_s on the returned value" do
+ -> {
+ "B\ufffd".encode(Encoding::US_ASCII, fallback: method(:replace_to_s))
+ }.should raise_error(TypeError, "no implicit conversion of Object into String")
+ end
+
+ it "raises an error if the returned value is itself invalid" do
+ -> {
+ "B\ufffd".encode(Encoding::US_ASCII, fallback: method(:replace_bad))
+ }.should raise_error(ArgumentError, "too big fallback string")
+ end
+ end
+ end
+
describe "given the xml: :text option" do
it "replaces all instances of '&' with '&amp;'" do
'& and &'.send(@method, "UTF-8", xml: :text).should == '&amp; and &amp;'
diff --git a/spec/ruby/core/string/shared/eql.rb b/spec/ruby/core/string/shared/eql.rb
index 6f268c929c..d5af337d53 100644
--- a/spec/ruby/core/string/shared/eql.rb
+++ b/spec/ruby/core/string/shared/eql.rb
@@ -1,4 +1,4 @@
-# -*- encoding: binary -*-
+# encoding: binary
require_relative '../../../spec_helper'
require_relative '../fixtures/classes'
@@ -13,15 +13,15 @@ describe :string_eql_value, shared: true do
end
it "ignores encoding difference of compatible string" do
- "hello".force_encoding("utf-8").send(@method, "hello".force_encoding("iso-8859-1")).should be_true
+ "hello".dup.force_encoding("utf-8").send(@method, "hello".dup.force_encoding("iso-8859-1")).should be_true
end
it "considers encoding difference of incompatible string" do
- "\xff".force_encoding("utf-8").send(@method, "\xff".force_encoding("iso-8859-1")).should be_false
+ "\xff".dup.force_encoding("utf-8").send(@method, "\xff".dup.force_encoding("iso-8859-1")).should be_false
end
it "considers encoding compatibility" do
- "abcd".force_encoding("utf-8").send(@method, "abcd".force_encoding("utf-32le")).should be_false
+ "abcd".dup.force_encoding("utf-8").send(@method, "abcd".dup.force_encoding("utf-32le")).should be_false
end
it "ignores subclass differences" do
@@ -33,6 +33,6 @@ describe :string_eql_value, shared: true do
end
it "returns true when comparing 2 empty strings but one is not ASCII-compatible" do
- "".send(@method, "".force_encoding('iso-2022-jp')).should == true
+ "".send(@method, "".dup.force_encoding('iso-2022-jp')).should == true
end
end
diff --git a/spec/ruby/core/string/shared/length.rb b/spec/ruby/core/string/shared/length.rb
index 94e5ec135b..ae572ba755 100644
--- a/spec/ruby/core/string/shared/length.rb
+++ b/spec/ruby/core/string/shared/length.rb
@@ -18,7 +18,7 @@ describe :string_length, shared: true do
end
it "returns the length of the new self after encoding is changed" do
- str = 'こにちわ'
+ str = +'こにちわ'
str.send(@method)
str.force_encoding('BINARY').send(@method).should == 12
@@ -44,12 +44,12 @@ describe :string_length, shared: true do
end
it "adds 1 (and not 2) for a incomplete surrogate in UTF-16" do
- "\x00\xd8".force_encoding("UTF-16LE").send(@method).should == 1
- "\xd8\x00".force_encoding("UTF-16BE").send(@method).should == 1
+ "\x00\xd8".dup.force_encoding("UTF-16LE").send(@method).should == 1
+ "\xd8\x00".dup.force_encoding("UTF-16BE").send(@method).should == 1
end
it "adds 1 for a broken sequence in UTF-32" do
- "\x04\x03\x02\x01".force_encoding("UTF-32LE").send(@method).should == 1
- "\x01\x02\x03\x04".force_encoding("UTF-32BE").send(@method).should == 1
+ "\x04\x03\x02\x01".dup.force_encoding("UTF-32LE").send(@method).should == 1
+ "\x01\x02\x03\x04".dup.force_encoding("UTF-32BE").send(@method).should == 1
end
end
diff --git a/spec/ruby/core/string/shared/replace.rb b/spec/ruby/core/string/shared/replace.rb
index a5108d9e7c..24dac0eb27 100644
--- a/spec/ruby/core/string/shared/replace.rb
+++ b/spec/ruby/core/string/shared/replace.rb
@@ -1,3 +1,4 @@
+# frozen_string_literal: false
describe :string_replace, shared: true do
it "returns self" do
a = "a"
diff --git a/spec/ruby/core/string/shared/slice.rb b/spec/ruby/core/string/shared/slice.rb
index 3ef4bc50d7..7b9b9f6a14 100644
--- a/spec/ruby/core/string/shared/slice.rb
+++ b/spec/ruby/core/string/shared/slice.rb
@@ -84,8 +84,8 @@ describe :string_slice_index_length, shared: true do
s = "hello there"
s.send(@method, 1, 9).encoding.should == s.encoding
- a = "hello".force_encoding("binary")
- b = " there".force_encoding("ISO-8859-1")
+ a = "hello".dup.force_encoding("binary")
+ b = " there".dup.force_encoding("ISO-8859-1")
c = (a + b).force_encoding(Encoding::US_ASCII)
c.send(@method, 0, 5).encoding.should == Encoding::US_ASCII
@@ -119,6 +119,18 @@ describe :string_slice_index_length, shared: true do
"hello there".send(@method, -4,-3).should == nil
end
+ platform_is pointer_size: 64 do
+ it "returns nil if the length is negative big value" do
+ "hello there".send(@method, 4, -(1 << 31)).should == nil
+
+ # by some reason length < -(1 << 31) on CI on Windows leads to
+ # 'RangeError: bignum too big to convert into `long'' error
+ platform_is_not :windows do
+ "hello there".send(@method, 4, -(1 << 63)).should == nil
+ end
+ end
+ end
+
it "calls to_int on the given index and the given length" do
"hello".send(@method, 0.5, 1).should == "h"
"hello".send(@method, 0.5, 2.5).should == "he"
@@ -152,6 +164,11 @@ describe :string_slice_index_length, shared: true do
-> { "hello".send(@method, 0, bignum_value) }.should raise_error(RangeError)
end
+ it "raises a RangeError if the index or length is too small" do
+ -> { "hello".send(@method, -bignum_value, 1) }.should raise_error(RangeError)
+ -> { "hello".send(@method, 0, -bignum_value) }.should raise_error(RangeError)
+ end
+
it "returns String instances" do
s = StringSpecs::MyString.new("hello")
s.send(@method, 0,0).should be_an_instance_of(String)
diff --git a/spec/ruby/core/string/shared/succ.rb b/spec/ruby/core/string/shared/succ.rb
index 24a729ce26..7c68345f10 100644
--- a/spec/ruby/core/string/shared/succ.rb
+++ b/spec/ruby/core/string/shared/succ.rb
@@ -1,4 +1,4 @@
-# -*- encoding: binary -*-
+# encoding: binary
describe :string_succ, shared: true do
it "returns an empty string for empty strings" do
"".send(@method).should == ""
@@ -73,6 +73,7 @@ end
describe :string_succ_bang, shared: true do
it "is equivalent to succ, but modifies self in place (still returns self)" do
["", "abcd", "THX1138"].each do |s|
+ s = +s
r = s.dup.send(@method)
s.send(@method).should equal(s)
s.should == r
diff --git a/spec/ruby/core/string/shared/to_sym.rb b/spec/ruby/core/string/shared/to_sym.rb
index 52d8314211..833eae100e 100644
--- a/spec/ruby/core/string/shared/to_sym.rb
+++ b/spec/ruby/core/string/shared/to_sym.rb
@@ -56,9 +56,9 @@ describe :string_to_sym, shared: true do
it "ignores existing symbols with different encoding" do
source = "fée"
- iso_symbol = source.force_encoding(Encoding::ISO_8859_1).send(@method)
+ iso_symbol = source.dup.force_encoding(Encoding::ISO_8859_1).send(@method)
iso_symbol.encoding.should == Encoding::ISO_8859_1
- binary_symbol = source.force_encoding(Encoding::BINARY).send(@method)
+ binary_symbol = source.dup.force_encoding(Encoding::BINARY).send(@method)
binary_symbol.encoding.should == Encoding::BINARY
end
generated by cgit v1.2.3 (git 2.25.1) at 2026-01-20 20:49:06 +0000