Add tests for RegExp modifiers

tc39 · Nov 15, 2023 · 060972b · 060972b
1 parent 99ac701
commit 060972b
Show file tree

Hide file tree

Showing 39 changed files with 1,333 additions and 0 deletions.
diff --git a/features.txt b/features.txt
@@ -101,6 +101,10 @@ iterator-helpers
 # https://github.com/tc39/proposal-promise-with-resolvers
 promise-with-resolvers
 
+# RegExp Modifiers
+# https://github.com/tc39/proposal-regexp-modifiers
+regexp-modifiers
+
 ## Standard language features
 #
 # Language features that have been included in a published version of the

diff --git a/test/built-ins/RegExp/regexp-modifiers/add-dotAll.js b/test/built-ins/RegExp/regexp-modifiers/add-dotAll.js
@@ -0,0 +1,72 @@
+// Copyright 2023 Ron Buckton. All rights reserved.
+// This code is governed by the BSD license found in the LICENSE file.
+
+/*---
+author: Ron Buckton
+description: >
+  Modifiers can be set via `(?ims:)`
+info: |
+  Runtime Semantics: CompileAtom
+  The syntax-directed operation CompileAtom takes arguments direction (forward or backward) and modifiers (a Modifiers Record) and returns a Matcher.
+
+  Atom :: `(` `?` RegularExpressionFlags `:` Disjunction `)`
+    1. Let addModifiers be the source text matched by RegularExpressionFlags.
+    2. Let removeModifiers be the empty String.
+    3. Let newModifiers be UpdateModifiers(modifiers, CodePointsToString(addModifiers), removeModifiers).
+    4. Return CompileSubpattern of Disjunction with arguments direction and newModifiers.
+
+  UpdateModifiers ( modifiers, add, remove )
+  The abstract operation UpdateModifiers takes arguments modifiers (a Modifiers Record), add (a String), and remove (a String) and returns a Modifiers. It performs the following steps when called:
+
+  1. Let dotAll be modifiers.[[DotAll]].
+  2. Let ignoreCase be modifiers.[[IgnoreCase]].
+  3. Let multiline be modifiers.[[Multiline]].
+  4. If add contains "s", set dotAll to true.
+  5. If add contains "i", set ignoreCase to true.
+  6. If add contains "m", set multiline to true.
+  7. If remove contains "s", set dotAll to false.
+  8. If remove contains "i", set ignoreCase to false.
+  9. If remove contains "m", set multiline to false.
+  10. Return the Modifiers Record { [[DotAll]]: dotAll, [[IgnoreCase]]: ignoreCase, [[Multiline]]: multiline }.
+
+esid: sec-compileatom
+features: [regexp-modifiers]
+---*/
+
+// dot-all (from RegExp literal)
+var re1 = /(?s:^.$)/;
+assert(!re1.dotAll, "RegExp instance dotAll flag should not be set");
+assert(re1.test("a"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re1.test("3"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re1.test("π"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re1.test("\u2027"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re1.test("\u0085"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re1.test("\v"), "Pattern character '.' should match mon-line terminators in modified group");
+assert(re1.test("\f"), "Pattern character '.' should match mon-line terminators in modified group");
+assert(re1.test("\u180E"), "Pattern character '.' should match non-line terminators in modified group");
+assert(!re1.test("\u{10300}"), "Supplementary plane not matched by a single .");
+assert(re1.test("\n"), "Pattern character '.' should match line terminators in modified group");
+assert(re1.test("\r"), "Pattern character '.' should match line terminators in modified group");
+assert(re1.test("\u2028"), "Pattern character '.' should match line terminators in modified group");
+assert(re1.test("\u2029"), "Pattern character '.' should match line terminators in modified group");
+assert(re1.test("\uD800"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re1.test("\uDFFF"), "Pattern character '.' should match non-line terminators in modified group");
+
+// dot-all (from RegExp constructor)
+var re2 = new RegExp("(?s:^.$)");
+assert(!re2.dotAll, "RegExp instance dotAll flag should not be set");
+assert(re2.test("a"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re2.test("3"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re2.test("π"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re2.test("\u2027"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re2.test("\u0085"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re2.test("\v"), "Pattern character '.' should match mon-line terminators in modified group");
+assert(re2.test("\f"), "Pattern character '.' should match mon-line terminators in modified group");
+assert(re2.test("\u180E"), "Pattern character '.' should match non-line terminators in modified group");
+assert(!re2.test("\u{10300}"), "Supplementary plane not matched by a single .");
+assert(re2.test("\n"), "Pattern character '.' should match line terminators in modified group");
+assert(re2.test("\r"), "Pattern character '.' should match line terminators in modified group");
+assert(re2.test("\u2028"), "Pattern character '.' should match line terminators in modified group");
+assert(re2.test("\u2029"), "Pattern character '.' should match line terminators in modified group");
+assert(re2.test("\uD800"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re2.test("\uDFFF"), "Pattern character '.' should match non-line terminators in modified group");
diff --git a/test/built-ins/RegExp/regexp-modifiers/add-ignoreCase.js b/test/built-ins/RegExp/regexp-modifiers/add-ignoreCase.js
@@ -0,0 +1,48 @@
+// Copyright 2023 Ron Buckton. All rights reserved.
+// This code is governed by the BSD license found in the LICENSE file.
+
+/*---
+author: Ron Buckton
+description: >
+  Modifiers can be set via `(?ims:)`
+info: |
+  Runtime Semantics: CompileAtom
+  The syntax-directed operation CompileAtom takes arguments direction (forward or backward) and modifiers (a Modifiers Record) and returns a Matcher.
+
+  Atom :: `(` `?` RegularExpressionFlags `:` Disjunction `)`
+    1. Let addModifiers be the source text matched by RegularExpressionFlags.
+    2. Let removeModifiers be the empty String.
+    3. Let newModifiers be UpdateModifiers(modifiers, CodePointsToString(addModifiers), removeModifiers).
+    4. Return CompileSubpattern of Disjunction with arguments direction and newModifiers.
+
+  UpdateModifiers ( modifiers, add, remove )
+  The abstract operation UpdateModifiers takes arguments modifiers (a Modifiers Record), add (a String), and remove (a String) and returns a Modifiers. It performs the following steps when called:
+
+  1. Let dotAll be modifiers.[[DotAll]].
+  2. Let ignoreCase be modifiers.[[IgnoreCase]].
+  3. Let multiline be modifiers.[[Multiline]].
+  4. If add contains "s", set dotAll to true.
+  5. If add contains "i", set ignoreCase to true.
+  6. If add contains "m", set multiline to true.
+  7. If remove contains "s", set dotAll to false.
+  8. If remove contains "i", set ignoreCase to false.
+  9. If remove contains "m", set multiline to false.
+  10. Return the Modifiers Record { [[DotAll]]: dotAll, [[IgnoreCase]]: ignoreCase, [[Multiline]]: multiline }.
+
+esid: sec-compileatom
+features: [regexp-modifiers]
+---*/
+
+// ignorecase (from RegExp literal)
+var re1 = /(?i:fo)o/;
+assert(!re1.ignoreCase, "RegExp instance ignoreCase flag should not be set");
+assert(!re1.test("FOO"), "Pattern should not match as atom does not ignore case");
+assert(re1.test("FOo"), "Pattern should ignore case in modified group");
+assert(re1.test("foo"), "Pattern should ignore case in modified group");
+
+// ignorecase (from RegExp constructor)
+var re2 = new RegExp("(?i:fo)o");
+assert(!re2.ignoreCase, "RegExp instance ignoreCase flag should not be set");
+assert(!re2.test("FOO"), "Pattern should not match as atom does not ignore case");
+assert(re2.test("FOo"), "Pattern should ignore case in modified group");
+assert(re2.test("foo"), "Pattern should ignore case in modified group");
diff --git a/test/built-ins/RegExp/regexp-modifiers/add-multiline.js b/test/built-ins/RegExp/regexp-modifiers/add-multiline.js
@@ -0,0 +1,44 @@
+// Copyright 2023 Ron Buckton. All rights reserved.
+// This code is governed by the BSD license found in the LICENSE file.
+
+/*---
+author: Ron Buckton
+description: >
+  Modifiers can be set via `(?ims:)`
+info: |
+  Runtime Semantics: CompileAtom
+  The syntax-directed operation CompileAtom takes arguments direction (forward or backward) and modifiers (a Modifiers Record) and returns a Matcher.
+
+  Atom :: `(` `?` RegularExpressionFlags `:` Disjunction `)`
+    1. Let addModifiers be the source text matched by RegularExpressionFlags.
+    2. Let removeModifiers be the empty String.
+    3. Let newModifiers be UpdateModifiers(modifiers, CodePointsToString(addModifiers), removeModifiers).
+    4. Return CompileSubpattern of Disjunction with arguments direction and newModifiers.
+
+  UpdateModifiers ( modifiers, add, remove )
+  The abstract operation UpdateModifiers takes arguments modifiers (a Modifiers Record), add (a String), and remove (a String) and returns a Modifiers. It performs the following steps when called:
+
+  1. Let dotAll be modifiers.[[DotAll]].
+  2. Let ignoreCase be modifiers.[[IgnoreCase]].
+  3. Let multiline be modifiers.[[Multiline]].
+  4. If add contains "s", set dotAll to true.
+  5. If add contains "i", set ignoreCase to true.
+  6. If add contains "m", set multiline to true.
+  7. If remove contains "s", set dotAll to false.
+  8. If remove contains "i", set ignoreCase to false.
+  9. If remove contains "m", set multiline to false.
+  10. Return the Modifiers Record { [[DotAll]]: dotAll, [[IgnoreCase]]: ignoreCase, [[Multiline]]: multiline }.
+
+esid: sec-compileatom
+features: [regexp-modifiers]
+---*/
+
+// multiline (from RegExp literal)
+var re1 = /(?m:es$)/;
+assert(!re1.multiline, "RegExp instance multiline flag should not be set");
+assert(re1.test("es\ns"), "$ should match newline in modified group");
+
+// multiline (from RegExp constructor)
+var re2 = new RegExp("(?m:es$)");
+assert(!re2.multiline, "RegExp instance multiline flag should not be set");
+assert(re2.test("es\ns"), "$ should match newline in modified group");
diff --git a/test/built-ins/RegExp/regexp-modifiers/add-remove-modifiers.js b/test/built-ins/RegExp/regexp-modifiers/add-remove-modifiers.js
@@ -0,0 +1,46 @@
+// Copyright 2023 Ron Buckton. All rights reserved.
+// This code is governed by the BSD license found in the LICENSE file.
+
+/*---
+author: Ron Buckton
+description: >
+  Modifiers can be both added and removed via `(?ims-ims:)`.
+info: |
+  Runtime Semantics: CompileAtom
+  The syntax-directed operation CompileAtom takes arguments direction (forward or backward) and modifiers (a Modifiers Record) and returns a Matcher.
+
+  Atom :: `(` `?` RegularExpressionFlags `-` RegularExpressionFlags `:` Disjunction `)`
+    1. Let addModifiers be the source text matched by the first RegularExpressionFlags.
+    2. Let removeModifiers be the source text matched by the second RegularExpressionFlags.
+    3. Let newModifiers be UpdateModifiers(modifiers, CodePointsToString(addModifiers), CodePointsToString(removeModifiers)).
+    4. Return CompileSubpattern of Disjunction with arguments direction and newModifiers.
+
+  UpdateModifiers ( modifiers, add, remove )
+  The abstract operation UpdateModifiers takes arguments modifiers (a Modifiers Record), add (a String), and remove (a String) and returns a Modifiers. It performs the following steps when called:
+
+  1. Let dotAll be modifiers.[[DotAll]].
+  2. Let ignoreCase be modifiers.[[IgnoreCase]].
+  3. Let multiline be modifiers.[[Multiline]].
+  4. If add contains "s", set dotAll to true.
+  5. If add contains "i", set ignoreCase to true.
+  6. If add contains "m", set multiline to true.
+  7. If remove contains "s", set dotAll to false.
+  8. If remove contains "i", set ignoreCase to false.
+  9. If remove contains "m", set multiline to false.
+  10. Return the Modifiers Record { [[DotAll]]: dotAll, [[IgnoreCase]]: ignoreCase, [[Multiline]]: multiline }.
+
+esid: sec-compileatom
+features: [regexp-modifiers]
+---*/
+
+var re1 = /(?m-i:^a$)/i;
+assert(re1.ignoreCase, "RegExp instance ignoreCase flag should be set");
+assert(!re1.multiline, "RegExp instance multiline flag should not be set");
+assert(!re1.test("A\n"), "Should not match 'A\\n'");
+assert(re1.test("a\n"), "Should match 'a\\n'");
+
+var re2 = new RegExp("(?m-i:^a$)", "i");
+assert(re2.ignoreCase, "RegExp instance ignoreCase flag should be set");
+assert(!re2.multiline, "RegExp instance multiline flag should not be set");
+assert(!re2.test("A\n"), "Should not match 'A\\n'");
+assert(re2.test("a\n"), "Should match 'a\\n'");
diff --git a/test/built-ins/RegExp/regexp-modifiers/nested-add-remove-modifiers.js b/test/built-ins/RegExp/regexp-modifiers/nested-add-remove-modifiers.js
@@ -0,0 +1,46 @@
+// Copyright 2023 Ron Buckton. All rights reserved.
+// This code is governed by the BSD license found in the LICENSE file.
+
+/*---
+author: Ron Buckton
+description: >
+  Modifiers can be nested.
+info: |
+  Runtime Semantics: CompileAtom
+  The syntax-directed operation CompileAtom takes arguments direction (forward or backward) and modifiers (a Modifiers Record) and returns a Matcher.
+
+  Atom :: `(` `?` RegularExpressionFlags `-` RegularExpressionFlags `:` Disjunction `)`
+    1. Let addModifiers be the source text matched by the first RegularExpressionFlags.
+    2. Let removeModifiers be the source text matched by the second RegularExpressionFlags.
+    3. Let newModifiers be UpdateModifiers(modifiers, CodePointsToString(addModifiers), CodePointsToString(removeModifiers)).
+    4. Return CompileSubpattern of Disjunction with arguments direction and newModifiers.
+
+  UpdateModifiers ( modifiers, add, remove )
+  The abstract operation UpdateModifiers takes arguments modifiers (a Modifiers Record), add (a String), and remove (a String) and returns a Modifiers. It performs the following steps when called:
+
+  1. Let dotAll be modifiers.[[DotAll]].
+  2. Let ignoreCase be modifiers.[[IgnoreCase]].
+  3. Let multiline be modifiers.[[Multiline]].
+  4. If add contains "s", set dotAll to true.
+  5. If add contains "i", set ignoreCase to true.
+  6. If add contains "m", set multiline to true.
+  7. If remove contains "s", set dotAll to false.
+  8. If remove contains "i", set ignoreCase to false.
+  9. If remove contains "m", set multiline to false.
+  10. Return the Modifiers Record { [[DotAll]]: dotAll, [[IgnoreCase]]: ignoreCase, [[Multiline]]: multiline }.
+
+esid: sec-compileatom
+features: [regexp-modifiers]
+---*/
+
+var re1 = /(?m:^(?-i:a)$)/i;
+assert(re1.ignoreCase, "RegExp instance ignoreCase flag should be set");
+assert(!re1.multiline, "RegExp instance multiline flag should not be set");
+assert(!re1.test("A\n"), "Should not match 'A\\n'");
+assert(re1.test("a\n"), "Should match 'a\\n'");
+
+var re2 = new RegExp("(?m:^(?-i:a)$)", "i");
+assert(re2.ignoreCase, "RegExp instance ignoreCase flag should be set");
+assert(!re2.multiline, "RegExp instance multiline flag should not be set");
+assert(!re2.test("A\n"), "Should not match 'A\\n'");
+assert(re2.test("a\n"), "Should match 'a\\n'");
diff --git a/test/built-ins/RegExp/regexp-modifiers/remove-dotAll.js b/test/built-ins/RegExp/regexp-modifiers/remove-dotAll.js
@@ -0,0 +1,72 @@
+// Copyright 2023 Ron Buckton. All rights reserved.
+// This code is governed by the BSD license found in the LICENSE file.
+
+/*---
+author: Ron Buckton
+description: >
+  Modifiers can be removed via `(?-ims:)`
+info: |
+  Runtime Semantics: CompileAtom
+  The syntax-directed operation CompileAtom takes arguments direction (forward or backward) and modifiers (a Modifiers Record) and returns a Matcher.
+
+  Atom :: `(` `?` RegularExpressionFlags `-` RegularExpressionFlags `:` Disjunction `)`
+    1. Let addModifiers be the source text matched by the first RegularExpressionFlags.
+    2. Let removeModifiers be the source text matched by the second RegularExpressionFlags.
+    3. Let newModifiers be UpdateModifiers(modifiers, CodePointsToString(addModifiers), CodePointsToString(removeModifiers)).
+    4. Return CompileSubpattern of Disjunction with arguments direction and newModifiers.
+
+  UpdateModifiers ( modifiers, add, remove )
+  The abstract operation UpdateModifiers takes arguments modifiers (a Modifiers Record), add (a String), and remove (a String) and returns a Modifiers. It performs the following steps when called:
+
+  1. Let dotAll be modifiers.[[DotAll]].
+  2. Let ignoreCase be modifiers.[[IgnoreCase]].
+  3. Let multiline be modifiers.[[Multiline]].
+  4. If add contains "s", set dotAll to true.
+  5. If add contains "i", set ignoreCase to true.
+  6. If add contains "m", set multiline to true.
+  7. If remove contains "s", set dotAll to false.
+  8. If remove contains "i", set ignoreCase to false.
+  9. If remove contains "m", set multiline to false.
+  10. Return the Modifiers Record { [[DotAll]]: dotAll, [[IgnoreCase]]: ignoreCase, [[Multiline]]: multiline }.
+
+esid: sec-compileatom
+features: [regexp-modifiers]
+---*/
+
+// dot-all (from RegExp literal)
+var re1 = /(?-s:^.$)/s;
+assert(re1.dotAll, "RegExp instance dotAll flag should be set");
+assert(re1.test("a"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re1.test("3"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re1.test("π"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re1.test("\u2027"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re1.test("\u0085"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re1.test("\v"), "Pattern character '.' should match mon-line terminators in modified group");
+assert(re1.test("\f"), "Pattern character '.' should match mon-line terminators in modified group");
+assert(re1.test("\u180E"), "Pattern character '.' should match non-line terminators in modified group");
+assert(!re1.test("\u{10300}"), "Supplementary plane not matched by a single .");
+assert(!re1.test("\n"), "Pattern character '.' should not match '\\n' in modified group");
+assert(!re1.test("\r"), "Pattern character '.' should not match '\\r' in modified group");
+assert(!re1.test("\u2028"), "Pattern character '.' should not match '\\u2028' in modified group");
+assert(!re1.test("\u2029"), "Pattern character '.' should not match '\\u2029' in modified group");
+assert(re1.test("\uD800"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re1.test("\uDFFF"), "Pattern character '.' should match non-line terminators in modified group");
+
+// dot-all (from RegExp constructor)
+var re2 = new RegExp("(?-s:^.$)", "s");
+assert(re2.dotAll, "RegExp instance dotAll flag should be set");
+assert(re2.test("a"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re2.test("3"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re2.test("π"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re2.test("\u2027"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re2.test("\u0085"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re2.test("\v"), "Pattern character '.' should match mon-line terminators in modified group");
+assert(re2.test("\f"), "Pattern character '.' should match mon-line terminators in modified group");
+assert(re2.test("\u180E"), "Pattern character '.' should match non-line terminators in modified group");
+assert(!re2.test("\u{10300}"), "Supplementary plane not matched by a single .");
+assert(!re2.test("\n"), "Pattern character '.' should not match '\\n' in modified group");
+assert(!re2.test("\r"), "Pattern character '.' should not match '\\r' in modified group");
+assert(!re2.test("\u2028"), "Pattern character '.' should not match '\\u2028' in modified group");
+assert(!re2.test("\u2029"), "Pattern character '.' should not match '\\u2029' in modified group");
+assert(re2.test("\uD800"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re2.test("\uDFFF"), "Pattern character '.' should match non-line terminators in modified group");