Add additional test cases for RegExp modifiers (#4321)

test/built-ins/RegExp/regexp-modifiers/add-dotAll-does-not-affect-alternatives-outside.js

@@ -0,0 +1,48 @@
+// Copyright 2024 Daniel Kwan. All rights reserved.
+// This code is governed by the BSD license found in the LICENSE file.
+
+/*---
+author: Daniel Kwan
+description: >
+  Adding dotAll (`s`) modifier should not affect alternatives outside.
+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]
+---*/
+
+var re1 = /a.a|b.b|(?s:c.c)|d.d|e.e/;
+assert(!re1.test("a\na"), "Alternative `a.a` should not match newline");
+assert(!re1.test("b\nb"), "Alternative `b.b` should not match newline");
+assert(re1.test("c\nc"), "Alternative `(?s:c.c)` should match newline in modified group");
+assert(!re1.test("d\nd"), "Alternative `d.d` should not match newline");
+assert(!re1.test("e\ne"), "Alternative `e.e` should not match newline");
+
+var re2 = /(a.a)|(?:b.b)|(?s:c.c)|(?:d.d)|(e.e)/;
+assert(!re2.test("a\na"), "Alternative `(a.a)` should not match newline");
+assert(!re2.test("b\nb"), "Alternative `(?:b.b)` should not match newline");
+assert(re2.test("c\nc"), "Alternative `(?s:c.c)` should match newline in modified group");
+assert(!re2.test("d\nd"), "Alternative `(?:d.d)` should not match newline");
+assert(!re2.test("e\ne"), "Alternative `(e.e)` should not match newline");

test/built-ins/RegExp/regexp-modifiers/add-dotAll.js

@@ -100,3 +100,17 @@ assert(re4.test("\u2028"), "Pattern character '.' should match line terminators
 assert(re4.test("\u2029"), "Pattern character '.' should match line terminators in modified group");
 assert(re4.test("\uD800"), "Pattern character '.' should match non-line terminators in modified group");
 assert(re4.test("\uDFFF"), "Pattern character '.' should match non-line terminators in modified group");
+
+var re5 = /a.(?s:b.b).c/;
+assert(re5.test("a,b,b,c"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re5.test("a,b\nb,c"), "Pattern character '.' should match line terminators in modified group");
+assert(!re5.test("a\nb\nb,c"), "Pattern character '.' should not match line terminators outside modified group");
+assert(!re5.test("a,b\nb\nc"), "Pattern character '.' should not match line terminators outside modified group");
+assert(!re5.test("a\nb\nb\nc"), "Pattern character '.' should not match line terminators outside modified group");
+
+var re6 = new RegExp("a.(?s:b.b).c");
+assert(re6.test("a,b,b,c"), "Pattern character '.' should match non-line terminators in modified group");
+assert(re6.test("a,b\nb,c"), "Pattern character '.' should match line terminators in modified group");
+assert(!re6.test("a\nb\nb,c"), "Pattern character '.' should not match line terminators outside modified group");
+assert(!re6.test("a,b\nb\nc"), "Pattern character '.' should not match line terminators outside modified group");
+assert(!re6.test("a\nb\nb\nc"), "Pattern character '.' should not match line terminators outside modified group");

test/built-ins/RegExp/regexp-modifiers/add-ignoreCase-does-not-affect-alternatives-outside.js

@@ -0,0 +1,48 @@
+// Copyright 2024 Daniel Kwan. All rights reserved.
+// This code is governed by the BSD license found in the LICENSE file.
+
+/*---
+author: Daniel Kwan
+description: >
+  Adding ignoreCase (`i`) modifier should not affect alternatives outside.
+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]
+---*/
+
+var re1 = /a|b|(?i:c)|d|e/;
+assert(!re1.test("A"), "Alternative `a` should not match `A`");
+assert(!re1.test("B"), "Alternative `b` should not match `B`");
+assert(re1.test("C"), "Alternative `(?i:c)` should match `C` in modified group");
+assert(!re1.test("D"), "Alternative `d` should not match `D`");
+assert(!re1.test("E"), "Alternative `e` should not match `E`");
+
+var re2 = /(a)|(?:b)|(?i:c)|(?:d)|(e)/;
+assert(!re2.test("A"), "Alternative `(a)` should not match `A`");
+assert(!re2.test("B"), "Alternative `(?:b)` should not match `B`");
+assert(re2.test("C"), "Alternative `(?i:c)` should match `C` in modified group");
+assert(!re2.test("D"), "Alternative `(?:d)` should not match `D`");
+assert(!re2.test("E"), "Alternative `(e)` should not match `E`");

test/built-ins/RegExp/regexp-modifiers/add-ignoreCase.js

@@ -42,3 +42,27 @@ var re2 = new RegExp("(?i:a)b");
 assert(!re2.test("AB"), "b should not match B in AB");
 assert(re2.test("Ab"), "a should match A in AB");
 assert(re2.test("ab"), "should match AB");
+
+var re3 = /b(?i:a)/;
+assert(!re3.test("BA"), "b should not match B in BA");
+assert(re3.test("bA"), "a should match A in BA");
+assert(re3.test("ba"), "should match BA");
+
+var re4 = new RegExp("b(?i:a)");
+assert(!re4.test("BA"), "b should not match B in BA");
+assert(re4.test("bA"), "a should match A in BA");
+assert(re4.test("ba"), "should match BA");
+
+var re5 = /a(?i:b)c/;
+assert(re5.test("abc"), "b should match b in abc");
+assert(re5.test("aBc"), "B should match b in abc");
+assert(!re5.test("ABc"), "A should not match a in abc");
+assert(!re5.test("aBC"), "C should not match c in abc");
+assert(!re5.test("ABC"), "should not match abc");
+
+var re6 = new RegExp("a(?i:b)c");
+assert(re6.test("abc"), "b should match b in abc");
+assert(re6.test("aBc"), "B should match b in abc");
+assert(!re6.test("ABc"), "A should not match a in abc");
+assert(!re6.test("aBC"), "C should not match c in abc");
+assert(!re6.test("ABC"), "should not match abc");

test/built-ins/RegExp/regexp-modifiers/add-multiline-does-not-affect-alternatives-outside.js

@@ -0,0 +1,48 @@
+// Copyright 2024 Daniel Kwan. All rights reserved.
+// This code is governed by the BSD license found in the LICENSE file.
+
+/*---
+author: Daniel Kwan
+description: >
+  Adding multiline (`m`) modifier should not affect alternatives outside.
+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]
+---*/
+
+var re1 = /^a$|^b$|(?m:^c$)|^d$|^e$/;
+assert(!re1.test("\na\n"), "Alternative `^a$` should not match newline");
+assert(!re1.test("\nb\n"), "Alternative `^b$` should not match newline");
+assert(re1.test("\nc\n"), "Alternative `(?m:^c$)` should match newline in modified group");
+assert(!re1.test("\nd\n"), "Alternative `^d$` should not match newline");
+assert(!re1.test("\ne\n"), "Alternative `^e$` should not match newline");
+
+var re2 = /(^a$)|(?:^b$)|(?m:^c$)|(?:^d$)|(^e$)/;
+assert(!re2.test("\na\n"), "Alternative `(^a$)` should not match newline");
+assert(!re2.test("\nb\n"), "Alternative `(?:^b$)` should not match newline");
+assert(re2.test("\nc\n"), "Alternative `(?m:^c$)` should match newline in modified group");
+assert(!re2.test("\nd\n"), "Alternative `(?:^d$)` should not match newline");
+assert(!re2.test("\ne\n"), "Alternative `(^e$)` should not match newline");

test/built-ins/RegExp/regexp-modifiers/add-multiline.js

@@ -44,3 +44,15 @@ assert(re3.test("es\ns"), "$ should match newline in modified group");
 
 var re4 = new RegExp("(?m-:es$)");
 assert(re4.test("es\ns"), "$ should match newline in modified group");
+
+var re5 = /^a\n(?m:^b$)\nc$/;
+assert(re5.test("a\nb\nc"), "^ and $ should match newline in modified group");
+assert(!re5.test("\na\nb\nc"), "^ should not match newline outside modified group");
+assert(!re5.test("a\nb\nc\n"), "$ should not match newline outside modified group");
+assert(!re5.test("\na\nb\nc\n"), "^ and $ should not match newline outside modified group");
+
+var re6 = new RegExp("^a\\n(?m:^b$)\\nc$");
+assert(re6.test("a\nb\nc"), "^ and $ should match newline in modified group");
+assert(!re6.test("\na\nb\nc"), "^ should not match newline outside modified group");
+assert(!re6.test("a\nb\nc\n"), "$ should not match newline outside modified group");
+assert(!re6.test("\na\nb\nc\n"), "^ and $ should not match newline outside modified group");

test/built-ins/RegExp/regexp-modifiers/nesting-dotAll-does-not-affect-alternatives-outside.js

@@ -0,0 +1,62 @@
+// Copyright 2024 Daniel Kwan. All rights reserved.
+// This code is governed by the BSD license found in the LICENSE file.
+
+/*---
+author: Daniel Kwan
+description: >
+  Nesting dotAll (`s`) modifier should not affect alternatives outside.
+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.
+
+  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 = /a.a|(?-s:b.b|(?s:c.c)|d.d|(?-s:e.e)|f.f)|g.g|(?s:h.h)|k.k/s;
+assert(re1.test("a\na"), "`a.a` should match newline");
+assert(!re1.test("b\nb"), "`b.b` should not match newline");
+assert(re1.test("c\nc"), "`c.c` should match newline");
+assert(!re1.test("d\nd"), "`d.d` should not match newline");
+assert(!re1.test("e\ne"), "`e.e` should not match newline");
+assert(!re1.test("f\nf"), "`f.f` should not match newline");
+assert(re1.test("g\ng"), "`g.g` should match newline");
+assert(re1.test("h\nh"), "`h.h` should match newline");
+assert(re1.test("k\nk"), "`k.k` should match newline");
+
+var re2 = /a.a|(?s:b.b|(?-s:c.c)|d.d|(?s:e.e)|f.f)|g.g|(?-s:h.h)|k.k/;
+assert(!re2.test("a\na"), "`a.a` should not match newline");
+assert(re2.test("b\nb"), "`b.b` should match newline");
+assert(!re2.test("c\nc"), "`c.c` should not match newline");
+assert(re2.test("d\nd"), "`d.d` should match newline");
+assert(re2.test("e\ne"), "`e.e` should match newline");
+assert(re2.test("f\nf"), "`f.f` should match newline");
+assert(!re2.test("g\ng"), "`g.g` should not match newline");
+assert(!re2.test("h\nh"), "`h.h` should not match newline");
+assert(!re2.test("k\nk"), "`k.k` should not match newline");

test/built-ins/RegExp/regexp-modifiers/nesting-ignoreCase-does-not-affect-alternatives-outside.js

@@ -0,0 +1,62 @@
+// Copyright 2024 Daniel Kwan. All rights reserved.
+// This code is governed by the BSD license found in the LICENSE file.
+
+/*---
+author: Daniel Kwan
+description: >
+  Nesting ignoreCase (`i`) modifier should not affect alternatives outside.
+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.
+
+  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 = /a|(?-i:b|(?i:c)|d|(?-i:e)|f)|g|(?i:h)|k/i;
+assert(re1.test("A"), "`a` should match `A`");
+assert(!re1.test("B"), "`b` should not match `B`");
+assert(re1.test("C"), "`c` should match `C`");
+assert(!re1.test("D"), "`d` should not match `D`");
+assert(!re1.test("E"), "`e` should not match `E`");
+assert(!re1.test("F"), "`f` should not match `F`");
+assert(re1.test("G"), "`g` should match `G`");
+assert(re1.test("H"), "`h` should match `H`");
+assert(re1.test("K"), "`k` should match `K`");
+
+var re2 = /a|(?i:b|(?-i:c)|d|(?i:e)|f)|g|(?-i:h)|k/;
+assert(!re2.test("A"), "`a` should not match `A`");
+assert(re2.test("B"), "`b` should match `B`");
+assert(!re2.test("C"), "`c` should not match `C`");
+assert(re2.test("D"), "`d` should match `D`");
+assert(re2.test("E"), "`e` should match `E`");
+assert(re2.test("F"), "`f` should match `F`");
+assert(!re2.test("G"), "`g` should not match `G`");
+assert(!re2.test("H"), "`h` should not match `H`");
+assert(!re2.test("K"), "`k` should not match `K`");