poe-code 14.0.73 → 14.0.75

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (47) hide show
  1. package/dist/index.js +2732 -1188
  2. package/dist/index.js.map +4 -4
  3. package/dist/metafile.json +1 -1
  4. package/package.json +1 -1
  5. package/packages/safe-js/dist/chunks/{chunk-I2Z37MKZ.js → chunk-BAJI3K4B.js} +2 -2
  6. package/packages/safe-js/dist/chunks/{chunk-XKCTB5FR.js → chunk-EOTY4PCT.js} +2902 -1425
  7. package/packages/safe-js/dist/chunks/chunk-EOTY4PCT.js.map +7 -0
  8. package/packages/safe-js/dist/cli.js +2 -2
  9. package/packages/safe-js/dist/core.js +1 -1
  10. package/packages/safe-js/dist/graph-depth.js +3 -0
  11. package/packages/safe-js/dist/index.js +2 -2
  12. package/packages/safe-js/dist/interp/cancel.js +3 -0
  13. package/packages/safe-js/dist/interp/globals/misc.js +2 -1
  14. package/packages/safe-js/dist/interp/globals/object.js +3 -0
  15. package/packages/safe-js/dist/interp/globals/regex.js +196 -9
  16. package/packages/safe-js/dist/interp/interpreter.js +14 -7
  17. package/packages/safe-js/dist/interp/iteration.js +9 -1
  18. package/packages/safe-js/dist/interp/methods/regex-split.d.ts +3 -0
  19. package/packages/safe-js/dist/interp/methods/regex-split.js +102 -0
  20. package/packages/safe-js/dist/interp/methods/regex.d.ts +8 -4
  21. package/packages/safe-js/dist/interp/methods/regex.js +141 -24
  22. package/packages/safe-js/dist/interp/methods/regexp-iterator.d.ts +12 -0
  23. package/packages/safe-js/dist/interp/methods/regexp-iterator.js +101 -0
  24. package/packages/safe-js/dist/interp/methods/string.d.ts +2 -0
  25. package/packages/safe-js/dist/interp/methods/string.js +415 -121
  26. package/packages/safe-js/dist/interp/object-model.d.ts +3 -0
  27. package/packages/safe-js/dist/interp/object-model.js +43 -10
  28. package/packages/safe-js/dist/interp/regex/compile-guard.js +26 -4
  29. package/packages/safe-js/dist/interp/regex/engine.d.ts +4 -0
  30. package/packages/safe-js/dist/interp/regex/engine.js +230 -30
  31. package/packages/safe-js/dist/interp/regex/parse.d.ts +42 -4
  32. package/packages/safe-js/dist/interp/regex/parse.js +429 -25
  33. package/packages/safe-js/dist/interp/regexp-iterator.d.ts +16 -0
  34. package/packages/safe-js/dist/interp/regexp-iterator.js +23 -0
  35. package/packages/safe-js/dist/interp/string-coercion.js +11 -7
  36. package/packages/safe-js/dist/interp/values.d.ts +2 -1
  37. package/packages/safe-js/dist/interp/values.js +39 -2
  38. package/packages/safe-js/dist/snapshot/dump-format.js +35 -3
  39. package/packages/safe-js/dist/snapshot/replay-data.d.ts +12 -1
  40. package/packages/safe-js/dist/snapshot/replay-data.js +42 -1
  41. package/packages/safe-js/dist/snapshot/replay-inputs.js +17 -0
  42. package/packages/safe-js/dist/snapshot/restore.js +23 -1
  43. package/packages/safe-js/dist/snapshot/serialize.d.ts +13 -2
  44. package/packages/safe-js/dist/snapshot/serialize.js +30 -11
  45. package/packages/safe-js/dist/snapshot/validation.js +22 -2
  46. package/packages/safe-js/dist/chunks/chunk-XKCTB5FR.js.map +0 -7
  47. /package/packages/safe-js/dist/chunks/{chunk-I2Z37MKZ.js.map → chunk-BAJI3K4B.js.map} +0 -0
@@ -9,11 +9,14 @@ export declare function getGuestFunctionProperty(closure: SandboxClosure, key: P
9
9
  export declare function installObjectPrototype(budget: Budget, prototype: SandboxObject, constructor: SandboxClosure): void;
10
10
  export declare function installBoxedPrototype(budget: Budget, prototype: SandboxObject, constructor: SandboxClosure, kind?: BoxedKind): void;
11
11
  export declare function getBoxedPrototype(value: BoxedPrimitive, budget: Budget): SandboxObject | undefined;
12
+ export declare function installRegexPrototype(budget: Budget, prototype: SandboxObject, constructor: SandboxClosure): void;
13
+ export declare function hasRegexPropertyOverride(value: SandboxValue, keys: readonly string[], budget: Budget): boolean;
12
14
  export declare function isDefaultBoxedMethod(value: BoxedPrimitive, key: string, budget: Budget): boolean;
13
15
  export declare function isIntrinsicConstructor(value: object): boolean;
14
16
  export declare function releaseObjectPrototype(budget: Budget): void;
15
17
  export declare function getSandboxPrototype(value: object, budget?: Budget): object | null;
16
18
  export declare function hasExplicitSandboxPrototype(value: object): boolean;
19
+ export declare function hasNullObjectPrototype(value: object): boolean;
17
20
  export declare function getSandboxPropertyDescriptor(value: SandboxValue, key: PropertyKey, budget?: Budget): PropertyDescriptor | undefined;
18
21
  export declare function getSandboxDataProperty(value: SandboxValue, key: PropertyKey, budget?: Budget): SandboxValue;
19
22
  export declare function setSandboxPrototype(value: object, prototype: object | null, budget?: Budget): void;
@@ -9,6 +9,8 @@ const functionProperties = new WeakMap();
9
9
  const prototypes = new WeakMap();
10
10
  const intrinsicPrototypes = new WeakMap();
11
11
  const boxedPrototypes = new WeakMap();
12
+ const regexPrototypes = new WeakMap();
13
+ const initialRegexDescriptors = new WeakMap();
12
14
  const intrinsicPrototypeRoots = new WeakMap();
13
15
  const intrinsicConstructors = new WeakMap();
14
16
  const initialBoxedMethods = new WeakMap();
@@ -80,6 +82,24 @@ export function installBoxedPrototype(budget, prototype, constructor, kind = typ
80
82
  export function getBoxedPrototype(value, budget) {
81
83
  return boxedPrototypes.get(budget)?.get(typeof value);
82
84
  }
85
+ export function installRegexPrototype(budget, prototype, constructor) {
86
+ regexPrototypes.set(budget, prototype);
87
+ initialRegexDescriptors.set(budget, Object.getOwnPropertyDescriptors(prototype));
88
+ registerIntrinsicPrototype(budget, prototype, constructor);
89
+ }
90
+ export function hasRegexPropertyOverride(value, keys, budget) {
91
+ const initial = initialRegexDescriptors.get(budget);
92
+ const prototype = regexPrototypes.get(budget);
93
+ const defaultPrototype = isSandboxRegex(value) && prototype !== undefined && getSandboxPrototype(value, budget) === prototype;
94
+ return keys.some(key => {
95
+ const current = defaultPrototype
96
+ ? Object.getOwnPropertyDescriptor(getRegexProperties(value), key) ?? Object.getOwnPropertyDescriptor(prototype, key)
97
+ : getSandboxPropertyDescriptor(value, key, budget);
98
+ const expected = initial?.[key];
99
+ return expected === undefined ? current !== undefined : current === undefined ||
100
+ current.value !== expected.value || current.get !== expected.get || current.set !== expected.set;
101
+ });
102
+ }
83
103
  export function isDefaultBoxedMethod(value, key, budget) {
84
104
  const prototype = getBoxedPrototype(value, budget);
85
105
  return prototype !== undefined && initialBoxedMethods.get(prototype)?.has(key) === true &&
@@ -93,7 +113,10 @@ function registerIntrinsicPrototype(budget, prototype, constructor) {
93
113
  if (roots === undefined)
94
114
  intrinsicPrototypeRoots.set(budget, (roots = new Set()));
95
115
  roots.add(prototype);
96
- const records = [prototype, constructor]
116
+ const methods = Reflect.ownKeys(prototype)
117
+ .map(key => Object.getOwnPropertyDescriptor(prototype, key)?.value)
118
+ .filter(isGuestClosure);
119
+ const records = [...new Set([prototype, constructor, ...methods])]
97
120
  .map((target) => ({
98
121
  target,
99
122
  value: isGuestClosure(target) ? materializeFunctionProperties(target) : target,
@@ -102,7 +125,7 @@ function registerIntrinsicPrototype(budget, prototype, constructor) {
102
125
  }))
103
126
  .map((record) => ({
104
127
  ...record,
105
- descriptors: new Map(Object.entries(Object.getOwnPropertyDescriptors(record.value)))
128
+ descriptors: new Map(Reflect.ownKeys(record.value).map(key => [key, Object.getOwnPropertyDescriptor(record.value, key)]))
106
129
  }));
107
130
  const unchanged = (before, after) => before !== undefined &&
108
131
  after !== undefined &&
@@ -116,16 +139,19 @@ function registerIntrinsicPrototype(budget, prototype, constructor) {
116
139
  const current = Object.getOwnPropertyDescriptors(value);
117
140
  return (getSandboxPrototype(target) === parent &&
118
141
  hasExplicitSandboxPrototype(target) === explicit &&
119
- Object.keys(current).length === descriptors.size &&
120
- Object.keys(current).every((key) => unchanged(descriptors.get(key), current[key])));
142
+ Reflect.ownKeys(current).length === descriptors.size &&
143
+ Reflect.ownKeys(value).every((key) => unchanged(descriptors.get(key), Object.getOwnPropertyDescriptor(value, key))));
121
144
  }));
122
145
  budget.setRetainedValues(prototype, () => records.flatMap(({ target, value, descriptors, prototype: parent }) => [
123
146
  ...(getSandboxPrototype(target) === parent
124
147
  ? []
125
148
  : [getSandboxPrototype(target)]),
126
- ...Object.entries(Object.getOwnPropertyDescriptors(value)).flatMap(([key, descriptor]) => unchanged(descriptors.get(key), descriptor)
127
- ? []
128
- : [key, descriptor.value, ...retainedAccessorClosures(descriptor)])
149
+ ...Reflect.ownKeys(value).flatMap(key => {
150
+ const descriptor = Object.getOwnPropertyDescriptor(value, key);
151
+ return unchanged(descriptors.get(key), descriptor)
152
+ ? []
153
+ : [key, descriptor.value, ...retainedAccessorClosures(descriptor)];
154
+ })
129
155
  ]));
130
156
  }
131
157
  export function releaseObjectPrototype(budget) {
@@ -133,11 +159,15 @@ export function releaseObjectPrototype(budget) {
133
159
  budget.setRetainedValues(prototype, undefined);
134
160
  intrinsicPrototypeRoots.delete(budget);
135
161
  boxedPrototypes.delete(budget);
162
+ regexPrototypes.delete(budget);
163
+ initialRegexDescriptors.delete(budget);
136
164
  intrinsicPrototypes.delete(budget);
137
165
  }
138
166
  export function getSandboxPrototype(value, budget) {
139
167
  if (prototypes.has(value))
140
168
  return prototypes.get(value) ?? null;
169
+ if (budget !== undefined && isSandboxRegex(value))
170
+ return regexPrototypes.get(budget) ?? null;
141
171
  if (budget !== undefined && isSandboxBox(value)) {
142
172
  const prototype = getBoxedPrototype(boxedValue(value), budget);
143
173
  if (prototype !== undefined && prototype !== value)
@@ -150,6 +180,9 @@ export function getSandboxPrototype(value, budget) {
150
180
  export function hasExplicitSandboxPrototype(value) {
151
181
  return prototypes.has(value);
152
182
  }
183
+ export function hasNullObjectPrototype(value) {
184
+ return prototypes.get(value) === null && !Array.isArray(value) && !isSandboxClosure(value) && !isSandboxRegex(value);
185
+ }
153
186
  export function getSandboxPropertyDescriptor(value, key, budget) {
154
187
  const hostProperties = isSandboxClosure(value) ? value.properties : undefined;
155
188
  if (isSandboxClosure(value) && !isGuestClosure(value))
@@ -213,13 +246,13 @@ export function setSandboxPrototype(value, prototype, budget) {
213
246
  if (budget !== undefined && intrinsicPrototypes.get(budget) === value && prototype !== null) {
214
247
  throw new TypeError("Object.prototype has an immutable null prototype.");
215
248
  }
216
- if ((!Array.isArray(value) && !isPrototypeRecord(value)) ||
249
+ if ((!Array.isArray(value) && !isSandboxRegex(value) && !isPrototypeRecord(value)) ||
217
250
  (prototype !== null && !Array.isArray(prototype) && !isPrototypeRecord(prototype))) {
218
251
  throw new TypeError("Prototype links require ordinary sandbox objects, arrays, or guest functions.");
219
252
  }
220
253
  if (prototypes.has(value) && getSandboxPrototype(value, budget) === prototype)
221
254
  return;
222
- if (!Object.isExtensible(isGuestClosure(value) ? materializeFunctionProperties(value) : value)) {
255
+ if (!Object.isExtensible(isGuestClosure(value) ? materializeFunctionProperties(value) : isSandboxRegex(value) ? getRegexProperties(value) : value)) {
223
256
  throw new TypeError("Cannot change the prototype of a non-extensible object.");
224
257
  }
225
258
  let depth = 0;
@@ -258,7 +291,7 @@ export function hasGuestObjectState(value) {
258
291
  return !intrinsicUnchanged();
259
292
  if (isLiveCapability(value))
260
293
  return true;
261
- if (functionProperties.has(value) || prototypes.has(value))
294
+ if (functionProperties.has(value) || (prototypes.has(value) && !hasNullObjectPrototype(value)))
262
295
  return true;
263
296
  if (isSandboxBox(value) || isSandboxDate(value))
264
297
  return false;
@@ -121,7 +121,7 @@ export class RegexCompileGuard {
121
121
  this.work(flags.length);
122
122
  }
123
123
  retain(pattern, valueUnits = 0) {
124
- const units = 10 + pattern.source.length + measureNode(pattern.body);
124
+ const units = measurePattern(pattern);
125
125
  if (this.ticket !== undefined) {
126
126
  this.ticket.owner.budget.resizeCompileTicket(this.ticket, units + valueUnits);
127
127
  }
@@ -139,7 +139,16 @@ export class RegexCompileGuard {
139
139
  }
140
140
  }
141
141
  export function regexCompiledData(pattern) {
142
- return (compiledData.get(pattern) ?? { units: 10 + pattern.source.length + measureNode(pattern.body) });
142
+ return (compiledData.get(pattern) ?? { units: measurePattern(pattern) });
143
+ }
144
+ function measurePattern(pattern) {
145
+ let units = 10 + pattern.source.length + measureNode(pattern.body);
146
+ if (pattern.groups !== undefined) {
147
+ units += 1;
148
+ for (const [name, indices] of Object.entries(pattern.groups))
149
+ units += name.length + 2 + indices.length;
150
+ }
151
+ return units;
143
152
  }
144
153
  function measureNode(node) {
145
154
  switch (node.type) {
@@ -148,18 +157,27 @@ function measureNode(node) {
148
157
  return 2;
149
158
  case "literal":
150
159
  return 3 + node.value.length;
160
+ case "namedBackreference":
161
+ return 3 + node.name.length;
151
162
  case "anchor":
152
163
  case "wordBoundary":
164
+ case "backreference":
153
165
  return 3;
154
166
  case "characterClass": {
155
167
  let usage = 5 + node.items.length;
168
+ if (node.operation !== undefined)
169
+ usage += 1 + node.operation.length;
170
+ if (node.strings !== undefined)
171
+ usage += 1;
156
172
  for (const item of node.items) {
157
173
  usage +=
158
174
  item.type === "character"
159
175
  ? 3 + item.value.length
160
176
  : item.type === "range"
161
177
  ? 4 + item.from.length + item.to.length
162
- : 4;
178
+ : item.type === "property" ? 4 + item.value.length + (item.strings === undefined ? 0 : 1)
179
+ : item.type === "set" ? 3 + measureNode({ type: "characterClass", ...item.value })
180
+ : item.type === "strings" ? 4 + item.values.reduce((total, value) => total + 1 + value.length, 0) : 4;
163
181
  }
164
182
  return usage;
165
183
  }
@@ -172,7 +190,11 @@ function measureNode(node) {
172
190
  return usage;
173
191
  }
174
192
  case "group":
175
- return 5 + measureNode(node.body);
193
+ return 5 + measureNode(node.body) + (node.name === undefined ? 0 : 1 + node.name.length) +
194
+ (node.modifiers === undefined ? 0 : 2 + Object.keys(node.modifiers).length);
195
+ case "lookahead":
196
+ case "lookbehind":
197
+ return 4 + measureNode(node.body);
176
198
  case "quantifier":
177
199
  return (Object.hasOwn(node, "max") ? 6 : 5) + measureNode(node.body);
178
200
  }
@@ -3,7 +3,11 @@ export type RegexMatch = {
3
3
  index: number;
4
4
  text: string;
5
5
  captures: (string | undefined)[];
6
+ indices?: ([number, number] | undefined)[];
7
+ groups?: Record<string, string | undefined>;
8
+ indicesGroups?: Record<string, [number, number] | undefined>;
6
9
  };
7
10
  export declare function matchRegex(pattern: RegexPattern, input: string, lastIndex?: number): RegexMatch | null;
8
11
  export declare function matchRegexFrom(pattern: RegexPattern, input: string, startIndex: number): RegexMatch | null;
9
12
  export declare function normalizeLastIndex(lastIndex: number): number;
13
+ export declare function advanceStringIndex(input: string, index: number, unicode: boolean): number;
@@ -1,14 +1,19 @@
1
1
  import { allocateRegexSteps } from "../budget.js";
2
2
  export function matchRegex(pattern, input, lastIndex = 0) {
3
- const startIndex = pattern.flags.global ? normalizeLastIndex(lastIndex) : 0;
3
+ const startIndex = pattern.flags.global || pattern.flags.sticky ? normalizeLastIndex(lastIndex) : 0;
4
4
  return matchRegexFrom(pattern, input, startIndex);
5
5
  }
6
6
  export function matchRegexFrom(pattern, input, startIndex) {
7
7
  if (startIndex > input.length) {
8
8
  return null;
9
9
  }
10
- for (let attempt = startIndex; attempt <= input.length; attempt += 1) {
11
- const context = { input, flags: pattern.flags, steps: 0 };
10
+ const unicode = pattern.flags.unicode || pattern.flags.unicodeSets;
11
+ if (unicode && startIndex > 0 && startIndex < input.length &&
12
+ input.charCodeAt(startIndex) >= 0xdc00 && input.charCodeAt(startIndex) <= 0xdfff &&
13
+ input.charCodeAt(startIndex - 1) >= 0xd800 && input.charCodeAt(startIndex - 1) <= 0xdbff)
14
+ startIndex--;
15
+ for (let attempt = startIndex; attempt <= input.length; attempt = advanceStringIndex(input, attempt, unicode)) {
16
+ const context = { input, flags: pattern.flags, groups: pattern.groups, direction: 1, work: { steps: 0 } };
12
17
  charge(context);
13
18
  const initialState = {
14
19
  position: attempt,
@@ -16,26 +21,63 @@ export function matchRegexFrom(pattern, input, startIndex) {
16
21
  };
17
22
  const result = matchNode(pattern.body, initialState, context).next();
18
23
  if (!result.done) {
19
- return toRegexMatch(input, attempt, result.value);
24
+ return toRegexMatch(input, attempt, result.value, pattern.flags.hasIndices, pattern.groups);
20
25
  }
26
+ if (pattern.flags.sticky)
27
+ break;
21
28
  }
22
29
  return null;
23
30
  }
24
31
  function* matchNode(node, state, context) {
25
32
  charge(context);
33
+ const unicode = context.flags.unicode || context.flags.unicodeSets;
34
+ const character = readCharacter(context.input, state.position, context.direction, unicode);
26
35
  switch (node.type) {
27
36
  case "empty":
28
37
  yield state;
29
38
  return;
30
39
  case "literal":
31
- if (charactersEqual(context.input[state.position], node.value, context.flags.ignoreCase)) {
32
- yield { ...state, position: state.position + 1 };
40
+ if (charactersEqual(character, node.value, context.flags.ignoreCase, unicode)) {
41
+ yield { ...state, position: state.position + context.direction * character.length };
33
42
  }
34
43
  return;
44
+ case "backreference":
45
+ case "namedBackreference": {
46
+ let capture;
47
+ if (node.type === "backreference")
48
+ capture = state.captures[node.index - 1];
49
+ else {
50
+ for (const index of context.groups?.[node.name] ?? []) {
51
+ charge(context);
52
+ if (state.captures[index - 1] !== undefined) {
53
+ capture = state.captures[index - 1];
54
+ break;
55
+ }
56
+ }
57
+ }
58
+ if (capture === undefined) {
59
+ yield state;
60
+ return;
61
+ }
62
+ let position = state.position;
63
+ let reference = context.direction === 1 ? capture.start : capture.end;
64
+ const end = context.direction === 1 ? capture.end : capture.start;
65
+ while (reference !== end) {
66
+ charge(context);
67
+ const expected = readCharacter(context.input, reference, context.direction, unicode);
68
+ const actual = readCharacter(context.input, position, context.direction, unicode);
69
+ if (!charactersEqual(actual, expected, context.flags.ignoreCase, unicode))
70
+ return;
71
+ reference += context.direction * expected.length;
72
+ position += context.direction * actual.length;
73
+ }
74
+ yield { ...state, position };
75
+ return;
76
+ }
35
77
  case "dot":
36
- if (state.position < context.input.length &&
37
- (context.flags.dotAll || !isLineTerminator(context.input[state.position]))) {
38
- yield { ...state, position: state.position + 1 };
78
+ if (character !== undefined &&
79
+ (context.flags.dotAll || !isLineTerminator(character))) {
80
+ yield { ...state, position: state.position + context.direction * character.length };
39
81
  }
40
82
  return;
41
83
  case "anchor":
@@ -44,18 +86,25 @@ function* matchNode(node, state, context) {
44
86
  }
45
87
  return;
46
88
  case "wordBoundary": {
47
- const previousWord = state.position > 0 && isWordCharacter(context.input[state.position - 1]);
48
- const nextWord = state.position < context.input.length && isWordCharacter(context.input[state.position]);
89
+ const previousWord = state.position > 0 && matchesCharacterClassItem(context.input[state.position - 1], { type: "kind", kind: "word", negated: false }, context.flags.ignoreCase, unicode);
90
+ const nextWord = state.position < context.input.length && matchesCharacterClassItem(context.input[state.position], { type: "kind", kind: "word", negated: false }, context.flags.ignoreCase, unicode);
49
91
  if ((previousWord !== nextWord) !== node.negated) {
50
92
  yield state;
51
93
  }
52
94
  return;
53
95
  }
54
96
  case "characterClass": {
55
- const character = context.input[state.position];
97
+ if (context.flags.unicodeSets) {
98
+ const lengths = [...matchUnicodeSet(node, state.position, context)].sort((a, b) => b - a);
99
+ for (const length of lengths) {
100
+ charge(context);
101
+ yield { ...state, position: state.position + context.direction * length };
102
+ }
103
+ return;
104
+ }
56
105
  if (character !== undefined &&
57
- matchesCharacterClass(character, node.items, node.negated, context.flags.ignoreCase)) {
58
- yield { ...state, position: state.position + 1 };
106
+ matchesCharacterClass(character, node.items, node.negated, context)) {
107
+ yield { ...state, position: state.position + context.direction * character.length };
59
108
  }
60
109
  return;
61
110
  }
@@ -67,14 +116,30 @@ function* matchNode(node, state, context) {
67
116
  yield* matchNode(alternative, cloneState(state), context);
68
117
  }
69
118
  return;
119
+ case "lookahead":
120
+ case "lookbehind": {
121
+ const result = matchNode(node.body, cloneState(state), {
122
+ ...context, direction: node.type === "lookbehind" ? -1 : 1
123
+ }).next();
124
+ if (node.negated) {
125
+ if (result.done)
126
+ yield state;
127
+ }
128
+ else if (!result.done) {
129
+ yield { position: state.position, captures: result.value.captures };
130
+ }
131
+ return;
132
+ }
70
133
  case "group":
71
- for (const result of matchNode(node.body, cloneState(state), context)) {
134
+ for (const result of matchNode(node.body, cloneState(state), node.modifiers === undefined ? context : {
135
+ ...context, flags: { ...context.flags, ...node.modifiers }
136
+ })) {
72
137
  if (!node.capturing || node.index === undefined) {
73
138
  yield result;
74
139
  continue;
75
140
  }
76
141
  const captures = result.captures.slice();
77
- captures[node.index - 1] = { start: state.position, end: result.position };
142
+ captures[node.index - 1] = { start: Math.min(state.position, result.position), end: Math.max(state.position, result.position) };
78
143
  yield { position: result.position, captures };
79
144
  }
80
145
  return;
@@ -88,7 +153,8 @@ function* matchSequence(elements, index, state, context) {
88
153
  yield state;
89
154
  return;
90
155
  }
91
- for (const result of matchNode(elements[index], state, context)) {
156
+ const element = elements[context.direction === 1 ? index : elements.length - 1 - index];
157
+ for (const result of matchNode(element, state, context)) {
92
158
  yield* matchSequence(elements, index + 1, result, context);
93
159
  }
94
160
  }
@@ -129,11 +195,35 @@ function matchesAnchor(kind, position, context) {
129
195
  position < context.input.length &&
130
196
  isLineTerminator(context.input[position])));
131
197
  }
132
- function matchesCharacterClass(character, items, negated, ignoreCase) {
133
- const matched = items.some((item) => matchesCharacterClassItem(character, item, ignoreCase));
198
+ function matchesCharacterClass(character, items, negated, context) {
199
+ const matched = items.some((item) => {
200
+ if (context.flags.unicode)
201
+ charge(context);
202
+ return matchesCharacterClassItem(character, item, context.flags.ignoreCase, context.flags.unicode);
203
+ });
134
204
  return negated ? !matched : matched;
135
205
  }
136
- function matchesCharacterClassItem(character, item, ignoreCase) {
206
+ function matchesCharacterClassItem(character, item, ignoreCase, unicode, unicodeSets = false) {
207
+ if (item.type === "strings" || item.type === "set")
208
+ return false;
209
+ if (unicode) {
210
+ // Generated atoms only: the host classifies one code point, never executes a guest pattern.
211
+ const hex = (value) => `\\u{${value.codePointAt(0).toString(16)}}`;
212
+ let atom;
213
+ if (item.type === "character")
214
+ atom = hex(item.value);
215
+ else if (item.type === "range")
216
+ atom = `[${hex(item.from)}-${hex(item.to)}]`;
217
+ else if (item.type === "property")
218
+ atom = `\\${item.negated ? "P" : "p"}{${item.value}}`;
219
+ else {
220
+ const kind = item.kind === "digit" ? "d" : item.kind === "word" ? "w" : "s";
221
+ atom = `\\${item.negated ? kind.toUpperCase() : kind}`;
222
+ }
223
+ return new RegExp(`^(?:${atom})$`, (ignoreCase ? "i" : "") + (unicodeSets ? "v" : "u")).test(character);
224
+ }
225
+ if (item.type === "property")
226
+ return false;
137
227
  if (item.type === "character") {
138
228
  return charactersEqual(character, item.value, ignoreCase);
139
229
  }
@@ -147,10 +237,10 @@ function matchesCharacterClassItem(character, item, ignoreCase) {
147
237
  if (!ignoreCase) {
148
238
  return false;
149
239
  }
150
- const foldedCandidate = foldCharacter(character, true).charCodeAt(0);
151
- const foldedFrom = foldCharacter(item.from, true).charCodeAt(0);
152
- const foldedTo = foldCharacter(item.to, true).charCodeAt(0);
153
- return foldedCandidate >= foldedFrom && foldedCandidate <= foldedTo;
240
+ // Folding endpoints does not preserve range membership. Classify one UTF-16
241
+ // code unit using generated hex endpoints, never a guest pattern.
242
+ return new RegExp(`^[\\u${from.toString(16).padStart(4, "0")}-\\u${to.toString(16).padStart(4, "0")}]$`, "i")
243
+ .test(character);
154
244
  }
155
245
  const matched = item.kind === "digit"
156
246
  ? isDigit(character)
@@ -159,12 +249,107 @@ function matchesCharacterClassItem(character, item, ignoreCase) {
159
249
  : isSpaceCharacter(character);
160
250
  return item.negated ? !matched : matched;
161
251
  }
162
- function toRegexMatch(input, start, state) {
163
- return {
252
+ function matchUnicodeSet(set, position, context) {
253
+ charge(context);
254
+ let lengths = new Set();
255
+ for (let index = 0; index < set.items.length; index++) {
256
+ const next = matchUnicodeSetItem(set.items[index], position, context);
257
+ if (index === 0)
258
+ lengths = next;
259
+ else if (set.operation === undefined) {
260
+ for (const length of next) {
261
+ charge(context);
262
+ lengths.add(length);
263
+ }
264
+ }
265
+ else {
266
+ for (const length of lengths) {
267
+ charge(context);
268
+ if (set.operation === "intersection" ? !next.has(length) : next.has(length))
269
+ lengths.delete(length);
270
+ }
271
+ }
272
+ }
273
+ if (set.negated) {
274
+ const character = readCharacter(context.input, position, context.direction, true);
275
+ return new Set(character !== undefined && !lengths.has(character.length) ? [character.length] : []);
276
+ }
277
+ return lengths;
278
+ }
279
+ function matchUnicodeSetItem(item, position, context) {
280
+ charge(context);
281
+ if (item.type === "set")
282
+ return matchUnicodeSet(item.value, position, context);
283
+ const lengths = new Set();
284
+ if (item.type === "strings") {
285
+ for (const value of item.values) {
286
+ charge(context);
287
+ let reference = context.direction === 1 ? 0 : value.length;
288
+ let cursor = position;
289
+ const end = context.direction === 1 ? value.length : 0;
290
+ while (reference !== end) {
291
+ charge(context);
292
+ const expected = readCharacter(value, reference, context.direction, true);
293
+ const actual = readCharacter(context.input, cursor, context.direction, true);
294
+ if (!charactersEqual(actual, expected, context.flags.ignoreCase, true))
295
+ break;
296
+ reference += context.direction * expected.length;
297
+ cursor += context.direction * actual.length;
298
+ }
299
+ if (reference === end)
300
+ lengths.add(Math.abs(cursor - position));
301
+ }
302
+ return lengths;
303
+ }
304
+ if (item.type === "property" && item.strings) {
305
+ // A validated property atom denotes a finite Unicode string set. The host
306
+ // finds its longest member only; guest alternation/backtracking stays here.
307
+ const atom = `\\p{${item.value}}`;
308
+ const flags = (context.flags.ignoreCase ? "i" : "") + "v";
309
+ const matcher = new RegExp(context.direction === 1 ? atom : `(?<=(${atom}))`, flags + "y");
310
+ matcher.lastIndex = position;
311
+ const match = matcher.exec(context.input);
312
+ if (match === null)
313
+ return lengths;
314
+ const longest = context.direction === 1 ? match[0] : match[1];
315
+ const member = new RegExp(`^(?:${atom})$`, flags);
316
+ let consumed = 0;
317
+ while (consumed < longest.length) {
318
+ charge(context);
319
+ const cursor = context.direction === 1 ? consumed : longest.length - consumed;
320
+ consumed += readCharacter(longest, cursor, context.direction, true).length;
321
+ const candidate = context.direction === 1 ? longest.slice(0, consumed) : longest.slice(longest.length - consumed);
322
+ if (member.test(candidate))
323
+ lengths.add(consumed);
324
+ }
325
+ return lengths;
326
+ }
327
+ const character = readCharacter(context.input, position, context.direction, true);
328
+ if (character !== undefined && matchesCharacterClassItem(character, item, context.flags.ignoreCase, true, true))
329
+ lengths.add(character.length);
330
+ return lengths;
331
+ }
332
+ function toRegexMatch(input, start, state, hasIndices, groups) {
333
+ const match = {
164
334
  index: start,
165
335
  text: input.slice(start, state.position),
166
336
  captures: state.captures.map((capture) => capture === undefined ? undefined : input.slice(capture.start, capture.end))
167
337
  };
338
+ if (hasIndices) {
339
+ match.indices = [[start, state.position], ...state.captures.map((capture) => capture === undefined ? undefined : [capture.start, capture.end])];
340
+ }
341
+ if (groups !== undefined) {
342
+ match.groups = Object.create(null);
343
+ if (hasIndices)
344
+ match.indicesGroups = Object.create(null);
345
+ for (const [name, indices] of Object.entries(groups)) {
346
+ const index = indices.find(index => state.captures[index - 1] !== undefined);
347
+ match.groups[name] = index === undefined ? undefined : match.captures[index - 1];
348
+ if (match.indicesGroups !== undefined)
349
+ match.indicesGroups[name] = index === undefined ? undefined : match.indices[index];
350
+ }
351
+ }
352
+ return match;
168
353
  }
169
354
  function cloneState(state) {
170
355
  return { position: state.position, captures: state.captures.slice() };
@@ -194,13 +379,13 @@ function clearNodeCaptures(node, captures) {
194
379
  }
195
380
  return;
196
381
  }
197
- if (node.type === "quantifier") {
382
+ if (node.type === "quantifier" || node.type === "lookahead" || node.type === "lookbehind") {
198
383
  clearNodeCaptures(node.body, captures);
199
384
  }
200
385
  }
201
386
  function charge(context) {
202
- context.steps += 1;
203
- allocateRegexSteps(context.steps);
387
+ context.work.steps += 1;
388
+ allocateRegexSteps(context.work.steps);
204
389
  }
205
390
  export function normalizeLastIndex(lastIndex) {
206
391
  if (Number.isNaN(lastIndex) || lastIndex <= 0) {
@@ -208,9 +393,24 @@ export function normalizeLastIndex(lastIndex) {
208
393
  }
209
394
  return Math.min(Math.floor(lastIndex), Number.MAX_SAFE_INTEGER);
210
395
  }
211
- function charactersEqual(left, right, ignoreCase) {
396
+ function charactersEqual(left, right, ignoreCase, unicode = false) {
397
+ if (unicode && ignoreCase && left !== undefined)
398
+ return new RegExp(`^\\u{${right.codePointAt(0).toString(16)}}$`, "iu").test(left);
212
399
  return left !== undefined && foldCharacter(left, ignoreCase) === foldCharacter(right, ignoreCase);
213
400
  }
401
+ export function advanceStringIndex(input, index, unicode) {
402
+ return index + (unicode && (input.codePointAt(index) ?? 0) > 0xffff ? 2 : 1);
403
+ }
404
+ function readCharacter(input, position, direction, unicode) {
405
+ let index = direction === 1 ? position : position - 1;
406
+ if (index < 0 || index >= input.length)
407
+ return undefined;
408
+ if (unicode && direction === -1 && index > 0 &&
409
+ input.charCodeAt(index) >= 0xdc00 && input.charCodeAt(index) <= 0xdfff &&
410
+ input.charCodeAt(index - 1) >= 0xd800 && input.charCodeAt(index - 1) <= 0xdbff)
411
+ index--;
412
+ return unicode ? String.fromCodePoint(input.codePointAt(index)) : input[index];
413
+ }
214
414
  function foldCharacter(character, ignoreCase) {
215
415
  if (!ignoreCase) {
216
416
  return character;