punctilio 3.0.7 → 3.0.9
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.
- package/README.md +5 -5
- package/dist/constants.d.ts +10 -0
- package/dist/constants.d.ts.map +1 -1
- package/dist/constants.js +19 -1
- package/dist/constants.js.map +1 -1
- package/dist/dashes.js +20 -20
- package/dist/dashes.js.map +1 -1
- package/dist/nbsp.d.ts.map +1 -1
- package/dist/nbsp.js +17 -23
- package/dist/nbsp.js.map +1 -1
- package/dist/quotes.js +19 -19
- package/dist/quotes.js.map +1 -1
- package/dist/rehype.d.ts.map +1 -1
- package/dist/rehype.js +3 -1
- package/dist/rehype.js.map +1 -1
- package/dist/symbols.d.ts.map +1 -1
- package/dist/symbols.js +12 -31
- package/dist/symbols.js.map +1 -1
- package/package.json +1 -1
- package/src/constants.ts +20 -1
- package/src/dashes.ts +20 -20
- package/src/nbsp.ts +17 -25
- package/src/quotes.ts +19 -19
- package/src/rehype.ts +2 -1
- package/src/symbols.ts +12 -31
- package/src/tests/utils.test.ts +14 -1
package/README.md
CHANGED
|
@@ -37,9 +37,9 @@ I tested `punctilio` 1.2.9 against [`smartypants`](https://www.npmjs.com/package
|
|
|
37
37
|
|
|
38
38
|
My [`benchmark.mjs`](https://github.com/alexander-turner/punctilio/blob/main/benchmark.mjs) measures how well libraries handle a [wide range of scenarios](https://github.com/alexander-turner/punctilio/blob/main/benchmark_cases.json). The benchmark normalizes stylistic differences (e.g. non-breaking vs regular space, British vs American dash spacing) for fair comparison.
|
|
39
39
|
|
|
40
|
-
| Package | Passed (of
|
|
40
|
+
| Package | Passed (of 157) |
|
|
41
41
|
|--------:|:----------------|
|
|
42
|
-
| `punctilio` |
|
|
42
|
+
| `punctilio` | 152 (97%) |
|
|
43
43
|
| `tipograph` | 92 (58%) |
|
|
44
44
|
| `typograf` | 74 (47%) |
|
|
45
45
|
| `smartquotes` | 72 (45%) |
|
|
@@ -76,7 +76,7 @@ My [`benchmark.mjs`](https://github.com/alexander-turner/punctilio/blob/main/ben
|
|
|
76
76
|
|
|
77
77
|
## Test suite
|
|
78
78
|
|
|
79
|
-
Setting aside the benchmark, `punctilio`’s test suite includes 1,
|
|
79
|
+
Setting aside the benchmark, `punctilio`’s test suite includes 1,400+ tests at 100% branch coverage, including edge cases derived from competitor libraries ([`smartquotes`](https://github.com/kellym/smartquotes.js), [`retext-smartypants`](https://github.com/retextjs/retext-smartypants), [`typograf`](https://github.com/typograf/typograf)) and the [Standard Ebooks typography manual](https://standardebooks.org/manual/). I also verify that all transformations are stable when applied multiple times.
|
|
80
80
|
|
|
81
81
|
## Works with HTML DOMs via separation boundaries
|
|
82
82
|
|
|
@@ -136,7 +136,7 @@ transform(text, {
|
|
|
136
136
|
superscript: false, // 1st → 1ˢᵗ
|
|
137
137
|
ligatures: false, // ??? → ⁇, ?! → ⁈, !? → ⁉, !!! → !
|
|
138
138
|
nbsp: true, // non-breaking spaces (after honorifics, between numbers and units, etc.)
|
|
139
|
-
checkIdempotency:
|
|
139
|
+
checkIdempotency: true, // verify transform(transform(x)) === transform(x)
|
|
140
140
|
})
|
|
141
141
|
```
|
|
142
142
|
|
|
@@ -148,5 +148,5 @@ transform(text, {
|
|
|
148
148
|
- Periods and commas go outside quotation marks (“Hello”, she said.)
|
|
149
149
|
- Spaced en-dashes between words (word – word)
|
|
150
150
|
- Setting either style to `none` skips the entire transform category: `punctuationStyle: 'none'` preserves straight quotes, apostrophes, and prime marks; `dashStyle: 'none'` preserves all hyphens, number ranges, date ranges, and minus signs.
|
|
151
|
-
- `punctilio` is idempotent by design: `transform(transform(text))` always equals `transform(text)`.
|
|
151
|
+
- `punctilio` is idempotent by design: `transform(transform(text))` always equals `transform(text)`. This is verified automatically by default (`checkIdempotency: true`). Set `checkIdempotency: false` to disable the check.
|
|
152
152
|
- Use `classifyApostrophes(text)` to distinguish apostrophes from closing single quotes. It returns text with apostrophes as U+02BC (MODIFIER LETTER APOSTROPHE) and closing quotes as U+2019 (RIGHT SINGLE QUOTATION MARK). Per the [Unicode Standard](https://www.unicode.org/versions/Unicode16.0.0/core-spec/chapter-6/#G30602), `transform()` and `niceQuotes()` use U+2019 for both in their output.
|
package/dist/constants.d.ts
CHANGED
|
@@ -99,6 +99,16 @@ export declare const LATIN_LETTERS = "A-Za-z\\u00C0-\\u00D6\\u00D8-\\u00F6\\u00F
|
|
|
99
99
|
*/
|
|
100
100
|
export declare const DEFAULT_SEPARATOR = "\uE000";
|
|
101
101
|
export declare const ESCAPED_DEFAULT_SEPARATOR: string;
|
|
102
|
+
/**
|
|
103
|
+
* Returns the regex-escaped separator string for the given options.
|
|
104
|
+
* Centralizes the common pattern of falling back to the pre-escaped default.
|
|
105
|
+
*
|
|
106
|
+
* @param options - Object with an optional `separator` field
|
|
107
|
+
* @returns Regex-escaped separator string
|
|
108
|
+
*/
|
|
109
|
+
export declare function getEscapedSeparator(options: {
|
|
110
|
+
separator?: string;
|
|
111
|
+
}): string;
|
|
102
112
|
/**
|
|
103
113
|
* Characters that have special meaning in regular expressions.
|
|
104
114
|
* Used for testing that separator escaping works correctly.
|
package/dist/constants.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"constants.d.ts","sourceRoot":"","sources":["../src/constants.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAIH;;GAEG;AACH,eAAO,MAAM,eAAe;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CAmElB,CAAA;AAEV;;;;;;GAMG;AACH,eAAO,MAAM,oBAAoB,8HAYvB,CAAA;AAEV;;;;;;;;;;;GAWG;AACH,eAAO,MAAM,aAAa,sFAAsF,CAAA;AAEhH;;;GAGG;AACH,eAAO,MAAM,iBAAiB,WAAW,CAAA;AACzC,eAAO,MAAM,yBAAyB,QAAwC,CAAA;AAE9E;;;GAGG;AACH,eAAO,MAAM,mBAAmB,wEAAyE,CAAA;AAEzG;;;;;;;;;;;;;;;GAeG;AACH,wBAAgB,iBAAiB,CAAC,gBAAgB,EAAE,MAAM,GAAG,MAAM,CAElE;AAED;;;;;;;;;;;;;;;GAeG;AACH,wBAAgB,eAAe,CAAC,gBAAgB,EAAE,MAAM,GAAG,MAAM,CAEhE;AAED;;;GAGG;AACH,eAAO,MAAM,WAAW,QAA6B,CAAA;AAErD;;;;;;GAMG;AACH,wBAAgB,kBAAkB,CAAC,gBAAgB,EAAE,MAAM,GAAG,MAAM,CAEnE;AAED;;;;;;GAMG;AACH,wBAAgB,gBAAgB,CAAC,gBAAgB,EAAE,MAAM,GAAG,MAAM,CAEjE;
|
|
1
|
+
{"version":3,"file":"constants.d.ts","sourceRoot":"","sources":["../src/constants.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAIH;;GAEG;AACH,eAAO,MAAM,eAAe;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CAmElB,CAAA;AAEV;;;;;;GAMG;AACH,eAAO,MAAM,oBAAoB,8HAYvB,CAAA;AAEV;;;;;;;;;;;GAWG;AACH,eAAO,MAAM,aAAa,sFAAsF,CAAA;AAEhH;;;GAGG;AACH,eAAO,MAAM,iBAAiB,WAAW,CAAA;AACzC,eAAO,MAAM,yBAAyB,QAAwC,CAAA;AAE9E;;;;;;GAMG;AACH,wBAAgB,mBAAmB,CAAC,OAAO,EAAE;IAAE,SAAS,CAAC,EAAE,MAAM,CAAA;CAAE,GAAG,MAAM,CAI3E;AAED;;;GAGG;AACH,eAAO,MAAM,mBAAmB,wEAAyE,CAAA;AAEzG;;;;;;;;;;;;;;;GAeG;AACH,wBAAgB,iBAAiB,CAAC,gBAAgB,EAAE,MAAM,GAAG,MAAM,CAElE;AAED;;;;;;;;;;;;;;;GAeG;AACH,wBAAgB,eAAe,CAAC,gBAAgB,EAAE,MAAM,GAAG,MAAM,CAEhE;AAED;;;GAGG;AACH,eAAO,MAAM,WAAW,QAA6B,CAAA;AAErD;;;;;;GAMG;AACH,wBAAgB,kBAAkB,CAAC,gBAAgB,EAAE,MAAM,GAAG,MAAM,CAEnE;AAED;;;;;;GAMG;AACH,wBAAgB,gBAAgB,CAAC,gBAAgB,EAAE,MAAM,GAAG,MAAM,CAEjE;AAUD;;;;GAIG;AACH,wBAAgB,YAAY,CAAC,OAAO,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,GAAG,MAAM,CAcnE"}
|
package/dist/constants.js
CHANGED
|
@@ -118,6 +118,18 @@ export const LATIN_LETTERS = "A-Za-z\\u00C0-\\u00D6\\u00D8-\\u00F6\\u00F8-\\u00F
|
|
|
118
118
|
*/
|
|
119
119
|
export const DEFAULT_SEPARATOR = "\uE000";
|
|
120
120
|
export const ESCAPED_DEFAULT_SEPARATOR = escapeStringRegexp(DEFAULT_SEPARATOR);
|
|
121
|
+
/**
|
|
122
|
+
* Returns the regex-escaped separator string for the given options.
|
|
123
|
+
* Centralizes the common pattern of falling back to the pre-escaped default.
|
|
124
|
+
*
|
|
125
|
+
* @param options - Object with an optional `separator` field
|
|
126
|
+
* @returns Regex-escaped separator string
|
|
127
|
+
*/
|
|
128
|
+
export function getEscapedSeparator(options) {
|
|
129
|
+
return options.separator
|
|
130
|
+
? escapeStringRegexp(options.separator)
|
|
131
|
+
: ESCAPED_DEFAULT_SEPARATOR;
|
|
132
|
+
}
|
|
121
133
|
/**
|
|
122
134
|
* Characters that have special meaning in regular expressions.
|
|
123
135
|
* Used for testing that separator escaping works correctly.
|
|
@@ -189,9 +201,10 @@ export function spaceBoundaryEnd(escapedSeparator) {
|
|
|
189
201
|
/**
|
|
190
202
|
* Cache for compiled RegExp objects keyed by `pattern + '\0' + flags`.
|
|
191
203
|
* Avoids recompiling identical regexes on every function call (common
|
|
192
|
-
* when using the default separator).
|
|
204
|
+
* when using the default separator). Capped to prevent unbounded growth.
|
|
193
205
|
*/
|
|
194
206
|
const regexCache = new Map();
|
|
207
|
+
const MAX_REGEX_CACHE_SIZE = 1000;
|
|
195
208
|
/**
|
|
196
209
|
* Returns a cached RegExp for the given pattern and flags.
|
|
197
210
|
* Resets `lastIndex` before returning to prevent stale state when
|
|
@@ -202,6 +215,11 @@ export function cachedRegExp(pattern, flags) {
|
|
|
202
215
|
let re = regexCache.get(key);
|
|
203
216
|
if (!re) {
|
|
204
217
|
re = new RegExp(pattern, flags);
|
|
218
|
+
if (regexCache.size >= MAX_REGEX_CACHE_SIZE) {
|
|
219
|
+
// Evict oldest entry (first key in insertion order)
|
|
220
|
+
const oldest = regexCache.keys().next().value;
|
|
221
|
+
regexCache.delete(oldest);
|
|
222
|
+
}
|
|
205
223
|
regexCache.set(key, re);
|
|
206
224
|
}
|
|
207
225
|
re.lastIndex = 0;
|
package/dist/constants.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"constants.js","sourceRoot":"","sources":["../src/constants.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAEH,OAAO,kBAAkB,MAAM,sBAAsB,CAAA;AAErD;;GAEG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG;IAC7B,QAAQ,EAAE,QAAQ;IAClB,cAAc,EAAE,QAAQ;IACxB,SAAS,EAAE,QAAQ;IACnB,UAAU,EAAE,QAAQ;IACpB,SAAS,EAAE,QAAQ;IACnB,UAAU,EAAE,QAAQ;IACpB,SAAS,EAAE,QAAQ;IACnB,MAAM,EAAE,QAAQ;IAChB,WAAW,EAAE,QAAQ;IACrB,UAAU,EAAE,QAAQ;IACpB,gBAAgB,EAAE,QAAQ;IAC1B,WAAW,EAAE,QAAQ;IACrB,UAAU,EAAE,QAAQ;IACpB,aAAa,EAAE,QAAQ;IACvB,KAAK,EAAE,QAAQ;IACf,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,OAAO,EAAE,QAAQ;IACjB,OAAO,EAAE,QAAQ;IACjB,KAAK,EAAE,QAAQ;IACf,iBAAiB,EAAE,QAAQ;IAC3B,kBAAkB,EAAE,QAAQ;IAC5B,iBAAiB,EAAE,QAAQ;IAC3B,kBAAkB,EAAE,QAAQ;IAC5B,0BAA0B,EAAE,QAAQ;IACpC,IAAI,EAAE,QAAQ;IACd,+BAA+B;IAC/B,cAAc,EAAE,cAAc,EAAE,KAAK;IACrC,cAAc,EAAE,cAAc,EAAE,KAAK;IACrC,cAAc,EAAE,cAAc,EAAE,KAAK;IACrC,cAAc,EAAE,cAAc,EAAE,KAAK;IACrC,wBAAwB;IACxB,eAAe,EAAE,QAAQ,EAAE,IAAI;IAC/B,oBAAoB,EAAE,QAAQ,EAAE,IAAI;IACpC,oBAAoB,EAAE,QAAQ,EAAE,IAAI;IACpC,kBAAkB,EAAE,QAAQ,EAAE,IAAI;IAClC,WAAW,EAAE,QAAQ,EAAE,IAAI;IAC3B,4BAA4B;IAC5B,qBAAqB,EAAE,QAAQ,EAAE,IAAI;IACrC,kBAAkB,EAAE,QAAQ,EAAE,IAAI;IAClC,gBAAgB,EAAE,QAAQ,EAAE,IAAI;IAChC,eAAe,EAAE,QAAQ,EAAE,IAAI;IAC/B,mBAAmB,EAAE,QAAQ,EAAE,IAAI;IACnC,eAAe,EAAE,QAAQ,EAAE,IAAI;IAC/B,8BAA8B;IAC9B,qBAAqB,EAAE,QAAQ,EAAE,IAAI;IACrC,iBAAiB,EAAE,QAAQ,EAAE,IAAI;IACjC,qBAAqB;IACrB,oBAAoB,EAAE,QAAQ,EAAE,IAAI;IACpC,gBAAgB,EAAE,QAAQ,EAAE,IAAI;IAChC,oBAAoB;IACpB,mBAAmB,EAAE,QAAQ,EAAE,4CAA4C;CACnE,CAAA;AAEV;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAG;IAClC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG;IAC5B,eAAe,CAAC,QAAQ;IACxB,eAAe,CAAC,eAAe,EAAE,eAAe,CAAC,oBAAoB;IACrE,eAAe,CAAC,oBAAoB,EAAE,eAAe,CAAC,kBAAkB;IACxE,eAAe,CAAC,WAAW;IAC3B,eAAe,CAAC,qBAAqB,EAAE,eAAe,CAAC,kBAAkB;IACzE,eAAe,CAAC,gBAAgB,EAAE,eAAe,CAAC,eAAe;IACjE,eAAe,CAAC,mBAAmB,EAAE,eAAe,CAAC,eAAe;IACpE,eAAe,CAAC,qBAAqB,EAAE,eAAe,CAAC,iBAAiB;IACxE,eAAe,CAAC,oBAAoB,EAAE,eAAe,CAAC,gBAAgB;IACtE,eAAe,CAAC,mBAAmB;CAC3B,CAAA;AAEV;;;;;;;;;;;GAWG;AACH,MAAM,CAAC,MAAM,aAAa,GAAG,mFAAmF,CAAA;AAEhH;;;GAGG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAAG,QAAQ,CAAA;AACzC,MAAM,CAAC,MAAM,yBAAyB,GAAG,kBAAkB,CAAC,iBAAiB,CAAC,CAAA;AAE9E;;;GAGG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAU,CAAA;AAEzG;;;;;;;;;;;;;;;GAeG;AACH,MAAM,UAAU,iBAAiB,CAAC,gBAAwB;IACxD,OAAO,UAAU,gBAAgB,OAAO,CAAA;AAC1C,CAAC;AAED;;;;;;;;;;;;;;;GAeG;AACH,MAAM,UAAU,eAAe,CAAC,gBAAwB;IACtD,OAAO,SAAS,gBAAgB,OAAO,CAAA;AACzC,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,IAAI,eAAe,CAAC,IAAI,EAAE,CAAA;AAErD;;;;;;GAMG;AACH,MAAM,UAAU,kBAAkB,CAAC,gBAAwB;IACzD,OAAO,WAAW,gBAAgB,MAAM,CAAA;AAC1C,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,gBAAgB,CAAC,gBAAwB;IACvD,OAAO,UAAU,gBAAgB,MAAM,CAAA;AACzC,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,GAAG,IAAI,GAAG,EAAkB,CAAA;
|
|
1
|
+
{"version":3,"file":"constants.js","sourceRoot":"","sources":["../src/constants.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAEH,OAAO,kBAAkB,MAAM,sBAAsB,CAAA;AAErD;;GAEG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG;IAC7B,QAAQ,EAAE,QAAQ;IAClB,cAAc,EAAE,QAAQ;IACxB,SAAS,EAAE,QAAQ;IACnB,UAAU,EAAE,QAAQ;IACpB,SAAS,EAAE,QAAQ;IACnB,UAAU,EAAE,QAAQ;IACpB,SAAS,EAAE,QAAQ;IACnB,MAAM,EAAE,QAAQ;IAChB,WAAW,EAAE,QAAQ;IACrB,UAAU,EAAE,QAAQ;IACpB,gBAAgB,EAAE,QAAQ;IAC1B,WAAW,EAAE,QAAQ;IACrB,UAAU,EAAE,QAAQ;IACpB,aAAa,EAAE,QAAQ;IACvB,KAAK,EAAE,QAAQ;IACf,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,YAAY,EAAE,QAAQ;IACtB,OAAO,EAAE,QAAQ;IACjB,OAAO,EAAE,QAAQ;IACjB,KAAK,EAAE,QAAQ;IACf,iBAAiB,EAAE,QAAQ;IAC3B,kBAAkB,EAAE,QAAQ;IAC5B,iBAAiB,EAAE,QAAQ;IAC3B,kBAAkB,EAAE,QAAQ;IAC5B,0BAA0B,EAAE,QAAQ;IACpC,IAAI,EAAE,QAAQ;IACd,+BAA+B;IAC/B,cAAc,EAAE,cAAc,EAAE,KAAK;IACrC,cAAc,EAAE,cAAc,EAAE,KAAK;IACrC,cAAc,EAAE,cAAc,EAAE,KAAK;IACrC,cAAc,EAAE,cAAc,EAAE,KAAK;IACrC,wBAAwB;IACxB,eAAe,EAAE,QAAQ,EAAE,IAAI;IAC/B,oBAAoB,EAAE,QAAQ,EAAE,IAAI;IACpC,oBAAoB,EAAE,QAAQ,EAAE,IAAI;IACpC,kBAAkB,EAAE,QAAQ,EAAE,IAAI;IAClC,WAAW,EAAE,QAAQ,EAAE,IAAI;IAC3B,4BAA4B;IAC5B,qBAAqB,EAAE,QAAQ,EAAE,IAAI;IACrC,kBAAkB,EAAE,QAAQ,EAAE,IAAI;IAClC,gBAAgB,EAAE,QAAQ,EAAE,IAAI;IAChC,eAAe,EAAE,QAAQ,EAAE,IAAI;IAC/B,mBAAmB,EAAE,QAAQ,EAAE,IAAI;IACnC,eAAe,EAAE,QAAQ,EAAE,IAAI;IAC/B,8BAA8B;IAC9B,qBAAqB,EAAE,QAAQ,EAAE,IAAI;IACrC,iBAAiB,EAAE,QAAQ,EAAE,IAAI;IACjC,qBAAqB;IACrB,oBAAoB,EAAE,QAAQ,EAAE,IAAI;IACpC,gBAAgB,EAAE,QAAQ,EAAE,IAAI;IAChC,oBAAoB;IACpB,mBAAmB,EAAE,QAAQ,EAAE,4CAA4C;CACnE,CAAA;AAEV;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAG;IAClC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG;IAC5B,eAAe,CAAC,QAAQ;IACxB,eAAe,CAAC,eAAe,EAAE,eAAe,CAAC,oBAAoB;IACrE,eAAe,CAAC,oBAAoB,EAAE,eAAe,CAAC,kBAAkB;IACxE,eAAe,CAAC,WAAW;IAC3B,eAAe,CAAC,qBAAqB,EAAE,eAAe,CAAC,kBAAkB;IACzE,eAAe,CAAC,gBAAgB,EAAE,eAAe,CAAC,eAAe;IACjE,eAAe,CAAC,mBAAmB,EAAE,eAAe,CAAC,eAAe;IACpE,eAAe,CAAC,qBAAqB,EAAE,eAAe,CAAC,iBAAiB;IACxE,eAAe,CAAC,oBAAoB,EAAE,eAAe,CAAC,gBAAgB;IACtE,eAAe,CAAC,mBAAmB;CAC3B,CAAA;AAEV;;;;;;;;;;;GAWG;AACH,MAAM,CAAC,MAAM,aAAa,GAAG,mFAAmF,CAAA;AAEhH;;;GAGG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAAG,QAAQ,CAAA;AACzC,MAAM,CAAC,MAAM,yBAAyB,GAAG,kBAAkB,CAAC,iBAAiB,CAAC,CAAA;AAE9E;;;;;;GAMG;AACH,MAAM,UAAU,mBAAmB,CAAC,OAA+B;IACjE,OAAO,OAAO,CAAC,SAAS;QACtB,CAAC,CAAC,kBAAkB,CAAC,OAAO,CAAC,SAAS,CAAC;QACvC,CAAC,CAAC,yBAAyB,CAAA;AAC/B,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAU,CAAA;AAEzG;;;;;;;;;;;;;;;GAeG;AACH,MAAM,UAAU,iBAAiB,CAAC,gBAAwB;IACxD,OAAO,UAAU,gBAAgB,OAAO,CAAA;AAC1C,CAAC;AAED;;;;;;;;;;;;;;;GAeG;AACH,MAAM,UAAU,eAAe,CAAC,gBAAwB;IACtD,OAAO,SAAS,gBAAgB,OAAO,CAAA;AACzC,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,IAAI,eAAe,CAAC,IAAI,EAAE,CAAA;AAErD;;;;;;GAMG;AACH,MAAM,UAAU,kBAAkB,CAAC,gBAAwB;IACzD,OAAO,WAAW,gBAAgB,MAAM,CAAA;AAC1C,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,gBAAgB,CAAC,gBAAwB;IACvD,OAAO,UAAU,gBAAgB,MAAM,CAAA;AACzC,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,GAAG,IAAI,GAAG,EAAkB,CAAA;AAC5C,MAAM,oBAAoB,GAAG,IAAI,CAAA;AAEjC;;;;GAIG;AACH,MAAM,UAAU,YAAY,CAAC,OAAe,EAAE,KAAa;IACzD,MAAM,GAAG,GAAG,GAAG,OAAO,KAAK,KAAK,EAAE,CAAA;IAClC,IAAI,EAAE,GAAG,UAAU,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;IAC5B,IAAI,CAAC,EAAE,EAAE,CAAC;QACR,EAAE,GAAG,IAAI,MAAM,CAAC,OAAO,EAAE,KAAK,CAAC,CAAA;QAC/B,IAAI,UAAU,CAAC,IAAI,IAAI,oBAAoB,EAAE,CAAC;YAC5C,oDAAoD;YACpD,MAAM,MAAM,GAAG,UAAU,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,CAAC,KAAM,CAAA;YAC9C,UAAU,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;QAC3B,CAAC;QACD,UAAU,CAAC,GAAG,CAAC,GAAG,EAAE,EAAE,CAAC,CAAA;IACzB,CAAC;IACD,EAAE,CAAC,SAAS,GAAG,CAAC,CAAA;IAChB,OAAO,EAAE,CAAA;AACX,CAAC"}
|
package/dist/dashes.js
CHANGED
|
@@ -2,7 +2,7 @@
|
|
|
2
2
|
* Dash transformation: hyphens → em-dashes, en-dashes, minus signs.
|
|
3
3
|
*/
|
|
4
4
|
import escapeStringRegexp from "escape-string-regexp";
|
|
5
|
-
import { UNICODE_SYMBOLS, DEFAULT_SEPARATOR,
|
|
5
|
+
import { UNICODE_SYMBOLS, DEFAULT_SEPARATOR, LATIN_LETTERS, wordBoundaryStart, wordBoundaryEnd, getEscapedSeparator, cachedRegExp } from "./constants.js";
|
|
6
6
|
const { EN_DASH, EM_DASH, MINUS, LEFT_DOUBLE_QUOTE, RIGHT_DOUBLE_QUOTE, LEFT_SINGLE_QUOTE, RIGHT_SINGLE_QUOTE } = UNICODE_SYMBOLS;
|
|
7
7
|
/**
|
|
8
8
|
* Characters that, when preceding a number, prevent it from being
|
|
@@ -18,7 +18,7 @@ export const months = [
|
|
|
18
18
|
].join("|");
|
|
19
19
|
/** Convert number ranges to en-dash (e.g., "1-5" → "1–5"). */
|
|
20
20
|
export function enDashNumberRange(text, options = {}) {
|
|
21
|
-
const chr =
|
|
21
|
+
const chr = getEscapedSeparator(options);
|
|
22
22
|
const wb = wordBoundaryStart(chr);
|
|
23
23
|
const wbe = wordBoundaryEnd(chr);
|
|
24
24
|
// Escape dash-like chars for lookbehind: prevents matching after dashes (e.g., Llama-2-7B)
|
|
@@ -36,11 +36,11 @@ export function enDashNumberRange(text, options = {}) {
|
|
|
36
36
|
const positiveRangePattern = [
|
|
37
37
|
phoneAreaCode, wb, notAfterDash, rangeStart, "-", rangeEnd, moreSegments, unitSuffix, wbe
|
|
38
38
|
].join("");
|
|
39
|
-
text = text.replace(
|
|
39
|
+
text = text.replace(cachedRegExp(positiveRangePattern, "g"), (match, precedingAreaCode, start, end, following, suffix = "") => {
|
|
40
40
|
if (following)
|
|
41
41
|
return match;
|
|
42
|
-
const startNum = start.replace(
|
|
43
|
-
const endNum = end.replace(
|
|
42
|
+
const startNum = start.replace(cachedRegExp(chr, "g"), "");
|
|
43
|
+
const endNum = end.replace(cachedRegExp(chr, "g"), "");
|
|
44
44
|
if (/^(?:19|20)\d{2}$/.test(startNum) && /^(?:0[1-9]|1[0-2])$/.test(endNum))
|
|
45
45
|
return match;
|
|
46
46
|
// Skip phone number patterns: 3 digits followed by 4 digits with preceding area code
|
|
@@ -55,7 +55,7 @@ export function enDashNumberRange(text, options = {}) {
|
|
|
55
55
|
});
|
|
56
56
|
// Negative ranges: −5-5 → −5–5, −5--2 → −5–−2
|
|
57
57
|
// Separate regex because MINUS isn't a word char, so \b in ${wb} would match after it
|
|
58
|
-
text = text.replace(
|
|
58
|
+
text = text.replace(cachedRegExp(`(?<![${LATIN_LETTERS}])(?<start>${MINUS}\\d[\\d.,]*${chr}?)-(?<neg>-)?(?<end>${chr}?\\d[\\d.,]*)(?<following>(?:${chr}?-${chr}?\\d+)*)(?<suffix>${chr}?[xKBTM])?${wbe}`, "g"), (match, start, neg, end, following, suffix = "") => {
|
|
59
59
|
if (following)
|
|
60
60
|
return match;
|
|
61
61
|
return `${start}${EN_DASH}${neg ? MINUS : ""}${end}${suffix || ""}`;
|
|
@@ -67,10 +67,10 @@ export function enDashDateRange(text, options = {}) {
|
|
|
67
67
|
const dashStyle = options.dashStyle ?? "american";
|
|
68
68
|
if (dashStyle === "none")
|
|
69
69
|
return text;
|
|
70
|
-
const chr =
|
|
70
|
+
const chr = getEscapedSeparator(options);
|
|
71
71
|
const wb = wordBoundaryStart(chr);
|
|
72
72
|
const wbe = wordBoundaryEnd(chr);
|
|
73
|
-
return text.replace(
|
|
73
|
+
return text.replace(cachedRegExp(`${wb}(?<startMonth>${months})(?<startYear>${chr}? \\d{4})?(?<preSep>${chr}?)(?<preSpace> ?)-(?<postSpace> ?)(?<postSep>${chr}?)(?<endMonth>${months})(?<endYear> \\d{4})?${wbe}`, "g"), (...args) => {
|
|
74
74
|
const g = args.at(-1);
|
|
75
75
|
const [pre, post] = dashStyle === "british" ? [" ", " "] : ["", ""];
|
|
76
76
|
return `${g.startMonth}${g.startYear || ""}${g.preSep}${pre}${EN_DASH}${post}${g.postSep}${g.endMonth}${g.endYear || ""}`;
|
|
@@ -78,21 +78,21 @@ export function enDashDateRange(text, options = {}) {
|
|
|
78
78
|
}
|
|
79
79
|
/** Convert hyphens to minus signs in numeric contexts (e.g., "-5" → "−5"). */
|
|
80
80
|
export function minusReplace(text, options = {}) {
|
|
81
|
-
const chr =
|
|
81
|
+
const chr = getEscapedSeparator(options);
|
|
82
82
|
// Pattern 1a: Subtraction of negative number (e.g., "5 - -3" → "5 − −3")
|
|
83
83
|
// Must come before Pattern 1b so the negative hyphen is consumed first
|
|
84
|
-
text = text.replaceAll(
|
|
84
|
+
text = text.replaceAll(cachedRegExp(`(?<=\\d${chr}?) - -(?<num>${chr}?\\d*\\.?\\d+)`, "g"), ` ${MINUS} ${MINUS}$<num>`);
|
|
85
85
|
// Pattern 1b: Spaced math subtraction (e.g., "5 - 3" → "5 − 3")
|
|
86
86
|
// Only when preceded by a digit - this distinguishes "5 - 3" from "Safari) - 9"
|
|
87
|
-
text = text.replaceAll(
|
|
87
|
+
text = text.replaceAll(cachedRegExp(`(?<=\\d${chr}?) - (?<num>${chr}?\\d*\\.?\\d+)`, "g"), ` ${MINUS} $<num>`);
|
|
88
88
|
// Pattern 2: Direct negative numbers (e.g., "-5" → "−5", "(-3)" → "(−3)")
|
|
89
89
|
// Match after: start of line, whitespace, (, or quotes (straight or curly)
|
|
90
90
|
// No space allowed between hyphen and digit
|
|
91
|
-
text = text.replaceAll(
|
|
91
|
+
text = text.replaceAll(cachedRegExp(`(?<before>^|[\\s\\("${LEFT_DOUBLE_QUOTE}${RIGHT_DOUBLE_QUOTE}])-(?<num>\\d*\\.?\\d+)`, "gm"), `$<before>${MINUS}$<num>`);
|
|
92
92
|
// Pattern 2b: Direct negative numbers after separator boundary (e.g., <em>-5</em>)
|
|
93
93
|
// Only when no word character precedes the separator — prevents both
|
|
94
94
|
// "2{SEP}-3" in "1-2-3" and "GPT{SEP}-3" from being misidentified as negative 3
|
|
95
|
-
text = text.replaceAll(
|
|
95
|
+
text = text.replaceAll(cachedRegExp(`(?<![\\d.,${LATIN_LETTERS}])(?<before>${chr})-(?<num>\\d*\\.?\\d+)`, "gm"), `$<before>${MINUS}$<num>`);
|
|
96
96
|
return text;
|
|
97
97
|
}
|
|
98
98
|
/**
|
|
@@ -108,7 +108,7 @@ function convertParentheticalDashes(text, sep, style) {
|
|
|
108
108
|
// A single plain hyphen directly before a word character (through optional separators) is a
|
|
109
109
|
// suspended/hanging hyphen ("Yes-men and -women"), not a parenthetical dash — skip it.
|
|
110
110
|
// Em/en dashes and multiple hyphens can have zero trailing spaces.
|
|
111
|
-
const spacedDashPattern =
|
|
111
|
+
const spacedDashPattern = cachedRegExp(`(?<=[^\\s]|^)(?<sepBefore>${escapedSep}?)[ ]+(?:[${EN_DASH}${EM_DASH}][-${EN_DASH}${EM_DASH}]*|-{2,}|-(?!${escapedSep}*[${LATIN_LETTERS}\\d]))(?!-*>)[ ]*(?<sepAfter>${escapedSep}?)(?<trailing>[ ]*)(?=\\S|$)`, "g");
|
|
112
112
|
text = text.replace(spacedDashPattern, (_match, sepBefore, sepAfter, trailing) => {
|
|
113
113
|
// For British style (spaced en-dash), preserve trailing spaces after separator
|
|
114
114
|
// For American style (unspaced em-dash), consume trailing spaces (em-dash replaces surrounding space)
|
|
@@ -117,15 +117,15 @@ function convertParentheticalDashes(text, sep, style) {
|
|
|
117
117
|
});
|
|
118
118
|
// Convert dashes at text node boundaries: separator alone precedes dash (e.g., "word{sep}– rest")
|
|
119
119
|
// Requires space after the dash to avoid matching number ranges like "1{sep}-{sep}5"
|
|
120
|
-
text = text.replace(
|
|
120
|
+
text = text.replace(cachedRegExp(`(?<=[^\\s])(?<sepBefore>${escapedSep})[${EN_DASH}${EM_DASH}-]+[ ]+(?<sepAfter>${escapedSep}?)`, "g"), `$<sepBefore>${maybeSpace}${localizedDash}${maybeSpace}$<sepAfter>`);
|
|
121
121
|
// Convert multiple dashes: "word--word" or "word---word" or "quote"--"quote"
|
|
122
122
|
const quoteChars = `"'${LEFT_DOUBLE_QUOTE}${RIGHT_DOUBLE_QUOTE}${LEFT_SINGLE_QUOTE}${RIGHT_SINGLE_QUOTE}`;
|
|
123
|
-
text = text.replace(
|
|
123
|
+
text = text.replace(cachedRegExp(`(?<=[${LATIN_LETTERS}\\d${quoteChars}])(?<sepBefore>${escapedSep}?)[${EN_DASH}${EM_DASH}-]{2,}(?<sepAfter>${escapedSep}?)(?=[${LATIN_LETTERS}${quoteChars} ])`, "g"), `$<sepBefore>${maybeSpace}${localizedDash}${maybeSpace}$<sepAfter>`);
|
|
124
124
|
// Convert dashes at start of line
|
|
125
|
-
text = text.replace(
|
|
125
|
+
text = text.replace(cachedRegExp(`^(?<leadingSep>${escapedSep})?[-]+ `, "gm"), `$<leadingSep>${localizedDash} `);
|
|
126
126
|
// British: convert unspaced em-dashes to spaced en-dashes (word—word → word – word)
|
|
127
127
|
if (style === "british") {
|
|
128
|
-
text = text.replace(
|
|
128
|
+
text = text.replace(cachedRegExp(`(?<=[${LATIN_LETTERS}.!?'"])(?<sepBefore>${escapedSep}?)${EM_DASH}(?<sepAfter>${escapedSep}?)(?=[${LATIN_LETTERS}])`, "g"), `$<sepBefore>${maybeSpace}${localizedDash}${maybeSpace}$<sepAfter>`);
|
|
129
129
|
}
|
|
130
130
|
return text;
|
|
131
131
|
}
|
|
@@ -139,9 +139,9 @@ function convertParentheticalDashes(text, sep, style) {
|
|
|
139
139
|
function normalizeEmDashSpacing(text, sep) {
|
|
140
140
|
const escapedSep = escapeStringRegexp(sep);
|
|
141
141
|
// Remove all spaces around em-dashes
|
|
142
|
-
text = text.replace(
|
|
142
|
+
text = text.replace(cachedRegExp(`(?<before>${escapedSep}?)[ ]*${EM_DASH}[ ]*(?<after>${escapedSep}?)`, "g"), `$<before>${EM_DASH}$<after>`);
|
|
143
143
|
// Preserve space after em-dash at start of line (e.g., attribution)
|
|
144
|
-
text = text.replace(
|
|
144
|
+
text = text.replace(cachedRegExp(`^(?<sep>${escapedSep}?)${EM_DASH}(?<after>[A-Z0-9])`, "gm"), `$<sep>${EM_DASH} $<after>`);
|
|
145
145
|
return text;
|
|
146
146
|
}
|
|
147
147
|
/** Full dash transformation. */
|
package/dist/dashes.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"dashes.js","sourceRoot":"","sources":["../src/dashes.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH,OAAO,kBAAkB,MAAM,sBAAsB,CAAA;AACrD,OAAO,EAAE,eAAe,EAAE,iBAAiB,EAAE,
|
|
1
|
+
{"version":3,"file":"dashes.js","sourceRoot":"","sources":["../src/dashes.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH,OAAO,kBAAkB,MAAM,sBAAsB,CAAA;AACrD,OAAO,EAAE,eAAe,EAAE,iBAAiB,EAAE,aAAa,EAAE,iBAAiB,EAAE,eAAe,EAAE,mBAAmB,EAAE,YAAY,EAAE,MAAM,gBAAgB,CAAA;AAWzJ,MAAM,EAAE,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,iBAAiB,EAAE,kBAAkB,EAAE,iBAAiB,EAAE,kBAAkB,EAAE,GAAG,eAAe,CAAA;AAEjI;;;;GAIG;AACH,MAAM,CAAC,MAAM,6BAA6B,GAAG,CAAC,GAAG,EAAE,OAAO,EAAE,OAAO,EAAE,KAAK,CAAU,CAAA;AAEpF,MAAM,CAAC,MAAM,MAAM,GAAG;IACpB,SAAS,EAAE,UAAU,EAAE,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM;IACtD,MAAM,EAAE,QAAQ,EAAE,WAAW,EAAE,SAAS,EAAE,UAAU,EAAE,UAAU;IAChE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK;IACjC,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK;CACzC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;AAEX,8DAA8D;AAC9D,MAAM,UAAU,iBAAiB,CAAC,IAAY,EAAE,UAAuB,EAAE;IACvE,MAAM,GAAG,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAA;IACxC,MAAM,EAAE,GAAG,iBAAiB,CAAC,GAAG,CAAC,CAAA;IACjC,MAAM,GAAG,GAAG,eAAe,CAAC,GAAG,CAAC,CAAA;IAEhC,2FAA2F;IAC3F,MAAM,UAAU,GAAG,6BAA6B,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,QAAQ,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;IAEzI,2CAA2C;IAC3C,MAAM,UAAU,GAAG,OAAO,CAAA;IAE1B,wDAAwD;IACxD,MAAM,aAAa,GAAG,+CAA+C,CAAA,CAAE,gBAAgB;IACvF,MAAM,YAAY,GAAG,QAAQ,UAAU,GAAG,aAAa,KAAK,CAAA,CAAW,qBAAqB;IAC5F,MAAM,UAAU,GAAG,sBAAsB,UAAU,iBAAiB,GAAG,IAAI,CAAA,CAAE,oBAAoB;IACjG,MAAM,QAAQ,GAAG,UAAU,GAAG,KAAK,UAAU,gBAAgB,CAAA,CAAU,kBAAkB;IACzF,MAAM,YAAY,GAAG,mBAAmB,GAAG,MAAM,KAAK,IAAI,GAAG,UAAU,CAAA,CAAC,yBAAyB;IACjG,MAAM,UAAU,GAAG,aAAa,GAAG,2BAA2B,CAAA,CAAS,eAAe;IAEtF,mDAAmD;IACnD,MAAM,oBAAoB,GAAG;QAC3B,aAAa,EAAE,EAAE,EAAE,YAAY,EAAE,UAAU,EAAE,GAAG,EAAE,QAAQ,EAAE,YAAY,EAAE,UAAU,EAAE,GAAG;KAC1F,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;IAEV,IAAI,GAAG,IAAI,CAAC,OAAO,CACjB,YAAY,CAAC,oBAAoB,EAAE,GAAG,CAAC,EACvC,CAAC,KAAK,EAAE,iBAAiB,EAAE,KAAK,EAAE,GAAG,EAAE,SAAS,EAAE,MAAM,GAAG,EAAE,EAAE,EAAE;QAC/D,IAAI,SAAS;YAAE,OAAO,KAAK,CAAA;QAC3B,MAAM,QAAQ,GAAG,KAAK,CAAC,OAAO,CAAC,YAAY,CAAC,GAAG,EAAE,GAAG,CAAC,EAAE,EAAE,CAAC,CAAA;QAC1D,MAAM,MAAM,GAAG,GAAG,CAAC,OAAO,CAAC,YAAY,CAAC,GAAG,EAAE,GAAG,CAAC,EAAE,EAAE,CAAC,CAAA;QACtD,IAAI,kBAAkB,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,qBAAqB,CAAC,IAAI,CAAC,MAAM,CAAC;YAAE,OAAO,KAAK,CAAA;QACzF,qFAAqF;QACrF,gFAAgF;QAChF,IAAI,iBAAiB,IAAI,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,SAAS,CAAC,IAAI,CAAC,MAAM,CAAC;YAAE,OAAO,KAAK,CAAA;QACzF,+DAA+D;QAC/D,sDAAsD;QACtD,IAAI,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,SAAS,CAAC,IAAI,CAAC,MAAM,CAAC;YAAE,OAAO,KAAK,CAAA;QAChE,OAAO,GAAG,iBAAiB,IAAI,EAAE,GAAG,KAAK,GAAG,OAAO,GAAG,GAAG,GAAG,MAAM,IAAI,EAAE,EAAE,CAAA;IAC5E,CAAC,CACF,CAAA;IAED,8CAA8C;IAC9C,sFAAsF;IACtF,IAAI,GAAG,IAAI,CAAC,OAAO,CACjB,YAAY,CACV,QAAQ,aAAa,cAAc,KAAK,cAAc,GAAG,uBAAuB,GAAG,gCAAgC,GAAG,KAAK,GAAG,qBAAqB,GAAG,aAAa,GAAG,EAAE,EACxK,GAAG,CACJ,EACD,CAAC,KAAK,EAAE,KAAK,EAAE,GAAG,EAAE,GAAG,EAAE,SAAS,EAAE,MAAM,GAAG,EAAE,EAAE,EAAE;QACjD,IAAI,SAAS;YAAE,OAAO,KAAK,CAAA;QAC3B,OAAO,GAAG,KAAK,GAAG,OAAO,GAAG,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,GAAG,GAAG,GAAG,MAAM,IAAI,EAAE,EAAE,CAAA;IACrE,CAAC,CACF,CAAA;IAED,OAAO,IAAI,CAAA;AACb,CAAC;AAED,iFAAiF;AACjF,MAAM,UAAU,eAAe,CAAC,IAAY,EAAE,UAAuB,EAAE;IACrE,MAAM,SAAS,GAAG,OAAO,CAAC,SAAS,IAAI,UAAU,CAAA;IACjD,IAAI,SAAS,KAAK,MAAM;QAAE,OAAO,IAAI,CAAA;IACrC,MAAM,GAAG,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAA;IACxC,MAAM,EAAE,GAAG,iBAAiB,CAAC,GAAG,CAAC,CAAA;IACjC,MAAM,GAAG,GAAG,eAAe,CAAC,GAAG,CAAC,CAAA;IAEhC,OAAO,IAAI,CAAC,OAAO,CACjB,YAAY,CAAC,GAAG,EAAE,iBAAiB,MAAM,iBAAiB,GAAG,uBAAuB,GAAG,gDAAgD,GAAG,iBAAiB,MAAM,wBAAwB,GAAG,EAAE,EAAE,GAAG,CAAC,EACpM,CAAC,GAAG,IAAI,EAAE,EAAE;QACV,MAAM,CAAC,GAAG,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,CAA2B,CAAA;QAC/C,MAAM,CAAC,GAAG,EAAE,IAAI,CAAC,GAAG,SAAS,KAAK,SAAS,CAAC,CAAC,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,CAAA;QACnE,OAAO,GAAG,CAAC,CAAC,UAAU,GAAG,CAAC,CAAC,SAAS,IAAI,EAAE,GAAG,CAAC,CAAC,MAAM,GAAG,GAAG,GAAG,OAAO,GAAG,IAAI,GAAG,CAAC,CAAC,OAAO,GAAG,CAAC,CAAC,QAAQ,GAAG,CAAC,CAAC,OAAO,IAAI,EAAE,EAAE,CAAA;IAC3H,CAAC,CACF,CAAA;AACH,CAAC;AAED,8EAA8E;AAC9E,MAAM,UAAU,YAAY,CAAC,IAAY,EAAE,UAAuB,EAAE;IAClE,MAAM,GAAG,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAA;IAExC,yEAAyE;IACzE,uEAAuE;IACvE,IAAI,GAAG,IAAI,CAAC,UAAU,CACpB,YAAY,CAAC,UAAU,GAAG,gBAAgB,GAAG,gBAAgB,EAAE,GAAG,CAAC,EACnE,IAAI,KAAK,IAAI,KAAK,QAAQ,CAC3B,CAAA;IAED,gEAAgE;IAChE,gFAAgF;IAChF,IAAI,GAAG,IAAI,CAAC,UAAU,CACpB,YAAY,CAAC,UAAU,GAAG,eAAe,GAAG,gBAAgB,EAAE,GAAG,CAAC,EAClE,IAAI,KAAK,SAAS,CACnB,CAAA;IAED,0EAA0E;IAC1E,2EAA2E;IAC3E,4CAA4C;IAC5C,IAAI,GAAG,IAAI,CAAC,UAAU,CACpB,YAAY,CAAC,uBAAuB,iBAAiB,GAAG,kBAAkB,yBAAyB,EAAE,IAAI,CAAC,EAC1G,YAAY,KAAK,QAAQ,CAC1B,CAAA;IAED,mFAAmF;IACnF,qEAAqE;IACrE,gFAAgF;IAChF,IAAI,GAAG,IAAI,CAAC,UAAU,CACpB,YAAY,CAAC,aAAa,aAAa,eAAe,GAAG,wBAAwB,EAAE,IAAI,CAAC,EACxF,YAAY,KAAK,QAAQ,CAC1B,CAAA;IAED,OAAO,IAAI,CAAA;AACb,CAAC;AAED;;;GAGG;AACH,SAAS,0BAA0B,CAAC,IAAY,EAAE,GAAW,EAAE,KAAgB;IAC7E,MAAM,aAAa,GAAG,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,OAAO,CAAA;IAC7D,MAAM,UAAU,GAAG,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAA;IACjD,MAAM,UAAU,GAAG,kBAAkB,CAAC,GAAG,CAAC,CAAA;IAE1C,wDAAwD;IACxD,sGAAsG;IACtG,4FAA4F;IAC5F,uFAAuF;IACvF,mEAAmE;IACnE,MAAM,iBAAiB,GAAG,YAAY,CACpC,6BAA6B,UAAU,aAAa,OAAO,GAAG,OAAO,MAAM,OAAO,GAAG,OAAO,gBAAgB,UAAU,KAAK,aAAa,gCAAgC,UAAU,8BAA8B,EAAE,GAAG,CACtN,CAAA;IACD,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,iBAAiB,EAAE,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,EAAE,QAAQ,EAAE,EAAE;QAC/E,+EAA+E;QAC/E,sGAAsG;QACtG,MAAM,YAAY,GAAG,UAAU,IAAI,QAAQ,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAA;QAC3D,OAAO,GAAG,SAAS,GAAG,UAAU,GAAG,aAAa,GAAG,UAAU,GAAG,QAAQ,GAAG,YAAY,EAAE,CAAA;IAC3F,CAAC,CAAC,CAAA;IACF,kGAAkG;IAClG,qFAAqF;IACrF,IAAI,GAAG,IAAI,CAAC,OAAO,CACjB,YAAY,CAAC,2BAA2B,UAAU,KAAK,OAAO,GAAG,OAAO,sBAAsB,UAAU,IAAI,EAAE,GAAG,CAAC,EAClH,eAAe,UAAU,GAAG,aAAa,GAAG,UAAU,aAAa,CACpE,CAAA;IACD,6EAA6E;IAC7E,MAAM,UAAU,GAAG,KAAK,iBAAiB,GAAG,kBAAkB,GAAG,iBAAiB,GAAG,kBAAkB,EAAE,CAAA;IACzG,IAAI,GAAG,IAAI,CAAC,OAAO,CACjB,YAAY,CAAC,QAAQ,aAAa,MAAM,UAAU,kBAAkB,UAAU,MAAM,OAAO,GAAG,OAAO,qBAAqB,UAAU,SAAS,aAAa,GAAG,UAAU,KAAK,EAAE,GAAG,CAAC,EAClL,eAAe,UAAU,GAAG,aAAa,GAAG,UAAU,aAAa,CACpE,CAAA;IACD,kCAAkC;IAClC,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,kBAAkB,UAAU,SAAS,EAAE,IAAI,CAAC,EAAE,gBAAgB,aAAa,GAAG,CAAC,CAAA;IAChH,oFAAoF;IACpF,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;QACxB,IAAI,GAAG,IAAI,CAAC,OAAO,CACjB,YAAY,CAAC,QAAQ,aAAa,uBAAuB,UAAU,KAAK,OAAO,eAAe,UAAU,SAAS,aAAa,IAAI,EAAE,GAAG,CAAC,EACxI,eAAe,UAAU,GAAG,aAAa,GAAG,UAAU,aAAa,CACpE,CAAA;IACH,CAAC;IACD,OAAO,IAAI,CAAA;AACb,CAAC;AAED;;;;;;GAMG;AACH,SAAS,sBAAsB,CAAC,IAAY,EAAE,GAAW;IACvD,MAAM,UAAU,GAAG,kBAAkB,CAAC,GAAG,CAAC,CAAA;IAE1C,qCAAqC;IACrC,IAAI,GAAG,IAAI,CAAC,OAAO,CACjB,YAAY,CAAC,aAAa,UAAU,SAAS,OAAO,gBAAgB,UAAU,IAAI,EAAE,GAAG,CAAC,EACxF,YAAY,OAAO,UAAU,CAC9B,CAAA;IAED,oEAAoE;IACpE,IAAI,GAAG,IAAI,CAAC,OAAO,CACjB,YAAY,CAAC,WAAW,UAAU,KAAK,OAAO,oBAAoB,EAAE,IAAI,CAAC,EACzE,SAAS,OAAO,WAAW,CAC5B,CAAA;IAED,OAAO,IAAI,CAAA;AACb,CAAC;AAED,gCAAgC;AAChC,MAAM,UAAU,aAAa,CAAC,IAAY,EAAE,UAAuB,EAAE;IACnE,MAAM,GAAG,GAAG,OAAO,CAAC,SAAS,IAAI,iBAAiB,CAAA;IAClD,MAAM,KAAK,GAAG,OAAO,CAAC,SAAS,IAAI,UAAU,CAAA;IAC7C,IAAI,KAAK,KAAK,MAAM;QAAE,OAAO,IAAI,CAAA;IACjC,IAAI,GAAG,YAAY,CAAC,IAAI,EAAE,OAAO,CAAC,CAAA;IAClC,IAAI,GAAG,eAAe,CAAC,IAAI,EAAE,OAAO,CAAC,CAAA;IACrC,IAAI,GAAG,0BAA0B,CAAC,IAAI,EAAE,GAAG,EAAE,KAAK,CAAC,CAAA;IACnD,IAAI,KAAK,KAAK,UAAU;QAAE,IAAI,GAAG,sBAAsB,CAAC,IAAI,EAAE,GAAG,CAAC,CAAA;IAClE,IAAI,GAAG,iBAAiB,CAAC,IAAI,EAAE,OAAO,CAAC,CAAA;IACvC,OAAO,IAAI,CAAA;AACb,CAAC"}
|
package/dist/nbsp.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"nbsp.d.ts","sourceRoot":"","sources":["../src/nbsp.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;
|
|
1
|
+
{"version":3,"file":"nbsp.d.ts","sourceRoot":"","sources":["../src/nbsp.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAGH,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,cAAc,CAAA;AAajD,MAAM,MAAM,WAAW,GAAG,IAAI,CAAC,aAAa,EAAE,WAAW,CAAC,CAAA;AAO1D;;;GAGG;AACH,eAAO,MAAM,KAAK,UA2BjB,CAAA;AAED,eAAO,MAAM,UAAU,UActB,CAAA;AAED,eAAO,MAAM,uBAAuB,UASnC,CAAA;AASD;;;;;GAKG;AACH,wBAAgB,mBAAmB,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,GAAE,WAAgB,GAAG,MAAM,CAOnF;AAED;;;;GAIG;AACH,wBAAgB,wBAAwB,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,GAAE,WAAgB,GAAG,MAAM,CAQxF;AAED;;;;;;GAMG;AACH,wBAAgB,kBAAkB,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,GAAE,WAAgB,GAAG,MAAM,CAOlF;AAED;;;;;GAKG;AACH,wBAAgB,+BAA+B,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,GAAE,WAAgB,GAAG,MAAM,CAO/F;AAED;;;;;GAKG;AACH,wBAAgB,uBAAuB,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,GAAE,WAAgB,GAAG,MAAM,CAOvF;AAED;;;;GAIG;AACH,wBAAgB,mBAAmB,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,GAAE,WAAgB,GAAG,MAAM,CAOnF;AAED;;;;;GAKG;AACH,wBAAgB,yBAAyB,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,GAAE,WAAgB,GAAG,MAAM,CAOzF;AAED;;;;GAIG;AACH,wBAAgB,mBAAmB,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,GAAE,WAAgB,GAAG,MAAM,CAOnF;AAsBD,gEAAgE;AAChE,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,GAAE,WAAgB,GAAG,MAAM,CAK7E"}
|
package/dist/nbsp.js
CHANGED
|
@@ -7,14 +7,8 @@
|
|
|
7
7
|
*
|
|
8
8
|
* @module nbsp
|
|
9
9
|
*/
|
|
10
|
-
import
|
|
11
|
-
import { UNICODE_SYMBOLS, ESCAPED_DEFAULT_SEPARATOR, LATIN_LETTERS, wordBoundaryEnd } from "./constants.js";
|
|
10
|
+
import { UNICODE_SYMBOLS, LATIN_LETTERS, wordBoundaryEnd, getEscapedSeparator, cachedRegExp } from "./constants.js";
|
|
12
11
|
const { NBSP, LEFT_DOUBLE_QUOTE, RIGHT_DOUBLE_QUOTE, LEFT_SINGLE_QUOTE, RIGHT_SINGLE_QUOTE, COPYRIGHT, REGISTERED, TRADEMARK, } = UNICODE_SYMBOLS;
|
|
13
|
-
function escapedSeparator(options) {
|
|
14
|
-
return options.separator
|
|
15
|
-
? escapeStringRegexp(options.separator)
|
|
16
|
-
: ESCAPED_DEFAULT_SEPARATOR;
|
|
17
|
-
}
|
|
18
12
|
/** Space pattern matching regular space, tab, and nbsp. */
|
|
19
13
|
const SPACE = `[ \\t${NBSP}]`;
|
|
20
14
|
const UNICODE_UPPERCASE = "\\p{Lu}";
|
|
@@ -88,8 +82,8 @@ const COPYRIGHT_SYMBOLS = `[${COPYRIGHT}${REGISTERED}${TRADEMARK}]`;
|
|
|
88
82
|
* @example "a cat" → "a\u00A0cat", "I am" → "I\u00A0am"
|
|
89
83
|
*/
|
|
90
84
|
export function nbspAfterShortWords(text, options = {}) {
|
|
91
|
-
const sep =
|
|
92
|
-
const pattern =
|
|
85
|
+
const sep = getEscapedSeparator(options);
|
|
86
|
+
const pattern = cachedRegExp(`(?<=^|${SPACE}|${PUNCTUATION_OR_QUOTE}|>)(?<shortWord>[${LATIN_LETTERS}]{1,2})(?<marker>${sep}?)${SPACE}`, "gmu");
|
|
93
87
|
return text.replace(pattern, `$<shortWord>$<marker>${NBSP}`);
|
|
94
88
|
}
|
|
95
89
|
/**
|
|
@@ -98,9 +92,9 @@ export function nbspAfterShortWords(text, options = {}) {
|
|
|
98
92
|
* @example "100 km" → "100\u00A0km", "5 kg" → "5\u00A0kg"
|
|
99
93
|
*/
|
|
100
94
|
export function nbspBetweenNumberAndUnit(text, options = {}) {
|
|
101
|
-
const sep =
|
|
95
|
+
const sep = getEscapedSeparator(options);
|
|
102
96
|
const wbe = wordBoundaryEnd(sep);
|
|
103
|
-
const pattern =
|
|
97
|
+
const pattern = cachedRegExp(`(?<digit>\\d)(?<marker1>${sep}?)${SPACE}(?<marker2>${sep}?)(?<unit>${UNIT_PATTERN})${wbe}`, "gm");
|
|
104
98
|
return text.replace(pattern, `$<digit>$<marker1>${NBSP}$<marker2>$<unit>`);
|
|
105
99
|
}
|
|
106
100
|
/**
|
|
@@ -111,8 +105,8 @@ export function nbspBetweenNumberAndUnit(text, options = {}) {
|
|
|
111
105
|
* true end of string.
|
|
112
106
|
*/
|
|
113
107
|
export function nbspBeforeLastWord(text, options = {}) {
|
|
114
|
-
const sep =
|
|
115
|
-
const pattern =
|
|
108
|
+
const sep = getEscapedSeparator(options);
|
|
109
|
+
const pattern = cachedRegExp(`(?<=[\\w${sep}])${SPACE}(?<lastWord>[^\\s${sep}]{1,10})(?<ending>${sep}?(?:\\n\\n|$))`, "g");
|
|
116
110
|
return text.replace(pattern, `${NBSP}$<lastWord>$<ending>`);
|
|
117
111
|
}
|
|
118
112
|
/**
|
|
@@ -122,8 +116,8 @@ export function nbspBeforeLastWord(text, options = {}) {
|
|
|
122
116
|
* @example "Fig. 1" → "Fig.\u00A01", "p. 42" → "p.\u00A042"
|
|
123
117
|
*/
|
|
124
118
|
export function nbspAfterReferenceAbbreviations(text, options = {}) {
|
|
125
|
-
const sep =
|
|
126
|
-
const pattern =
|
|
119
|
+
const sep = getEscapedSeparator(options);
|
|
120
|
+
const pattern = cachedRegExp(`(?<abbrev>${ABBREVIATION_PATTERN})(?<marker>${sep}?)${SPACE}(?=${sep}?\\d)`, "g");
|
|
127
121
|
return text.replace(pattern, `$<abbrev>$<marker>${NBSP}`);
|
|
128
122
|
}
|
|
129
123
|
/**
|
|
@@ -133,8 +127,8 @@ export function nbspAfterReferenceAbbreviations(text, options = {}) {
|
|
|
133
127
|
* @example "§ 5" → "§\u00A05"
|
|
134
128
|
*/
|
|
135
129
|
export function nbspAfterSectionSymbols(text, options = {}) {
|
|
136
|
-
const sep =
|
|
137
|
-
const pattern =
|
|
130
|
+
const sep = getEscapedSeparator(options);
|
|
131
|
+
const pattern = cachedRegExp(`(?<symbol>[§¶])(?<marker>${sep}?)${SPACE}(?=${sep}?\\d)`, "g");
|
|
138
132
|
return text.replace(pattern, `$<symbol>$<marker>${NBSP}`);
|
|
139
133
|
}
|
|
140
134
|
/**
|
|
@@ -143,8 +137,8 @@ export function nbspAfterSectionSymbols(text, options = {}) {
|
|
|
143
137
|
* @example "Dr. Smith" → "Dr.\u00A0Smith", "Mr. Jones" → "Mr.\u00A0Jones"
|
|
144
138
|
*/
|
|
145
139
|
export function nbspAfterHonorifics(text, options = {}) {
|
|
146
|
-
const sep =
|
|
147
|
-
const pattern =
|
|
140
|
+
const sep = getEscapedSeparator(options);
|
|
141
|
+
const pattern = cachedRegExp(`(?<honorific>${HONORIFIC_PATTERN})(?<marker>${sep}?)${SPACE}(?=${sep}?${UNICODE_UPPERCASE})`, "gu");
|
|
148
142
|
return text.replace(pattern, `$<honorific>$<marker>${NBSP}`);
|
|
149
143
|
}
|
|
150
144
|
/**
|
|
@@ -154,8 +148,8 @@ export function nbspAfterHonorifics(text, options = {}) {
|
|
|
154
148
|
* @example "© 2024" → "©\u00A02024", "® Brand" → "®\u00A0Brand"
|
|
155
149
|
*/
|
|
156
150
|
export function nbspAfterCopyrightSymbols(text, options = {}) {
|
|
157
|
-
const sep =
|
|
158
|
-
const pattern =
|
|
151
|
+
const sep = getEscapedSeparator(options);
|
|
152
|
+
const pattern = cachedRegExp(`(?<symbol>${COPYRIGHT_SYMBOLS})(?<marker>${sep}?)${SPACE}(?=${sep}?[\\d${UNICODE_UPPERCASE}])`, "gu");
|
|
159
153
|
return text.replace(pattern, `$<symbol>$<marker>${NBSP}`);
|
|
160
154
|
}
|
|
161
155
|
/**
|
|
@@ -164,8 +158,8 @@ export function nbspAfterCopyrightSymbols(text, options = {}) {
|
|
|
164
158
|
* @example "J. K. Rowling" → "J.\u00A0K.\u00A0Rowling"
|
|
165
159
|
*/
|
|
166
160
|
export function nbspBetweenInitials(text, options = {}) {
|
|
167
|
-
const sep =
|
|
168
|
-
const pattern =
|
|
161
|
+
const sep = getEscapedSeparator(options);
|
|
162
|
+
const pattern = cachedRegExp(`(?<initial>${UNICODE_UPPERCASE}\\.)(?<marker>${sep}?)${SPACE}(?=${sep}?${UNICODE_UPPERCASE})`, "gu");
|
|
169
163
|
return text.replace(pattern, `$<initial>$<marker>${NBSP}`);
|
|
170
164
|
}
|
|
171
165
|
/**
|
package/dist/nbsp.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"nbsp.js","sourceRoot":"","sources":["../src/nbsp.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAEH,OAAO,
|
|
1
|
+
{"version":3,"file":"nbsp.js","sourceRoot":"","sources":["../src/nbsp.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAEH,OAAO,EAAE,eAAe,EAAE,aAAa,EAAE,eAAe,EAAE,mBAAmB,EAAE,YAAY,EAAE,MAAM,gBAAgB,CAAA;AAGnH,MAAM,EACJ,IAAI,EACJ,iBAAiB,EACjB,kBAAkB,EAClB,iBAAiB,EACjB,kBAAkB,EAClB,SAAS,EACT,UAAU,EACV,SAAS,GACV,GAAG,eAAe,CAAA;AAInB,2DAA2D;AAC3D,MAAM,KAAK,GAAG,QAAQ,IAAI,GAAG,CAAA;AAE7B,MAAM,iBAAiB,GAAG,SAAS,CAAA;AAEnC;;;GAGG;AACH,MAAM,CAAC,MAAM,KAAK,GAAG;IACnB,SAAS;IACT,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG;IACzD,OAAO;IACP,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG;IACvC,SAAS;IACT,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG;IACjC,OAAO;IACP,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,GAAG;IAClC,oBAAoB;IACpB,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI;IACvC,UAAU;IACV,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI;IACtE,iBAAiB;IACjB,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG;IAC1D,cAAc;IACd,GAAG;IACH,aAAa;IACb,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG;IAC1B,kBAAkB;IAClB,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI;IACtC,mBAAmB;IACnB,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI;IAC1C,UAAU;IACV,IAAI,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG;IACnB,OAAO;IACP,MAAM,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI;CAC3B,CAAA;AAED,MAAM,CAAC,MAAM,UAAU,GAAG;IACxB,UAAU;IACV,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK;IACtC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK;IAC5C,SAAS;IACT,KAAK,EAAE,MAAM,EAAE,KAAK;IACpB,kBAAkB;IAClB,IAAI,EAAE,IAAI;IACV,uBAAuB;IACvB,KAAK,EAAE,MAAM;IACb,UAAU;IACV,KAAK,EAAE,MAAM;IACb,QAAQ;IACR,KAAK,EAAE,MAAM;CACd,CAAA;AAED,MAAM,CAAC,MAAM,uBAAuB,GAAG;IACrC,kBAAkB;IAClB,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK;IACjC,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK;IAC9B,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI;IAC/B,SAAS;IACT,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK;IACxB,2CAA2C;IAC3C,KAAK;CACN,CAAA;AAED,8BAA8B;AAC9B,MAAM,YAAY,GAAG,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;AACpC,MAAM,iBAAiB,GAAG,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;AACpE,MAAM,oBAAoB,GAAG,uBAAuB,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;AACpF,MAAM,oBAAoB,GAAG,YAAY,iBAAiB,GAAG,kBAAkB,eAAe,iBAAiB,GAAG,kBAAkB,IAAI,CAAA;AACxI,MAAM,iBAAiB,GAAG,IAAI,SAAS,GAAG,UAAU,GAAG,SAAS,GAAG,CAAA;AAEnE;;;;;GAKG;AACH,MAAM,UAAU,mBAAmB,CAAC,IAAY,EAAE,UAAuB,EAAE;IACzE,MAAM,GAAG,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAA;IACxC,MAAM,OAAO,GAAG,YAAY,CAC1B,SAAS,KAAK,IAAI,oBAAoB,oBAAoB,aAAa,oBAAoB,GAAG,KAAK,KAAK,EAAE,EAC1G,KAAK,CACN,CAAA;IACD,OAAO,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,wBAAwB,IAAI,EAAE,CAAC,CAAA;AAC9D,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,wBAAwB,CAAC,IAAY,EAAE,UAAuB,EAAE;IAC9E,MAAM,GAAG,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAA;IACxC,MAAM,GAAG,GAAG,eAAe,CAAC,GAAG,CAAC,CAAA;IAChC,MAAM,OAAO,GAAG,YAAY,CAC1B,2BAA2B,GAAG,KAAK,KAAK,cAAc,GAAG,aAAa,YAAY,IAAI,GAAG,EAAE,EAC3F,IAAI,CACL,CAAA;IACD,OAAO,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,qBAAqB,IAAI,mBAAmB,CAAC,CAAA;AAC5E,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,kBAAkB,CAAC,IAAY,EAAE,UAAuB,EAAE;IACxE,MAAM,GAAG,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAA;IACxC,MAAM,OAAO,GAAG,YAAY,CAC1B,WAAW,GAAG,KAAK,KAAK,oBAAoB,GAAG,qBAAqB,GAAG,gBAAgB,EACvF,GAAG,CACJ,CAAA;IACD,OAAO,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,GAAG,IAAI,sBAAsB,CAAC,CAAA;AAC7D,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,+BAA+B,CAAC,IAAY,EAAE,UAAuB,EAAE;IACrF,MAAM,GAAG,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAA;IACxC,MAAM,OAAO,GAAG,YAAY,CAC1B,aAAa,oBAAoB,cAAc,GAAG,KAAK,KAAK,MAAM,GAAG,OAAO,EAC5E,GAAG,CACJ,CAAA;IACD,OAAO,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,qBAAqB,IAAI,EAAE,CAAC,CAAA;AAC3D,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,uBAAuB,CAAC,IAAY,EAAE,UAAuB,EAAE;IAC7E,MAAM,GAAG,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAA;IACxC,MAAM,OAAO,GAAG,YAAY,CAC1B,4BAA4B,GAAG,KAAK,KAAK,MAAM,GAAG,OAAO,EACzD,GAAG,CACJ,CAAA;IACD,OAAO,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,qBAAqB,IAAI,EAAE,CAAC,CAAA;AAC3D,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,mBAAmB,CAAC,IAAY,EAAE,UAAuB,EAAE;IACzE,MAAM,GAAG,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAA;IACxC,MAAM,OAAO,GAAG,YAAY,CAC1B,gBAAgB,iBAAiB,cAAc,GAAG,KAAK,KAAK,MAAM,GAAG,IAAI,iBAAiB,GAAG,EAC7F,IAAI,CACL,CAAA;IACD,OAAO,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,wBAAwB,IAAI,EAAE,CAAC,CAAA;AAC9D,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,yBAAyB,CAAC,IAAY,EAAE,UAAuB,EAAE;IAC/E,MAAM,GAAG,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAA;IACxC,MAAM,OAAO,GAAG,YAAY,CAC1B,aAAa,iBAAiB,cAAc,GAAG,KAAK,KAAK,MAAM,GAAG,QAAQ,iBAAiB,IAAI,EAC/F,IAAI,CACL,CAAA;IACD,OAAO,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,qBAAqB,IAAI,EAAE,CAAC,CAAA;AAC3D,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,mBAAmB,CAAC,IAAY,EAAE,UAAuB,EAAE;IACzE,MAAM,GAAG,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAA;IACxC,MAAM,OAAO,GAAG,YAAY,CAC1B,cAAc,iBAAiB,iBAAiB,GAAG,KAAK,KAAK,MAAM,GAAG,IAAI,iBAAiB,GAAG,EAC9F,IAAI,CACL,CAAA;IACD,OAAO,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,sBAAsB,IAAI,EAAE,CAAC,CAAA;AAC5D,CAAC;AAID;;;;;;GAMG;AACH,MAAM,eAAe,GAAa;IAChC,mBAAmB;IACnB,+BAA+B;IAC/B,uBAAuB;IACvB,yBAAyB;IACzB,mBAAmB;IACnB,wBAAwB;IACxB,mBAAmB;IACnB,kBAAkB;CACnB,CAAA;AAED,gEAAgE;AAChE,MAAM,UAAU,aAAa,CAAC,IAAY,EAAE,UAAuB,EAAE;IACnE,KAAK,MAAM,EAAE,IAAI,eAAe,EAAE,CAAC;QACjC,IAAI,GAAG,EAAE,CAAC,IAAI,EAAE,OAAO,CAAC,CAAA;IAC1B,CAAC;IACD,OAAO,IAAI,CAAA;AACb,CAAC"}
|
package/dist/quotes.js
CHANGED
|
@@ -2,7 +2,7 @@
|
|
|
2
2
|
* Smart quote transformation: straight quotes → curly quotes.
|
|
3
3
|
*/
|
|
4
4
|
import escapeStringRegexp from "escape-string-regexp";
|
|
5
|
-
import { UNICODE_SYMBOLS, DEFAULT_SEPARATOR, LATIN_LETTERS, TERMINAL_PUNCTUATION } from "./constants.js";
|
|
5
|
+
import { UNICODE_SYMBOLS, DEFAULT_SEPARATOR, LATIN_LETTERS, TERMINAL_PUNCTUATION, cachedRegExp } from "./constants.js";
|
|
6
6
|
const { EM_DASH, LEFT_DOUBLE_QUOTE, RIGHT_DOUBLE_QUOTE, LEFT_SINGLE_QUOTE, RIGHT_SINGLE_QUOTE, MODIFIER_LETTER_APOSTROPHE, } = UNICODE_SYMBOLS;
|
|
7
7
|
/** Joined string of all terminal punctuation characters for use in regex character classes. */
|
|
8
8
|
const TERMINAL_PUNCTUATION_CLASS = TERMINAL_PUNCTUATION.join("");
|
|
@@ -13,8 +13,8 @@ function convertSingleQuotes(text, sep) {
|
|
|
13
13
|
// Only match straight quotes, not already-converted curly quotes
|
|
14
14
|
const singleQuoteChars = `'${LEFT_SINGLE_QUOTE}${RIGHT_SINGLE_QUOTE}${MODIFIER_LETTER_APOSTROPHE}`;
|
|
15
15
|
const singleQuoteOrWord = `[${singleQuoteChars}\\w]`;
|
|
16
|
-
text = text.replace(
|
|
17
|
-
text = text.replace(
|
|
16
|
+
text = text.replace(cachedRegExp(`(?<!${singleQuoteOrWord})''(?!${singleQuoteOrWord})`, "g"), `${LEFT_SINGLE_QUOTE}${RIGHT_SINGLE_QUOTE}`);
|
|
17
|
+
text = text.replace(cachedRegExp(`(?<!${singleQuoteOrWord})'(\\s+)'(?!${singleQuoteOrWord})`, "g"), `${LEFT_SINGLE_QUOTE}$1${RIGHT_SINGLE_QUOTE}`);
|
|
18
18
|
const afterEndingSinglePatterns = `\\s\\.!?;,\\)${EM_DASH}\\-\\]"`;
|
|
19
19
|
// Full pattern with optional 's' for lookahead detection in apostropheRegex
|
|
20
20
|
const afterEndingSingle = `(?=${escapedSep}?(?:s${escapedSep}?)?(?:[${afterEndingSinglePatterns}]|$))`;
|
|
@@ -22,24 +22,24 @@ function convertSingleQuotes(text, sep) {
|
|
|
22
22
|
// the general rules produced LSQ+n+RSQ which was fine. But MLA requires both
|
|
23
23
|
// quotes to be MLA (semantic apostrophes), and the general rules can't achieve
|
|
24
24
|
// that: neither quote is in a contraction context (no Latin letter on both sides).
|
|
25
|
-
text = text.replace(
|
|
25
|
+
text = text.replace(cachedRegExp(`(?<=\\w${escapedSep}? )['${RIGHT_SINGLE_QUOTE}]n['${RIGHT_SINGLE_QUOTE}](?= ${escapedSep}?\\w)`, "gm"), `${MODIFIER_LETTER_APOSTROPHE}n${MODIFIER_LETTER_APOSTROPHE}`);
|
|
26
26
|
// Possessive: 's followed by ending context (e.g., dog's) → U+02BC
|
|
27
27
|
const afterPossessive = `(?=${escapedSep}?s${escapedSep}?(?:[${afterEndingSinglePatterns}]|$))`;
|
|
28
28
|
const possessiveSingle = `(?<=[^\\s${LEFT_DOUBLE_QUOTE}'])['${RIGHT_SINGLE_QUOTE}]${afterPossessive}`;
|
|
29
|
-
text = text.replace(
|
|
29
|
+
text = text.replace(cachedRegExp(possessiveSingle, "gm"), MODIFIER_LETTER_APOSTROPHE);
|
|
30
30
|
// Closing single quote: ending context without 's' → U+2019
|
|
31
31
|
const afterClosingSingle = `(?=${escapedSep}?(?:[${afterEndingSinglePatterns}]|$))`;
|
|
32
32
|
const closingSingle = `(?<=[^\\s${LEFT_DOUBLE_QUOTE}'])[']${afterClosingSingle}`;
|
|
33
|
-
text = text.replace(
|
|
33
|
+
text = text.replace(cachedRegExp(closingSingle, "gm"), RIGHT_SINGLE_QUOTE);
|
|
34
34
|
const contraction = `(?<=[${LATIN_LETTERS}])['${RIGHT_SINGLE_QUOTE}${MODIFIER_LETTER_APOSTROPHE}](?=${escapedSep}?[${LATIN_LETTERS}])`;
|
|
35
|
-
text = text.replace(
|
|
35
|
+
text = text.replace(cachedRegExp(contraction, "gm"), MODIFIER_LETTER_APOSTROPHE);
|
|
36
36
|
const apostropheWhitelist = `(?=n${MODIFIER_LETTER_APOSTROPHE} )`;
|
|
37
37
|
const endQuoteNotContraction = `(?!${contraction})[${RIGHT_SINGLE_QUOTE}${MODIFIER_LETTER_APOSTROPHE}]${afterEndingSingle}`;
|
|
38
38
|
// Limit lookahead scan to 1000 chars to prevent catastrophic backtracking on pathological inputs
|
|
39
|
-
const apostropheRegex =
|
|
39
|
+
const apostropheRegex = cachedRegExp(`(?<=^|[^\\w])['](${apostropheWhitelist}|(?![^${LEFT_SINGLE_QUOTE}'\\n]{0,1000}${endQuoteNotContraction}))`, "gm");
|
|
40
40
|
text = text.replace(apostropheRegex, MODIFIER_LETTER_APOSTROPHE);
|
|
41
41
|
const beginningSingle = `(?<beforeContext>(?:^|[\\s${LEFT_DOUBLE_QUOTE}${RIGHT_DOUBLE_QUOTE}${EM_DASH}\\-\\(])${escapedSep}?)['](?=${escapedSep}?\\S)`;
|
|
42
|
-
text = text.replace(
|
|
42
|
+
text = text.replace(cachedRegExp(beginningSingle, "gm"), `$<beforeContext>${LEFT_SINGLE_QUOTE}`);
|
|
43
43
|
text = convertUnmatchedPluralPossessives(text, sep);
|
|
44
44
|
return text;
|
|
45
45
|
}
|
|
@@ -50,7 +50,7 @@ function convertSingleQuotes(text, sep) {
|
|
|
50
50
|
*/
|
|
51
51
|
function convertUnmatchedPluralPossessives(text, sep) {
|
|
52
52
|
let singleQuoteBalance = 0;
|
|
53
|
-
return text.replace(
|
|
53
|
+
return text.replace(cachedRegExp(`[${LEFT_SINGLE_QUOTE}${RIGHT_SINGLE_QUOTE}]`, "g"), (match, offset) => {
|
|
54
54
|
if (match === LEFT_SINGLE_QUOTE) {
|
|
55
55
|
singleQuoteBalance++;
|
|
56
56
|
return match;
|
|
@@ -76,13 +76,13 @@ function convertDoubleQuotes(text, sep) {
|
|
|
76
76
|
text = text.replace(/(?<=^|[\s([{])""(?=$|[\s)\]}.!?,;:])/g, `${LEFT_DOUBLE_QUOTE}${RIGHT_DOUBLE_QUOTE}`);
|
|
77
77
|
// Handle whitespace-only quotes " " - require non-quote chars on both sides
|
|
78
78
|
text = text.replace(/(?<=^|[\s([{])"(?<whitespace>\s+)"(?=$|[\s)\]}.!?,;:])/g, `${LEFT_DOUBLE_QUOTE}$<whitespace>${RIGHT_DOUBLE_QUOTE}`);
|
|
79
|
-
const beginningDouble =
|
|
79
|
+
const beginningDouble = cachedRegExp(`(?<=^|[\\s\\(\\/\\[\\{\\-${EM_DASH}${escapedSep}])(?<beforeChr>${escapedSep}?)["](?<afterChr>(?<sepWithPunct>${escapedSep}[ .,])|(?=${escapedSep}?\\.{3}|${escapedSep}?[^\\s\\)\\${EM_DASH},!?${escapedSep};:.\\}]))`, "gm");
|
|
80
80
|
text = text.replace(beginningDouble, `$<beforeChr>${LEFT_DOUBLE_QUOTE}$<afterChr>`);
|
|
81
|
-
text = text.replace(
|
|
81
|
+
text = text.replace(cachedRegExp(`(?<=\\{)(?<sepSpace>${escapedSep}? )?["]`, "g"), `$<sepSpace>${LEFT_DOUBLE_QUOTE}`);
|
|
82
82
|
const endingDouble = `(?<beforeQuote>[^\\s\\(])["]((?<sepAfter>${escapedSep})?)(?=${escapedSep}|[\\s/\\).,;${EM_DASH}:\\-\\}!?s]|$)`;
|
|
83
|
-
text = text.replace(
|
|
84
|
-
text = text.replace(
|
|
85
|
-
text = text.replace(
|
|
83
|
+
text = text.replace(cachedRegExp(endingDouble, "g"), `$<beforeQuote>${RIGHT_DOUBLE_QUOTE}$<sepAfter>`);
|
|
84
|
+
text = text.replace(cachedRegExp(`["](?<sepEnd>${escapedSep}?)$`, "g"), `${RIGHT_DOUBLE_QUOTE}$<sepEnd>`);
|
|
85
|
+
text = text.replace(cachedRegExp(`'(?=${RIGHT_DOUBLE_QUOTE})`, "gu"), RIGHT_SINGLE_QUOTE);
|
|
86
86
|
return text;
|
|
87
87
|
}
|
|
88
88
|
/** Apply American or British punctuation style */
|
|
@@ -90,19 +90,19 @@ function applyPunctuationStyle(text, sep, style) {
|
|
|
90
90
|
const escapedSep = escapeStringRegexp(sep);
|
|
91
91
|
if (style === "american") {
|
|
92
92
|
// Period outside → inside: "Hello". → "Hello."
|
|
93
|
-
const periodOutsideRegex =
|
|
93
|
+
const periodOutsideRegex = cachedRegExp(`(?<![${TERMINAL_PUNCTUATION_CLASS}])(?<sepBefore>${escapedSep}?)(?<quote>[${RIGHT_SINGLE_QUOTE}${RIGHT_DOUBLE_QUOTE}])(?<sepAfter>${escapedSep}?)(?!\\.\\.\\.)\\.`, "g");
|
|
94
94
|
text = text.replace(periodOutsideRegex, "$<sepBefore>.$<quote>$<sepAfter>");
|
|
95
95
|
// Comma outside → inside: "Hello", → "Hello,"
|
|
96
|
-
const commaOutsideRegex =
|
|
96
|
+
const commaOutsideRegex = cachedRegExp(`(?<![${TERMINAL_PUNCTUATION_CLASS}])(?<sepBefore>${escapedSep}?)(?<quote>[${RIGHT_SINGLE_QUOTE}${RIGHT_DOUBLE_QUOTE}])(?<sepAfter>${escapedSep}?),`, "g");
|
|
97
97
|
text = text.replace(commaOutsideRegex, "$<sepBefore>,$<quote>$<sepAfter>");
|
|
98
98
|
}
|
|
99
99
|
else if (style === "british") {
|
|
100
100
|
// Period inside → outside: "Hello." → "Hello".
|
|
101
101
|
// No terminal punctuation guard — "Stop!." inside is always wrong; move the period out.
|
|
102
|
-
const periodInsideRegex =
|
|
102
|
+
const periodInsideRegex = cachedRegExp(`(?<sepBefore>${escapedSep}?)\\.(?<sepMiddle>${escapedSep}?)(?<quote>[${RIGHT_SINGLE_QUOTE}${RIGHT_DOUBLE_QUOTE}])`, "g");
|
|
103
103
|
text = text.replace(periodInsideRegex, "$<sepBefore>$<sepMiddle>$<quote>.");
|
|
104
104
|
// Comma inside → outside: "Hello," → "Hello",
|
|
105
|
-
const commaInsideRegex =
|
|
105
|
+
const commaInsideRegex = cachedRegExp(`,(?<sepAndQuote>${escapedSep}?[${RIGHT_DOUBLE_QUOTE}${RIGHT_SINGLE_QUOTE}])`, "g");
|
|
106
106
|
text = text.replace(commaInsideRegex, "$<sepAndQuote>,");
|
|
107
107
|
}
|
|
108
108
|
return text;
|
package/dist/quotes.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"quotes.js","sourceRoot":"","sources":["../src/quotes.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH,OAAO,kBAAkB,MAAM,sBAAsB,CAAA;AACrD,OAAO,EAAE,eAAe,EAAE,iBAAiB,EAAE,aAAa,EAAE,oBAAoB,EAAE,MAAM,gBAAgB,CAAA;
|
|
1
|
+
{"version":3,"file":"quotes.js","sourceRoot":"","sources":["../src/quotes.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH,OAAO,kBAAkB,MAAM,sBAAsB,CAAA;AACrD,OAAO,EAAE,eAAe,EAAE,iBAAiB,EAAE,aAAa,EAAE,oBAAoB,EAAE,YAAY,EAAE,MAAM,gBAAgB,CAAA;AAEtH,MAAM,EACJ,OAAO,EACP,iBAAiB,EACjB,kBAAkB,EAClB,iBAAiB,EACjB,kBAAkB,EAClB,0BAA0B,GAC3B,GAAG,eAAe,CAAA;AAEnB,+FAA+F;AAC/F,MAAM,0BAA0B,GAAG,oBAAoB,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;AAWhE,qEAAqE;AACrE,SAAS,mBAAmB,CAAC,IAAY,EAAE,GAAW;IACpD,MAAM,UAAU,GAAG,kBAAkB,CAAC,GAAG,CAAC,CAAA;IAE1C,qEAAqE;IACrE,iEAAiE;IACjE,MAAM,gBAAgB,GAAG,IAAI,iBAAiB,GAAG,kBAAkB,GAAG,0BAA0B,EAAE,CAAA;IAClG,MAAM,iBAAiB,GAAG,IAAI,gBAAgB,MAAM,CAAA;IACpD,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,OAAO,iBAAiB,SAAS,iBAAiB,GAAG,EAAE,GAAG,CAAC,EAAE,GAAG,iBAAiB,GAAG,kBAAkB,EAAE,CAAC,CAAA;IAC1I,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,OAAO,iBAAiB,eAAe,iBAAiB,GAAG,EAAE,GAAG,CAAC,EAAE,GAAG,iBAAiB,KAAK,kBAAkB,EAAE,CAAC,CAAA;IAElJ,MAAM,yBAAyB,GAAG,gBAAgB,OAAO,SAAS,CAAA;IAClE,4EAA4E;IAC5E,MAAM,iBAAiB,GAAG,MAAM,UAAU,QAAQ,UAAU,UAAU,yBAAyB,OAAO,CAAA;IAEtG,2EAA2E;IAC3E,6EAA6E;IAC7E,+EAA+E;IAC/E,mFAAmF;IACnF,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,UAAU,UAAU,QAAQ,kBAAkB,OAAO,kBAAkB,QAAQ,UAAU,OAAO,EAAE,IAAI,CAAC,EAAE,GAAG,0BAA0B,IAAI,0BAA0B,EAAE,CAAC,CAAA;IAExM,mEAAmE;IACnE,MAAM,eAAe,GAAG,MAAM,UAAU,KAAK,UAAU,QAAQ,yBAAyB,OAAO,CAAA;IAC/F,MAAM,gBAAgB,GAAG,YAAY,iBAAiB,QAAQ,kBAAkB,IAAI,eAAe,EAAE,CAAA;IACrG,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,gBAAgB,EAAE,IAAI,CAAC,EAAE,0BAA0B,CAAC,CAAA;IAErF,4DAA4D;IAC5D,MAAM,kBAAkB,GAAG,MAAM,UAAU,QAAQ,yBAAyB,OAAO,CAAA;IACnF,MAAM,aAAa,GAAG,YAAY,iBAAiB,SAAS,kBAAkB,EAAE,CAAA;IAChF,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,aAAa,EAAE,IAAI,CAAC,EAAE,kBAAkB,CAAC,CAAA;IAE1E,MAAM,WAAW,GAAG,QAAQ,aAAa,OAAO,kBAAkB,GAAG,0BAA0B,OAAO,UAAU,KAAK,aAAa,IAAI,CAAA;IACtI,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,WAAW,EAAE,IAAI,CAAC,EAAE,0BAA0B,CAAC,CAAA;IAEhF,MAAM,mBAAmB,GAAG,OAAO,0BAA0B,IAAI,CAAA;IACjE,MAAM,sBAAsB,GAAG,MAAM,WAAW,KAAK,kBAAkB,GAAG,0BAA0B,IAAI,iBAAiB,EAAE,CAAA;IAC3H,iGAAiG;IACjG,MAAM,eAAe,GAAG,YAAY,CAClC,oBAAoB,mBAAmB,SAAS,iBAAiB,gBAAgB,sBAAsB,IAAI,EAC3G,IAAI,CACL,CAAA;IACD,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,eAAe,EAAE,0BAA0B,CAAC,CAAA;IAEhE,MAAM,eAAe,GAAG,6BAA6B,iBAAiB,GAAG,kBAAkB,GAAG,OAAO,WAAW,UAAU,WAAW,UAAU,OAAO,CAAA;IACtJ,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,eAAe,EAAE,IAAI,CAAC,EAAE,mBAAmB,iBAAiB,EAAE,CAAC,CAAA;IAEhG,IAAI,GAAG,iCAAiC,CAAC,IAAI,EAAE,GAAG,CAAC,CAAA;IAEnD,OAAO,IAAI,CAAA;AACb,CAAC;AAED;;;;GAIG;AACH,SAAS,iCAAiC,CAAC,IAAY,EAAE,GAAW;IAClE,IAAI,kBAAkB,GAAG,CAAC,CAAA;IAC1B,OAAO,IAAI,CAAC,OAAO,CACjB,YAAY,CAAC,IAAI,iBAAiB,GAAG,kBAAkB,GAAG,EAAE,GAAG,CAAC,EAChE,CAAC,KAAK,EAAE,MAAM,EAAE,EAAE;QAChB,IAAI,KAAK,KAAK,iBAAiB,EAAE,CAAC;YAChC,kBAAkB,EAAE,CAAA;YACpB,OAAO,KAAK,CAAA;QACd,CAAC;QACD,IAAI,kBAAkB,GAAG,CAAC,EAAE,CAAC;YAC3B,kBAAkB,EAAE,CAAA;YACpB,OAAO,KAAK,CAAA;QACd,CAAC;QACD,IAAI,CAAC,GAAG,MAAM,GAAG,CAAC,CAAA;QAClB,OAAO,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,GAAG;YAAE,CAAC,EAAE,CAAA;QACrC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,EAAE,CAAC;YACnD,OAAO,0BAA0B,CAAA;QACnC,CAAC;QACD,OAAO,KAAK,CAAA;IACd,CAAC,CACF,CAAA;AACH,CAAC;AAED,qDAAqD;AACrD,SAAS,mBAAmB,CAAC,IAAY,EAAE,GAAW;IACpD,MAAM,UAAU,GAAG,kBAAkB,CAAC,GAAG,CAAC,CAAA;IAE1C,6EAA6E;IAC7E,oEAAoE;IACpE,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,uCAAuC,EAAE,GAAG,iBAAiB,GAAG,kBAAkB,EAAE,CAAC,CAAA;IACzG,4EAA4E;IAC5E,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,yDAAyD,EAAE,GAAG,iBAAiB,gBAAgB,kBAAkB,EAAE,CAAC,CAAA;IAExI,MAAM,eAAe,GAAG,YAAY,CAClC,4BAA4B,OAAO,GAAG,UAAU,kBAAkB,UAAU,oCAAoC,UAAU,aAAa,UAAU,WAAW,UAAU,cAAc,OAAO,MAAM,UAAU,WAAW,EACtN,IAAI,CACL,CAAA;IACD,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,eAAe,EAAE,eAAe,iBAAiB,aAAa,CAAC,CAAA;IAEnF,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,uBAAuB,UAAU,SAAS,EAAE,GAAG,CAAC,EAAE,cAAc,iBAAiB,EAAE,CAAC,CAAA;IAErH,MAAM,YAAY,GAAG,4CAA4C,UAAU,SAAS,UAAU,eAAe,OAAO,gBAAgB,CAAA;IACpI,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,YAAY,EAAE,GAAG,CAAC,EAAE,iBAAiB,kBAAkB,aAAa,CAAC,CAAA;IAEtG,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,gBAAgB,UAAU,KAAK,EAAE,GAAG,CAAC,EAAE,GAAG,kBAAkB,WAAW,CAAC,CAAA;IACzG,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,OAAO,kBAAkB,GAAG,EAAE,IAAI,CAAC,EAAE,kBAAkB,CAAC,CAAA;IAEzF,OAAO,IAAI,CAAA;AACb,CAAC;AAED,kDAAkD;AAClD,SAAS,qBAAqB,CAAC,IAAY,EAAE,GAAW,EAAE,KAAuB;IAC/E,MAAM,UAAU,GAAG,kBAAkB,CAAC,GAAG,CAAC,CAAA;IAE1C,IAAI,KAAK,KAAK,UAAU,EAAE,CAAC;QACzB,+CAA+C;QAC/C,MAAM,kBAAkB,GAAG,YAAY,CACrC,QAAQ,0BAA0B,kBAAkB,UAAU,eAAe,kBAAkB,GAAG,kBAAkB,iBAAiB,UAAU,oBAAoB,EACnK,GAAG,CACJ,CAAA;QACD,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,kBAAkB,EAAE,kCAAkC,CAAC,CAAA;QAE3E,8CAA8C;QAC9C,MAAM,iBAAiB,GAAG,YAAY,CACpC,QAAQ,0BAA0B,kBAAkB,UAAU,eAAe,kBAAkB,GAAG,kBAAkB,iBAAiB,UAAU,KAAK,EACpJ,GAAG,CACJ,CAAA;QACD,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,iBAAiB,EAAE,kCAAkC,CAAC,CAAA;IAC5E,CAAC;SAAM,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;QAC/B,+CAA+C;QAC/C,wFAAwF;QACxF,MAAM,iBAAiB,GAAG,YAAY,CACpC,gBAAgB,UAAU,qBAAqB,UAAU,eAAe,kBAAkB,GAAG,kBAAkB,IAAI,EACnH,GAAG,CACJ,CAAA;QACD,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,iBAAiB,EAAE,mCAAmC,CAAC,CAAA;QAE3E,8CAA8C;QAC9C,MAAM,gBAAgB,GAAG,YAAY,CACnC,mBAAmB,UAAU,KAAK,kBAAkB,GAAG,kBAAkB,IAAI,EAC7E,GAAG,CACJ,CAAA;QACD,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,gBAAgB,EAAE,iBAAiB,CAAC,CAAA;IAC1D,CAAC;IACD,OAAO,IAAI,CAAA;AACb,CAAC;AAED,+EAA+E;AAC/E,SAAS,aAAa,CAAC,IAAY,EAAE,OAAqB;IACxD,MAAM,GAAG,GAAG,OAAO,CAAC,SAAS,IAAI,iBAAiB,CAAA;IAClD,MAAM,gBAAgB,GAAG,OAAO,CAAC,gBAAgB,IAAI,UAAU,CAAA;IAC/D,IAAI,gBAAgB,KAAK,MAAM;QAAE,OAAO,IAAI,CAAA;IAE5C,IAAI,GAAG,mBAAmB,CAAC,IAAI,EAAE,GAAG,CAAC,CAAA;IACrC,IAAI,GAAG,mBAAmB,CAAC,IAAI,EAAE,GAAG,CAAC,CAAA;IACrC,IAAI,GAAG,qBAAqB,CAAC,IAAI,EAAE,GAAG,EAAE,gBAAgB,CAAC,CAAA;IAEzD,OAAO,IAAI,CAAA;AACb,CAAC;AAED,+CAA+C;AAC/C,MAAM,UAAU,UAAU,CAAC,IAAY,EAAE,UAAwB,EAAE;IACjE,kEAAkE;IAClE,oEAAoE;IACpE,8DAA8D;IAC9D,OAAO,aAAa,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC,UAAU,CAAC,0BAA0B,EAAE,kBAAkB,CAAC,CAAA;AAChG,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,UAAU,mBAAmB,CAAC,IAAY,EAAE,UAAwB,EAAE;IAC1E,OAAO,aAAa,CAAC,IAAI,EAAE,OAAO,CAAC,CAAA;AACrC,CAAC"}
|
package/dist/rehype.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"rehype.d.ts","sourceRoot":"","sources":["../src/rehype.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,KAAK,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,cAAc,EAAE,MAAM,EAAE,WAAW,EAAE,MAAM,MAAM,CAAA;AACpF,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,SAAS,CAAA;AAI1C,OAAO,EAAa,KAAK,gBAAgB,EAAE,MAAM,YAAY,CAAA;AAI7D;;GAEG;AACH,MAAM,WAAW,sBAAuB,SAAQ,gBAAgB;IAC9D;;;;;OAKG;IACH,QAAQ,CAAC,EAAE,MAAM,EAAE,CAAA;IAEnB;;;;;;OAMG;IACH,WAAW,CAAC,EAAE,MAAM,EAAE,CAAA;CACvB;AAwCD;;;;;;;;;;;;GAYG;AACH,wBAAgB,gBAAgB,CAC9B,IAAI,EAAE,OAAO,GAAG,cAAc,EAC9B,UAAU,EAAE,CAAC,CAAC,EAAE,OAAO,KAAK,OAAO,EACnC,KAAK,GAAE,MAAU,GAChB,IAAI,EAAE,CAkBR;AAED;;;;;;;;;;;;GAYG;AACH,wBAAgB,cAAc,CAC5B,IAAI,EAAE,OAAO,EACb,UAAU,GAAE,CAAC,CAAC,EAAE,OAAO,KAAK,OAAqB,GAChD,MAAM,CAIR;AAED;;;;;;;;;;;;GAYG;AACH,wBAAgB,gBAAgB,CAC9B,IAAI,EAAE,MAAM,GAAG,WAAW,EAC1B,KAAK,GAAE,MAAU,GAChB,IAAI,GAAG,IAAI,CAeb;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,sBAAsB,CAAC,KAAK,EAAE,MAAM,GAAG,IAAI,CAuB1D;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAiCG;AACH,wBAAgB,gBAAgB,CAC9B,IAAI,EAAE,OAAO,EACb,WAAW,EAAE,CAAC,KAAK,EAAE,MAAM,KAAK,MAAM,EACtC,UAAU,EAAE,CAAC,KAAK,EAAE,OAAO,KAAK,OAAO,EACvC,SAAS,EAAE,MAAM,EACjB,eAAe,GAAE,OAAe,GAC/B,IAAI,CAgDN;AA6DD;;;;;;;;GAQG;AACH,wBAAgB,4BAA4B,CAC1C,IAAI,EAAE,OAAO,EACb,UAAU,EAAE,CAAC,CAAC,EAAE,OAAO,KAAK,OAAO,EACnC,KAAK,GAAE,MAAU,GAChB,OAAO,EAAE,
|
|
1
|
+
{"version":3,"file":"rehype.d.ts","sourceRoot":"","sources":["../src/rehype.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,KAAK,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,cAAc,EAAE,MAAM,EAAE,WAAW,EAAE,MAAM,MAAM,CAAA;AACpF,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,SAAS,CAAA;AAI1C,OAAO,EAAa,KAAK,gBAAgB,EAAE,MAAM,YAAY,CAAA;AAI7D;;GAEG;AACH,MAAM,WAAW,sBAAuB,SAAQ,gBAAgB;IAC9D;;;;;OAKG;IACH,QAAQ,CAAC,EAAE,MAAM,EAAE,CAAA;IAEnB;;;;;;OAMG;IACH,WAAW,CAAC,EAAE,MAAM,EAAE,CAAA;CACvB;AAwCD;;;;;;;;;;;;GAYG;AACH,wBAAgB,gBAAgB,CAC9B,IAAI,EAAE,OAAO,GAAG,cAAc,EAC9B,UAAU,EAAE,CAAC,CAAC,EAAE,OAAO,KAAK,OAAO,EACnC,KAAK,GAAE,MAAU,GAChB,IAAI,EAAE,CAkBR;AAED;;;;;;;;;;;;GAYG;AACH,wBAAgB,cAAc,CAC5B,IAAI,EAAE,OAAO,EACb,UAAU,GAAE,CAAC,CAAC,EAAE,OAAO,KAAK,OAAqB,GAChD,MAAM,CAIR;AAED;;;;;;;;;;;;GAYG;AACH,wBAAgB,gBAAgB,CAC9B,IAAI,EAAE,MAAM,GAAG,WAAW,EAC1B,KAAK,GAAE,MAAU,GAChB,IAAI,GAAG,IAAI,CAeb;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,sBAAsB,CAAC,KAAK,EAAE,MAAM,GAAG,IAAI,CAuB1D;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAiCG;AACH,wBAAgB,gBAAgB,CAC9B,IAAI,EAAE,OAAO,EACb,WAAW,EAAE,CAAC,KAAK,EAAE,MAAM,KAAK,MAAM,EACtC,UAAU,EAAE,CAAC,KAAK,EAAE,OAAO,KAAK,OAAO,EACvC,SAAS,EAAE,MAAM,EACjB,eAAe,GAAE,OAAe,GAC/B,IAAI,CAgDN;AA6DD;;;;;;;;GAQG;AACH,wBAAgB,4BAA4B,CAC1C,IAAI,EAAE,OAAO,EACb,UAAU,EAAE,CAAC,CAAC,EAAE,OAAO,KAAK,OAAO,EACnC,KAAK,GAAE,MAAU,GAChB,OAAO,EAAE,CA6BX;AAqBD;;;;;;;;;;;;;;;;;;;;;;;;;;GA0BG;AACH,wBAAgB,eAAe,CAC7B,OAAO,GAAE,sBAA2B,GACnC,WAAW,CAAC,IAAI,EAAE,IAAI,CAAC,CAoEzB;AAED,eAAe,eAAe,CAAA"}
|
package/dist/rehype.js
CHANGED
|
@@ -306,7 +306,9 @@ export function collectTransformableElements(node, shouldSkip, depth = 0) {
|
|
|
306
306
|
// Otherwise, recurse into children
|
|
307
307
|
for (const child of node.children) {
|
|
308
308
|
if (child.type === "element") {
|
|
309
|
-
|
|
309
|
+
const childResults = collectTransformableElements(child, shouldSkip, depth + 1);
|
|
310
|
+
for (const r of childResults)
|
|
311
|
+
results.push(r);
|
|
310
312
|
}
|
|
311
313
|
}
|
|
312
314
|
}
|