codekingpro/portable-devtools
115k
1import Diff from './base.js';
2import { longestCommonPrefix, longestCommonSuffix, replacePrefix, replaceSuffix, removePrefix, removeSuffix, maximumOverlap, leadingWs, trailingWs, leadingAndTrailingWs, segment } from '../util/string.js';
3// Based on https://en.wikipedia.org/wiki/Latin_script_in_Unicode
4//
5// Chars/ranges counted as "word" characters by this regex are as follows:
6//
7// + U+00AD Soft hyphen
8// + 00C0–00FF (letters with diacritics from the Latin-1 Supplement), except:
9// - U+00D7 × Multiplication sign
10// - U+00F7 ÷ Division sign
11// + Latin Extended-A, 0100–017F
12// + Latin Extended-B, 0180–024F
13// + IPA Extensions, 0250–02AF
14// + Spacing Modifier Letters, 02B0–02FF, except:
15// - U+02C7 ˇ ˇ Caron
16// - U+02D8 ˘ ˘ Breve
17// - U+02D9 ˙ ˙ Dot Above
18// - U+02DA ˚ ˚ Ring Above
19// - U+02DB ˛ ˛ Ogonek
20// - U+02DC ˜ ˜ Small Tilde
21// - U+02DD ˝ ˝ Double Acute Accent
22// + Latin Extended Additional, 1E00–1EFF
23const extendedWordChars = 'a-zA-Z0-9_\\u{AD}\\u{C0}-\\u{D6}\\u{D8}-\\u{F6}\\u{F8}-\\u{2C6}\\u{2C8}-\\u{2D7}\\u{2DE}-\\u{2FF}\\u{1E00}-\\u{1EFF}';
24// Each token is one of the following:
25// - A punctuation mark plus the surrounding whitespace
26// - A word plus the surrounding whitespace
27// - Pure whitespace (but only in the special case where the entire text
28// is just whitespace)
29//
30// We have to include surrounding whitespace in the tokens because the two
31// alternative approaches produce horribly broken results:
32// * If we just discard the whitespace, we can't fully reproduce the original
33// text from the sequence of tokens and any attempt to render the diff will
34// get the whitespace wrong.
35// * If we have separate tokens for whitespace, then in a typical text every
36// second token will be a single space character. But this often results in
37// the optimal diff between two texts being a perverse one that preserves
38// the spaces between words but deletes and reinserts actual common words.
39// See https://github.com/kpdecker/jsdiff/issues/160#issuecomment-1866099640
40// for an example.
41//
42// Keeping the surrounding whitespace of course has implications for .equals
43// and .join, not just .tokenize.
44// This regex does NOT fully implement the tokenization rules described above.
45// Instead, it gives runs of whitespace their own "token". The tokenize method
46// then handles stitching whitespace tokens onto adjacent word or punctuation
47// tokens.
48const tokenizeIncludingWhitespace = new RegExp(`[${extendedWordChars}]+|\\s+|[^${extendedWordChars}]`, 'ug');
49class WordDiff extends Diff {
50 equals(left, right, options) {
51 if (options.ignoreCase) {
52 left = left.toLowerCase();
53 right = right.toLowerCase();
54 }
55 return left.trim() === right.trim();
56 }
57 tokenize(value, options = {}) {
58 let parts;
59 if (options.intlSegmenter) {
60 const segmenter = options.intlSegmenter;
61 if (segmenter.resolvedOptions().granularity != 'word') {
62 throw new Error('The segmenter passed must have a granularity of "word"');
63 }
64 // We want `parts` to be an array whose elements alternate between being
65 // pure whitespace and being pure non-whitespace. This is ALMOST what the
66 // segments returned by a word-based Intl.Segmenter already look like,
67 // but not quite - see explanation in the docs of our custom segment()
68 // function.
69 parts = segment(value, segmenter);
70 }
71 else {
72 parts = value.match(tokenizeIncludingWhitespace) || [];
73 }
74 const tokens = [];
75 let prevPart = null;
76 parts.forEach(part => {
77 if ((/\s/).test(part)) {
78 if (prevPart == null) {
79 tokens.push(part);
80 }
81 else {
82 tokens.push(tokens.pop() + part);
83 }
84 }
85 else if (prevPart != null && (/\s/).test(prevPart)) {
86 if (tokens[tokens.length - 1] == prevPart) {
87 tokens.push(tokens.pop() + part);
88 }
89 else {
90 tokens.push(prevPart + part);
91 }
92 }
93 else {
94 tokens.push(part);
95 }
96 prevPart = part;
97 });
98 return tokens;
99 }
100 join(tokens) {
101 // Tokens being joined here will always have appeared consecutively in the
102 // same text, so we can simply strip off the leading whitespace from all the
103 // tokens except the first (and except any whitespace-only tokens - but such
104 // a token will always be the first and only token anyway) and then join them
105 // and the whitespace around words and punctuation will end up correct.
106 return tokens.map((token, i) => {
107 if (i == 0) {
108 return token;
109 }
110 else {
111 return token.replace((/^\s+/), '');
112 }
113 }).join('');
114 }
115 postProcess(changes, options) {
116 if (!changes || options.oneChangePerToken) {
117 return changes;
118 }
119 let lastKeep = null;
120 // Change objects representing any insertion or deletion since the last
121 // "keep" change object. There can be at most one of each.
122 let insertion = null;
123 let deletion = null;
124 changes.forEach(change => {
125 if (change.added) {
126 insertion = change;
127 }
128 else if (change.removed) {
129 deletion = change;
130 }
131 else {
132 if (insertion || deletion) { // May be false at start of text
133 dedupeWhitespaceInChangeObjects(lastKeep, deletion, insertion, change, options.intlSegmenter);
134 }
135 lastKeep = change;
136 insertion = null;
137 deletion = null;
138 }
139 });
140 if (insertion || deletion) {
141 dedupeWhitespaceInChangeObjects(lastKeep, deletion, insertion, null, options.intlSegmenter);
142 }
143 return changes;
144 }
145}
146export const wordDiff = new WordDiff();
147export function diffWords(oldStr, newStr, options) {
148 // This option has never been documented and never will be (it's clearer to
149 // just call `diffWordsWithSpace` directly if you need that behavior), but
150 // has existed in jsdiff for a long time, so we retain support for it here
151 // for the sake of backwards compatibility.
152 if ((options === null || options === void 0 ? void 0 : options.ignoreWhitespace) != null && !options.ignoreWhitespace) {
153 return diffWordsWithSpace(oldStr, newStr, options);
154 }
155 return wordDiff.diff(oldStr, newStr, options);
156}
157function dedupeWhitespaceInChangeObjects(startKeep, deletion, insertion, endKeep, segmenter) {
158 // Before returning, we tidy up the leading and trailing whitespace of the
159 // change objects to eliminate cases where trailing whitespace in one object
160 // is repeated as leading whitespace in the next.
161 // Below are examples of the outcomes we want here to explain the code.
162 // I=insert, K=keep, D=delete
163 // 1. diffing 'foo bar baz' vs 'foo baz'
164 // Prior to cleanup, we have K:'foo ' D:' bar ' K:' baz'
165 // After cleanup, we want: K:'foo ' D:'bar ' K:'baz'
166 //
167 // 2. Diffing 'foo bar baz' vs 'foo qux baz'
168 // Prior to cleanup, we have K:'foo ' D:' bar ' I:' qux ' K:' baz'
169 // After cleanup, we want K:'foo ' D:'bar' I:'qux' K:' baz'
170 //
171 // 3. Diffing 'foo\nbar baz' vs 'foo baz'
172 // Prior to cleanup, we have K:'foo ' D:'\nbar ' K:' baz'
173 // After cleanup, we want K'foo' D:'\nbar' K:' baz'
174 //
175 // 4. Diffing 'foo baz' vs 'foo\nbar baz'
176 // Prior to cleanup, we have K:'foo\n' I:'\nbar ' K:' baz'
177 // After cleanup, we ideally want K'foo' I:'\nbar' K:' baz'
178 // but don't actually manage this currently (the pre-cleanup change
179 // objects don't contain enough information to make it possible).
180 //
181 // 5. Diffing 'foo bar baz' vs 'foo baz'
182 // Prior to cleanup, we have K:'foo ' D:' bar ' K:' baz'
183 // After cleanup, we want K:'foo ' D:' bar ' K:'baz'
184 //
185 // Our handling is unavoidably imperfect in the case where there's a single
186 // indel between keeps and the whitespace has changed. For instance, consider
187 // diffing 'foo\tbar\nbaz' vs 'foo baz'. Unless we create an extra change
188 // object to represent the insertion of the space character (which isn't even
189 // a token), we have no way to avoid losing information about the texts'
190 // original whitespace in the result we return. Still, we do our best to
191 // output something that will look sensible if we e.g. print it with
192 // insertions in green and deletions in red.
193 // Between two "keep" change objects (or before the first or after the last
194 // change object), we can have either:
195 // * A "delete" followed by an "insert"
196 // * Just an "insert"
197 // * Just a "delete"
198 // We handle the three cases separately.
199 if (deletion && insertion) {
200 const [oldWsPrefix, oldWsSuffix] = leadingAndTrailingWs(deletion.value, segmenter);
201 const [newWsPrefix, newWsSuffix] = leadingAndTrailingWs(insertion.value, segmenter);
202 if (startKeep) {
203 const commonWsPrefix = longestCommonPrefix(oldWsPrefix, newWsPrefix);
204 startKeep.value = replaceSuffix(startKeep.value, newWsPrefix, commonWsPrefix);
205 deletion.value = removePrefix(deletion.value, commonWsPrefix);
206 insertion.value = removePrefix(insertion.value, commonWsPrefix);
207 }
208 if (endKeep) {
209 const commonWsSuffix = longestCommonSuffix(oldWsSuffix, newWsSuffix);
210 endKeep.value = replacePrefix(endKeep.value, newWsSuffix, commonWsSuffix);
211 deletion.value = removeSuffix(deletion.value, commonWsSuffix);
212 insertion.value = removeSuffix(insertion.value, commonWsSuffix);
213 }
214 }
215 else if (insertion) {
216 // The whitespaces all reflect what was in the new text rather than
217 // the old, so we essentially have no information about whitespace
218 // insertion or deletion. We just want to dedupe the whitespace.
219 // We do that by having each change object keep its trailing
220 // whitespace and deleting duplicate leading whitespace where
221 // present.
222 if (startKeep) {
223 const ws = leadingWs(insertion.value, segmenter);
224 insertion.value = insertion.value.substring(ws.length);
225 }
226 if (endKeep) {
227 const ws = leadingWs(endKeep.value, segmenter);
228 endKeep.value = endKeep.value.substring(ws.length);
229 }
230 // otherwise we've got a deletion and no insertion
231 }
232 else if (startKeep && endKeep) {
233 const newWsFull = leadingWs(endKeep.value, segmenter), [delWsStart, delWsEnd] = leadingAndTrailingWs(deletion.value, segmenter);
234 // Any whitespace that comes straight after startKeep in both the old and
235 // new texts, assign to startKeep and remove from the deletion.
236 const newWsStart = longestCommonPrefix(newWsFull, delWsStart);
237 deletion.value = removePrefix(deletion.value, newWsStart);
238 // Any whitespace that comes straight before endKeep in both the old and
239 // new texts, and hasn't already been assigned to startKeep, assign to
240 // endKeep and remove from the deletion.
241 const newWsEnd = longestCommonSuffix(removePrefix(newWsFull, newWsStart), delWsEnd);
242 deletion.value = removeSuffix(deletion.value, newWsEnd);
243 endKeep.value = replacePrefix(endKeep.value, newWsFull, newWsEnd);
244 // If there's any whitespace from the new text that HASN'T already been
245 // assigned, assign it to the start:
246 startKeep.value = replaceSuffix(startKeep.value, newWsFull, newWsFull.slice(0, newWsFull.length - newWsEnd.length));
247 }
248 else if (endKeep) {
249 // We are at the start of the text. Preserve all the whitespace on
250 // endKeep, and just remove whitespace from the end of deletion to the
251 // extent that it overlaps with the start of endKeep.
252 const endKeepWsPrefix = leadingWs(endKeep.value, segmenter);
253 const deletionWsSuffix = trailingWs(deletion.value, segmenter);
254 const overlap = maximumOverlap(deletionWsSuffix, endKeepWsPrefix);
255 deletion.value = removeSuffix(deletion.value, overlap);
256 }
257 else if (startKeep) {
258 // We are at the END of the text. Preserve all the whitespace on
259 // startKeep, and just remove whitespace from the start of deletion to
260 // the extent that it overlaps with the end of startKeep.
261 const startKeepWsSuffix = trailingWs(startKeep.value, segmenter);
262 const deletionWsPrefix = leadingWs(deletion.value, segmenter);
263 const overlap = maximumOverlap(startKeepWsSuffix, deletionWsPrefix);
264 deletion.value = removePrefix(deletion.value, overlap);
265 }
266}
267class WordsWithSpaceDiff extends Diff {
268 tokenize(value) {
269 // Slightly different to the tokenizeIncludingWhitespace regex used above in
270 // that this one treats each individual newline as a distinct token, rather
271 // than merging them into other surrounding whitespace. This was requested
272 // in https://github.com/kpdecker/jsdiff/issues/180 &
273 // https://github.com/kpdecker/jsdiff/issues/211
274 const regex = new RegExp(`(\\r?\\n)|[${extendedWordChars}]+|[^\\S\\n\\r]+|[^${extendedWordChars}]`, 'ug');
275 return value.match(regex) || [];
276 }
277}
278export const wordsWithSpaceDiff = new WordsWithSpaceDiff();
279export function diffWordsWithSpace(oldStr, newStr, options) {
280 return wordsWithSpaceDiff.diff(oldStr, newStr, options);
281}
282