slopless 0.1.0
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 +64 -0
- package/dist/families/metrics/avg-sentence-length.js +22 -0
- package/dist/families/metrics/avg-sentence-length.js.map +1 -0
- package/dist/families/metrics/coleman-liau.js +27 -0
- package/dist/families/metrics/coleman-liau.js.map +1 -0
- package/dist/families/metrics/flesch-kincaid.js +27 -0
- package/dist/families/metrics/flesch-kincaid.js.map +1 -0
- package/dist/families/metrics/gunning-fog.js +27 -0
- package/dist/families/metrics/gunning-fog.js.map +1 -0
- package/dist/families/metrics/paragraph-length.js +21 -0
- package/dist/families/metrics/paragraph-length.js.map +1 -0
- package/dist/families/metrics/word-repetition.js +46 -0
- package/dist/families/metrics/word-repetition.js.map +1 -0
- package/dist/families/orthography/colon-dramatic.js +117 -0
- package/dist/families/orthography/colon-dramatic.js.map +1 -0
- package/dist/families/orthography/em-dashes.js +41 -0
- package/dist/families/orthography/em-dashes.js.map +1 -0
- package/dist/families/orthography/exclamation-density.js +51 -0
- package/dist/families/orthography/exclamation-density.js.map +1 -0
- package/dist/families/orthography/fake-timestamps.js +103 -0
- package/dist/families/orthography/fake-timestamps.js.map +1 -0
- package/dist/families/orthography/sentence-case.js +66 -0
- package/dist/families/orthography/sentence-case.js.map +1 -0
- package/dist/families/orthography/smart-quotes.js +53 -0
- package/dist/families/orthography/smart-quotes.js.map +1 -0
- package/dist/families/phrases/cliches.js +17 -0
- package/dist/families/phrases/cliches.js.map +1 -0
- package/dist/families/phrases/corporate-speak.js +23 -0
- package/dist/families/phrases/corporate-speak.js.map +1 -0
- package/dist/families/phrases/data/cliches.json +699 -0
- package/dist/families/phrases/data/corporate-speak.json +27 -0
- package/dist/families/phrases/data/prohibited-phrases.json +8 -0
- package/dist/families/phrases/data/skunked-terms.json +10 -0
- package/dist/families/phrases/data/uncomparables.json +57 -0
- package/dist/families/phrases/humble-bragger.js +26 -0
- package/dist/families/phrases/humble-bragger.js.map +1 -0
- package/dist/families/phrases/jargon-faker.js +84 -0
- package/dist/families/phrases/jargon-faker.js.map +1 -0
- package/dist/families/phrases/llm-disclaimer.js +54 -0
- package/dist/families/phrases/llm-disclaimer.js.map +1 -0
- package/dist/families/phrases/prohibited-phrases.js +17 -0
- package/dist/families/phrases/prohibited-phrases.js.map +1 -0
- package/dist/families/phrases/skunked-terms.js +23 -0
- package/dist/families/phrases/skunked-terms.js.map +1 -0
- package/dist/families/phrases/uncomparables.js +78 -0
- package/dist/families/phrases/uncomparables.js.map +1 -0
- package/dist/families/semantic-thinness/patterns/abstract-contrast.json +85 -0
- package/dist/families/semantic-thinness/patterns/abstract-metaphor-claim.json +112 -0
- package/dist/families/semantic-thinness/patterns/body-emotion-shorthand.json +77 -0
- package/dist/families/semantic-thinness/patterns/body-knows.json +80 -0
- package/dist/families/semantic-thinness/patterns/deictic-summary.json +109 -0
- package/dist/families/semantic-thinness/patterns/empty-atmosphere-shift.json +70 -0
- package/dist/families/semantic-thinness/patterns/empty-emotional-weather.json +104 -0
- package/dist/families/semantic-thinness/patterns/empty-scene-state.json +117 -0
- package/dist/families/semantic-thinness/patterns/empty-scene-transition.json +98 -0
- package/dist/families/semantic-thinness/patterns/gaze-choreography.json +87 -0
- package/dist/families/semantic-thinness/patterns/generic-lesson-extraction.json +82 -0
- package/dist/families/semantic-thinness/patterns/generic-pressure-or-stakes.json +107 -0
- package/dist/families/semantic-thinness/patterns/generic-realization.json +113 -0
- package/dist/families/semantic-thinness/patterns/hollow-significance.json +103 -0
- package/dist/families/semantic-thinness/patterns/low-information-physical-blocking.json +113 -0
- package/dist/families/semantic-thinness/patterns/point-is-frame.json +81 -0
- package/dist/families/semantic-thinness/patterns/real-work-begins.json +66 -0
- package/dist/families/semantic-thinness/patterns/silence-as-actor.json +69 -0
- package/dist/families/semantic-thinness/patterns/something-shifted.json +73 -0
- package/dist/families/semantic-thinness/patterns/truth-answer-moves.json +75 -0
- package/dist/families/semantic-thinness/patterns/vague-connective-payoff.json +90 -0
- package/dist/families/semantic-thinness/patterns/vague-threshold-change.json +98 -0
- package/dist/families/semantic-thinness/private/pattern-data-a.js +15 -0
- package/dist/families/semantic-thinness/private/pattern-data-a.js.map +1 -0
- package/dist/families/semantic-thinness/private/pattern-data-b.js +15 -0
- package/dist/families/semantic-thinness/private/pattern-data-b.js.map +1 -0
- package/dist/families/semantic-thinness/private/pattern-data-c.js +9 -0
- package/dist/families/semantic-thinness/private/pattern-data-c.js.map +1 -0
- package/dist/families/semantic-thinness/private/pattern-data-d.js +17 -0
- package/dist/families/semantic-thinness/private/pattern-data-d.js.map +1 -0
- package/dist/families/semantic-thinness/private/pattern-data.js +11 -0
- package/dist/families/semantic-thinness/private/pattern-data.js.map +1 -0
- package/dist/families/semantic-thinness/private/pattern-matcher.js +172 -0
- package/dist/families/semantic-thinness/private/pattern-matcher.js.map +1 -0
- package/dist/families/semantic-thinness/semantic-thinness.js +25 -0
- package/dist/families/semantic-thinness/semantic-thinness.js.map +1 -0
- package/dist/families/syntactic-patterns/authority/authority-padding.js +81 -0
- package/dist/families/syntactic-patterns/authority/authority-padding.js.map +1 -0
- package/dist/families/syntactic-patterns/closers/affirmation-closers.js +48 -0
- package/dist/families/syntactic-patterns/closers/affirmation-closers.js.map +1 -0
- package/dist/families/syntactic-patterns/closers/boilerplate-conclusion.js +131 -0
- package/dist/families/syntactic-patterns/closers/boilerplate-conclusion.js.map +1 -0
- package/dist/families/syntactic-patterns/closers/false-question.js +30 -0
- package/dist/families/syntactic-patterns/closers/false-question.js.map +1 -0
- package/dist/families/syntactic-patterns/closers/summative-closer.js +33 -0
- package/dist/families/syntactic-patterns/closers/summative-closer.js.map +1 -0
- package/dist/families/syntactic-patterns/contrast/blame-reframe.js +59 -0
- package/dist/families/syntactic-patterns/contrast/blame-reframe.js.map +1 -0
- package/dist/families/syntactic-patterns/contrast/contrastive-aphorism.js +203 -0
- package/dist/families/syntactic-patterns/contrast/contrastive-aphorism.js.map +1 -0
- package/dist/families/syntactic-patterns/contrast/negation-reframe.js +20 -0
- package/dist/families/syntactic-patterns/contrast/negation-reframe.js.map +1 -0
- package/dist/families/syntactic-patterns/contrast/private/negation-reframe-matcher.js +218 -0
- package/dist/families/syntactic-patterns/contrast/private/negation-reframe-matcher.js.map +1 -0
- package/dist/families/syntactic-patterns/contrast/private/negation-reframe-parts.js +136 -0
- package/dist/families/syntactic-patterns/contrast/private/negation-reframe-parts.js.map +1 -0
- package/dist/families/syntactic-patterns/generalization/softening-language.js +143 -0
- package/dist/families/syntactic-patterns/generalization/softening-language.js.map +1 -0
- package/dist/families/syntactic-patterns/generalization/universalizing-claims.js +137 -0
- package/dist/families/syntactic-patterns/generalization/universalizing-claims.js.map +1 -0
- package/dist/families/syntactic-patterns/lead-ins/boilerplate-framing.js +104 -0
- package/dist/families/syntactic-patterns/lead-ins/boilerplate-framing.js.map +1 -0
- package/dist/families/syntactic-patterns/lead-ins/generic-signposting.js +230 -0
- package/dist/families/syntactic-patterns/lead-ins/generic-signposting.js.map +1 -0
- package/dist/families/syntactic-patterns/lead-ins/lesson-framing.js +76 -0
- package/dist/families/syntactic-patterns/lead-ins/lesson-framing.js.map +1 -0
- package/dist/families/syntactic-patterns/lead-ins/llm-openers.js +24 -0
- package/dist/families/syntactic-patterns/lead-ins/llm-openers.js.map +1 -0
- package/dist/families/syntactic-patterns/lead-ins/observer-guidance.js +95 -0
- package/dist/families/syntactic-patterns/lead-ins/observer-guidance.js.map +1 -0
- package/dist/families/syntactic-patterns/llm-artifacts/response-wrapper.js +106 -0
- package/dist/families/syntactic-patterns/llm-artifacts/response-wrapper.js.map +1 -0
- package/dist/families/syntactic-patterns/repetition/demonstrative-emphasis.js +251 -0
- package/dist/families/syntactic-patterns/repetition/demonstrative-emphasis.js.map +1 -0
- package/dist/families/syntactic-patterns/repetition/empty-emphasis.js +108 -0
- package/dist/families/syntactic-patterns/repetition/empty-emphasis.js.map +1 -0
- package/dist/families/syntactic-patterns/repetition/fragment-stacking.js +306 -0
- package/dist/families/syntactic-patterns/repetition/fragment-stacking.js.map +1 -0
- package/dist/families/syntactic-patterns/repetition/triple-repeat.js +149 -0
- package/dist/families/syntactic-patterns/repetition/triple-repeat.js.map +1 -0
- package/dist/families/term-policy/recommended-terms.js +87 -0
- package/dist/families/term-policy/recommended-terms.js.map +1 -0
- package/dist/families/term-policy/required-terms.js +32 -0
- package/dist/families/term-policy/required-terms.js.map +1 -0
- package/dist/families/words/data/prohibited-words.json +12 -0
- package/dist/families/words/data/simplicity-pairs.json +7 -0
- package/dist/families/words/hedge-stacking.js +43 -0
- package/dist/families/words/hedge-stacking.js.map +1 -0
- package/dist/families/words/llm-vocabulary.js +33 -0
- package/dist/families/words/llm-vocabulary.js.map +1 -0
- package/dist/families/words/prohibited-words.js +17 -0
- package/dist/families/words/prohibited-words.js.map +1 -0
- package/dist/families/words/simplicity.js +27 -0
- package/dist/families/words/simplicity.js.map +1 -0
- package/dist/index.js +26 -0
- package/dist/index.js.map +1 -0
- package/dist/presets/everything.js +52 -0
- package/dist/presets/everything.js.map +1 -0
- package/dist/registries/metrics.js +15 -0
- package/dist/registries/metrics.js.map +1 -0
- package/dist/registries/orthography.js +15 -0
- package/dist/registries/orthography.js.map +1 -0
- package/dist/registries/phrases.js +19 -0
- package/dist/registries/phrases.js.map +1 -0
- package/dist/registries/semantic-thinness.js +5 -0
- package/dist/registries/semantic-thinness.js.map +1 -0
- package/dist/registries/syntactic-patterns/authority.js +5 -0
- package/dist/registries/syntactic-patterns/authority.js.map +1 -0
- package/dist/registries/syntactic-patterns/closers.js +11 -0
- package/dist/registries/syntactic-patterns/closers.js.map +1 -0
- package/dist/registries/syntactic-patterns/contrast.js +9 -0
- package/dist/registries/syntactic-patterns/contrast.js.map +1 -0
- package/dist/registries/syntactic-patterns/generalization.js +7 -0
- package/dist/registries/syntactic-patterns/generalization.js.map +1 -0
- package/dist/registries/syntactic-patterns/lead-ins.js +13 -0
- package/dist/registries/syntactic-patterns/lead-ins.js.map +1 -0
- package/dist/registries/syntactic-patterns/llm-artifacts.js +5 -0
- package/dist/registries/syntactic-patterns/llm-artifacts.js.map +1 -0
- package/dist/registries/syntactic-patterns/repetition.js +11 -0
- package/dist/registries/syntactic-patterns/repetition.js.map +1 -0
- package/dist/registries/syntactic-patterns.js +17 -0
- package/dist/registries/syntactic-patterns.js.map +1 -0
- package/dist/registries/term-policy.js +7 -0
- package/dist/registries/term-policy.js.map +1 -0
- package/dist/registries/words.js +11 -0
- package/dist/registries/words.js.map +1 -0
- package/dist/shared/matchers/phrases.js +71 -0
- package/dist/shared/matchers/phrases.js.map +1 -0
- package/dist/shared/matchers/prose-patterns.js +104 -0
- package/dist/shared/matchers/prose-patterns.js.map +1 -0
- package/dist/shared/text/document.js +18 -0
- package/dist/shared/text/document.js.map +1 -0
- package/dist/shared/text/normalize.js +30 -0
- package/dist/shared/text/normalize.js.map +1 -0
- package/dist/shared/text/quotes.js +20 -0
- package/dist/shared/text/quotes.js.map +1 -0
- package/dist/shared/text/sections.js +119 -0
- package/dist/shared/text/sections.js.map +1 -0
- package/dist/shared/text/sentences.js +79 -0
- package/dist/shared/text/sentences.js.map +1 -0
- package/dist/shared/text/tokens.js +18 -0
- package/dist/shared/text/tokens.js.map +1 -0
- package/dist/shared/text/traverse.js +10 -0
- package/dist/shared/text/traverse.js.map +1 -0
- package/dist/shared/text/whitespace.js +25 -0
- package/dist/shared/text/whitespace.js.map +1 -0
- package/package.json +115 -0
|
@@ -0,0 +1,108 @@
|
|
|
1
|
+
import { cleanSentence, tokens } from "../../../shared/matchers/prose-patterns.js";
|
|
2
|
+
import { allParagraphSentences } from "../../../shared/text/sections.js";
|
|
3
|
+
const PREFIXES = ["and ", "but ", "so ", "because "];
|
|
4
|
+
const EMPHASIS_REFERENTS = ["part", "bit"];
|
|
5
|
+
const EMPHASIS_QUALIFIERS = ["last", "first", "main"];
|
|
6
|
+
const WEAKENING_REFERENTS = ["pattern", "cycle", "loop"];
|
|
7
|
+
const EMPTY_VIRTUE_LABELS = ["discipline"];
|
|
8
|
+
function isDeictic(token) {
|
|
9
|
+
return token === "that" || token === "this";
|
|
10
|
+
}
|
|
11
|
+
function matchesPartMatters(words) {
|
|
12
|
+
const [first, second, third, fourth, fifth, sixth] = words;
|
|
13
|
+
return ((words.length === 3 &&
|
|
14
|
+
isDeictic(first) &&
|
|
15
|
+
second !== undefined &&
|
|
16
|
+
EMPHASIS_REFERENTS.includes(second) &&
|
|
17
|
+
third === "matters") ||
|
|
18
|
+
(words.length === 4 &&
|
|
19
|
+
isDeictic(first) &&
|
|
20
|
+
second !== undefined &&
|
|
21
|
+
third !== undefined &&
|
|
22
|
+
EMPHASIS_QUALIFIERS.includes(second) &&
|
|
23
|
+
EMPHASIS_REFERENTS.includes(third) &&
|
|
24
|
+
fourth === "matters") ||
|
|
25
|
+
(words.length === 6 &&
|
|
26
|
+
isDeictic(first) &&
|
|
27
|
+
second === "one" &&
|
|
28
|
+
third === "change" &&
|
|
29
|
+
fourth === "helped" &&
|
|
30
|
+
fifth === "a" &&
|
|
31
|
+
sixth === "lot"));
|
|
32
|
+
}
|
|
33
|
+
function matchDeicticIsFrame(words) {
|
|
34
|
+
const [first, second, third, fourth, fifth, sixth] = words;
|
|
35
|
+
const startsWithDeicticIs = isDeictic(first) && second === "is";
|
|
36
|
+
if (words.length === 5 &&
|
|
37
|
+
startsWithDeicticIs &&
|
|
38
|
+
third === "telling" &&
|
|
39
|
+
fourth === "you" &&
|
|
40
|
+
fifth === "something") {
|
|
41
|
+
return "deictic-telling-you-something";
|
|
42
|
+
}
|
|
43
|
+
if (words.length === 5 &&
|
|
44
|
+
startsWithDeicticIs &&
|
|
45
|
+
third === "still" &&
|
|
46
|
+
fourth === "real" &&
|
|
47
|
+
fifth === "change") {
|
|
48
|
+
return "deictic-real-change";
|
|
49
|
+
}
|
|
50
|
+
if (words.length === 6 &&
|
|
51
|
+
startsWithDeicticIs &&
|
|
52
|
+
third === "how" &&
|
|
53
|
+
fourth === "the" &&
|
|
54
|
+
fifth !== undefined &&
|
|
55
|
+
WEAKENING_REFERENTS.includes(fifth) &&
|
|
56
|
+
sixth === "weakens") {
|
|
57
|
+
return "deictic-pattern-weakens";
|
|
58
|
+
}
|
|
59
|
+
if (words.length === 3 &&
|
|
60
|
+
startsWithDeicticIs &&
|
|
61
|
+
third !== undefined &&
|
|
62
|
+
EMPTY_VIRTUE_LABELS.includes(third)) {
|
|
63
|
+
return "deictic-empty-virtue-label";
|
|
64
|
+
}
|
|
65
|
+
return undefined;
|
|
66
|
+
}
|
|
67
|
+
function matchWhatHelpsFrame(words) {
|
|
68
|
+
const [first, second, third, fourth, fifth] = words;
|
|
69
|
+
if (words.length === 5 &&
|
|
70
|
+
first === "what" &&
|
|
71
|
+
second === "helps" &&
|
|
72
|
+
third === "is" &&
|
|
73
|
+
fourth === "not" &&
|
|
74
|
+
fifth === "brilliant") {
|
|
75
|
+
return "what-helps-not-brilliant";
|
|
76
|
+
}
|
|
77
|
+
return undefined;
|
|
78
|
+
}
|
|
79
|
+
function matchEmptyEmphasis(sentence) {
|
|
80
|
+
const words = tokens(cleanSentence(sentence, PREFIXES));
|
|
81
|
+
if (matchesPartMatters(words)) {
|
|
82
|
+
return words.length === 6
|
|
83
|
+
? "deictic-change-helped"
|
|
84
|
+
: "deictic-part-matters";
|
|
85
|
+
}
|
|
86
|
+
return matchDeicticIsFrame(words) ?? matchWhatHelpsFrame(words);
|
|
87
|
+
}
|
|
88
|
+
const rule = (context) => {
|
|
89
|
+
const { Syntax, RuleError, locator, report } = context;
|
|
90
|
+
return {
|
|
91
|
+
[Syntax.Document](node) {
|
|
92
|
+
for (const item of allParagraphSentences(node)) {
|
|
93
|
+
const matched = matchEmptyEmphasis(item.sentence.text);
|
|
94
|
+
if (matched === undefined) {
|
|
95
|
+
continue;
|
|
96
|
+
}
|
|
97
|
+
report(item.paragraph, new RuleError(`Empty emphasis found: ${matched}. Replace the filler line with the actual point.`, {
|
|
98
|
+
padding: locator.range([
|
|
99
|
+
item.source.originalStartFor(item.sentence.start),
|
|
100
|
+
item.source.originalEndFor(item.sentence.end)
|
|
101
|
+
])
|
|
102
|
+
}));
|
|
103
|
+
}
|
|
104
|
+
}
|
|
105
|
+
};
|
|
106
|
+
};
|
|
107
|
+
export default rule;
|
|
108
|
+
//# sourceMappingURL=empty-emphasis.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"empty-emphasis.js","sourceRoot":"","sources":["../../../../src/families/syntactic-patterns/repetition/empty-emphasis.ts"],"names":[],"mappings":"AAEA,OAAO,EACL,aAAa,EACb,MAAM,EACP,MAAM,4CAA4C,CAAC;AACpD,OAAO,EAAE,qBAAqB,EAAE,MAAM,kCAAkC,CAAC;AAEzE,MAAM,QAAQ,GAAG,CAAC,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,UAAU,CAAC,CAAC;AACrD,MAAM,kBAAkB,GAAG,CAAC,MAAM,EAAE,KAAK,CAAC,CAAC;AAC3C,MAAM,mBAAmB,GAAG,CAAC,MAAM,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC;AACtD,MAAM,mBAAmB,GAAG,CAAC,SAAS,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC;AACzD,MAAM,mBAAmB,GAAG,CAAC,YAAY,CAAC,CAAC;AAE3C,SAAS,SAAS,CAAC,KAAyB;IAC1C,OAAO,KAAK,KAAK,MAAM,IAAI,KAAK,KAAK,MAAM,CAAC;AAC9C,CAAC;AAED,SAAS,kBAAkB,CAAC,KAAwB;IAClD,MAAM,CAAC,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,GAAG,KAAK,CAAC;IAE3D,OAAO,CACL,CAAC,KAAK,CAAC,MAAM,KAAK,CAAC;QACjB,SAAS,CAAC,KAAK,CAAC;QAChB,MAAM,KAAK,SAAS;QACpB,kBAAkB,CAAC,QAAQ,CAAC,MAAM,CAAC;QACnC,KAAK,KAAK,SAAS,CAAC;QACtB,CAAC,KAAK,CAAC,MAAM,KAAK,CAAC;YACjB,SAAS,CAAC,KAAK,CAAC;YAChB,MAAM,KAAK,SAAS;YACpB,KAAK,KAAK,SAAS;YACnB,mBAAmB,CAAC,QAAQ,CAAC,MAAM,CAAC;YACpC,kBAAkB,CAAC,QAAQ,CAAC,KAAK,CAAC;YAClC,MAAM,KAAK,SAAS,CAAC;QACvB,CAAC,KAAK,CAAC,MAAM,KAAK,CAAC;YACjB,SAAS,CAAC,KAAK,CAAC;YAChB,MAAM,KAAK,KAAK;YAChB,KAAK,KAAK,QAAQ;YAClB,MAAM,KAAK,QAAQ;YACnB,KAAK,KAAK,GAAG;YACb,KAAK,KAAK,KAAK,CAAC,CACnB,CAAC;AACJ,CAAC;AAED,SAAS,mBAAmB,CAAC,KAAwB;IACnD,MAAM,CAAC,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,GAAG,KAAK,CAAC;IAC3D,MAAM,mBAAmB,GAAG,SAAS,CAAC,KAAK,CAAC,IAAI,MAAM,KAAK,IAAI,CAAC;IAEhE,IACE,KAAK,CAAC,MAAM,KAAK,CAAC;QAClB,mBAAmB;QACnB,KAAK,KAAK,SAAS;QACnB,MAAM,KAAK,KAAK;QAChB,KAAK,KAAK,WAAW,EACrB,CAAC;QACD,OAAO,+BAA+B,CAAC;IACzC,CAAC;IAED,IACE,KAAK,CAAC,MAAM,KAAK,CAAC;QAClB,mBAAmB;QACnB,KAAK,KAAK,OAAO;QACjB,MAAM,KAAK,MAAM;QACjB,KAAK,KAAK,QAAQ,EAClB,CAAC;QACD,OAAO,qBAAqB,CAAC;IAC/B,CAAC;IAED,IACE,KAAK,CAAC,MAAM,KAAK,CAAC;QAClB,mBAAmB;QACnB,KAAK,KAAK,KAAK;QACf,MAAM,KAAK,KAAK;QAChB,KAAK,KAAK,SAAS;QACnB,mBAAmB,CAAC,QAAQ,CAAC,KAAK,CAAC;QACnC,KAAK,KAAK,SAAS,EACnB,CAAC;QACD,OAAO,yBAAyB,CAAC;IACnC,CAAC;IAED,IACE,KAAK,CAAC,MAAM,KAAK,CAAC;QAClB,mBAAmB;QACnB,KAAK,KAAK,SAAS;QACnB,mBAAmB,CAAC,QAAQ,CAAC,KAAK,CAAC,EACnC,CAAC;QACD,OAAO,4BAA4B,CAAC;IACtC,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,mBAAmB,CAAC,KAAwB;IACnD,MAAM,CAAC,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,CAAC,GAAG,KAAK,CAAC;IAEpD,IACE,KAAK,CAAC,MAAM,KAAK,CAAC;QAClB,KAAK,KAAK,MAAM;QAChB,MAAM,KAAK,OAAO;QAClB,KAAK,KAAK,IAAI;QACd,MAAM,KAAK,KAAK;QAChB,KAAK,KAAK,WAAW,EACrB,CAAC;QACD,OAAO,0BAA0B,CAAC;IACpC,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,kBAAkB,CAAC,QAAgB;IAC1C,MAAM,KAAK,GAAG,MAAM,CAAC,aAAa,CAAC,QAAQ,EAAE,QAAQ,CAAC,CAAC,CAAC;IAExD,IAAI,kBAAkB,CAAC,KAAK,CAAC,EAAE,CAAC;QAC9B,OAAO,KAAK,CAAC,MAAM,KAAK,CAAC;YACvB,CAAC,CAAC,uBAAuB;YACzB,CAAC,CAAC,sBAAsB,CAAC;IAC7B,CAAC;IAED,OAAO,mBAAmB,CAAC,KAAK,CAAC,IAAI,mBAAmB,CAAC,KAAK,CAAC,CAAC;AAClE,CAAC;AAED,MAAM,IAAI,GAAuB,CAAC,OAAO,EAAE,EAAE;IAC3C,MAAM,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC;IAEvD,OAAO;QACL,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,IAAqB;YACrC,KAAK,MAAM,IAAI,IAAI,qBAAqB,CAAC,IAAI,CAAC,EAAE,CAAC;gBAC/C,MAAM,OAAO,GAAG,kBAAkB,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;gBACvD,IAAI,OAAO,KAAK,SAAS,EAAE,CAAC;oBAC1B,SAAS;gBACX,CAAC;gBAED,MAAM,CACJ,IAAI,CAAC,SAAS,EACd,IAAI,SAAS,CACX,yBAAyB,OAAO,kDAAkD,EAClF;oBACE,OAAO,EAAE,OAAO,CAAC,KAAK,CAAC;wBACrB,IAAI,CAAC,MAAM,CAAC,gBAAgB,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC;wBACjD,IAAI,CAAC,MAAM,CAAC,cAAc,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC;qBAC9C,CAAC;iBACH,CACF,CACF,CAAC;YACJ,CAAC;QACH,CAAC;KACF,CAAC;AACJ,CAAC,CAAC;AAEF,eAAe,IAAI,CAAC"}
|
|
@@ -0,0 +1,306 @@
|
|
|
1
|
+
import { RuleHelper } from "textlint-rule-helper";
|
|
2
|
+
import { splitSentences } from "../../../shared/text/sentences.js";
|
|
3
|
+
import { sourceText } from "../../../shared/text/traverse.js";
|
|
4
|
+
import { splitWhitespace } from "../../../shared/text/whitespace.js";
|
|
5
|
+
const MAX_FRAGMENT_WORDS = 6;
|
|
6
|
+
const MAX_PAYOFF_WORDS = 28;
|
|
7
|
+
const SUBJECT_WORDS = new Set([
|
|
8
|
+
"i",
|
|
9
|
+
"you",
|
|
10
|
+
"we",
|
|
11
|
+
"they",
|
|
12
|
+
"he",
|
|
13
|
+
"she",
|
|
14
|
+
"it",
|
|
15
|
+
"this",
|
|
16
|
+
"that",
|
|
17
|
+
"these",
|
|
18
|
+
"those"
|
|
19
|
+
]);
|
|
20
|
+
const OBJECT_WORDS = new Set([
|
|
21
|
+
"me",
|
|
22
|
+
"him",
|
|
23
|
+
"her",
|
|
24
|
+
"us",
|
|
25
|
+
"them",
|
|
26
|
+
"it",
|
|
27
|
+
"the",
|
|
28
|
+
"a",
|
|
29
|
+
"an",
|
|
30
|
+
"my",
|
|
31
|
+
"your",
|
|
32
|
+
"our",
|
|
33
|
+
"their"
|
|
34
|
+
]);
|
|
35
|
+
const FINITE_VERBS = new Set([
|
|
36
|
+
"is",
|
|
37
|
+
"are",
|
|
38
|
+
"was",
|
|
39
|
+
"were",
|
|
40
|
+
"am",
|
|
41
|
+
"be",
|
|
42
|
+
"been",
|
|
43
|
+
"being",
|
|
44
|
+
"have",
|
|
45
|
+
"has",
|
|
46
|
+
"had",
|
|
47
|
+
"do",
|
|
48
|
+
"does",
|
|
49
|
+
"did",
|
|
50
|
+
"can",
|
|
51
|
+
"could",
|
|
52
|
+
"will",
|
|
53
|
+
"would",
|
|
54
|
+
"should",
|
|
55
|
+
"may",
|
|
56
|
+
"might",
|
|
57
|
+
"must",
|
|
58
|
+
"shall"
|
|
59
|
+
]);
|
|
60
|
+
const FRAGMENT_LEADS = new Set([
|
|
61
|
+
"too",
|
|
62
|
+
"most",
|
|
63
|
+
"more",
|
|
64
|
+
"less",
|
|
65
|
+
"deeply",
|
|
66
|
+
"completely",
|
|
67
|
+
"possibly",
|
|
68
|
+
"probably",
|
|
69
|
+
"maybe",
|
|
70
|
+
"weird",
|
|
71
|
+
"strange",
|
|
72
|
+
"odd",
|
|
73
|
+
"pure",
|
|
74
|
+
"total",
|
|
75
|
+
"incredibly"
|
|
76
|
+
]);
|
|
77
|
+
const PAYOFF_STARTS = ["more like ", "most ", "then ", "instead "];
|
|
78
|
+
const IMPERATIVE_STARTS = new Set([
|
|
79
|
+
"feed",
|
|
80
|
+
"leave",
|
|
81
|
+
"notice",
|
|
82
|
+
"stop",
|
|
83
|
+
"start",
|
|
84
|
+
"take",
|
|
85
|
+
"keep",
|
|
86
|
+
"get",
|
|
87
|
+
"look",
|
|
88
|
+
"think",
|
|
89
|
+
"try",
|
|
90
|
+
"make",
|
|
91
|
+
"let",
|
|
92
|
+
"give",
|
|
93
|
+
"accept",
|
|
94
|
+
"hold",
|
|
95
|
+
"reduce"
|
|
96
|
+
]);
|
|
97
|
+
const SIMPLE_PAST_VERBS = new Set([
|
|
98
|
+
"ran",
|
|
99
|
+
"went",
|
|
100
|
+
"came",
|
|
101
|
+
"felt",
|
|
102
|
+
"heard",
|
|
103
|
+
"found",
|
|
104
|
+
"made",
|
|
105
|
+
"took",
|
|
106
|
+
"kept",
|
|
107
|
+
"left",
|
|
108
|
+
"thought",
|
|
109
|
+
"knew",
|
|
110
|
+
"got",
|
|
111
|
+
"put",
|
|
112
|
+
"said",
|
|
113
|
+
"told",
|
|
114
|
+
"held",
|
|
115
|
+
"stood",
|
|
116
|
+
"sat",
|
|
117
|
+
"became",
|
|
118
|
+
"wrote",
|
|
119
|
+
"spoke",
|
|
120
|
+
"won",
|
|
121
|
+
"lost",
|
|
122
|
+
"paid",
|
|
123
|
+
"met",
|
|
124
|
+
"read",
|
|
125
|
+
"saw",
|
|
126
|
+
"grew",
|
|
127
|
+
"fell",
|
|
128
|
+
"broke"
|
|
129
|
+
]);
|
|
130
|
+
function isAlphanumeric(character) {
|
|
131
|
+
const lower = character.toLocaleLowerCase("en");
|
|
132
|
+
const upper = character.toLocaleUpperCase("en");
|
|
133
|
+
return (character >= "0" && character <= "9") || lower !== upper;
|
|
134
|
+
}
|
|
135
|
+
function cleanWord(word) {
|
|
136
|
+
const normalized = word.toLocaleLowerCase("en").replaceAll("\u2019", "'");
|
|
137
|
+
let start = 0;
|
|
138
|
+
let end = normalized.length;
|
|
139
|
+
while (start < end &&
|
|
140
|
+
normalized[start] !== "'" &&
|
|
141
|
+
!isAlphanumeric(normalized[start] ?? "")) {
|
|
142
|
+
start += 1;
|
|
143
|
+
}
|
|
144
|
+
while (end > start &&
|
|
145
|
+
normalized[end - 1] !== "'" &&
|
|
146
|
+
!isAlphanumeric(normalized[end - 1] ?? "")) {
|
|
147
|
+
end -= 1;
|
|
148
|
+
}
|
|
149
|
+
return normalized.slice(start, end);
|
|
150
|
+
}
|
|
151
|
+
function sentenceWords(sentence) {
|
|
152
|
+
return splitWhitespace(sentence).map(cleanWord);
|
|
153
|
+
}
|
|
154
|
+
function isFunctionWord(word) {
|
|
155
|
+
return (word === "the" ||
|
|
156
|
+
word === "a" ||
|
|
157
|
+
word === "an" ||
|
|
158
|
+
word === "and" ||
|
|
159
|
+
word === "or" ||
|
|
160
|
+
word === "but" ||
|
|
161
|
+
word === "to" ||
|
|
162
|
+
word === "of" ||
|
|
163
|
+
word === "in" ||
|
|
164
|
+
word === "on" ||
|
|
165
|
+
word === "at" ||
|
|
166
|
+
word === "for");
|
|
167
|
+
}
|
|
168
|
+
function looksLikeModifierPhrase(first, second) {
|
|
169
|
+
return (first.endsWith("ly") ||
|
|
170
|
+
second.endsWith("ive") ||
|
|
171
|
+
second.endsWith("ous") ||
|
|
172
|
+
second.endsWith("al") ||
|
|
173
|
+
second.endsWith("ful") ||
|
|
174
|
+
second.endsWith("less"));
|
|
175
|
+
}
|
|
176
|
+
function looksLikeSubjectDrop(first, second) {
|
|
177
|
+
return (((first.endsWith("ed") && !first.endsWith("eed")) ||
|
|
178
|
+
first.endsWith("ing")) &&
|
|
179
|
+
!isFunctionWord(second));
|
|
180
|
+
}
|
|
181
|
+
function looksLikeSimpleClause(first, second) {
|
|
182
|
+
return (!isFunctionWord(first) &&
|
|
183
|
+
(second.endsWith("ed") || SIMPLE_PAST_VERBS.has(second)));
|
|
184
|
+
}
|
|
185
|
+
function looksLikeBriefImperative(first, second) {
|
|
186
|
+
return (IMPERATIVE_STARTS.has(first) &&
|
|
187
|
+
(OBJECT_WORDS.has(second) || second.endsWith("er") || second.endsWith("ly")));
|
|
188
|
+
}
|
|
189
|
+
function classifyFragment(sentence) {
|
|
190
|
+
const words = sentenceWords(sentence.text);
|
|
191
|
+
if (words.length < 2 || words.length > MAX_FRAGMENT_WORDS) {
|
|
192
|
+
return undefined;
|
|
193
|
+
}
|
|
194
|
+
if (words.some((word) => word === "")) {
|
|
195
|
+
return undefined;
|
|
196
|
+
}
|
|
197
|
+
if (words.some((word) => SUBJECT_WORDS.has(word))) {
|
|
198
|
+
return undefined;
|
|
199
|
+
}
|
|
200
|
+
if (words.some((word) => FINITE_VERBS.has(word))) {
|
|
201
|
+
return undefined;
|
|
202
|
+
}
|
|
203
|
+
const first = words[0];
|
|
204
|
+
const second = words[1];
|
|
205
|
+
if (first === undefined || second === undefined) {
|
|
206
|
+
return undefined;
|
|
207
|
+
}
|
|
208
|
+
if (FRAGMENT_LEADS.has(first) || looksLikeModifierPhrase(first, second)) {
|
|
209
|
+
return "modifier-fragment";
|
|
210
|
+
}
|
|
211
|
+
if (looksLikeSimpleClause(first, second) ||
|
|
212
|
+
looksLikeBriefImperative(first, second)) {
|
|
213
|
+
return undefined;
|
|
214
|
+
}
|
|
215
|
+
if (looksLikeSubjectDrop(first, second)) {
|
|
216
|
+
return "subject-drop";
|
|
217
|
+
}
|
|
218
|
+
return "noun-fragment";
|
|
219
|
+
}
|
|
220
|
+
function looksLikePayoffSentence(sentence) {
|
|
221
|
+
const words = sentenceWords(sentence.text);
|
|
222
|
+
if (words.length <= 2 || words.length > MAX_PAYOFF_WORDS) {
|
|
223
|
+
return false;
|
|
224
|
+
}
|
|
225
|
+
const lowered = sentence.text.toLocaleLowerCase("en");
|
|
226
|
+
if (!PAYOFF_STARTS.some((start) => lowered.startsWith(start))) {
|
|
227
|
+
return false;
|
|
228
|
+
}
|
|
229
|
+
return !words.some((word) => SUBJECT_WORDS.has(word));
|
|
230
|
+
}
|
|
231
|
+
function findFragmentStacks(text) {
|
|
232
|
+
const sentences = splitSentences(text);
|
|
233
|
+
const classifications = sentences.map(classifyFragment);
|
|
234
|
+
const matches = [];
|
|
235
|
+
let sentenceIndex = 0;
|
|
236
|
+
while (sentenceIndex < sentences.length) {
|
|
237
|
+
const firstType = classifications[sentenceIndex];
|
|
238
|
+
const firstSentence = sentences[sentenceIndex];
|
|
239
|
+
if (firstType === undefined || firstSentence === undefined) {
|
|
240
|
+
sentenceIndex += 1;
|
|
241
|
+
continue;
|
|
242
|
+
}
|
|
243
|
+
const runSentences = [firstSentence];
|
|
244
|
+
const fragmentTypes = [firstType];
|
|
245
|
+
let cursor = sentenceIndex + 1;
|
|
246
|
+
while (cursor < sentences.length) {
|
|
247
|
+
const sentence = sentences[cursor];
|
|
248
|
+
const nextType = classifications[cursor];
|
|
249
|
+
if (sentence === undefined) {
|
|
250
|
+
break;
|
|
251
|
+
}
|
|
252
|
+
if (nextType !== undefined) {
|
|
253
|
+
runSentences.push(sentence);
|
|
254
|
+
fragmentTypes.push(nextType);
|
|
255
|
+
cursor += 1;
|
|
256
|
+
continue;
|
|
257
|
+
}
|
|
258
|
+
if (looksLikePayoffSentence(sentence)) {
|
|
259
|
+
runSentences.push(sentence);
|
|
260
|
+
cursor += 1;
|
|
261
|
+
}
|
|
262
|
+
break;
|
|
263
|
+
}
|
|
264
|
+
if (fragmentTypes.length >= 2 && runSentences.length >= 3) {
|
|
265
|
+
const last = runSentences[runSentences.length - 1];
|
|
266
|
+
if (last !== undefined) {
|
|
267
|
+
matches.push({
|
|
268
|
+
end: last.end,
|
|
269
|
+
fragmentTypes,
|
|
270
|
+
sentences: runSentences.map((sentence) => sentence.text),
|
|
271
|
+
start: firstSentence.start
|
|
272
|
+
});
|
|
273
|
+
}
|
|
274
|
+
}
|
|
275
|
+
sentenceIndex = Math.max(cursor, sentenceIndex + 1);
|
|
276
|
+
}
|
|
277
|
+
return matches;
|
|
278
|
+
}
|
|
279
|
+
const rule = (context) => {
|
|
280
|
+
const { Syntax, RuleError, locator, report } = context;
|
|
281
|
+
const helper = new RuleHelper(context);
|
|
282
|
+
const ignoredParents = [
|
|
283
|
+
Syntax.List,
|
|
284
|
+
Syntax.ListItem,
|
|
285
|
+
Syntax.Table,
|
|
286
|
+
Syntax.TableCell
|
|
287
|
+
];
|
|
288
|
+
return {
|
|
289
|
+
[Syntax.Paragraph](node) {
|
|
290
|
+
if (helper.isChildNode(node, ignoredParents)) {
|
|
291
|
+
return;
|
|
292
|
+
}
|
|
293
|
+
const source = sourceText(node);
|
|
294
|
+
for (const match of findFragmentStacks(source.text)) {
|
|
295
|
+
report(node, new RuleError(`Fragment stack found: ${match.sentences.join(" ")} Rewrite the clipped cadence as normal prose.`, {
|
|
296
|
+
padding: locator.range([
|
|
297
|
+
source.originalStartFor(match.start),
|
|
298
|
+
source.originalEndFor(match.end)
|
|
299
|
+
])
|
|
300
|
+
}));
|
|
301
|
+
}
|
|
302
|
+
}
|
|
303
|
+
};
|
|
304
|
+
};
|
|
305
|
+
export default rule;
|
|
306
|
+
//# sourceMappingURL=fragment-stacking.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"fragment-stacking.js","sourceRoot":"","sources":["../../../../src/families/syntactic-patterns/repetition/fragment-stacking.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,UAAU,EAAE,MAAM,sBAAsB,CAAC;AAClD,OAAO,EAEL,cAAc,EACf,MAAM,mCAAmC,CAAC;AAC3C,OAAO,EAAE,UAAU,EAAE,MAAM,kCAAkC,CAAC;AAC9D,OAAO,EAAE,eAAe,EAAE,MAAM,oCAAoC,CAAC;AAErE,MAAM,kBAAkB,GAAG,CAAC,CAAC;AAC7B,MAAM,gBAAgB,GAAG,EAAE,CAAC;AAE5B,MAAM,aAAa,GAAG,IAAI,GAAG,CAAC;IAC5B,GAAG;IACH,KAAK;IACL,IAAI;IACJ,MAAM;IACN,IAAI;IACJ,KAAK;IACL,IAAI;IACJ,MAAM;IACN,MAAM;IACN,OAAO;IACP,OAAO;CACR,CAAC,CAAC;AAEH,MAAM,YAAY,GAAG,IAAI,GAAG,CAAC;IAC3B,IAAI;IACJ,KAAK;IACL,KAAK;IACL,IAAI;IACJ,MAAM;IACN,IAAI;IACJ,KAAK;IACL,GAAG;IACH,IAAI;IACJ,IAAI;IACJ,MAAM;IACN,KAAK;IACL,OAAO;CACR,CAAC,CAAC;AAEH,MAAM,YAAY,GAAG,IAAI,GAAG,CAAC;IAC3B,IAAI;IACJ,KAAK;IACL,KAAK;IACL,MAAM;IACN,IAAI;IACJ,IAAI;IACJ,MAAM;IACN,OAAO;IACP,MAAM;IACN,KAAK;IACL,KAAK;IACL,IAAI;IACJ,MAAM;IACN,KAAK;IACL,KAAK;IACL,OAAO;IACP,MAAM;IACN,OAAO;IACP,QAAQ;IACR,KAAK;IACL,OAAO;IACP,MAAM;IACN,OAAO;CACR,CAAC,CAAC;AAEH,MAAM,cAAc,GAAG,IAAI,GAAG,CAAC;IAC7B,KAAK;IACL,MAAM;IACN,MAAM;IACN,MAAM;IACN,QAAQ;IACR,YAAY;IACZ,UAAU;IACV,UAAU;IACV,OAAO;IACP,OAAO;IACP,SAAS;IACT,KAAK;IACL,MAAM;IACN,OAAO;IACP,YAAY;CACb,CAAC,CAAC;AAEH,MAAM,aAAa,GAAG,CAAC,YAAY,EAAE,OAAO,EAAE,OAAO,EAAE,UAAU,CAAC,CAAC;AACnE,MAAM,iBAAiB,GAAG,IAAI,GAAG,CAAC;IAChC,MAAM;IACN,OAAO;IACP,QAAQ;IACR,MAAM;IACN,OAAO;IACP,MAAM;IACN,MAAM;IACN,KAAK;IACL,MAAM;IACN,OAAO;IACP,KAAK;IACL,MAAM;IACN,KAAK;IACL,MAAM;IACN,QAAQ;IACR,MAAM;IACN,QAAQ;CACT,CAAC,CAAC;AAEH,MAAM,iBAAiB,GAAG,IAAI,GAAG,CAAC;IAChC,KAAK;IACL,MAAM;IACN,MAAM;IACN,MAAM;IACN,OAAO;IACP,OAAO;IACP,MAAM;IACN,MAAM;IACN,MAAM;IACN,MAAM;IACN,SAAS;IACT,MAAM;IACN,KAAK;IACL,KAAK;IACL,MAAM;IACN,MAAM;IACN,MAAM;IACN,OAAO;IACP,KAAK;IACL,QAAQ;IACR,OAAO;IACP,OAAO;IACP,KAAK;IACL,MAAM;IACN,MAAM;IACN,KAAK;IACL,MAAM;IACN,KAAK;IACL,MAAM;IACN,MAAM;IACN,OAAO;CACR,CAAC,CAAC;AASH,SAAS,cAAc,CAAC,SAAiB;IACvC,MAAM,KAAK,GAAG,SAAS,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC;IAChD,MAAM,KAAK,GAAG,SAAS,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC;IAEhD,OAAO,CAAC,SAAS,IAAI,GAAG,IAAI,SAAS,IAAI,GAAG,CAAC,IAAI,KAAK,KAAK,KAAK,CAAC;AACnE,CAAC;AAED,SAAS,SAAS,CAAC,IAAY;IAC7B,MAAM,UAAU,GAAG,IAAI,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,UAAU,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC;IAC1E,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,IAAI,GAAG,GAAG,UAAU,CAAC,MAAM,CAAC;IAE5B,OACE,KAAK,GAAG,GAAG;QACX,UAAU,CAAC,KAAK,CAAC,KAAK,GAAG;QACzB,CAAC,cAAc,CAAC,UAAU,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,EACxC,CAAC;QACD,KAAK,IAAI,CAAC,CAAC;IACb,CAAC;IAED,OACE,GAAG,GAAG,KAAK;QACX,UAAU,CAAC,GAAG,GAAG,CAAC,CAAC,KAAK,GAAG;QAC3B,CAAC,cAAc,CAAC,UAAU,CAAC,GAAG,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAC1C,CAAC;QACD,GAAG,IAAI,CAAC,CAAC;IACX,CAAC;IAED,OAAO,UAAU,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC;AACtC,CAAC;AAED,SAAS,aAAa,CAAC,QAAgB;IACrC,OAAO,eAAe,CAAC,QAAQ,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;AAClD,CAAC;AAED,SAAS,cAAc,CAAC,IAAY;IAClC,OAAO,CACL,IAAI,KAAK,KAAK;QACd,IAAI,KAAK,GAAG;QACZ,IAAI,KAAK,IAAI;QACb,IAAI,KAAK,KAAK;QACd,IAAI,KAAK,IAAI;QACb,IAAI,KAAK,KAAK;QACd,IAAI,KAAK,IAAI;QACb,IAAI,KAAK,IAAI;QACb,IAAI,KAAK,IAAI;QACb,IAAI,KAAK,IAAI;QACb,IAAI,KAAK,IAAI;QACb,IAAI,KAAK,KAAK,CACf,CAAC;AACJ,CAAC;AAED,SAAS,uBAAuB,CAAC,KAAa,EAAE,MAAc;IAC5D,OAAO,CACL,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC;QACpB,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC;QACtB,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC;QACtB,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC;QACrB,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC;QACtB,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,CACxB,CAAC;AACJ,CAAC;AAED,SAAS,oBAAoB,CAAC,KAAa,EAAE,MAAc;IACzD,OAAO,CACL,CAAC,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;QAC/C,KAAK,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;QACxB,CAAC,cAAc,CAAC,MAAM,CAAC,CACxB,CAAC;AACJ,CAAC;AAED,SAAS,qBAAqB,CAAC,KAAa,EAAE,MAAc;IAC1D,OAAO,CACL,CAAC,cAAc,CAAC,KAAK,CAAC;QACtB,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,iBAAiB,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CACzD,CAAC;AACJ,CAAC;AAED,SAAS,wBAAwB,CAAC,KAAa,EAAE,MAAc;IAC7D,OAAO,CACL,iBAAiB,CAAC,GAAG,CAAC,KAAK,CAAC;QAC5B,CAAC,YAAY,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,CAC7E,CAAC;AACJ,CAAC;AAED,SAAS,gBAAgB,CAAC,QAAuB;IAC/C,MAAM,KAAK,GAAG,aAAa,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;IAC3C,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,MAAM,GAAG,kBAAkB,EAAE,CAAC;QAC1D,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,EAAE,CAAC,EAAE,CAAC;QACtC,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,aAAa,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;QAClD,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;QACjD,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;IACvB,MAAM,MAAM,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;IACxB,IAAI,KAAK,KAAK,SAAS,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;QAChD,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,IAAI,cAAc,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,uBAAuB,CAAC,KAAK,EAAE,MAAM,CAAC,EAAE,CAAC;QACxE,OAAO,mBAAmB,CAAC;IAC7B,CAAC;IAED,IACE,qBAAqB,CAAC,KAAK,EAAE,MAAM,CAAC;QACpC,wBAAwB,CAAC,KAAK,EAAE,MAAM,CAAC,EACvC,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,IAAI,oBAAoB,CAAC,KAAK,EAAE,MAAM,CAAC,EAAE,CAAC;QACxC,OAAO,cAAc,CAAC;IACxB,CAAC;IAED,OAAO,eAAe,CAAC;AACzB,CAAC;AAED,SAAS,uBAAuB,CAAC,QAAuB;IACtD,MAAM,KAAK,GAAG,aAAa,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;IAC3C,IAAI,KAAK,CAAC,MAAM,IAAI,CAAC,IAAI,KAAK,CAAC,MAAM,GAAG,gBAAgB,EAAE,CAAC;QACzD,OAAO,KAAK,CAAC;IACf,CAAC;IAED,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC;IACtD,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,OAAO,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC;QAC9D,OAAO,KAAK,CAAC;IACf,CAAC;IAED,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,aAAa,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC;AACxD,CAAC;AAED,SAAS,kBAAkB,CAAC,IAAY;IACtC,MAAM,SAAS,GAAG,cAAc,CAAC,IAAI,CAAC,CAAC;IACvC,MAAM,eAAe,GAAG,SAAS,CAAC,GAAG,CAAC,gBAAgB,CAAC,CAAC;IACxD,MAAM,OAAO,GAAoB,EAAE,CAAC;IACpC,IAAI,aAAa,GAAG,CAAC,CAAC;IAEtB,OAAO,aAAa,GAAG,SAAS,CAAC,MAAM,EAAE,CAAC;QACxC,MAAM,SAAS,GAAG,eAAe,CAAC,aAAa,CAAC,CAAC;QACjD,MAAM,aAAa,GAAG,SAAS,CAAC,aAAa,CAAC,CAAC;QAC/C,IAAI,SAAS,KAAK,SAAS,IAAI,aAAa,KAAK,SAAS,EAAE,CAAC;YAC3D,aAAa,IAAI,CAAC,CAAC;YACnB,SAAS;QACX,CAAC;QAED,MAAM,YAAY,GAAG,CAAC,aAAa,CAAC,CAAC;QACrC,MAAM,aAAa,GAAG,CAAC,SAAS,CAAC,CAAC;QAClC,IAAI,MAAM,GAAG,aAAa,GAAG,CAAC,CAAC;QAE/B,OAAO,MAAM,GAAG,SAAS,CAAC,MAAM,EAAE,CAAC;YACjC,MAAM,QAAQ,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC;YACnC,MAAM,QAAQ,GAAG,eAAe,CAAC,MAAM,CAAC,CAAC;YACzC,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;gBAC3B,MAAM;YACR,CAAC;YAED,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;gBAC3B,YAAY,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBAC5B,aAAa,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBAC7B,MAAM,IAAI,CAAC,CAAC;gBACZ,SAAS;YACX,CAAC;YAED,IAAI,uBAAuB,CAAC,QAAQ,CAAC,EAAE,CAAC;gBACtC,YAAY,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBAC5B,MAAM,IAAI,CAAC,CAAC;YACd,CAAC;YACD,MAAM;QACR,CAAC;QAED,IAAI,aAAa,CAAC,MAAM,IAAI,CAAC,IAAI,YAAY,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;YAC1D,MAAM,IAAI,GAAG,YAAY,CAAC,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;YACnD,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;gBACvB,OAAO,CAAC,IAAI,CAAC;oBACX,GAAG,EAAE,IAAI,CAAC,GAAG;oBACb,aAAa;oBACb,SAAS,EAAE,YAAY,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,QAAQ,CAAC,IAAI,CAAC;oBACxD,KAAK,EAAE,aAAa,CAAC,KAAK;iBAC3B,CAAC,CAAC;YACL,CAAC;QACH,CAAC;QAED,aAAa,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,EAAE,aAAa,GAAG,CAAC,CAAC,CAAC;IACtD,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,MAAM,IAAI,GAAuB,CAAC,OAAO,EAAE,EAAE;IAC3C,MAAM,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC;IACvD,MAAM,MAAM,GAAG,IAAI,UAAU,CAAC,OAAO,CAAC,CAAC;IACvC,MAAM,cAAc,GAAG;QACrB,MAAM,CAAC,IAAI;QACX,MAAM,CAAC,QAAQ;QACf,MAAM,CAAC,KAAK;QACZ,MAAM,CAAC,SAAS;KACjB,CAAC;IAEF,OAAO;QACL,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAmB;YACpC,IAAI,MAAM,CAAC,WAAW,CAAC,IAAI,EAAE,cAAc,CAAC,EAAE,CAAC;gBAC7C,OAAO;YACT,CAAC;YAED,MAAM,MAAM,GAAG,UAAU,CAAC,IAAI,CAAC,CAAC;YAChC,KAAK,MAAM,KAAK,IAAI,kBAAkB,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC;gBACpD,MAAM,CACJ,IAAI,EACJ,IAAI,SAAS,CACX,yBAAyB,KAAK,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,+CAA+C,EACjG;oBACE,OAAO,EAAE,OAAO,CAAC,KAAK,CAAC;wBACrB,MAAM,CAAC,gBAAgB,CAAC,KAAK,CAAC,KAAK,CAAC;wBACpC,MAAM,CAAC,cAAc,CAAC,KAAK,CAAC,GAAG,CAAC;qBACjC,CAAC;iBACH,CACF,CACF,CAAC;YACJ,CAAC;QACH,CAAC;KACF,CAAC;AACJ,CAAC,CAAC;AAEF,eAAe,IAAI,CAAC"}
|
|
@@ -0,0 +1,149 @@
|
|
|
1
|
+
import { RuleHelper } from "textlint-rule-helper";
|
|
2
|
+
import { splitSentences } from "../../../shared/text/sentences.js";
|
|
3
|
+
import { sourceText } from "../../../shared/text/traverse.js";
|
|
4
|
+
import { splitWhitespace } from "../../../shared/text/whitespace.js";
|
|
5
|
+
const REPEATED_FRAME_PREFIXES = [
|
|
6
|
+
["the", "answer", "is"],
|
|
7
|
+
["the", "fix", "is"],
|
|
8
|
+
["the", "goal", "is"],
|
|
9
|
+
["the", "lesson", "is"],
|
|
10
|
+
["the", "point", "is"],
|
|
11
|
+
["the", "result", "is"],
|
|
12
|
+
["the", "trick", "is"],
|
|
13
|
+
["the", "useful", "move"],
|
|
14
|
+
["what", "helps", "is"],
|
|
15
|
+
["what", "matters", "is"]
|
|
16
|
+
];
|
|
17
|
+
function isWordCharacter(character) {
|
|
18
|
+
return ((character >= "a" && character <= "z") ||
|
|
19
|
+
(character >= "0" && character <= "9") ||
|
|
20
|
+
character === "'");
|
|
21
|
+
}
|
|
22
|
+
function cleanWord(word) {
|
|
23
|
+
return word
|
|
24
|
+
.toLocaleLowerCase("en")
|
|
25
|
+
.replaceAll("\u2019", "'")
|
|
26
|
+
.replaceAll("\u2018", "'");
|
|
27
|
+
}
|
|
28
|
+
function trimWordEdges(word) {
|
|
29
|
+
const cleaned = cleanWord(word);
|
|
30
|
+
let start = 0;
|
|
31
|
+
let end = cleaned.length;
|
|
32
|
+
while (start < end && !isWordCharacter(cleaned[start] ?? "")) {
|
|
33
|
+
start += 1;
|
|
34
|
+
}
|
|
35
|
+
while (end > start && !isWordCharacter(cleaned[end - 1] ?? "")) {
|
|
36
|
+
end -= 1;
|
|
37
|
+
}
|
|
38
|
+
return cleaned.slice(start, end);
|
|
39
|
+
}
|
|
40
|
+
function normalizedWords(sentence) {
|
|
41
|
+
return splitWhitespace(sentence).map(trimWordEdges);
|
|
42
|
+
}
|
|
43
|
+
function firstWord(sentence) {
|
|
44
|
+
const word = normalizedWords(sentence)[0];
|
|
45
|
+
if (word === undefined) {
|
|
46
|
+
return undefined;
|
|
47
|
+
}
|
|
48
|
+
return word;
|
|
49
|
+
}
|
|
50
|
+
function framePrefix(sentence) {
|
|
51
|
+
const words = normalizedWords(sentence);
|
|
52
|
+
const matched = REPEATED_FRAME_PREFIXES.find((prefix) => prefix.every((word, index) => words[index] === word));
|
|
53
|
+
return matched?.join(" ");
|
|
54
|
+
}
|
|
55
|
+
function findTripleRepeats(text) {
|
|
56
|
+
const sentences = splitSentences(text);
|
|
57
|
+
const matches = [];
|
|
58
|
+
if (sentences.length < 3) {
|
|
59
|
+
return matches;
|
|
60
|
+
}
|
|
61
|
+
for (let index = 0; index <= sentences.length - 3; index += 1) {
|
|
62
|
+
const first = sentences[index];
|
|
63
|
+
const second = sentences[index + 1];
|
|
64
|
+
const third = sentences[index + 2];
|
|
65
|
+
if (first === undefined || second === undefined || third === undefined) {
|
|
66
|
+
continue;
|
|
67
|
+
}
|
|
68
|
+
const firstOpener = firstWord(first.text);
|
|
69
|
+
const secondOpener = firstWord(second.text);
|
|
70
|
+
const thirdOpener = firstWord(third.text);
|
|
71
|
+
if (firstOpener === undefined ||
|
|
72
|
+
firstOpener !== secondOpener ||
|
|
73
|
+
secondOpener !== thirdOpener) {
|
|
74
|
+
continue;
|
|
75
|
+
}
|
|
76
|
+
matches.push({
|
|
77
|
+
end: third.end,
|
|
78
|
+
kind: "triple",
|
|
79
|
+
opener: firstOpener,
|
|
80
|
+
sentences: [first.text, second.text, third.text],
|
|
81
|
+
start: first.start
|
|
82
|
+
});
|
|
83
|
+
}
|
|
84
|
+
return matches;
|
|
85
|
+
}
|
|
86
|
+
function findRepeatedFrames(text) {
|
|
87
|
+
const sentences = splitSentences(text);
|
|
88
|
+
const matches = [];
|
|
89
|
+
if (sentences.length < 2) {
|
|
90
|
+
return matches;
|
|
91
|
+
}
|
|
92
|
+
for (let index = 0; index <= sentences.length - 2; index += 1) {
|
|
93
|
+
const first = sentences[index];
|
|
94
|
+
const second = sentences[index + 1];
|
|
95
|
+
if (first === undefined || second === undefined) {
|
|
96
|
+
continue;
|
|
97
|
+
}
|
|
98
|
+
const firstFrame = framePrefix(first.text);
|
|
99
|
+
const secondFrame = framePrefix(second.text);
|
|
100
|
+
if (firstFrame === undefined ||
|
|
101
|
+
firstFrame !== secondFrame ||
|
|
102
|
+
first.text.length > 120 ||
|
|
103
|
+
second.text.length > 120) {
|
|
104
|
+
continue;
|
|
105
|
+
}
|
|
106
|
+
matches.push({
|
|
107
|
+
end: second.end,
|
|
108
|
+
kind: "frame",
|
|
109
|
+
opener: firstFrame,
|
|
110
|
+
sentences: [first.text, second.text],
|
|
111
|
+
start: first.start
|
|
112
|
+
});
|
|
113
|
+
}
|
|
114
|
+
return matches;
|
|
115
|
+
}
|
|
116
|
+
const rule = (context) => {
|
|
117
|
+
const { Syntax, RuleError, locator, report } = context;
|
|
118
|
+
const helper = new RuleHelper(context);
|
|
119
|
+
const ignoredParents = [
|
|
120
|
+
Syntax.List,
|
|
121
|
+
Syntax.ListItem,
|
|
122
|
+
Syntax.Table,
|
|
123
|
+
Syntax.TableCell
|
|
124
|
+
];
|
|
125
|
+
return {
|
|
126
|
+
[Syntax.Paragraph](node) {
|
|
127
|
+
if (helper.isChildNode(node, ignoredParents)) {
|
|
128
|
+
return;
|
|
129
|
+
}
|
|
130
|
+
const source = sourceText(node);
|
|
131
|
+
const matches = [
|
|
132
|
+
...findTripleRepeats(source.text),
|
|
133
|
+
...findRepeatedFrames(source.text)
|
|
134
|
+
];
|
|
135
|
+
for (const match of matches) {
|
|
136
|
+
report(node, new RuleError(match.kind === "triple"
|
|
137
|
+
? `Triple repeat opener found: "${match.opener}". Vary the sentence openers.`
|
|
138
|
+
: `Repeated sentence frame found: "${match.opener}". Vary the sentence frame.`, {
|
|
139
|
+
padding: locator.range([
|
|
140
|
+
source.originalStartFor(match.start),
|
|
141
|
+
source.originalEndFor(match.end)
|
|
142
|
+
])
|
|
143
|
+
}));
|
|
144
|
+
}
|
|
145
|
+
}
|
|
146
|
+
};
|
|
147
|
+
};
|
|
148
|
+
export default rule;
|
|
149
|
+
//# sourceMappingURL=triple-repeat.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"triple-repeat.js","sourceRoot":"","sources":["../../../../src/families/syntactic-patterns/repetition/triple-repeat.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,UAAU,EAAE,MAAM,sBAAsB,CAAC;AAClD,OAAO,EAAE,cAAc,EAAE,MAAM,mCAAmC,CAAC;AACnE,OAAO,EAAE,UAAU,EAAE,MAAM,kCAAkC,CAAC;AAC9D,OAAO,EAAE,eAAe,EAAE,MAAM,oCAAoC,CAAC;AAUrE,MAAM,uBAAuB,GAAG;IAC9B,CAAC,KAAK,EAAE,QAAQ,EAAE,IAAI,CAAC;IACvB,CAAC,KAAK,EAAE,KAAK,EAAE,IAAI,CAAC;IACpB,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,CAAC;IACrB,CAAC,KAAK,EAAE,QAAQ,EAAE,IAAI,CAAC;IACvB,CAAC,KAAK,EAAE,OAAO,EAAE,IAAI,CAAC;IACtB,CAAC,KAAK,EAAE,QAAQ,EAAE,IAAI,CAAC;IACvB,CAAC,KAAK,EAAE,OAAO,EAAE,IAAI,CAAC;IACtB,CAAC,KAAK,EAAE,QAAQ,EAAE,MAAM,CAAC;IACzB,CAAC,MAAM,EAAE,OAAO,EAAE,IAAI,CAAC;IACvB,CAAC,MAAM,EAAE,SAAS,EAAE,IAAI,CAAC;CACjB,CAAC;AAEX,SAAS,eAAe,CAAC,SAAiB;IACxC,OAAO,CACL,CAAC,SAAS,IAAI,GAAG,IAAI,SAAS,IAAI,GAAG,CAAC;QACtC,CAAC,SAAS,IAAI,GAAG,IAAI,SAAS,IAAI,GAAG,CAAC;QACtC,SAAS,KAAK,GAAG,CAClB,CAAC;AACJ,CAAC;AAED,SAAS,SAAS,CAAC,IAAY;IAC7B,OAAO,IAAI;SACR,iBAAiB,CAAC,IAAI,CAAC;SACvB,UAAU,CAAC,QAAQ,EAAE,GAAG,CAAC;SACzB,UAAU,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC;AAC/B,CAAC;AAED,SAAS,aAAa,CAAC,IAAY;IACjC,MAAM,OAAO,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC;IAChC,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,IAAI,GAAG,GAAG,OAAO,CAAC,MAAM,CAAC;IAEzB,OAAO,KAAK,GAAG,GAAG,IAAI,CAAC,eAAe,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;QAC7D,KAAK,IAAI,CAAC,CAAC;IACb,CAAC;IAED,OAAO,GAAG,GAAG,KAAK,IAAI,CAAC,eAAe,CAAC,OAAO,CAAC,GAAG,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;QAC/D,GAAG,IAAI,CAAC,CAAC;IACX,CAAC;IAED,OAAO,OAAO,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC;AACnC,CAAC;AAED,SAAS,eAAe,CAAC,QAAgB;IACvC,OAAO,eAAe,CAAC,QAAQ,CAAC,CAAC,GAAG,CAAC,aAAa,CAAC,CAAC;AACtD,CAAC;AAED,SAAS,SAAS,CAAC,QAAgB;IACjC,MAAM,IAAI,GAAG,eAAe,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;IAC1C,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;QACvB,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,OAAO,IAAI,CAAC;AACd,CAAC;AAED,SAAS,WAAW,CAAC,QAAgB;IACnC,MAAM,KAAK,GAAG,eAAe,CAAC,QAAQ,CAAC,CAAC;IACxC,MAAM,OAAO,GAAG,uBAAuB,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,EAAE,CACtD,MAAM,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,IAAI,CAAC,CACrD,CAAC;IAEF,OAAO,OAAO,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC;AAC5B,CAAC;AAED,SAAS,iBAAiB,CAAC,IAAY;IACrC,MAAM,SAAS,GAAG,cAAc,CAAC,IAAI,CAAC,CAAC;IACvC,MAAM,OAAO,GAAkB,EAAE,CAAC;IAElC,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACzB,OAAO,OAAO,CAAC;IACjB,CAAC;IAED,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;QAC9D,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,CAAC,CAAC;QAC/B,MAAM,MAAM,GAAG,SAAS,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;QACpC,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;QACnC,IAAI,KAAK,KAAK,SAAS,IAAI,MAAM,KAAK,SAAS,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;YACvE,SAAS;QACX,CAAC;QAED,MAAM,WAAW,GAAG,SAAS,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QAC1C,MAAM,YAAY,GAAG,SAAS,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;QAC5C,MAAM,WAAW,GAAG,SAAS,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QAC1C,IACE,WAAW,KAAK,SAAS;YACzB,WAAW,KAAK,YAAY;YAC5B,YAAY,KAAK,WAAW,EAC5B,CAAC;YACD,SAAS;QACX,CAAC;QAED,OAAO,CAAC,IAAI,CAAC;YACX,GAAG,EAAE,KAAK,CAAC,GAAG;YACd,IAAI,EAAE,QAAQ;YACd,MAAM,EAAE,WAAW;YACnB,SAAS,EAAE,CAAC,KAAK,CAAC,IAAI,EAAE,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC;YAChD,KAAK,EAAE,KAAK,CAAC,KAAK;SACnB,CAAC,CAAC;IACL,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,SAAS,kBAAkB,CAAC,IAAY;IACtC,MAAM,SAAS,GAAG,cAAc,CAAC,IAAI,CAAC,CAAC;IACvC,MAAM,OAAO,GAAkB,EAAE,CAAC;IAElC,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACzB,OAAO,OAAO,CAAC;IACjB,CAAC;IAED,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;QAC9D,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,CAAC,CAAC;QAC/B,MAAM,MAAM,GAAG,SAAS,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;QACpC,IAAI,KAAK,KAAK,SAAS,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;YAChD,SAAS;QACX,CAAC;QAED,MAAM,UAAU,GAAG,WAAW,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QAC3C,MAAM,WAAW,GAAG,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;QAC7C,IACE,UAAU,KAAK,SAAS;YACxB,UAAU,KAAK,WAAW;YAC1B,KAAK,CAAC,IAAI,CAAC,MAAM,GAAG,GAAG;YACvB,MAAM,CAAC,IAAI,CAAC,MAAM,GAAG,GAAG,EACxB,CAAC;YACD,SAAS;QACX,CAAC;QAED,OAAO,CAAC,IAAI,CAAC;YACX,GAAG,EAAE,MAAM,CAAC,GAAG;YACf,IAAI,EAAE,OAAO;YACb,MAAM,EAAE,UAAU;YAClB,SAAS,EAAE,CAAC,KAAK,CAAC,IAAI,EAAE,MAAM,CAAC,IAAI,CAAC;YACpC,KAAK,EAAE,KAAK,CAAC,KAAK;SACnB,CAAC,CAAC;IACL,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,MAAM,IAAI,GAAuB,CAAC,OAAO,EAAE,EAAE;IAC3C,MAAM,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC;IACvD,MAAM,MAAM,GAAG,IAAI,UAAU,CAAC,OAAO,CAAC,CAAC;IACvC,MAAM,cAAc,GAAG;QACrB,MAAM,CAAC,IAAI;QACX,MAAM,CAAC,QAAQ;QACf,MAAM,CAAC,KAAK;QACZ,MAAM,CAAC,SAAS;KACjB,CAAC;IAEF,OAAO;QACL,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAmB;YACpC,IAAI,MAAM,CAAC,WAAW,CAAC,IAAI,EAAE,cAAc,CAAC,EAAE,CAAC;gBAC7C,OAAO;YACT,CAAC;YAED,MAAM,MAAM,GAAG,UAAU,CAAC,IAAI,CAAC,CAAC;YAChC,MAAM,OAAO,GAAG;gBACd,GAAG,iBAAiB,CAAC,MAAM,CAAC,IAAI,CAAC;gBACjC,GAAG,kBAAkB,CAAC,MAAM,CAAC,IAAI,CAAC;aACnC,CAAC;YAEF,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;gBAC5B,MAAM,CACJ,IAAI,EACJ,IAAI,SAAS,CACX,KAAK,CAAC,IAAI,KAAK,QAAQ;oBACrB,CAAC,CAAC,gCAAgC,KAAK,CAAC,MAAM,+BAA+B;oBAC7E,CAAC,CAAC,mCAAmC,KAAK,CAAC,MAAM,6BAA6B,EAChF;oBACE,OAAO,EAAE,OAAO,CAAC,KAAK,CAAC;wBACrB,MAAM,CAAC,gBAAgB,CAAC,KAAK,CAAC,KAAK,CAAC;wBACpC,MAAM,CAAC,cAAc,CAAC,KAAK,CAAC,GAAG,CAAC;qBACjC,CAAC;iBACH,CACF,CACF,CAAC;YACJ,CAAC;QACH,CAAC;KACF,CAAC;AACJ,CAAC,CAAC;AAEF,eAAe,IAAI,CAAC"}
|
|
@@ -0,0 +1,87 @@
|
|
|
1
|
+
import { documentText } from "../../shared/text/document.js";
|
|
2
|
+
import { normalizeForMatch } from "../../shared/text/normalize.js";
|
|
3
|
+
import { wordTokens } from "../../shared/text/tokens.js";
|
|
4
|
+
const SUFFIXES = [
|
|
5
|
+
"ation",
|
|
6
|
+
"tion",
|
|
7
|
+
"ment",
|
|
8
|
+
"ness",
|
|
9
|
+
"ing",
|
|
10
|
+
"ies",
|
|
11
|
+
"ied",
|
|
12
|
+
"es",
|
|
13
|
+
"ed",
|
|
14
|
+
"er",
|
|
15
|
+
"ly",
|
|
16
|
+
"s"
|
|
17
|
+
];
|
|
18
|
+
function roughStem(word) {
|
|
19
|
+
for (const suffix of SUFFIXES) {
|
|
20
|
+
if (!word.endsWith(suffix)) {
|
|
21
|
+
continue;
|
|
22
|
+
}
|
|
23
|
+
const stripped = word.slice(0, -suffix.length);
|
|
24
|
+
if (stripped.length >= 3) {
|
|
25
|
+
return stripped;
|
|
26
|
+
}
|
|
27
|
+
}
|
|
28
|
+
return word;
|
|
29
|
+
}
|
|
30
|
+
function tokenSet(text) {
|
|
31
|
+
return new Set(wordTokens(text).map((token) => token.normalized));
|
|
32
|
+
}
|
|
33
|
+
function configuredPolicy(options) {
|
|
34
|
+
const terms = options.terms?.map(normalizeForMatch).filter((term) => term.length > 0) ??
|
|
35
|
+
[];
|
|
36
|
+
const minCount = options.minCount ?? 0;
|
|
37
|
+
if (terms.length === 0 || minCount <= 0) {
|
|
38
|
+
return undefined;
|
|
39
|
+
}
|
|
40
|
+
return {
|
|
41
|
+
allowInflections: options.allowInflections === true,
|
|
42
|
+
minCount,
|
|
43
|
+
terms
|
|
44
|
+
};
|
|
45
|
+
}
|
|
46
|
+
function hasTerm(tokens, term, allowInflections) {
|
|
47
|
+
if (!allowInflections) {
|
|
48
|
+
return tokens.has(term);
|
|
49
|
+
}
|
|
50
|
+
const stem = roughStem(term);
|
|
51
|
+
for (const token of tokens) {
|
|
52
|
+
if (token.startsWith(stem)) {
|
|
53
|
+
return true;
|
|
54
|
+
}
|
|
55
|
+
}
|
|
56
|
+
return false;
|
|
57
|
+
}
|
|
58
|
+
function countPresentTerms(text, policy) {
|
|
59
|
+
const tokens = tokenSet(text);
|
|
60
|
+
let count = 0;
|
|
61
|
+
for (const term of policy.terms) {
|
|
62
|
+
if (hasTerm(tokens, term, policy.allowInflections)) {
|
|
63
|
+
count += 1;
|
|
64
|
+
}
|
|
65
|
+
}
|
|
66
|
+
return count;
|
|
67
|
+
}
|
|
68
|
+
const rule = (context, options = {}) => {
|
|
69
|
+
const { Syntax, RuleError, locator, report } = context;
|
|
70
|
+
return {
|
|
71
|
+
[Syntax.Document](node) {
|
|
72
|
+
const policy = configuredPolicy(options);
|
|
73
|
+
if (policy === undefined) {
|
|
74
|
+
return;
|
|
75
|
+
}
|
|
76
|
+
const count = countPresentTerms(documentText(node), policy);
|
|
77
|
+
if (count >= policy.minCount) {
|
|
78
|
+
return;
|
|
79
|
+
}
|
|
80
|
+
report(node, new RuleError(`Recommended terms present: ${count}. Include at least ${policy.minCount} terms from the policy pool.`, {
|
|
81
|
+
padding: locator.at(0)
|
|
82
|
+
}));
|
|
83
|
+
}
|
|
84
|
+
};
|
|
85
|
+
};
|
|
86
|
+
export default rule;
|
|
87
|
+
//# sourceMappingURL=recommended-terms.js.map
|