chaffjs 0.35.0 → 0.36.0
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.ja.md +12 -0
- package/README.md +15 -0
- package/dist/announced-count.d.ts +2 -0
- package/dist/announced-count.d.ts.map +1 -1
- package/dist/announced-count.js +29 -7
- package/dist/announced-count.js.map +1 -1
- package/dist/chronology-rows.d.ts +9 -0
- package/dist/chronology-rows.d.ts.map +1 -0
- package/dist/chronology-rows.js +79 -0
- package/dist/chronology-rows.js.map +1 -0
- package/dist/company-name-run.d.ts +8 -0
- package/dist/company-name-run.d.ts.map +1 -0
- package/dist/company-name-run.js +30 -0
- package/dist/company-name-run.js.map +1 -0
- package/dist/compound-head.d.ts +7 -0
- package/dist/compound-head.d.ts.map +1 -0
- package/dist/compound-head.js +19 -0
- package/dist/compound-head.js.map +1 -0
- package/dist/date-format.d.ts +8 -3
- package/dist/date-format.d.ts.map +1 -1
- package/dist/date-format.js +36 -3
- package/dist/date-format.js.map +1 -1
- package/dist/derived/count-per-period.d.ts +23 -0
- package/dist/derived/count-per-period.d.ts.map +1 -0
- package/dist/derived/count-per-period.js +14 -0
- package/dist/derived/count-per-period.js.map +1 -0
- package/dist/detectors/acronym-context.d.ts +4 -0
- package/dist/detectors/acronym-context.d.ts.map +1 -1
- package/dist/detectors/acronym-context.js +29 -1
- package/dist/detectors/acronym-context.js.map +1 -1
- package/dist/detectors/acronym-expansion.d.ts +5 -1
- package/dist/detectors/acronym-expansion.d.ts.map +1 -1
- package/dist/detectors/acronym-expansion.js +35 -3
- package/dist/detectors/acronym-expansion.js.map +1 -1
- package/dist/detectors/acronym-known.d.ts +19 -0
- package/dist/detectors/acronym-known.d.ts.map +1 -0
- package/dist/detectors/acronym-known.js +32 -0
- package/dist/detectors/acronym-known.js.map +1 -0
- package/dist/detectors/acronym-link.d.ts +14 -0
- package/dist/detectors/acronym-link.d.ts.map +1 -0
- package/dist/detectors/acronym-link.js +32 -0
- package/dist/detectors/acronym-link.js.map +1 -0
- package/dist/detectors/acronym-name.d.ts +8 -1
- package/dist/detectors/acronym-name.d.ts.map +1 -1
- package/dist/detectors/acronym-name.js +23 -4
- package/dist/detectors/acronym-name.js.map +1 -1
- package/dist/detectors/addressee-edge.d.ts +13 -0
- package/dist/detectors/addressee-edge.d.ts.map +1 -0
- package/dist/detectors/addressee-edge.js +13 -0
- package/dist/detectors/addressee-edge.js.map +1 -0
- package/dist/detectors/announced-count.d.ts.map +1 -1
- package/dist/detectors/announced-count.js +1 -0
- package/dist/detectors/announced-count.js.map +1 -1
- package/dist/detectors/bare-comparative.d.ts +4 -0
- package/dist/detectors/bare-comparative.d.ts.map +1 -1
- package/dist/detectors/bare-comparative.js +28 -3
- package/dist/detectors/bare-comparative.js.map +1 -1
- package/dist/detectors/blank-structure.d.ts +7 -1
- package/dist/detectors/blank-structure.d.ts.map +1 -1
- package/dist/detectors/blank-structure.js +13 -8
- package/dist/detectors/blank-structure.js.map +1 -1
- package/dist/detectors/bracket-specified.d.ts +6 -0
- package/dist/detectors/bracket-specified.d.ts.map +1 -0
- package/dist/detectors/bracket-specified.js +23 -0
- package/dist/detectors/bracket-specified.js.map +1 -0
- package/dist/detectors/broken-link.d.ts +6 -0
- package/dist/detectors/broken-link.d.ts.map +1 -1
- package/dist/detectors/broken-link.js +23 -2
- package/dist/detectors/broken-link.js.map +1 -1
- package/dist/detectors/caps-caption.d.ts +7 -6
- package/dist/detectors/caps-caption.d.ts.map +1 -1
- package/dist/detectors/caps-caption.js +38 -0
- package/dist/detectors/caps-caption.js.map +1 -1
- package/dist/detectors/caps-phrase.d.ts +6 -0
- package/dist/detectors/caps-phrase.d.ts.map +1 -0
- package/dist/detectors/caps-phrase.js +16 -0
- package/dist/detectors/caps-phrase.js.map +1 -0
- package/dist/detectors/char-shape.d.ts +1 -0
- package/dist/detectors/char-shape.d.ts.map +1 -1
- package/dist/detectors/char-shape.js +29 -16
- package/dist/detectors/char-shape.js.map +1 -1
- package/dist/detectors/cited-title.d.ts +3 -2
- package/dist/detectors/cited-title.d.ts.map +1 -1
- package/dist/detectors/cited-title.js +14 -3
- package/dist/detectors/cited-title.js.map +1 -1
- package/dist/detectors/counter-edge.d.ts +2 -1
- package/dist/detectors/counter-edge.d.ts.map +1 -1
- package/dist/detectors/counter-edge.js +17 -2
- package/dist/detectors/counter-edge.js.map +1 -1
- package/dist/detectors/date-any-year.d.ts +7 -3
- package/dist/detectors/date-any-year.d.ts.map +1 -1
- package/dist/detectors/date-any-year.js +25 -5
- package/dist/detectors/date-any-year.js.map +1 -1
- package/dist/detectors/date-format.d.ts.map +1 -1
- package/dist/detectors/date-format.js +2 -1
- package/dist/detectors/date-format.js.map +1 -1
- package/dist/detectors/date-year.d.ts +9 -3
- package/dist/detectors/date-year.d.ts.map +1 -1
- package/dist/detectors/date-year.js +45 -9
- package/dist/detectors/date-year.js.map +1 -1
- package/dist/detectors/deadline-unit.d.ts +20 -0
- package/dist/detectors/deadline-unit.d.ts.map +1 -0
- package/dist/detectors/deadline-unit.js +19 -0
- package/dist/detectors/deadline-unit.js.map +1 -0
- package/dist/detectors/defined-term-form.d.ts.map +1 -1
- package/dist/detectors/defined-term-form.js +33 -7
- package/dist/detectors/defined-term-form.js.map +1 -1
- package/dist/detectors/definition-use.d.ts.map +1 -1
- package/dist/detectors/definition-use.js +6 -3
- package/dist/detectors/definition-use.js.map +1 -1
- package/dist/detectors/derived-numbers.d.ts.map +1 -1
- package/dist/detectors/derived-numbers.js +17 -4
- package/dist/detectors/derived-numbers.js.map +1 -1
- package/dist/detectors/doubled-across-aside.d.ts +9 -0
- package/dist/detectors/doubled-across-aside.d.ts.map +1 -0
- package/dist/detectors/doubled-across-aside.js +27 -0
- package/dist/detectors/doubled-across-aside.js.map +1 -0
- package/dist/detectors/doubled-punctuation.d.ts +3 -3
- package/dist/detectors/doubled-punctuation.d.ts.map +1 -1
- package/dist/detectors/doubled-punctuation.js +23 -4
- package/dist/detectors/doubled-punctuation.js.map +1 -1
- package/dist/detectors/doubled-word.d.ts.map +1 -1
- package/dist/detectors/doubled-word.js +2 -1
- package/dist/detectors/doubled-word.js.map +1 -1
- package/dist/detectors/ellipsis.d.ts.map +1 -1
- package/dist/detectors/ellipsis.js +13 -1
- package/dist/detectors/ellipsis.js.map +1 -1
- package/dist/detectors/email-letter.d.ts +31 -1
- package/dist/detectors/email-letter.d.ts.map +1 -1
- package/dist/detectors/email-letter.js +76 -14
- package/dist/detectors/email-letter.js.map +1 -1
- package/dist/detectors/field-label.d.ts +7 -0
- package/dist/detectors/field-label.d.ts.map +1 -0
- package/dist/detectors/field-label.js +23 -0
- package/dist/detectors/field-label.js.map +1 -0
- package/dist/detectors/footnote-mismatch.d.ts +6 -1
- package/dist/detectors/footnote-mismatch.d.ts.map +1 -1
- package/dist/detectors/footnote-mismatch.js +28 -7
- package/dist/detectors/footnote-mismatch.js.map +1 -1
- package/dist/detectors/graded-step.d.ts +20 -0
- package/dist/detectors/graded-step.d.ts.map +1 -0
- package/dist/detectors/graded-step.js +33 -0
- package/dist/detectors/graded-step.js.map +1 -0
- package/dist/detectors/gram-openers.d.ts +8 -0
- package/dist/detectors/gram-openers.d.ts.map +1 -0
- package/dist/detectors/gram-openers.js +16 -0
- package/dist/detectors/gram-openers.js.map +1 -0
- package/dist/detectors/gram-place.d.ts +3 -1
- package/dist/detectors/gram-place.d.ts.map +1 -1
- package/dist/detectors/gram-place.js +8 -0
- package/dist/detectors/gram-place.js.map +1 -1
- package/dist/detectors/gram-predicate.d.ts.map +1 -1
- package/dist/detectors/gram-predicate.js +37 -3
- package/dist/detectors/gram-predicate.js.map +1 -1
- package/dist/detectors/heading-echo.d.ts.map +1 -1
- package/dist/detectors/heading-echo.js +28 -2
- package/dist/detectors/heading-echo.js.map +1 -1
- package/dist/detectors/ime-residue.d.ts +1 -1
- package/dist/detectors/ime-residue.d.ts.map +1 -1
- package/dist/detectors/ime-residue.js +4 -3
- package/dist/detectors/ime-residue.js.map +1 -1
- package/dist/detectors/kanji-compound-slip.d.ts +3 -0
- package/dist/detectors/kanji-compound-slip.d.ts.map +1 -0
- package/dist/detectors/kanji-compound-slip.js +50 -0
- package/dist/detectors/kanji-compound-slip.js.map +1 -0
- package/dist/detectors/kutoten-consistency.d.ts +10 -2
- package/dist/detectors/kutoten-consistency.d.ts.map +1 -1
- package/dist/detectors/kutoten-consistency.js +76 -12
- package/dist/detectors/kutoten-consistency.js.map +1 -1
- package/dist/detectors/lexicon.d.ts.map +1 -1
- package/dist/detectors/lexicon.js +8 -2
- package/dist/detectors/lexicon.js.map +1 -1
- package/dist/detectors/link-shape.d.ts.map +1 -1
- package/dist/detectors/link-shape.js +9 -3
- package/dist/detectors/link-shape.js.map +1 -1
- package/dist/detectors/list-item-ending.d.ts +18 -0
- package/dist/detectors/list-item-ending.d.ts.map +1 -0
- package/dist/detectors/list-item-ending.js +35 -0
- package/dist/detectors/list-item-ending.js.map +1 -0
- package/dist/detectors/list-parallel.d.ts +18 -5
- package/dist/detectors/list-parallel.d.ts.map +1 -1
- package/dist/detectors/list-parallel.js +116 -36
- package/dist/detectors/list-parallel.js.map +1 -1
- package/dist/detectors/list-sentence.d.ts +4 -0
- package/dist/detectors/list-sentence.d.ts.map +1 -1
- package/dist/detectors/list-sentence.js +1 -1
- package/dist/detectors/list-sentence.js.map +1 -1
- package/dist/detectors/mark-shape.d.ts +11 -4
- package/dist/detectors/mark-shape.d.ts.map +1 -1
- package/dist/detectors/mark-shape.js +58 -10
- package/dist/detectors/mark-shape.js.map +1 -1
- package/dist/detectors/mentioned-phrase.d.ts +9 -0
- package/dist/detectors/mentioned-phrase.d.ts.map +1 -0
- package/dist/detectors/mentioned-phrase.js +25 -0
- package/dist/detectors/mentioned-phrase.js.map +1 -0
- package/dist/detectors/middle-dot.d.ts +2 -2
- package/dist/detectors/middle-dot.d.ts.map +1 -1
- package/dist/detectors/middle-dot.js +4 -2
- package/dist/detectors/middle-dot.js.map +1 -1
- package/dist/detectors/modal-conflict.d.ts.map +1 -1
- package/dist/detectors/modal-conflict.js +5 -2
- package/dist/detectors/modal-conflict.js.map +1 -1
- package/dist/detectors/notice-paragraph.d.ts +9 -0
- package/dist/detectors/notice-paragraph.d.ts.map +1 -0
- package/dist/detectors/notice-paragraph.js +21 -0
- package/dist/detectors/notice-paragraph.js.map +1 -0
- package/dist/detectors/number-style.d.ts.map +1 -1
- package/dist/detectors/number-style.js +15 -2
- package/dist/detectors/number-style.js.map +1 -1
- package/dist/detectors/number-unit.d.ts.map +1 -1
- package/dist/detectors/number-unit.js +24 -1
- package/dist/detectors/number-unit.js.map +1 -1
- package/dist/detectors/numeral-opener.d.ts +6 -0
- package/dist/detectors/numeral-opener.d.ts.map +1 -1
- package/dist/detectors/numeral-opener.js +24 -1
- package/dist/detectors/numeral-opener.js.map +1 -1
- package/dist/detectors/obligation-without-deadline.d.ts +5 -1
- package/dist/detectors/obligation-without-deadline.d.ts.map +1 -1
- package/dist/detectors/obligation-without-deadline.js +44 -23
- package/dist/detectors/obligation-without-deadline.js.map +1 -1
- package/dist/detectors/open-end-class.d.ts +16 -0
- package/dist/detectors/open-end-class.d.ts.map +1 -0
- package/dist/detectors/open-end-class.js +65 -0
- package/dist/detectors/open-end-class.js.map +1 -0
- package/dist/detectors/open-end-condition.d.ts +7 -0
- package/dist/detectors/open-end-condition.d.ts.map +1 -0
- package/dist/detectors/open-end-condition.js +35 -0
- package/dist/detectors/open-end-condition.js.map +1 -0
- package/dist/detectors/open-end-names.d.ts +9 -3
- package/dist/detectors/open-end-names.d.ts.map +1 -1
- package/dist/detectors/open-end-names.js +5 -2
- package/dist/detectors/open-end-names.js.map +1 -1
- package/dist/detectors/open-end-question.d.ts +15 -0
- package/dist/detectors/open-end-question.d.ts.map +1 -0
- package/dist/detectors/open-end-question.js +37 -0
- package/dist/detectors/open-end-question.js.map +1 -0
- package/dist/detectors/open-end-scope.d.ts +16 -0
- package/dist/detectors/open-end-scope.d.ts.map +1 -0
- package/dist/detectors/open-end-scope.js +47 -0
- package/dist/detectors/open-end-scope.js.map +1 -0
- package/dist/detectors/opening-date.d.ts +10 -0
- package/dist/detectors/opening-date.d.ts.map +1 -0
- package/dist/detectors/opening-date.js +33 -0
- package/dist/detectors/opening-date.js.map +1 -0
- package/dist/detectors/ordinary-phrase.d.ts +20 -0
- package/dist/detectors/ordinary-phrase.d.ts.map +1 -0
- package/dist/detectors/ordinary-phrase.js +54 -0
- package/dist/detectors/ordinary-phrase.js.map +1 -0
- package/dist/detectors/organisation-run.d.ts +5 -1
- package/dist/detectors/organisation-run.d.ts.map +1 -1
- package/dist/detectors/organisation-run.js +7 -0
- package/dist/detectors/organisation-run.js.map +1 -1
- package/dist/detectors/orthographic-variant.d.ts.map +1 -1
- package/dist/detectors/orthographic-variant.js +56 -20
- package/dist/detectors/orthographic-variant.js.map +1 -1
- package/dist/detectors/orthography.d.ts.map +1 -1
- package/dist/detectors/orthography.js +6 -1
- package/dist/detectors/orthography.js.map +1 -1
- package/dist/detectors/oxford-comma.d.ts.map +1 -1
- package/dist/detectors/oxford-comma.js +38 -1
- package/dist/detectors/oxford-comma.js.map +1 -1
- package/dist/detectors/person-consistency.d.ts +2 -1
- package/dist/detectors/person-consistency.d.ts.map +1 -1
- package/dist/detectors/person-consistency.js +5 -3
- package/dist/detectors/person-consistency.js.map +1 -1
- package/dist/detectors/phrase-match.d.ts +2 -0
- package/dist/detectors/phrase-match.d.ts.map +1 -1
- package/dist/detectors/phrase-match.js +11 -4
- package/dist/detectors/phrase-match.js.map +1 -1
- package/dist/detectors/placeholder-text.d.ts.map +1 -1
- package/dist/detectors/placeholder-text.js +18 -3
- package/dist/detectors/placeholder-text.js.map +1 -1
- package/dist/detectors/placeholder.d.ts +3 -1
- package/dist/detectors/placeholder.d.ts.map +1 -1
- package/dist/detectors/placeholder.js +8 -3
- package/dist/detectors/placeholder.js.map +1 -1
- package/dist/detectors/postcode-end.d.ts +3 -0
- package/dist/detectors/postcode-end.d.ts.map +1 -0
- package/dist/detectors/postcode-end.js +15 -0
- package/dist/detectors/postcode-end.js.map +1 -0
- package/dist/detectors/prose-total-words.d.ts +5 -0
- package/dist/detectors/prose-total-words.d.ts.map +1 -0
- package/dist/detectors/prose-total-words.js +14 -0
- package/dist/detectors/prose-total-words.js.map +1 -0
- package/dist/detectors/registry/kanji-compound-slip.d.ts +3 -0
- package/dist/detectors/registry/kanji-compound-slip.d.ts.map +1 -0
- package/dist/detectors/registry/kanji-compound-slip.js +3 -0
- package/dist/detectors/registry/kanji-compound-slip.js.map +1 -0
- package/dist/detectors/registry/phrase-in-use.d.ts +3 -0
- package/dist/detectors/registry/phrase-in-use.d.ts.map +1 -0
- package/dist/detectors/registry/phrase-in-use.js +3 -0
- package/dist/detectors/registry/phrase-in-use.js.map +1 -0
- package/dist/detectors/requirement-smell.d.ts.map +1 -1
- package/dist/detectors/requirement-smell.js +29 -5
- package/dist/detectors/requirement-smell.js.map +1 -1
- package/dist/detectors/scaled-figure-unit.d.ts +28 -0
- package/dist/detectors/scaled-figure-unit.d.ts.map +1 -0
- package/dist/detectors/scaled-figure-unit.js +78 -0
- package/dist/detectors/scaled-figure-unit.js.map +1 -0
- package/dist/detectors/sentence-counts.d.ts +2 -0
- package/dist/detectors/sentence-counts.d.ts.map +1 -1
- package/dist/detectors/sentence-counts.js +18 -7
- package/dist/detectors/sentence-counts.js.map +1 -1
- package/dist/detectors/sentence-ending.d.ts +2 -0
- package/dist/detectors/sentence-ending.d.ts.map +1 -1
- package/dist/detectors/sentence-ending.js +10 -3
- package/dist/detectors/sentence-ending.js.map +1 -1
- package/dist/detectors/sentence-load.d.ts +4 -1
- package/dist/detectors/sentence-load.d.ts.map +1 -1
- package/dist/detectors/sentence-load.js +9 -2
- package/dist/detectors/sentence-load.js.map +1 -1
- package/dist/detectors/signals.d.ts.map +1 -1
- package/dist/detectors/signals.js +30 -50
- package/dist/detectors/signals.js.map +1 -1
- package/dist/detectors/structure-tree.d.ts.map +1 -1
- package/dist/detectors/structure-tree.js +14 -13
- package/dist/detectors/structure-tree.js.map +1 -1
- package/dist/detectors/structure.d.ts.map +1 -1
- package/dist/detectors/structure.js +2 -1
- package/dist/detectors/structure.js.map +1 -1
- package/dist/detectors/subject-name-line.d.ts +8 -0
- package/dist/detectors/subject-name-line.d.ts.map +1 -0
- package/dist/detectors/subject-name-line.js +53 -0
- package/dist/detectors/subject-name-line.js.map +1 -0
- package/dist/detectors/table-header.d.ts.map +1 -1
- package/dist/detectors/table-header.js +13 -1
- package/dist/detectors/table-header.js.map +1 -1
- package/dist/detectors/technical-docs.d.ts.map +1 -1
- package/dist/detectors/technical-docs.js +4 -1
- package/dist/detectors/technical-docs.js.map +1 -1
- package/dist/detectors/term-mention-sentence.d.ts +4 -0
- package/dist/detectors/term-mention-sentence.d.ts.map +1 -0
- package/dist/detectors/term-mention-sentence.js +28 -0
- package/dist/detectors/term-mention-sentence.js.map +1 -0
- package/dist/detectors/titled-run.d.ts +2 -1
- package/dist/detectors/titled-run.d.ts.map +1 -1
- package/dist/detectors/titled-run.js +9 -4
- package/dist/detectors/titled-run.js.map +1 -1
- package/dist/detectors/token-shape.d.ts.map +1 -1
- package/dist/detectors/token-shape.js +13 -5
- package/dist/detectors/token-shape.js.map +1 -1
- package/dist/detectors/vague-density.d.ts +5 -2
- package/dist/detectors/vague-density.d.ts.map +1 -1
- package/dist/detectors/vague-density.js +17 -4
- package/dist/detectors/vague-density.js.map +1 -1
- package/dist/document-paragraphs.d.ts +4 -0
- package/dist/document-paragraphs.d.ts.map +1 -0
- package/dist/document-paragraphs.js +24 -0
- package/dist/document-paragraphs.js.map +1 -0
- package/dist/document.d.ts.map +1 -1
- package/dist/document.js +2 -19
- package/dist/document.js.map +1 -1
- package/dist/figure-references.d.ts +3 -0
- package/dist/figure-references.d.ts.map +1 -1
- package/dist/figure-references.js +6 -2
- package/dist/figure-references.js.map +1 -1
- package/dist/format-template.d.ts +3 -0
- package/dist/format-template.d.ts.map +1 -0
- package/dist/format-template.js +13 -0
- package/dist/format-template.js.map +1 -0
- package/dist/formula-brackets.d.ts +7 -0
- package/dist/formula-brackets.d.ts.map +1 -0
- package/dist/formula-brackets.js +29 -0
- package/dist/formula-brackets.js.map +1 -0
- package/dist/html/html-markdown.d.ts.map +1 -1
- package/dist/html/html-markdown.js +2 -1
- package/dist/html/html-markdown.js.map +1 -1
- package/dist/html/html-site-chrome.d.ts +5 -0
- package/dist/html/html-site-chrome.d.ts.map +1 -0
- package/dist/html/html-site-chrome.js +33 -0
- package/dist/html/html-site-chrome.js.map +1 -0
- package/dist/joined-view.d.ts.map +1 -1
- package/dist/joined-view.js +1 -0
- package/dist/joined-view.js.map +1 -1
- package/dist/kanji-compound-slip.d.ts +32 -0
- package/dist/kanji-compound-slip.d.ts.map +1 -0
- package/dist/kanji-compound-slip.js +63 -0
- package/dist/kanji-compound-slip.js.map +1 -0
- package/dist/latin-company-form.d.ts +6 -0
- package/dist/latin-company-form.d.ts.map +1 -0
- package/dist/latin-company-form.js +32 -0
- package/dist/latin-company-form.js.map +1 -0
- package/dist/latin-word-boundary.d.ts +15 -0
- package/dist/latin-word-boundary.d.ts.map +1 -0
- package/dist/latin-word-boundary.js +40 -0
- package/dist/latin-word-boundary.js.map +1 -0
- package/dist/lexicon-genres.d.ts +2 -0
- package/dist/lexicon-genres.d.ts.map +1 -1
- package/dist/lexicon-genres.js +1 -1
- package/dist/lexicon-genres.js.map +1 -1
- package/dist/link-shape.d.ts +5 -0
- package/dist/link-shape.d.ts.map +1 -1
- package/dist/link-shape.js +18 -0
- package/dist/link-shape.js.map +1 -1
- package/dist/link-title.d.ts +4 -0
- package/dist/link-title.d.ts.map +1 -0
- package/dist/link-title.js +33 -0
- package/dist/link-title.js.map +1 -0
- package/dist/list-item-paragraphs.d.ts +15 -0
- package/dist/list-item-paragraphs.d.ts.map +1 -0
- package/dist/list-item-paragraphs.js +19 -0
- package/dist/list-item-paragraphs.js.map +1 -0
- package/dist/literal-quote.d.ts +12 -0
- package/dist/literal-quote.d.ts.map +1 -0
- package/dist/literal-quote.js +43 -0
- package/dist/literal-quote.js.map +1 -0
- package/dist/nado-closed-list.d.ts.map +1 -1
- package/dist/nado-closed-list.js +39 -5
- package/dist/nado-closed-list.js.map +1 -1
- package/dist/name-separator.d.ts.map +1 -1
- package/dist/name-separator.js +25 -2
- package/dist/name-separator.js.map +1 -1
- package/dist/nominal-stem.d.ts +14 -1
- package/dist/nominal-stem.d.ts.map +1 -1
- package/dist/nominal-stem.js +42 -4
- package/dist/nominal-stem.js.map +1 -1
- package/dist/person-forms.d.ts +8 -0
- package/dist/person-forms.d.ts.map +1 -1
- package/dist/person-forms.js +12 -0
- package/dist/person-forms.js.map +1 -1
- package/dist/plugin.d.ts +5 -0
- package/dist/plugin.d.ts.map +1 -1
- package/dist/preamble-paragraphs.d.ts +3 -2
- package/dist/preamble-paragraphs.d.ts.map +1 -1
- package/dist/preamble-paragraphs.js +8 -2
- package/dist/preamble-paragraphs.js.map +1 -1
- package/dist/spacing-code.d.ts +1 -0
- package/dist/spacing-code.d.ts.map +1 -1
- package/dist/spacing-code.js +4 -1
- package/dist/spacing-code.js.map +1 -1
- package/dist/split-noun.d.ts +10 -0
- package/dist/split-noun.d.ts.map +1 -0
- package/dist/split-noun.js +50 -0
- package/dist/split-noun.js.map +1 -0
- package/dist/structure/attached-parts.d.ts +11 -0
- package/dist/structure/attached-parts.d.ts.map +1 -0
- package/dist/structure/attached-parts.js +31 -0
- package/dist/structure/attached-parts.js.map +1 -0
- package/dist/structure/build.d.ts.map +1 -1
- package/dist/structure/build.js +7 -4
- package/dist/structure/build.js.map +1 -1
- package/dist/structure/contents-lines.d.ts +7 -0
- package/dist/structure/contents-lines.d.ts.map +1 -0
- package/dist/structure/contents-lines.js +51 -0
- package/dist/structure/contents-lines.js.map +1 -0
- package/dist/structure/defining-names.d.ts +37 -0
- package/dist/structure/defining-names.d.ts.map +1 -0
- package/dist/structure/defining-names.js +122 -0
- package/dist/structure/defining-names.js.map +1 -0
- package/dist/structure/definition-use.d.ts +3 -2
- package/dist/structure/definition-use.d.ts.map +1 -1
- package/dist/structure/definition-use.js +99 -31
- package/dist/structure/definition-use.js.map +1 -1
- package/dist/structure/issues.d.ts +4 -1
- package/dist/structure/issues.d.ts.map +1 -1
- package/dist/structure/issues.js +5 -2
- package/dist/structure/issues.js.map +1 -1
- package/dist/structure/list-lead-in.d.ts +19 -0
- package/dist/structure/list-lead-in.d.ts.map +1 -0
- package/dist/structure/list-lead-in.js +85 -0
- package/dist/structure/list-lead-in.js.map +1 -0
- package/dist/structure/modal-conflict.d.ts +7 -2
- package/dist/structure/modal-conflict.d.ts.map +1 -1
- package/dist/structure/modal-conflict.js +36 -7
- package/dist/structure/modal-conflict.js.map +1 -1
- package/dist/structure/negated-act.d.ts +17 -0
- package/dist/structure/negated-act.d.ts.map +1 -0
- package/dist/structure/negated-act.js +32 -0
- package/dist/structure/negated-act.js.map +1 -0
- package/dist/structure/numbering.d.ts +3 -1
- package/dist/structure/numbering.d.ts.map +1 -1
- package/dist/structure/numbering.js +20 -8
- package/dist/structure/numbering.js.map +1 -1
- package/dist/structure/prose-total.d.ts +4 -0
- package/dist/structure/prose-total.d.ts.map +1 -1
- package/dist/structure/prose-total.js +16 -6
- package/dist/structure/prose-total.js.map +1 -1
- package/dist/structure/repeated-number.d.ts +24 -0
- package/dist/structure/repeated-number.d.ts.map +1 -0
- package/dist/structure/repeated-number.js +89 -0
- package/dist/structure/repeated-number.js.map +1 -0
- package/dist/structure/rounded-amount.d.ts +8 -0
- package/dist/structure/rounded-amount.d.ts.map +1 -0
- package/dist/structure/rounded-amount.js +48 -0
- package/dist/structure/rounded-amount.js.map +1 -0
- package/dist/structure/side-step.d.ts +8 -0
- package/dist/structure/side-step.d.ts.map +1 -0
- package/dist/structure/side-step.js +21 -0
- package/dist/structure/side-step.js.map +1 -0
- package/dist/structure/spanned-label-blank.d.ts +8 -0
- package/dist/structure/spanned-label-blank.d.ts.map +1 -0
- package/dist/structure/spanned-label-blank.js +37 -0
- package/dist/structure/spanned-label-blank.js.map +1 -0
- package/dist/structure/universal.d.ts +3 -1
- package/dist/structure/universal.d.ts.map +1 -1
- package/dist/structure/universal.js +9 -2
- package/dist/structure/universal.js.map +1 -1
- package/dist/structure/unspaced-heading.d.ts +10 -0
- package/dist/structure/unspaced-heading.d.ts.map +1 -0
- package/dist/structure/unspaced-heading.js +24 -0
- package/dist/structure/unspaced-heading.js.map +1 -0
- package/dist/structure/written-list-path.d.ts +10 -0
- package/dist/structure/written-list-path.d.ts.map +1 -0
- package/dist/structure/written-list-path.js +37 -0
- package/dist/structure/written-list-path.js.map +1 -0
- package/dist/structure/written-resolve.d.ts +1 -1
- package/dist/structure/written-resolve.d.ts.map +1 -1
- package/dist/structure/written-resolve.js +15 -9
- package/dist/structure/written-resolve.js.map +1 -1
- package/dist/title-line.d.ts +21 -0
- package/dist/title-line.d.ts.map +1 -0
- package/dist/title-line.js +27 -0
- package/dist/title-line.js.map +1 -0
- package/package.json +3 -3
- package/rules/all-caps-shouting.yaml +2 -2
- package/rules/announced-count-mismatch.yaml +2 -2
- package/rules/bare-url-mix.yaml +2 -2
- package/rules/bracket-nesting.yaml +2 -2
- package/rules/bracket-width-consistency.yaml +2 -2
- package/rules/broken-link.yaml +2 -2
- package/rules/cliche.yaml +4 -3
- package/rules/code-fence-language-mix.yaml +2 -2
- package/rules/comparative-without-baseline.yaml +2 -2
- package/rules/dangling-figure-reference.yaml +2 -2
- package/rules/date-without-year.yaml +10 -2
- package/rules/defined-term-form.yaml +2 -2
- package/rules/doubled-punctuation.yaml +4 -4
- package/rules/doubled-word.yaml +2 -2
- package/rules/duplicate-definition.yaml +2 -2
- package/rules/duration-mismatch.yaml +2 -2
- package/rules/email-greeting-closing.yaml +2 -2
- package/rules/email-subject-length.yaml +2 -2
- package/rules/empty-table-cell.yaml +2 -2
- package/rules/footnote-mismatch.yaml +2 -2
- package/rules/heading-echo.yaml +2 -2
- package/rules/ime-residue.yaml +2 -2
- package/rules/kanji-compound-slip.yaml +67 -0
- package/rules/kutoten-consistency.yaml +4 -4
- package/rules/latin-spacing.yaml +2 -2
- package/rules/list-item-form-mix.yaml +2 -2
- package/rules/max-kanji-continuous.yaml +6 -4
- package/rules/modal-conflict.yaml +1 -0
- package/rules/nado-closed-list.yaml +2 -2
- package/rules/ngram-repetition.yaml +2 -2
- package/rules/no-doubled-joshi.yaml +2 -2
- package/rules/no-mixed-desumasu.yaml +3 -3
- package/rules/no-nakaguro-parallel.yaml +2 -0
- package/rules/number-without-unit.yaml +9 -4
- package/rules/numbering-gap.yaml +7 -2
- package/rules/numeral-sentence-start.yaml +2 -2
- package/rules/obligation-without-deadline.yaml +2 -2
- package/rules/orthographic-variant.yaml +2 -2
- package/rules/oxford-comma-consistency.yaml +3 -3
- package/rules/preamble-length.yaml +2 -2
- package/rules/requirement-smell.yaml +3 -3
- package/rules/table-header-variant.yaml +2 -2
- package/rules/teki-overuse.yaml +3 -2
- package/rules/total-mismatch.yaml +2 -2
- package/rules/translationese-density.yaml +2 -2
- package/rules/undefined-acronym.yaml +4 -2
- package/rules/unfilled-placeholder.yaml +4 -4
- package/rules/use-before-definition.yaml +2 -2
- package/rules/vague-word-density.yaml +2 -2
- package/skills/chaff/SKILL.md +7 -1
- package/src/announced-count.ts +35 -8
- package/src/chronology-rows.ts +86 -0
- package/src/company-name-run.ts +37 -0
- package/src/compound-head.ts +25 -0
- package/src/date-format.ts +43 -4
- package/src/derived/count-per-period.ts +28 -0
- package/src/detectors/acronym-context.ts +30 -1
- package/src/detectors/acronym-expansion.ts +47 -6
- package/src/detectors/acronym-known.ts +40 -0
- package/src/detectors/acronym-link.ts +37 -0
- package/src/detectors/acronym-name.ts +27 -4
- package/src/detectors/addressee-edge.ts +16 -0
- package/src/detectors/announced-count.ts +1 -0
- package/src/detectors/bare-comparative.ts +44 -3
- package/src/detectors/blank-structure.ts +21 -10
- package/src/detectors/bracket-specified.ts +25 -0
- package/src/detectors/broken-link.ts +26 -2
- package/src/detectors/caps-caption.ts +42 -0
- package/src/detectors/caps-phrase.ts +17 -0
- package/src/detectors/char-shape.ts +30 -9
- package/src/detectors/cited-title.ts +15 -3
- package/src/detectors/counter-edge.ts +17 -2
- package/src/detectors/date-any-year.ts +27 -6
- package/src/detectors/date-format.ts +2 -1
- package/src/detectors/date-year.ts +64 -16
- package/src/detectors/deadline-unit.ts +40 -0
- package/src/detectors/defined-term-form.ts +32 -12
- package/src/detectors/definition-use.ts +7 -3
- package/src/detectors/derived-numbers.ts +19 -4
- package/src/detectors/doubled-across-aside.ts +36 -0
- package/src/detectors/doubled-punctuation.ts +28 -7
- package/src/detectors/doubled-word.ts +2 -1
- package/src/detectors/ellipsis.ts +16 -1
- package/src/detectors/email-letter.ts +97 -21
- package/src/detectors/field-label.ts +25 -0
- package/src/detectors/footnote-mismatch.ts +39 -7
- package/src/detectors/graded-step.ts +52 -0
- package/src/detectors/gram-openers.ts +21 -0
- package/src/detectors/gram-place.ts +10 -1
- package/src/detectors/gram-predicate.ts +43 -3
- package/src/detectors/heading-echo.ts +31 -3
- package/src/detectors/ime-residue.ts +4 -3
- package/src/detectors/kanji-compound-slip.ts +61 -0
- package/src/detectors/kutoten-consistency.ts +93 -14
- package/src/detectors/lexicon.ts +12 -3
- package/src/detectors/link-shape.ts +10 -3
- package/src/detectors/list-item-ending.ts +51 -0
- package/src/detectors/list-parallel.ts +158 -33
- package/src/detectors/list-sentence.ts +5 -1
- package/src/detectors/mark-shape.ts +64 -10
- package/src/detectors/mentioned-phrase.ts +38 -0
- package/src/detectors/middle-dot.ts +7 -2
- package/src/detectors/modal-conflict.ts +5 -2
- package/src/detectors/notice-paragraph.ts +33 -0
- package/src/detectors/number-style.ts +14 -2
- package/src/detectors/number-unit.ts +28 -1
- package/src/detectors/numeral-opener.ts +27 -1
- package/src/detectors/obligation-without-deadline.ts +29 -4
- package/src/detectors/open-end-class.ts +78 -0
- package/src/detectors/open-end-condition.ts +40 -0
- package/src/detectors/open-end-names.ts +9 -3
- package/src/detectors/open-end-question.ts +50 -0
- package/src/detectors/open-end-scope.ts +62 -0
- package/src/detectors/opening-date.ts +47 -0
- package/src/detectors/ordinary-phrase.ts +62 -0
- package/src/detectors/organisation-run.ts +13 -2
- package/src/detectors/orthographic-variant.ts +85 -25
- package/src/detectors/orthography.ts +10 -1
- package/src/detectors/oxford-comma.ts +40 -1
- package/src/detectors/person-consistency.ts +5 -3
- package/src/detectors/phrase-match.ts +16 -5
- package/src/detectors/placeholder-text.ts +18 -3
- package/src/detectors/placeholder.ts +10 -3
- package/src/detectors/postcode-end.ts +16 -0
- package/src/detectors/prose-total-words.ts +18 -0
- package/src/detectors/registry/kanji-compound-slip.ts +4 -0
- package/src/detectors/registry/phrase-in-use.ts +4 -0
- package/src/detectors/requirement-smell.ts +38 -5
- package/src/detectors/scaled-figure-unit.ts +103 -0
- package/src/detectors/sentence-counts.ts +24 -8
- package/src/detectors/sentence-ending.ts +12 -3
- package/src/detectors/sentence-load.ts +9 -2
- package/src/detectors/signals.ts +33 -51
- package/src/detectors/structure-tree.ts +14 -14
- package/src/detectors/structure.ts +2 -1
- package/src/detectors/subject-name-line.ts +66 -0
- package/src/detectors/table-header.ts +15 -1
- package/src/detectors/technical-docs.ts +5 -1
- package/src/detectors/term-mention-sentence.ts +31 -0
- package/src/detectors/titled-run.ts +9 -4
- package/src/detectors/token-shape.ts +14 -5
- package/src/detectors/vague-density.ts +20 -4
- package/src/document-paragraphs.ts +27 -0
- package/src/document.ts +3 -23
- package/src/figure-references.ts +7 -2
- package/src/format-template.ts +15 -0
- package/src/formula-brackets.ts +36 -0
- package/src/html/html-markdown.ts +2 -1
- package/src/html/html-site-chrome.ts +47 -0
- package/src/joined-view.ts +1 -0
- package/src/kanji-compound-slip.ts +98 -0
- package/src/latin-company-form.ts +38 -0
- package/src/latin-word-boundary.ts +46 -0
- package/src/lexicon-genres.ts +1 -1
- package/src/link-shape.ts +19 -0
- package/src/link-title.ts +38 -0
- package/src/list-item-paragraphs.ts +22 -0
- package/src/literal-quote.ts +55 -0
- package/src/nado-closed-list.ts +48 -5
- package/src/name-separator.ts +27 -2
- package/src/nominal-stem.ts +62 -5
- package/src/person-forms.ts +15 -0
- package/src/plugin.ts +5 -0
- package/src/preamble-paragraphs.ts +13 -3
- package/src/spacing-code.ts +5 -1
- package/src/split-noun.ts +63 -0
- package/src/structure/attached-parts.ts +32 -0
- package/src/structure/build.ts +7 -4
- package/src/structure/contents-lines.ts +63 -0
- package/src/structure/defining-names.ts +155 -0
- package/src/structure/definition-use.ts +116 -25
- package/src/structure/issues.ts +7 -3
- package/src/structure/list-lead-in.ts +94 -0
- package/src/structure/modal-conflict.ts +42 -7
- package/src/structure/negated-act.ts +49 -0
- package/src/structure/numbering.ts +24 -9
- package/src/structure/prose-total.ts +20 -6
- package/src/structure/repeated-number.ts +110 -0
- package/src/structure/rounded-amount.ts +55 -0
- package/src/structure/side-step.ts +21 -0
- package/src/structure/spanned-label-blank.ts +39 -0
- package/src/structure/universal.ts +11 -2
- package/src/structure/unspaced-heading.ts +27 -0
- package/src/structure/written-list-path.ts +46 -0
- package/src/structure/written-resolve.ts +16 -9
- package/src/title-line.ts +52 -0
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"phrase-in-use.js","sourceRoot":"","sources":["../../../src/detectors/registry/phrase-in-use.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AAEjD,MAAM,CAAC,MAAM,QAAQ,GAAa,WAAW,CAAC"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"requirement-smell.d.ts","sourceRoot":"","sources":["../../src/detectors/requirement-smell.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAW,OAAO,EAA+B,QAAQ,EAAE,MAAM,cAAc,CAAC;
|
|
1
|
+
{"version":3,"file":"requirement-smell.d.ts","sourceRoot":"","sources":["../../src/detectors/requirement-smell.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAW,OAAO,EAA+B,QAAQ,EAAE,MAAM,cAAc,CAAC;AAsCtG,eAAO,MAAM,aAAa,GAAI,UAAU,QAAQ,EAAE,SAAS,OAAO,EAAE,YAAY,OAAO,KAAG,OACwB,CAAC;AAsEnH,eAAO,MAAM,gBAAgB,EAAE,QAQ9B,CAAC"}
|
|
@@ -1,11 +1,19 @@
|
|
|
1
1
|
import { entryCloses, entryIn, entryRanges } from "./lexicon-match.js";
|
|
2
|
+
import { classWordsOf, isClassCue, namesClassAfter } from "./open-end-class.js";
|
|
3
|
+
import { glossesCondition } from "./open-end-condition.js";
|
|
2
4
|
import { openEndAt, openEndNames } from "./open-end-names.js";
|
|
5
|
+
import { asksAfter, questionWordsOf } from "./open-end-question.js";
|
|
6
|
+
import { isScopeCue, openOnPurpose, scopeWordsOf } from "./open-end-scope.js";
|
|
3
7
|
import { statesConcrete } from "./vague-deadline.js";
|
|
4
8
|
// Requirements smells (Femmer et al. 2017; Berry et al. 2003; ISO/IEC/IEEE 29148): a loophole, an open-ended list and
|
|
5
9
|
// "and/or", in a sentence that states a requirement. What makes a sentence a requirement, and each smell's words, come
|
|
6
10
|
// from the language's lexicons; this file knows neither language.
|
|
7
11
|
const OPEN_END = "requirement-open-end";
|
|
8
12
|
const LOOPHOLE = "requirement-loophole";
|
|
13
|
+
/** The words of a question that an open list gives example answers to. */
|
|
14
|
+
const OPEN_QUESTION = "requirement-open-question";
|
|
15
|
+
/** The words of a condition whose bracketed examples an open list may give. */
|
|
16
|
+
const OPEN_CONDITION = "requirement-open-condition";
|
|
9
17
|
const TIME = "time";
|
|
10
18
|
const CONCRETE_DEADLINE = "concrete-deadline";
|
|
11
19
|
/** Each smell: the lexicon of its words, and the finding's variant (the rule's messages). */
|
|
@@ -40,18 +48,34 @@ const findingOf = (sentence, entry, variant, offset) => ({
|
|
|
40
48
|
const boundedInTime = (doc, sentence, entry, lexicon) => lexicon === LOOPHOLE &&
|
|
41
49
|
entry.group === TIME &&
|
|
42
50
|
statesConcrete({ start: sentence.span.start, text: sentence.text }, (doc.lexicons[CONCRETE_DEADLINE] ?? []).map((limit) => limit.pattern));
|
|
43
|
-
/**
|
|
51
|
+
/** A lexicon's words; the open-end lexicon also holds the cues of a class named after the list and of what holds it, which are no smell. */
|
|
52
|
+
const smellWords = (doc, lexicon) => (doc.lexicons[lexicon] ?? []).filter((entry) => !isClassCue(entry) && !isScopeCue(entry));
|
|
53
|
+
/**
|
|
54
|
+
* Where the word stands as a smell. An open-end word inside a name the document gives (業務執行取締役等), giving examples
|
|
55
|
+
* of a class named right after it (印刷等、二次的に利用), or ending a list a condition, a permission or a prohibition holds
|
|
56
|
+
* (転載等をしてはならない), is none; without tokens none of these can be told apart.
|
|
57
|
+
*/
|
|
44
58
|
const smellAt = (doc, sentence, entry, lexicon, names) => {
|
|
45
59
|
if (!entryIn(sentence, entry) || boundedInTime(doc, sentence, entry, lexicon))
|
|
46
60
|
return undefined;
|
|
47
61
|
const tokenized = sentence.tokens !== undefined && (entry.tokens?.length ?? 0) > 0;
|
|
48
|
-
|
|
62
|
+
if (lexicon !== OPEN_END || !tokenized)
|
|
63
|
+
return offsetOf(sentence, entry);
|
|
64
|
+
const classWords = classWordsOf(doc.lexicons[OPEN_END] ?? []);
|
|
65
|
+
const scopeWords = scopeWordsOf(doc.lexicons[OPEN_END] ?? []);
|
|
66
|
+
return openEndAt(doc, sentence, entry, names(), (tokens, range) => namesClassAfter(tokens, range, classWords) || openOnPurpose(tokens, range, scopeWords));
|
|
49
67
|
};
|
|
68
|
+
/** An open list right before a question the reader answers (…措置など、どのような措置を求めるのか): its items are example answers. */
|
|
69
|
+
const answersQuestion = (doc, sentence, entry, lexicon, offset) => lexicon === OPEN_END && asksAfter(sentence, offset + entry.pattern.length, questionWordsOf(doc.lexicons[OPEN_QUESTION] ?? []));
|
|
70
|
+
/** An open list in brackets right after a condition word (場合(単一テーブルへの SELECT 等)): examples of when, not of what. */
|
|
71
|
+
const glossesConditionAt = (doc, sentence, lexicon, offset) => lexicon === OPEN_END &&
|
|
72
|
+
glossesCondition(sentence, offset, (doc.lexicons[OPEN_CONDITION] ?? []).map((entry) => entry.pattern));
|
|
50
73
|
/** The first word of each smell in one requirement sentence: one finding per smell, not per word. */
|
|
51
|
-
const smellsIn = (doc, sentence, names) => SMELLS.flatMap(({ lexicon, variant }) => (doc
|
|
74
|
+
const smellsIn = (doc, sentence, names) => SMELLS.flatMap(({ lexicon, variant }) => smellWords(doc, lexicon)
|
|
52
75
|
.flatMap((entry) => {
|
|
53
76
|
const offset = smellAt(doc, sentence, entry, lexicon, names);
|
|
54
|
-
|
|
77
|
+
const examples = offset !== undefined && (answersQuestion(doc, sentence, entry, lexicon, offset) || glossesConditionAt(doc, sentence, lexicon, offset));
|
|
78
|
+
return offset === undefined || examples ? [] : [findingOf(sentence, entry, variant, offset)];
|
|
55
79
|
})
|
|
56
80
|
.slice(0, 1));
|
|
57
81
|
export const requirementSmell = (doc) => {
|
|
@@ -60,7 +84,7 @@ export const requirementSmell = (doc) => {
|
|
|
60
84
|
const statements = (doc.lexicons[STATEMENTS] ?? []).map((entry) => entry.pattern);
|
|
61
85
|
// Read once, and only when a requirement has an open-end word to tell apart from a name.
|
|
62
86
|
const cache = {};
|
|
63
|
-
const names = () => (cache.names ??= openEndNames(doc, doc
|
|
87
|
+
const names = () => (cache.names ??= openEndNames(doc, smellWords(doc, OPEN_END), statements));
|
|
64
88
|
return doc.sentences.filter((sentence) => isRequirement(sentence, markers, notMarkers)).flatMap((sentence) => smellsIn(doc, sentence, names));
|
|
65
89
|
};
|
|
66
90
|
//# sourceMappingURL=requirement-smell.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"requirement-smell.js","sourceRoot":"","sources":["../../src/detectors/requirement-smell.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,WAAW,EAAE,OAAO,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AACvE,OAAO,EAAE,SAAS,EAAE,YAAY,EAAqB,MAAM,qBAAqB,CAAC;AACjF,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAErD,sHAAsH;AACtH,uHAAuH;AACvH,kEAAkE;AAElE,MAAM,QAAQ,GAAG,sBAAsB,CAAC;AACxC,MAAM,QAAQ,GAAG,sBAAsB,CAAC;AACxC,MAAM,IAAI,GAAG,MAAM,CAAC;AACpB,MAAM,iBAAiB,GAAG,mBAAmB,CAAC;AAE9C,6FAA6F;AAC7F,MAAM,MAAM,GAAG;IACb,EAAE,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,UAAU,EAAE;IAC1C,EAAE,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,UAAU,EAAE;IAC1C,EAAE,OAAO,EAAE,oBAAoB,EAAE,OAAO,EAAE,QAAQ,EAAE;CAC5C,CAAC;AAEX,MAAM,OAAO,GAAG,oBAAoB,CAAC;AACrC,kHAAkH;AAClH,MAAM,WAAW,GAAG,wBAAwB,CAAC;AAC7C,oHAAoH;AACpH,MAAM,UAAU,GAAG,sBAAsB,CAAC;AAE1C,mHAAmH;AACnH,MAAM,UAAU,GAAG,CAAC,QAAkB,EAAE,KAAmB,EAAW,EAAE,CAAC,CAAC,KAAK,CAAC,QAAQ,KAAK,OAAO,CAAC,CAAC,CAAC,WAAW,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,CAAC;AAEhK,MAAM,CAAC,MAAM,aAAa,GAAG,CAAC,QAAkB,EAAE,OAAgB,EAAE,UAAmB,EAAW,EAAE,CAClG,OAAO,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,CAAC;AAEnH,wGAAwG;AACxG,MAAM,QAAQ,GAAG,CAAC,QAAkB,EAAE,KAAmB,EAAU,EAAE;IACnE,MAAM,KAAK,GAAG,WAAW,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;IAC9C,OAAO,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,MAAM,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,IAAI,CAAC,KAAK,IAAI,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;AACzH,CAAC,CAAC;AAEF,MAAM,SAAS,GAAG,CAAC,QAAkB,EAAE,KAAmB,EAAE,OAAe,EAAE,MAAc,EAAW,EAAE,CAAC,CAAC;IACxG,IAAI,EAAE,EAAE;IACR,QAAQ,EAAE,SAAS;IACnB,IAAI,EAAE,CAAC;IACP,MAAM,EAAE,CAAC;IACT,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE;IAC3B,OAAO;IACP,MAAM,EAAE,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,MAAM,EAAE;CAC3C,CAAC,CAAC;AAEH,kJAAkJ;AAClJ,MAAM,aAAa,GAAG,CAAC,GAAkB,EAAE,QAAkB,EAAE,KAAmB,EAAE,OAAe,EAAW,EAAE,CAC9G,OAAO,KAAK,QAAQ;IACpB,KAAK,CAAC,KAAK,KAAK,IAAI;IACpB,cAAc,CACZ,EAAE,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,EAAE,QAAQ,CAAC,IAAI,EAAE,EACnD,CAAC,GAAG,CAAC,QAAQ,CAAC,iBAAiB,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,CACtE,CAAC;AAEJ,
|
|
1
|
+
{"version":3,"file":"requirement-smell.js","sourceRoot":"","sources":["../../src/detectors/requirement-smell.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,WAAW,EAAE,OAAO,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AACvE,OAAO,EAAE,YAAY,EAAE,UAAU,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAChF,OAAO,EAAE,gBAAgB,EAAE,MAAM,yBAAyB,CAAC;AAC3D,OAAO,EAAE,SAAS,EAAE,YAAY,EAAqB,MAAM,qBAAqB,CAAC;AACjF,OAAO,EAAE,SAAS,EAAE,eAAe,EAAE,MAAM,wBAAwB,CAAC;AACpE,OAAO,EAAE,UAAU,EAAE,aAAa,EAAE,YAAY,EAAE,MAAM,qBAAqB,CAAC;AAC9E,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAErD,sHAAsH;AACtH,uHAAuH;AACvH,kEAAkE;AAElE,MAAM,QAAQ,GAAG,sBAAsB,CAAC;AACxC,MAAM,QAAQ,GAAG,sBAAsB,CAAC;AACxC,0EAA0E;AAC1E,MAAM,aAAa,GAAG,2BAA2B,CAAC;AAClD,+EAA+E;AAC/E,MAAM,cAAc,GAAG,4BAA4B,CAAC;AACpD,MAAM,IAAI,GAAG,MAAM,CAAC;AACpB,MAAM,iBAAiB,GAAG,mBAAmB,CAAC;AAE9C,6FAA6F;AAC7F,MAAM,MAAM,GAAG;IACb,EAAE,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,UAAU,EAAE;IAC1C,EAAE,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,UAAU,EAAE;IAC1C,EAAE,OAAO,EAAE,oBAAoB,EAAE,OAAO,EAAE,QAAQ,EAAE;CAC5C,CAAC;AAEX,MAAM,OAAO,GAAG,oBAAoB,CAAC;AACrC,kHAAkH;AAClH,MAAM,WAAW,GAAG,wBAAwB,CAAC;AAC7C,oHAAoH;AACpH,MAAM,UAAU,GAAG,sBAAsB,CAAC;AAE1C,mHAAmH;AACnH,MAAM,UAAU,GAAG,CAAC,QAAkB,EAAE,KAAmB,EAAW,EAAE,CAAC,CAAC,KAAK,CAAC,QAAQ,KAAK,OAAO,CAAC,CAAC,CAAC,WAAW,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,CAAC;AAEhK,MAAM,CAAC,MAAM,aAAa,GAAG,CAAC,QAAkB,EAAE,OAAgB,EAAE,UAAmB,EAAW,EAAE,CAClG,OAAO,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,CAAC;AAEnH,wGAAwG;AACxG,MAAM,QAAQ,GAAG,CAAC,QAAkB,EAAE,KAAmB,EAAU,EAAE;IACnE,MAAM,KAAK,GAAG,WAAW,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;IAC9C,OAAO,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,MAAM,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,IAAI,CAAC,KAAK,IAAI,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;AACzH,CAAC,CAAC;AAEF,MAAM,SAAS,GAAG,CAAC,QAAkB,EAAE,KAAmB,EAAE,OAAe,EAAE,MAAc,EAAW,EAAE,CAAC,CAAC;IACxG,IAAI,EAAE,EAAE;IACR,QAAQ,EAAE,SAAS;IACnB,IAAI,EAAE,CAAC;IACP,MAAM,EAAE,CAAC;IACT,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE;IAC3B,OAAO;IACP,MAAM,EAAE,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,MAAM,EAAE;CAC3C,CAAC,CAAC;AAEH,kJAAkJ;AAClJ,MAAM,aAAa,GAAG,CAAC,GAAkB,EAAE,QAAkB,EAAE,KAAmB,EAAE,OAAe,EAAW,EAAE,CAC9G,OAAO,KAAK,QAAQ;IACpB,KAAK,CAAC,KAAK,KAAK,IAAI;IACpB,cAAc,CACZ,EAAE,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,EAAE,QAAQ,CAAC,IAAI,EAAE,EACnD,CAAC,GAAG,CAAC,QAAQ,CAAC,iBAAiB,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,CACtE,CAAC;AAEJ,4IAA4I;AAC5I,MAAM,UAAU,GAAG,CAAC,GAAkB,EAAE,OAAe,EAAkB,EAAE,CACzE,CAAC,GAAG,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC;AAE5F;;;;GAIG;AACH,MAAM,OAAO,GAAG,CAAC,GAAkB,EAAE,QAAkB,EAAE,KAAmB,EAAE,OAAe,EAAE,KAAyB,EAAsB,EAAE;IAC9I,IAAI,CAAC,OAAO,CAAC,QAAQ,EAAE,KAAK,CAAC,IAAI,aAAa,CAAC,GAAG,EAAE,QAAQ,EAAE,KAAK,EAAE,OAAO,CAAC;QAAE,OAAO,SAAS,CAAC;IAChG,MAAM,SAAS,GAAG,QAAQ,CAAC,MAAM,KAAK,SAAS,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,MAAM,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC;IACnF,IAAI,OAAO,KAAK,QAAQ,IAAI,CAAC,SAAS;QAAE,OAAO,QAAQ,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC;IACzE,MAAM,UAAU,GAAG,YAAY,CAAC,GAAG,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAC;IAC9D,MAAM,UAAU,GAAG,YAAY,CAAC,GAAG,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAC;IAC9D,OAAO,SAAS,CAAC,GAAG,EAAE,QAAQ,EAAE,KAAK,EAAE,KAAK,EAAE,EAAE,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC,eAAe,CAAC,MAAM,EAAE,KAAK,EAAE,UAAU,CAAC,IAAI,aAAa,CAAC,MAAM,EAAE,KAAK,EAAE,UAAU,CAAC,CAAC,CAAC;AAC7J,CAAC,CAAC;AAEF,oHAAoH;AACpH,MAAM,eAAe,GAAG,CAAC,GAAkB,EAAE,QAAkB,EAAE,KAAmB,EAAE,OAAe,EAAE,MAAc,EAAW,EAAE,CAChI,OAAO,KAAK,QAAQ,IAAI,SAAS,CAAC,QAAQ,EAAE,MAAM,GAAG,KAAK,CAAC,OAAO,CAAC,MAAM,EAAE,eAAe,CAAC,GAAG,CAAC,QAAQ,CAAC,aAAa,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC;AAEjI,oHAAoH;AACpH,MAAM,kBAAkB,GAAG,CAAC,GAAkB,EAAE,QAAkB,EAAE,OAAe,EAAE,MAAc,EAAW,EAAE,CAC9G,OAAO,KAAK,QAAQ;IACpB,gBAAgB,CACd,QAAQ,EACR,MAAM,EACN,CAAC,GAAG,CAAC,QAAQ,CAAC,cAAc,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,CACnE,CAAC;AAEJ,qGAAqG;AACrG,MAAM,QAAQ,GAAG,CAAC,GAAkB,EAAE,QAAkB,EAAE,KAAyB,EAAa,EAAE,CAChG,MAAM,CAAC,OAAO,CAAC,CAAC,EAAE,OAAO,EAAE,OAAO,EAAE,EAAE,EAAE,CACtC,UAAU,CAAC,GAAG,EAAE,OAAO,CAAC;KACrB,OAAO,CAAC,CAAC,KAAK,EAAE,EAAE;IACjB,MAAM,MAAM,GAAG,OAAO,CAAC,GAAG,EAAE,QAAQ,EAAE,KAAK,EAAE,OAAO,EAAE,KAAK,CAAC,CAAC;IAC7D,MAAM,QAAQ,GAAG,MAAM,KAAK,SAAS,IAAI,CAAC,eAAe,CAAC,GAAG,EAAE,QAAQ,EAAE,KAAK,EAAE,OAAO,EAAE,MAAM,CAAC,IAAI,kBAAkB,CAAC,GAAG,EAAE,QAAQ,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC,CAAC;IACxJ,OAAO,MAAM,KAAK,SAAS,IAAI,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,QAAQ,EAAE,KAAK,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC,CAAC;AAC/F,CAAC,CAAC;KACD,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CACf,CAAC;AAEJ,MAAM,CAAC,MAAM,gBAAgB,GAAa,CAAC,GAAG,EAAa,EAAE;IAC3D,MAAM,OAAO,GAAG,GAAG,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC;IAC5C,MAAM,UAAU,GAAG,GAAG,CAAC,QAAQ,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC;IACnD,MAAM,UAAU,GAAG,CAAC,GAAG,CAAC,QAAQ,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;IAClF,yFAAyF;IACzF,MAAM,KAAK,GAA6B,EAAE,CAAC;IAC3C,MAAM,KAAK,GAAG,GAAiB,EAAE,CAAC,CAAC,KAAK,CAAC,KAAK,KAAK,YAAY,CAAC,GAAG,EAAE,UAAU,CAAC,GAAG,EAAE,QAAQ,CAAC,EAAE,UAAU,CAAC,CAAC,CAAC;IAC7G,OAAO,GAAG,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,aAAa,CAAC,QAAQ,EAAE,OAAO,EAAE,UAAU,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,QAAQ,CAAC,GAAG,EAAE,QAAQ,EAAE,KAAK,CAAC,CAAC,CAAC;AAChJ,CAAC,CAAC"}
|
|
@@ -0,0 +1,28 @@
|
|
|
1
|
+
import type { CurrencyForm } from "./currency-notation.ts";
|
|
2
|
+
/** A number with a word of scale: from its currency mark (if any) to the end of its unit, and the unit ("" when bare). */
|
|
3
|
+
export type ScaledFigure = {
|
|
4
|
+
readonly offset: number;
|
|
5
|
+
readonly end: number;
|
|
6
|
+
readonly written: string;
|
|
7
|
+
readonly scale: string;
|
|
8
|
+
readonly unit: string;
|
|
9
|
+
};
|
|
10
|
+
export type ScaleWords = {
|
|
11
|
+
readonly scales: readonly string[];
|
|
12
|
+
readonly forms: readonly CurrencyForm[];
|
|
13
|
+
readonly joiners: readonly string[];
|
|
14
|
+
};
|
|
15
|
+
export type ScaleGap = {
|
|
16
|
+
readonly figure: ScaledFigure;
|
|
17
|
+
readonly unit: string;
|
|
18
|
+
};
|
|
19
|
+
/** How many numbers must carry the unit before the one without it reads as a slip rather than a way of writing. */
|
|
20
|
+
export declare const MIN_WITH_UNIT = 5;
|
|
21
|
+
/** Every number with a word of scale in a text, in order; one that runs on into another number (1億2,000万円) is left out. */
|
|
22
|
+
export declare const scaledFiguresIn: (text: string, words: ScaleWords) => ScaledFigure[];
|
|
23
|
+
/**
|
|
24
|
+
* The one bare number of a word of scale whose other numbers all carry one unit, at least MIN_WITH_UNIT times. Two bare
|
|
25
|
+
* numbers, or two units (百万円 and 百万株), and the document is read as writing the word that way on purpose.
|
|
26
|
+
*/
|
|
27
|
+
export declare const scaleGaps: (text: string, figures: readonly ScaledFigure[], joiners: readonly string[]) => ScaleGap[];
|
|
28
|
+
//# sourceMappingURL=scaled-figure-unit.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"scaled-figure-unit.d.ts","sourceRoot":"","sources":["../../src/detectors/scaled-figure-unit.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,wBAAwB,CAAC;AAG3D,0HAA0H;AAC1H,MAAM,MAAM,YAAY,GAAG;IAAE,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;IAAC,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAC;IAAC,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IAAC,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;IAAC,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAA;CAAE,CAAC;AAEtJ,MAAM,MAAM,UAAU,GAAG;IAAE,QAAQ,CAAC,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC;IAAC,QAAQ,CAAC,KAAK,EAAE,SAAS,YAAY,EAAE,CAAC;IAAC,QAAQ,CAAC,OAAO,EAAE,SAAS,MAAM,EAAE,CAAA;CAAE,CAAC;AAE9I,MAAM,MAAM,QAAQ,GAAG;IAAE,QAAQ,CAAC,MAAM,EAAE,YAAY,CAAC;IAAC,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAA;CAAE,CAAC;AAEhF,mHAAmH;AACnH,eAAO,MAAM,aAAa,IAAI,CAAC;AAiD/B,2HAA2H;AAC3H,eAAO,MAAM,eAAe,GAAI,MAAM,MAAM,EAAE,OAAO,UAAU,KAAG,YAAY,EAC2D,CAAC;AAqB1I;;;GAGG;AACH,eAAO,MAAM,SAAS,GAAI,MAAM,MAAM,EAAE,SAAS,SAAS,YAAY,EAAE,EAAE,SAAS,SAAS,MAAM,EAAE,KAAG,QAAQ,EAW9G,CAAC"}
|
|
@@ -0,0 +1,78 @@
|
|
|
1
|
+
// number-without-unit, in prose: a number written with a word of scale and no unit (83,494百万, 83 million) where the
|
|
2
|
+
// document writes that word of scale with one unit many times and without one nowhere else (83,494百万円, $83 million).
|
|
3
|
+
// Pure; the words of scale, the currency marks and the joiners come from the language's lexicons.
|
|
4
|
+
import { escapeRegExp } from "../orthography.js";
|
|
5
|
+
import { AMOUNT } from "./currency-notation.js";
|
|
6
|
+
/** How many numbers must carry the unit before the one without it reads as a slip rather than a way of writing. */
|
|
7
|
+
export const MIN_WITH_UNIT = 5;
|
|
8
|
+
const LATIN = /[A-Za-z]/u;
|
|
9
|
+
const longestFirst = (words) => words.toSorted((left, right) => right.length - left.length);
|
|
10
|
+
const figurePattern = (scales) => {
|
|
11
|
+
const scale = longestFirst(scales)
|
|
12
|
+
.map((word) => `${escapeRegExp(word)}${LATIN.test(word) ? "(?![A-Za-z])" : ""}`)
|
|
13
|
+
.join("|");
|
|
14
|
+
return new RegExp(`(?<![\\w.,,.0-9])${AMOUNT}\\s?(?:${scale})`, "gu");
|
|
15
|
+
};
|
|
16
|
+
const formMatches = (text, pattern, side) => {
|
|
17
|
+
if (side === "before")
|
|
18
|
+
return text.endsWith(pattern) && !(LATIN.test(pattern.charAt(0)) && LATIN.test(text.charAt(text.length - pattern.length - 1)));
|
|
19
|
+
return text.startsWith(pattern) && !(LATIN.test(pattern.at(-1) ?? "") && LATIN.test(text.charAt(pattern.length)));
|
|
20
|
+
};
|
|
21
|
+
/** The currency mark written before a number ($, US$, ¥), with the one space it may have. */
|
|
22
|
+
const markBefore = (text, offset, forms) => {
|
|
23
|
+
const before = text.slice(Math.max(0, offset - 8), offset);
|
|
24
|
+
const spaced = before.endsWith(" ") ? before.slice(0, -1) : before;
|
|
25
|
+
const form = longestFirst(forms.filter((entry) => entry.position === "before").map((entry) => entry.pattern)).find((pattern) => formMatches(spaced, pattern, "before"));
|
|
26
|
+
if (form === undefined)
|
|
27
|
+
return "";
|
|
28
|
+
return spaced === before ? form : `${form} `;
|
|
29
|
+
};
|
|
30
|
+
/** What follows a word of scale as its unit: a currency word (円, dollars), else the next word ("people", 人, ドル); "" when none. */
|
|
31
|
+
const unitAfter = (rest, forms) => {
|
|
32
|
+
const spaced = rest.startsWith(" ") ? rest.slice(1) : rest;
|
|
33
|
+
const form = longestFirst(forms.filter((entry) => entry.position === "after").map((entry) => entry.pattern)).find((pattern) => formMatches(spaced, pattern, "after"));
|
|
34
|
+
const word = form ?? /^(?:[A-Za-z]+|[\p{Script=Katakana}ー]+|\p{Script=Han})/u.exec(spaced)?.[0];
|
|
35
|
+
if (word === undefined)
|
|
36
|
+
return "";
|
|
37
|
+
return spaced === rest ? word : ` ${word}`;
|
|
38
|
+
};
|
|
39
|
+
const figureAt = (text, match, words) => {
|
|
40
|
+
const end = match.index + match[0].length;
|
|
41
|
+
const rest = text.slice(end);
|
|
42
|
+
if (/^\s?[0-90-9]/u.test(rest))
|
|
43
|
+
return undefined;
|
|
44
|
+
const mark = markBefore(text, match.index, words.forms);
|
|
45
|
+
const after = mark === "" ? unitAfter(rest, words.forms) : "";
|
|
46
|
+
const scale = longestFirst(words.scales).find((word) => match[0].endsWith(word)) ?? "";
|
|
47
|
+
return { offset: match.index - mark.length, end: end + after.length, written: match[0], scale, unit: (mark || after).trim() };
|
|
48
|
+
};
|
|
49
|
+
/** Every number with a word of scale in a text, in order; one that runs on into another number (1億2,000万円) is left out. */
|
|
50
|
+
export const scaledFiguresIn = (text, words) => words.scales.length === 0 ? [] : [...text.matchAll(figurePattern(words.scales))].flatMap((match) => figureAt(text, match, words) ?? []);
|
|
51
|
+
/** A bare number joined to its neighbour by a range or list word, so that one unit serves both (100百万〜200百万円). */
|
|
52
|
+
const sharesUnit = (text, left, right, joiners) => {
|
|
53
|
+
if (left === undefined || right === undefined)
|
|
54
|
+
return false;
|
|
55
|
+
const gap = text.slice(left.end, right.offset).trim();
|
|
56
|
+
return gap === "" || joiners.includes(gap) || joiners.includes(gap.toLowerCase());
|
|
57
|
+
};
|
|
58
|
+
const gapOfScale = (text, figures, joiners) => {
|
|
59
|
+
const bare = figures.filter((figure, index) => figure.unit === "" && !sharesUnit(text, figure, figures[index + 1], joiners) && !sharesUnit(text, figures[index - 1], figure, joiners));
|
|
60
|
+
const withUnit = figures.filter((figure) => figure.unit !== "");
|
|
61
|
+
const units = new Set(withUnit.map((figure) => figure.unit));
|
|
62
|
+
const [unit] = units;
|
|
63
|
+
const [figure] = bare;
|
|
64
|
+
if (figure === undefined || unit === undefined || bare.length !== 1 || units.size !== 1 || withUnit.length < MIN_WITH_UNIT)
|
|
65
|
+
return [];
|
|
66
|
+
return [{ figure, unit }];
|
|
67
|
+
};
|
|
68
|
+
/**
|
|
69
|
+
* The one bare number of a word of scale whose other numbers all carry one unit, at least MIN_WITH_UNIT times. Two bare
|
|
70
|
+
* numbers, or two units (百万円 and 百万株), and the document is read as writing the word that way on purpose.
|
|
71
|
+
*/
|
|
72
|
+
export const scaleGaps = (text, figures, joiners) => {
|
|
73
|
+
const scales = [...new Set(figures.map((figure) => figure.scale))];
|
|
74
|
+
return scales
|
|
75
|
+
.flatMap((scale) => gapOfScale(text, figures.filter((figure) => figure.scale === scale), joiners))
|
|
76
|
+
.toSorted((left, right) => left.figure.offset - right.figure.offset);
|
|
77
|
+
};
|
|
78
|
+
//# sourceMappingURL=scaled-figure-unit.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"scaled-figure-unit.js","sourceRoot":"","sources":["../../src/detectors/scaled-figure-unit.ts"],"names":[],"mappings":"AAAA,oHAAoH;AACpH,qHAAqH;AACrH,kGAAkG;AAClG,OAAO,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AAEjD,OAAO,EAAE,MAAM,EAAE,MAAM,wBAAwB,CAAC;AAShD,mHAAmH;AACnH,MAAM,CAAC,MAAM,aAAa,GAAG,CAAC,CAAC;AAE/B,MAAM,KAAK,GAAG,WAAW,CAAC;AAC1B,MAAM,YAAY,GAAG,CAAC,KAAwB,EAAY,EAAE,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC;AAEzH,MAAM,aAAa,GAAG,CAAC,MAAyB,EAAU,EAAE;IAC1D,MAAM,KAAK,GAAG,YAAY,CAAC,MAAM,CAAC;SAC/B,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,GAAG,YAAY,CAAC,IAAI,CAAC,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,cAAc,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC;SAC/E,IAAI,CAAC,GAAG,CAAC,CAAC;IACb,OAAO,IAAI,MAAM,CAAC,oBAAoB,MAAM,UAAU,KAAK,GAAG,EAAE,IAAI,CAAC,CAAC;AACxE,CAAC,CAAC;AAEF,MAAM,WAAW,GAAG,CAAC,IAAY,EAAE,OAAe,EAAE,IAAwB,EAAW,EAAE;IACvF,IAAI,IAAI,KAAK,QAAQ;QAAE,OAAO,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IACtJ,OAAO,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC;AACpH,CAAC,CAAC;AAEF,6FAA6F;AAC7F,MAAM,UAAU,GAAG,CAAC,IAAY,EAAE,MAAc,EAAE,KAA8B,EAAU,EAAE;IAC1F,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,MAAM,GAAG,CAAC,CAAC,EAAE,MAAM,CAAC,CAAC;IAC3D,MAAM,MAAM,GAAG,MAAM,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC;IACnE,MAAM,IAAI,GAAG,YAAY,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,QAAQ,KAAK,QAAQ,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,EAAE,CAC7H,WAAW,CAAC,MAAM,EAAE,OAAO,EAAE,QAAQ,CAAC,CACvC,CAAC;IACF,IAAI,IAAI,KAAK,SAAS;QAAE,OAAO,EAAE,CAAC;IAClC,OAAO,MAAM,KAAK,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,IAAI,GAAG,CAAC;AAC/C,CAAC,CAAC;AAEF,kIAAkI;AAClI,MAAM,SAAS,GAAG,CAAC,IAAY,EAAE,KAA8B,EAAU,EAAE;IACzE,MAAM,MAAM,GAAG,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;IAC3D,MAAM,IAAI,GAAG,YAAY,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,QAAQ,KAAK,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,EAAE,CAC5H,WAAW,CAAC,MAAM,EAAE,OAAO,EAAE,OAAO,CAAC,CACtC,CAAC;IACF,MAAM,IAAI,GAAG,IAAI,IAAI,wDAAwD,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;IAChG,IAAI,IAAI,KAAK,SAAS;QAAE,OAAO,EAAE,CAAC;IAClC,OAAO,MAAM,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,IAAI,EAAE,CAAC;AAC7C,CAAC,CAAC;AAEF,MAAM,QAAQ,GAAG,CAAC,IAAY,EAAE,KAAsB,EAAE,KAAiB,EAA4B,EAAE;IACrG,MAAM,GAAG,GAAG,KAAK,CAAC,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC;IAC1C,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;IAC7B,IAAI,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,SAAS,CAAC;IACjD,MAAM,IAAI,GAAG,UAAU,CAAC,IAAI,EAAE,KAAK,CAAC,KAAK,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC;IACxD,MAAM,KAAK,GAAG,IAAI,KAAK,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC,IAAI,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;IAC9D,MAAM,KAAK,GAAG,YAAY,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;IACvF,OAAO,EAAE,MAAM,EAAE,KAAK,CAAC,KAAK,GAAG,IAAI,CAAC,MAAM,EAAE,GAAG,EAAE,GAAG,GAAG,KAAK,CAAC,MAAM,EAAE,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,IAAI,IAAI,KAAK,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC;AAChI,CAAC,CAAC;AAEF,2HAA2H;AAC3H,MAAM,CAAC,MAAM,eAAe,GAAG,CAAC,IAAY,EAAE,KAAiB,EAAkB,EAAE,CACjF,KAAK,CAAC,MAAM,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,QAAQ,CAAC,aAAa,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,QAAQ,CAAC,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC;AAE1I,kHAAkH;AAClH,MAAM,UAAU,GAAG,CAAC,IAAY,EAAE,IAA8B,EAAE,KAA+B,EAAE,OAA0B,EAAW,EAAE;IACxI,IAAI,IAAI,KAAK,SAAS,IAAI,KAAK,KAAK,SAAS;QAAE,OAAO,KAAK,CAAC;IAC5D,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,CAAC;IACtD,OAAO,GAAG,KAAK,EAAE,IAAI,OAAO,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,OAAO,CAAC,QAAQ,CAAC,GAAG,CAAC,WAAW,EAAE,CAAC,CAAC;AACpF,CAAC,CAAC;AAEF,MAAM,UAAU,GAAG,CAAC,IAAY,EAAE,OAAgC,EAAE,OAA0B,EAAc,EAAE;IAC5G,MAAM,IAAI,GAAG,OAAO,CAAC,MAAM,CACzB,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC,MAAM,CAAC,IAAI,KAAK,EAAE,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,CAAC,KAAK,GAAG,CAAC,CAAC,EAAE,OAAO,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,OAAO,CAAC,KAAK,GAAG,CAAC,CAAC,EAAE,MAAM,EAAE,OAAO,CAAC,CAC1J,CAAC;IACF,MAAM,QAAQ,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,IAAI,KAAK,EAAE,CAAC,CAAC;IAChE,MAAM,KAAK,GAAG,IAAI,GAAG,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC;IAC7D,MAAM,CAAC,IAAI,CAAC,GAAG,KAAK,CAAC;IACrB,MAAM,CAAC,MAAM,CAAC,GAAG,IAAI,CAAC;IACtB,IAAI,MAAM,KAAK,SAAS,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,IAAI,KAAK,CAAC,IAAI,KAAK,CAAC,IAAI,QAAQ,CAAC,MAAM,GAAG,aAAa;QAAE,OAAO,EAAE,CAAC;IACtI,OAAO,CAAC,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC,CAAC;AAC5B,CAAC,CAAC;AAEF;;;GAGG;AACH,MAAM,CAAC,MAAM,SAAS,GAAG,CAAC,IAAY,EAAE,OAAgC,EAAE,OAA0B,EAAc,EAAE;IAClH,MAAM,MAAM,GAAG,CAAC,GAAG,IAAI,GAAG,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;IACnE,OAAO,MAAM;SACV,OAAO,CAAC,CAAC,KAAK,EAAE,EAAE,CACjB,UAAU,CACR,IAAI,EACJ,OAAO,CAAC,MAAM,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,KAAK,KAAK,KAAK,CAAC,EAClD,OAAO,CACR,CACF;SACA,QAAQ,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;AACzE,CAAC,CAAC"}
|
|
@@ -1,4 +1,6 @@
|
|
|
1
1
|
import type { Detector, Lexicon, Sentence, Token } from "../plugin.ts";
|
|
2
|
+
/** The terms whose counted word belongs to the term itself (電磁的方法), not to the writer's style. */
|
|
3
|
+
export declare const FIXED_TERM_LEXICON = "teki-fixed-term";
|
|
2
4
|
/** A sentence holding more than limit tokens written as one of the lexicon's words (the suffix 的). */
|
|
3
5
|
export declare const sentenceWordCount: Detector;
|
|
4
6
|
/** A comma between two nouns separates listed items (名前、会社名、連絡先), not clauses. A bracketed note after an item is part of it. */
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"sentence-counts.d.ts","sourceRoot":"","sources":["../../src/detectors/sentence-counts.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAA4B,OAAO,EAAiB,QAAQ,
|
|
1
|
+
{"version":3,"file":"sentence-counts.d.ts","sourceRoot":"","sources":["../../src/detectors/sentence-counts.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAA4B,OAAO,EAAiB,QAAQ,EAAQ,KAAK,EAAE,MAAM,cAAc,CAAC;AAStH,kGAAkG;AAClG,eAAO,MAAM,kBAAkB,oBAAoB,CAAC;AAmCpD,sGAAsG;AACtG,eAAO,MAAM,iBAAiB,EAAE,QAG/B,CAAC;AAsBF,gIAAgI;AAChI,eAAO,MAAM,cAAc,GAAI,IAAI,MAAM,EAAE,QAAQ,SAAS,KAAK,EAAE,KAAG,OACyB,CAAC;AAEhG,mHAAmH;AACnH,eAAO,MAAM,kBAAkB,EAAE,QAGhC,CAAC;AAIF,yIAAyI;AACzI,eAAO,MAAM,wBAAwB,EAAE,QAGtC,CAAC;AAEF,KAAK,MAAM,GAAG;IAAE,QAAQ,CAAC,QAAQ,EAAE,QAAQ,CAAC;IAAC,QAAQ,CAAC,MAAM,EAAE,MAAM,GAAG,SAAS,CAAA;CAAE,CAAC;AASnF,6GAA6G;AAC7G,eAAO,MAAM,kBAAkB,6BAA6B,CAAC;AAE7D,yFAAyF;AACzF,eAAO,MAAM,UAAU,GAAI,WAAW,SAAS,QAAQ,EAAE,EAAE,SAAS,OAAO,EAAE,aAAY,OAAY,KAAG,MAAM,EAAE,EAW5E,CAAC;AAcrC,mHAAmH;AACnH,eAAO,MAAM,SAAS,EAAE,QAQvB,CAAC"}
|
|
@@ -1,10 +1,21 @@
|
|
|
1
|
+
import { escapeRegExp } from "../orthography.js";
|
|
1
2
|
import { isWithinAny, quotedIn } from "../quoted-span.js";
|
|
2
3
|
import { entryOpens } from "./lexicon-match.js";
|
|
3
4
|
const wordsOf = (lexicon) => new Set((lexicon ?? []).map((entry) => entry.pattern));
|
|
4
|
-
/** The
|
|
5
|
-
const
|
|
6
|
-
|
|
7
|
-
|
|
5
|
+
/** The terms whose counted word belongs to the term itself (電磁的方法), not to the writer's style. */
|
|
6
|
+
export const FIXED_TERM_LEXICON = "teki-fixed-term";
|
|
7
|
+
/** Where the sentence holds one of the terms, in the document's coordinates. */
|
|
8
|
+
const termSpans = (sentence, terms) => terms.flatMap((entry) => [...sentence.text.matchAll(new RegExp(escapeRegExp(entry.pattern), "gu"))].map((match) => ({
|
|
9
|
+
start: sentence.span.start + match.index,
|
|
10
|
+
end: sentence.span.start + match.index + entry.pattern.length,
|
|
11
|
+
})));
|
|
12
|
+
/**
|
|
13
|
+
* The tokens of a sentence that pass the test, outside anything quoted in 「」『』 and outside the fixed terms: neither a
|
|
14
|
+
* quotation nor a term the law defines (電磁的方法) is the writer's to rephrase.
|
|
15
|
+
*/
|
|
16
|
+
const countedTokens = (sentence, test, fixedTerms) => {
|
|
17
|
+
const kept = [...quotedIn(sentence), ...termSpans(sentence, fixedTerms)];
|
|
18
|
+
return (sentence.tokens ?? []).filter((token, at, tokens) => test(token, at, tokens) && !isWithinAny(kept, token.span));
|
|
8
19
|
};
|
|
9
20
|
const findingOf = (sentence, counted, limit) => ({
|
|
10
21
|
rule: "",
|
|
@@ -14,14 +25,14 @@ const findingOf = (sentence, counted, limit) => ({
|
|
|
14
25
|
quote: sentence.text.trim(),
|
|
15
26
|
values: { word: counted[0]?.surface ?? "", count: counted.length, limit, offset: sentence.span.start },
|
|
16
27
|
});
|
|
17
|
-
const sentencesOver = (doc, limit, test) => doc.sentences.flatMap((sentence) => {
|
|
18
|
-
const counted = countedTokens(sentence, test);
|
|
28
|
+
const sentencesOver = (doc, limit, test, fixedTerms = []) => doc.sentences.flatMap((sentence) => {
|
|
29
|
+
const counted = countedTokens(sentence, test, fixedTerms);
|
|
19
30
|
return counted.length > limit ? [findingOf(sentence, counted, limit)] : [];
|
|
20
31
|
});
|
|
21
32
|
/** A sentence holding more than limit tokens written as one of the lexicon's words (the suffix 的). */
|
|
22
33
|
export const sentenceWordCount = (doc, options) => {
|
|
23
34
|
const words = wordsOf(options.lexicon);
|
|
24
|
-
return sentencesOver(doc, options.limit, (token) => words.has(token.surface));
|
|
35
|
+
return sentencesOver(doc, options.limit, (token) => words.has(token.surface), doc.lexicons[FIXED_TERM_LEXICON] ?? []);
|
|
25
36
|
};
|
|
26
37
|
const NOMINAL = new Set(["NOUN", "PROPN", "NUM"]);
|
|
27
38
|
const isNominal = (token) => token !== undefined && NOMINAL.has(token.pos);
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"sentence-counts.js","sourceRoot":"","sources":["../../src/detectors/sentence-counts.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,WAAW,EAAE,QAAQ,EAAE,MAAM,mBAAmB,CAAC;AAC1D,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAIhD,MAAM,OAAO,GAAG,CAAC,OAA4B,EAAuB,EAAE,CAAC,IAAI,GAAG,CAAC,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC;AAE9H,
|
|
1
|
+
{"version":3,"file":"sentence-counts.js","sourceRoot":"","sources":["../../src/detectors/sentence-counts.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AACjD,OAAO,EAAE,WAAW,EAAE,QAAQ,EAAE,MAAM,mBAAmB,CAAC;AAC1D,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAIhD,MAAM,OAAO,GAAG,CAAC,OAA4B,EAAuB,EAAE,CAAC,IAAI,GAAG,CAAC,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC;AAE9H,kGAAkG;AAClG,MAAM,CAAC,MAAM,kBAAkB,GAAG,iBAAiB,CAAC;AAEpD,gFAAgF;AAChF,MAAM,SAAS,GAAG,CAAC,QAAkB,EAAE,KAAc,EAAU,EAAE,CAC/D,KAAK,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,EAAE,CACtB,CAAC,GAAG,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,MAAM,CAAC,YAAY,CAAC,KAAK,CAAC,OAAO,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;IACzF,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC,KAAK;IACxC,GAAG,EAAE,QAAQ,CAAC,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC,KAAK,GAAG,KAAK,CAAC,OAAO,CAAC,MAAM;CAC9D,CAAC,CAAC,CACJ,CAAC;AAEJ;;;GAGG;AACH,MAAM,aAAa,GAAG,CAAC,QAAkB,EAAE,IAAe,EAAE,UAAmB,EAAW,EAAE;IAC1F,MAAM,IAAI,GAAG,CAAC,GAAG,QAAQ,CAAC,QAAQ,CAAC,EAAE,GAAG,SAAS,CAAC,QAAQ,EAAE,UAAU,CAAC,CAAC,CAAC;IACzE,OAAO,CAAC,QAAQ,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,EAAE,MAAM,EAAE,EAAE,CAAC,IAAI,CAAC,KAAK,EAAE,EAAE,EAAE,MAAM,CAAC,IAAI,CAAC,WAAW,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC;AAC1H,CAAC,CAAC;AAEF,MAAM,SAAS,GAAG,CAAC,QAAkB,EAAE,OAAyB,EAAE,KAAa,EAAW,EAAE,CAAC,CAAC;IAC5F,IAAI,EAAE,EAAE;IACR,QAAQ,EAAE,MAAM;IAChB,IAAI,EAAE,CAAC;IACP,MAAM,EAAE,CAAC;IACT,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE;IAC3B,MAAM,EAAE,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC,EAAE,OAAO,IAAI,EAAE,EAAE,KAAK,EAAE,OAAO,CAAC,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE;CACvG,CAAC,CAAC;AAEH,MAAM,aAAa,GAAG,CAAC,GAAkB,EAAE,KAAa,EAAE,IAAe,EAAE,aAAsB,EAAE,EAAa,EAAE,CAChH,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,QAAQ,EAAE,EAAE;IACjC,MAAM,OAAO,GAAG,aAAa,CAAC,QAAQ,EAAE,IAAI,EAAE,UAAU,CAAC,CAAC;IAC1D,OAAO,OAAO,CAAC,MAAM,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,QAAQ,EAAE,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;AAC7E,CAAC,CAAC,CAAC;AAEL,sGAAsG;AACtG,MAAM,CAAC,MAAM,iBAAiB,GAAa,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE;IACrE,MAAM,KAAK,GAAG,OAAO,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC;IACvC,OAAO,aAAa,CAAC,GAAG,EAAE,OAAO,CAAC,KAAK,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,OAAO,CAAC,EAAE,GAAG,CAAC,QAAQ,CAAC,kBAAkB,CAAC,IAAI,EAAE,CAAC,CAAC;AACxH,CAAC,CAAC;AAEF,MAAM,OAAO,GAAwB,IAAI,GAAG,CAAC,CAAC,MAAM,EAAE,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC;AAEvE,MAAM,SAAS,GAAG,CAAC,KAAwB,EAAW,EAAE,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;AAEvG,MAAM,kBAAkB,GAAqC,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC;AAElH,MAAM,OAAO,GAAG,CAAC,KAAwB,EAAW,EAAE,CAAC,KAAK,KAAK,SAAS,IAAI,KAAK,CAAC,OAAO,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC;AAE1G,yGAAyG;AACzG,MAAM,aAAa,GAAG,CAAC,EAAU,EAAE,MAAwB,EAAU,EAAE;IACrE,MAAM,MAAM,GAAG,MAAM,CAAC,aAAa,CAAC,CAAC,KAAK,EAAE,KAAK,EAAE,EAAE,CAAC,KAAK,GAAG,EAAE,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC;IACrF,MAAM,OAAO,GAAG,kBAAkB,CAAC,MAAM,CAAC,MAAM,CAAC,EAAE,OAAO,IAAI,EAAE,CAAC,CAAC;IAClE,IAAI,OAAO,KAAK,SAAS;QAAE,OAAO,MAAM,CAAC;IACzC,MAAM,MAAM,GAAG,MAAM,CAAC,aAAa,CAAC,CAAC,KAAK,EAAE,KAAK,EAAE,EAAE,CAAC,KAAK,GAAG,MAAM,IAAI,KAAK,CAAC,OAAO,KAAK,OAAO,CAAC,CAAC;IACnG,OAAO,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,aAAa,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;AAChE,CAAC,CAAC;AAEF,qEAAqE;AACrE,MAAM,cAAc,GAAG,CAAC,EAAU,EAAE,MAAwB,EAAU,EAAE,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,KAAK,EAAE,KAAK,EAAE,EAAE,CAAC,KAAK,GAAG,EAAE,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC;AAE3I,gIAAgI;AAChI,MAAM,CAAC,MAAM,cAAc,GAAG,CAAC,EAAU,EAAE,MAAwB,EAAW,EAAE,CAC9E,SAAS,CAAC,MAAM,CAAC,aAAa,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC,CAAC,IAAI,SAAS,CAAC,MAAM,CAAC,cAAc,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC;AAEhG,mHAAmH;AACnH,MAAM,CAAC,MAAM,kBAAkB,GAAa,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE;IACtE,MAAM,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC;IACxC,OAAO,aAAa,CAAC,GAAG,EAAE,OAAO,CAAC,KAAK,EAAE,CAAC,KAAK,EAAE,EAAE,EAAE,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,cAAc,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC,CAAC;AAC5H,CAAC,CAAC;AAEF,MAAM,WAAW,GAAG,OAAO,CAAC;AAE5B,yIAAyI;AACzI,MAAM,CAAC,MAAM,wBAAwB,GAAa,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE;IAC5E,MAAM,KAAK,GAAG,OAAO,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC;IACvC,OAAO,aAAa,CAAC,GAAG,EAAE,OAAO,CAAC,KAAK,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,GAAG,KAAK,WAAW,IAAI,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC;AAC7G,CAAC,CAAC;AAIF;;;GAGG;AACH,MAAM,QAAQ,GAAG,CAAC,QAAkB,EAAE,OAAgB,EAAE,UAAmB,EAAsB,EAAE,CACjG,UAAU,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,EAAE,OAAO,CAAC;AAEtI,6GAA6G;AAC7G,MAAM,CAAC,MAAM,kBAAkB,GAAG,0BAA0B,CAAC;AAE7D,yFAAyF;AACzF,MAAM,CAAC,MAAM,UAAU,GAAG,CAAC,SAA8B,EAAE,OAAgB,EAAE,aAAsB,EAAE,EAAc,EAAE,CACnH,SAAS;KACN,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,EAAE,QAAQ,EAAE,MAAM,EAAE,QAAQ,CAAC,QAAQ,EAAE,OAAO,EAAE,UAAU,CAAC,EAAE,CAAC,CAAC;KAClF,MAAM,CACL,CAAC,IAAI,EAAE,MAAM,EAAE,EAAE;IACf,IAAI,MAAM,CAAC,MAAM,KAAK,SAAS;QAAE,OAAO,CAAC,GAAG,IAAI,EAAE,EAAE,CAAC,CAAC;IACtD,MAAM,IAAI,GAAG,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;IAC/B,OAAO,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,GAAG,IAAI,EAAE,MAAM,CAAC,CAAC,CAAC;AACnD,CAAC,EACD,CAAC,EAAE,CAAC,CACL;KACA,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;AAErC,MAAM,UAAU,GAAG,CAAC,GAAsB,EAAE,OAAwB,EAAW,EAAE;IAC/E,MAAM,KAAK,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC;IACrB,OAAO;QACL,IAAI,EAAE,EAAE;QACR,QAAQ,EAAE,MAAM;QAChB,IAAI,EAAE,CAAC;QACP,MAAM,EAAE,CAAC;QACT,KAAK,EAAE,GAAG,CAAC,GAAG,CAAC,CAAC,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC;QAC/D,MAAM,EAAE,EAAE,IAAI,EAAE,KAAK,EAAE,MAAM,IAAI,EAAE,EAAE,KAAK,EAAE,GAAG,CAAC,MAAM,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,KAAK,IAAI,CAAC,EAAE;KACxH,CAAC;AACJ,CAAC,CAAC;AAEF,mHAAmH;AACnH,MAAM,CAAC,MAAM,SAAS,GAAa,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE;IAC7D,MAAM,OAAO,GAAG,OAAO,CAAC,OAAO,IAAI,EAAE,CAAC;IACtC,MAAM,UAAU,GAAG,GAAG,CAAC,QAAQ,CAAC,kBAAkB,CAAC,IAAI,EAAE,CAAC;IAC1D,OAAO,GAAG,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC,SAAS,EAAE,EAAE,CAC1C,UAAU,CAAC,SAAS,CAAC,SAAS,EAAE,OAAO,EAAE,UAAU,CAAC;SACjD,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,MAAM,GAAG,OAAO,CAAC,KAAK,CAAC;SAC3C,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,UAAU,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC,CAC1C,CAAC;AACJ,CAAC,CAAC"}
|
|
@@ -6,6 +6,8 @@ import type { Detector, LexiconEntry, ProseDocument, Sentence } from "../plugin.
|
|
|
6
6
|
* どちらが正しいかは決めない。決めると、方針の違う書き手にそのまま無視される。
|
|
7
7
|
* 見るのは文書の中での一貫性だけで、少数派のほうを指摘する。spec §12.3 の方針を日本語にも。
|
|
8
8
|
* 終止符で終わらないもの(見出しの下の `MaaSサービス` のような名前だけの行)と、項目や段落の頭の太字の札(**速く書ける。**)は文として数えない。
|
|
9
|
+
* 年表の行(1921年信号装置の製作を開始する。)も、文書の文体と揃えない書き方なので数えない。
|
|
10
|
+
* 断り書きの段落(…ご了承ください。)も、本文とは別の声で書く決まり文句なので数えない(notice-paragraph.ts)。
|
|
9
11
|
*/
|
|
10
12
|
type Entry = {
|
|
11
13
|
readonly sentence: Sentence;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"sentence-ending.d.ts","sourceRoot":"","sources":["../../src/detectors/sentence-ending.ts"],"names":[],"mappings":"AAEA,OAAO,
|
|
1
|
+
{"version":3,"file":"sentence-ending.d.ts","sourceRoot":"","sources":["../../src/detectors/sentence-ending.ts"],"names":[],"mappings":"AAEA,OAAO,EAA2D,KAAK,QAAQ,EAAE,MAAM,eAAe,CAAC;AAIvG,OAAO,KAAK,EAAE,QAAQ,EAAW,YAAY,EAAE,aAAa,EAAE,QAAQ,EAAQ,MAAM,cAAc,CAAC;AAEnG;;;;;;;;GAQG;AACH,KAAK,KAAK,GAAG;IAAE,QAAQ,CAAC,QAAQ,EAAE,QAAQ,CAAC;IAAC,QAAQ,CAAC,QAAQ,EAAE,QAAQ,CAAC;IAAC,QAAQ,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,CAAA;CAAE,CAAC;AAoB9G,2DAA2D;AAC3D,eAAO,MAAM,eAAe,GAAI,KAAK,aAAa,EAAE,QAAQ,SAAS,YAAY,EAAE,KAAG,KAAK,EAkB1F,CAAC;AAEF,eAAO,MAAM,cAAc,EAAE,QAY5B,CAAC"}
|
|
@@ -1,7 +1,9 @@
|
|
|
1
1
|
import { endingTokens, isClosed } from "../sentence-shape.js";
|
|
2
2
|
import { enumeratedRuns, enumeratorStarts, numberedStarts } from "./enumerated-runs.js";
|
|
3
|
-
import { continuesInto, groupOf, registerOf, slipsOf } from "./register.js";
|
|
3
|
+
import { continuesInto, groupOf, majorityOf, registerOf, slipsOf } from "./register.js";
|
|
4
4
|
import { isRunInLabel } from "./run-in-label.js";
|
|
5
|
+
import { noticeSentences } from "./notice-paragraph.js";
|
|
6
|
+
import { chronologyRows } from "../chronology-rows.js";
|
|
5
7
|
const registerOfSentence = (sentence, polite, neutral) => {
|
|
6
8
|
const ending = endingTokens(sentence);
|
|
7
9
|
const tokens = sentence.tokens ?? [];
|
|
@@ -19,12 +21,17 @@ const runsOf = (doc, registers) => {
|
|
|
19
21
|
export const judgedSentences = (doc, polite) => {
|
|
20
22
|
const neutral = doc.lexicons["neutral-ending"] ?? [];
|
|
21
23
|
const lists = doc.lists.map((list) => list.span);
|
|
22
|
-
const
|
|
23
|
-
|
|
24
|
+
const rows = chronologyRows(doc);
|
|
25
|
+
const all = doc.sentences.flatMap((sentence, index) => {
|
|
26
|
+
if (!isClosed(sentence) || rows.has(sentence) || continuesInto(sentence, doc.sentences[index + 1]) || isRunInLabel(doc.source, sentence.span))
|
|
24
27
|
return [];
|
|
25
28
|
const register = registerOfSentence(sentence, polite, neutral);
|
|
26
29
|
return register === undefined ? [] : [{ sentence, register }];
|
|
27
30
|
});
|
|
31
|
+
const registers = new Map(all.map((entry) => [entry.sentence, entry.register]));
|
|
32
|
+
const majority = majorityOf(all.map((entry) => ({ register: entry.register, group: undefined })));
|
|
33
|
+
const notices = noticeSentences(doc.paragraphs, doc.lexicons["notice-phrase"] ?? [], (sentence) => registers.get(sentence), majority);
|
|
34
|
+
const found = all.filter((entry) => !notices.has(entry.sentence));
|
|
28
35
|
const runs = runsOf(doc, found.map((entry) => entry.register));
|
|
29
36
|
return found.map((entry) => ({ ...entry, group: groupOf(entry.sentence.span.start, lists, runs) }));
|
|
30
37
|
};
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"sentence-ending.js","sourceRoot":"","sources":["../../src/detectors/sentence-ending.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,QAAQ,EAAE,MAAM,sBAAsB,CAAC;AAC9D,OAAO,EAAE,cAAc,EAAE,gBAAgB,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACxF,OAAO,EAAE,aAAa,EAAE,OAAO,EAAE,UAAU,EAAE,OAAO,EAAiB,MAAM,eAAe,CAAC;
|
|
1
|
+
{"version":3,"file":"sentence-ending.js","sourceRoot":"","sources":["../../src/detectors/sentence-ending.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,QAAQ,EAAE,MAAM,sBAAsB,CAAC;AAC9D,OAAO,EAAE,cAAc,EAAE,gBAAgB,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACxF,OAAO,EAAE,aAAa,EAAE,OAAO,EAAE,UAAU,EAAE,UAAU,EAAE,OAAO,EAAiB,MAAM,eAAe,CAAC;AACvG,OAAO,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AACjD,OAAO,EAAE,eAAe,EAAE,MAAM,uBAAuB,CAAC;AACxD,OAAO,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAC;AAcvD,MAAM,kBAAkB,GAAG,CAAC,QAAkB,EAAE,MAA+B,EAAE,OAAgC,EAAwB,EAAE;IACzI,MAAM,MAAM,GAAG,YAAY,CAAC,QAAQ,CAAC,CAAC;IACtC,MAAM,MAAM,GAAG,QAAQ,CAAC,MAAM,IAAI,EAAE,CAAC;IACrC,MAAM,IAAI,GAAG,MAAM,CAAC,CAAC,CAAC,KAAK,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;IACtE,OAAO,UAAU,CAAC,MAAM,EAAE,IAAI,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC;AACpF,CAAC,CAAC;AAEF,yDAAyD;AACzD,MAAM,MAAM,GAAG,CAAC,GAAkB,EAAE,SAA8B,EAAU,EAAE;IAC5E,MAAM,IAAI,GAAG,IAAI,GAAG,CAAC,SAAS,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC;IACrE,IAAI,IAAI,KAAK,SAAS;QAAE,OAAO,EAAE,CAAC;IAClC,OAAO,cAAc,CACnB,GAAG,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE,CAAC,SAAS,CAAC,IAAI,CAAC,EACjD,gBAAgB,CAAC,cAAc,CAAC,IAAI,CAAC,EAAE,GAAG,CAAC,SAAS,EAAE,GAAG,CAAC,MAAM,CAAC,EACjE,GAAG,CAAC,MAAM,CACX,CAAC;AACJ,CAAC,CAAC;AAEF,2DAA2D;AAC3D,MAAM,CAAC,MAAM,eAAe,GAAG,CAAC,GAAkB,EAAE,MAA+B,EAAW,EAAE;IAC9F,MAAM,OAAO,GAAG,GAAG,CAAC,QAAQ,CAAC,gBAAgB,CAAC,IAAI,EAAE,CAAC;IACrD,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACjD,MAAM,IAAI,GAAG,cAAc,CAAC,GAAG,CAAC,CAAC;IACjC,MAAM,GAAG,GAAG,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,QAAQ,EAAE,KAAK,EAAgD,EAAE;QAClG,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,aAAa,CAAC,QAAQ,EAAE,GAAG,CAAC,SAAS,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,IAAI,YAAY,CAAC,GAAG,CAAC,MAAM,EAAE,QAAQ,CAAC,IAAI,CAAC;YAAE,OAAO,EAAE,CAAC;QACzJ,MAAM,QAAQ,GAAG,kBAAkB,CAAC,QAAQ,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC;QAC/D,OAAO,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,QAAQ,EAAE,CAAC,CAAC;IAChE,CAAC,CAAC,CAAC;IACH,MAAM,SAAS,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,QAAQ,EAAE,KAAK,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC;IAChF,MAAM,QAAQ,GAAG,UAAU,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,EAAE,QAAQ,EAAE,KAAK,CAAC,QAAQ,EAAE,KAAK,EAAE,SAAS,EAAE,CAAC,CAAC,CAAC,CAAC;IAClG,MAAM,OAAO,GAAG,eAAe,CAAC,GAAG,CAAC,UAAU,EAAE,GAAG,CAAC,QAAQ,CAAC,eAAe,CAAC,IAAI,EAAE,EAAE,CAAC,QAAQ,EAAE,EAAE,CAAC,SAAS,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAE,QAAQ,CAAC,CAAC;IACtI,MAAM,KAAK,GAAG,GAAG,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,CAAC;IAClE,MAAM,IAAI,GAAG,MAAM,CACjB,GAAG,EACH,KAAK,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,QAAQ,CAAC,CACrC,CAAC;IACF,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,KAAK,EAAS,EAAE,CAAC,CAAC,EAAE,GAAG,KAAK,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE,KAAK,EAAE,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC;AAC7G,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,cAAc,GAAa,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE;IAClE,MAAM,MAAM,GAAG,eAAe,CAAC,GAAG,EAAE,OAAO,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC;IAC3D,wDAAwD;IACxD,OAAO,OAAO,CAAC,MAAM,EAAE,OAAO,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC,CAAC;QACpF,IAAI,EAAE,mBAAmB;QACzB,QAAQ,EAAE,SAAS;QACnB,IAAI,EAAE,CAAC;QACP,MAAM,EAAE,CAAC;QACT,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE;QAC3B,GAAG,CAAC,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC;QACnD,MAAM,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,EAAE,MAAM,EAAE,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE;KACrE,CAAC,CAAC,CAAC;AACN,CAAC,CAAC"}
|
|
@@ -1,5 +1,8 @@
|
|
|
1
1
|
import type { Detector } from "../plugin.ts";
|
|
2
|
-
/**
|
|
2
|
+
/**
|
|
3
|
+
* Brackets inside brackets: 「(注(a)参照)」. The depth is read from the text, so it works in every language. A quotation of a
|
|
4
|
+
* string to type as it stands (件名に「応募(総務課)」と記載) counts as one bracket: its own brackets cannot be taken out.
|
|
5
|
+
*/
|
|
3
6
|
export declare const bracketNesting: Detector;
|
|
4
7
|
/** Clauses chained in one sentence (調べて、まとめ、報告したが、返事がないので、…). Needs the language's tokens. */
|
|
5
8
|
export declare const clauseChain: Detector;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"sentence-load.d.ts","sourceRoot":"","sources":["../../src/detectors/sentence-load.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAA4B,MAAM,cAAc,CAAC;
|
|
1
|
+
{"version":3,"file":"sentence-load.d.ts","sourceRoot":"","sources":["../../src/detectors/sentence-load.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAA4B,MAAM,cAAc,CAAC;AAsCvE;;;GAGG;AACH,eAAO,MAAM,cAAc,EAAE,QAG5B,CAAC;AAEF,8FAA8F;AAC9F,eAAO,MAAM,WAAW,EAAE,QAAkH,CAAC;AAE7I,mHAAmH;AACnH,eAAO,MAAM,sBAAsB,EAAE,QAAoH,CAAC"}
|
|
@@ -1,5 +1,6 @@
|
|
|
1
1
|
import { proseText } from "../measure.js";
|
|
2
2
|
import { quotedSpans } from "../quoted-span.js";
|
|
3
|
+
import { literalCues, withoutLiteralQuotes } from "../literal-quote.js";
|
|
3
4
|
import { bracketDepth, clauseCount, topicDistance } from "../sentence-load.js";
|
|
4
5
|
const findingOf = (sentence, count, limit) => ({
|
|
5
6
|
rule: "",
|
|
@@ -26,8 +27,14 @@ const ownTokens = (sentence) => {
|
|
|
26
27
|
const inQuote = (token) => mask[token.span.start - sentence.span.start] === 1 && mask[token.span.end - 1 - sentence.span.start] === 1;
|
|
27
28
|
return (sentence.tokens ?? []).filter((token) => !inQuote(token));
|
|
28
29
|
};
|
|
29
|
-
/**
|
|
30
|
-
|
|
30
|
+
/**
|
|
31
|
+
* Brackets inside brackets: 「(注(a)参照)」. The depth is read from the text, so it works in every language. A quotation of a
|
|
32
|
+
* string to type as it stands (件名に「応募(総務課)」と記載) counts as one bracket: its own brackets cannot be taken out.
|
|
33
|
+
*/
|
|
34
|
+
export const bracketNesting = (doc, options) => {
|
|
35
|
+
const cues = literalCues(doc.lexicons["literal-quote-cue"] ?? []);
|
|
36
|
+
return overLimit((sentence) => bracketDepth(withoutLiteralQuotes(sentence.text, cues)))(doc, options);
|
|
37
|
+
};
|
|
31
38
|
/** Clauses chained in one sentence (調べて、まとめ、報告したが、返事がないので、…). Needs the language's tokens. */
|
|
32
39
|
export const clauseChain = overLimit((sentence) => (sentence.tokens === undefined ? undefined : clauseCount(ownTokens(sentence))));
|
|
33
40
|
/** The distance from the sentence's topic (〜は) to the end, where its predicate is. Needs the language's tokens. */
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"sentence-load.js","sourceRoot":"","sources":["../../src/detectors/sentence-load.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,SAAS,EAAE,MAAM,eAAe,CAAC;AAC1C,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,YAAY,EAAE,WAAW,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AAE/E,MAAM,SAAS,GAAG,CAAC,QAAkB,EAAE,KAAa,EAAE,KAAa,EAAW,EAAE,CAAC,CAAC;IAChF,IAAI,EAAE,EAAE;IACR,QAAQ,EAAE,MAAM;IAChB,IAAI,EAAE,CAAC;IACP,MAAM,EAAE,CAAC;IACT,KAAK,EAAE,SAAS,CAAC,QAAQ,CAAC;IAC1B,MAAM,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE;CACtD,CAAC,CAAC;AAEH,mHAAmH;AACnH,MAAM,SAAS,GACb,CAAC,OAAmD,EAAY,EAAE,CAClE,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE,CAC1B,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,QAAQ,EAAE,EAAE;IACjC,MAAM,KAAK,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAC;IAChC,OAAO,KAAK,KAAK,SAAS,IAAI,KAAK,IAAI,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,QAAQ,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC;AAC1G,CAAC,CAAC,CAAC;AAEP,oIAAoI;AACpI,MAAM,UAAU,GAAG,CAAC,IAAY,EAAc,EAAE;IAC9C,MAAM,IAAI,GAAG,IAAI,UAAU,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IACzC,WAAW,CAAC,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;IACxE,OAAO,IAAI,CAAC;AACd,CAAC,CAAC;AAEF,iIAAiI;AACjI,MAAM,SAAS,GAAG,CAAC,QAAkB,EAAW,EAAE;IAChD,MAAM,IAAI,GAAG,UAAU,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;IACvC,MAAM,OAAO,GAAG,CAAC,KAAY,EAAW,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,GAAG,CAAC,GAAG,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;IACtJ,OAAO,CAAC,QAAQ,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC;AACpE,CAAC,CAAC;AAEF
|
|
1
|
+
{"version":3,"file":"sentence-load.js","sourceRoot":"","sources":["../../src/detectors/sentence-load.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,SAAS,EAAE,MAAM,eAAe,CAAC;AAC1C,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,WAAW,EAAE,oBAAoB,EAAE,MAAM,qBAAqB,CAAC;AACxE,OAAO,EAAE,YAAY,EAAE,WAAW,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AAE/E,MAAM,SAAS,GAAG,CAAC,QAAkB,EAAE,KAAa,EAAE,KAAa,EAAW,EAAE,CAAC,CAAC;IAChF,IAAI,EAAE,EAAE;IACR,QAAQ,EAAE,MAAM;IAChB,IAAI,EAAE,CAAC;IACP,MAAM,EAAE,CAAC;IACT,KAAK,EAAE,SAAS,CAAC,QAAQ,CAAC;IAC1B,MAAM,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE;CACtD,CAAC,CAAC;AAEH,mHAAmH;AACnH,MAAM,SAAS,GACb,CAAC,OAAmD,EAAY,EAAE,CAClE,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE,CAC1B,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,QAAQ,EAAE,EAAE;IACjC,MAAM,KAAK,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAC;IAChC,OAAO,KAAK,KAAK,SAAS,IAAI,KAAK,IAAI,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,QAAQ,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC;AAC1G,CAAC,CAAC,CAAC;AAEP,oIAAoI;AACpI,MAAM,UAAU,GAAG,CAAC,IAAY,EAAc,EAAE;IAC9C,MAAM,IAAI,GAAG,IAAI,UAAU,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IACzC,WAAW,CAAC,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;IACxE,OAAO,IAAI,CAAC;AACd,CAAC,CAAC;AAEF,iIAAiI;AACjI,MAAM,SAAS,GAAG,CAAC,QAAkB,EAAW,EAAE;IAChD,MAAM,IAAI,GAAG,UAAU,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;IACvC,MAAM,OAAO,GAAG,CAAC,KAAY,EAAW,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,GAAG,CAAC,GAAG,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;IACtJ,OAAO,CAAC,QAAQ,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC;AACpE,CAAC,CAAC;AAEF;;;GAGG;AACH,MAAM,CAAC,MAAM,cAAc,GAAa,CAAC,GAAG,EAAE,OAAO,EAAE,EAAE;IACvD,MAAM,IAAI,GAAG,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,mBAAmB,CAAC,IAAI,EAAE,CAAC,CAAC;IAClE,OAAO,SAAS,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,YAAY,CAAC,oBAAoB,CAAC,QAAQ,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;AACxG,CAAC,CAAC;AAEF,8FAA8F;AAC9F,MAAM,CAAC,MAAM,WAAW,GAAa,SAAS,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,QAAQ,CAAC,MAAM,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,WAAW,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;AAE7I,mHAAmH;AACnH,MAAM,CAAC,MAAM,sBAAsB,GAAa,SAAS,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,QAAQ,CAAC,MAAM,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,aAAa,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"signals.d.ts","sourceRoot":"","sources":["../../src/detectors/signals.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"signals.d.ts","sourceRoot":"","sources":["../../src/detectors/signals.ts"],"names":[],"mappings":"AAqBA,OAAO,KAAK,EAAE,QAAQ,EAA6C,MAAM,cAAc,CAAC;AAIxF;;;GAGG;AACH,eAAO,MAAM,mBAAmB;;;CAA2B,CAAC;AAoB5D,eAAO,MAAM,YAAY,EAAE,QAe1B,CAAC;AA8FF,eAAO,MAAM,eAAe,EAAE,QAiB7B,CAAC;AAwIF;;;GAGG;AACH,eAAO,MAAM,gBAAgB,EAAE,QAsC9B,CAAC;AAqCF,eAAO,MAAM,gBAAgB,EAAE,QA2B9B,CAAC;AAUF,eAAO,MAAM,WAAW,EAAE,QAezB,CAAC"}
|