chaffjs 0.2.0 → 0.4.0

package/src/run-semantic.ts

@@ -1,6 +1,7 @@
 import { severityOf, type UserCheck } from "./checks.ts";
 import { ask, type JudgeOptions } from "./judge.ts";
 import { FILTERS, type Candidate } from "./semantic.ts";
+import { narrow, type Narrowing } from "./look-at.ts";
 import { localized } from "./render/text.ts";
 import type { Finding, Level, ProseDocument, RuleDefinition, Severity } from "./plugin.ts";
 import { lineStarts, placeOf } from "./position.ts";
@@ -10,15 +11,18 @@ export type SemanticResult = {
   readonly skipped: readonly { rule: string; why: string }[];
   readonly asked: number;
   readonly sentencesSeen: number;
+  /** 利用者の検査が、どこまで絞り込めたか。絞り込めていないものを黙って通さない。 */
+  readonly narrowed: readonly { readonly name: string; readonly narrowing: Narrowing }[];
 };
 
-type Job = {
+export type Job = {
   readonly rule: string;
   readonly name: string;
   readonly rubric: string;
   readonly howToFix: string;
   readonly severity: Severity;
   readonly candidates: readonly Candidate[];
+  readonly narrowing?: Narrowing;
 };
 
 const filterFor = (name: string | undefined): (typeof FILTERS)[string] => FILTERS[name ?? "whole-document"] ?? FILTERS["whole-document"] ?? (() => []);
@@ -52,15 +56,19 @@ const builtInJobs = (doc: ProseDocument, rules: readonly RuleDefinition[], setti
 const userJobs = (doc: ProseDocument, checks: readonly UserCheck[], genre: string): Job[] =>
   checks
     .filter((check) => check.level !== "off" && check.use_for.some((target) => genre.startsWith(target)))
-    .map((check) => ({
-      rule: check.id,
-      name: check.name,
-      // look_at は自然文なので、まだ絞り込みに変換できない。文書全体を渡す。spec §26-7。
-      rubric: check.look_at === undefined ? check.check : `${check.check}\n\n見るところ: ${check.look_at}`,
-      howToFix: check.how_to_fix,
-      severity: severityOf(check.level),
-      candidates: filterFor("whole-document")(doc),
-    }));
+    .map((check) => {
+      // look_at の括弧の中を取り出して絞り込む。取り出せなければ文書全体。spec §14。
+      const { candidates, narrowing } = narrow(doc, check.look_at, filterFor("whole-document"));
+      return {
+        rule: check.id,
+        name: check.name,
+        rubric: check.look_at === undefined ? check.check : `${check.check}\n\n見るところ: ${check.look_at}`,
+        howToFix: check.how_to_fix,
+        severity: severityOf(check.level),
+        candidates,
+        narrowing,
+      };
+    });
 
 const findingOf = (job: Job, candidate: Candidate, reason: string, confidence: number, threshold: number, starts: readonly number[]): Finding => {
   const at = placeOf(starts, candidate.offset);
@@ -78,6 +86,18 @@ const findingOf = (job: Job, candidate: Candidate, reason: string, confidence: n
 
 export type SemanticOptions = JudgeOptions & { readonly confidenceThreshold: number };
 
+/**
+ * 何を問い合わせるかを決めるところまで。API を呼ばない。
+ * 鍵が無くても「何が送られるのか」を確かめられるようにするため、run から切り出してある。
+ */
+export const planSemantic = (
+  doc: ProseDocument,
+  rules: readonly RuleDefinition[],
+  checks: readonly UserCheck[],
+  settings: Readonly<Record<string, Level>>,
+  genre: string,
+): Job[] => [...builtInJobs(doc, rules, settings, genre), ...userJobs(doc, checks, genre)];
+
 export const runSemantic = async (
   doc: ProseDocument,
   rules: readonly RuleDefinition[],
@@ -86,7 +106,7 @@ export const runSemantic = async (
   genre: string,
   options: SemanticOptions,
 ): Promise<SemanticResult> => {
-  const jobs = [...builtInJobs(doc, rules, settings, genre), ...userJobs(doc, checks, genre)];
+  const jobs = planSemantic(doc, rules, checks, settings, genre);
   const starts = lineStarts(doc.source);
   const skipped = jobs.filter((job) => job.candidates.length === 0).map((job) => ({ rule: job.rule, why: "見るところが無かったため" }));
   const live = jobs.filter((job) => job.candidates.length > 0);
@@ -103,5 +123,6 @@ export const runSemantic = async (
     skipped,
     asked: live.reduce((sum, job) => sum + job.candidates.length, 0),
     sentencesSeen: doc.sentences.length,
+    narrowed: jobs.flatMap((job) => (job.narrowing === undefined ? [] : [{ name: job.name, narrowing: job.narrowing }])),
   };
 };

package/src/run.ts

@@ -1,6 +1,6 @@
 import { DETECTORS } from "./detectors/index.ts";
 import { resolve } from "./levels.ts";
-import type { Finding, Level, ProseDocument, RuleDefinition } from "./plugin.ts";
+import type { AdapterNeeds, Finding, Level, ProseDocument, RuleDefinition } from "./plugin.ts";
 import { lineStarts, placeOf } from "./position.ts";
 
 export type Skipped = { readonly rule: string; readonly why: string };
@@ -22,6 +22,37 @@ const levelFor = (rule: RuleDefinition, settings: Settings, experimental: boolea
   return "normal";
 };
 
+const CAPABILITY_NAME: Readonly<Record<string, string>> = { pos: "品詞解析", lemma: "原形" };
+
+/** 知らない要求は満たされていないものとして扱う。黙って無視すると、要求なしで動いてしまう。 */
+const has = (capabilities: ProseDocument["capabilities"], need: string): boolean => {
+  if (need === "pos") return capabilities.pos;
+  if (need === "lemma") return capabilities.lemma;
+  return false;
+};
+
+/** 要求を満たさない rule は動かせない。満たさないまま動かすと「指摘 0 件」が保証に見える。 */
+const unmet = (rule: RuleDefinition, doc: ProseDocument): string | undefined => {
+  if (rule.languages !== undefined && !rule.languages.includes(doc.language)) return `${doc.language} 向けの rule ではないため`;
+  const missing = rule.requires.find((need) => !has(doc.capabilities, need));
+  if (missing === undefined) return undefined;
+  return `この言語では${CAPABILITY_NAME[missing] ?? missing}が使えないため`;
+};
+
+const forGenre = (rules: readonly RuleDefinition[], genre: string): RuleDefinition[] =>
+  rules.filter((rule) => rule.use_for.some((target) => genre.startsWith(target)));
+
+/**
+ * 解析器の初期化に払う代金を決める。動く rule が 1 本も要求しないなら読み込まない。
+ * capabilities は「払えばできる」の宣言なので、ここでは見ない。
+ */
+export const neededBy = (rules: readonly RuleDefinition[], settings: Settings, experimental: boolean, genre: string, language: string): AdapterNeeds => ({
+  pos: forGenre(rules, genre)
+    .filter((rule) => rule.layer !== "L4" && levelFor(rule, settings, experimental) !== "off")
+    .filter((rule) => rule.languages === undefined || rule.languages.includes(language))
+    .some((rule) => rule.requires.includes("pos") || rule.requires.includes("lemma")),
+});
+
 const place = (starts: readonly number[], finding: Finding): Finding => {
   const offset = finding.values["offset"];
   const at = placeOf(starts, typeof offset === "number" ? offset : 0);
@@ -30,7 +61,7 @@ const place = (starts: readonly number[], finding: Finding): Finding => {
 
 export const runRules = (doc: ProseDocument, rules: readonly RuleDefinition[], settings: Settings, experimental: boolean, genre: string): RunResult => {
   const starts = lineStarts(doc.source);
-  const applicable = rules.filter((rule) => rule.use_for.some((target) => genre.startsWith(target)));
+  const applicable = forGenre(rules, genre);
   const forced = applicable
     .filter((rule) => rule.status === "experimental" && settings[rule.id] !== undefined && settings[rule.id] !== "off")
     .map((rule) => rule.id);
@@ -40,6 +71,8 @@ export const runRules = (doc: ProseDocument, rules: readonly RuleDefinition[], s
       if (rule.layer === "L4") {
         return { findings: acc.findings, skipped: [...acc.skipped, { rule: rule.id, why: "意味を読む検査のため（npx chaff test で動きます）" }] };
       }
+      const blocked = unmet(rule, doc);
+      if (blocked !== undefined) return { findings: acc.findings, skipped: [...acc.skipped, { rule: rule.id, why: blocked }] };
       const level = levelFor(rule, settings, experimental);
       if (level === "off") {
         const why = rule.status === "experimental" && settings[rule.id] === undefined ? "まだ試験中のため" : "設定で止めているため";
@@ -48,7 +81,7 @@ export const runRules = (doc: ProseDocument, rules: readonly RuleDefinition[], s
       const detector = DETECTORS[rule.how_to_find];
       if (detector === undefined) return { findings: acc.findings, skipped: [...acc.skipped, { rule: rule.id, why: `検出器 ${rule.how_to_find} がないため` }] };
       const options = {
-        limit: resolve(rule, level).limit,
+        limit: resolve(rule, level, genre).limit,
         lexicon: rule.word_list === undefined ? undefined : doc.lexicons[rule.word_list],
         where: rule.where,
       };

package/src/sentence-shape.ts

@@ -0,0 +1,22 @@
+import type { Sentence, Token } from "./plugin.ts";
+
+/**
+ * 「これは文か」の判定を 1 箇所に置く。
+ *
+ * 段落として解析されるものには、見出しの下の名前だけの行（`OpenAI`）、引用の出典行、
+ * 表題行が混ざる。文として数えると、文書全体を分母にする rule がそこで狂う。
+ * 終止符で終わることを条件にすると、そのほとんどが落ちる。
+ */
+const ENDS = /[。．.！？!?][")）」』\s]*$/u;
+
+export const isClosed = (sentence: Sentence): boolean => ENDS.test(sentence.text);
+
+const PREDICATE = new Set(["AUX", "VERB", "ADJ"]);
+
+export const hasPredicate = (sentence: Sentence): boolean => (sentence.tokens ?? []).some((token) => PREDICATE.has(token.pos));
+
+export const hasParticle = (sentence: Sentence): boolean => (sentence.tokens ?? []).some((token) => token.pos === "ADP" || token.pos === "SCONJ");
+
+const SKIP = new Set(["PUNCT", "PART", "SYM"]);
+
+export const lastContent = (sentence: Sentence): Token | undefined => [...(sentence.tokens ?? [])].reverse().find((token) => !SKIP.has(token.pos));