hanlint 0.0.7 → 0.0.8

package/src/rules/shared/repeat.js

@@ -0,0 +1,48 @@
+// @ts-check
+/**
+ * 반복 기제. 같은 모양이 창 안에서 이어지거나 (runsOf) 한 모양이 창을 채우는 (shareOf) 자리를 센다.
+ * 파이썬 rules/shared/repeat.py 와 같다. 규칙은 열쇠와 창과 임계만 선언하고 셈은 여기서 한다.
+ */
+
+/**
+ * 같은 값이 minLength 개 이상 이어지는 구간을 [시작 위치, 길이, 값] 으로 준다.
+ * @template T
+ * @param {T[]} values
+ * @param {number} minLength
+ * @returns {[number, number, T][]}
+ */
+export function runsOf(values, minLength) {
+  /** @type {[number, number, T][]} */
+  const runs = [];
+  let start = 0;
+  while (start < values.length) {
+    let end = start;
+    while (end + 1 < values.length && values[end + 1] === values[start]) end += 1;
+    const length = end - start + 1;
+    if (length >= minLength) runs.push([start, length, values[start]]);
+    start = end + 1;
+  }
+  return runs;
+}
+
+/**
+ * 가장 많은 값과 그 수와 전체 수. 같은 수면 먼저 나온 값이 이긴다. 빈 열이면 [null, 0, 0].
+ * @template T
+ * @param {T[]} values
+ * @returns {[T | null, number, number]}
+ */
+export function shareOf(values) {
+  /** @type {Map<T, number>} */
+  const counts = new Map();
+  for (const value of values) counts.set(value, (counts.get(value) ?? 0) + 1);
+  /** @type {T | null} */
+  let best = null;
+  let count = 0;
+  for (const [value, seen] of counts) {
+    if (seen > count) {
+      best = value;
+      count = seen;
+    }
+  }
+  return [best, count, values.length];
+}

package/src/rules/structure/blockUnread.js

@@ -0,0 +1,33 @@
+// @ts-check
+import { PROSE } from "../../document/model.js";
+import { NOTICE, SECTION, finding } from "../finding.js";
+
+export const name = "blockUnread";
+export const mechanism = "threshold";
+const OUTPUT_LANGUAGES = ["", "text", "output", "console"];
+
+/**
+ * @param {import("../../fingerprint/build.js").DocumentPrint} doc
+ * @param {import("../../config/settings.js").Config} config
+ */
+export function run(doc, config) {
+  void config;
+  /** @type {Map<number, string>} */
+  const languages = new Map(doc.codeBlocks.map((block) => [block.index, block.language]));
+  const outputs = new Set([...languages].filter(([, language]) => OUTPUT_LANGUAGES.includes(language)).map(([index]) => index));
+  if (!outputs.size) return [];
+  const blocks = [...doc.blocks].sort((a, b) => a.index - b.index);
+  const findings = [];
+  for (let position = 0; position < blocks.length; position += 1) {
+    const block = blocks[position];
+    if (!outputs.has(block.index) || position === 0) continue;
+    const previous = languages.get(blocks[position - 1].index);
+    if (previous === undefined || OUTPUT_LANGUAGES.includes(previous)) continue;
+    const following = position + 1 < blocks.length ? blocks[position + 1] : null;
+    if (following !== null && following.kind === PROSE) continue;
+    findings.push(
+      finding(name, block.startLine, block.text.includes("\n") ? block.text.split("\n")[1] : block.text, "출력을 붙여 놓고 읽어 주지 않았다. 어느 부분이 무엇인지 짚는 문장을 바로 아래에 붙인다", null, NOTICE, SECTION, block.index),
+    );
+  }
+  return findings;
+}

package/src/rules/structure/bridgeRepeat.js

@@ -0,0 +1,40 @@
+// @ts-check
+import { matchedTexts } from "../../fingerprint/markers.js";
+import { DOCUMENT, finding } from "../finding.js";
+
+export const name = "bridgeRepeat";
+export const mechanism = "repeat";
+
+/**
+ * 절을 닫는 문장이 이음 표지 (이번에는, 이제, 다음으로) 로 시작하는 절이 bridgeRepeatMin 개 이상인 글.
+ * 파이썬 rules/structure/bridgeRepeat.py 와 같다. 표지 목록은 data/bridgeOpeners.txt.
+ * @param {import("../../fingerprint/build.js").DocumentPrint} doc
+ * @param {import("../../config/settings.js").Config} config
+ */
+export function run(doc, config) {
+  /** @type {[import("../../fingerprint/build.js").SentencePrint, string][]} */
+  const closers = [];
+  for (const section of doc.bodySections) {
+    if (!section.paragraphs.length) continue;
+    const sentences = section.paragraphs[section.paragraphs.length - 1].sentences;
+    if (!sentences.length) continue;
+    const last = sentences[sentences.length - 1];
+    const found = matchedTexts(last.text, "bridgeOpeners.txt");
+    if (found.length) closers.push([last, found[0]]);
+  }
+  if (closers.length < config.bridgeRepeatMin) return [];
+  const markers = [...new Set(closers.map(([, marker]) => marker))].sort();
+  const [first] = closers[0];
+  return [
+    finding(
+      name,
+      first.line,
+      first.text,
+      `절 ${closers.length}개가 \`${markers.join(", ")}\` 같은 이음 표지로 시작하는 문장으로 닫힌다. 예고가 절마다 되풀이되면 틀이다. 방금 만든 결과를 이름으로 부르고 아직 못 하는 일을 적는다`,
+      null,
+      "error",
+      DOCUMENT,
+      -1,
+    ),
+  ];
+}

package/src/rules/structure/emojiBullet.js

@@ -3,6 +3,7 @@ import { LIST } from "../../document/model.js";
 import { DOCUMENT, finding } from "../finding.js";
 
 export const name = "emojiBullet";
+export const mechanism = "dictionary";
 // 이모지 범위. 코드포인트로 만든다 (도구가 이스케이프를 건드리지 않게). 아스트랄 범위라 u 플래그가 필요하다.
 const EMOJI =
   "[" +

package/src/rules/structure/headingQuestion.js

@@ -0,0 +1,19 @@
+// @ts-check
+import { DOCUMENT, finding } from "../finding.js";
+
+export const name = "headingQuestion";
+export const mechanism = "repeat";
+
+/**
+ * @param {import("../../fingerprint/build.js").DocumentPrint} doc
+ * @param {import("../../config/settings.js").Config} config
+ */
+export function run(doc, config) {
+  const headings = doc.headings.filter(([level]) => level === 2);
+  if (headings.length < 3) return [];
+  const asking = headings.filter(([, text]) => text.trimEnd().endsWith("?")).map(([, text]) => text);
+  if (asking.length / headings.length <= config.headingQuestionRatio) return [];
+  return [
+    finding(name, headings[0][2], asking.join(" / "), `H2 ${headings.length}개 중 ${asking.length}개가 물음표로 끝난다. 목차가 과정이 아니라 FAQ 로 읽힌다. 형태를 섞는다`, null, "error", DOCUMENT, -1),
+  ];
+}

package/src/rules/structure/headingSentence.js

@@ -2,12 +2,17 @@
 import { DOCUMENT, finding } from "../finding.js";
 
 export const name = "headingSentence";
-const SENTENCE_ENDINGS = ["니다", "세요", "십시오", "합시다", "."];
+export const mechanism = "threshold";
+const SENTENCE_ENDINGS = ["니다", "한다", "해요", "세요", "십시오", "합시다", "."];
 
-/** @param {import("../../fingerprint/build.js").DocumentPrint} doc */
-export function run(doc) {
+/**
+ * @param {import("../../fingerprint/build.js").DocumentPrint} doc
+ * @param {import("../../config/settings.js").Config} config
+ */
+export function run(doc, config) {
   const findings = [];
-  for (const [, text, line] of doc.headings) {
+  for (const [level, text, line] of doc.headings) {
+    if (level > config.headingSentenceMaxLevel) continue;
     const trimmed = text.trimEnd();
     if (SENTENCE_ENDINGS.some((ending) => trimmed.endsWith(ending))) {
       findings.push(finding(name, line, text, "절 제목이 본문 문장을 복사했다. 대상이나 질문을 짧게 쓴다", null, "error", DOCUMENT, -1));

package/src/rules/structure/headingSkip.js

@@ -2,6 +2,7 @@
 import { DOCUMENT, finding } from "../finding.js";
 
 export const name = "headingSkip";
+export const mechanism = "threshold";
 
 /** @param {import("../../fingerprint/build.js").DocumentPrint} doc */
 export function run(doc) {

package/src/rules/structure/headingUniform.js

@@ -1,7 +1,9 @@
 // @ts-check
 import { DOCUMENT, finding } from "../finding.js";
+import { shareOf } from "../shared/repeat.js";
 
 export const name = "headingUniform";
+export const mechanism = "repeat";
 
 /**
  * @param {import("../../fingerprint/build.js").DocumentPrint} doc
@@ -17,23 +19,13 @@ export function run(doc, config) {
     return !(char >= "0" && char <= "9");
   });
   if (eligible.length < 3) return [];
-  // 파이썬 Counter.most_common 은 같은 수면 먼저 들어온 것이 이긴다. 삽입 순서를 지킨다.
-  /** @type {Map<string, number>} */
-  const lastChars = new Map();
-  for (const [, text] of eligible) {
-    const trimmed = text.trimEnd();
-    const char = trimmed[trimmed.length - 1];
-    lastChars.set(char, (lastChars.get(char) ?? 0) + 1);
-  }
-  let best = "";
-  let count = -1;
-  for (const [char, n] of lastChars) {
-    if (n > count) {
-      best = char;
-      count = n;
-    }
-  }
-  if (count / eligible.length >= config.headingUniformRatio) {
+  const [best, count, total] = shareOf(
+    eligible.map(([, text]) => {
+      const trimmed = text.trimEnd();
+      return trimmed[trimmed.length - 1];
+    }),
+  );
+  if (count / total >= config.headingUniformRatio) {
     return [
       finding(name, eligible[0][2], eligible.map(([, text]) => text).join(" / "), `H2 ${eligible.length}개 중 ${count}개가 \`${best}\` 로 끝난다. 목차가 한 어미로 끝나면 과정이 아니라 나열로 읽힌다. 형태를 섞는다`, null, "error", DOCUMENT, -1),
     ];

package/src/rules/structure/introImage.js

@@ -0,0 +1,19 @@
+// @ts-check
+import { IMAGE } from "../../document/model.js";
+import { SECTION, finding } from "../finding.js";
+
+export const name = "introImage";
+export const mechanism = "threshold";
+
+/**
+ * @param {import("../../fingerprint/build.js").DocumentPrint} doc
+ * @param {import("../../config/settings.js").Config} config
+ */
+export function run(doc, config) {
+  if (!doc.bodySections.length) return [];
+  const count = doc.intro.blockKinds.filter((kind) => kind === IMAGE).length;
+  if (count <= config.introMaxImages) return [];
+  return [
+    finding(name, doc.intro.startLine, doc.intro.title || "도입", `도입에 그림이 ${count}장이다. ${config.introMaxImages}장을 넘지 않는다. 나머지는 그것을 설명하는 절로 내린다`, null, "error", SECTION, doc.intro.index),
+  ];
+}

package/src/rules/structure/introLong.js

@@ -2,6 +2,7 @@
 import { PARAGRAPH, finding } from "../finding.js";
 
 export const name = "introLong";
+export const mechanism = "threshold";
 
 /**
  * @param {import("../../fingerprint/build.js").DocumentPrint} doc

package/src/rules/structure/loneSubheading.js

@@ -0,0 +1,42 @@
+// @ts-check
+import { topicsOf } from "../../fingerprint/topics.js";
+import { NOTICE, SECTION, finding } from "../finding.js";
+
+export const name = "loneSubheading";
+export const mechanism = "contrast";
+
+/** (부모 H2 제목, 외동 H3 제목, 그 줄). H2 하나 아래 H3 이 정확히 하나일 때만. @param {[number, string, number][]} headings */
+function onlyChild(headings) {
+  /** @type {[string, string, number][]} */
+  const found = [];
+  for (let position = 0; position < headings.length; position += 1) {
+    const [level, text] = headings[position];
+    if (level !== 2) continue;
+    /** @type {[string, number][]} */
+    const children = [];
+    for (const [childLevel, childText, childLine] of headings.slice(position + 1)) {
+      if (childLevel <= 2) break;
+      if (childLevel === 3) children.push([childText, childLine]);
+    }
+    if (children.length === 1) found.push([text, children[0][0], children[0][1]]);
+  }
+  return found;
+}
+
+/**
+ * @param {import("../../fingerprint/build.js").DocumentPrint} doc
+ * @param {import("../../config/settings.js").Config} config
+ */
+export function run(doc, config) {
+  void config;
+  const findings = [];
+  for (const [parent, child, line] of onlyChild(doc.headings)) {
+    const childTopics = topicsOf(child);
+    const parentTopics = topicsOf(parent);
+    if (!childTopics.size || ![...childTopics].every((word) => parentTopics.has(word))) continue;
+    findings.push(
+      finding(name, line, child, `\`${parent}\` 아래 소제목이 이것 하나뿐이고 절 제목에 없는 말이 하나도 없다. 소제목을 지우고 본문을 올린다`, null, NOTICE, SECTION, -1),
+    );
+  }
+  return findings;
+}

package/src/rules/structure/moreLater.js

@@ -0,0 +1,48 @@
+// @ts-check
+import { LIST } from "../../document/model.js";
+import { plainText } from "../../document/plainText.js";
+import { NOTICE, SECTION, finding } from "../finding.js";
+
+export const name = "moreLater";
+export const mechanism = "threshold";
+const BULLET = /^\s*(?:[-*+]|\d+[.)])\s+/;
+
+/** (블록 안 줄 오프셋, 항목 전체). 이어지는 줄은 앞 항목에 붙인다. @param {string} text @returns {[number, string][]} */
+function itemsOf(text) {
+  /** @type {[number, string][]} */
+  const items = [];
+  text.split("\n").forEach((line, offset) => {
+    if (BULLET.test(line)) {
+      items.push([offset, line.replace(BULLET, "").trim()]);
+    } else if (items.length && line.trim()) {
+      const last = items[items.length - 1];
+      last[1] = `${last[1]} ${line.trim()}`;
+    }
+  });
+  return items;
+}
+
+/**
+ * @param {import("../../fingerprint/build.js").DocumentPrint} doc
+ * @param {import("../../config/settings.js").Config} config
+ */
+export function run(doc, config) {
+  if (doc.sections.length < 2) return [];
+  const last = doc.sections[doc.sections.length - 1];
+  if (last.isIntro) return [];
+  /** @type {[number, string][]} */
+  const over = [];
+  for (const block of doc.blocks) {
+    if (block.kind !== LIST || block.startLine < last.startLine) continue;
+    for (const [offset, item] of itemsOf(block.text)) {
+      const visible = plainText(item);
+      if (visible.length > config.moreLaterMaxChars) over.push([block.startLine + offset, visible]);
+    }
+  }
+  if (!over.length) return [];
+  let best = over[0];
+  for (const pair of over) if (pair[1].length > best[1].length) best = pair;
+  return [
+    finding(name, best[0], best[1], `마지막 절의 목록 항목 ${over.length}개가 ${config.moreLaterMaxChars}자를 넘는다 (가장 긴 것 ${best[1].length}자). 미룬 것이 아니라 옮겨 적은 것이다. 한 줄과 링크로 줄인다`, null, NOTICE, SECTION, last.index),
+  ];
+}

package/src/rules/structure/sectionNoProse.js

@@ -3,6 +3,7 @@ import { LIST, PROSE } from "../../document/model.js";
 import { SECTION, finding } from "../finding.js";
 
 export const name = "sectionNoProse";
+export const mechanism = "threshold";
 
 /** @param {import("../../fingerprint/build.js").DocumentPrint} doc */
 export function run(doc) {

package/src/rules/structure/sectionResult.js

@@ -1,9 +1,10 @@
 // @ts-check
-import { CODE, IMAGE, TABLE } from "../../document/model.js";
+import { CODE, EMBED, IMAGE, TABLE } from "../../document/model.js";
 import { NOTICE, SECTION, finding } from "../finding.js";
 
 export const name = "sectionResult";
-const RESULT_KINDS = [CODE, IMAGE, TABLE];
+export const mechanism = "threshold";
+const RESULT_KINDS = [CODE, EMBED, IMAGE, TABLE];
 const FILE_NAME = /[\w가-힣-]+\.(?:csv|xlsx|xls|parquet|json|png|jpg|svg|db|sqlite|txt|py|pdf|html|md)\b/;
 const OUTPUT_WORDS = ["출력", "화면에", "찍힙니다", "나옵니다", "뜹니다", "보입니다", "만들어집니다", "생깁니다", "저장됩니다"];
 

package/src/text.js

@@ -61,3 +61,22 @@ export function countIn(text, needle, end = text.length) {
     from = at + needle.length;
   }
 }
+
+/**
+ * 파이썬 round(value, digits). 정확한 이진 값을 십진으로 풀어 반올림하고 정확히 절반이면 짝수로 간다.
+ * Math.round 는 절반을 위로 올려 13/16 을 0.813 으로 만들고 파이썬은 0.812 다. 두 판의 글자가 같으려면 파이썬 쪽이다.
+ * toFixed(100) 은 1e21 아래의 수를 정확한 십진 전개로 준다. 지문의 비율과 평균이 사는 범위다.
+ * @param {number} value
+ * @param {number} digits
+ */
+export function roundHalfEven(value, digits) {
+  if (!Number.isFinite(value) || Math.abs(value) >= 1e21) return value;
+  const sign = value < 0 ? -1 : 1;
+  const [whole, fraction] = Math.abs(value).toFixed(100).split(".");
+  const rest = fraction.slice(digits);
+  let units = BigInt(whole + fraction.slice(0, digits));
+  const next = rest.charCodeAt(0) - 48;
+  const halfway = next === 5 && /^0*$/.test(rest.slice(1));
+  if (next > 5 || (next === 5 && !halfway) || (halfway && units % 2n === 1n)) units += 1n;
+  return sign * (Number(units) / 10 ** digits);
+}

package/src/rules/document/readerAbsent.js

@@ -1,11 +0,0 @@
-// @ts-check
-import { DOCUMENT, NOTICE, finding } from "../finding.js";
-
-export const name = "readerAbsent";
-
-/** @param {import("../../fingerprint/build.js").DocumentPrint} doc */
-export function run(doc) {
-  if (doc.bodySections.length < 2 || doc.questionCount > 0 || doc.readerCallCount > 0) return [];
-  const first = doc.sentences.length ? doc.sentences[0] : null;
-  return [finding(name, first ? first.line : 1, first ? first.text : "", "질문도 독자를 부르는 문장도 없다. 독자가 할 행동을 동사로 끝내는 문장을 한 번은 넣는다", null, NOTICE, DOCUMENT, -1)];
-}

package/src/rules/shared/runs.js

@@ -1,22 +0,0 @@
-// @ts-check
-/** 연속 구간 찾기. 같은 값이 이어지는 자리를 [시작 위치, 길이, 값] 으로 준다. */
-
-/**
- * @template T
- * @param {T[]} values
- * @param {number} minLength
- * @returns {[number, number, T][]}
- */
-export function runsOf(values, minLength) {
-  /** @type {[number, number, T][]} */
-  const runs = [];
-  let start = 0;
-  while (start < values.length) {
-    let end = start;
-    while (end + 1 < values.length && values[end + 1] === values[start]) end += 1;
-    const length = end - start + 1;
-    if (length >= minLength) runs.push([start, length, values[start]]);
-    start = end + 1;
-  }
-  return runs;
-}