@wdprlib/ast 4.3.0 → 5.0.0

package/dist/index.cjs

@@ -47,11 +47,16 @@ __export(exports_src, {
   horizontalRule: () => horizontalRule,
   heading: () => heading,
   formatExprValue: () => formatExprValue,
+  extractReadableText: () => extractReadableText,
+  extractFirstParagraph: () => extractFirstParagraph,
+  excerptText: () => excerptText,
   evaluateExpression: () => evaluateExpression,
   createSettings: () => createSettings,
   createPosition: () => createPosition,
   createPoint: () => createPoint,
+  countCharacters: () => countCharacters,
   container: () => container,
+  compileTextExcerpt: () => compileTextExcerpt,
   buildInfo: () => buildInfo,
   bold: () => bold,
   STYLE_SLOT_PREFIX: () => STYLE_SLOT_PREFIX,
@@ -63,320 +68,43 @@ module.exports = __toCommonJS(exports_src);
 
 // packages/ast/src/build-info.generated.ts
 var buildInfo = Object.freeze({
-  version: "4.3.0",
-  sha: "5e221ddeccc585e64dd1a2eb49b14817c3c1be5d",
+  version: "5.0.0",
+  sha: "705ff97e22abd7bfb49d7f8f30087e6f09654c91",
   dirty: false
 });
-// packages/ast/src/position.ts
-function createPoint(line, column, offset) {
-  return { line, column, offset };
-}
-function createPosition(start, end) {
-  return { start, end };
-}
-// packages/ast/src/element.ts
-function isStringContainerType(type) {
-  return typeof type === "string";
-}
-function isHeaderType(type) {
-  return typeof type === "object" && type !== null && "header" in type;
-}
-function isAlignType(type) {
-  return typeof type === "object" && type !== null && "align" in type;
-}
-function text(value) {
-  return { element: "text", data: value };
-}
-function container(type, elements, attributes = {}) {
-  return {
-    element: "container",
-    data: { type, attributes, elements }
-  };
-}
-function paragraph(elements, attributes = {}) {
-  return container("paragraph", elements, attributes);
-}
-function bold(elements, attributes = {}) {
-  return container("bold", elements, attributes);
-}
-function italics(elements, attributes = {}) {
-  return container("italics", elements, attributes);
-}
-function heading(level, elements, hasToc = true, attributes = {}) {
-  return container({ header: { level, "has-toc": hasToc } }, elements, attributes);
+// packages/ast/src/expr-eval.ts
+var FALSE_VALUES = new Set(["false", "null", "", "0"]);
+function isTruthy(value) {
+  return !FALSE_VALUES.has(value.toLowerCase().trim());
 }
-function lineBreak() {
-  return { element: "line-break" };
+var MAX_EXPRESSION_LENGTH = 256;
+function isTruthyNum(n) {
+  return n !== 0 && !Number.isNaN(n);
 }
-function horizontalRule() {
-  return { element: "horizontal-rule" };
+function formatExprValue(n) {
+  return String(n);
 }
-function link(linkLocation, label, options = {}) {
-  return {
-    element: "link",
-    data: {
-      type: options.type ?? (typeof linkLocation === "string" ? "direct" : "page"),
-      link: linkLocation,
-      extra: options.extra ?? null,
-      label,
-      target: options.target ?? null
+function evaluateExpression(expr) {
+  try {
+    if (expr.length > MAX_EXPRESSION_LENGTH) {
+      return { success: false, error: "expression too long" };
     }
-  };
-}
-function list(type, items, attributes = {}) {
-  return {
-    element: "list",
-    data: { type, attributes, items }
-  };
-}
-function listItemElements(elements, attributes = {}) {
-  return {
-    "item-type": "elements",
-    attributes,
-    elements
-  };
-}
-function listItemSubList(data) {
-  return {
-    "item-type": "sub-list",
-    element: "list",
-    data
-  };
-}
-function isContainerTypeParagraphSafe(type) {
-  if (isHeaderType(type))
-    return false;
-  if (isAlignType(type))
-    return false;
-  switch (type) {
-    case "bold":
-    case "italics":
-    case "underline":
-    case "superscript":
-    case "subscript":
-    case "strikethrough":
-    case "monospace":
-    case "span":
-    case "size":
-      return true;
-    case "div":
-    case "note":
-    case "blockquote":
-    case "paragraph":
-    case "heading":
-    case "collapsible":
-    case "definition-list":
-    case "definition-list-item":
-    case "definition-list-key":
-    case "definition-list-value":
-    case "table-row":
-    case "table-cell":
-      return false;
-    default:
-      return false;
-  }
-}
-function isParagraphSafe(element) {
-  switch (element.element) {
-    case "container": {
-      const data = element.data;
-      return isContainerTypeParagraphSafe(data.type);
+    if (expr.trim() === "") {
+      return { success: false, error: "empty expression" };
     }
-    case "module":
-      return false;
-    case "text":
-    case "raw":
-    case "variable":
-    case "email":
-      return true;
-    case "table":
-      return false;
-    case "tab-view":
-      return false;
-    case "anchor":
-    case "anchor-name":
-    case "link":
-      return true;
-    case "image":
-      return true;
-    case "list":
-      return false;
-    case "definition-list":
-      return false;
-    case "collapsible":
-      return false;
-    case "table-of-contents":
-      return false;
-    case "footnote":
-      return true;
-    case "footnote-ref":
-      return true;
-    case "footnote-block":
-      return false;
-    case "bibliography-cite":
-      return true;
-    case "bibliography-block":
-      return false;
-    case "user":
-      return true;
-    case "date":
-    case "social":
-      return true;
-    case "button":
-      return isPageButtonAction(element.data.action);
-    case "color":
-      return true;
-    case "code":
-      return false;
-    case "math":
-      return false;
-    case "math-inline":
-      return true;
-    case "embed":
-      return false;
-    case "embed-block":
-      return true;
-    case "html":
-    case "iframe":
-      return false;
-    case "include": {
-      const data = element.data;
-      return data["paragraph-safe"];
+    const tokens = tokenize(expr);
+    if (tokens.length <= 1) {
+      return { success: false, error: "empty expression" };
     }
-    case "style":
-      return false;
-    case "line-break":
-    case "line-breaks":
-      return true;
-    case "clear-float":
-      return false;
-    case "horizontal-rule":
-      return false;
-    case "content-separator":
-      return false;
-    case "if-tags":
-      return false;
-    case "expr":
-    case "if":
-    case "ifexpr":
-      return true;
-    default:
-      return false;
-  }
-}
-function isPageButtonAction(action) {
-  switch (action) {
-    case "edit":
-    case "edit-append":
-    case "edit-sections":
-    case "history":
-    case "print":
-    case "files":
-    case "tags":
-    case "source":
-    case "backlinks":
-    case "talk":
-    case "delete":
-    case "rename":
-    case "site-tools":
-    case "edit-meta":
-    case "watchers":
-    case "parent":
-    case "lock-page":
-      return true;
-    default:
-      return false;
-  }
-}
-// packages/ast/src/constants.ts
-var STYLE_SLOT_PREFIX = "\x00__IFTAGS_SLOT__";
-var STYLE_ANCHOR_PREFIX = "\x00__STYLE_ANCHOR__";
-// packages/ast/src/css.ts
-var CSS_LENGTH_UNITS = [
-  "cqmax",
-  "cqmin",
-  "cqw",
-  "cqh",
-  "cqi",
-  "cqb",
-  "svmin",
-  "svmax",
-  "lvmin",
-  "lvmax",
-  "dvmin",
-  "dvmax",
-  "vmin",
-  "vmax",
-  "svw",
-  "svh",
-  "svi",
-  "svb",
-  "lvw",
-  "lvh",
-  "lvi",
-  "lvb",
-  "dvw",
-  "dvh",
-  "dvi",
-  "dvb",
-  "vw",
-  "vh",
-  "vi",
-  "vb",
-  "rcap",
-  "rem",
-  "rex",
-  "rch",
-  "ric",
-  "rlh",
-  "cap",
-  "em",
-  "ex",
-  "ch",
-  "ic",
-  "lh",
-  "cm",
-  "mm",
-  "in",
-  "pc",
-  "pt",
-  "px",
-  "q",
-  "%"
-];
-// packages/ast/src/expr-eval.ts
-var FALSE_VALUES = new Set(["false", "null", "", "0"]);
-function isTruthy(value) {
-  return !FALSE_VALUES.has(value.toLowerCase().trim());
-}
-var MAX_EXPRESSION_LENGTH = 256;
-function isTruthyNum(n) {
-  return n !== 0 && !Number.isNaN(n);
-}
-function formatExprValue(n) {
-  return String(n);
-}
-function evaluateExpression(expr) {
-  try {
-    if (expr.length > MAX_EXPRESSION_LENGTH) {
-      return { success: false, error: "expression too long" };
-    }
-    if (expr.trim() === "") {
-      return { success: false, error: "empty expression" };
-    }
-    const tokens = tokenize(expr);
-    if (tokens.length <= 1) {
-      return { success: false, error: "empty expression" };
-    }
-    const parser = new ExprParser(tokens);
-    const result = parser.parse();
-    if (!Number.isFinite(result)) {
-      return { success: false, error: "division by zero" };
-    }
-    return { success: true, value: result };
-  } catch (e) {
-    const msg = e instanceof Error ? e.message : "unknown error";
-    return { success: false, error: msg };
+    const parser = new ExprParser(tokens);
+    const result = parser.parse();
+    if (!Number.isFinite(result)) {
+      return { success: false, error: "division by zero" };
+    }
+    return { success: true, value: result };
+  } catch (e) {
+    const msg = e instanceof Error ? e.message : "unknown error";
+    return { success: false, error: msg };
   }
 }
 function tokenize(expr) {
@@ -732,6 +460,574 @@ class ExprParser {
     }
   }
 }
+
+// packages/ast/src/readable-text.ts
+var segmenter = new Intl.Segmenter("und", { granularity: "grapheme" });
+function countCharacters(text) {
+  let count = 0;
+  for (const _segment of segmenter.segment(text))
+    count++;
+  return count;
+}
+function extractReadableText(ast, options = {}) {
+  return extractText(ast, options, false);
+}
+function extractFirstParagraph(ast, options = {}) {
+  return extractText(ast, options, true);
+}
+function extractText(ast, options, firstParagraphOnly) {
+  const parts = [];
+  const notes = new Set;
+  let footnoteIndex = 0;
+  let firstParagraph;
+  const emit = (value, omit) => {
+    if (!omit)
+      parts.push(value);
+  };
+  const block = (elements, omit) => {
+    emit(`
+
+`, omit);
+    visit(elements, omit);
+    emit(`
+
+`, omit);
+  };
+  const visitBranch = (elements, omit) => {
+    const end = elements.findLastIndex((element) => element.element !== "text" || element.data.trim() !== "") + 1;
+    visit(elements.slice(0, end), omit);
+  };
+  const visit = (elements, inheritedOmit = false) => {
+    for (const element of elements) {
+      if (firstParagraphOnly && firstParagraph !== undefined)
+        return;
+      const omit = inheritedOmit || options.exclude?.(element) === true;
+      switch (element.element) {
+        case "text":
+        case "raw":
+        case "email":
+          emit(element.data, omit);
+          break;
+        case "container": {
+          const start = parts.length;
+          const type = element.data.type;
+          const kind = typeof type === "string" ? type : "block";
+          const hidden = omit || ["ruby-text", "hidden", "invisible"].includes(kind);
+          if ([
+            "paragraph",
+            "div",
+            "blockquote",
+            "note",
+            "heading",
+            "block",
+            "table-row",
+            "definition-list"
+          ].includes(kind))
+            block(element.data.elements, hidden);
+          else
+            visit(element.data.elements, hidden);
+          if (firstParagraphOnly && kind === "paragraph" && firstParagraph === undefined) {
+            const text = normalizeText(parts.slice(start).join(""));
+            if (text)
+              firstParagraph = text;
+          }
+          break;
+        }
+        case "color":
+        case "anchor":
+        case "include":
+          visit(element.data.elements, omit);
+          break;
+        case "collapsible":
+          block(element.data.elements, omit);
+          break;
+        case "tab-view":
+          for (const tab of element.data) {
+            emit(`
+
+${tab.label}
+`, omit);
+            block(tab.elements, omit);
+          }
+          break;
+        case "table":
+          emit(`
+
+`, omit);
+          for (const row of element.data.rows) {
+            for (const cell of row.cells) {
+              visit(cell.elements, omit);
+              emit(" ", omit);
+            }
+            emit(`
+`, omit);
+          }
+          emit(`
+`, omit);
+          break;
+        case "definition-list":
+          for (const entry of element.data) {
+            block(entry.key, omit);
+            visit(entry.value, omit);
+          }
+          break;
+        case "list":
+          for (const item of element.data.items) {
+            emit(`
+`, omit);
+            if (item["item-type"] === "elements")
+              visit(item.elements, omit);
+            else
+              visit([{ element: "list", data: item.data }], omit);
+          }
+          emit(`
+`, omit);
+          break;
+        case "link": {
+          const label = element.data.label;
+          const destination = typeof element.data.link === "string" ? element.data.link : element.data.link.page;
+          const fallback = label === "page" ? destination : ("text" in label) ? label.text : label.url ?? destination;
+          emit(options.resolveText?.(element) ?? fallback, omit);
+          break;
+        }
+        case "user":
+          emit(options.resolveText?.(element) ?? element.data.name, omit);
+          break;
+        case "date":
+        case "math":
+        case "math-inline":
+          emit(options.resolveText?.(element) ?? "", omit);
+          break;
+        case "image":
+          emit(element.data.attributes.alt ?? "", omit);
+          break;
+        case "gallery":
+          if (element.data.content.type === "items")
+            for (const item of element.data.content.items)
+              emit(`${item.alt ?? ""}
+`, omit);
+          break;
+        case "code":
+          emit(`
+
+${element.data.contents}
+
+`, omit);
+          break;
+        case "footnote":
+        case "footnote-ref": {
+          const index = element.element === "footnote" ? footnoteIndex++ : element.data - 1;
+          if (!omit)
+            notes.add(index);
+          break;
+        }
+        case "bibliography-block":
+          for (const entry of element.data.entries)
+            block(entry.value, omit);
+          break;
+        case "if":
+          visitBranch(isTruthy(element.data.condition) ? element.data.then : element.data.else, omit);
+          break;
+        case "ifexpr": {
+          const result = evaluateExpression(element.data.expression);
+          if (result.success)
+            visitBranch(result.value !== 0 ? element.data.then : element.data.else, omit);
+          break;
+        }
+        case "expr": {
+          const result = evaluateExpression(element.data.expression);
+          if (result.success)
+            emit(formatExprValue(result.value), omit);
+          break;
+        }
+        case "line-break":
+        case "line-breaks":
+          emit(`
+`, omit);
+          break;
+        case "horizontal-rule":
+        case "content-separator":
+          emit(`
+
+`, omit);
+          break;
+        default:
+          break;
+      }
+    }
+  };
+  visit(ast.elements);
+  if (firstParagraphOnly)
+    return firstParagraph ?? "";
+  for (const index of notes) {
+    const note = ast.footnotes?.[index];
+    if (note)
+      block(note, false);
+  }
+  return normalizeText(parts.join(""));
+}
+function normalizeText(text) {
+  return text.replace(/[^\S\n]+/g, " ").replace(/ ?\n ?/g, `
+`).replace(/\n{3,}/g, `
+
+`).trim();
+}
+// packages/ast/src/text-excerpt.ts
+var import_re2js = require("re2js");
+var segmenter2 = new Intl.Segmenter("und", { granularity: "grapheme" });
+var emptyExcerpt = (_text) => "";
+var MAX_PATTERN_LENGTH = 4096;
+var MAX_INPUT_LENGTH = 1e6;
+var MAX_PROGRAM_SIZE = 4096;
+var MAX_MATCHES = 1e4;
+var SEARCH_BUDGET = 16000000;
+function compileTextExcerpt(options = {}) {
+  const limit = options.maxLength ?? 200;
+  if (!Number.isSafeInteger(limit) || limit <= 0)
+    return emptyExcerpt;
+  if (options.pattern === undefined) {
+    if (options.flags !== undefined || options.group !== undefined || options.match !== undefined) {
+      return emptyExcerpt;
+    }
+    return (text) => truncateText(text, limit);
+  }
+  const occurrence = options.match ?? 1;
+  const group = options.group ?? 0;
+  if (options.pattern.length > MAX_PATTERN_LENGTH || !Number.isSafeInteger(occurrence) || occurrence < 1 || occurrence > MAX_MATCHES || typeof group === "number" && (!Number.isSafeInteger(group) || group < 0))
+    return emptyExcerpt;
+  let flags = 0;
+  const seenFlags = new Set;
+  for (const flag of options.flags ?? "") {
+    if (seenFlags.has(flag))
+      return emptyExcerpt;
+    seenFlags.add(flag);
+    switch (flag) {
+      case "i":
+        flags |= import_re2js.RE2JS.CASE_INSENSITIVE;
+        break;
+      case "m":
+        flags |= import_re2js.RE2JS.MULTILINE;
+        break;
+      case "s":
+        flags |= import_re2js.RE2JS.DOTALL;
+        break;
+      default:
+        return emptyExcerpt;
+    }
+  }
+  let compiled;
+  try {
+    compiled = import_re2js.RE2JS.compile(options.pattern, flags);
+  } catch {
+    return emptyExcerpt;
+  }
+  const programSize = compiled.programSize();
+  const groupCount = compiled.groupCount();
+  if (programSize > MAX_PROGRAM_SIZE || (typeof group === "number" ? group > groupCount : !Object.hasOwn(compiled.namedGroups(), group)))
+    return emptyExcerpt;
+  return (text) => {
+    if (text.length > MAX_INPUT_LENGTH)
+      return "";
+    const matcher = compiled.matcher(text);
+    let budget = SEARCH_BUDGET;
+    let from = 0;
+    for (let index = 1;index <= occurrence; index++) {
+      budget -= programSize * (groupCount + 1) * (text.length - from + 1);
+      if (budget < 0 || !matcher.find())
+        return "";
+      from = matcher.end();
+    }
+    return truncateText(matcher.group(group) ?? "", limit);
+  };
+}
+function excerptText(text, options = {}) {
+  return compileTextExcerpt(options)(text);
+}
+function truncateText(text, limit) {
+  let count = 0;
+  for (const segment of segmenter2.segment(text)) {
+    if (count++ === limit)
+      return text.slice(0, segment.index);
+  }
+  return text;
+}
+// packages/ast/src/position.ts
+function createPoint(line, column, offset) {
+  return { line, column, offset };
+}
+function createPosition(start, end) {
+  return { start, end };
+}
+// packages/ast/src/element.ts
+function isStringContainerType(type) {
+  return typeof type === "string";
+}
+function isHeaderType(type) {
+  return typeof type === "object" && type !== null && "header" in type;
+}
+function isAlignType(type) {
+  return typeof type === "object" && type !== null && "align" in type;
+}
+function text(value) {
+  return { element: "text", data: value };
+}
+function container(type, elements, attributes = {}) {
+  return {
+    element: "container",
+    data: { type, attributes, elements }
+  };
+}
+function paragraph(elements, attributes = {}) {
+  return container("paragraph", elements, attributes);
+}
+function bold(elements, attributes = {}) {
+  return container("bold", elements, attributes);
+}
+function italics(elements, attributes = {}) {
+  return container("italics", elements, attributes);
+}
+function heading(level, elements, hasToc = true, attributes = {}) {
+  return container({ header: { level, "has-toc": hasToc } }, elements, attributes);
+}
+function lineBreak() {
+  return { element: "line-break" };
+}
+function horizontalRule() {
+  return { element: "horizontal-rule" };
+}
+function link(linkLocation, label, options = {}) {
+  return {
+    element: "link",
+    data: {
+      type: options.type ?? (typeof linkLocation === "string" ? "direct" : "page"),
+      link: linkLocation,
+      extra: options.extra ?? null,
+      label,
+      target: options.target ?? null
+    }
+  };
+}
+function list(type, items, attributes = {}) {
+  return {
+    element: "list",
+    data: { type, attributes, items }
+  };
+}
+function listItemElements(elements, attributes = {}) {
+  return {
+    "item-type": "elements",
+    attributes,
+    elements
+  };
+}
+function listItemSubList(data) {
+  return {
+    "item-type": "sub-list",
+    element: "list",
+    data
+  };
+}
+function isContainerTypeParagraphSafe(type) {
+  if (isHeaderType(type))
+    return false;
+  if (isAlignType(type))
+    return false;
+  switch (type) {
+    case "bold":
+    case "italics":
+    case "underline":
+    case "superscript":
+    case "subscript":
+    case "strikethrough":
+    case "monospace":
+    case "span":
+    case "size":
+      return true;
+    case "div":
+    case "note":
+    case "blockquote":
+    case "paragraph":
+    case "heading":
+    case "collapsible":
+    case "definition-list":
+    case "definition-list-item":
+    case "definition-list-key":
+    case "definition-list-value":
+    case "table-row":
+    case "table-cell":
+      return false;
+    default:
+      return false;
+  }
+}
+function isParagraphSafe(element) {
+  switch (element.element) {
+    case "container": {
+      const data = element.data;
+      return isContainerTypeParagraphSafe(data.type);
+    }
+    case "module":
+      return false;
+    case "text":
+    case "raw":
+    case "variable":
+    case "email":
+      return true;
+    case "table":
+      return false;
+    case "tab-view":
+      return false;
+    case "anchor":
+    case "anchor-name":
+    case "link":
+      return true;
+    case "image":
+      return true;
+    case "list":
+      return false;
+    case "definition-list":
+      return false;
+    case "collapsible":
+      return false;
+    case "table-of-contents":
+      return false;
+    case "footnote":
+      return true;
+    case "footnote-ref":
+      return true;
+    case "footnote-block":
+      return false;
+    case "bibliography-cite":
+      return true;
+    case "bibliography-block":
+      return false;
+    case "user":
+      return true;
+    case "date":
+    case "social":
+      return true;
+    case "button":
+      return isPageButtonAction(element.data.action);
+    case "color":
+      return true;
+    case "code":
+      return false;
+    case "math":
+      return false;
+    case "math-inline":
+      return true;
+    case "embed":
+      return false;
+    case "embed-block":
+      return true;
+    case "html":
+    case "iframe":
+      return false;
+    case "include": {
+      const data = element.data;
+      return data["paragraph-safe"];
+    }
+    case "style":
+      return false;
+    case "line-break":
+    case "line-breaks":
+      return true;
+    case "clear-float":
+      return false;
+    case "horizontal-rule":
+      return false;
+    case "content-separator":
+      return false;
+    case "if-tags":
+      return false;
+    case "expr":
+    case "if":
+    case "ifexpr":
+      return true;
+    default:
+      return false;
+  }
+}
+function isPageButtonAction(action) {
+  switch (action) {
+    case "edit":
+    case "edit-append":
+    case "edit-sections":
+    case "history":
+    case "print":
+    case "files":
+    case "tags":
+    case "source":
+    case "backlinks":
+    case "talk":
+    case "delete":
+    case "rename":
+    case "site-tools":
+    case "edit-meta":
+    case "watchers":
+    case "parent":
+    case "lock-page":
+      return true;
+    default:
+      return false;
+  }
+}
+// packages/ast/src/constants.ts
+var STYLE_SLOT_PREFIX = "\x00__IFTAGS_SLOT__";
+var STYLE_ANCHOR_PREFIX = "\x00__STYLE_ANCHOR__";
+// packages/ast/src/css.ts
+var CSS_LENGTH_UNITS = [
+  "cqmax",
+  "cqmin",
+  "cqw",
+  "cqh",
+  "cqi",
+  "cqb",
+  "svmin",
+  "svmax",
+  "lvmin",
+  "lvmax",
+  "dvmin",
+  "dvmax",
+  "vmin",
+  "vmax",
+  "svw",
+  "svh",
+  "svi",
+  "svb",
+  "lvw",
+  "lvh",
+  "lvi",
+  "lvb",
+  "dvw",
+  "dvh",
+  "dvi",
+  "dvb",
+  "vw",
+  "vh",
+  "vi",
+  "vb",
+  "rcap",
+  "rem",
+  "rex",
+  "rch",
+  "ric",
+  "rlh",
+  "cap",
+  "em",
+  "ex",
+  "ch",
+  "ic",
+  "lh",
+  "cm",
+  "mm",
+  "in",
+  "pc",
+  "pt",
+  "px",
+  "q",
+  "%"
+];
 // packages/ast/src/settings.ts
 function createSettings(mode) {
   switch (mode) {