flappa-doormal 2.18.0 → 2.20.0

package/dist/index.mjs

@@ -148,7 +148,7 @@ numbered: "{{raqms}} {{dash}} " };
 const expandCompositeTokensInTemplate = (template) => {
 	let out = template;
 	for (let i = 0; i < 10; i++) {
-		const next = out.replace(/\{\{(\w+)\}\}/g, (m, tokenName) => COMPOSITE_TOKENS[tokenName] ?? m);
+		const next = out.replace(/\{\{(\w+)\}\}/g, (m, tokenName) => tokenName in COMPOSITE_TOKENS ? COMPOSITE_TOKENS[tokenName] : m);
 		if (next === out) break;
 		out = next;
 	}
@@ -162,7 +162,8 @@ const expandCompositeTokensInTemplate = (template) => {
 * @returns Expanded pattern with base tokens replaced
 * @internal
 */
-const expandBaseTokens = (template) => template.replace(/\{\{(\w+)\}\}/g, (_, tokenName) => BASE_TOKENS[tokenName] ?? `{{${tokenName}}}`);
+const expandBaseTokens = (template) => template.replace(/\{\{(\w+)\}\}/g, (_, tokenName) => tokenName in BASE_TOKENS ? BASE_TOKENS[tokenName] : `{{${tokenName}}}`);
+const EXPANDED_COMPOSITE_TOKENS = Object.fromEntries(Object.entries(COMPOSITE_TOKENS).map(([key, value]) => [key, expandBaseTokens(value)]));
 /**
 * Token definitions mapping human-readable token names to regex patterns.
 *
@@ -190,7 +191,7 @@ const expandBaseTokens = (template) => template.replace(/\{\{(\w+)\}\}/g, (_, to
 */
 const TOKEN_PATTERNS = {
 	...BASE_TOKENS,
-	...Object.fromEntries(Object.entries(COMPOSITE_TOKENS).map(([k, v]) => [k, expandBaseTokens(v)]))
+	...EXPANDED_COMPOSITE_TOKENS
 };
 /**
 * Regex pattern for matching tokens with optional named capture syntax.
@@ -283,8 +284,8 @@ const expandTokenLiteral = (literal, opts) => {
 	if (!parsed) return literal;
 	const { tokenName, captureName } = parsed;
 	if (!tokenName && captureName) return `(?<${opts.registerCapture(captureName)}>.+)`;
+	if (!(tokenName in TOKEN_PATTERNS)) return literal;
 	let tokenPattern = TOKEN_PATTERNS[tokenName];
-	if (!tokenPattern) return literal;
 	tokenPattern = maybeApplyFuzzyToTokenPattern(tokenPattern, opts.fuzzyTransform);
 	if (captureName) return `(?<${opts.registerCapture(captureName)}>${tokenPattern})`;
 	return tokenPattern;
@@ -490,7 +491,7 @@ const applyTokenMappings = (template, mappings) => {
 * // → '{{raqms}} {{dash}}'
 */
 const stripTokenMappings = (template) => {
-	return template.replace(/\{\{([^:}]+):[^}]+\}\}/g, "{{$1}}");
+	return template.replace(/\{\{([^:}]*)?:[^}]+\}\}/g, (_match, tokenName) => `{{${tokenName ?? ""}}}`);
 };
 //#endregion
 //#region src/utils/textUtils.ts
@@ -1279,6 +1280,451 @@ const analyzeTextForRule = (text) => {
 	};
 };
 //#endregion
+//#region src/dictionary/arabic-dictionary-rule.ts
+const uniqueCanonicalWords = (words) => {
+	const seen = /* @__PURE__ */ new Set();
+	const result = [];
+	for (const word of words) {
+		const normalized = normalizeArabicForComparison(word);
+		if (!normalized || seen.has(normalized)) continue;
+		seen.add(normalized);
+		result.push(word);
+	}
+	return result;
+};
+const buildStopAlternation = (stopWords) => {
+	const unique = uniqueCanonicalWords(stopWords);
+	if (unique.length === 0) return "";
+	return unique.map((word) => makeDiacriticInsensitive(normalizeArabicForComparison(word))).join("|");
+};
+const buildHeadwordBody = ({ allowCommaSeparated, colonPattern, stopAlternation, stopwordBody, unit }) => {
+	if (!stopAlternation) return allowCommaSeparated ? `${unit}(?:\\s*[،,]\\s*${unit})*` : unit;
+	const guardedUnit = `(?!(?:${stopwordBody})${allowCommaSeparated ? `(?:\\s*[،,]\\s*|${colonPattern})` : colonPattern})${unit}`;
+	return allowCommaSeparated ? `${guardedUnit}(?:\\s*[،,]\\s*${guardedUnit})*` : guardedUnit;
+};
+const buildBalancedMarker = ({ allowParenthesized, allowWhitespaceBeforeColon, captureName, headwordBody }) => {
+	const colon = allowWhitespaceBeforeColon ? "\\s*:" : ":";
+	const withCapture = `(?<${captureName}>${headwordBody})`;
+	if (!allowParenthesized) return `${withCapture}${colon}`;
+	return `(?:\\(\\s*${withCapture}\\s*\\)|${withCapture})${colon}`;
+};
+const validateDictionaryEntryOptions = ({ captureName = "lemma", maxLetters = 10, minLetters = 2 }) => {
+	if (!Number.isInteger(minLetters) || minLetters < 1) throw new Error(`createArabicDictionaryEntryRule: minLetters must be an integer >= 1, got ${minLetters}`);
+	if (!Number.isInteger(maxLetters) || maxLetters < minLetters) throw new Error(`createArabicDictionaryEntryRule: maxLetters must be an integer >= minLetters, got ${maxLetters}`);
+	if (!/^[A-Za-z_]\w*$/.test(captureName)) throw new Error(`createArabicDictionaryEntryRule: invalid captureName "${captureName}"`);
+};
+const buildArabicDictionaryEntryRegexSource = ({ allowCommaSeparated = false, allowParenthesized = false, allowWhitespaceBeforeColon = false, captureName = "lemma", maxLetters = 10, midLineSubentries = true, minLetters = 2, stopWords }, capturePrefix) => {
+	validateDictionaryEntryOptions({
+		captureName,
+		maxLetters,
+		minLetters
+	});
+	const zeroWidthPrefix = "[\\u200E\\u200F\\u061C\\u200B\\u200C\\u200D\\uFEFF]*";
+	const wawWithMarks = `و${ARABIC_MARKS_CLASS}*`;
+	const alWithMarks = `ا${ARABIC_MARKS_CLASS}*ل${ARABIC_MARKS_CLASS}*`;
+	const lemmaUnit = `(?:${wawWithMarks})?(?:${alWithMarks})?${`${ARABIC_LETTER_WITH_OPTIONAL_MARKS_PATTERN}(?:${ARABIC_LETTER_WITH_OPTIONAL_MARKS_PATTERN}){${minLetters - 1},${maxLetters - 1}}`}`;
+	const stopAlternation = buildStopAlternation(stopWords);
+	const lemmaBody = buildHeadwordBody({
+		allowCommaSeparated,
+		colonPattern: allowWhitespaceBeforeColon ? "\\s*:" : ":",
+		stopAlternation,
+		stopwordBody: stopAlternation ? `(?:${wawWithMarks})?(?:${stopAlternation})` : "",
+		unit: lemmaUnit
+	});
+	const lineStartBoundary = `(?:(?<=^)|(?<=\\n))${zeroWidthPrefix}`;
+	const midLineTrigger = allowParenthesized ? `(?<=\\s)(?=(?:\\(\\s*)?${wawWithMarks}(?:${alWithMarks})?)` : `(?<=\\s)(?=${wawWithMarks}(?:${alWithMarks})?)`;
+	const prefixedCaptureName = capturePrefix ? `${capturePrefix}${captureName}` : captureName;
+	const regex = `(?:${lineStartBoundary}${midLineSubentries ? `|${midLineTrigger}` : ""})` + buildBalancedMarker({
+		allowParenthesized,
+		allowWhitespaceBeforeColon,
+		captureName: prefixedCaptureName,
+		headwordBody: lemmaBody
+	});
+	return {
+		captureNames: [prefixedCaptureName],
+		regex
+	};
+};
+/**
+* Creates a reusable split rule for Arabic dictionary entries.
+*
+* The returned rule preserves authoring intent as a serializable
+* `{ dictionaryEntry: ... }` pattern rather than eagerly compiling to a raw
+* regex string.
+*
+* @example
+* createArabicDictionaryEntryRule({
+*   stopWords: ['وقيل', 'ويقال', 'قال'],
+*   pageStartPrevWordStoplist: ['قال', 'وقيل', 'ويقال'],
+* })
+*
+* @example
+* createArabicDictionaryEntryRule({
+*   allowParenthesized: true,
+*   allowWhitespaceBeforeColon: true,
+*   allowCommaSeparated: true,
+*   stopWords: ['الليث', 'العجاج'],
+* })
+*/
+/**
+* @deprecated Prefer the top-level `SegmentationOptions.dictionary` profile for
+* whole-book dictionary segmentation. Keep this helper for advanced single-rule
+* composition inside a broader `SplitRule[]` pipeline.
+*/
+const createArabicDictionaryEntryRule = ({ allowCommaSeparated = false, allowParenthesized = false, allowWhitespaceBeforeColon = false, captureName = "lemma", maxLetters = 10, meta, midLineSubentries = true, minLetters = 2, pageStartPrevWordStoplist, samePagePrevWordStoplist, stopWords }) => {
+	validateDictionaryEntryOptions({
+		captureName,
+		maxLetters,
+		minLetters
+	});
+	return {
+		dictionaryEntry: {
+			allowCommaSeparated,
+			allowParenthesized,
+			allowWhitespaceBeforeColon,
+			captureName,
+			maxLetters,
+			midLineSubentries,
+			minLetters,
+			stopWords: uniqueCanonicalWords(stopWords)
+		},
+		meta,
+		pageStartPrevWordStoplist,
+		samePagePrevWordStoplist
+	};
+};
+//#endregion
+//#region src/dictionary/heading-classifier.ts
+const HEADING_PREFIX$1 = "## ";
+const CODE_LINE_PATTERN$1 = getTokenPattern("harfs").replaceAll("\\s+", "[ \\t]+");
+const ARABIC_WORD_PATTERN = ARABIC_WORD_WITH_OPTIONAL_MARKS_PATTERN;
+const PLAIN_ENTRY_RE = new RegExp(`^(?<lemma>${ARABIC_WORD_PATTERN}(?:\\s+${ARABIC_WORD_PATTERN}){0,1}|[([{]${ARABIC_WORD_PATTERN}(?:\\s+${ARABIC_WORD_PATTERN}){0,1}[)\\]}])\\s*:`, "u");
+const INLINE_SUBENTRY_RE = new RegExp(`(^|[\\s،؛,:.])(?<lemma>و${ARABIC_WORD_PATTERN})\\s*:`, "gu");
+const CODE_LINE_RE = new RegExp(`^(?:[[(])?(?<codes>${CODE_LINE_PATTERN$1})(?:[)\\]])?$`, "u");
+const PAIRED_FORMS_RE = new RegExp(`^(?<forms>${ARABIC_WORD_PATTERN}(?:\\s*[،,]\\s*${ARABIC_WORD_PATTERN})+)\\s*:`, "u");
+const ARABIC_BOUNDARY_OR_PUNCTUATION = "(?=$|[\\s:،؛()\\[\\]{}\\-–—]|[^\\p{Script=Arabic}])";
+const CHAPTER_HEADING_RE = new RegExp(`^(?:[([{]\\s*)?(?:باب|فصل|كتاب|حرف|أبواب)${ARABIC_BOUNDARY_OR_PUNCTUATION}`, "u");
+const CLUSTER_HEADING_RE = new RegExp(`^(?:\\(?\\s*)?(?:أبواب|أبنية)${ARABIC_BOUNDARY_OR_PUNCTUATION}|^(?=.{1,80}$).+?[،,].+?(?:مستعمل|مهمل|مستعملة|مستعملان)(?=$|[.،,:؛\\s])`, "u");
+const STATUS_HEADING_RE = new RegExp(`^(?:${CODE_LINE_PATTERN$1}|(?:(?:${ARABIC_WORD_PATTERN}\\s+){1,3}${ARABIC_WORD_PATTERN}|${ARABIC_WORD_PATTERN}(?:\\s*[،,]\\s*${ARABIC_WORD_PATTERN})+))\\s*:?[\\s]*(?:مستعمل|مستعملة|مستعملان|مهمل|مهملة)(?=$|[.،,:؛\\s])`, "u");
+const CODE_NOTE_HEADING_RE = new RegExp(`^(?:${ARABIC_WORD_PATTERN}\\s+){1,3}\\(.+\\)$`, "u");
+const COLON_NOISE_RE = /^.+:\s*.+$/u;
+const CHAPTER_TERMS = [
+	"باب",
+	"فصل",
+	"كتاب",
+	"حرف",
+	"أبواب"
+];
+const MARKER_PREFIXES = [
+	"بسم الله",
+	"توكلت على الله",
+	"آخر كتاب",
+	"ويتلوه"
+];
+const NOISE_TOKENS = [
+	"قال",
+	"وقيل",
+	"ويقال",
+	"وفي",
+	"يعني",
+	"فإذا"
+];
+const emptyCounts = () => ({
+	chapter: 0,
+	cluster: 0,
+	codeLine: 0,
+	entry: 0,
+	inlineSubentry: 0,
+	lineEntry: 0,
+	marker: 0,
+	noise: 0,
+	pairedForms: 0
+});
+const extractWrappedLemma = (lemma) => lemma.replace(/^[[{(]+|[\])}]+$/gu, "").trim();
+const stripLeadingWrappers = (text) => text.replace(/^[[{(]+\s*/u, "").trim();
+const isDelimitedPrefixMatch$1 = (text, prefix) => {
+	if (text === prefix) return true;
+	if (!text.startsWith(prefix)) return false;
+	const nextChar = text[prefix.length];
+	return nextChar === void 0 || /[\s:،؛()[\]{}\-–—]/u.test(nextChar);
+};
+const isCodeHeading = (text) => {
+	if (CODE_LINE_RE.test(text)) return true;
+	const words = text.trim().split(/\s+/u).filter(Boolean);
+	return words.length === 1 && (words[0]?.length ?? 0) === 1;
+};
+const looksLikeNoiseHeading = (text) => {
+	const normalized = normalizeArabicForComparison(text);
+	const wordCount = text.trim().split(/\s+/u).filter(Boolean).length;
+	if (/(?:مستعمل|مهمل|مستعملة|مستعملان)(?=$|[.،,:؛\s])/u.test(text)) return false;
+	if (wordCount >= 8 && COLON_NOISE_RE.test(text)) return true;
+	return NOISE_TOKENS.some((token) => normalized.includes(normalizeArabicForComparison(token))) && wordCount >= 4;
+};
+/**
+* Classifies a markdown heading line produced by `convertContentToMarkdown()`.
+*/
+const classifyDictionaryHeading = (line) => {
+	const text = line.startsWith(HEADING_PREFIX$1) ? line.slice(3).trim() : line.trim();
+	const unwrapped = stripLeadingWrappers(text);
+	if (!text) return "noise";
+	if (CHAPTER_HEADING_RE.test(text) || CHAPTER_TERMS.some((term) => isDelimitedPrefixMatch$1(normalizeArabicForComparison(unwrapped), normalizeArabicForComparison(term)))) return "chapter";
+	if (looksLikeNoiseHeading(text)) return "noise";
+	if (isCodeHeading(text)) return "marker";
+	if (MARKER_PREFIXES.some((token) => normalizeArabicForComparison(unwrapped).startsWith(normalizeArabicForComparison(token)))) return "marker";
+	if (STATUS_HEADING_RE.test(text) || CODE_NOTE_HEADING_RE.test(text)) return "marker";
+	if (CLUSTER_HEADING_RE.test(text)) return "cluster";
+	return "entry";
+};
+const createHeadingMatch = (kind, page, rawLine, lineNumber) => ({
+	kind,
+	lemma: kind === "entry" ? rawLine.slice(3).trim() : void 0,
+	line: lineNumber,
+	pageId: page.id,
+	text: rawLine
+});
+const createSurfaceMatch = (kind, page, text, lineNumber, lemma) => ({
+	kind,
+	lemma,
+	line: lineNumber,
+	pageId: page.id,
+	text
+});
+const scanHeadingLine = (page, rawLine, lineNumber, matches) => {
+	if (!rawLine.startsWith(HEADING_PREFIX$1)) return false;
+	const kind = classifyDictionaryHeading(rawLine);
+	matches.push(createHeadingMatch(kind, page, rawLine, lineNumber));
+	return true;
+};
+const scanLineEntry = (page, rawLine, lineNumber, matches) => {
+	const lineEntry = rawLine.match(PLAIN_ENTRY_RE);
+	if (!lineEntry?.groups?.lemma) return;
+	matches.push(createSurfaceMatch("lineEntry", page, rawLine, lineNumber, extractWrappedLemma(lineEntry.groups.lemma)));
+};
+const scanPairedForms = (page, rawLine, lineNumber, matches) => {
+	const pairedForms = rawLine.match(PAIRED_FORMS_RE);
+	if (!pairedForms?.groups?.forms) return;
+	matches.push(createSurfaceMatch("pairedForms", page, rawLine, lineNumber, pairedForms.groups.forms));
+};
+const scanCodeLine = (page, rawLine, lineNumber, matches) => {
+	const codeLine = rawLine.match(CODE_LINE_RE);
+	if (!codeLine?.groups?.codes) return;
+	matches.push(createSurfaceMatch("codeLine", page, rawLine, lineNumber, codeLine.groups.codes));
+};
+const scanInlineSubentries = (page, rawLine, lineNumber, matches) => {
+	for (const match of rawLine.matchAll(INLINE_SUBENTRY_RE)) {
+		if (!match.groups?.lemma) continue;
+		matches.push(createSurfaceMatch("inlineSubentry", page, match.groups.lemma, lineNumber, match.groups.lemma));
+	}
+};
+/**
+* Extracts dictionary surface matches from a markdown page.
+*/
+const scanDictionaryMarkdownPage = (page) => {
+	const lines = page.content.split(/\n/u);
+	const matches = [];
+	for (let index = 0; index < lines.length; index++) {
+		const rawLine = lines[index]?.trim() ?? "";
+		if (!rawLine) continue;
+		if (scanHeadingLine(page, rawLine, index + 1, matches)) continue;
+		scanLineEntry(page, rawLine, index + 1, matches);
+		scanPairedForms(page, rawLine, index + 1, matches);
+		scanCodeLine(page, rawLine, index + 1, matches);
+		scanInlineSubentries(page, rawLine, index + 1, matches);
+	}
+	return matches;
+};
+/**
+* Aggregates dictionary surface counts across markdown pages.
+*/
+const analyzeDictionaryMarkdownPages = (pages) => {
+	const counts = emptyCounts();
+	const matches = [];
+	for (const page of pages) {
+		const pageMatches = scanDictionaryMarkdownPage(page);
+		for (const match of pageMatches) {
+			counts[match.kind] += 1;
+			matches.push(match);
+		}
+	}
+	return {
+		counts,
+		matches
+	};
+};
+//#endregion
+//#region src/dictionary/profile.ts
+const normalizedProfileCache = /* @__PURE__ */ new WeakMap();
+const normalizeStopLemmaWord = (word) => normalizeArabicForComparison(word).replace(/^[\s:؛،,.!?؟()[\]{}«»"'“”‘’]+/gu, "").replace(/[\s:؛،,.!?؟()[\]{}«»"'“”‘’]+$/gu, "").trim();
+const uniqueNormalizedSet = (values, normalize) => new Set(values.map(normalize).filter(Boolean));
+const assertNever$2 = (value) => {
+	throw new Error(`Unhandled dictionary profile variant: ${JSON.stringify(value)}`);
+};
+const normalizeFamily = (family) => {
+	switch (family.use) {
+		case "heading": return {
+			...family,
+			allowNextLineColon: family.allowNextLineColon ?? false,
+			allowSingleLetter: family.allowSingleLetter ?? false
+		};
+		case "lineEntry": return {
+			...family,
+			allowMultiWord: family.allowMultiWord ?? false,
+			allowWhitespaceBeforeColon: family.allowWhitespaceBeforeColon ?? false,
+			wrappers: family.wrappers ?? "none"
+		};
+		case "inlineSubentry": return {
+			...family,
+			prefixes: family.prefixes ?? ["و"],
+			stripPrefixesFromLemma: family.stripPrefixesFromLemma ?? true
+		};
+		case "codeLine": return {
+			...family,
+			wrappers: family.wrappers ?? "either"
+		};
+		case "pairedForms": return {
+			...family,
+			requireStatusTail: family.requireStatusTail ?? false,
+			separator: family.separator ?? "comma"
+		};
+		default: return assertNever$2(family);
+	}
+};
+const normalizeBlocker = (blocker) => {
+	switch (blocker.use) {
+		case "authorityIntro": return {
+			...blocker,
+			precision: blocker.precision ?? "high"
+		};
+		case "stopLemma": return {
+			...blocker,
+			normalizedWords: uniqueNormalizedSet(blocker.words, normalizeStopLemmaWord)
+		};
+		case "previousWord": return {
+			...blocker,
+			normalizedWords: uniqueNormalizedSet(blocker.words, normalizeArabicForComparison)
+		};
+		case "previousChar": return {
+			...blocker,
+			charSet: new Set(blocker.chars)
+		};
+		case "intro":
+		case "pageContinuation": return blocker;
+		default: return assertNever$2(blocker);
+	}
+};
+const normalizeZone = (zone) => ({
+	blockers: (zone.blockers ?? []).map(normalizeBlocker),
+	families: zone.families.map(normalizeFamily),
+	name: zone.name,
+	when: zone.when ? {
+		activateAfter: zone.when.activateAfter,
+		maxPageId: zone.when.maxPageId,
+		minPageId: zone.when.minPageId
+	} : void 0
+});
+const createIssue$1 = (code, path, message, zoneName) => ({
+	code,
+	message,
+	path,
+	...zoneName ? { zoneName } : {}
+});
+const validateGate = (gate, zone, gateIndex, seenActivateAfterKeys, issues) => {
+	const gatePath = `zones[].when.activateAfter[${gateIndex}]`.replace("[]", `[${zone.name}]`);
+	if (gate.use === "headingText") {
+		if (!gate.match.trim()) issues.push(createIssue$1("invalid_gate_match", `${gatePath}.match`, `dictionary gate match must be non-empty`, zone.name));
+		if (gate.fuzzy !== void 0 && typeof gate.fuzzy !== "boolean") issues.push(createIssue$1("invalid_gate_fuzzy", `${gatePath}.fuzzy`, `dictionary gate fuzzy must be a boolean when provided`, zone.name));
+	}
+	const dedupeKey = `${gate.use}:${JSON.stringify(gate)}`;
+	if (seenActivateAfterKeys.has(dedupeKey)) issues.push(createIssue$1("duplicate_activate_after_gate", gatePath, `dictionary zone "${zone.name}" has duplicate activateAfter gates`, zone.name));
+	seenActivateAfterKeys.add(dedupeKey);
+};
+const validateFamily = (family, zone, familyIndex, issues) => {
+	const familyPath = `zones[].families[${familyIndex}]`.replace("[]", `[${zone.name}]`);
+	switch (family.use) {
+		case "heading":
+			if (family.classes.length === 0) issues.push(createIssue$1("empty_heading_classes", `${familyPath}.classes`, `dictionary heading family in zone "${zone.name}" must include at least one class`, zone.name));
+			if (family.emit === "chapter" && !family.classes.includes("chapter")) issues.push(createIssue$1("inert_heading_family", familyPath, `dictionary heading family in zone "${zone.name}" emits "chapter" but never matches chapter headings`, zone.name));
+			if (family.emit === "marker" && !family.classes.includes("marker")) issues.push(createIssue$1("inert_heading_family", familyPath, `dictionary heading family in zone "${zone.name}" emits "marker" but never matches marker headings`, zone.name));
+			if (family.emit === "entry" && !family.classes.includes("entry")) issues.push(createIssue$1("inert_heading_family", familyPath, `dictionary heading family in zone "${zone.name}" emits "entry" but never matches entry headings`, zone.name));
+			break;
+		case "lineEntry": break;
+		case "inlineSubentry":
+			if (family.prefixes?.some((prefix) => !prefix.trim())) issues.push(createIssue$1("empty_inline_prefixes", `${familyPath}.prefixes`, `inlineSubentry prefixes must be non-empty strings`, zone.name));
+			break;
+		case "codeLine": break;
+		case "pairedForms": break;
+		default: assertNever$2(family);
+	}
+};
+const validateBlocker = (blocker, zone, blockerIndex, issues) => {
+	const blockerPath = `zones[].blockers[${blockerIndex}]`.replace("[]", `[${zone.name}]`);
+	switch (blocker.use) {
+		case "stopLemma":
+			if (blocker.words.length === 0 || blocker.words.some((word) => !word.trim())) issues.push(createIssue$1("invalid_stop_words", `${blockerPath}.words`, `stopLemma blocker in zone "${zone.name}" must include non-empty words`, zone.name));
+			break;
+		case "previousWord":
+			if (blocker.words.length === 0 || blocker.words.some((word) => !word.trim())) issues.push(createIssue$1("invalid_previous_words", `${blockerPath}.words`, `previousWord blocker in zone "${zone.name}" must include non-empty words`, zone.name));
+			break;
+		case "previousChar":
+			if (blocker.chars.length === 0 || blocker.chars.some((char) => !char)) issues.push(createIssue$1("invalid_previous_chars", `${blockerPath}.chars`, `previousChar blocker in zone "${zone.name}" must include chars`, zone.name));
+			break;
+		case "authorityIntro":
+		case "intro":
+		case "pageContinuation": break;
+		default: assertNever$2(blocker);
+	}
+};
+var DictionaryProfileValidationError = class extends Error {
+	issues;
+	constructor(issues) {
+		super(issues.length === 1 ? issues[0].message : `Dictionary profile validation failed with ${issues.length} issues`);
+		this.name = "DictionaryProfileValidationError";
+		this.issues = issues;
+	}
+};
+const validateZone = (zone, zoneIndex, seenZoneNames, issues) => {
+	const zonePath = `zones[${zoneIndex}]`;
+	const trimmedName = zone.name.trim();
+	if (!trimmedName) issues.push(createIssue$1("empty_zone_name", `${zonePath}.name`, `dictionary zone name must be non-empty`));
+	else if (seenZoneNames.has(trimmedName)) issues.push(createIssue$1("duplicate_zone_name", `${zonePath}.name`, `dictionary zone names must be unique; duplicated "${trimmedName}"`, trimmedName));
+	else seenZoneNames.add(trimmedName);
+	if (zone.families.length === 0) issues.push(createIssue$1("empty_zone_families", `${zonePath}.families`, `dictionary zone "${zone.name}" must declare at least one family`, zone.name));
+	if (zone.when?.minPageId !== void 0 && zone.when?.maxPageId !== void 0 && zone.when.minPageId > zone.when.maxPageId) issues.push(createIssue$1("invalid_zone_page_range", `${zonePath}.when`, `dictionary zone "${zone.name}" has minPageId greater than maxPageId`, zone.name));
+	const seenActivateAfterKeys = /* @__PURE__ */ new Set();
+	for (let gateIndex = 0; gateIndex < (zone.when?.activateAfter?.length ?? 0); gateIndex++) validateGate(zone.when.activateAfter[gateIndex], zone, gateIndex, seenActivateAfterKeys, issues);
+	for (let familyIndex = 0; familyIndex < zone.families.length; familyIndex++) validateFamily(zone.families[familyIndex], zone, familyIndex, issues);
+	for (let blockerIndex = 0; blockerIndex < (zone.blockers?.length ?? 0); blockerIndex++) validateBlocker(zone.blockers[blockerIndex], zone, blockerIndex, issues);
+};
+/**
+* Validates a dictionary profile without normalizing it.
+*/
+const validateDictionaryProfile = (profile) => {
+	const issues = [];
+	if (profile.version !== 2) issues.push(createIssue$1("invalid_version", "version", `dictionary profile version must be 2, got ${profile.version}`));
+	if (profile.zones.length === 0) {
+		issues.push(createIssue$1("missing_zones", "zones", `dictionary profile must contain at least one zone`));
+		return issues;
+	}
+	const seenZoneNames = /* @__PURE__ */ new Set();
+	for (let zoneIndex = 0; zoneIndex < profile.zones.length; zoneIndex++) validateZone(profile.zones[zoneIndex], zoneIndex, seenZoneNames, issues);
+	return issues;
+};
+/**
+* Normalizes and validates a dictionary profile before runtime matching.
+*/
+const normalizeDictionaryProfile = (profile) => {
+	const cached = normalizedProfileCache.get(profile);
+	if (cached) return cached;
+	const issues = validateDictionaryProfile(profile);
+	if (issues.length > 0) throw new DictionaryProfileValidationError(issues);
+	const normalized = {
+		version: 2,
+		zones: profile.zones.map(normalizeZone)
+	};
+	normalizedProfileCache.set(profile, normalized);
+	return normalized;
+};
+//#endregion
 //#region src/types/rules.ts
 /**
 * Pattern type key names for split rules.
@@ -1300,9 +1746,850 @@ const PATTERN_TYPE_KEYS = [
 	"lineStartsAfter",
 	"lineEndsWith",
 	"template",
-	"regex"
+	"regex",
+	"dictionaryEntry"
 ];
 //#endregion
+//#region src/segmentation/debug-meta.ts
+const resolveDebugConfig = (debug) => {
+	if (debug === true) return {
+		includeBreakpoint: true,
+		includeRule: true,
+		metaKey: "_flappa"
+	};
+	if (!debug || typeof debug !== "object") return null;
+	const { metaKey, include } = debug;
+	const includeRule = Array.isArray(include) ? include.includes("rule") : true;
+	return {
+		includeBreakpoint: Array.isArray(include) ? include.includes("breakpoint") : true,
+		includeRule,
+		metaKey: typeof metaKey === "string" && metaKey ? metaKey : "_flappa"
+	};
+};
+const getRulePatternType = (rule) => {
+	return PATTERN_TYPE_KEYS.find((key) => key in rule) ?? "regex";
+};
+const isPlainObject$1 = (v) => Boolean(v) && typeof v === "object" && !Array.isArray(v);
+const mergeDebugIntoMeta = (meta, metaKey, patch) => {
+	const out = meta ? { ...meta } : {};
+	const existing = out[metaKey];
+	out[metaKey] = {
+		...isPlainObject$1(existing) ? existing : {},
+		...patch
+	};
+	return out;
+};
+const buildRuleDebugPatch = (ruleIndex, rule, wordIndex) => {
+	const patternType = getRulePatternType(rule);
+	const patterns = rule[patternType];
+	const word = wordIndex !== void 0 && Array.isArray(patterns) && patterns[wordIndex] !== void 0 ? patterns[wordIndex] : void 0;
+	return { rule: {
+		index: ruleIndex,
+		patternType,
+		...wordIndex !== void 0 ? { wordIndex } : {},
+		...word !== void 0 ? { word } : {}
+	} };
+};
+const buildBreakpointDebugPatch = (breakpointIndex, rule, wordIndex) => ({ breakpoint: {
+	index: breakpointIndex,
+	kind: rule.pattern === "" ? "pageBoundary" : rule.regex ? "regex" : "pattern",
+	pattern: rule.pattern ?? rule.regex,
+	...wordIndex !== void 0 ? { wordIndex } : {},
+	...wordIndex !== void 0 && rule.words ? { word: rule.words[wordIndex] } : {}
+} });
+/**
+* Helper to format the debug info into a human-readable string.
+* @param meta - The segment metadata object
+* @param options - Formatting options
+*/
+const formatRuleReason = (rule, concise) => {
+	const { index, patternType, wordIndex, word } = rule;
+	if (concise) return `Rule: ${word ? `"${word}"` : patternType}`;
+	const wordInfo = word ? ` (Matched: "${word}")` : "";
+	return `Rule #${index} (${patternType})${wordIndex !== void 0 ? ` [idx:${wordIndex}]` : ""}${wordInfo}`;
+};
+const formatBreakpointReason = (breakpoint, concise) => {
+	const { index, kind, pattern, wordIndex, word } = breakpoint;
+	if (kind === "pageBoundary") return concise ? "Breakpoint: <page-boundary>" : "Page Boundary (Fallback)";
+	if (concise) return `Breakpoint: ${word ? `"${word}"` : `"${pattern}"`}`;
+	if (word) return `Breakpoint #${index} (Words) [idx:${wordIndex}] - "${word}"`;
+	return `Breakpoint #${index} (${kind}) - "${pattern}"`;
+};
+const formatContentLengthReason = (split, concise) => {
+	const { maxContentLength, splitReason } = split;
+	if (concise) return `> ${maxContentLength} (${splitReason})`;
+	return `Safety Split (${splitReason}) > ${maxContentLength}`;
+};
+/**
+* Helper to format the debug info into a human-readable string.
+* @param meta - The segment metadata object
+* @param options - Formatting options
+*/
+const getDebugReason = (meta, options) => {
+	const debug = meta?._flappa;
+	if (!debug) return "-";
+	const concise = options?.concise;
+	if (debug.rule) return formatRuleReason(debug.rule, concise);
+	if (debug.breakpoint) return formatBreakpointReason(debug.breakpoint, concise);
+	if (debug.contentLengthSplit) return formatContentLengthReason(debug.contentLengthSplit, concise);
+	return "Unknown";
+};
+/**
+* Convenience helper to get the formatted debug reason directly from a segment.
+* @param segment - The segment object
+* @param options - Formatting options
+*/
+const getSegmentDebugReason = (segment, options) => {
+	return getDebugReason(segment.meta, options);
+};
+//#endregion
+//#region src/dictionary/runtime.ts
+const INTRO_PHRASES = [
+	"وقال",
+	"قال",
+	"وفي الحديث",
+	"في الحديث",
+	"وفي حديث",
+	"في حديث",
+	"وفي رواية",
+	"في رواية",
+	"وفي قراءة",
+	"في قراءة",
+	"وفي قول",
+	"في قول",
+	"وفي كلام",
+	"في كلام",
+	"ومنه قول",
+	"ومنها قول",
+	"وقرأ",
+	"قرأ",
+	"قراءة",
+	"حديث",
+	"ويقال",
+	"وقيل",
+	"قلت",
+	"فقال",
+	"قال الشاعر",
+	"أنشد",
+	"وأنشد"
+];
+const INTRO_TAIL_PHRASES = [
+	"بفتح",
+	"بالفتح",
+	"بكسر",
+	"بالكسر",
+	"بضم",
+	"بالضم",
+	"بالتحريك",
+	"حديث",
+	"الحديث",
+	"في التنزيل",
+	"وفي التنزيل",
+	"في التنزيل العزيز",
+	"وفي التنزيل العزيز",
+	"في مقتل",
+	"وفي مقتل",
+	"في المجاز",
+	"وفي المجاز",
+	"من المجاز",
+	"ومن المجاز",
+	"في رواية",
+	"وفي رواية",
+	"في قراءة",
+	"وفي قراءة",
+	"في قول",
+	"وفي قول",
+	"في كلام",
+	"وفي كلام",
+	"في صفة",
+	"وفي صفة",
+	"في خطبته",
+	"وفي خطبته",
+	"ومنه قول",
+	"ومنها قول",
+	"يقال لرقبة",
+	"على جهتين",
+	"قوله جل",
+	"قوله جل وعز",
+	"جل وعز",
+	"ومنه حديث",
+	"ومنه الحديث",
+	"كرم الله",
+	"صلى الله عليه",
+	"رضي الله عنه",
+	"رضي الله عنها",
+	"رضي الله عنهما",
+	"قال ابو",
+	"وقال ابو",
+	"عن ابي",
+	"قال ابن",
+	"وقال ابن",
+	"عن ابن"
+];
+const INTRO_TAIL_PATTERNS = [
+	/(?:^|\s)(?:في|وفي|ومنه|ومنها)\s+(?:حديث|الحديث|رواية|قراءة|قول|كلام|مقتل|صفة|خطبته)(?:\s+\S+){0,8}$/u,
+	/(?:^|\s)(?:حديث|الحديث|رواية|قراءة|قول|كلام)(?:\s+\S+){1,8}$/u,
+	/(?:^|\s)(?:قوله|قول(?:ه|هم)?|قال(?:\s+قائل)?|وقرأ|قرأ|قراءة)\s+(?:جل(?:\s+وعز)?|[^\s]+)$/u,
+	/(?:^|\s)(?:ابو|ابي|ابا|ابن|بن|بنت)(?:\s+\S+){1,4}$/u,
+	/(?:^|\s)(?:قال|وقال|انشد|وانشد|روي|وروي|اخبر|واخبر)(?:\s+\S+){0,4}$/u
+];
+const QUALIFIER_TAIL_PREFIXES = [
+	"أي",
+	"قال",
+	"تقول",
+	"يقال",
+	"يقول",
+	"يريد",
+	"يُريد",
+	"ويقال",
+	"ويقول",
+	"وجمعه",
+	"وجمعها",
+	"والجميع",
+	"والجمع"
+];
+const STRUCTURAL_LEMMA_PREFIXES = [
+	"لجزء",
+	"جزء",
+	"ومما يستدرك عليه",
+	"آخر حرف",
+	"كتاب حرف"
+];
+const STRUCTURAL_LINE_PATTERNS = [
+	/^\d+\s*-\s*\(.+\)$/u,
+	/^\(.+\)$/u,
+	/^\(.+\)\s*##\s*/u
+];
+const STRUCTURAL_LINE_KEYWORDS = [
+	"باب",
+	"فصل",
+	"حرف",
+	"أبواب",
+	"كتاب",
+	"المعجمة",
+	"المهملة",
+	"المثناة"
+];
+const CONTINUATION_PREV_WORDS = [
+	"بفتح",
+	"بالفتح",
+	"بكسر",
+	"بالكسر",
+	"بضم",
+	"بالضم",
+	"بالتحريك",
+	"قال",
+	"وقال",
+	"وقيل",
+	"ويقال",
+	"يقال",
+	"قلت",
+	"فقال",
+	"قالوا",
+	"من",
+	"في",
+	"على",
+	"إذا",
+	"نحو",
+	"ثم",
+	"وجل"
+];
+const AUTHORITY_RE = /^(?:(?:و)?قال\s+(?:أبو|ابن|ثعلب|الليث|الأزهري|الجوهري|الفراء)\b|(?:أبو|ابن|ثعلب|الليث|الأزهري|الجوهري|الفراء)\s+\S+)/u;
+const AUTHORITY_HEAD_WORDS = [
+	"الأزهري",
+	"الأصمعي",
+	"الأشجعي",
+	"الأموي",
+	"الأمويّ",
+	"الجوهري",
+	"الرياشي",
+	"الزجاج",
+	"الزجاجي",
+	"الشيباني",
+	"الفراء",
+	"الكسائي",
+	"اللحياني",
+	"الليث",
+	"المبرد",
+	"المنذري",
+	"ثعلب",
+	"شمر"
+];
+const STRONG_SENTENCE_TERMINATORS$1 = /[.!?؟؛۔…]$/u;
+const TRAILING_PAGE_WRAP_NOISE$1 = /[\s\u0660-\u0669\d«»"“”'‘’()[\]{}<>]+$/u;
+const TRAILING_WORD_DELIMITERS$1 = /[\s\u0660-\u0669\d«»"“”'‘’()[\]{}<>.,!?؟؛،:]+$/u;
+const ARABIC_WORD_REGEX$1 = new RegExp(ARABIC_WORD_WITH_OPTIONAL_MARKS_PATTERN, "gu");
+const HEADING_PREFIX = "## ";
+const CODE_LINE_PATTERN = getTokenPattern("harfs").replaceAll("\\s+", "[ \\t]+");
+const BARE_CODE_LEMMA_RE = new RegExp(`^(?:${CODE_LINE_PATTERN})$`, "u");
+const STATUS_TAIL_PATTERN = "(?:مستعمل|مستعملة|مستعملان|مهمل|مهملة)";
+const GATE_TOKEN_MAP = {
+	bab: "باب",
+	fasl: "فصل",
+	kitab: "كتاب"
+};
+const GATE_DELIMITER_RE = /[\s:،؛()[\]{}\-–—]/u;
+const assertNever$1 = (value) => {
+	throw new Error(`Unhandled dictionary runtime variant: ${JSON.stringify(value)}`);
+};
+const lineEntryRegexCache = /* @__PURE__ */ new WeakMap();
+const inlineSubentryRegexCache = /* @__PURE__ */ new WeakMap();
+const pairedFormsRegexCache = /* @__PURE__ */ new WeakMap();
+const trimTrailingPageWrapNoise$1 = (text) => text.trimEnd().replace(TRAILING_PAGE_WRAP_NOISE$1, "");
+const endsWithStrongSentenceTerminator$1 = (pageContent) => {
+	return STRONG_SENTENCE_TERMINATORS$1.test(trimTrailingPageWrapNoise$1(pageContent));
+};
+const extractLastArabicWord$1 = (text, endExclusive = text.length) => {
+	const windowStart = Math.max(0, endExclusive - 256);
+	const withoutTrailingDelimiters = trimTrailingPageWrapNoise$1(text.slice(windowStart, endExclusive)).replace(TRAILING_WORD_DELIMITERS$1, "");
+	let lastMatch = "";
+	ARABIC_WORD_REGEX$1.lastIndex = 0;
+	for (const match of withoutTrailingDelimiters.matchAll(ARABIC_WORD_REGEX$1)) lastMatch = match[0];
+	return lastMatch;
+};
+const previousNonWhitespaceChar = (text, endExclusive = text.length) => {
+	for (let index = endExclusive - 1; index >= 0; index--) {
+		const char = text[index];
+		if (char && !/\s/u.test(char)) return char;
+	}
+	return "";
+};
+const normalizedEquals = (left, right) => normalizeArabicForComparison(left) === normalizeArabicForComparison(right);
+const normalizedStartsWith = (text, prefix) => normalizeArabicForComparison(text).startsWith(normalizeArabicForComparison(prefix));
+const normalizeStopLemma = (text) => normalizeArabicForComparison(text).replace(/^[\s:؛،,.!?؟()[\]{}«»"'“”‘’]+/gu, "").replace(/[\s:؛،,.!?؟()[\]{}«»"'“”‘’]+$/gu, "").trim();
+const getTrailingContext = (text, endExclusive, maxChars = 240) => text.slice(Math.max(0, endExclusive - maxChars), endExclusive);
+const isDelimitedPrefixMatch = (text, prefix) => {
+	if (text === prefix) return true;
+	if (!text.startsWith(prefix)) return false;
+	const nextChar = text[prefix.length];
+	return nextChar === void 0 || GATE_DELIMITER_RE.test(nextChar);
+};
+const createPageContexts = (pages, pageMap, normalizedPages) => {
+	if (normalizedPages && normalizedPages.length !== pages.length) throw new Error(`Dictionary runtime expected ${pages.length} normalized pages, received ${normalizedPages.length}`);
+	if (pageMap.boundaries.length !== pages.length) throw new Error(`Dictionary runtime expected ${pages.length} page boundaries, received ${pageMap.boundaries.length}`);
+	const contexts = [];
+	for (let index = 0; index < pages.length; index++) {
+		const page = pages[index];
+		const boundary = pageMap.boundaries[index];
+		if (!page || !boundary) throw new Error(`Dictionary runtime encountered a missing page or boundary at index ${index}`);
+		const content = normalizedPages?.[index] ?? normalizeLineEndings(page.content);
+		contexts.push({
+			boundary,
+			content,
+			index,
+			lines: buildPageLines(content),
+			page
+		});
+	}
+	return contexts;
+};
+const normalizeIntroContextText = (text) => normalizeArabicForComparison(text).replace(/[\\/]+/gu, " ").replace(/[«»"“”'‘’()[\]{}]+/gu, " ").replace(/\s+/gu, " ").trim();
+const startsWithConfiguredWord = (words, candidate) => words.some((word) => normalizedStartsWith(candidate, word));
+const buildPageLines = (content) => {
+	const parts = content.split("\n");
+	const lines = [];
+	let offset = 0;
+	for (let index = 0; index < parts.length; index++) {
+		const text = parts[index] ?? "";
+		lines.push({
+			lineNumber: index + 1,
+			start: offset,
+			text
+		});
+		offset += text.length + 1;
+	}
+	return lines;
+};
+const headingMatchesGate = (headingText, gate) => {
+	if (gate.use === "headingText") {
+		const useFuzzy = gate.fuzzy ?? false;
+		const source = useFuzzy ? normalizeArabicForComparison(headingText) : headingText.trim();
+		const match = useFuzzy ? normalizeArabicForComparison(gate.match) : gate.match.trim();
+		return !!match && isDelimitedPrefixMatch(source, match);
+	}
+	return normalizedStartsWith(headingText, GATE_TOKEN_MAP[gate.token]);
+};
+const pageMatchesAnyGate = (page, gates) => page.lines.some((line) => {
+	const trimmed = line.text.trim();
+	if (!trimmed.startsWith(HEADING_PREFIX)) return false;
+	const headingText = trimmed.replace(/^##\s+/u, "").trim();
+	return gates.some((gate) => headingMatchesGate(headingText, gate));
+});
+const pageWithinZoneBounds = (zone, pageId) => {
+	if (zone.when?.minPageId !== void 0 && pageId < zone.when.minPageId) return false;
+	if (zone.when?.maxPageId !== void 0 && pageId > zone.when.maxPageId) return false;
+	return true;
+};
+const findActivationPageId = (zone, pages) => {
+	for (const page of pages) {
+		if (!pageWithinZoneBounds(zone, page.page.id)) continue;
+		if (pageMatchesAnyGate(page, zone.when?.activateAfter ?? [])) return page.page.id;
+	}
+	return null;
+};
+const createZoneActivationMap = (profile, pages) => {
+	const activation = /* @__PURE__ */ new Map();
+	for (const zone of profile.zones) {
+		if (!zone.when?.activateAfter?.length) {
+			activation.set(zone.name, null);
+			continue;
+		}
+		activation.set(zone.name, findActivationPageId(zone, pages));
+	}
+	return activation;
+};
+const pageMatchesZone = (zone, activationMap, pageId) => {
+	if (zone.when?.minPageId !== void 0 && pageId < zone.when.minPageId) return false;
+	if (zone.when?.maxPageId !== void 0 && pageId > zone.when.maxPageId) return false;
+	if (!zone.when?.activateAfter?.length) return true;
+	const activatedAt = activationMap.get(zone.name);
+	return activatedAt !== null && activatedAt !== void 0 && pageId >= activatedAt;
+};
+const resolveActiveZone = (profile, activationMap, pageId) => {
+	let activeZone = null;
+	for (const zone of profile.zones) if (pageMatchesZone(zone, activationMap, pageId)) activeZone = zone;
+	return activeZone;
+};
+const createHeadingCandidate = (pageStartOffset, line, nextLine, family, headingClass) => {
+	if (!family.classes.includes(headingClass)) return null;
+	const headingText = line.text.trim().slice(3).trim();
+	if (!family.allowSingleLetter && headingClass === "entry" && headingText.length <= 1) return null;
+	if (headingClass === "entry" && !family.allowNextLineColon && nextLine?.text.trimStart().startsWith(":")) return null;
+	return {
+		absoluteIndex: pageStartOffset + line.start,
+		contentStartOffset: 3,
+		family: "heading",
+		headingClass,
+		kind: family.emit,
+		lemma: family.emit === "entry" ? headingText : void 0,
+		lineNumber: line.lineNumber,
+		localIndex: line.start,
+		probeText: line.text.trim(),
+		text: line.text.trim()
+	};
+};
+const optionalSecondWord = (allowMultiWord) => allowMultiWord ? `(?:\\s+${ARABIC_WORD_WITH_OPTIONAL_MARKS_PATTERN})?` : "";
+const wrappedWordPattern = (open, close, allowMultiWord) => `${open}${ARABIC_WORD_WITH_OPTIONAL_MARKS_PATTERN}${optionalSecondWord(allowMultiWord)}${close}`;
+const bareWordPattern = (allowMultiWord) => `${ARABIC_WORD_WITH_OPTIONAL_MARKS_PATTERN}${optionalSecondWord(allowMultiWord)}`;
+const STATUS_LINE_RE = new RegExp(`^(?:${CODE_LINE_PATTERN}|${ARABIC_WORD_WITH_OPTIONAL_MARKS_PATTERN}(?:\\s*[،,]\\s*${ARABIC_WORD_WITH_OPTIONAL_MARKS_PATTERN})+)\\s*:?[\\s]*${STATUS_TAIL_PATTERN}(?=$|[.،,:؛\\s])`, "u");
+const createLineEntryRegex = (family) => {
+	const cached = lineEntryRegexCache.get(family);
+	if (cached) return cached;
+	const wrapperPattern = family.wrappers === "parentheses" ? wrappedWordPattern("\\(", "\\)", family.allowMultiWord) : family.wrappers === "brackets" ? wrappedWordPattern("\\[", "\\]", family.allowMultiWord) : family.wrappers === "curly" ? wrappedWordPattern("\\{", "\\}", family.allowMultiWord) : family.wrappers === "any" ? `(?:${wrappedWordPattern("\\(", "\\)", family.allowMultiWord)}|${wrappedWordPattern("\\[", "\\]", family.allowMultiWord)}|${wrappedWordPattern("\\{", "\\}", family.allowMultiWord)})` : bareWordPattern(family.allowMultiWord);
+	const colonSpacing = family.allowWhitespaceBeforeColon ? "\\s*:" : ":";
+	const regex = new RegExp(`^(?<lemma>${wrapperPattern})${colonSpacing}`, "u");
+	lineEntryRegexCache.set(family, regex);
+	return regex;
+};
+const collectLineEntryCandidates = (pageStartOffset, line, family) => {
+	const trimmed = line.text.trim();
+	if (STATUS_LINE_RE.test(trimmed)) return [];
+	const match = trimmed.match(createLineEntryRegex(family));
+	if (!match?.groups?.lemma) return [];
+	return [{
+		absoluteIndex: pageStartOffset + line.start,
+		family: "lineEntry",
+		kind: "entry",
+		lemma: match.groups.lemma.replace(/^[[{(]+|[\])}]+$/gu, "").trim(),
+		lineNumber: line.lineNumber,
+		localIndex: line.start,
+		probeText: trimmed,
+		text: trimmed
+	}];
+};
+const collectInlineSubentryCandidates = (pageStartOffset, line, family) => {
+	const cached = inlineSubentryRegexCache.get(family);
+	const prefixes = family.prefixes.length > 0 ? family.prefixes.map(escapeRegex).join("|") : escapeRegex("و");
+	const regex = cached ?? new RegExp(`(^|[\\s،؛,:.])(?<lemma>(?:${prefixes})${ARABIC_WORD_WITH_OPTIONAL_MARKS_PATTERN})\\s*:`, "gu");
+	if (!cached) inlineSubentryRegexCache.set(family, regex);
+	const candidates = [];
+	for (const match of line.text.matchAll(regex)) {
+		if (!match.groups?.lemma || match.index === void 0) continue;
+		const lemmaIndex = match[0].indexOf(match.groups.lemma);
+		if (lemmaIndex < 0) continue;
+		const candidateStart = match.index + lemmaIndex;
+		const lemma = family.stripPrefixesFromLemma ? match.groups.lemma.replace(new RegExp(`^(?:${prefixes})`, "u"), "") : match.groups.lemma;
+		candidates.push({
+			absoluteIndex: pageStartOffset + line.start + candidateStart,
+			family: "inlineSubentry",
+			kind: "entry",
+			lemma,
+			lineNumber: line.lineNumber,
+			localIndex: line.start + candidateStart,
+			probeText: line.text.slice(candidateStart).trimStart(),
+			text: line.text.trim()
+		});
+	}
+	return candidates;
+};
+const CODE_CORE_RE = new RegExp(`^${CODE_LINE_PATTERN}$`, "u");
+const STATUS_SUFFIX_RE = new RegExp(`(?:\\s*:?[\\s]*${STATUS_TAIL_PATTERN}.*)?$`, "u");
+const parseWrappedCode = (text) => {
+	const paired = text.match(/^(?<open>[[(])(?<inner>.+)(?<close>[\])])$/u);
+	if (!paired?.groups?.inner || !paired.groups.open || !paired.groups.close) return null;
+	return {
+		close: paired.groups.close,
+		inner: paired.groups.inner.trim(),
+		open: paired.groups.open,
+		paired: paired.groups.open === "(" && paired.groups.close === ")" || paired.groups.open === "[" && paired.groups.close === "]"
+	};
+};
+const collectCodeLineCandidates = (pageStartOffset, line, family) => {
+	const trimmed = line.text.trim();
+	const bare = trimmed.replace(STATUS_SUFFIX_RE, "").trim();
+	const wrapped = parseWrappedCode(bare);
+	const inner = wrapped?.inner ?? bare;
+	if (!CODE_CORE_RE.test(inner)) return [];
+	if (!(family.wrappers === "either" ? true : family.wrappers === "none" ? wrapped === null : family.wrappers === "paired" ? wrapped?.paired === true : wrapped !== null && !wrapped.paired)) return [];
+	return [{
+		absoluteIndex: pageStartOffset + line.start,
+		family: "codeLine",
+		kind: "marker",
+		lemma: inner,
+		lineNumber: line.lineNumber,
+		localIndex: line.start,
+		probeText: trimmed,
+		text: trimmed
+	}];
+};
+const collectPairedFormsCandidates = (pageStartOffset, line, family) => {
+	const cached = pairedFormsRegexCache.get(family);
+	const separator = family.separator === "space" ? "\\s+" : "\\s*[،,]\\s*";
+	const statusTail = family.requireStatusTail ? "\\s*:\\s*(?:مستعمل|مستعملة|مستعملان|مهمل|مهملة).*" : "\\s*:";
+	const regex = cached ?? new RegExp(`^(?<forms>${ARABIC_WORD_WITH_OPTIONAL_MARKS_PATTERN}(?:${separator}${ARABIC_WORD_WITH_OPTIONAL_MARKS_PATTERN})+)${statusTail}`, "u");
+	if (!cached) pairedFormsRegexCache.set(family, regex);
+	const match = line.text.trim().match(regex);
+	if (!match?.groups?.forms) return [];
+	return [{
+		absoluteIndex: pageStartOffset + line.start,
+		family: "pairedForms",
+		kind: family.emit,
+		lemma: family.emit === "entry" ? match.groups.forms : void 0,
+		lineNumber: line.lineNumber,
+		localIndex: line.start,
+		probeText: line.text.trim(),
+		text: line.text.trim()
+	}];
+};
+const blockerApplies = (blocker, family) => !blocker.appliesTo || blocker.appliesTo.includes(family);
+const isIntroCandidate = (text) => {
+	const normalized = normalizeIntroContextText(text);
+	return INTRO_PHRASES.some((phrase) => normalized.startsWith(normalizeArabicForComparison(phrase)));
+};
+const endsWithIntroPhrase = (text) => {
+	const trimmed = text.trimEnd();
+	if (STRONG_SENTENCE_TERMINATORS$1.test(trimmed)) return false;
+	const normalized = normalizeIntroContextText(trimmed).trimEnd().replace(/[:؛،,.!?؟]+$/u, "").trimEnd();
+	return INTRO_PHRASES.some((phrase) => normalized.endsWith(normalizeArabicForComparison(phrase)));
+};
+const endsWithIntroContext = (text) => {
+	const trimmed = text.trimEnd();
+	if (STRONG_SENTENCE_TERMINATORS$1.test(trimmed)) return false;
+	const normalized = normalizeIntroContextText(trimmed).trimEnd().replace(/[:؛،,.!?؟]+$/u, "").trimEnd();
+	if (!normalized) return false;
+	if (INTRO_PHRASES.some((phrase) => normalized.endsWith(normalizeArabicForComparison(phrase)))) return true;
+	if (INTRO_TAIL_PHRASES.some((phrase) => normalized.endsWith(normalizeArabicForComparison(phrase)))) return true;
+	return INTRO_TAIL_PATTERNS.some((pattern) => pattern.test(normalized));
+};
+const isAuthorityCandidate = (text, precision) => {
+	const head = normalizeStopLemma(text.split(":", 1)[0] ?? text);
+	if (head && AUTHORITY_HEAD_WORDS.some((term) => normalizeStopLemma(term) === head)) return true;
+	if (AUTHORITY_RE.test(text)) return true;
+	if (precision === "aggressive") {
+		const normalized = normalizeIntroContextText(text);
+		return [
+			"الليث",
+			"الأزهري",
+			"الأصمعي",
+			"الجوهري",
+			"الفراء",
+			"ثعلب",
+			"شمر"
+		].some((term) => normalized.startsWith(normalizeArabicForComparison(term)));
+	}
+	return false;
+};
+const hasBlockedQualifierTail = (lemma) => {
+	const parts = lemma.split(/[،,]/u).map((part) => part.trim()).filter(Boolean);
+	if (parts.length < 2) return false;
+	return startsWithConfiguredWord(QUALIFIER_TAIL_PREFIXES, parts.slice(1).join(" "));
+};
+const looksLikeStructuralLeak = (candidate) => {
+	if (!candidate.lemma) return false;
+	const normalizedLemma = normalizeArabicForComparison(candidate.lemma);
+	if (candidate.kind === "entry" && (/^[^\p{Script=Arabic}\d]+/u.test(candidate.lemma) || candidate.lemma.includes("{") || candidate.lemma.includes("}") || candidate.lemma.includes("##"))) return true;
+	if (candidate.kind === "entry" && BARE_CODE_LEMMA_RE.test(candidate.lemma) && (candidate.text === candidate.lemma || candidate.text === `${HEADING_PREFIX}${candidate.lemma}` || candidate.text.startsWith(`${HEADING_PREFIX}${candidate.lemma}`) || candidate.text.startsWith(`${candidate.lemma}\n${HEADING_PREFIX}`))) return true;
+	if (candidate.family !== "pairedForms" && candidate.lemma.split(/\s+/u).filter(Boolean).length > 4) return true;
+	if (startsWithConfiguredWord(STRUCTURAL_LEMMA_PREFIXES, candidate.lemma)) return true;
+	if (normalizedLemma.startsWith(normalizeArabicForComparison("ولل"))) return true;
+	const structuralText = candidate.text.startsWith(HEADING_PREFIX) ? candidate.text.slice(3).trim() : candidate.text;
+	if (/^[\d\u0660-\u0669]+\s*-\s*\([^)]+\)(?:\s+##.*)?$/u.test(structuralText)) return true;
+	const normalizedText = normalizeArabicForComparison(structuralText);
+	if (STRUCTURAL_LINE_PATTERNS.some((pattern) => pattern.test(structuralText))) return STRUCTURAL_LINE_KEYWORDS.some((keyword) => normalizedText.includes(normalizeArabicForComparison(keyword)));
+	return false;
+};
+const countLemma = (map, lemma) => {
+	if (!lemma) return;
+	map.set(lemma, (map.get(lemma) ?? 0) + 1);
+};
+const createInitialKindCounts = () => ({
+	chapter: 0,
+	entry: 0,
+	marker: 0
+});
+const createInitialReasonCounts = () => ({
+	authorityIntro: 0,
+	intro: 0,
+	pageContinuation: 0,
+	previousChar: 0,
+	previousWord: 0,
+	qualifierTail: 0,
+	stopLemma: 0,
+	structuralLeak: 0
+});
+const createInitialFamilyCounts = () => ({
+	codeLine: {
+		accepted: 0,
+		rejected: 0
+	},
+	heading: {
+		accepted: 0,
+		rejected: 0
+	},
+	inlineSubentry: {
+		accepted: 0,
+		rejected: 0
+	},
+	lineEntry: {
+		accepted: 0,
+		rejected: 0
+	},
+	pairedForms: {
+		accepted: 0,
+		rejected: 0
+	}
+});
+const rejectsViaIntroBlocker = (candidate, blocker, localBeforeCandidate) => {
+	if (blocker.use !== "intro") return false;
+	return isIntroCandidate(candidate.probeText) || endsWithIntroPhrase(localBeforeCandidate) || endsWithIntroContext(localBeforeCandidate);
+};
+const rejectsViaAuthorityBlocker = (candidate, blocker) => blocker.use === "authorityIntro" && isAuthorityCandidate(candidate.probeText, blocker.precision);
+const rejectsViaStopLemmaBlocker = (candidate, blocker) => blocker.use === "stopLemma" && !!candidate.lemma && !!normalizeStopLemma(candidate.lemma) && blocker.normalizedWords.has(normalizeStopLemma(candidate.lemma));
+const rejectsViaPreviousWordBlocker = (pageContent, localIndex, blocker) => {
+	if (blocker.use !== "previousWord") return false;
+	const lastWord = extractLastArabicWord$1(pageContent, localIndex);
+	return !!lastWord && blocker.normalizedWords.has(normalizeArabicForComparison(lastWord));
+};
+const rejectsViaPreviousCharBlocker = (pageContent, localIndex, blocker) => {
+	if (blocker.use !== "previousChar") return false;
+	const previousChar = previousNonWhitespaceChar(pageContent, localIndex);
+	return !!previousChar && blocker.charSet.has(previousChar);
+};
+const rejectsViaPageContinuationBlocker = (candidate, blocker, localBeforeCandidate, pageIndex, pages) => {
+	if (blocker.use !== "pageContinuation") return false;
+	if (!(localBeforeCandidate.trim().length === 0) || pageIndex === 0) return false;
+	const previousPage = pages[pageIndex - 1];
+	if (!previousPage || endsWithStrongSentenceTerminator$1(previousPage.content)) return false;
+	const previousWord = extractLastArabicWord$1(previousPage.content);
+	return !!previousWord && CONTINUATION_PREV_WORDS.some((word) => normalizedEquals(word, previousWord)) || endsWithIntroContext(previousPage.content) || isIntroCandidate(candidate.probeText) || isAuthorityCandidate(candidate.probeText, "high");
+};
+const getBlockerRejectionReason = (blocker, candidate, localBeforeCandidate, pageContent, pageIndex, pages) => {
+	if (rejectsViaIntroBlocker(candidate, blocker, localBeforeCandidate)) return "intro";
+	if (rejectsViaAuthorityBlocker(candidate, blocker)) return "authorityIntro";
+	if (rejectsViaStopLemmaBlocker(candidate, blocker)) return "stopLemma";
+	if (rejectsViaPreviousWordBlocker(pageContent, candidate.localIndex, blocker)) return "previousWord";
+	if (rejectsViaPreviousCharBlocker(pageContent, candidate.localIndex, blocker)) return "previousChar";
+	if (rejectsViaPageContinuationBlocker(candidate, blocker, localBeforeCandidate, pageIndex, pages)) return "pageContinuation";
+	return null;
+};
+const getCandidateRejection = (candidate, zone, pageContext, pages) => {
+	const hasQualifierTail = hasBlockedQualifierTail(candidate.lemma ?? "");
+	if (hasQualifierTail || looksLikeStructuralLeak(candidate)) return { reason: hasQualifierTail ? "qualifierTail" : "structuralLeak" };
+	const localBeforeCandidate = getTrailingContext(pageContext.content, candidate.localIndex);
+	for (const blocker of zone.blockers) {
+		if (!blockerApplies(blocker, candidate.family)) continue;
+		const reason = getBlockerRejectionReason(blocker, candidate, localBeforeCandidate, pageContext.content, pageContext.index, pages);
+		if (reason) return { reason };
+	}
+	return null;
+};
+const shouldRejectCandidate = (candidate, zone, pageContext, pages) => {
+	return getCandidateRejection(candidate, zone, pageContext, pages) !== null;
+};
+const collectHeadingCandidates = (pageStartOffset, line, nextLine, family, trimmed) => {
+	if (!trimmed.startsWith(HEADING_PREFIX)) return [];
+	const headingClass = classifyDictionaryHeading(trimmed);
+	if (headingClass === "noise") return [];
+	const candidate = createHeadingCandidate(pageStartOffset, line, nextLine, family, headingClass);
+	return candidate ? [candidate] : [];
+};
+const collectCandidatesForFamily = (pageStartOffset, line, nextLine, family, trimmed) => {
+	switch (family.use) {
+		case "heading": return collectHeadingCandidates(pageStartOffset, line, nextLine, family, trimmed);
+		case "lineEntry": return collectLineEntryCandidates(pageStartOffset, line, family);
+		case "inlineSubentry": return collectInlineSubentryCandidates(pageStartOffset, line, family);
+		case "codeLine": return collectCodeLineCandidates(pageStartOffset, line, family);
+		case "pairedForms": return collectPairedFormsCandidates(pageStartOffset, line, family);
+		default: return assertNever$1(family);
+	}
+};
+const collectCandidatesForLine = (pageStartOffset, line, nextLine, zone) => {
+	const trimmed = line.text.trim();
+	const candidates = [];
+	if (!trimmed) return candidates;
+	for (const family of zone.families) candidates.push(...collectCandidatesForFamily(pageStartOffset, line, nextLine, family, trimmed));
+	return candidates;
+};
+const candidateToSplitPoint = (candidate, debugMetaKey) => {
+	const baseMeta = candidate.lemma ? {
+		kind: candidate.kind,
+		lemma: candidate.lemma
+	} : { kind: candidate.kind };
+	const meta = debugMetaKey === void 0 ? baseMeta : mergeDebugIntoMeta(baseMeta, debugMetaKey, { dictionary: {
+		family: candidate.family,
+		...candidate.headingClass ? { headingClass: candidate.headingClass } : {}
+	} });
+	return {
+		contentStartOffset: candidate.contentStartOffset,
+		index: candidate.absoluteIndex,
+		meta
+	};
+};
+const pushDiagnosticSample = (samples, sampleLimit, sample) => {
+	if (samples.length < sampleLimit) samples.push(sample);
+};
+/**
+* Collects dictionary-profile split points using the pages-only markdown surface.
+*/
+const collectDictionarySplitPoints = (pages, profile, pageMap, normalizedPages, logger, debugMetaKey) => {
+	const normalizedProfile = normalizeDictionaryProfile(profile);
+	const pageContexts = createPageContexts(pages, pageMap, normalizedPages);
+	const activationMap = createZoneActivationMap(normalizedProfile, pageContexts);
+	const splitPoints = [];
+	logger?.debug?.("[dictionary] collecting split points", {
+		pageCount: pages.length,
+		zoneCount: normalizedProfile.zones.length
+	});
+	for (const pageContext of pageContexts) {
+		const zone = resolveActiveZone(normalizedProfile, activationMap, pageContext.page.id);
+		if (!zone) continue;
+		for (let lineIndex = 0; lineIndex < pageContext.lines.length; lineIndex++) {
+			const line = pageContext.lines[lineIndex];
+			const nextLine = pageContext.lines[lineIndex + 1];
+			const candidates = collectCandidatesForLine(pageContext.boundary.start, line, nextLine, zone);
+			for (const candidate of candidates) {
+				if (shouldRejectCandidate(candidate, zone, pageContext, pageContexts)) continue;
+				splitPoints.push(candidateToSplitPoint(candidate, debugMetaKey));
+			}
+		}
+	}
+	logger?.debug?.("[dictionary] collected split points", { splitPointCount: splitPoints.length });
+	return splitPoints;
+};
+/**
+* Collects authoring diagnostics for a dictionary profile without creating segments.
+*
+* This is useful when tuning blockers and family choices for a new dictionary.
+*/
+const diagnoseDictionaryProfile = (pages, profile, options = {}) => {
+	const normalizedProfile = normalizeDictionaryProfile(profile);
+	const pageMap = {
+		boundaries: [],
+		getId: (offset) => {
+			for (const boundary of pageMap.boundaries) if (offset >= boundary.start && offset <= boundary.end) return boundary.id;
+			return pageMap.boundaries.at(-1)?.id ?? 0;
+		},
+		pageBreaks: [],
+		pageIds: pages.map((page) => page.id)
+	};
+	let offset = 0;
+	const pageContexts = createPageContexts(pages, pageMap, pages.map((page, pageIndex) => {
+		const normalized = normalizeLineEndings(page.content);
+		pageMap.boundaries.push({
+			end: offset + normalized.length,
+			id: page.id,
+			start: offset
+		});
+		if (pageIndex < pages.length - 1) {
+			pageMap.pageBreaks.push(offset + normalized.length);
+			offset += normalized.length + 1;
+		} else offset += normalized.length;
+		return normalized;
+	}));
+	const activationMap = createZoneActivationMap(normalizedProfile, pageContexts);
+	const sampleLimit = options.sampleLimit ?? 50;
+	const acceptedKinds = createInitialKindCounts();
+	const blockerHits = createInitialReasonCounts();
+	const familyCounts = createInitialFamilyCounts();
+	const zoneCounts = {};
+	const rejectedLemmaCounts = /* @__PURE__ */ new Map();
+	const samples = [];
+	let acceptedCount = 0;
+	let rejectedCount = 0;
+	for (const pageContext of pageContexts) {
+		const zone = resolveActiveZone(normalizedProfile, activationMap, pageContext.page.id);
+		if (!zone) continue;
+		zoneCounts[zone.name] ??= {
+			accepted: 0,
+			rejected: 0
+		};
+		for (let lineIndex = 0; lineIndex < pageContext.lines.length; lineIndex++) {
+			const line = pageContext.lines[lineIndex];
+			const nextLine = pageContext.lines[lineIndex + 1];
+			const candidates = collectCandidatesForLine(pageContext.boundary.start, line, nextLine, zone);
+			for (const candidate of candidates) {
+				const rejection = getCandidateRejection(candidate, zone, pageContext, pageContexts);
+				const sampleBase = {
+					absoluteIndex: candidate.absoluteIndex,
+					family: candidate.family,
+					kind: candidate.kind,
+					lemma: candidate.lemma,
+					line: candidate.lineNumber,
+					pageId: pageContext.page.id,
+					text: candidate.text,
+					zone: zone.name
+				};
+				if (rejection) {
+					rejectedCount += 1;
+					blockerHits[rejection.reason] += 1;
+					familyCounts[candidate.family].rejected += 1;
+					zoneCounts[zone.name].rejected += 1;
+					countLemma(rejectedLemmaCounts, candidate.lemma);
+					pushDiagnosticSample(samples, sampleLimit, {
+						...sampleBase,
+						accepted: false,
+						reason: rejection.reason
+					});
+					continue;
+				}
+				acceptedCount += 1;
+				acceptedKinds[candidate.kind] += 1;
+				familyCounts[candidate.family].accepted += 1;
+				zoneCounts[zone.name].accepted += 1;
+				pushDiagnosticSample(samples, sampleLimit, {
+					...sampleBase,
+					accepted: true
+				});
+			}
+		}
+	}
+	const rejectedLemmas = [...rejectedLemmaCounts.entries()].sort((left, right) => right[1] - left[1] || left[0].localeCompare(right[0])).map(([lemma, count]) => ({
+		count,
+		lemma
+	}));
+	return {
+		acceptedCount,
+		acceptedKinds,
+		blockerHits,
+		familyCounts,
+		pageCount: pages.length,
+		rejectedCount,
+		rejectedLemmas,
+		samples,
+		zoneCounts
+	};
+};
+//#endregion
 //#region src/optimization/optimize-rules.ts
 const MERGEABLE_KEYS = new Set([
 	"lineStartsWith",
@@ -1319,11 +2606,17 @@ const getPatternArray = (rule, key) => {
 };
 const getPatternString = (rule, key) => {
 	const value = rule[key];
-	return typeof value === "string" ? value : Array.isArray(value) ? value.join("\n") : "";
+	return typeof value === "string" ? value : Array.isArray(value) ? value.join("\n") : value ? JSON.stringify(value) : "";
 };
 const normalizePatterns = (patterns) => [...new Set(patterns)].sort((a, b) => b.length - a.length || a.localeCompare(b));
+const getDictionaryEntrySpecificityScore = (rule) => {
+	if (!("dictionaryEntry" in rule) || !rule.dictionaryEntry) return 0;
+	const { allowCommaSeparated = false, allowParenthesized = false, allowWhitespaceBeforeColon = false, maxLetters = 10, midLineSubentries = true, minLetters = 2, stopWords } = rule.dictionaryEntry;
+	return minLetters * 20 + maxLetters + (allowCommaSeparated ? 0 : 120) + (allowParenthesized ? 0 : 60) + (allowWhitespaceBeforeColon ? 0 : 20) + (midLineSubentries ? 0 : 160) + Math.min(stopWords.length, 25);
+};
 const getSpecificityScore = (rule) => {
 	const key = getPatternKey(rule);
+	if (key === "dictionaryEntry") return getDictionaryEntrySpecificityScore(rule);
 	return MERGEABLE_KEYS.has(key) ? getPatternArray(rule, key).reduce((max, p) => Math.max(max, p.length), 0) : getPatternString(rule, key).length;
 };
 const createMergeKey = (rule) => {
@@ -1468,89 +2761,6 @@ const applyPreprocessToPage = (content, pageId, transforms) => {
 	}
 	return result;
 };
-//#endregion
-//#region src/segmentation/arabic-dictionary-rule.ts
-const uniqueNormalizedWords = (words) => {
-	const seen = /* @__PURE__ */ new Set();
-	const result = [];
-	for (const word of words) {
-		const normalized = normalizeArabicForComparison(word);
-		if (!normalized || seen.has(normalized)) continue;
-		seen.add(normalized);
-		result.push(normalized);
-	}
-	return result;
-};
-const buildStopAlternation = (stopWords) => {
-	const unique = uniqueNormalizedWords(stopWords);
-	if (unique.length === 0) return "";
-	return unique.map((word) => makeDiacriticInsensitive(word)).join("|");
-};
-const buildHeadwordBody = ({ allowCommaSeparated, colonPattern, stopAlternation, stopwordBody, unit }) => {
-	if (!stopAlternation) return allowCommaSeparated ? `${unit}(?:\\s*[،,]\\s*${unit})*` : unit;
-	const guardedUnit = `(?!(?:${stopwordBody})${allowCommaSeparated ? `(?:\\s*[،,]\\s*|${colonPattern})` : colonPattern})${unit}`;
-	return allowCommaSeparated ? `${guardedUnit}(?:\\s*[،,]\\s*${guardedUnit})*` : guardedUnit;
-};
-const buildBalancedMarker = ({ allowParenthesized, allowWhitespaceBeforeColon, captureName, headwordBody }) => {
-	const colon = allowWhitespaceBeforeColon ? "\\s*:" : ":";
-	const withCapture = captureName ? `(?<${captureName}>${headwordBody})` : `(?:${headwordBody})`;
-	if (!allowParenthesized) return `${withCapture}${colon}`;
-	return `(?:\\(\\s*${withCapture}\\s*\\)|${withCapture})${colon}`;
-};
-/**
-* Creates a reusable split rule for Arabic dictionary entries.
-*
-* The generated rule:
-* - keeps the lemma marker in `segment.content`
-* - stores the lemma in `segment.meta[captureName]`
-* - matches root entries at true line/page starts
-* - matches mid-line subentries conservatively when they begin with `و`
-* - can optionally support parenthesized headwords like `(عنبر) :`
-* - can optionally support comma-separated headword lists like `سبد، دبس:`
-*
-* @example
-* createArabicDictionaryEntryRule({
-*   stopWords: ['وقيل', 'ويقال', 'قال'],
-*   pageStartPrevWordStoplist: ['قال', 'وقيل', 'ويقال'],
-* })
-*
-* @example
-* createArabicDictionaryEntryRule({
-*   allowParenthesized: true,
-*   allowWhitespaceBeforeColon: true,
-*   allowCommaSeparated: true,
-*   stopWords: ['الليث', 'العجاج'],
-* })
-*/
-const createArabicDictionaryEntryRule = ({ allowCommaSeparated = false, allowParenthesized = false, allowWhitespaceBeforeColon = false, captureName = "lemma", maxLetters = 10, meta, minLetters = 2, pageStartPrevWordStoplist, samePagePrevWordStoplist, stopWords }) => {
-	if (!Number.isInteger(minLetters) || minLetters < 1) throw new Error(`createArabicDictionaryEntryRule: minLetters must be an integer >= 1, got ${minLetters}`);
-	if (!Number.isInteger(maxLetters) || maxLetters < minLetters) throw new Error(`createArabicDictionaryEntryRule: maxLetters must be an integer >= minLetters, got ${maxLetters}`);
-	if (!captureName.match(/^[A-Za-z_]\w*$/)) throw new Error(`createArabicDictionaryEntryRule: invalid captureName "${captureName}"`);
-	const zeroWidthPrefix = "[\\u200E\\u200F\\u061C\\u200B\\u200C\\u200D\\uFEFF]*";
-	const wawWithMarks = `و${ARABIC_MARKS_CLASS}*`;
-	const alWithMarks = `ا${ARABIC_MARKS_CLASS}*ل${ARABIC_MARKS_CLASS}*`;
-	const lemmaUnit = `(?:${wawWithMarks})?(?:${alWithMarks})?${`${ARABIC_LETTER_WITH_OPTIONAL_MARKS_PATTERN}(?:${ARABIC_LETTER_WITH_OPTIONAL_MARKS_PATTERN}){${minLetters - 1},${maxLetters - 1}}`}`;
-	const stopAlternation = buildStopAlternation(stopWords);
-	const lemmaBody = buildHeadwordBody({
-		allowCommaSeparated,
-		colonPattern: allowWhitespaceBeforeColon ? "\\s*:" : ":",
-		stopAlternation,
-		stopwordBody: stopAlternation ? `(?:${wawWithMarks})?(?:${stopAlternation})` : "",
-		unit: lemmaUnit
-	});
-	return {
-		meta,
-		pageStartPrevWordStoplist,
-		regex: `(?:${`(?:(?<=^)|(?<=\\n))${zeroWidthPrefix}`}|${allowParenthesized ? `(?<=\\s)(?=(?:\\(\\s*)?${wawWithMarks}(?:${alWithMarks})?)` : `(?<=\\s)(?=${wawWithMarks}(?:${alWithMarks})?)`})` + buildBalancedMarker({
-			allowParenthesized,
-			allowWhitespaceBeforeColon,
-			captureName,
-			headwordBody: lemmaBody
-		}),
-		samePagePrevWordStoplist,
-		split: "at"
-	};
-};
 const WINDOW_PREFIX_LENGTHS = [
 	80,
 	60,
@@ -2458,106 +3668,37 @@ const findSafeBreakPosition = (content, targetPosition, lookbackChars = 100) =>
 	return -1;
 };
 //#endregion
-//#region src/segmentation/debug-meta.ts
-const resolveDebugConfig = (debug) => {
-	if (debug === true) return {
-		includeBreakpoint: true,
-		includeRule: true,
-		metaKey: "_flappa"
-	};
-	if (!debug || typeof debug !== "object") return null;
-	const { metaKey, include } = debug;
-	const includeRule = Array.isArray(include) ? include.includes("rule") : true;
-	return {
-		includeBreakpoint: Array.isArray(include) ? include.includes("breakpoint") : true,
-		includeRule,
-		metaKey: typeof metaKey === "string" && metaKey ? metaKey : "_flappa"
-	};
-};
-const getRulePatternType = (rule) => {
-	return PATTERN_TYPE_KEYS.find((key) => key in rule) ?? "regex";
-};
-const isPlainObject = (v) => Boolean(v) && typeof v === "object" && !Array.isArray(v);
-const mergeDebugIntoMeta = (meta, metaKey, patch) => {
-	const out = meta ? { ...meta } : {};
-	const existing = out[metaKey];
-	out[metaKey] = {
-		...isPlainObject(existing) ? existing : {},
-		...patch
-	};
-	return out;
-};
-const buildRuleDebugPatch = (ruleIndex, rule, wordIndex) => {
-	const patternType = getRulePatternType(rule);
-	const patterns = rule[patternType];
-	const word = wordIndex !== void 0 && Array.isArray(patterns) && patterns[wordIndex] !== void 0 ? patterns[wordIndex] : void 0;
-	return { rule: {
-		index: ruleIndex,
-		patternType,
-		...wordIndex !== void 0 ? { wordIndex } : {},
-		...word !== void 0 ? { word } : {}
-	} };
-};
-const buildBreakpointDebugPatch = (breakpointIndex, rule, wordIndex) => ({ breakpoint: {
-	index: breakpointIndex,
-	kind: rule.pattern === "" ? "pageBoundary" : rule.regex ? "regex" : "pattern",
-	pattern: rule.pattern ?? rule.regex,
-	...wordIndex !== void 0 ? { wordIndex } : {},
-	...wordIndex !== void 0 && rule.words ? { word: rule.words[wordIndex] } : {}
-} });
-/**
-* Helper to format the debug info into a human-readable string.
-* @param meta - The segment metadata object
-* @param options - Formatting options
-*/
-const formatRuleReason = (rule, concise) => {
-	const { index, patternType, wordIndex, word } = rule;
-	if (concise) return `Rule: ${word ? `"${word}"` : patternType}`;
-	const wordInfo = word ? ` (Matched: "${word}")` : "";
-	return `Rule #${index} (${patternType})${wordIndex !== void 0 ? ` [idx:${wordIndex}]` : ""}${wordInfo}`;
-};
-const formatBreakpointReason = (breakpoint, concise) => {
-	const { index, kind, pattern, wordIndex, word } = breakpoint;
-	if (kind === "pageBoundary") return concise ? "Breakpoint: <page-boundary>" : "Page Boundary (Fallback)";
-	if (concise) return `Breakpoint: ${word ? `"${word}"` : `"${pattern}"`}`;
-	if (word) return `Breakpoint #${index} (Words) [idx:${wordIndex}] - "${word}"`;
-	return `Breakpoint #${index} (${kind}) - "${pattern}"`;
-};
-const formatContentLengthReason = (split, concise) => {
-	const { maxContentLength, splitReason } = split;
-	if (concise) return `> ${maxContentLength} (${splitReason})`;
-	return `Safety Split (${splitReason}) > ${maxContentLength}`;
-};
-/**
-* Helper to format the debug info into a human-readable string.
-* @param meta - The segment metadata object
-* @param options - Formatting options
-*/
-const getDebugReason = (meta, options) => {
-	const debug = meta?._flappa;
-	if (!debug) return "-";
-	const concise = options?.concise;
-	if (debug.rule) return formatRuleReason(debug.rule, concise);
-	if (debug.breakpoint) return formatBreakpointReason(debug.breakpoint, concise);
-	if (debug.contentLengthSplit) return formatContentLengthReason(debug.contentLengthSplit, concise);
-	return "Unknown";
-};
-/**
-* Convenience helper to get the formatted debug reason directly from a segment.
-* @param segment - The segment object
-* @param options - Formatting options
-*/
-const getSegmentDebugReason = (segment, options) => {
-	return getDebugReason(segment.meta, options);
-};
-//#endregion
 //#region src/segmentation/pattern-validator.ts
 const KNOWN_TOKENS = new Set(getAvailableTokens());
 const TOKEN_INSIDE_BRACES = /\{\{(\w+)(?::\w+)?\}\}/g;
-const buildBareTokenRegex = () => {
+const BARE_TOKEN_REGEX = (() => {
 	const tokens = [...KNOWN_TOKENS].sort((a, b) => b.length - a.length);
 	return new RegExp(`(?<!\\{\\{)(${tokens.join("|")})(?::\\w+)?(?!\\}\\})`, "g");
+})();
+const createMalformedTokenIssue = (tokenLiteral, side) => {
+	const token = tokenLiteral.split(":", 1)[0] || void 0;
+	return {
+		message: `Token "${tokenLiteral || "unknown"}" appears to be missing ${side} braces.`,
+		suggestion: tokenLiteral ? `{{${tokenLiteral}}}` : void 0,
+		token,
+		type: "missing_braces"
+	};
+};
+const detectMalformedLeftToken = (pattern) => {
+	for (let index = 0; index < pattern.length - 1; index++) {
+		if (pattern.slice(index, index + 2) !== "{{") continue;
+		const closeIndex = pattern.indexOf("}}", index + 2);
+		if (closeIndex === -1) return createMalformedTokenIssue(pattern.slice(index + 2).match(/^\w+(?::\w+)?/u)?.[0] ?? "", "closing");
+		index = closeIndex + 1;
+	}
 };
+const detectMalformedRightToken = (pattern) => {
+	for (let index = 0; index < pattern.length - 1; index++) {
+		if (pattern.slice(index, index + 2) !== "}}") continue;
+		if (pattern.lastIndexOf("{{", index) === -1) return createMalformedTokenIssue(pattern.slice(0, index).match(/(\w+(?::\w+)?)$/u)?.[1] ?? "", "opening");
+	}
+};
+const detectMalformedToken = (pattern) => detectMalformedLeftToken(pattern) ?? detectMalformedRightToken(pattern);
 /**
 * Validates a single pattern for common issues.
 */
@@ -2575,14 +3716,16 @@ const validatePattern = (pattern, seenPatterns) => {
 	TOKEN_INSIDE_BRACES.lastIndex = 0;
 	for (const match of pattern.matchAll(TOKEN_INSIDE_BRACES)) {
 		const name = match[1];
-		if (!KNOWN_TOKENS.has(name)) return {
+		if (name && !KNOWN_TOKENS.has(name)) return {
 			message: `Unknown token: {{${name}}}. Available tokens: ${[...KNOWN_TOKENS].slice(0, 5).join(", ")}...`,
 			suggestion: "Check spelling or use a known token",
 			token: name,
 			type: "unknown_token"
 		};
 	}
-	for (const match of pattern.matchAll(buildBareTokenRegex())) {
+	const malformed = detectMalformedToken(pattern);
+	if (malformed) return malformed;
+	for (const match of pattern.matchAll(BARE_TOKEN_REGEX)) {
 		const [full, name] = match;
 		const idx = match.index;
 		if (pattern.slice(Math.max(0, idx - 2), idx) !== "{{" || pattern.slice(idx + full.length, idx + full.length + 2) !== "}}") return {
@@ -2609,14 +3752,14 @@ const applyRulePatternValidation = (result, key, patterns) => {
 	return true;
 };
 const validateTemplateRule = (rule, result) => {
-	if (rule.template === void 0) return false;
+	if (!("template" in rule)) return false;
 	const issue = validatePattern(rule.template, /* @__PURE__ */ new Set());
 	if (!issue) return false;
 	result.template = issue;
 	return true;
 };
 const validateRegexRule = (rule, result) => {
-	if (rule.regex === void 0) return false;
+	if (!("regex" in rule)) return false;
 	if (!rule.regex.trim()) {
 		result.regex = {
 			message: "Empty pattern is not allowed",
@@ -2636,6 +3779,39 @@ const validateRegexRule = (rule, result) => {
 		return true;
 	}
 };
+const invalidDictionaryEntryIssue = (message) => ({
+	message,
+	type: "invalid_option"
+});
+const addBooleanDictionaryEntryIssue = (issues, key, value) => {
+	if (value !== void 0 && typeof value !== "boolean") issues[key] = invalidDictionaryEntryIssue(`${key} must be a boolean`);
+};
+const addCaptureNameIssue = (issues, captureName) => {
+	if (captureName !== void 0 && !/^[A-Za-z_]\w*$/.test(captureName)) issues.captureName = invalidDictionaryEntryIssue(`captureName must match /^[A-Za-z_]\\w*$/, got "${captureName}"`);
+};
+const addMinLettersIssue = (issues, minLetters) => {
+	if (minLetters !== void 0 && (!Number.isInteger(minLetters) || minLetters < 1)) issues.minLetters = invalidDictionaryEntryIssue("minLetters must be an integer >= 1");
+};
+const addMaxLettersIssue = (issues, maxLetters, minLetters) => {
+	const min = minLetters ?? 2;
+	if (maxLetters !== void 0 && (!Number.isInteger(maxLetters) || maxLetters < min)) issues.maxLetters = invalidDictionaryEntryIssue(`maxLetters must be an integer >= ${min}`);
+};
+const validateDictionaryEntryRule = (rule, result) => {
+	if (!("dictionaryEntry" in rule) || !rule.dictionaryEntry) return false;
+	const issues = {};
+	const { allowCommaSeparated, allowParenthesized, allowWhitespaceBeforeColon, captureName, maxLetters, midLineSubentries, minLetters, stopWords } = rule.dictionaryEntry;
+	if (!Array.isArray(stopWords) || stopWords.some((word) => typeof word !== "string" || !word.trim())) issues.stopWords = invalidDictionaryEntryIssue("stopWords must be a string[] with non-empty entries");
+	addBooleanDictionaryEntryIssue(issues, "allowCommaSeparated", allowCommaSeparated);
+	addBooleanDictionaryEntryIssue(issues, "allowParenthesized", allowParenthesized);
+	addBooleanDictionaryEntryIssue(issues, "allowWhitespaceBeforeColon", allowWhitespaceBeforeColon);
+	addBooleanDictionaryEntryIssue(issues, "midLineSubentries", midLineSubentries);
+	addCaptureNameIssue(issues, captureName);
+	addMinLettersIssue(issues, minLetters);
+	addMaxLettersIssue(issues, maxLetters, minLetters);
+	if (Object.keys(issues).length === 0) return false;
+	result.dictionaryEntry = issues;
+	return true;
+};
 const formatValidationIssue = (_type, issue, loc) => {
 	if (!issue) return null;
 	if (issue.type === "missing_braces") return `${loc}: Missing {{}} around token "${issue.token}"`;
@@ -2665,12 +3841,13 @@ const formatValidationIssue = (_type, issue, loc) => {
 */
 const validateRules = (rules) => rules.map((rule) => {
 	const result = {};
-	const startsWithIssues = applyRulePatternValidation(result, "lineStartsWith", rule.lineStartsWith);
-	const startsAfterIssues = applyRulePatternValidation(result, "lineStartsAfter", rule.lineStartsAfter);
-	const endsWithIssues = applyRulePatternValidation(result, "lineEndsWith", rule.lineEndsWith);
+	const startsWithIssues = applyRulePatternValidation(result, "lineStartsWith", "lineStartsWith" in rule ? rule.lineStartsWith : void 0);
+	const startsAfterIssues = applyRulePatternValidation(result, "lineStartsAfter", "lineStartsAfter" in rule ? rule.lineStartsAfter : void 0);
+	const endsWithIssues = applyRulePatternValidation(result, "lineEndsWith", "lineEndsWith" in rule ? rule.lineEndsWith : void 0);
 	const templateIssues = validateTemplateRule(rule, result);
 	const regexIssues = validateRegexRule(rule, result);
-	return startsWithIssues || startsAfterIssues || endsWithIssues || templateIssues || regexIssues ? result : void 0;
+	const dictionaryEntryIssues = validateDictionaryEntryRule(rule, result);
+	return startsWithIssues || startsAfterIssues || endsWithIssues || templateIssues || regexIssues || dictionaryEntryIssues ? result : void 0;
 });
 /**
 * Formats a validation result array into a list of human-readable error messages.
@@ -2687,8 +3864,12 @@ const validateRules = (rules) => rules.map((rule) => {
 */
 const formatValidationReport = (results) => results.flatMap((result, i) => {
 	if (!result) return [];
-	return Object.entries(result).flatMap(([type, issues]) => (Array.isArray(issues) ? issues : [issues]).map((issue) => formatValidationIssue(type, issue, `Rule ${i + 1}, ${type}`)).filter((msg) => msg !== null));
+	return Object.entries(result).flatMap(([type, issues]) => formatValidationIssues(type, issues, i + 1));
 });
+const formatValidationIssues = (type, issues, ruleNumber) => {
+	if (type === "dictionaryEntry" && issues && typeof issues === "object" && !Array.isArray(issues)) return Object.entries(issues).map(([field, issue]) => formatValidationIssue(type, issue, `Rule ${ruleNumber}, ${type}.${field}`)).filter((msg) => msg !== null);
+	return (Array.isArray(issues) ? issues : [issues]).map((issue) => formatValidationIssue(type, issue, `Rule ${ruleNumber}, ${type}`)).filter((msg) => msg !== null);
+};
 //#endregion
 //#region src/segmentation/breakpoint-processor.ts
 const buildPageIdToIndexMap = (pageIds) => new Map(pageIds.map((id, i) => [id, i]));
@@ -3336,6 +4517,7 @@ const buildLineBasedRuleRegex = (rule, fuzzy, capturePrefix) => {
 	if ("lineStartsWith" in rule && Array.isArray(rule.lineStartsWith) && rule.lineStartsWith.length > 0) return buildLineStartsWithRegexSource(rule.lineStartsWith, fuzzy, capturePrefix);
 	if ("lineEndsWith" in rule && Array.isArray(rule.lineEndsWith) && rule.lineEndsWith.length > 0) return buildLineEndsWithRegexSource(rule.lineEndsWith, fuzzy, capturePrefix);
 	if ("template" in rule && typeof rule.template === "string") return buildTemplateRegexSource(rule.template, capturePrefix);
+	if ("dictionaryEntry" in rule && rule.dictionaryEntry) return buildArabicDictionaryEntryRegexSource(rule.dictionaryEntry, capturePrefix);
 	return null;
 };
 /**
@@ -3358,7 +4540,7 @@ const buildRuleRegex = (rule, capturePrefix) => {
 	let finalRegex = ruleRegexSource?.regex;
 	let allCaptureNames = ruleRegexSource?.captureNames ?? [];
 	if (!finalRegex && "regex" in rule && typeof rule.regex === "string") finalRegex = rule.regex;
-	if (!finalRegex) throw new Error("Rule must specify exactly one pattern type: regex, template, lineStartsWith, lineStartsAfter, or lineEndsWith");
+	if (!finalRegex) throw new Error("Rule must specify exactly one pattern type: regex, template, lineStartsWith, lineStartsAfter, lineEndsWith, or dictionaryEntry");
 	if (allCaptureNames.length === 0) allCaptureNames = extractNamedCaptureNames(finalRegex);
 	return {
 		captureNames: allCaptureNames,
@@ -3902,14 +5084,20 @@ const mergeRecord = (existing, incoming) => existing || incoming ? {
 	...existing ?? {},
 	...incoming ?? {}
 } : void 0;
+const isPlainObject = (value) => typeof value === "object" && value !== null && !Array.isArray(value);
 const mergeSplitPoints = (existing, incoming) => {
 	const preferred = prefersIncomingSplitPoint(existing, incoming) ? incoming : existing;
 	const fallback = preferred === incoming ? existing : incoming;
+	const meta = mergeRecord(existing.meta, incoming.meta);
+	if (meta && isPlainObject(existing.meta?._flappa) && isPlainObject(incoming.meta?._flappa)) meta._flappa = {
+		...existing.meta._flappa,
+		...incoming.meta._flappa
+	};
 	return {
 		...fallback,
 		...preferred,
 		contentStartOffset: preferred.contentStartOffset ?? fallback.contentStartOffset,
-		meta: mergeRecord(existing.meta, incoming.meta),
+		meta,
 		namedCaptures: mergeRecord(existing.namedCaptures, incoming.namedCaptures)
 	};
 };
@@ -4035,7 +5223,7 @@ const convertPageBreaks = (content, startOffset, pageBreaks, pageJoiner) => {
 * });
 */
 const segmentPages = (pages, options) => {
-	const { rules = [], breakpoints = [], prefer = "longer", pageJoiner = "space", logger, maxContentLength, preprocess } = options;
+	const { dictionary, rules = [], breakpoints = [], prefer = "longer", pageJoiner = "space", logger, maxContentLength, preprocess } = options;
 	if (maxContentLength && maxContentLength < 50) throw new Error(`maxContentLength must be at least 50 characters.`);
 	const maxPages = options.maxPages ?? Number.MAX_SAFE_INTEGER;
 	const hasLimits = options.maxPages !== void 0 || maxContentLength !== void 0;
@@ -4059,13 +5247,17 @@ const segmentPages = (pages, options) => {
 		pageIds: pageMap.pageIds,
 		totalContentLength: matchContent.length
 	});
-	const splitPoints = collectSplitPointsFromRules(rules, matchContent, pageMap, debugMetaKey, logger);
+	const splitPointsFromRules = collectSplitPointsFromRules(rules, matchContent, pageMap, debugMetaKey, logger);
+	const splitPointsFromDictionary = dictionary ? collectDictionarySplitPoints(preprocessedPages, dictionary, pageMap, normalizedContent, logger, debugMetaKey) : [];
+	const splitPoints = [...splitPointsFromRules, ...splitPointsFromDictionary];
 	const unique = dedupeSplitPoints(splitPoints);
 	logger?.debug?.("[segmenter] split points collected", {
+		dictionarySplitPoints: splitPointsFromDictionary.length,
 		rawSplitPoints: splitPoints.length,
+		ruleSplitPoints: splitPointsFromRules.length,
 		uniqueSplitPoints: unique.length
 	});
-	let segments = buildSegments(unique, matchContent, pageMap, rules, pageJoiner);
+	let segments = buildSegments(unique, matchContent, pageMap, rules, pageJoiner, dictionary !== void 0);
 	logger?.debug?.("[segmenter] structural segments built", { segmentCount: segments.length });
 	segments = ensureFallbackSegment(segments, preprocessedPages, normalizedContent, pageJoiner);
 	if (hasLimits) {
@@ -4092,7 +5284,7 @@ const segmentPages = (pages, options) => {
 * @param rules - Original rules (for constraint checking on first segment)
 * @returns Array of segment objects
 */
-const buildSegments = (splitPoints, content, pageMap, rules, pageJoiner) => {
+const buildSegments = (splitPoints, content, pageMap, rules, pageJoiner, hasDictionaryProfile) => {
 	const getActualStart = (start, contentStartOffset) => start + (contentStartOffset ?? 0);
 	const trimSegmentText = (sliced, capturedContent, contentStartOffset) => capturedContent?.trim() ?? (contentStartOffset ? sliced.trim() : sliced.replace(/[\s\n]+$/, ""));
 	const getAdjustedStart = (actualStart, sliced, contentStartOffset) => actualStart + (contentStartOffset ? sliced.length - sliced.trimStart().length : 0);
@@ -4136,14 +5328,16 @@ const buildSegments = (splitPoints, content, pageMap, rules, pageJoiner) => {
 	};
 	const segments = [];
 	if (!splitPoints.length) {
-		if (anyRuleAllowsId(rules, pageMap.getId(0))) {
+		const firstId = pageMap.getId(0);
+		if (hasDictionaryProfile || anyRuleAllowsId(rules, firstId)) {
 			const s = createSegment(0, content.length);
 			if (s) segments.push(s);
 		}
 		return segments;
 	}
 	if (splitPoints[0].index > 0) {
-		if (anyRuleAllowsId(rules, pageMap.getId(0))) {
+		const firstId = pageMap.getId(0);
+		if (hasDictionaryProfile || anyRuleAllowsId(rules, firstId)) {
 			const s = createSegment(0, splitPoints[0].index);
 			if (s) segments.push(s);
 		}
@@ -4544,6 +5738,6 @@ const validateSegments = (pages, options, segments, validationOptions) => {
 	};
 };
 //#endregion
-export { ARABIC_BASE_LETTER_CLASS, ARABIC_LETTER_WITH_OPTIONAL_MARKS_PATTERN, ARABIC_MARKS_CLASS, ARABIC_WORD_WITH_OPTIONAL_MARKS_PATTERN, PATTERN_TYPE_KEYS, TOKEN_PATTERNS, Token, analyzeCommonLineStarts, analyzeRepeatingSequences, analyzeTextForRule, applyPreprocessToPage, applyTokenMappings, condenseEllipsis, containsTokens, createArabicDictionaryEntryRule, detectTokenPatterns, escapeRegex, escapeTemplateBrackets, escapeWordsOutsideTokens, expandCompositeTokensInTemplate, expandTokens, expandTokensWithCaptures, fixTrailingWaw, formatValidationReport, generateTemplateFromText, getAvailableTokens, getDebugReason, getSegmentDebugReason, getTokenPattern, makeDiacriticInsensitive, normalizeArabicForComparison, optimizeRules, removeZeroWidth, segmentPages, shouldDefaultToFuzzy, stripTokenMappings, suggestPatternConfig, templateToRegex, validateRules, validateSegments, withCapture };
+export { ARABIC_BASE_LETTER_CLASS, ARABIC_LETTER_WITH_OPTIONAL_MARKS_PATTERN, ARABIC_MARKS_CLASS, ARABIC_WORD_WITH_OPTIONAL_MARKS_PATTERN, DictionaryProfileValidationError, PATTERN_TYPE_KEYS, TOKEN_PATTERNS, Token, analyzeCommonLineStarts, analyzeDictionaryMarkdownPages, analyzeRepeatingSequences, analyzeTextForRule, applyPreprocessToPage, applyTokenMappings, classifyDictionaryHeading, condenseEllipsis, containsTokens, createArabicDictionaryEntryRule, detectTokenPatterns, diagnoseDictionaryProfile, escapeRegex, escapeTemplateBrackets, escapeWordsOutsideTokens, expandCompositeTokensInTemplate, expandTokens, expandTokensWithCaptures, fixTrailingWaw, formatValidationReport, generateTemplateFromText, getAvailableTokens, getDebugReason, getSegmentDebugReason, getTokenPattern, makeDiacriticInsensitive, normalizeArabicForComparison, optimizeRules, removeZeroWidth, scanDictionaryMarkdownPage, segmentPages, shouldDefaultToFuzzy, stripTokenMappings, suggestPatternConfig, templateToRegex, validateDictionaryProfile, validateRules, validateSegments, withCapture };
 
 //# sourceMappingURL=index.mjs.map