opencode-rehydra-core 0.14.0

package/dist/pipeline/tagger.js

@@ -0,0 +1,545 @@
+/**
+ * Replacement Tagger
+ * Replaces PII spans with placeholder tags and builds the PII map
+ */
+import { PIIType, DEFAULT_TAG_FORMAT, } from "../types/index.js";
+import { sortSpansByPosition } from "../utils/offsets.js";
+import { escapeRegExp } from "../utils/regex.js";
+/**
+ * Character class for tag IDs.
+ *
+ * Lowercase alphanumerics only: decimal digits keep counter-assigned IDs
+ * working unchanged, while letters give callers enough space to seed
+ * value-derived IDs without relying on long decimal strings. Uppercase is
+ * excluded so a PII map key `${TYPE}_${id}` stays unambiguous — types are
+ * `[A-Z_]+` and may themselves contain underscores.
+ */
+const ID_CHARS = "[0-9a-z]+";
+/**
+ * Converts a raw id attribute value into a {@link TagId}.
+ * Decimal-only IDs become numbers (preserving the existing numeric API);
+ * everything else is folded to a lowercase string.
+ */
+function parseTagId(raw) {
+    return /^[0-9]+$/.test(raw) ? parseInt(raw, 10) : raw.toLowerCase();
+}
+/**
+ * Generates a PII placeholder tag
+ * Format depends on TagFormat configuration:
+ * - Default: <PII type="TYPE" id="N"/>
+ * - Custom:  [[PII type="TYPE" id="N"]]
+ *
+ * Semantic attributes (gender, scope) are included when provided and not 'unknown'
+ */
+export function generateTag(type, id, semantic, tagFormat = DEFAULT_TAG_FORMAT) {
+    const keyword = tagFormat.keyword ?? "PII";
+    let attrs = `type="${type}"`;
+    // Add semantic attributes if present and meaningful
+    if (semantic?.gender && semantic.gender !== "unknown") {
+        attrs += ` gender="${semantic.gender}"`;
+    }
+    if (semantic?.scope && semantic.scope !== "unknown") {
+        attrs += ` scope="${semantic.scope}"`;
+    }
+    attrs += ` id="${id}"`;
+    return `${tagFormat.open}${keyword} ${attrs}${tagFormat.close}`;
+}
+/**
+ * Parses a PII tag to extract type, id, and semantic attributes
+ * Returns null if not a valid tag
+ *
+ * Supports formats (with configurable delimiters):
+ * - <PII type="TYPE" id="N"/>       (default)
+ * - [[PII type="TYPE" id="N"]]      (custom)
+ * - Plus optional gender/scope attributes
+ */
+export function parseTag(tag, tagFormat = DEFAULT_TAG_FORMAT) {
+    const keyword = tagFormat.keyword ?? "PII";
+    const open = escapeRegExp(tagFormat.open);
+    const close = escapeRegExp(tagFormat.close);
+    // Build regex dynamically from tag format
+    const match = tag.match(new RegExp(`^${open}${escapeRegExp(keyword)}\\s+type="([A-Z_]+)"(?:\\s+gender="(\\w+)")?(?:\\s+scope="([\\w-]+)")?\\s+id="(${ID_CHARS})"\\s*${close}$`));
+    if (match === null) {
+        return null;
+    }
+    const [, typeStr, genderStr, scopeStr, idStr] = match;
+    if (typeStr === undefined || idStr === undefined) {
+        return null;
+    }
+    // Allow any [A-Z_]+ type string — custom recognizers (createCustomIdRecognizer)
+    // may emit types that are not members of the PIIType enum. Downstream code
+    // treats `type` as an opaque string when building PII map keys.
+    const type = typeStr;
+    const id = parseTagId(idStr);
+    // Build semantic attributes if present
+    let semantic;
+    if ((genderStr !== undefined && genderStr !== "") ||
+        (scopeStr !== undefined && scopeStr !== "")) {
+        semantic = {};
+        if (genderStr !== undefined &&
+            genderStr !== "" &&
+            ["male", "female", "neutral", "unknown"].includes(genderStr)) {
+            semantic.gender = genderStr;
+        }
+        if (scopeStr !== undefined &&
+            scopeStr !== "" &&
+            ["city", "country", "region", "macro-region", "unknown"].includes(scopeStr)) {
+            semantic.scope = scopeStr;
+        }
+    }
+    return { type, id, semantic };
+}
+/**
+ * Creates a key for the PII map
+ */
+export function createPIIMapKey(type, id) {
+    return `${type}_${id}`;
+}
+/**
+ * Parses a PII map key to extract type and id
+ * @param key - Key in format "TYPE_ID" (e.g., "PERSON_1")
+ * @returns Parsed type and id, or null if invalid
+ */
+function parsePIIMapKey(key) {
+    const match = key.match(new RegExp(`^([A-Z_]+)_(${ID_CHARS})$`));
+    if (!match || match[1] === undefined || match[2] === undefined) {
+        return null;
+    }
+    // Allow any [A-Z_]+ type — matches what parseTag accepts, so custom-type
+    // entries survive `buildExistingEntityLookup` and session-level ID reuse.
+    const type = match[1];
+    const id = parseTagId(match[2]);
+    return { type, id };
+}
+/**
+ * Builds a reverse lookup and max ID from an existing PII map
+ * @param existingPiiMap - Existing PII map (key → value)
+ * @returns Reverse lookup (type:value → id) and global max numeric ID
+ */
+function buildExistingEntityLookup(existingPiiMap) {
+    const reverseLookup = new Map(); // "TYPE:value" → id
+    let maxId = 0;
+    for (const [key, value] of existingPiiMap) {
+        const parsed = parsePIIMapKey(key);
+        if (parsed !== null) {
+            // Build reverse lookup: "PERSON:Tom" → 1
+            const lookupKey = `${parsed.type}:${value}`;
+            reverseLookup.set(lookupKey, parsed.id);
+            // Track global max ID. Only numeric IDs feed the counter — an
+            // alphanumeric ID has no successor, so it must not affect `nextId`.
+            if (typeof parsed.id === "number" && parsed.id > maxId) {
+                maxId = parsed.id;
+            }
+        }
+    }
+    return { reverseLookup, maxId };
+}
+/**
+ * Tags PII spans in text and builds the PII map
+ * @param text - Input text to tag
+ * @param matches - Detected PII spans
+ * @param policy - Anonymization policy
+ * @param existingPiiMap - Optional existing PII map for session-level ID reuse
+ * @param tagFormat - Tag format configuration (defaults to XML-style)
+ */
+export function tagEntities(text, matches, policy, existingPiiMap, tagFormat = DEFAULT_TAG_FORMAT) {
+    if (matches.length === 0) {
+        return {
+            anonymizedText: text,
+            entities: [],
+            piiMap: new Map(),
+        };
+    }
+    // Sort by start position ascending for ID assignment
+    const sortedAscending = sortSpansByPosition(matches);
+    // Build lookup from existing PII map (if provided)
+    const { reverseLookup, maxId } = existingPiiMap
+        ? buildExistingEntityLookup(existingPiiMap)
+        : { reverseLookup: new Map(), maxId: 0 };
+    // Assign IDs
+    const entitiesWithIds = [];
+    // Global ID counter (starts from max existing + 1)
+    let nextId = maxId + 1;
+    // Track seen text for ID reuse within this call (if enabled)
+    const seenText = new Map(); // "type:text" -> id
+    for (const match of sortedAscending) {
+        let id;
+        const lookupKey = `${match.type}:${match.text}`;
+        // First, check if this value exists in the existing PII map (session-level reuse)
+        const existingId = reverseLookup.get(lookupKey);
+        if (existingId !== undefined) {
+            id = existingId;
+            // Also add to seenText so repeated occurrences in this call use the same ID
+            seenText.set(lookupKey, id);
+        }
+        else if (policy.reuseIdsForRepeatedPII) {
+            // Check if we've seen this value earlier in this call
+            const seenId = seenText.get(lookupKey);
+            if (seenId !== undefined) {
+                id = seenId;
+            }
+            else {
+                id = nextId++;
+                seenText.set(lookupKey, id);
+            }
+        }
+        else {
+            id = nextId++;
+        }
+        entitiesWithIds.push({ ...match, id });
+    }
+    // Build PII map
+    const piiMap = new Map();
+    for (const entity of entitiesWithIds) {
+        const key = createPIIMapKey(entity.type, entity.id);
+        piiMap.set(key, entity.text);
+    }
+    // Sort by start position descending for replacement
+    // (replacing from end to start preserves earlier offsets)
+    const sortedDescending = [...entitiesWithIds].sort((a, b) => b.start - a.start);
+    // Perform replacements
+    let anonymizedText = text;
+    for (const entity of sortedDescending) {
+        const tag = generateTag(entity.type, entity.id, entity.semantic, tagFormat);
+        anonymizedText =
+            anonymizedText.slice(0, entity.start) +
+                tag +
+                anonymizedText.slice(entity.end);
+    }
+    // Build final entities list (sorted by position)
+    const entities = entitiesWithIds.map((e) => ({
+        type: e.type,
+        id: e.id,
+        start: e.start,
+        end: e.end,
+        confidence: e.confidence,
+        source: e.source,
+        original: e.text,
+        semantic: e.semantic,
+    }));
+    return {
+        anonymizedText,
+        entities: sortSpansByPosition(entities),
+        piiMap,
+    };
+}
+/**
+ * Validates that a tag is well-formed
+ */
+export function isValidTag(tag, tagFormat = DEFAULT_TAG_FORMAT) {
+    return parseTag(tag, tagFormat) !== null;
+}
+/**
+ * Quote characters that might appear after translation
+ * Includes: standard quotes, smart quotes, German quotes, French quotes, etc.
+ *
+ * Unicode references:
+ * - \u0022 (") Standard double quote
+ * - \u0027 (') Standard single quote
+ * - \u0060 (`) Backtick
+ * - \u00AB («) Left guillemet
+ * - \u00BB (») Right guillemet
+ * - \u2018 (') Left single curly quote
+ * - \u2019 (') Right single curly quote
+ * - \u201A (‚) Single low-9 quote
+ * - \u201C (") Left double curly quote
+ * - \u201D (") Right double curly quote
+ * - \u201E („) Double low-9 quote (German)
+ */
+const QUOTE_CHARS = "(?:\\\\\"|[\"'`\u00AB\u00BB\u2018\u2019\u201A\u201C\u201D\u201E])";
+/**
+ * Whitespace pattern including various unicode spaces
+ */
+const FLEXIBLE_WS = `[\\s\\u00A0\\u2000-\\u200B]*`;
+const FLEXIBLE_WS_REQUIRED = `[\\s\\u00A0\\u2000-\\u200B]+`;
+/**
+ * Builds the opening bracket pattern for fuzzy matching.
+ * For XML-style delimiters, also matches HTML-encoded variants (&lt; / &gt;).
+ * For custom delimiters, matches the literal (escaped) delimiter with flexible internal whitespace.
+ */
+function buildOpenBracketPattern(tagFormat) {
+    const open = tagFormat.open;
+    if (open === "<") {
+        return `(?:<|&lt;)`;
+    }
+    // For multi-character delimiters, allow flexible whitespace between characters
+    // e.g., "[[" → "\[[\s]*\["
+    if (open.length > 1) {
+        return open
+            .split("")
+            .map((ch) => escapeRegExp(ch))
+            .join(FLEXIBLE_WS);
+    }
+    return escapeRegExp(open);
+}
+/**
+ * Builds the closing bracket/self-closing pattern for fuzzy matching.
+ * For XML-style "/>", handles various mangled forms (/>, / >, >, etc.).
+ * For custom delimiters, matches the literal delimiter with flexible internal whitespace.
+ */
+function buildClosingPattern(tagFormat) {
+    const close = tagFormat.close;
+    if (close === "/>") {
+        // XML self-closing: />, / >, >, /&gt;, &gt;, or nothing if already closed inside quotes
+        const CLOSE_BRACKET = `(?:>|&gt;)`;
+        return `(?:${FLEXIBLE_WS}(?:\\/(?:${FLEXIBLE_WS}${CLOSE_BRACKET})?|${CLOSE_BRACKET}))?`;
+    }
+    // For custom close delimiters, allow flexible whitespace between characters
+    if (close.length > 1) {
+        const pattern = close
+            .split("")
+            .map((ch) => escapeRegExp(ch))
+            .join(FLEXIBLE_WS);
+        return `${FLEXIBLE_WS}${pattern}`;
+    }
+    return `${FLEXIBLE_WS}${escapeRegExp(close)}`;
+}
+/**
+ * Builds the malformed id-close pattern for fuzzy matching.
+ * For XML-style tags, handles cases where /> got placed inside the id quotes (e.g., id="7/>").
+ * For custom formats, handles the close delimiter leaking into the id value.
+ */
+function buildIdCloseLeakPattern(tagFormat) {
+    const close = tagFormat.close;
+    if (close === "/>") {
+        return `(?:\\/?(?:>|&gt;)?)?`;
+    }
+    // For custom delimiters, allow individual close-delimiter characters to
+    // partially leak into the id value (e.g., id="7]" for close="]]"),
+    // but not the full delimiter which would consume a valid closing.
+    if (close.length > 1) {
+        const partial = close
+            .slice(0, -1)
+            .split("")
+            .map((ch) => `${escapeRegExp(ch)}?`)
+            .join("");
+        return `(?:${partial})?`;
+    }
+    return "";
+}
+// Cache compiled fuzzy patterns keyed by "open\0close\0keyword"
+const fuzzyPatternCache = new Map();
+function fuzzyPatternCacheKey(tagFormat) {
+    return `${tagFormat.open}\0${tagFormat.close}\0${tagFormat.keyword ?? "PII"}`;
+}
+/**
+ * Builds patterns for fuzzy PII tag matching
+ * Handles various translation artifacts and optional semantic attributes
+ *
+ * For XML-style tags, also handles HTML-encoded brackets (&lt; / &gt;).
+ * For custom formats, handles flexible whitespace within multi-character delimiters.
+ *
+ * Results are cached per unique TagFormat configuration.
+ */
+function buildFuzzyTagPatterns(tagFormat = DEFAULT_TAG_FORMAT) {
+    const cacheKey = fuzzyPatternCacheKey(tagFormat);
+    const cached = fuzzyPatternCache.get(cacheKey);
+    if (cached !== undefined)
+        return cached;
+    const keyword = tagFormat.keyword ?? "PII";
+    const openPattern = buildOpenBracketPattern(tagFormat);
+    const closingPattern = buildClosingPattern(tagFormat);
+    const idCloseLeak = buildIdCloseLeakPattern(tagFormat);
+    // Pattern for type attribute: type = "VALUE" (flexible spacing and quotes)
+    const typeAttr = `type${FLEXIBLE_WS}=${FLEXIBLE_WS}${QUOTE_CHARS}([A-Z_]+)${QUOTE_CHARS}`;
+    // Pattern for id attribute: id = "VALUE" (flexible spacing and quotes)
+    // Also handles malformed cases where close delimiter got placed inside the quotes
+    const idAttr = `id${FLEXIBLE_WS}=${FLEXIBLE_WS}${QUOTE_CHARS}(${ID_CHARS})${idCloseLeak}${QUOTE_CHARS}`;
+    // Optional gender attribute
+    const genderAttr = `(?:${FLEXIBLE_WS}gender${FLEXIBLE_WS}=${FLEXIBLE_WS}${QUOTE_CHARS}(\\w+)${QUOTE_CHARS})?`;
+    // Optional scope attribute
+    const scopeAttr = `(?:${FLEXIBLE_WS}scope${FLEXIBLE_WS}=${FLEXIBLE_WS}${QUOTE_CHARS}([\\w-]+)${QUOTE_CHARS})?`;
+    const escapedKeyword = escapeRegExp(keyword);
+    const patterns = [
+        // type first with optional gender/scope
+        // Groups: type=1, gender=2, scope=3, id=4
+        new RegExp(`${openPattern}${FLEXIBLE_WS}${escapedKeyword}${FLEXIBLE_WS_REQUIRED}${typeAttr}${genderAttr}${scopeAttr}${FLEXIBLE_WS_REQUIRED}${idAttr}${closingPattern}`, "gi"),
+        // id first
+        // Groups: id=1, type=2
+        new RegExp(`${openPattern}${FLEXIBLE_WS}${escapedKeyword}${FLEXIBLE_WS_REQUIRED}${idAttr}${FLEXIBLE_WS_REQUIRED}${typeAttr}${closingPattern}`, "gi"),
+    ];
+    fuzzyPatternCache.set(cacheKey, patterns);
+    return patterns;
+}
+/**
+ * Extracts all PII tags from anonymized text using fuzzy matching
+ * Handles mangled tags that may occur after translation
+ *
+ * Translation can mangle tags by:
+ * - Changing quote types (" → " or „ or « etc.)
+ * - Adding/removing whitespace
+ * - Changing case (type → Type, PII → pii)
+ * - Reordering attributes (id before type)
+ * - Modifying self-closing syntax (/> → / > or >)
+ */
+export function extractTags(anonymizedText, tagFormat = DEFAULT_TAG_FORMAT) {
+    const tags = [];
+    const patterns = buildFuzzyTagPatterns(tagFormat);
+    // Track positions we've already matched to avoid duplicates from overlapping patterns
+    const matchedPositions = new Set();
+    for (let patternIndex = 0; patternIndex < patterns.length; patternIndex++) {
+        const pattern = patterns[patternIndex];
+        if (pattern === undefined)
+            continue;
+        let match;
+        // Reset lastIndex for each pattern
+        pattern.lastIndex = 0;
+        while ((match = pattern.exec(anonymizedText)) !== null) {
+            if (matchedPositions.has(match.index)) {
+                continue; // Skip duplicates from overlapping patterns
+            }
+            // Extract type, id, and semantic attributes based on which pattern matched
+            // Pattern 0: type first with optional gender/scope (groups: type=1, gender=2, scope=3, id=4)
+            // Pattern 1: id first (groups: id=1, type=2)
+            let typeStr;
+            let idStr;
+            let genderStr;
+            let scopeStr;
+            if (patternIndex === 0) {
+                typeStr = match[1];
+                genderStr = match[2];
+                scopeStr = match[3];
+                idStr = match[4];
+            }
+            else {
+                idStr = match[1];
+                typeStr = match[2];
+            }
+            if (typeStr !== undefined && idStr !== undefined) {
+                // Accept any [A-Z_]+ type. Custom recognizers registered via
+                // createCustomIdRecognizer may emit types that are not PIIType enum
+                // members; rehydrate() must still be able to restore them.
+                const type = typeStr.toUpperCase();
+                // Fuzzy matching is case-insensitive, so fold the id back to lowercase
+                // (lossless: ids never contain uppercase).
+                const id = parseTagId(idStr);
+                // Build semantic attributes if present
+                let semantic;
+                if ((genderStr !== undefined && genderStr !== "") ||
+                    (scopeStr !== undefined && scopeStr !== "")) {
+                    semantic = {};
+                    if (genderStr !== undefined &&
+                        genderStr !== "" &&
+                        ["male", "female", "neutral", "unknown"].includes(genderStr.toLowerCase())) {
+                        semantic.gender =
+                            genderStr.toLowerCase();
+                    }
+                    if (scopeStr !== undefined &&
+                        scopeStr !== "" &&
+                        ["city", "country", "region", "macro-region", "unknown"].includes(scopeStr.toLowerCase())) {
+                        semantic.scope =
+                            scopeStr.toLowerCase();
+                    }
+                }
+                tags.push({
+                    type,
+                    id,
+                    position: match.index,
+                    matchedText: match[0],
+                    semantic,
+                });
+                matchedPositions.add(match.index);
+            }
+        }
+    }
+    // Sort by position ascending
+    tags.sort((a, b) => a.position - b.position);
+    return tags;
+}
+/**
+ * Extracts tags using strict matching (original behavior)
+ * Useful when you know tags haven't been mangled
+ * Supports optional gender and scope attributes
+ */
+export function extractTagsStrict(anonymizedText, tagFormat = DEFAULT_TAG_FORMAT) {
+    const tags = [];
+    const keyword = tagFormat.keyword ?? "PII";
+    const open = escapeRegExp(tagFormat.open);
+    const close = escapeRegExp(tagFormat.close);
+    // Pattern matches: OPEN KEYWORD type="X" [gender="Y"] [scope="Z"] id="N" CLOSE
+    const tagPattern = new RegExp(`${open}${escapeRegExp(keyword)}\\s+type="([A-Z_]+)"(?:\\s+gender="(\\w+)")?(?:\\s+scope="([\\w-]+)")?\\s+id="(${ID_CHARS})"\\s*${close}`, "g");
+    let match;
+    while ((match = tagPattern.exec(anonymizedText)) !== null) {
+        const typeStr = match[1];
+        const genderStr = match[2];
+        const scopeStr = match[3];
+        const idStr = match[4];
+        if (typeStr !== undefined && idStr !== undefined) {
+            // Accept any [A-Z_]+ type (see extractTags for rationale).
+            const type = typeStr;
+            const id = parseTagId(idStr);
+            // Build semantic attributes if present
+            let semantic;
+            if ((genderStr !== undefined && genderStr !== "") ||
+                (scopeStr !== undefined && scopeStr !== "")) {
+                semantic = {};
+                if (genderStr !== undefined &&
+                    genderStr !== "" &&
+                    ["male", "female", "neutral", "unknown"].includes(genderStr)) {
+                    semantic.gender = genderStr;
+                }
+                if (scopeStr !== undefined &&
+                    scopeStr !== "" &&
+                    ["city", "country", "region", "macro-region", "unknown"].includes(scopeStr)) {
+                    semantic.scope = scopeStr;
+                }
+            }
+            tags.push({
+                type,
+                id,
+                position: match.index,
+                matchedText: match[0],
+                semantic,
+            });
+        }
+    }
+    return tags;
+}
+/**
+ * Counts entities by type
+ */
+export function countEntitiesByType(entities) {
+    const counts = {};
+    // Initialize all types to 0
+    for (const type of Object.values(PIIType)) {
+        counts[type] = 0;
+    }
+    // Count entities
+    for (const entity of entities) {
+        counts[entity.type] = (counts[entity.type] ?? 0) + 1;
+    }
+    return counts;
+}
+/**
+ * Rehydrates anonymized text using the PII map
+ * Uses fuzzy matching to handle tags that may have been mangled by translation
+ *
+ * @param anonymizedText - Text containing PII tags (possibly mangled)
+ * @param piiMap - Map of PII keys to original values
+ * @param strict - If true, use strict matching (original behavior). Default: false
+ * @param tagFormat - Tag format configuration (defaults to XML-style)
+ * @returns Text with PII tags replaced by original values
+ */
+export function rehydrate(anonymizedText, piiMap, strict = false, tagFormat = DEFAULT_TAG_FORMAT) {
+    let result = anonymizedText;
+    const tags = strict
+        ? extractTagsStrict(anonymizedText, tagFormat)
+        : extractTags(anonymizedText, tagFormat);
+    // Sort by position descending for replacement
+    // (replacing from end to start preserves earlier offsets)
+    tags.sort((a, b) => b.position - a.position);
+    for (const { type, id, position, matchedText } of tags) {
+        const key = createPIIMapKey(type, id);
+        const original = piiMap.get(key);
+        if (original !== undefined) {
+            // Use the actual matched text length for replacement
+            // This handles mangled tags where the length differs from the canonical form
+            result =
+                result.slice(0, position) +
+                    original +
+                    result.slice(position + matchedText.length);
+        }
+    }
+    return result;
+}
+//# sourceMappingURL=tagger.js.map

package/dist/pipeline/tagger.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"tagger.js","sourceRoot":"","sources":["../../src/pipeline/tagger.ts"],"names":[],"mappings":"AAAA;;;GAGG;AAEH,OAAO,EACL,OAAO,EAOP,kBAAkB,GACnB,MAAM,mBAAmB,CAAC;AAC3B,OAAO,EAAE,mBAAmB,EAAE,MAAM,qBAAqB,CAAC;AAC1D,OAAO,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AAEjD;;;;;;;;GAQG;AACH,MAAM,QAAQ,GAAG,WAAW,CAAC;AAE7B;;;;GAIG;AACH,SAAS,UAAU,CAAC,GAAW;IAC7B,OAAO,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,WAAW,EAAE,CAAC;AACtE,CAAC;AA+BD;;;;;;;GAOG;AACH,MAAM,UAAU,WAAW,CACzB,IAAa,EACb,EAAS,EACT,QAA6B,EAC7B,YAAuB,kBAAkB;IAEzC,MAAM,OAAO,GAAG,SAAS,CAAC,OAAO,IAAI,KAAK,CAAC;IAC3C,IAAI,KAAK,GAAG,SAAS,IAAI,GAAG,CAAC;IAE7B,oDAAoD;IACpD,IAAI,QAAQ,EAAE,MAAM,IAAI,QAAQ,CAAC,MAAM,KAAK,SAAS,EAAE,CAAC;QACtD,KAAK,IAAI,YAAY,QAAQ,CAAC,MAAM,GAAG,CAAC;IAC1C,CAAC;IACD,IAAI,QAAQ,EAAE,KAAK,IAAI,QAAQ,CAAC,KAAK,KAAK,SAAS,EAAE,CAAC;QACpD,KAAK,IAAI,WAAW,QAAQ,CAAC,KAAK,GAAG,CAAC;IACxC,CAAC;IAED,KAAK,IAAI,QAAQ,EAAE,GAAG,CAAC;IAEvB,OAAO,GAAG,SAAS,CAAC,IAAI,GAAG,OAAO,IAAI,KAAK,GAAG,SAAS,CAAC,KAAK,EAAE,CAAC;AAClE,CAAC;AAWD;;;;;;;;GAQG;AACH,MAAM,UAAU,QAAQ,CACtB,GAAW,EACX,YAAuB,kBAAkB;IAEzC,MAAM,OAAO,GAAG,SAAS,CAAC,OAAO,IAAI,KAAK,CAAC;IAC3C,MAAM,IAAI,GAAG,YAAY,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC;IAC1C,MAAM,KAAK,GAAG,YAAY,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;IAE5C,0CAA0C;IAC1C,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CACrB,IAAI,MAAM,CACR,IAAI,IAAI,GAAG,YAAY,CAAC,OAAO,CAAC,kFAAkF,QAAQ,SAAS,KAAK,GAAG,CAC5I,CACF,CAAC;IAEF,IAAI,KAAK,KAAK,IAAI,EAAE,CAAC;QACnB,OAAO,IAAI,CAAC;IACd,CAAC;IAED,MAAM,CAAC,EAAE,OAAO,EAAE,SAAS,EAAE,QAAQ,EAAE,KAAK,CAAC,GAAG,KAAK,CAAC;IACtD,IAAI,OAAO,KAAK,SAAS,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;QACjD,OAAO,IAAI,CAAC;IACd,CAAC;IAED,gFAAgF;IAChF,2EAA2E;IAC3E,gEAAgE;IAChE,MAAM,IAAI,GAAG,OAAkB,CAAC;IAChC,MAAM,EAAE,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC;IAE7B,uCAAuC;IACvC,IAAI,QAAwC,CAAC;IAC7C,IACE,CAAC,SAAS,KAAK,SAAS,IAAI,SAAS,KAAK,EAAE,CAAC;QAC7C,CAAC,QAAQ,KAAK,SAAS,IAAI,QAAQ,KAAK,EAAE,CAAC,EAC3C,CAAC;QACD,QAAQ,GAAG,EAAE,CAAC;QACd,IACE,SAAS,KAAK,SAAS;YACvB,SAAS,KAAK,EAAE;YAChB,CAAC,MAAM,EAAE,QAAQ,EAAE,SAAS,EAAE,SAAS,CAAC,CAAC,QAAQ,CAAC,SAAS,CAAC,EAC5D,CAAC;YACD,QAAQ,CAAC,MAAM,GAAG,SAAyC,CAAC;QAC9D,CAAC;QACD,IACE,QAAQ,KAAK,SAAS;YACtB,QAAQ,KAAK,EAAE;YACf,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,EAAE,cAAc,EAAE,SAAS,CAAC,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAC3E,CAAC;YACD,QAAQ,CAAC,KAAK,GAAG,QAAuC,CAAC;QAC3D,CAAC;IACH,CAAC;IAED,OAAO,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,CAAC;AAChC,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,eAAe,CAAC,IAAa,EAAE,EAAS;IACtD,OAAO,GAAG,IAAI,IAAI,EAAE,EAAE,CAAC;AACzB,CAAC;AAED;;;;GAIG;AACH,SAAS,cAAc,CAAC,GAAW;IACjC,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,IAAI,MAAM,CAAC,eAAe,QAAQ,IAAI,CAAC,CAAC,CAAC;IACjE,IAAI,CAAC,KAAK,IAAI,KAAK,CAAC,CAAC,CAAC,KAAK,SAAS,IAAI,KAAK,CAAC,CAAC,CAAC,KAAK,SAAS,EAAE,CAAC;QAC/D,OAAO,IAAI,CAAC;IACd,CAAC;IACD,yEAAyE;IACzE,0EAA0E;IAC1E,MAAM,IAAI,GAAG,KAAK,CAAC,CAAC,CAAY,CAAC;IACjC,MAAM,EAAE,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;IAChC,OAAO,EAAE,IAAI,EAAE,EAAE,EAAE,CAAC;AACtB,CAAC;AAED;;;;GAIG;AACH,SAAS,yBAAyB,CAAC,cAAyB;IAI1D,MAAM,aAAa,GAAG,IAAI,GAAG,EAAiB,CAAC,CAAC,oBAAoB;IACpE,IAAI,KAAK,GAAG,CAAC,CAAC;IAEd,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,cAAc,EAAE,CAAC;QAC1C,MAAM,MAAM,GAAG,cAAc,CAAC,GAAG,CAAC,CAAC;QACnC,IAAI,MAAM,KAAK,IAAI,EAAE,CAAC;YACpB,yCAAyC;YACzC,MAAM,SAAS,GAAG,GAAG,MAAM,CAAC,IAAI,IAAI,KAAK,EAAE,CAAC;YAC5C,aAAa,CAAC,GAAG,CAAC,SAAS,EAAE,MAAM,CAAC,EAAE,CAAC,CAAC;YAExC,8DAA8D;YAC9D,oEAAoE;YACpE,IAAI,OAAO,MAAM,CAAC,EAAE,KAAK,QAAQ,IAAI,MAAM,CAAC,EAAE,GAAG,KAAK,EAAE,CAAC;gBACvD,KAAK,GAAG,MAAM,CAAC,EAAE,CAAC;YACpB,CAAC;QACH,CAAC;IACH,CAAC;IAED,OAAO,EAAE,aAAa,EAAE,KAAK,EAAE,CAAC;AAClC,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,WAAW,CACzB,IAAY,EACZ,OAAoB,EACpB,MAA2B,EAC3B,cAA0B,EAC1B,YAAuB,kBAAkB;IAEzC,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACzB,OAAO;YACL,cAAc,EAAE,IAAI;YACpB,QAAQ,EAAE,EAAE;YACZ,MAAM,EAAE,IAAI,GAAG,EAAE;SAClB,CAAC;IACJ,CAAC;IAED,qDAAqD;IACrD,MAAM,eAAe,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAC;IAErD,mDAAmD;IACnD,MAAM,EAAE,aAAa,EAAE,KAAK,EAAE,GAAG,cAAc;QAC7C,CAAC,CAAC,yBAAyB,CAAC,cAAc,CAAC;QAC3C,CAAC,CAAC,EAAE,aAAa,EAAE,IAAI,GAAG,EAAiB,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC;IAE1D,aAAa;IACb,MAAM,eAAe,GAAqC,EAAE,CAAC;IAE7D,mDAAmD;IACnD,IAAI,MAAM,GAAG,KAAK,GAAG,CAAC,CAAC;IAEvB,6DAA6D;IAC7D,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAiB,CAAC,CAAC,oBAAoB;IAE/D,KAAK,MAAM,KAAK,IAAI,eAAe,EAAE,CAAC;QACpC,IAAI,EAAS,CAAC;QACd,MAAM,SAAS,GAAG,GAAG,KAAK,CAAC,IAAI,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC;QAEhD,kFAAkF;QAClF,MAAM,UAAU,GAAG,aAAa,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QAChD,IAAI,UAAU,KAAK,SAAS,EAAE,CAAC;YAC7B,EAAE,GAAG,UAAU,CAAC;YAChB,4EAA4E;YAC5E,QAAQ,CAAC,GAAG,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC;QAC9B,CAAC;aAAM,IAAI,MAAM,CAAC,sBAAsB,EAAE,CAAC;YACzC,sDAAsD;YACtD,MAAM,MAAM,GAAG,QAAQ,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;YACvC,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;gBACzB,EAAE,GAAG,MAAM,CAAC;YACd,CAAC;iBAAM,CAAC;gBACN,EAAE,GAAG,MAAM,EAAE,CAAC;gBACd,QAAQ,CAAC,GAAG,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC;YAC9B,CAAC;QACH,CAAC;aAAM,CAAC;YACN,EAAE,GAAG,MAAM,EAAE,CAAC;QAChB,CAAC;QAED,eAAe,CAAC,IAAI,CAAC,EAAE,GAAG,KAAK,EAAE,EAAE,EAAE,CAAC,CAAC;IACzC,CAAC;IAED,gBAAgB;IAChB,MAAM,MAAM,GAAc,IAAI,GAAG,EAAE,CAAC;IACpC,KAAK,MAAM,MAAM,IAAI,eAAe,EAAE,CAAC;QACrC,MAAM,GAAG,GAAG,eAAe,CAAC,MAAM,CAAC,IAAI,EAAE,MAAM,CAAC,EAAE,CAAC,CAAC;QACpD,MAAM,CAAC,GAAG,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,CAAC,CAAC;IAC/B,CAAC;IAED,oDAAoD;IACpD,0DAA0D;IAC1D,MAAM,gBAAgB,GAAG,CAAC,GAAG,eAAe,CAAC,CAAC,IAAI,CAChD,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,KAAK,CAC5B,CAAC;IAEF,uBAAuB;IACvB,IAAI,cAAc,GAAG,IAAI,CAAC;IAC1B,KAAK,MAAM,MAAM,IAAI,gBAAgB,EAAE,CAAC;QACtC,MAAM,GAAG,GAAG,WAAW,CAAC,MAAM,CAAC,IAAI,EAAE,MAAM,CAAC,EAAE,EAAE,MAAM,CAAC,QAAQ,EAAE,SAAS,CAAC,CAAC;QAC5E,cAAc;YACZ,cAAc,CAAC,KAAK,CAAC,CAAC,EAAE,MAAM,CAAC,KAAK,CAAC;gBACrC,GAAG;gBACH,cAAc,CAAC,KAAK,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;IACrC,CAAC;IAED,iDAAiD;IACjD,MAAM,QAAQ,GAAqB,eAAe,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;QAC7D,IAAI,EAAE,CAAC,CAAC,IAAI;QACZ,EAAE,EAAE,CAAC,CAAC,EAAE;QACR,KAAK,EAAE,CAAC,CAAC,KAAK;QACd,GAAG,EAAE,CAAC,CAAC,GAAG;QACV,UAAU,EAAE,CAAC,CAAC,UAAU;QACxB,MAAM,EAAE,CAAC,CAAC,MAAM;QAChB,QAAQ,EAAE,CAAC,CAAC,IAAI;QAChB,QAAQ,EAAE,CAAC,CAAC,QAAQ;KACrB,CAAC,CAAC,CAAC;IAEJ,OAAO;QACL,cAAc;QACd,QAAQ,EAAE,mBAAmB,CAAC,QAAQ,CAAC;QACvC,MAAM;KACP,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,UAAU,CACxB,GAAW,EACX,YAAuB,kBAAkB;IAEzC,OAAO,QAAQ,CAAC,GAAG,EAAE,SAAS,CAAC,KAAK,IAAI,CAAC;AAC3C,CAAC;AAeD;;;;;;;;;;;;;;;;GAgBG;AACH,MAAM,WAAW,GAAG,mEAAmE,CAAC;AAExF;;GAEG;AACH,MAAM,WAAW,GAAG,8BAA8B,CAAC;AACnD,MAAM,oBAAoB,GAAG,8BAA8B,CAAC;AAE5D;;;;GAIG;AACH,SAAS,uBAAuB,CAAC,SAAoB;IACnD,MAAM,IAAI,GAAG,SAAS,CAAC,IAAI,CAAC;IAC5B,IAAI,IAAI,KAAK,GAAG,EAAE,CAAC;QACjB,OAAO,YAAY,CAAC;IACtB,CAAC;IACD,+EAA+E;IAC/E,2BAA2B;IAC3B,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACpB,OAAO,IAAI;aACR,KAAK,CAAC,EAAE,CAAC;aACT,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC;aAC7B,IAAI,CAAC,WAAW,CAAC,CAAC;IACvB,CAAC;IACD,OAAO,YAAY,CAAC,IAAI,CAAC,CAAC;AAC5B,CAAC;AAED;;;;GAIG;AACH,SAAS,mBAAmB,CAAC,SAAoB;IAC/C,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,CAAC;IAC9B,IAAI,KAAK,KAAK,IAAI,EAAE,CAAC;QACnB,wFAAwF;QACxF,MAAM,aAAa,GAAG,YAAY,CAAC;QACnC,OAAO,MAAM,WAAW,YAAY,WAAW,GAAG,aAAa,MAAM,aAAa,KAAK,CAAC;IAC1F,CAAC;IACD,4EAA4E;IAC5E,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACrB,MAAM,OAAO,GAAG,KAAK;aAClB,KAAK,CAAC,EAAE,CAAC;aACT,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC;aAC7B,IAAI,CAAC,WAAW,CAAC,CAAC;QACrB,OAAO,GAAG,WAAW,GAAG,OAAO,EAAE,CAAC;IACpC,CAAC;IACD,OAAO,GAAG,WAAW,GAAG,YAAY,CAAC,KAAK,CAAC,EAAE,CAAC;AAChD,CAAC;AAED;;;;GAIG;AACH,SAAS,uBAAuB,CAAC,SAAoB;IACnD,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,CAAC;IAC9B,IAAI,KAAK,KAAK,IAAI,EAAE,CAAC;QACnB,OAAO,sBAAsB,CAAC;IAChC,CAAC;IACD,wEAAwE;IACxE,mEAAmE;IACnE,kEAAkE;IAClE,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACrB,MAAM,OAAO,GAAG,KAAK;aAClB,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;aACZ,KAAK,CAAC,EAAE,CAAC;aACT,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,GAAG,YAAY,CAAC,EAAE,CAAC,GAAG,CAAC;aACnC,IAAI,CAAC,EAAE,CAAC,CAAC;QACZ,OAAO,MAAM,OAAO,IAAI,CAAC;IAC3B,CAAC;IACD,OAAO,EAAE,CAAC;AACZ,CAAC;AAED,gEAAgE;AAChE,MAAM,iBAAiB,GAAG,IAAI,GAAG,EAAoB,CAAC;AAEtD,SAAS,oBAAoB,CAAC,SAAoB;IAChD,OAAO,GAAG,SAAS,CAAC,IAAI,KAAK,SAAS,CAAC,KAAK,KAAK,SAAS,CAAC,OAAO,IAAI,KAAK,EAAE,CAAC;AAChF,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,qBAAqB,CAC5B,YAAuB,kBAAkB;IAEzC,MAAM,QAAQ,GAAG,oBAAoB,CAAC,SAAS,CAAC,CAAC;IACjD,MAAM,MAAM,GAAG,iBAAiB,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;IAC/C,IAAI,MAAM,KAAK,SAAS;QAAE,OAAO,MAAM,CAAC;IACxC,MAAM,OAAO,GAAG,SAAS,CAAC,OAAO,IAAI,KAAK,CAAC;IAC3C,MAAM,WAAW,GAAG,uBAAuB,CAAC,SAAS,CAAC,CAAC;IACvD,MAAM,cAAc,GAAG,mBAAmB,CAAC,SAAS,CAAC,CAAC;IACtD,MAAM,WAAW,GAAG,uBAAuB,CAAC,SAAS,CAAC,CAAC;IAEvD,2EAA2E;IAC3E,MAAM,QAAQ,GAAG,OAAO,WAAW,IAAI,WAAW,GAAG,WAAW,YAAY,WAAW,EAAE,CAAC;IAC1F,uEAAuE;IACvE,kFAAkF;IAClF,MAAM,MAAM,GAAG,KAAK,WAAW,IAAI,WAAW,GAAG,WAAW,IAAI,QAAQ,IAAI,WAAW,GAAG,WAAW,EAAE,CAAC;IACxG,4BAA4B;IAC5B,MAAM,UAAU,GAAG,MAAM,WAAW,SAAS,WAAW,IAAI,WAAW,GAAG,WAAW,SAAS,WAAW,IAAI,CAAC;IAC9G,2BAA2B;IAC3B,MAAM,SAAS,GAAG,MAAM,WAAW,QAAQ,WAAW,IAAI,WAAW,GAAG,WAAW,YAAY,WAAW,IAAI,CAAC;IAE/G,MAAM,cAAc,GAAG,YAAY,CAAC,OAAO,CAAC,CAAC;IAE7C,MAAM,QAAQ,GAAG;QACf,wCAAwC;QACxC,0CAA0C;QAC1C,IAAI,MAAM,CACR,GAAG,WAAW,GAAG,WAAW,GAAG,cAAc,GAAG,oBAAoB,GAAG,QAAQ,GAAG,UAAU,GAAG,SAAS,GAAG,oBAAoB,GAAG,MAAM,GAAG,cAAc,EAAE,EAC3J,IAAI,CACL;QACD,WAAW;QACX,uBAAuB;QACvB,IAAI,MAAM,CACR,GAAG,WAAW,GAAG,WAAW,GAAG,cAAc,GAAG,oBAAoB,GAAG,MAAM,GAAG,oBAAoB,GAAG,QAAQ,GAAG,cAAc,EAAE,EAClI,IAAI,CACL;KACF,CAAC;IAEF,iBAAiB,CAAC,GAAG,CAAC,QAAQ,EAAE,QAAQ,CAAC,CAAC;IAC1C,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED;;;;;;;;;;GAUG;AACH,MAAM,UAAU,WAAW,CACzB,cAAsB,EACtB,YAAuB,kBAAkB;IAEzC,MAAM,IAAI,GAAmB,EAAE,CAAC;IAChC,MAAM,QAAQ,GAAG,qBAAqB,CAAC,SAAS,CAAC,CAAC;IAElD,sFAAsF;IACtF,MAAM,gBAAgB,GAAG,IAAI,GAAG,EAAU,CAAC;IAE3C,KAAK,IAAI,YAAY,GAAG,CAAC,EAAE,YAAY,GAAG,QAAQ,CAAC,MAAM,EAAE,YAAY,EAAE,EAAE,CAAC;QAC1E,MAAM,OAAO,GAAG,QAAQ,CAAC,YAAY,CAAC,CAAC;QACvC,IAAI,OAAO,KAAK,SAAS;YAAE,SAAS;QAEpC,IAAI,KAA6B,CAAC;QAClC,mCAAmC;QACnC,OAAO,CAAC,SAAS,GAAG,CAAC,CAAC;QAEtB,OAAO,CAAC,KAAK,GAAG,OAAO,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;YACvD,IAAI,gBAAgB,CAAC,GAAG,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,CAAC;gBACtC,SAAS,CAAC,4CAA4C;YACxD,CAAC;YAED,2EAA2E;YAC3E,6FAA6F;YAC7F,6CAA6C;YAC7C,IAAI,OAA2B,CAAC;YAChC,IAAI,KAAyB,CAAC;YAC9B,IAAI,SAA6B,CAAC;YAClC,IAAI,QAA4B,CAAC;YAEjC,IAAI,YAAY,KAAK,CAAC,EAAE,CAAC;gBACvB,OAAO,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;gBACnB,SAAS,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;gBACrB,QAAQ,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;gBACpB,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;YACnB,CAAC;iBAAM,CAAC;gBACN,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;gBACjB,OAAO,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;YACrB,CAAC;YAED,IAAI,OAAO,KAAK,SAAS,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;gBACjD,6DAA6D;gBAC7D,oEAAoE;gBACpE,2DAA2D;gBAC3D,MAAM,IAAI,GAAG,OAAO,CAAC,WAAW,EAAa,CAAC;gBAC9C,uEAAuE;gBACvE,2CAA2C;gBAC3C,MAAM,EAAE,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC;gBAE7B,uCAAuC;gBACvC,IAAI,QAAwC,CAAC;gBAC7C,IACE,CAAC,SAAS,KAAK,SAAS,IAAI,SAAS,KAAK,EAAE,CAAC;oBAC7C,CAAC,QAAQ,KAAK,SAAS,IAAI,QAAQ,KAAK,EAAE,CAAC,EAC3C,CAAC;oBACD,QAAQ,GAAG,EAAE,CAAC;oBACd,IACE,SAAS,KAAK,SAAS;wBACvB,SAAS,KAAK,EAAE;wBAChB,CAAC,MAAM,EAAE,QAAQ,EAAE,SAAS,EAAE,SAAS,CAAC,CAAC,QAAQ,CAC/C,SAAS,CAAC,WAAW,EAAE,CACxB,EACD,CAAC;wBACD,QAAQ,CAAC,MAAM;4BACb,SAAS,CAAC,WAAW,EAAkC,CAAC;oBAC5D,CAAC;oBACD,IACE,QAAQ,KAAK,SAAS;wBACtB,QAAQ,KAAK,EAAE;wBACf,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,EAAE,cAAc,EAAE,SAAS,CAAC,CAAC,QAAQ,CAC/D,QAAQ,CAAC,WAAW,EAAE,CACvB,EACD,CAAC;wBACD,QAAQ,CAAC,KAAK;4BACZ,QAAQ,CAAC,WAAW,EAAiC,CAAC;oBAC1D,CAAC;gBACH,CAAC;gBAED,IAAI,CAAC,IAAI,CAAC;oBACR,IAAI;oBACJ,EAAE;oBACF,QAAQ,EAAE,KAAK,CAAC,KAAK;oBACrB,WAAW,EAAE,KAAK,CAAC,CAAC,CAAC;oBACrB,QAAQ;iBACT,CAAC,CAAC;gBACH,gBAAgB,CAAC,GAAG,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;YACpC,CAAC;QACH,CAAC;IACH,CAAC;IAED,6BAA6B;IAC7B,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,GAAG,CAAC,CAAC,QAAQ,CAAC,CAAC;IAE7C,OAAO,IAAI,CAAC;AACd,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,iBAAiB,CAC/B,cAAsB,EACtB,YAAuB,kBAAkB;IAEzC,MAAM,IAAI,GAAmB,EAAE,CAAC;IAChC,MAAM,OAAO,GAAG,SAAS,CAAC,OAAO,IAAI,KAAK,CAAC;IAC3C,MAAM,IAAI,GAAG,YAAY,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC;IAC1C,MAAM,KAAK,GAAG,YAAY,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;IAC5C,+EAA+E;IAC/E,MAAM,UAAU,GAAG,IAAI,MAAM,CAC3B,GAAG,IAAI,GAAG,YAAY,CAAC,OAAO,CAAC,kFAAkF,QAAQ,SAAS,KAAK,EAAE,EACzI,GAAG,CACJ,CAAC;IAEF,IAAI,KAA6B,CAAC;IAClC,OAAO,CAAC,KAAK,GAAG,UAAU,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;QAC1D,MAAM,OAAO,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;QACzB,MAAM,SAAS,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;QAC3B,MAAM,QAAQ,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;QAC1B,MAAM,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;QAEvB,IAAI,OAAO,KAAK,SAAS,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;YACjD,2DAA2D;YAC3D,MAAM,IAAI,GAAG,OAAkB,CAAC;YAChC,MAAM,EAAE,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC;YAE7B,uCAAuC;YACvC,IAAI,QAAwC,CAAC;YAC7C,IACE,CAAC,SAAS,KAAK,SAAS,IAAI,SAAS,KAAK,EAAE,CAAC;gBAC7C,CAAC,QAAQ,KAAK,SAAS,IAAI,QAAQ,KAAK,EAAE,CAAC,EAC3C,CAAC;gBACD,QAAQ,GAAG,EAAE,CAAC;gBACd,IACE,SAAS,KAAK,SAAS;oBACvB,SAAS,KAAK,EAAE;oBAChB,CAAC,MAAM,EAAE,QAAQ,EAAE,SAAS,EAAE,SAAS,CAAC,CAAC,QAAQ,CAAC,SAAS,CAAC,EAC5D,CAAC;oBACD,QAAQ,CAAC,MAAM,GAAG,SAAyC,CAAC;gBAC9D,CAAC;gBACD,IACE,QAAQ,KAAK,SAAS;oBACtB,QAAQ,KAAK,EAAE;oBACf,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,EAAE,cAAc,EAAE,SAAS,CAAC,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAC3E,CAAC;oBACD,QAAQ,CAAC,KAAK,GAAG,QAAuC,CAAC;gBAC3D,CAAC;YACH,CAAC;YAED,IAAI,CAAC,IAAI,CAAC;gBACR,IAAI;gBACJ,EAAE;gBACF,QAAQ,EAAE,KAAK,CAAC,KAAK;gBACrB,WAAW,EAAE,KAAK,CAAC,CAAC,CAAC;gBACrB,QAAQ;aACT,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAED,OAAO,IAAI,CAAC;AACd,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,mBAAmB,CACjC,QAA0B;IAE1B,MAAM,MAAM,GAA4B,EAA6B,CAAC;IAEtE,4BAA4B;IAC5B,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,MAAM,CAAC,OAAO,CAAC,EAAE,CAAC;QAC1C,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IACnB,CAAC;IAED,iBAAiB;IACjB,KAAK,MAAM,MAAM,IAAI,QAAQ,EAAE,CAAC;QAC9B,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC;IACvD,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,UAAU,SAAS,CACvB,cAAsB,EACtB,MAAiB,EACjB,SAAkB,KAAK,EACvB,YAAuB,kBAAkB;IAEzC,IAAI,MAAM,GAAG,cAAc,CAAC;IAC5B,MAAM,IAAI,GAAG,MAAM;QACjB,CAAC,CAAC,iBAAiB,CAAC,cAAc,EAAE,SAAS,CAAC;QAC9C,CAAC,CAAC,WAAW,CAAC,cAAc,EAAE,SAAS,CAAC,CAAC;IAE3C,8CAA8C;IAC9C,0DAA0D;IAC1D,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,GAAG,CAAC,CAAC,QAAQ,CAAC,CAAC;IAE7C,KAAK,MAAM,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,WAAW,EAAE,IAAI,IAAI,EAAE,CAAC;QACvD,MAAM,GAAG,GAAG,eAAe,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;QACtC,MAAM,QAAQ,GAAG,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QAEjC,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC3B,qDAAqD;YACrD,6EAA6E;YAC7E,MAAM;gBACJ,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC;oBACzB,QAAQ;oBACR,MAAM,CAAC,KAAK,CAAC,QAAQ,GAAG,WAAW,CAAC,MAAM,CAAC,CAAC;QAChD,CAAC;IACH,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC"}

package/dist/pipeline/title-extractor.d.ts

@@ -0,0 +1,79 @@
+/**
+ * Title Extractor
+ * Extracts and strips honorific titles/prefixes from PERSON entities
+ * so that titles remain visible in anonymized text for translation.
+ *
+ * Supported languages: ar, de, en, es, fr, it, lv, nl, pt, zh
+ */
+import { SpanMatch, SemanticAttributes } from "../types/index.js";
+/**
+ * Title extraction result
+ */
+export interface TitleExtractionResult {
+    /** The extracted title (e.g., "Dr.", "Mr.") or undefined if no title */
+    title?: string;
+    /** The name without the title */
+    nameWithoutTitle: string;
+    /** Character offset where the name starts (after title + space) */
+    titleLength: number;
+}
+/**
+ * Extracts a title from the beginning of a name
+ *
+ * @param name - Full name potentially starting with a title
+ * @returns Extraction result with title, remaining name, and offset
+ *
+ * @example
+ * extractTitle("Dr. John Smith") // { title: "Dr.", nameWithoutTitle: "John Smith", titleLength: 4 }
+ * extractTitle("John Smith") // { title: undefined, nameWithoutTitle: "John Smith", titleLength: 0 }
+ */
+export declare function extractTitle(name: string): TitleExtractionResult;
+/**
+ * Extended semantic attributes including title
+ */
+export interface SemanticAttributesWithTitle extends SemanticAttributes {
+    /** Extracted title prefix (e.g., "Dr.", "Mrs.") */
+    title?: string;
+}
+/**
+ * Processes PERSON spans to extract titles
+ * Titles are removed from the span and stored in semantic attributes
+ * The span boundaries are adjusted so the title remains visible
+ *
+ * @param spans - Array of detected PII spans
+ * @param originalText - The original text (needed to verify span boundaries)
+ * @returns Array of spans with titles extracted from PERSON entities
+ */
+export declare function extractTitlesFromSpans(spans: SpanMatch[], originalText: string): SpanMatch[];
+/**
+ * Gets all supported titles for a specific language
+ */
+export declare function getTitlesForLanguage(langCode: "ar" | "de" | "en" | "es" | "fr" | "it" | "lv" | "nl" | "pt" | "zh"): string[];
+/**
+ * Gets all supported titles across all languages
+ */
+export declare function getAllTitles(): string[];
+/**
+ * Checks if a string starts with a known title
+ */
+export declare function startsWithTitle(text: string): boolean;
+/**
+ * Checks if a text consists entirely of a title (with optional punctuation)
+ */
+export declare function isOnlyTitle(text: string): boolean;
+/**
+ * Merges adjacent PERSON spans when one is a title
+ *
+ * This fixes issues where NER models split "Mrs. Smith" into two entities:
+ * - PERSON: "Mrs" (or "Mrs.")
+ * - PERSON: "Smith"
+ *
+ * After merging: PERSON: "Mrs. Smith"
+ *
+ * @param spans - Array of detected PII spans
+ * @param originalText - The original text
+ * @param maxGap - Maximum characters between spans to consider them adjacent (default: 3)
+ * @returns Array of spans with adjacent title+name PERSON entities merged
+ */
+export declare function mergeAdjacentTitleSpans(spans: SpanMatch[], originalText: string, maxGap?: number): SpanMatch[];
+//# sourceMappingURL=title-extractor.d.ts.map

package/dist/pipeline/title-extractor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"title-extractor.d.ts","sourceRoot":"","sources":["../../src/pipeline/title-extractor.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH,OAAO,EACL,SAAS,EAET,kBAAkB,EAEnB,MAAM,mBAAmB,CAAC;AAE3B;;GAEG;AACH,MAAM,WAAW,qBAAqB;IACpC,wEAAwE;IACxE,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,iCAAiC;IACjC,gBAAgB,EAAE,MAAM,CAAC;IACzB,mEAAmE;IACnE,WAAW,EAAE,MAAM,CAAC;CACrB;AAqjBD;;;;;;;;;GASG;AACH,wBAAgB,YAAY,CAAC,IAAI,EAAE,MAAM,GAAG,qBAAqB,CAyBhE;AAED;;GAEG;AACH,MAAM,WAAW,2BAA4B,SAAQ,kBAAkB;IACrE,mDAAmD;IACnD,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED;;;;;;;;GAQG;AACH,wBAAgB,sBAAsB,CACpC,KAAK,EAAE,SAAS,EAAE,EAClB,YAAY,EAAE,MAAM,GACnB,SAAS,EAAE,CAiEb;AAED;;GAEG;AACH,wBAAgB,oBAAoB,CAClC,QAAQ,EAAE,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,GAC5E,MAAM,EAAE,CAcV;AAED;;GAEG;AACH,wBAAgB,YAAY,IAAI,MAAM,EAAE,CAEvC;AAED;;GAEG;AACH,wBAAgB,eAAe,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAGrD;AAED;;GAEG;AACH,wBAAgB,WAAW,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CA0BjD;AAED;;;;;;;;;;;;;GAaG;AACH,wBAAgB,uBAAuB,CACrC,KAAK,EAAE,SAAS,EAAE,EAClB,YAAY,EAAE,MAAM,EACpB,MAAM,GAAE,MAAU,GACjB,SAAS,EAAE,CAuFb"}