@karmaniverous/jeeves-watcher 0.2.4 → 0.2.6

package/dist/index.d.ts

@@ -15,6 +15,7 @@ declare const watchConfigSchema: z.ZodObject<{
     usePolling: z.ZodOptional<z.ZodBoolean>;
     debounceMs: z.ZodOptional<z.ZodNumber>;
     stabilityThresholdMs: z.ZodOptional<z.ZodNumber>;
+    respectGitignore: z.ZodOptional<z.ZodBoolean>;
 }, z.core.$strip>;
 /** Watch configuration for file system monitoring paths, ignore patterns, and debounce/stability settings. */
 type WatchConfig = z.infer<typeof watchConfigSchema>;
@@ -91,6 +92,7 @@ declare const jeevesWatcherConfigSchema: z.ZodObject<{
         usePolling: z.ZodOptional<z.ZodBoolean>;
         debounceMs: z.ZodOptional<z.ZodNumber>;
         stabilityThresholdMs: z.ZodOptional<z.ZodNumber>;
+        respectGitignore: z.ZodOptional<z.ZodBoolean>;
     }, z.core.$strip>;
     configWatch: z.ZodOptional<z.ZodObject<{
         enabled: z.ZodOptional<z.ZodBoolean>;
@@ -543,6 +545,42 @@ declare function createApiServer(options: ApiServerOptions): FastifyInstance;
  */
 declare function createLogger(config?: LoggingConfig): pino.Logger;
 
+/**
+ * @module gitignore
+ * Processor-level gitignore filtering. Scans watched paths for `.gitignore` files in git repos, caches parsed patterns, and exposes `isIgnored()` for path checking.
+ */
+/**
+ * Processor-level gitignore filter. Checks file paths against the nearest
+ * `.gitignore` chain in git repositories.
+ */
+declare class GitignoreFilter {
+    private repos;
+    /**
+     * Create a GitignoreFilter by scanning watched paths for `.gitignore` files.
+     *
+     * @param watchPaths - Absolute paths being watched (directories or globs resolved to roots).
+     */
+    constructor(watchPaths: string[]);
+    /**
+     * Scan paths for git repos and their `.gitignore` files.
+     */
+    private scan;
+    /**
+     * Check whether a file path is ignored by any applicable `.gitignore`.
+     *
+     * @param filePath - Absolute file path to check.
+     * @returns `true` if the file should be ignored.
+     */
+    isIgnored(filePath: string): boolean;
+    /**
+     * Invalidate and re-parse a specific `.gitignore` file.
+     * Call when a `.gitignore` file is added, changed, or removed.
+     *
+     * @param gitignorePath - Absolute path to the `.gitignore` file that changed.
+     */
+    invalidate(gitignorePath: string): void;
+}
+
 /**
  * @module health
  * Tracks consecutive system-level failures and applies exponential backoff.
@@ -612,6 +650,8 @@ interface FileSystemWatcherOptions {
     maxBackoffMs?: number;
     /** Callback invoked on unrecoverable system error. If not set, throws. */
     onFatalError?: (error: unknown) => void;
+    /** Optional gitignore filter for processor-level filtering. */
+    gitignoreFilter?: GitignoreFilter;
 }
 /**
  * Filesystem watcher that maps chokidar events to the processing queue.
@@ -622,6 +662,7 @@ declare class FileSystemWatcher {
     private readonly processor;
     private readonly logger;
     private readonly health;
+    private readonly gitignoreFilter?;
     private watcher;
     /**
      * Create a new FileSystemWatcher.
@@ -645,6 +686,14 @@ declare class FileSystemWatcher {
      * Get the system health tracker.
      */
     get systemHealth(): SystemHealth;
+    /**
+     * Check if a path is gitignored and should be skipped.
+     */
+    private isGitignored;
+    /**
+     * If the changed file is a `.gitignore`, invalidate the filter cache.
+     */
+    private handleGitignoreChange;
     /**
      * Wrap a processing operation with health tracking.
      * On success, resets the failure counter.
@@ -812,5 +861,5 @@ declare function deleteMetadata(filePath: string, metadataDir: string): Promise<
  */
 declare function pointId(filePath: string, chunkIndex?: number): string;
 
-export { DocumentProcessor, EventQueue, FileSystemWatcher, JeevesWatcher, SystemHealth, VectorStoreClient, apiConfigSchema, applyRules, buildAttributes, compileRules, configWatchConfigSchema, contentHash, createApiServer, createEmbeddingProvider, createLogger, deleteMetadata, embeddingConfigSchema, extractText, inferenceRuleSchema, jeevesWatcherConfigSchema, loadConfig, loggingConfigSchema, metadataPath, pointId, readMetadata, startFromConfig, vectorStoreConfigSchema, watchConfigSchema, writeMetadata };
+export { DocumentProcessor, EventQueue, FileSystemWatcher, GitignoreFilter, JeevesWatcher, SystemHealth, VectorStoreClient, apiConfigSchema, applyRules, buildAttributes, compileRules, configWatchConfigSchema, contentHash, createApiServer, createEmbeddingProvider, createLogger, deleteMetadata, embeddingConfigSchema, extractText, inferenceRuleSchema, jeevesWatcherConfigSchema, loadConfig, loggingConfigSchema, metadataPath, pointId, readMetadata, startFromConfig, vectorStoreConfigSchema, watchConfigSchema, writeMetadata };
 export type { ApiConfig, ApiServerOptions, CompiledRule, ConfigWatchConfig, EmbeddingConfig, EmbeddingProvider, EventQueueOptions, ExtractedText, FileAttributes, FileSystemWatcherOptions, InferenceRule, JeevesWatcherConfig, JeevesWatcherFactories, JeevesWatcherRuntimeOptions, LoggingConfig, ProcessFn, ProcessorConfig, RuleLogger, ScrolledPoint, SearchResult, SystemHealthOptions, VectorPoint, VectorStoreConfig, WatchConfig, WatchEvent };

package/dist/index.iife.js

@@ -1,4 +1,4 @@
-(function (exports, Fastify, promises, node_path, picomatch, radash, node_crypto, cosmiconfig, zod, jsonmap, googleGenai, pino, uuid, cheerio, yaml, mammoth, Ajv, addFormats, textsplitters, jsClientRest, chokidar) {
+(function (exports, Fastify, promises, node_path, picomatch, radash, node_crypto, cosmiconfig, zod, jsonmap, googleGenai, node_fs, ignore, pino, uuid, cheerio, yaml, mammoth, Ajv, addFormats, textsplitters, jsClientRest, chokidar) {
     'use strict';
 
     function _interopNamespaceDefault(e) {
@@ -434,6 +434,7 @@
         stabilityThresholdMs: 500,
         usePolling: false,
         pollIntervalMs: 1000,
+        respectGitignore: true,
     };
     /** Default embedding configuration. */
     const EMBEDDING_DEFAULTS = {
@@ -478,6 +479,11 @@
             .number()
             .optional()
             .describe('Time in milliseconds a file must remain unchanged before processing.'),
+        /** Whether to respect .gitignore files when processing. */
+        respectGitignore: zod.z
+            .boolean()
+            .optional()
+            .describe('Skip files ignored by .gitignore in git repositories. Only applies to repos with a .git directory. Default: true.'),
     });
     /**
      * Configuration watch settings.
@@ -943,6 +949,212 @@
         return factory(config, logger);
     }
 
+    /**
+     * @module gitignore
+     * Processor-level gitignore filtering. Scans watched paths for `.gitignore` files in git repos, caches parsed patterns, and exposes `isIgnored()` for path checking.
+     */
+    /**
+     * Find the git repo root by walking up from `startDir` looking for `.git/`.
+     * Returns `undefined` if no repo is found.
+     */
+    function findRepoRoot(startDir) {
+        let dir = node_path.resolve(startDir);
+        const root = node_path.resolve('/');
+        while (dir !== root) {
+            if (node_fs.existsSync(node_path.join(dir, '.git')) &&
+                node_fs.statSync(node_path.join(dir, '.git')).isDirectory()) {
+                return dir;
+            }
+            const parent = node_path.dirname(dir);
+            if (parent === dir)
+                break;
+            dir = parent;
+        }
+        return undefined;
+    }
+    /**
+     * Convert a watch path (directory, file path, or glob) to a concrete directory
+     * that can be scanned for a repo root.
+     */
+    function watchPathToScanDir(watchPath) {
+        const absPath = node_path.resolve(watchPath);
+        try {
+            return node_fs.statSync(absPath).isDirectory() ? absPath : node_path.dirname(absPath);
+        }
+        catch {
+            // ignore
+        }
+        // If this is a glob, fall back to the non-glob prefix.
+        const globMatch = /[*?[{]/.exec(watchPath);
+        if (!globMatch)
+            return undefined;
+        const prefix = watchPath.slice(0, globMatch.index);
+        const trimmed = prefix.trim();
+        const baseDir = trimmed.length === 0
+            ? '.'
+            : trimmed.endsWith('/') || trimmed.endsWith('\\')
+                ? trimmed
+                : node_path.dirname(trimmed);
+        const resolved = node_path.resolve(baseDir);
+        if (!node_fs.existsSync(resolved))
+            return undefined;
+        return resolved;
+    }
+    /**
+     * Recursively find all `.gitignore` files under `dir`.
+     * Skips `.git` and `node_modules` directories for performance.
+     */
+    function findGitignoreFiles(dir) {
+        const results = [];
+        const gitignorePath = node_path.join(dir, '.gitignore');
+        if (node_fs.existsSync(gitignorePath)) {
+            results.push(gitignorePath);
+        }
+        let entries;
+        try {
+            entries = node_fs.readdirSync(dir);
+        }
+        catch {
+            return results;
+        }
+        for (const entry of entries) {
+            if (entry === '.git' || entry === 'node_modules')
+                continue;
+            const fullPath = node_path.join(dir, entry);
+            try {
+                if (node_fs.statSync(fullPath).isDirectory()) {
+                    results.push(...findGitignoreFiles(fullPath));
+                }
+            }
+            catch {
+                // Skip inaccessible entries
+            }
+        }
+        return results;
+    }
+    /**
+     * Parse a `.gitignore` file into an `ignore` instance.
+     */
+    function parseGitignore(gitignorePath) {
+        const content = node_fs.readFileSync(gitignorePath, 'utf8');
+        return ignore().add(content);
+    }
+    /**
+     * Normalize a path to use forward slashes (required by `ignore` package).
+     */
+    function toForwardSlash(p) {
+        return p.replace(/\\/g, '/');
+    }
+    /**
+     * Processor-level gitignore filter. Checks file paths against the nearest
+     * `.gitignore` chain in git repositories.
+     */
+    class GitignoreFilter {
+        repos = new Map();
+        /**
+         * Create a GitignoreFilter by scanning watched paths for `.gitignore` files.
+         *
+         * @param watchPaths - Absolute paths being watched (directories or globs resolved to roots).
+         */
+        constructor(watchPaths) {
+            this.scan(watchPaths);
+        }
+        /**
+         * Scan paths for git repos and their `.gitignore` files.
+         */
+        scan(watchPaths) {
+            this.repos.clear();
+            const scannedDirs = new Set();
+            for (const watchPath of watchPaths) {
+                const scanDir = watchPathToScanDir(watchPath);
+                if (!scanDir)
+                    continue;
+                if (scannedDirs.has(scanDir))
+                    continue;
+                scannedDirs.add(scanDir);
+                const repoRoot = findRepoRoot(scanDir);
+                if (!repoRoot)
+                    continue;
+                if (this.repos.has(repoRoot))
+                    continue;
+                const gitignoreFiles = findGitignoreFiles(repoRoot);
+                const entries = gitignoreFiles.map((gf) => ({
+                    dir: node_path.dirname(gf),
+                    ig: parseGitignore(gf),
+                }));
+                // Sort deepest-first so nested `.gitignore` files are checked first
+                entries.sort((a, b) => b.dir.length - a.dir.length);
+                this.repos.set(repoRoot, { root: repoRoot, entries });
+            }
+        }
+        /**
+         * Check whether a file path is ignored by any applicable `.gitignore`.
+         *
+         * @param filePath - Absolute file path to check.
+         * @returns `true` if the file should be ignored.
+         */
+        isIgnored(filePath) {
+            const absPath = node_path.resolve(filePath);
+            for (const [, repo] of this.repos) {
+                // Check if file is within this repo
+                const relToRepo = node_path.relative(repo.root, absPath);
+                if (relToRepo.startsWith('..') || relToRepo.startsWith(node_path.resolve('/'))) {
+                    continue;
+                }
+                // Check each `.gitignore` entry (deepest-first)
+                for (const entry of repo.entries) {
+                    const relToEntry = node_path.relative(entry.dir, absPath);
+                    if (relToEntry.startsWith('..'))
+                        continue;
+                    const normalized = toForwardSlash(relToEntry);
+                    if (entry.ig.ignores(normalized)) {
+                        return true;
+                    }
+                }
+            }
+            return false;
+        }
+        /**
+         * Invalidate and re-parse a specific `.gitignore` file.
+         * Call when a `.gitignore` file is added, changed, or removed.
+         *
+         * @param gitignorePath - Absolute path to the `.gitignore` file that changed.
+         */
+        invalidate(gitignorePath) {
+            const absPath = node_path.resolve(gitignorePath);
+            const gitignoreDir = node_path.dirname(absPath);
+            for (const [, repo] of this.repos) {
+                const relToRepo = node_path.relative(repo.root, gitignoreDir);
+                if (relToRepo.startsWith('..'))
+                    continue;
+                // Remove old entry for this directory
+                repo.entries = repo.entries.filter((e) => e.dir !== gitignoreDir);
+                // Re-parse if file still exists
+                if (node_fs.existsSync(absPath)) {
+                    repo.entries.push({ dir: gitignoreDir, ig: parseGitignore(absPath) });
+                    // Re-sort deepest-first
+                    repo.entries.sort((a, b) => b.dir.length - a.dir.length);
+                }
+                return;
+            }
+            // If not in any known repo, check if it's in a repo we haven't scanned
+            const repoRoot = findRepoRoot(gitignoreDir);
+            if (repoRoot && node_fs.existsSync(absPath)) {
+                const entries = [
+                    { dir: gitignoreDir, ig: parseGitignore(absPath) },
+                ];
+                if (this.repos.has(repoRoot)) {
+                    const repo = this.repos.get(repoRoot);
+                    repo.entries.push(entries[0]);
+                    repo.entries.sort((a, b) => b.dir.length - a.dir.length);
+                }
+                else {
+                    this.repos.set(repoRoot, { root: repoRoot, entries });
+                }
+            }
+        }
+    }
+
     /**
      * @module logger
      * Creates pino logger instances. I/O: optionally writes logs to file via pino/file transport. Defaults to stdout at info level.
@@ -2034,6 +2246,7 @@
         processor;
         logger;
         health;
+        gitignoreFilter;
         watcher;
         /**
          * Create a new FileSystemWatcher.
@@ -2049,6 +2262,7 @@
             this.queue = queue;
             this.processor = processor;
             this.logger = logger;
+            this.gitignoreFilter = options.gitignoreFilter;
             const healthOptions = {
                 maxRetries: options.maxRetries,
                 maxBackoffMs: options.maxBackoffMs,
@@ -2071,14 +2285,23 @@
                 ignoreInitial: false,
             });
             this.watcher.on('add', (path) => {
+                this.handleGitignoreChange(path);
+                if (this.isGitignored(path))
+                    return;
                 this.logger.debug({ path }, 'File added');
                 this.queue.enqueue({ type: 'create', path, priority: 'normal' }, () => this.wrapProcessing(() => this.processor.processFile(path)));
             });
             this.watcher.on('change', (path) => {
+                this.handleGitignoreChange(path);
+                if (this.isGitignored(path))
+                    return;
                 this.logger.debug({ path }, 'File changed');
                 this.queue.enqueue({ type: 'modify', path, priority: 'normal' }, () => this.wrapProcessing(() => this.processor.processFile(path)));
             });
             this.watcher.on('unlink', (path) => {
+                this.handleGitignoreChange(path);
+                if (this.isGitignored(path))
+                    return;
                 this.logger.debug({ path }, 'File removed');
                 this.queue.enqueue({ type: 'delete', path, priority: 'normal' }, () => this.wrapProcessing(() => this.processor.deleteFile(path)));
             });
@@ -2105,6 +2328,29 @@
         get systemHealth() {
             return this.health;
         }
+        /**
+         * Check if a path is gitignored and should be skipped.
+         */
+        isGitignored(path) {
+            if (!this.gitignoreFilter)
+                return false;
+            const ignored = this.gitignoreFilter.isIgnored(path);
+            if (ignored) {
+                this.logger.debug({ path }, 'Skipping gitignored file');
+            }
+            return ignored;
+        }
+        /**
+         * If the changed file is a `.gitignore`, invalidate the filter cache.
+         */
+        handleGitignoreChange(path) {
+            if (!this.gitignoreFilter)
+                return;
+            if (path.endsWith('.gitignore')) {
+                this.logger.info({ path }, 'Gitignore file changed, refreshing filter');
+                this.gitignoreFilter.invalidate(path);
+            }
+        }
         /**
          * Wrap a processing operation with health tracking.
          * On success, resets the failure counter.
@@ -2260,10 +2506,15 @@
                 rateLimitPerMinute: this.config.embedding.rateLimitPerMinute,
             });
             this.queue = queue;
+            const respectGitignore = this.config.watch.respectGitignore ?? true;
+            const gitignoreFilter = respectGitignore
+                ? new GitignoreFilter(this.config.watch.paths)
+                : undefined;
             const watcher = this.factories.createFileSystemWatcher(this.config.watch, queue, processor, logger, {
                 maxRetries: this.config.maxRetries,
                 maxBackoffMs: this.config.maxBackoffMs,
                 onFatalError: this.runtimeOptions.onFatalError,
+                gitignoreFilter,
             });
             this.watcher = watcher;
             const server = this.factories.createApiServer({
@@ -2372,6 +2623,7 @@
     exports.DocumentProcessor = DocumentProcessor;
     exports.EventQueue = EventQueue;
     exports.FileSystemWatcher = FileSystemWatcher;
+    exports.GitignoreFilter = GitignoreFilter;
     exports.JeevesWatcher = JeevesWatcher;
     exports.SystemHealth = SystemHealth;
     exports.VectorStoreClient = VectorStoreClient;
@@ -2399,4 +2651,4 @@
     exports.watchConfigSchema = watchConfigSchema;
     exports.writeMetadata = writeMetadata;
 
-})(this["jeeves-watcher"] = this["jeeves-watcher"] || {}, Fastify, promises, node_path, picomatch, radash, node_crypto, cosmiconfig, zod, jsonmap, googleGenai, pino, uuid, cheerio, yaml, mammoth, Ajv, addFormats, textsplitters, jsClientRest, chokidar);
+})(this["jeeves-watcher"] = this["jeeves-watcher"] || {}, Fastify, promises, node_path, picomatch, radash, node_crypto, cosmiconfig, zod, jsonmap, googleGenai, node_fs, ignore, pino, uuid, cheerio, yaml, mammoth, Ajv, addFormats, textsplitters, jsClientRest, chokidar);

package/dist/index.iife.min.js

@@ -1 +1 @@
-!function(e,t,r,i,o,n,s,a,c,l,h,u,d,f,g,m,p,y,w,b,v){"use strict";function M(e){var t=Object.create(null);return e&&Object.keys(e).forEach((function(r){if("default"!==r){var i=Object.getOwnPropertyDescriptor(e,r);Object.defineProperty(t,r,i.get?i:{enumerable:!0,get:function(){return e[r]}})}})),t.default=e,Object.freeze(t)}var x=M(f);function P(e){if(e instanceof Error)return e;if("string"==typeof e)return new Error(e);const t=String("object"==typeof e&&null!==e&&"message"in e?e.message:e),r=new Error(t);return r.cause=e,r}function k(e){const t=e.replace(/\\/g,"/"),r=t.search(/[*?\[]/);if(-1===r)return i.resolve(e);const o=t.slice(0,r),n=o.endsWith("/")?o.slice(0,-1):i.dirname(o);return i.resolve(n)}async function*S(e){let t;try{t=(await r.readdir(e,{withFileTypes:!0})).map((e=>({name:e.name,isDirectory:e.isDirectory()})))}catch{return}for(const o of t){const t=i.resolve(e,o.name);if(o.isDirectory)yield*S(t);else try{(await r.stat(t)).isFile()&&(yield t)}catch{}}}async function F(e,t,r,i){const n=await async function(e,t=[]){const r=e.map((e=>e.replace(/\\/g,"/"))),i=t.map((e=>e.replace(/\\/g,"/"))),n=o(r,{dot:!0}),s=i.length?o(i,{dot:!0}):()=>!1,a=Array.from(new Set(e.map(k))),c=new Set;for(const e of a)for await(const t of S(e)){const e=t.replace(/\\/g,"/");s(e)||n(e)&&c.add(t)}return Array.from(c)}(e,t);for(const e of n)await r[i](e);return n.length}function z(e,t=!1){let r=e.replace(/\\/g,"/").toLowerCase();return t&&(r=r.replace(/^([a-z]):/,((e,t)=>t))),r}function j(e,t){const r=z(e,!0),o=s.createHash("sha256").update(r,"utf8").digest("hex");return i.join(t,`${o}.meta.json`)}async function R(e,t){try{const i=await r.readFile(j(e,t),"utf8");return JSON.parse(i)}catch{return null}}async function E(e,t,o){const n=j(e,t);await r.mkdir(i.dirname(n),{recursive:!0}),await r.writeFile(n,JSON.stringify(o,null,2),"utf8")}async function C(e,t){try{await r.rm(j(e,t))}catch{}}const T=["file_path","chunk_index","total_chunks","content_hash","chunk_text"];function I(e){const{processor:r,vectorStore:i,embeddingProvider:o,logger:s,config:a}=e,c=t({logger:!1});var l;return c.get("/status",(()=>({status:"ok",uptime:process.uptime()}))),c.post("/metadata",(l={processor:r,logger:s},async(e,t)=>{try{const{path:t,metadata:r}=e.body;return await l.processor.processMetadataUpdate(t,r),{ok:!0}}catch(e){return l.logger.error({err:P(e)},"Metadata update failed"),t.status(500).send({error:"Internal server error"})}})),c.post("/search",function(e){return async(t,r)=>{try{const{query:r,limit:i=10}=t.body,o=await e.embeddingProvider.embed([r]);return await e.vectorStore.search(o[0],i)}catch(t){return e.logger.error({err:P(t)},"Search failed"),r.status(500).send({error:"Internal server error"})}}}({embeddingProvider:o,vectorStore:i,logger:s})),c.post("/reindex",function(e){return async(t,r)=>{try{const t=await F(e.config.watch.paths,e.config.watch.ignored,e.processor,"processFile");return await r.status(200).send({ok:!0,filesIndexed:t})}catch(t){return e.logger.error({err:P(t)},"Reindex failed"),await r.status(500).send({error:"Internal server error"})}}}({config:a,processor:r,logger:s})),c.post("/rebuild-metadata",function(e){return async(t,r)=>{try{const t=e.config.metadataDir??".jeeves-metadata",i=[...T];for await(const r of e.vectorStore.scroll()){const e=r.payload,o=e.file_path;if("string"!=typeof o||0===o.length)continue;const s=n.omit(e,i);await E(o,t,s)}return await r.status(200).send({ok:!0})}catch(t){return e.logger.error({err:P(t)},"Rebuild metadata failed"),await r.status(500).send({error:"Internal server error"})}}}({config:a,vectorStore:i,logger:s})),c.post("/config-reindex",function(e){return async(t,r)=>{try{const i=t.body.scope??"rules";return(async()=>{try{if("rules"===i){const t=await F(e.config.watch.paths,e.config.watch.ignored,e.processor,"processRulesUpdate");e.logger.info({scope:i,filesProcessed:t},"Config reindex (rules) completed")}else{const t=await F(e.config.watch.paths,e.config.watch.ignored,e.processor,"processFile");e.logger.info({scope:i,filesProcessed:t},"Config reindex (full) completed")}}catch(t){e.logger.error({err:P(t),scope:i},"Config reindex failed")}})(),await r.status(200).send({status:"started",scope:i})}catch(t){return e.logger.error({err:P(t)},"Config reindex request failed"),await r.status(500).send({error:"Internal server error"})}}}({config:a,processor:r,logger:s})),c}const D={metadataDir:".jeeves-watcher",shutdownTimeoutMs:1e4},N={enabled:!0,debounceMs:1e3},A={host:"127.0.0.1",port:3456},O={level:"info"},W={debounceMs:300,stabilityThresholdMs:500,usePolling:!1,pollIntervalMs:1e3},_={chunkSize:1e3,chunkOverlap:200,dimensions:3072,rateLimitPerMinute:300,concurrency:5},q=c.z.object({paths:c.z.array(c.z.string()).min(1).describe('Glob patterns for files to watch (e.g., "**/*.md"). At least one required.'),ignored:c.z.array(c.z.string()).optional().describe('Glob patterns to exclude from watching (e.g., "**/node_modules/**").'),pollIntervalMs:c.z.number().optional().describe("Polling interval in milliseconds when usePolling is enabled."),usePolling:c.z.boolean().optional().describe("Use polling instead of native file system events (for network drives)."),debounceMs:c.z.number().optional().describe("Debounce delay in milliseconds for file change events."),stabilityThresholdMs:c.z.number().optional().describe("Time in milliseconds a file must remain unchanged before processing.")}),L=c.z.object({enabled:c.z.boolean().optional().describe("Enable automatic reloading when config file changes."),debounceMs:c.z.number().optional().describe("Debounce delay in milliseconds for config file change detection.")}),Q=c.z.object({provider:c.z.string().default("gemini").describe('Embedding provider name (e.g., "gemini", "openai").'),model:c.z.string().default("gemini-embedding-001").describe('Embedding model identifier (e.g., "gemini-embedding-001", "text-embedding-3-small").'),chunkSize:c.z.number().optional().describe("Maximum chunk size in characters for text splitting."),chunkOverlap:c.z.number().optional().describe("Character overlap between consecutive chunks."),dimensions:c.z.number().optional().describe("Embedding vector dimensions (must match model output)."),apiKey:c.z.string().optional().describe("API key for embedding provider (supports ${ENV_VAR} substitution)."),rateLimitPerMinute:c.z.number().optional().describe("Maximum embedding API requests per minute (rate limiting)."),concurrency:c.z.number().optional().describe("Maximum concurrent embedding requests.")}),$=c.z.object({url:c.z.string().describe('Qdrant server URL (e.g., "http://localhost:6333").'),collectionName:c.z.string().describe("Qdrant collection name for vector storage."),apiKey:c.z.string().optional().describe("Qdrant API key for authentication (supports ${ENV_VAR} substitution).")}),B=c.z.object({host:c.z.string().optional().describe('Host address for API server (e.g., "127.0.0.1", "0.0.0.0").'),port:c.z.number().optional().describe("Port for API server (e.g., 3456).")}),K=c.z.object({level:c.z.string().optional().describe("Logging level (trace, debug, info, warn, error, fatal)."),file:c.z.string().optional().describe("Path to log file (logs to stdout if omitted).")}),J=c.z.object({match:c.z.record(c.z.string(),c.z.unknown()).describe("JSON Schema object to match against file attributes."),set:c.z.record(c.z.string(),c.z.unknown()).describe("Metadata fields to set when match succeeds."),map:c.z.union([l.jsonMapMapSchema,c.z.string()]).optional().describe("JsonMap transformation (inline definition or named map reference).")}),G=c.z.object({watch:q.describe("File system watch configuration."),configWatch:L.optional().describe("Configuration file watch settings."),embedding:Q.describe("Embedding model configuration."),vectorStore:$.describe("Qdrant vector store configuration."),metadataDir:c.z.string().optional().describe("Directory for persisted metadata sidecar files."),api:B.optional().describe("API server configuration."),extractors:c.z.record(c.z.string(),c.z.unknown()).optional().describe("Extractor configurations keyed by name."),inferenceRules:c.z.array(J).optional().describe("Rules for inferring metadata from file attributes."),maps:c.z.record(c.z.string(),l.jsonMapMapSchema).optional().describe("Reusable named JsonMap transformations."),logging:K.optional().describe("Logging configuration."),shutdownTimeoutMs:c.z.number().optional().describe("Timeout in milliseconds for graceful shutdown."),maxRetries:c.z.number().optional().describe("Maximum consecutive system-level failures before triggering fatal error. Default: Infinity."),maxBackoffMs:c.z.number().optional().describe("Maximum backoff delay in milliseconds for system errors. Default: 60000.")}),V=/\$\{([^}]+)\}/g;function U(e){if("string"==typeof e)return function(e){return e.replace(V,((e,t)=>{const r=process.env[t];return void 0===r?e:r}))}(e);if(Array.isArray(e))return e.map((e=>U(e)));if(null!==e&&"object"==typeof e){const t={};for(const[r,i]of Object.entries(e))t[r]=U(i);return t}return e}const H="jeeves-watcher";async function Y(e){const t=a.cosmiconfig(H),r=e?await t.load(e):await t.search();if(!r||r.isEmpty)throw new Error("No jeeves-watcher configuration found. Create a .jeeves-watcherrc or jeeves-watcher.config.{js,ts,json,yaml} file.");try{const e=G.parse(r.config);return U((i=e,{...D,...i,watch:{...W,...i.watch},configWatch:{...N,...i.configWatch},embedding:{..._,...i.embedding},api:{...A,...i.api},logging:{...O,...i.logging}}))}catch(e){if(e instanceof c.ZodError){const t=e.issues.map((e=>`${e.path.join(".")}: ${e.message}`)).join("; ");throw new Error(`Invalid jeeves-watcher configuration: ${t}`)}throw e}var i}function Z(e){return e||{warn(e,t){t?console.warn(e,t):console.warn(e)}}}function X(e,t){return e<=0?Promise.resolve():new Promise(((r,i)=>{const o=setTimeout((()=>{s(),r()}),e),n=()=>{s(),i(new Error("Retry sleep aborted"))},s=()=>{clearTimeout(o),t&&t.removeEventListener("abort",n)};if(t){if(t.aborted)return void n();t.addEventListener("abort",n,{once:!0})}}))}function ee(e,t,r,i=0){const o=Math.max(0,e-1),n=Math.min(r,t*2**o),s=i>0?1+Math.random()*i:1;return Math.round(n*s)}async function te(e,t){const r=Math.max(1,t.attempts);let i;for(let o=1;o<=r;o++)try{return await e(o)}catch(e){i=e;if(o>=r)break;const n=ee(o,t.baseDelayMs,t.maxDelayMs,t.jitter);t.onRetry?.({attempt:o,attempts:r,delayMs:n,error:e}),await X(n,t.signal)}throw i}const re=new Map([["mock",function(e){return function(e){return{dimensions:e,embed:t=>Promise.resolve(t.map((t=>{const r=s.createHash("sha256").update(t,"utf8").digest(),i=[];for(let t=0;t<e;t++){const e=r[t%r.length];i.push(e/127.5-1)}return i})))}}(e.dimensions??768)}],["gemini",function(e,t){if(!e.apiKey)throw new Error("Gemini embedding provider requires config.embedding.apiKey");const r=e.dimensions??3072,i=Z(t),o=new h.GoogleGenerativeAIEmbeddings({apiKey:e.apiKey,model:e.model});return{dimensions:r,async embed(t){const n=await te((async r=>(r>1&&i.warn({attempt:r,provider:"gemini",model:e.model},"Retrying embedding request"),o.embedDocuments(t))),{attempts:5,baseDelayMs:500,maxDelayMs:1e4,jitter:.2,onRetry:({attempt:t,delayMs:r,error:o})=>{i.warn({attempt:t,delayMs:r,provider:"gemini",model:e.model,err:P(o)},"Embedding call failed; will retry")}});for(const e of n)if(e.length!==r)throw new Error(`Gemini embedding returned invalid dimensions: expected ${String(r)}, got ${String(e.length)}`);return n}}}]]);function ie(e,t){const r=re.get(e.provider);if(!r)throw new Error(`Unsupported embedding provider: ${e.provider}`);return r(e,t)}function oe(e){const t=e?.level??"info";if(e?.file){const r=u.transport({target:"pino/file",options:{destination:e.file,mkdir:!0}});return u({level:t},r)}return u({level:t})}function ne(e){return s.createHash("sha256").update(e,"utf8").digest("hex")}const se="6a6f686e-6761-4c74-ad6a-656576657321";function ae(e,t){const r=void 0!==t?`${z(e)}#${String(t)}`:z(e);return d.v5(r,se)}const ce=["content","body","text","snippet","subject","description","summary","transcript"];function le(e){if(!e||"object"!=typeof e)return JSON.stringify(e);const t=e;for(const e of ce){const r=t[e];if("string"==typeof r&&r.trim())return r}return JSON.stringify(e)}async function he(e){const t=await r.readFile(e,"utf8"),{frontmatter:i,body:o}=function(e){const t=e.replace(/^\uFEFF/,"");if(!/^\s*---/.test(t))return{body:e};const r=/^---\s*\n([\s\S]*?)\n---\s*\n?([\s\S]*)$/m.exec(t);if(!r)return{body:e};const[,i,o]=r,n=g.load(i);return{frontmatter:n&&"object"==typeof n&&!Array.isArray(n)?n:void 0,body:o}}(t);return{text:o,frontmatter:i}}async function ue(e){return{text:(await r.readFile(e,"utf8")).replace(/^\uFEFF/,"")}}async function de(e){const t=await r.readFile(e,"utf8"),i=x.load(t.replace(/^\uFEFF/,""));i("script, style").remove();return{text:i("body").text().trim()||i.text().trim()}}const fe=new Map([[".md",he],[".markdown",he],[".txt",ue],[".text",ue],[".json",async function(e){const t=await r.readFile(e,"utf8"),i=JSON.parse(t.replace(/^\uFEFF/,"")),o=i&&"object"==typeof i&&!Array.isArray(i)?i:void 0;return{text:le(i),json:o}}],[".pdf",async function(e){const t=await r.readFile(e),i=new Uint8Array(t),{extractText:o}=await import("unpdf"),{text:n}=await o(i);return{text:Array.isArray(n)?n.join("\n\n"):n}}],[".docx",async function(e){const t=await r.readFile(e);return{text:(await m.extractRawText({buffer:t})).value}}],[".html",de],[".htm",de]]);async function ge(e,t){const r=fe.get(t.toLowerCase());return r?r(e):ue(e)}function me(e,t){return"string"!=typeof e?e:e.replace(/\$\{([^}]+)\}/g,((e,r)=>{const i=n.get(t,r);return null==i?"":"string"==typeof i?i:JSON.stringify(i)}))}function pe(e,t){const r={};for(const[i,o]of Object.entries(e))r[i]=me(o,t);return r}async function ye(e,t,r,i){const o={split:(e,t)=>e.split(t),slice:(e,t,r)=>e.slice(t,r),join:(e,t)=>e.join(t),toLowerCase:e=>e.toLowerCase(),replace:(e,t,r)=>e.replace(t,r),get:(e,t)=>n.get(e,t)};let s={};const a=i??console;for(const{rule:i,validate:n}of e)if(n(t)){const e=pe(i.set,t);if(s={...s,...e},i.map){let e;if("string"==typeof i.map){if(e=r?.[i.map],!e){a.warn(`Map reference "${i.map}" not found in named maps. Skipping map transformation.`);continue}}else e=i.map;try{const r=new l.JsonMap(e,o),i=await r.transform(t);i&&"object"==typeof i&&!Array.isArray(i)?s={...s,...i}:a.warn("JsonMap transformation did not return an object; skipping merge.")}catch(e){a.warn(`JsonMap transformation failed: ${e instanceof Error?e.message:String(e)}`)}}}return s}function we(e,t,r,o){const n=e.replace(/\\/g,"/"),s={file:{path:n,directory:i.dirname(n).replace(/\\/g,"/"),filename:i.basename(n),extension:i.extname(n),sizeBytes:t.size,modified:t.mtime.toISOString()}};return r&&(s.frontmatter=r),o&&(s.json=o),s}function be(e){const t=function(){const e=new p({allErrors:!0});return y(e),e.addKeyword({keyword:"glob",type:"string",schemaType:"string",validate:(e,t)=>o.isMatch(t,e)}),e}();return e.map(((e,r)=>({rule:e,validate:t.compile({$id:`rule-${String(r)}`,...e.match})})))}async function ve(e,t,o,n,s){const a=i.extname(e),c=await r.stat(e),l=await ge(e,a),h=we(e,c,l.frontmatter,l.json),u=await ye(t,h,n,s),d=await R(e,o);return{inferred:u,enrichment:d,metadata:{...u,...d??{}},attributes:h,extracted:l}}function Me(e,t){const r=[];for(let i=0;i<t;i++)r.push(ae(e,i));return r}function xe(e,t=1){if(!e)return t;const r=e.total_chunks;return"number"==typeof r?r:t}class Pe{config;embeddingProvider;vectorStore;compiledRules;logger;constructor(e,t,r,i,o){this.config=e,this.embeddingProvider=t,this.vectorStore=r,this.compiledRules=i,this.logger=o}async processFile(e){try{const t=i.extname(e),{metadata:r,extracted:o}=await ve(e,this.compiledRules,this.config.metadataDir,this.config.maps,this.logger);if(!o.text.trim())return void this.logger.debug({filePath:e},"Skipping empty file");const n=ne(o.text),s=ae(e,0),a=await this.vectorStore.getPayload(s);if(a&&a.content_hash===n)return void this.logger.debug({filePath:e},"Content unchanged, skipping");const c=xe(a),l=this.config.chunkSize??1e3,h=function(e,t,r){const i=e.toLowerCase();return".md"===i||".markdown"===i?new w.MarkdownTextSplitter({chunkSize:t,chunkOverlap:r}):new w.RecursiveCharacterTextSplitter({chunkSize:t,chunkOverlap:r})}(t,l,this.config.chunkOverlap??200),u=await h.splitText(o.text),d=await this.embeddingProvider.embed(u),f=u.map(((t,i)=>({id:ae(e,i),vector:d[i],payload:{...r,file_path:e.replace(/\\/g,"/"),chunk_index:i,total_chunks:u.length,content_hash:n,chunk_text:t}})));if(await this.vectorStore.upsert(f),c>u.length){const t=Me(e,c).slice(u.length);await this.vectorStore.delete(t)}this.logger.info({filePath:e,chunks:u.length},"File processed successfully")}catch(t){this.logger.error({filePath:e,err:P(t)},"Failed to process file")}}async deleteFile(e){try{const t=ae(e,0),r=await this.vectorStore.getPayload(t),i=Me(e,xe(r));await this.vectorStore.delete(i),await C(e,this.config.metadataDir),this.logger.info({filePath:e},"File deleted from index")}catch(t){this.logger.error({filePath:e,err:P(t)},"Failed to delete file")}}async processMetadataUpdate(e,t){try{const r={...await R(e,this.config.metadataDir)??{},...t};await E(e,this.config.metadataDir,r);const i=ae(e,0),o=await this.vectorStore.getPayload(i);if(!o)return null;const n=xe(o),s=Me(e,n);return await this.vectorStore.setPayload(s,r),this.logger.info({filePath:e,chunks:n},"Metadata updated"),r}catch(t){return this.logger.error({filePath:e,err:P(t)},"Failed to update metadata"),null}}async processRulesUpdate(e){try{const t=ae(e,0),r=await this.vectorStore.getPayload(t);if(!r)return this.logger.debug({filePath:e},"File not indexed, skipping"),null;const{metadata:i}=await ve(e,this.compiledRules,this.config.metadataDir,this.config.maps,this.logger),o=xe(r),n=Me(e,o);return await this.vectorStore.setPayload(n,i),this.logger.info({filePath:e,chunks:o},"Rules re-applied"),i}catch(t){return this.logger.error({filePath:e,err:P(t)},"Failed to re-apply rules"),null}}updateRules(e){this.compiledRules=e,this.logger.info({rules:e.length},"Inference rules updated")}}class ke{debounceMs;concurrency;rateLimitPerMinute;started=!1;active=0;debounceTimers=new Map;latestByKey=new Map;normalQueue=[];lowQueue=[];tokens;lastRefillMs=Date.now();drainWaiters=[];constructor(e){this.debounceMs=e.debounceMs,this.concurrency=e.concurrency,this.rateLimitPerMinute=e.rateLimitPerMinute,this.tokens=this.rateLimitPerMinute??Number.POSITIVE_INFINITY}enqueue(e,t){const r=`${e.priority}:${e.path}`;this.latestByKey.set(r,{event:e,fn:t});const i=this.debounceTimers.get(r);i&&clearTimeout(i);const o=setTimeout((()=>{this.debounceTimers.delete(r);const e=this.latestByKey.get(r);e&&(this.latestByKey.delete(r),this.push(e),this.pump())}),this.debounceMs);this.debounceTimers.set(r,o)}process(){this.started=!0,this.pump()}async drain(){this.isIdle()||await new Promise((e=>{this.drainWaiters.push(e)}))}push(e){"low"===e.event.priority?this.lowQueue.push(e):this.normalQueue.push(e)}refillTokens(e){if(void 0===this.rateLimitPerMinute)return;const t=Math.max(0,e-this.lastRefillMs)*(this.rateLimitPerMinute/6e4);this.tokens=Math.min(this.rateLimitPerMinute,this.tokens+t),this.lastRefillMs=e}takeToken(){const e=Date.now();return this.refillTokens(e),!(this.tokens<1)&&(this.tokens-=1,!0)}nextItem(){return this.normalQueue.shift()??this.lowQueue.shift()}pump(){if(this.started){for(;this.active<this.concurrency;){const e=this.nextItem();if(!e)break;if(!this.takeToken()){"low"===e.event.priority?this.lowQueue.unshift(e):this.normalQueue.unshift(e),setTimeout((()=>{this.pump()}),250);break}this.active+=1,Promise.resolve().then((()=>e.fn(e.event))).finally((()=>{this.active-=1,this.pump(),this.maybeResolveDrain()}))}this.maybeResolveDrain()}}isIdle(){return 0===this.active&&0===this.normalQueue.length&&0===this.lowQueue.length&&0===this.debounceTimers.size&&0===this.latestByKey.size}maybeResolveDrain(){if(!this.isIdle())return;const e=this.drainWaiters;this.drainWaiters=[];for(const t of e)t()}}class Se{client;collectionName;dims;log;constructor(e,t,r){this.client=new b.QdrantClient({url:e.url,apiKey:e.apiKey,checkCompatibility:!1}),this.collectionName=e.collectionName,this.dims=t,this.log=Z(r)}async ensureCollection(){try{const e=await this.client.getCollections();e.collections.some((e=>e.name===this.collectionName))||await this.client.createCollection(this.collectionName,{vectors:{size:this.dims,distance:"Cosine"}})}catch(e){throw new Error(`Failed to ensure collection "${this.collectionName}": ${String(e)}`)}}async upsert(e){0!==e.length&&await te((async t=>{t>1&&this.log.warn({attempt:t,operation:"qdrant.upsert",points:e.length},"Retrying Qdrant upsert"),await this.client.upsert(this.collectionName,{wait:!0,points:e.map((e=>({id:e.id,vector:e.vector,payload:e.payload})))})}),{attempts:5,baseDelayMs:500,maxDelayMs:1e4,jitter:.2,onRetry:({attempt:e,delayMs:t,error:r})=>{this.log.warn({attempt:e,delayMs:t,operation:"qdrant.upsert",err:P(r)},"Qdrant upsert failed; will retry")}})}async delete(e){0!==e.length&&await te((async t=>{t>1&&this.log.warn({attempt:t,operation:"qdrant.delete",ids:e.length},"Retrying Qdrant delete"),await this.client.delete(this.collectionName,{wait:!0,points:e})}),{attempts:5,baseDelayMs:500,maxDelayMs:1e4,jitter:.2,onRetry:({attempt:e,delayMs:t,error:r})=>{this.log.warn({attempt:e,delayMs:t,operation:"qdrant.delete",err:P(r)},"Qdrant delete failed; will retry")}})}async setPayload(e,t){0!==e.length&&await this.client.setPayload(this.collectionName,{wait:!0,points:e,payload:t})}async getPayload(e){try{const t=await this.client.retrieve(this.collectionName,{ids:[e],with_payload:!0,with_vector:!1});return 0===t.length?null:t[0].payload}catch{return null}}async search(e,t,r){return(await this.client.search(this.collectionName,{vector:e,limit:t,with_payload:!0,...r?{filter:r}:{}})).map((e=>({id:String(e.id),score:e.score,payload:e.payload})))}async*scroll(e,t=100){let r;for(;;){const i=await this.client.scroll(this.collectionName,{limit:t,with_payload:!0,with_vector:!1,...e?{filter:e}:{},...void 0!==r?{offset:r}:{}});for(const e of i.points)yield{id:String(e.id),payload:e.payload};const o=i.next_page_offset;if(null==o)break;if("string"!=typeof o&&"number"!=typeof o)break;r=o}}}class Fe{consecutiveFailures=0;maxRetries;maxBackoffMs;baseDelayMs;onFatalError;logger;constructor(e){this.maxRetries=e.maxRetries??Number.POSITIVE_INFINITY,this.maxBackoffMs=e.maxBackoffMs??6e4,this.baseDelayMs=e.baseDelayMs??1e3,this.onFatalError=e.onFatalError,this.logger=e.logger}recordSuccess(){this.consecutiveFailures>0&&this.logger.info({previousFailures:this.consecutiveFailures},"System health recovered"),this.consecutiveFailures=0}recordFailure(e){if(this.consecutiveFailures+=1,this.logger.error({consecutiveFailures:this.consecutiveFailures,maxRetries:this.maxRetries,err:P(e)},"System-level failure recorded"),this.consecutiveFailures>=this.maxRetries){if(this.logger.fatal({consecutiveFailures:this.consecutiveFailures},"Maximum retries exceeded, triggering fatal error"),this.onFatalError)return this.onFatalError(e),!1;throw e instanceof Error?e:new Error(`Fatal system error: ${String(e)}`)}return!0}get currentBackoffMs(){if(0===this.consecutiveFailures)return 0;const e=Math.max(0,this.consecutiveFailures-1);return Math.min(this.maxBackoffMs,this.baseDelayMs*2**e)}async backoff(e){const t=this.currentBackoffMs;t<=0||(this.logger.warn({delayMs:t,consecutiveFailures:this.consecutiveFailures},"Backing off before next attempt"),await new Promise(((r,i)=>{const o=setTimeout((()=>{s(),r()}),t),n=()=>{s(),i(new Error("Backoff aborted"))},s=()=>{clearTimeout(o),e&&e.removeEventListener("abort",n)};if(e){if(e.aborted)return void n();e.addEventListener("abort",n,{once:!0})}})))}get failures(){return this.consecutiveFailures}}class ze{config;queue;processor;logger;health;watcher;constructor(e,t,r,i,o={}){this.config=e,this.queue=t,this.processor=r,this.logger=i;const n={maxRetries:o.maxRetries,maxBackoffMs:o.maxBackoffMs,onFatalError:o.onFatalError,logger:i};this.health=new Fe(n)}start(){this.watcher=v.watch(this.config.paths,{ignored:this.config.ignored,usePolling:this.config.usePolling,interval:this.config.pollIntervalMs,awaitWriteFinish:!!this.config.stabilityThresholdMs&&{stabilityThreshold:this.config.stabilityThresholdMs},ignoreInitial:!1}),this.watcher.on("add",(e=>{this.logger.debug({path:e},"File added"),this.queue.enqueue({type:"create",path:e,priority:"normal"},(()=>this.wrapProcessing((()=>this.processor.processFile(e)))))})),this.watcher.on("change",(e=>{this.logger.debug({path:e},"File changed"),this.queue.enqueue({type:"modify",path:e,priority:"normal"},(()=>this.wrapProcessing((()=>this.processor.processFile(e)))))})),this.watcher.on("unlink",(e=>{this.logger.debug({path:e},"File removed"),this.queue.enqueue({type:"delete",path:e,priority:"normal"},(()=>this.wrapProcessing((()=>this.processor.deleteFile(e)))))})),this.watcher.on("error",(e=>{this.logger.error({err:P(e)},"Watcher error"),this.health.recordFailure(e)})),this.queue.process(),this.logger.info({paths:this.config.paths},"Filesystem watcher started")}async stop(){this.watcher&&(await this.watcher.close(),this.watcher=void 0,this.logger.info("Filesystem watcher stopped"))}get systemHealth(){return this.health}async wrapProcessing(e){try{await this.health.backoff(),await e(),this.health.recordSuccess()}catch(e){this.health.recordFailure(e)||await this.stop()}}}class je{options;watcher;debounce;constructor(e){this.options=e}start(){this.options.enabled&&(this.watcher=v.watch(this.options.configPath,{ignoreInitial:!0}),this.watcher.on("change",(()=>{this.debounce&&clearTimeout(this.debounce),this.debounce=setTimeout((()=>{this.options.onChange()}),this.options.debounceMs)})),this.watcher.on("error",(e=>{this.options.logger.error({err:P(e)},"Config watcher error")})),this.options.logger.info({configPath:this.options.configPath,debounceMs:this.options.debounceMs},"Config watcher started"))}async stop(){this.debounce&&(clearTimeout(this.debounce),this.debounce=void 0),this.watcher&&(await this.watcher.close(),this.watcher=void 0)}}const Re={loadConfig:Y,createLogger:oe,createEmbeddingProvider:ie,createVectorStoreClient:(e,t,r)=>new Se(e,t,r),compileRules:be,createDocumentProcessor:(e,t,r,i,o)=>new Pe(e,t,r,i,o),createEventQueue:e=>new ke(e),createFileSystemWatcher:(e,t,r,i,o)=>new ze(e,t,r,i,o),createApiServer:I};class Ee{config;configPath;factories;runtimeOptions;logger;watcher;queue;server;processor;configWatcher;constructor(e,t,r={},i={}){this.config=e,this.configPath=t,this.factories={...Re,...r},this.runtimeOptions=i}async start(){const e=this.factories.createLogger(this.config.logging);let t;this.logger=e;try{t=this.factories.createEmbeddingProvider(this.config.embedding,e)}catch(t){throw e.fatal({err:P(t)},"Failed to create embedding provider"),t}const r=this.factories.createVectorStoreClient(this.config.vectorStore,t.dimensions,e);await r.ensureCollection();const i=this.factories.compileRules(this.config.inferenceRules??[]),o={metadataDir:this.config.metadataDir??".jeeves-metadata",chunkSize:this.config.embedding.chunkSize,chunkOverlap:this.config.embedding.chunkOverlap,maps:this.config.maps},n=this.factories.createDocumentProcessor(o,t,r,i,e);this.processor=n;const s=this.factories.createEventQueue({debounceMs:this.config.watch.debounceMs??2e3,concurrency:this.config.embedding.concurrency??5,rateLimitPerMinute:this.config.embedding.rateLimitPerMinute});this.queue=s;const a=this.factories.createFileSystemWatcher(this.config.watch,s,n,e,{maxRetries:this.config.maxRetries,maxBackoffMs:this.config.maxBackoffMs,onFatalError:this.runtimeOptions.onFatalError});this.watcher=a;const c=this.factories.createApiServer({processor:n,vectorStore:r,embeddingProvider:t,queue:s,config:this.config,logger:e});this.server=c,await c.listen({host:this.config.api?.host??"127.0.0.1",port:this.config.api?.port??3456}),a.start(),this.startConfigWatch(),e.info("jeeves-watcher started")}async stop(){if(await this.stopConfigWatch(),this.watcher&&await this.watcher.stop(),this.queue){const e=this.config.shutdownTimeoutMs??1e4;await Promise.race([this.queue.drain().then((()=>!0)),new Promise((t=>{setTimeout((()=>{t(!1)}),e)}))])||this.logger?.warn({timeoutMs:e},"Queue drain timeout hit, forcing shutdown")}this.server&&await this.server.close(),this.logger?.info("jeeves-watcher stopped")}startConfigWatch(){const e=this.logger;if(!e)return;const t=this.config.configWatch?.enabled??!0;if(!t)return;if(!this.configPath)return void e.debug("Config watch enabled, but no config path was provided");const r=this.config.configWatch?.debounceMs??1e4;this.configWatcher=new je({configPath:this.configPath,enabled:t,debounceMs:r,logger:e,onChange:async()=>this.reloadConfig()}),this.configWatcher.start()}async stopConfigWatch(){this.configWatcher&&(await this.configWatcher.stop(),this.configWatcher=void 0)}async reloadConfig(){const e=this.logger,t=this.processor;if(e&&t&&this.configPath){e.info({configPath:this.configPath},"Config change detected, reloading...");try{const r=await this.factories.loadConfig(this.configPath);this.config=r;const i=this.factories.compileRules(r.inferenceRules??[]);t.updateRules(i),e.info({configPath:this.configPath,rules:i.length},"Config reloaded")}catch(t){e.error({err:P(t)},"Failed to reload config")}}}}e.DocumentProcessor=Pe,e.EventQueue=ke,e.FileSystemWatcher=ze,e.JeevesWatcher=Ee,e.SystemHealth=Fe,e.VectorStoreClient=Se,e.apiConfigSchema=B,e.applyRules=ye,e.buildAttributes=we,e.compileRules=be,e.configWatchConfigSchema=L,e.contentHash=ne,e.createApiServer=I,e.createEmbeddingProvider=ie,e.createLogger=oe,e.deleteMetadata=C,e.embeddingConfigSchema=Q,e.extractText=ge,e.inferenceRuleSchema=J,e.jeevesWatcherConfigSchema=G,e.loadConfig=Y,e.loggingConfigSchema=K,e.metadataPath=j,e.pointId=ae,e.readMetadata=R,e.startFromConfig=async function(e){const t=await Y(e),r=new Ee(t,e);return function(e){const t=async()=>{await e(),process.exit(0)};process.on("SIGTERM",(()=>{t()})),process.on("SIGINT",(()=>{t()}))}((()=>r.stop())),await r.start(),r},e.vectorStoreConfigSchema=$,e.watchConfigSchema=q,e.writeMetadata=E}(this["jeeves-watcher"]=this["jeeves-watcher"]||{},Fastify,promises,node_path,picomatch,radash,node_crypto,cosmiconfig,zod,jsonmap,googleGenai,pino,uuid,cheerio,yaml,mammoth,Ajv,addFormats,textsplitters,jsClientRest,chokidar);
+!function(e,t,i,r,o,n,s,a,c,l,h,d,u,g,f,p,m,y,w,b,v,M,x){"use strict";function S(e){var t=Object.create(null);return e&&Object.keys(e).forEach((function(i){if("default"!==i){var r=Object.getOwnPropertyDescriptor(e,i);Object.defineProperty(t,i,r.get?r:{enumerable:!0,get:function(){return e[i]}})}})),t.default=e,Object.freeze(t)}var F=S(p);function P(e){if(e instanceof Error)return e;if("string"==typeof e)return new Error(e);const t=String("object"==typeof e&&null!==e&&"message"in e?e.message:e),i=new Error(t);return i.cause=e,i}function k(e){const t=e.replace(/\\/g,"/"),i=t.search(/[*?\[]/);if(-1===i)return r.resolve(e);const o=t.slice(0,i),n=o.endsWith("/")?o.slice(0,-1):r.dirname(o);return r.resolve(n)}async function*z(e){let t;try{t=(await i.readdir(e,{withFileTypes:!0})).map((e=>({name:e.name,isDirectory:e.isDirectory()})))}catch{return}for(const o of t){const t=r.resolve(e,o.name);if(o.isDirectory)yield*z(t);else try{(await i.stat(t)).isFile()&&(yield t)}catch{}}}async function j(e,t,i,r){const n=await async function(e,t=[]){const i=e.map((e=>e.replace(/\\/g,"/"))),r=t.map((e=>e.replace(/\\/g,"/"))),n=o(i,{dot:!0}),s=r.length?o(r,{dot:!0}):()=>!1,a=Array.from(new Set(e.map(k))),c=new Set;for(const e of a)for await(const t of z(e)){const e=t.replace(/\\/g,"/");s(e)||n(e)&&c.add(t)}return Array.from(c)}(e,t);for(const e of n)await i[r](e);return n.length}function R(e,t=!1){let i=e.replace(/\\/g,"/").toLowerCase();return t&&(i=i.replace(/^([a-z]):/,((e,t)=>t))),i}function E(e,t){const i=R(e,!0),o=s.createHash("sha256").update(i,"utf8").digest("hex");return r.join(t,`${o}.meta.json`)}async function C(e,t){try{const r=await i.readFile(E(e,t),"utf8");return JSON.parse(r)}catch{return null}}async function D(e,t,o){const n=E(e,t);await i.mkdir(r.dirname(n),{recursive:!0}),await i.writeFile(n,JSON.stringify(o,null,2),"utf8")}async function T(e,t){try{await i.rm(E(e,t))}catch{}}const I=["file_path","chunk_index","total_chunks","content_hash","chunk_text"];function W(e){const{processor:i,vectorStore:r,embeddingProvider:o,logger:s,config:a}=e,c=t({logger:!1});var l;return c.get("/status",(()=>({status:"ok",uptime:process.uptime()}))),c.post("/metadata",(l={processor:i,logger:s},async(e,t)=>{try{const{path:t,metadata:i}=e.body;return await l.processor.processMetadataUpdate(t,i),{ok:!0}}catch(e){return l.logger.error({err:P(e)},"Metadata update failed"),t.status(500).send({error:"Internal server error"})}})),c.post("/search",function(e){return async(t,i)=>{try{const{query:i,limit:r=10}=t.body,o=await e.embeddingProvider.embed([i]);return await e.vectorStore.search(o[0],r)}catch(t){return e.logger.error({err:P(t)},"Search failed"),i.status(500).send({error:"Internal server error"})}}}({embeddingProvider:o,vectorStore:r,logger:s})),c.post("/reindex",function(e){return async(t,i)=>{try{const t=await j(e.config.watch.paths,e.config.watch.ignored,e.processor,"processFile");return await i.status(200).send({ok:!0,filesIndexed:t})}catch(t){return e.logger.error({err:P(t)},"Reindex failed"),await i.status(500).send({error:"Internal server error"})}}}({config:a,processor:i,logger:s})),c.post("/rebuild-metadata",function(e){return async(t,i)=>{try{const t=e.config.metadataDir??".jeeves-metadata",r=[...I];for await(const i of e.vectorStore.scroll()){const e=i.payload,o=e.file_path;if("string"!=typeof o||0===o.length)continue;const s=n.omit(e,r);await D(o,t,s)}return await i.status(200).send({ok:!0})}catch(t){return e.logger.error({err:P(t)},"Rebuild metadata failed"),await i.status(500).send({error:"Internal server error"})}}}({config:a,vectorStore:r,logger:s})),c.post("/config-reindex",function(e){return async(t,i)=>{try{const r=t.body.scope??"rules";return(async()=>{try{if("rules"===r){const t=await j(e.config.watch.paths,e.config.watch.ignored,e.processor,"processRulesUpdate");e.logger.info({scope:r,filesProcessed:t},"Config reindex (rules) completed")}else{const t=await j(e.config.watch.paths,e.config.watch.ignored,e.processor,"processFile");e.logger.info({scope:r,filesProcessed:t},"Config reindex (full) completed")}}catch(t){e.logger.error({err:P(t),scope:r},"Config reindex failed")}})(),await i.status(200).send({status:"started",scope:r})}catch(t){return e.logger.error({err:P(t)},"Config reindex request failed"),await i.status(500).send({error:"Internal server error"})}}}({config:a,processor:i,logger:s})),c}const N={metadataDir:".jeeves-watcher",shutdownTimeoutMs:1e4},O={enabled:!0,debounceMs:1e3},_={host:"127.0.0.1",port:3456},A={level:"info"},q={debounceMs:300,stabilityThresholdMs:500,usePolling:!1,pollIntervalMs:1e3,respectGitignore:!0},L={chunkSize:1e3,chunkOverlap:200,dimensions:3072,rateLimitPerMinute:300,concurrency:5},Q=c.z.object({paths:c.z.array(c.z.string()).min(1).describe('Glob patterns for files to watch (e.g., "**/*.md"). At least one required.'),ignored:c.z.array(c.z.string()).optional().describe('Glob patterns to exclude from watching (e.g., "**/node_modules/**").'),pollIntervalMs:c.z.number().optional().describe("Polling interval in milliseconds when usePolling is enabled."),usePolling:c.z.boolean().optional().describe("Use polling instead of native file system events (for network drives)."),debounceMs:c.z.number().optional().describe("Debounce delay in milliseconds for file change events."),stabilityThresholdMs:c.z.number().optional().describe("Time in milliseconds a file must remain unchanged before processing."),respectGitignore:c.z.boolean().optional().describe("Skip files ignored by .gitignore in git repositories. Only applies to repos with a .git directory. Default: true.")}),$=c.z.object({enabled:c.z.boolean().optional().describe("Enable automatic reloading when config file changes."),debounceMs:c.z.number().optional().describe("Debounce delay in milliseconds for config file change detection.")}),G=c.z.object({provider:c.z.string().default("gemini").describe('Embedding provider name (e.g., "gemini", "openai").'),model:c.z.string().default("gemini-embedding-001").describe('Embedding model identifier (e.g., "gemini-embedding-001", "text-embedding-3-small").'),chunkSize:c.z.number().optional().describe("Maximum chunk size in characters for text splitting."),chunkOverlap:c.z.number().optional().describe("Character overlap between consecutive chunks."),dimensions:c.z.number().optional().describe("Embedding vector dimensions (must match model output)."),apiKey:c.z.string().optional().describe("API key for embedding provider (supports ${ENV_VAR} substitution)."),rateLimitPerMinute:c.z.number().optional().describe("Maximum embedding API requests per minute (rate limiting)."),concurrency:c.z.number().optional().describe("Maximum concurrent embedding requests.")}),B=c.z.object({url:c.z.string().describe('Qdrant server URL (e.g., "http://localhost:6333").'),collectionName:c.z.string().describe("Qdrant collection name for vector storage."),apiKey:c.z.string().optional().describe("Qdrant API key for authentication (supports ${ENV_VAR} substitution).")}),K=c.z.object({host:c.z.string().optional().describe('Host address for API server (e.g., "127.0.0.1", "0.0.0.0").'),port:c.z.number().optional().describe("Port for API server (e.g., 3456).")}),J=c.z.object({level:c.z.string().optional().describe("Logging level (trace, debug, info, warn, error, fatal)."),file:c.z.string().optional().describe("Path to log file (logs to stdout if omitted).")}),V=c.z.object({match:c.z.record(c.z.string(),c.z.unknown()).describe("JSON Schema object to match against file attributes."),set:c.z.record(c.z.string(),c.z.unknown()).describe("Metadata fields to set when match succeeds."),map:c.z.union([l.jsonMapMapSchema,c.z.string()]).optional().describe("JsonMap transformation (inline definition or named map reference).")}),U=c.z.object({watch:Q.describe("File system watch configuration."),configWatch:$.optional().describe("Configuration file watch settings."),embedding:G.describe("Embedding model configuration."),vectorStore:B.describe("Qdrant vector store configuration."),metadataDir:c.z.string().optional().describe("Directory for persisted metadata sidecar files."),api:K.optional().describe("API server configuration."),extractors:c.z.record(c.z.string(),c.z.unknown()).optional().describe("Extractor configurations keyed by name."),inferenceRules:c.z.array(V).optional().describe("Rules for inferring metadata from file attributes."),maps:c.z.record(c.z.string(),l.jsonMapMapSchema).optional().describe("Reusable named JsonMap transformations."),logging:J.optional().describe("Logging configuration."),shutdownTimeoutMs:c.z.number().optional().describe("Timeout in milliseconds for graceful shutdown."),maxRetries:c.z.number().optional().describe("Maximum consecutive system-level failures before triggering fatal error. Default: Infinity."),maxBackoffMs:c.z.number().optional().describe("Maximum backoff delay in milliseconds for system errors. Default: 60000.")}),H=/\$\{([^}]+)\}/g;function Y(e){if("string"==typeof e)return function(e){return e.replace(H,((e,t)=>{const i=process.env[t];return void 0===i?e:i}))}(e);if(Array.isArray(e))return e.map((e=>Y(e)));if(null!==e&&"object"==typeof e){const t={};for(const[i,r]of Object.entries(e))t[i]=Y(r);return t}return e}const Z="jeeves-watcher";async function X(e){const t=a.cosmiconfig(Z),i=e?await t.load(e):await t.search();if(!i||i.isEmpty)throw new Error("No jeeves-watcher configuration found. Create a .jeeves-watcherrc or jeeves-watcher.config.{js,ts,json,yaml} file.");try{const e=U.parse(i.config);return Y((r=e,{...N,...r,watch:{...q,...r.watch},configWatch:{...O,...r.configWatch},embedding:{...L,...r.embedding},api:{..._,...r.api},logging:{...A,...r.logging}}))}catch(e){if(e instanceof c.ZodError){const t=e.issues.map((e=>`${e.path.join(".")}: ${e.message}`)).join("; ");throw new Error(`Invalid jeeves-watcher configuration: ${t}`)}throw e}var r}function ee(e){return e||{warn(e,t){t?console.warn(e,t):console.warn(e)}}}function te(e,t){return e<=0?Promise.resolve():new Promise(((i,r)=>{const o=setTimeout((()=>{s(),i()}),e),n=()=>{s(),r(new Error("Retry sleep aborted"))},s=()=>{clearTimeout(o),t&&t.removeEventListener("abort",n)};if(t){if(t.aborted)return void n();t.addEventListener("abort",n,{once:!0})}}))}function ie(e,t,i,r=0){const o=Math.max(0,e-1),n=Math.min(i,t*2**o),s=r>0?1+Math.random()*r:1;return Math.round(n*s)}async function re(e,t){const i=Math.max(1,t.attempts);let r;for(let o=1;o<=i;o++)try{return await e(o)}catch(e){r=e;if(o>=i)break;const n=ie(o,t.baseDelayMs,t.maxDelayMs,t.jitter);t.onRetry?.({attempt:o,attempts:i,delayMs:n,error:e}),await te(n,t.signal)}throw r}const oe=new Map([["mock",function(e){return function(e){return{dimensions:e,embed:t=>Promise.resolve(t.map((t=>{const i=s.createHash("sha256").update(t,"utf8").digest(),r=[];for(let t=0;t<e;t++){const e=i[t%i.length];r.push(e/127.5-1)}return r})))}}(e.dimensions??768)}],["gemini",function(e,t){if(!e.apiKey)throw new Error("Gemini embedding provider requires config.embedding.apiKey");const i=e.dimensions??3072,r=ee(t),o=new h.GoogleGenerativeAIEmbeddings({apiKey:e.apiKey,model:e.model});return{dimensions:i,async embed(t){const n=await re((async i=>(i>1&&r.warn({attempt:i,provider:"gemini",model:e.model},"Retrying embedding request"),o.embedDocuments(t))),{attempts:5,baseDelayMs:500,maxDelayMs:1e4,jitter:.2,onRetry:({attempt:t,delayMs:i,error:o})=>{r.warn({attempt:t,delayMs:i,provider:"gemini",model:e.model,err:P(o)},"Embedding call failed; will retry")}});for(const e of n)if(e.length!==i)throw new Error(`Gemini embedding returned invalid dimensions: expected ${String(i)}, got ${String(e.length)}`);return n}}}]]);function ne(e,t){const i=oe.get(e.provider);if(!i)throw new Error(`Unsupported embedding provider: ${e.provider}`);return i(e,t)}function se(e){let t=r.resolve(e);const i=r.resolve("/");for(;t!==i;){if(d.existsSync(r.join(t,".git"))&&d.statSync(r.join(t,".git")).isDirectory())return t;const e=r.dirname(t);if(e===t)break;t=e}}function ae(e){const t=r.resolve(e);try{return d.statSync(t).isDirectory()?t:r.dirname(t)}catch{}const i=/[*?[{]/.exec(e);if(!i)return;const o=e.slice(0,i.index).trim(),n=0===o.length?".":o.endsWith("/")||o.endsWith("\\")?o:r.dirname(o),s=r.resolve(n);return d.existsSync(s)?s:void 0}function ce(e){const t=[],i=r.join(e,".gitignore");let o;d.existsSync(i)&&t.push(i);try{o=d.readdirSync(e)}catch{return t}for(const i of o){if(".git"===i||"node_modules"===i)continue;const o=r.join(e,i);try{d.statSync(o).isDirectory()&&t.push(...ce(o))}catch{}}return t}function le(e){const t=d.readFileSync(e,"utf8");return u().add(t)}class he{repos=new Map;constructor(e){this.scan(e)}scan(e){this.repos.clear();const t=new Set;for(const i of e){const e=ae(i);if(!e)continue;if(t.has(e))continue;t.add(e);const o=se(e);if(!o)continue;if(this.repos.has(o))continue;const n=ce(o).map((e=>({dir:r.dirname(e),ig:le(e)})));n.sort(((e,t)=>t.dir.length-e.dir.length)),this.repos.set(o,{root:o,entries:n})}}isIgnored(e){const t=r.resolve(e);for(const[,e]of this.repos){const i=r.relative(e.root,t);if(!i.startsWith("..")&&!i.startsWith(r.resolve("/")))for(const i of e.entries){const e=r.relative(i.dir,t);if(e.startsWith(".."))continue;const o=e.replace(/\\/g,"/");if(i.ig.ignores(o))return!0}}return!1}invalidate(e){const t=r.resolve(e),i=r.dirname(t);for(const[,e]of this.repos){if(!r.relative(e.root,i).startsWith(".."))return e.entries=e.entries.filter((e=>e.dir!==i)),void(d.existsSync(t)&&(e.entries.push({dir:i,ig:le(t)}),e.entries.sort(((e,t)=>t.dir.length-e.dir.length))))}const o=se(i);if(o&&d.existsSync(t)){const e=[{dir:i,ig:le(t)}];if(this.repos.has(o)){const t=this.repos.get(o);t.entries.push(e[0]),t.entries.sort(((e,t)=>t.dir.length-e.dir.length))}else this.repos.set(o,{root:o,entries:e})}}}function de(e){const t=e?.level??"info";if(e?.file){const i=g.transport({target:"pino/file",options:{destination:e.file,mkdir:!0}});return g({level:t},i)}return g({level:t})}function ue(e){return s.createHash("sha256").update(e,"utf8").digest("hex")}const ge="6a6f686e-6761-4c74-ad6a-656576657321";function fe(e,t){const i=void 0!==t?`${R(e)}#${String(t)}`:R(e);return f.v5(i,ge)}const pe=["content","body","text","snippet","subject","description","summary","transcript"];function me(e){if(!e||"object"!=typeof e)return JSON.stringify(e);const t=e;for(const e of pe){const i=t[e];if("string"==typeof i&&i.trim())return i}return JSON.stringify(e)}async function ye(e){const t=await i.readFile(e,"utf8"),{frontmatter:r,body:o}=function(e){const t=e.replace(/^\uFEFF/,"");if(!/^\s*---/.test(t))return{body:e};const i=/^---\s*\n([\s\S]*?)\n---\s*\n?([\s\S]*)$/m.exec(t);if(!i)return{body:e};const[,r,o]=i,n=m.load(r);return{frontmatter:n&&"object"==typeof n&&!Array.isArray(n)?n:void 0,body:o}}(t);return{text:o,frontmatter:r}}async function we(e){return{text:(await i.readFile(e,"utf8")).replace(/^\uFEFF/,"")}}async function be(e){const t=await i.readFile(e,"utf8"),r=F.load(t.replace(/^\uFEFF/,""));r("script, style").remove();return{text:r("body").text().trim()||r.text().trim()}}const ve=new Map([[".md",ye],[".markdown",ye],[".txt",we],[".text",we],[".json",async function(e){const t=await i.readFile(e,"utf8"),r=JSON.parse(t.replace(/^\uFEFF/,"")),o=r&&"object"==typeof r&&!Array.isArray(r)?r:void 0;return{text:me(r),json:o}}],[".pdf",async function(e){const t=await i.readFile(e),r=new Uint8Array(t),{extractText:o}=await import("unpdf"),{text:n}=await o(r);return{text:Array.isArray(n)?n.join("\n\n"):n}}],[".docx",async function(e){const t=await i.readFile(e);return{text:(await y.extractRawText({buffer:t})).value}}],[".html",be],[".htm",be]]);async function Me(e,t){const i=ve.get(t.toLowerCase());return i?i(e):we(e)}function xe(e,t){return"string"!=typeof e?e:e.replace(/\$\{([^}]+)\}/g,((e,i)=>{const r=n.get(t,i);return null==r?"":"string"==typeof r?r:JSON.stringify(r)}))}function Se(e,t){const i={};for(const[r,o]of Object.entries(e))i[r]=xe(o,t);return i}async function Fe(e,t,i,r){const o={split:(e,t)=>e.split(t),slice:(e,t,i)=>e.slice(t,i),join:(e,t)=>e.join(t),toLowerCase:e=>e.toLowerCase(),replace:(e,t,i)=>e.replace(t,i),get:(e,t)=>n.get(e,t)};let s={};const a=r??console;for(const{rule:r,validate:n}of e)if(n(t)){const e=Se(r.set,t);if(s={...s,...e},r.map){let e;if("string"==typeof r.map){if(e=i?.[r.map],!e){a.warn(`Map reference "${r.map}" not found in named maps. Skipping map transformation.`);continue}}else e=r.map;try{const i=new l.JsonMap(e,o),r=await i.transform(t);r&&"object"==typeof r&&!Array.isArray(r)?s={...s,...r}:a.warn("JsonMap transformation did not return an object; skipping merge.")}catch(e){a.warn(`JsonMap transformation failed: ${e instanceof Error?e.message:String(e)}`)}}}return s}function Pe(e,t,i,o){const n=e.replace(/\\/g,"/"),s={file:{path:n,directory:r.dirname(n).replace(/\\/g,"/"),filename:r.basename(n),extension:r.extname(n),sizeBytes:t.size,modified:t.mtime.toISOString()}};return i&&(s.frontmatter=i),o&&(s.json=o),s}function ke(e){const t=function(){const e=new w({allErrors:!0});return b(e),e.addKeyword({keyword:"glob",type:"string",schemaType:"string",validate:(e,t)=>o.isMatch(t,e)}),e}();return e.map(((e,i)=>({rule:e,validate:t.compile({$id:`rule-${String(i)}`,...e.match})})))}async function ze(e,t,o,n,s){const a=r.extname(e),c=await i.stat(e),l=await Me(e,a),h=Pe(e,c,l.frontmatter,l.json),d=await Fe(t,h,n,s),u=await C(e,o);return{inferred:d,enrichment:u,metadata:{...d,...u??{}},attributes:h,extracted:l}}function je(e,t){const i=[];for(let r=0;r<t;r++)i.push(fe(e,r));return i}function Re(e,t=1){if(!e)return t;const i=e.total_chunks;return"number"==typeof i?i:t}class Ee{config;embeddingProvider;vectorStore;compiledRules;logger;constructor(e,t,i,r,o){this.config=e,this.embeddingProvider=t,this.vectorStore=i,this.compiledRules=r,this.logger=o}async processFile(e){try{const t=r.extname(e),{metadata:i,extracted:o}=await ze(e,this.compiledRules,this.config.metadataDir,this.config.maps,this.logger);if(!o.text.trim())return void this.logger.debug({filePath:e},"Skipping empty file");const n=ue(o.text),s=fe(e,0),a=await this.vectorStore.getPayload(s);if(a&&a.content_hash===n)return void this.logger.debug({filePath:e},"Content unchanged, skipping");const c=Re(a),l=this.config.chunkSize??1e3,h=function(e,t,i){const r=e.toLowerCase();return".md"===r||".markdown"===r?new v.MarkdownTextSplitter({chunkSize:t,chunkOverlap:i}):new v.RecursiveCharacterTextSplitter({chunkSize:t,chunkOverlap:i})}(t,l,this.config.chunkOverlap??200),d=await h.splitText(o.text),u=await this.embeddingProvider.embed(d),g=d.map(((t,r)=>({id:fe(e,r),vector:u[r],payload:{...i,file_path:e.replace(/\\/g,"/"),chunk_index:r,total_chunks:d.length,content_hash:n,chunk_text:t}})));if(await this.vectorStore.upsert(g),c>d.length){const t=je(e,c).slice(d.length);await this.vectorStore.delete(t)}this.logger.info({filePath:e,chunks:d.length},"File processed successfully")}catch(t){this.logger.error({filePath:e,err:P(t)},"Failed to process file")}}async deleteFile(e){try{const t=fe(e,0),i=await this.vectorStore.getPayload(t),r=je(e,Re(i));await this.vectorStore.delete(r),await T(e,this.config.metadataDir),this.logger.info({filePath:e},"File deleted from index")}catch(t){this.logger.error({filePath:e,err:P(t)},"Failed to delete file")}}async processMetadataUpdate(e,t){try{const i={...await C(e,this.config.metadataDir)??{},...t};await D(e,this.config.metadataDir,i);const r=fe(e,0),o=await this.vectorStore.getPayload(r);if(!o)return null;const n=Re(o),s=je(e,n);return await this.vectorStore.setPayload(s,i),this.logger.info({filePath:e,chunks:n},"Metadata updated"),i}catch(t){return this.logger.error({filePath:e,err:P(t)},"Failed to update metadata"),null}}async processRulesUpdate(e){try{const t=fe(e,0),i=await this.vectorStore.getPayload(t);if(!i)return this.logger.debug({filePath:e},"File not indexed, skipping"),null;const{metadata:r}=await ze(e,this.compiledRules,this.config.metadataDir,this.config.maps,this.logger),o=Re(i),n=je(e,o);return await this.vectorStore.setPayload(n,r),this.logger.info({filePath:e,chunks:o},"Rules re-applied"),r}catch(t){return this.logger.error({filePath:e,err:P(t)},"Failed to re-apply rules"),null}}updateRules(e){this.compiledRules=e,this.logger.info({rules:e.length},"Inference rules updated")}}class Ce{debounceMs;concurrency;rateLimitPerMinute;started=!1;active=0;debounceTimers=new Map;latestByKey=new Map;normalQueue=[];lowQueue=[];tokens;lastRefillMs=Date.now();drainWaiters=[];constructor(e){this.debounceMs=e.debounceMs,this.concurrency=e.concurrency,this.rateLimitPerMinute=e.rateLimitPerMinute,this.tokens=this.rateLimitPerMinute??Number.POSITIVE_INFINITY}enqueue(e,t){const i=`${e.priority}:${e.path}`;this.latestByKey.set(i,{event:e,fn:t});const r=this.debounceTimers.get(i);r&&clearTimeout(r);const o=setTimeout((()=>{this.debounceTimers.delete(i);const e=this.latestByKey.get(i);e&&(this.latestByKey.delete(i),this.push(e),this.pump())}),this.debounceMs);this.debounceTimers.set(i,o)}process(){this.started=!0,this.pump()}async drain(){this.isIdle()||await new Promise((e=>{this.drainWaiters.push(e)}))}push(e){"low"===e.event.priority?this.lowQueue.push(e):this.normalQueue.push(e)}refillTokens(e){if(void 0===this.rateLimitPerMinute)return;const t=Math.max(0,e-this.lastRefillMs)*(this.rateLimitPerMinute/6e4);this.tokens=Math.min(this.rateLimitPerMinute,this.tokens+t),this.lastRefillMs=e}takeToken(){const e=Date.now();return this.refillTokens(e),!(this.tokens<1)&&(this.tokens-=1,!0)}nextItem(){return this.normalQueue.shift()??this.lowQueue.shift()}pump(){if(this.started){for(;this.active<this.concurrency;){const e=this.nextItem();if(!e)break;if(!this.takeToken()){"low"===e.event.priority?this.lowQueue.unshift(e):this.normalQueue.unshift(e),setTimeout((()=>{this.pump()}),250);break}this.active+=1,Promise.resolve().then((()=>e.fn(e.event))).finally((()=>{this.active-=1,this.pump(),this.maybeResolveDrain()}))}this.maybeResolveDrain()}}isIdle(){return 0===this.active&&0===this.normalQueue.length&&0===this.lowQueue.length&&0===this.debounceTimers.size&&0===this.latestByKey.size}maybeResolveDrain(){if(!this.isIdle())return;const e=this.drainWaiters;this.drainWaiters=[];for(const t of e)t()}}class De{client;collectionName;dims;log;constructor(e,t,i){this.client=new M.QdrantClient({url:e.url,apiKey:e.apiKey,checkCompatibility:!1}),this.collectionName=e.collectionName,this.dims=t,this.log=ee(i)}async ensureCollection(){try{const e=await this.client.getCollections();e.collections.some((e=>e.name===this.collectionName))||await this.client.createCollection(this.collectionName,{vectors:{size:this.dims,distance:"Cosine"}})}catch(e){throw new Error(`Failed to ensure collection "${this.collectionName}": ${String(e)}`)}}async upsert(e){0!==e.length&&await re((async t=>{t>1&&this.log.warn({attempt:t,operation:"qdrant.upsert",points:e.length},"Retrying Qdrant upsert"),await this.client.upsert(this.collectionName,{wait:!0,points:e.map((e=>({id:e.id,vector:e.vector,payload:e.payload})))})}),{attempts:5,baseDelayMs:500,maxDelayMs:1e4,jitter:.2,onRetry:({attempt:e,delayMs:t,error:i})=>{this.log.warn({attempt:e,delayMs:t,operation:"qdrant.upsert",err:P(i)},"Qdrant upsert failed; will retry")}})}async delete(e){0!==e.length&&await re((async t=>{t>1&&this.log.warn({attempt:t,operation:"qdrant.delete",ids:e.length},"Retrying Qdrant delete"),await this.client.delete(this.collectionName,{wait:!0,points:e})}),{attempts:5,baseDelayMs:500,maxDelayMs:1e4,jitter:.2,onRetry:({attempt:e,delayMs:t,error:i})=>{this.log.warn({attempt:e,delayMs:t,operation:"qdrant.delete",err:P(i)},"Qdrant delete failed; will retry")}})}async setPayload(e,t){0!==e.length&&await this.client.setPayload(this.collectionName,{wait:!0,points:e,payload:t})}async getPayload(e){try{const t=await this.client.retrieve(this.collectionName,{ids:[e],with_payload:!0,with_vector:!1});return 0===t.length?null:t[0].payload}catch{return null}}async search(e,t,i){return(await this.client.search(this.collectionName,{vector:e,limit:t,with_payload:!0,...i?{filter:i}:{}})).map((e=>({id:String(e.id),score:e.score,payload:e.payload})))}async*scroll(e,t=100){let i;for(;;){const r=await this.client.scroll(this.collectionName,{limit:t,with_payload:!0,with_vector:!1,...e?{filter:e}:{},...void 0!==i?{offset:i}:{}});for(const e of r.points)yield{id:String(e.id),payload:e.payload};const o=r.next_page_offset;if(null==o)break;if("string"!=typeof o&&"number"!=typeof o)break;i=o}}}class Te{consecutiveFailures=0;maxRetries;maxBackoffMs;baseDelayMs;onFatalError;logger;constructor(e){this.maxRetries=e.maxRetries??Number.POSITIVE_INFINITY,this.maxBackoffMs=e.maxBackoffMs??6e4,this.baseDelayMs=e.baseDelayMs??1e3,this.onFatalError=e.onFatalError,this.logger=e.logger}recordSuccess(){this.consecutiveFailures>0&&this.logger.info({previousFailures:this.consecutiveFailures},"System health recovered"),this.consecutiveFailures=0}recordFailure(e){if(this.consecutiveFailures+=1,this.logger.error({consecutiveFailures:this.consecutiveFailures,maxRetries:this.maxRetries,err:P(e)},"System-level failure recorded"),this.consecutiveFailures>=this.maxRetries){if(this.logger.fatal({consecutiveFailures:this.consecutiveFailures},"Maximum retries exceeded, triggering fatal error"),this.onFatalError)return this.onFatalError(e),!1;throw e instanceof Error?e:new Error(`Fatal system error: ${String(e)}`)}return!0}get currentBackoffMs(){if(0===this.consecutiveFailures)return 0;const e=Math.max(0,this.consecutiveFailures-1);return Math.min(this.maxBackoffMs,this.baseDelayMs*2**e)}async backoff(e){const t=this.currentBackoffMs;t<=0||(this.logger.warn({delayMs:t,consecutiveFailures:this.consecutiveFailures},"Backing off before next attempt"),await new Promise(((i,r)=>{const o=setTimeout((()=>{s(),i()}),t),n=()=>{s(),r(new Error("Backoff aborted"))},s=()=>{clearTimeout(o),e&&e.removeEventListener("abort",n)};if(e){if(e.aborted)return void n();e.addEventListener("abort",n,{once:!0})}})))}get failures(){return this.consecutiveFailures}}class Ie{config;queue;processor;logger;health;gitignoreFilter;watcher;constructor(e,t,i,r,o={}){this.config=e,this.queue=t,this.processor=i,this.logger=r,this.gitignoreFilter=o.gitignoreFilter;const n={maxRetries:o.maxRetries,maxBackoffMs:o.maxBackoffMs,onFatalError:o.onFatalError,logger:r};this.health=new Te(n)}start(){this.watcher=x.watch(this.config.paths,{ignored:this.config.ignored,usePolling:this.config.usePolling,interval:this.config.pollIntervalMs,awaitWriteFinish:!!this.config.stabilityThresholdMs&&{stabilityThreshold:this.config.stabilityThresholdMs},ignoreInitial:!1}),this.watcher.on("add",(e=>{this.handleGitignoreChange(e),this.isGitignored(e)||(this.logger.debug({path:e},"File added"),this.queue.enqueue({type:"create",path:e,priority:"normal"},(()=>this.wrapProcessing((()=>this.processor.processFile(e))))))})),this.watcher.on("change",(e=>{this.handleGitignoreChange(e),this.isGitignored(e)||(this.logger.debug({path:e},"File changed"),this.queue.enqueue({type:"modify",path:e,priority:"normal"},(()=>this.wrapProcessing((()=>this.processor.processFile(e))))))})),this.watcher.on("unlink",(e=>{this.handleGitignoreChange(e),this.isGitignored(e)||(this.logger.debug({path:e},"File removed"),this.queue.enqueue({type:"delete",path:e,priority:"normal"},(()=>this.wrapProcessing((()=>this.processor.deleteFile(e))))))})),this.watcher.on("error",(e=>{this.logger.error({err:P(e)},"Watcher error"),this.health.recordFailure(e)})),this.queue.process(),this.logger.info({paths:this.config.paths},"Filesystem watcher started")}async stop(){this.watcher&&(await this.watcher.close(),this.watcher=void 0,this.logger.info("Filesystem watcher stopped"))}get systemHealth(){return this.health}isGitignored(e){if(!this.gitignoreFilter)return!1;const t=this.gitignoreFilter.isIgnored(e);return t&&this.logger.debug({path:e},"Skipping gitignored file"),t}handleGitignoreChange(e){this.gitignoreFilter&&e.endsWith(".gitignore")&&(this.logger.info({path:e},"Gitignore file changed, refreshing filter"),this.gitignoreFilter.invalidate(e))}async wrapProcessing(e){try{await this.health.backoff(),await e(),this.health.recordSuccess()}catch(e){this.health.recordFailure(e)||await this.stop()}}}class We{options;watcher;debounce;constructor(e){this.options=e}start(){this.options.enabled&&(this.watcher=x.watch(this.options.configPath,{ignoreInitial:!0}),this.watcher.on("change",(()=>{this.debounce&&clearTimeout(this.debounce),this.debounce=setTimeout((()=>{this.options.onChange()}),this.options.debounceMs)})),this.watcher.on("error",(e=>{this.options.logger.error({err:P(e)},"Config watcher error")})),this.options.logger.info({configPath:this.options.configPath,debounceMs:this.options.debounceMs},"Config watcher started"))}async stop(){this.debounce&&(clearTimeout(this.debounce),this.debounce=void 0),this.watcher&&(await this.watcher.close(),this.watcher=void 0)}}const Ne={loadConfig:X,createLogger:de,createEmbeddingProvider:ne,createVectorStoreClient:(e,t,i)=>new De(e,t,i),compileRules:ke,createDocumentProcessor:(e,t,i,r,o)=>new Ee(e,t,i,r,o),createEventQueue:e=>new Ce(e),createFileSystemWatcher:(e,t,i,r,o)=>new Ie(e,t,i,r,o),createApiServer:W};class Oe{config;configPath;factories;runtimeOptions;logger;watcher;queue;server;processor;configWatcher;constructor(e,t,i={},r={}){this.config=e,this.configPath=t,this.factories={...Ne,...i},this.runtimeOptions=r}async start(){const e=this.factories.createLogger(this.config.logging);let t;this.logger=e;try{t=this.factories.createEmbeddingProvider(this.config.embedding,e)}catch(t){throw e.fatal({err:P(t)},"Failed to create embedding provider"),t}const i=this.factories.createVectorStoreClient(this.config.vectorStore,t.dimensions,e);await i.ensureCollection();const r=this.factories.compileRules(this.config.inferenceRules??[]),o={metadataDir:this.config.metadataDir??".jeeves-metadata",chunkSize:this.config.embedding.chunkSize,chunkOverlap:this.config.embedding.chunkOverlap,maps:this.config.maps},n=this.factories.createDocumentProcessor(o,t,i,r,e);this.processor=n;const s=this.factories.createEventQueue({debounceMs:this.config.watch.debounceMs??2e3,concurrency:this.config.embedding.concurrency??5,rateLimitPerMinute:this.config.embedding.rateLimitPerMinute});this.queue=s;const a=this.config.watch.respectGitignore??!0?new he(this.config.watch.paths):void 0,c=this.factories.createFileSystemWatcher(this.config.watch,s,n,e,{maxRetries:this.config.maxRetries,maxBackoffMs:this.config.maxBackoffMs,onFatalError:this.runtimeOptions.onFatalError,gitignoreFilter:a});this.watcher=c;const l=this.factories.createApiServer({processor:n,vectorStore:i,embeddingProvider:t,queue:s,config:this.config,logger:e});this.server=l,await l.listen({host:this.config.api?.host??"127.0.0.1",port:this.config.api?.port??3456}),c.start(),this.startConfigWatch(),e.info("jeeves-watcher started")}async stop(){if(await this.stopConfigWatch(),this.watcher&&await this.watcher.stop(),this.queue){const e=this.config.shutdownTimeoutMs??1e4;await Promise.race([this.queue.drain().then((()=>!0)),new Promise((t=>{setTimeout((()=>{t(!1)}),e)}))])||this.logger?.warn({timeoutMs:e},"Queue drain timeout hit, forcing shutdown")}this.server&&await this.server.close(),this.logger?.info("jeeves-watcher stopped")}startConfigWatch(){const e=this.logger;if(!e)return;const t=this.config.configWatch?.enabled??!0;if(!t)return;if(!this.configPath)return void e.debug("Config watch enabled, but no config path was provided");const i=this.config.configWatch?.debounceMs??1e4;this.configWatcher=new We({configPath:this.configPath,enabled:t,debounceMs:i,logger:e,onChange:async()=>this.reloadConfig()}),this.configWatcher.start()}async stopConfigWatch(){this.configWatcher&&(await this.configWatcher.stop(),this.configWatcher=void 0)}async reloadConfig(){const e=this.logger,t=this.processor;if(e&&t&&this.configPath){e.info({configPath:this.configPath},"Config change detected, reloading...");try{const i=await this.factories.loadConfig(this.configPath);this.config=i;const r=this.factories.compileRules(i.inferenceRules??[]);t.updateRules(r),e.info({configPath:this.configPath,rules:r.length},"Config reloaded")}catch(t){e.error({err:P(t)},"Failed to reload config")}}}}e.DocumentProcessor=Ee,e.EventQueue=Ce,e.FileSystemWatcher=Ie,e.GitignoreFilter=he,e.JeevesWatcher=Oe,e.SystemHealth=Te,e.VectorStoreClient=De,e.apiConfigSchema=K,e.applyRules=Fe,e.buildAttributes=Pe,e.compileRules=ke,e.configWatchConfigSchema=$,e.contentHash=ue,e.createApiServer=W,e.createEmbeddingProvider=ne,e.createLogger=de,e.deleteMetadata=T,e.embeddingConfigSchema=G,e.extractText=Me,e.inferenceRuleSchema=V,e.jeevesWatcherConfigSchema=U,e.loadConfig=X,e.loggingConfigSchema=J,e.metadataPath=E,e.pointId=fe,e.readMetadata=C,e.startFromConfig=async function(e){const t=await X(e),i=new Oe(t,e);return function(e){const t=async()=>{await e(),process.exit(0)};process.on("SIGTERM",(()=>{t()})),process.on("SIGINT",(()=>{t()}))}((()=>i.stop())),await i.start(),i},e.vectorStoreConfigSchema=B,e.watchConfigSchema=Q,e.writeMetadata=D}(this["jeeves-watcher"]=this["jeeves-watcher"]||{},Fastify,promises,node_path,picomatch,radash,node_crypto,cosmiconfig,zod,jsonmap,googleGenai,node_fs,ignore,pino,uuid,cheerio,yaml,mammoth,Ajv,addFormats,textsplitters,jsClientRest,chokidar);