2.0.5

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@apiclient.xyz/elasticsearch",
-  "version": "1.0.43",
+  "version": "2.0.5",
   "private": false,
   "description": "log to elasticsearch in a kibana compatible format",
   "main": "dist_ts/index.js",

test/test.nonci.ts

@@ -2,10 +2,11 @@ import { expect, tap } from '@pushrocks/tapbundle';
 import { Qenv } from '@pushrocks/qenv';
 import * as elasticsearch from '../ts/index.js';
 
-let testElasticLog: elasticsearch.ElasticSearch<any>;
+let testElasticLog: elasticsearch.ElsSmartlogDestination<any>;
+let testElasticDoc: elasticsearch.ElasticDoc;
 
 tap.test('first test', async () => {
-  testElasticLog = new elasticsearch.ElasticSearch({
+  testElasticLog = new elasticsearch.ElsSmartlogDestination({
     indexPrefix: 'testprefix',
     indexRetention: 7,
     node: 'http://localhost:9200',
@@ -14,7 +15,7 @@ tap.test('first test', async () => {
       password: 'YourPassword'
     }
   });
-  expect(testElasticLog).toBeInstanceOf(elasticsearch.ElasticSearch);
+  expect(testElasticLog).toBeInstanceOf(elasticsearch.ElsSmartlogDestination);
 });
 
 tap.test('should send a message to Elasticsearch', async () => {
@@ -35,4 +36,54 @@ tap.test('should send a message to Elasticsearch', async () => {
   });
 });
 
+tap.test('should create an ElasticDoc instance', async () => {
+  testElasticDoc = new elasticsearch.ElasticDoc({
+    index: 'testindex',
+    node: 'http://localhost:9200',
+    auth: {
+      username: 'elastic',
+      password: 'YourPassword'
+    }
+  });
+  expect(testElasticDoc).toBeInstanceOf(elasticsearch.ElasticDoc);
+});
+
+tap.test('should add and update documents in a piping session', async () => {
+  await testElasticDoc.startPipingSession({});
+  await testElasticDoc.pipeDocument({
+    docId: '1',
+    timestamp: new Date().toISOString(),
+    doc: { name: 'doc1' }
+  });
+  await testElasticDoc.pipeDocument({
+    docId: '2',
+    timestamp: new Date().toISOString(),
+    doc: { name: 'doc2' }
+  });
+  await testElasticDoc.pipeDocument({
+    docId: '1',
+    timestamp: new Date().toISOString(),
+    doc: { name: 'updated doc1' }
+  });
+});
+
+tap.test('should delete documents not part of the piping session', async () => {
+  await testElasticDoc.endPipingSession();
+});
+
+tap.test('should take and store snapshot', async () => {
+  await testElasticDoc.takeSnapshot(async (iterator, prevSnapshot) => {
+    const aggregationData = [];
+    for await (const doc of iterator) {
+      // Sample aggregation: counting documents
+      aggregationData.push(doc);
+    }
+    const snapshot = {
+      date: new Date().toISOString(),
+      aggregationData,
+    };
+    return snapshot;
+  });
+});
+
 tap.start();

ts/00_commitinfo_data.ts

@@ -3,6 +3,6 @@
  */
 export const commitinfo = {
   name: '@apiclient.xyz/elasticsearch',
-  version: '1.0.43',
+  version: '2.0.5',
   description: 'log to elasticsearch in a kibana compatible format'
 }

ts/elasticsearch.classes.elasticscheduler.ts

@@ -1,44 +0,0 @@
-import { ElasticSearch, type IStandardLogParams } from './elasticsearch.classes.elasticsearch.js';
-
-export class ElasticScheduler {
-  elasticSearchRef: ElasticSearch<any>;
-  docsScheduled = false;
-  docsStorage: any[] = [];
-
-  constructor(elasticLogRefArg: ElasticSearch<any>) {
-    this.elasticSearchRef = elasticLogRefArg;
-  }
-
-  public addFailedDoc(objectArg: any | IStandardLogParams) {
-    this.docsStorage.push(objectArg);
-    this.setRetry();
-  }
-  public scheduleDoc(logObject: any) {
-    this.docsStorage.push(logObject);
-  }
-
-  public setRetry() {
-    setTimeout(() => {
-      const oldStorage = this.docsStorage;
-      this.docsStorage = [];
-      for (let logObject of oldStorage) {
-        this.elasticSearchRef.log(logObject, true);
-      }
-      if (this.docsStorage.length === 0) {
-        console.log('ElasticLog retry success!!!');
-        this.docsScheduled = false;
-      } else {
-        console.log('ElasticLog retry failed');
-        this.setRetry();
-      }
-    }, 5000);
-  }
-
-  public deferSend() {
-    if (!this.docsScheduled) {
-      console.log('Retry ElasticLog in 5 seconds!');
-      this.docsScheduled = true;
-      this.setRetry();
-    }
-  }
-}

ts/els.classes.elasticdoc.ts

@@ -0,0 +1,250 @@
+import { Client as ElasticClient } from '@elastic/elasticsearch';
+
+export interface IElasticDocConstructorOptions {
+  index: string;
+  node: string;
+  auth?: {
+    username: string;
+    password: string;
+  };
+}
+
+export interface ISnapshot {
+  date: string;
+  aggregationData: any;
+}
+
+export type SnapshotProcessor = (
+  iterator: AsyncIterable<any>,
+  prevSnapshot: ISnapshot | null
+) => Promise<ISnapshot>;
+
+export class ElasticDoc {
+  public client: ElasticClient;
+  public index: string;
+  private sessionDocs: Set<string> = new Set();
+  private indexInitialized: boolean = false;
+  private latestTimestamp: string | null = null; // Store the latest timestamp
+  private onlyNew: boolean = false; // Whether to only pipe new docs
+
+  private BATCH_SIZE = 1000;
+
+  constructor(options: IElasticDocConstructorOptions) {
+    this.client = new ElasticClient({
+      node: options.node,
+      ...(options.auth && { auth: options.auth }),
+    });
+    this.index = options.index;
+  }
+
+  private async ensureIndexExists(doc: any) {
+    if (!this.indexInitialized) {
+      const { body: indexExists } = await this.client.indices.exists({ index: this.index });
+      if (!indexExists) {
+        const mappings = this.createMappingsFromDoc(doc);
+        await this.client.indices.create({
+          index: this.index,
+          body: {
+            mappings,
+            settings: {
+              // You can define the settings according to your requirements here
+            },
+          },
+        });
+      }
+      this.indexInitialized = true;
+    }
+  }
+
+  private createMappingsFromDoc(doc: any): any {
+    const properties: any = {};
+    for (const key in doc) {
+      if (key === '@timestamp') {
+        properties[key] = { type: 'date' };
+        continue;
+      }
+      properties[key] = { type: typeof doc[key] === 'number' ? 'float' : 'text' };
+    }
+    return { properties };
+  }
+
+  async startPipingSession(options: { onlyNew?: boolean }) {
+    this.sessionDocs.clear();
+    this.onlyNew = options.onlyNew;
+  
+    if (this.onlyNew) {
+      try {
+        const response = await this.client.search({
+          index: this.index,
+          sort: '@timestamp:desc',
+          size: 1,
+        });
+  
+        // If the search query succeeded, the index exists.
+        const hit = response.body.hits.hits[0];
+        this.latestTimestamp = hit?._source?.['@timestamp'] || null;
+  
+        if (this.latestTimestamp) {
+          console.log(`Working in "onlyNew" mode. Hence we are omitting documents prior to ${this.latestTimestamp}`);
+        } else {
+          console.log(`Working in "onlyNew" mode, but no documents found in index ${this.index}. Hence processing all documents now.`);
+        }
+      } catch (e) {
+        // If the search query failed, the index likely doesn't exist or some other error occurred.
+        if (e.meta && e.meta.statusCode === 404) {
+          console.log(`Index ${this.index} does not exist. Working in "onlyNew" mode, but will process all documents as the index is empty.`);
+        } else {
+          console.log(`An error occurred while trying to retrieve the latest timestamp: ${e}`);
+        }
+  
+        this.latestTimestamp = null;
+      }
+    }
+  }
+  
+  
+
+  async pipeDocument(optionsArg: { docId: string; timestamp?: string | number; doc: any }) {
+    await this.ensureIndexExists(optionsArg.doc);
+
+    const documentBody = {
+      ...optionsArg.doc,
+      ...(optionsArg.timestamp && { '@timestamp': optionsArg.timestamp }),
+    };
+
+    // If 'onlyNew' is true, compare the document timestamp with the latest timestamp
+    if (this.onlyNew) {
+      if (this.latestTimestamp && optionsArg.timestamp <= this.latestTimestamp) {
+        // Omit the document
+        return;
+      } else {
+        await this.client.index({
+          index: this.index,
+          id: optionsArg.docId,
+          body: documentBody,
+        });
+      }
+    }
+    this.sessionDocs.add(optionsArg.docId);
+  }
+
+  async endPipingSession() {
+    const allDocIds: string[] = [];
+    const responseQueue = [];
+    let response = await this.client.search({
+      index: this.index,
+      scroll: '1m',
+      size: this.BATCH_SIZE,
+    });
+    while (true) {
+      response.body.hits.hits.forEach((hit: any) => allDocIds.push(hit._id));
+      if (!response.body.hits.hits.length) {
+        break;
+      }
+      response = await this.client.scroll({ scroll_id: response.body._scroll_id, scroll: '1m' });
+    }
+
+    for (const docId of allDocIds) {
+      if (!this.sessionDocs.has(docId)) {
+        responseQueue.push({
+          delete: {
+            _index: this.index,
+            _id: docId,
+          },
+        });
+
+        if (responseQueue.length >= this.BATCH_SIZE) {
+          await this.client.bulk({ refresh: true, body: responseQueue });
+          responseQueue.length = 0;
+        }
+      }
+    }
+
+    if (responseQueue.length > 0) {
+      await this.client.bulk({ refresh: true, body: responseQueue });
+    }
+
+    this.sessionDocs.clear();
+  }
+
+  async takeSnapshot(processIterator: SnapshotProcessor) {
+    const snapshotIndex = `${this.index}_snapshots`;
+
+    const { body: indexExists } = await this.client.indices.exists({ index: snapshotIndex });
+    if (!indexExists) {
+      await this.client.indices.create({
+        index: snapshotIndex,
+        body: {
+          mappings: {
+            properties: {
+              date: {
+                type: 'date',
+              },
+              aggregationData: {
+                type: 'object',
+                enabled: true,
+              },
+            },
+          },
+        },
+      });
+    }
+
+    const documentIterator = this.getDocumentIterator();
+
+    const newSnapshot = await processIterator(documentIterator, await this.getLastSnapshot());
+
+    await this.storeSnapshot(newSnapshot);
+  }
+
+  private async getLastSnapshot(): Promise<ISnapshot | null> {
+    const snapshotIndex = `${this.index}_snapshots`;
+    const { body: indexExists } = await this.client.indices.exists({ index: snapshotIndex });
+
+    if (!indexExists) {
+      return null;
+    }
+
+    const response = await this.client.search({
+      index: snapshotIndex,
+      sort: 'date:desc',
+      size: 1,
+    });
+
+    if (response.body.hits.hits.length > 0) {
+      const hit = response.body.hits.hits[0];
+      return {
+        date: hit._source.date,
+        aggregationData: hit._source.aggregationData,
+      };
+    } else {
+      return null;
+    }
+  }
+
+  private async *getDocumentIterator() {
+    let response = await this.client.search({
+      index: this.index,
+      scroll: '1m',
+      size: this.BATCH_SIZE,
+    });
+    while (true) {
+      for (const hit of response.body.hits.hits) {
+        yield hit._source;
+      }
+
+      if (!response.body.hits.hits.length) {
+        break;
+      }
+
+      response = await this.client.scroll({ scroll_id: response.body._scroll_id, scroll: '1m' });
+    }
+  }
+
+  private async storeSnapshot(snapshot: ISnapshot) {
+    await this.client.index({
+      index: `${this.index}_snapshots`,
+      body: snapshot,
+    });
+  }
+}

ts/els.classes.elasticindex.ts

@@ -1,21 +1,21 @@
-import * as plugins from './elasticsearch.plugins.js';
-import { ElasticSearch } from './elasticsearch.classes.elasticsearch.js';
+import * as plugins from './els.plugins.js';
+import { ElsSmartlogDestination } from './els.classes.smartlogdestination.js';
 import { type ILogPackage } from '@pushrocks/smartlog-interfaces';
-
 import { Stringmap } from '@pushrocks/lik';
 
 export class ElasticIndex {
   private stringmap = new Stringmap();
-  private elasticSearchRef: ElasticSearch<any>;
+  private elasticSearchRef: ElsSmartlogDestination<any>;
 
-  constructor(elasticSearchInstanceArg: ElasticSearch<ILogPackage>) {
+  constructor(elasticSearchInstanceArg: ElsSmartlogDestination<ILogPackage>) {
     this.elasticSearchRef = elasticSearchInstanceArg;
   }
 
   public async ensureIndex(prefixArg: string, indexNameArg: string) {
     if (this.stringmap.checkString(indexNameArg)) {
-      return;
+      return indexNameArg;
     }
+
     const responseArg = await this.elasticSearchRef.client.cat.indices({
       format: 'json',
       bytes: 'm',
@@ -27,7 +27,6 @@ export class ElasticIndex {
       throw new Error('Could not get valid response from elastic search');
     }
 
-    // lets delete indexes that violate the retention
     if (Array.isArray(responseArg.body)) {
       const filteredIndices = responseArg.body.filter((indexObjectArg) => {
         return indexObjectArg.index.startsWith(prefixArg);
@@ -49,6 +48,7 @@ export class ElasticIndex {
     if (!index) {
       await this.createNewIndex(indexNameArg);
     }
+
     this.stringmap.addString(indexNameArg);
     return index;
   }
@@ -57,6 +57,23 @@ export class ElasticIndex {
     const response = await this.elasticSearchRef.client.indices.create({
       wait_for_active_shards: '1',
       index: indexNameArg,
+      body: {
+        mappings: {
+          properties: {
+            '@timestamp': {
+              type: 'date',
+            },
+            logPackageArg: {
+              properties: {
+                payload: {
+                  type: 'object',
+                  dynamic: true
+                }
+              }
+            },
+          },
+        },
+      },
     });
   }
 

ts/els.classes.elasticscheduler.ts

@@ -0,0 +1,63 @@
+import { ElsSmartlogDestination, type IStandardLogParams } from './els.classes.smartlogdestination.js';
+
+export class ElasticScheduler {
+  elasticSearchRef: ElsSmartlogDestination<any>;
+  docsScheduled = false;
+  docsStorage: any[] = [];
+
+  // maximum size of the buffer
+  maxBufferSize = 500;
+
+  constructor(elasticLogRefArg: ElsSmartlogDestination<any>) {
+    this.elasticSearchRef = elasticLogRefArg;
+  }
+
+  public addFailedDoc(objectArg: any | IStandardLogParams) {
+    this.addToStorage(objectArg);
+    this.setRetry();
+  }
+  
+  public scheduleDoc(logObject: any) {
+    this.addToStorage(logObject);
+  }
+
+  private addToStorage(logObject: any) {
+    this.docsStorage.push(logObject);
+    
+    // if buffer is full, send logs immediately
+    if (this.docsStorage.length >= this.maxBufferSize) {
+      this.flushLogsToElasticSearch();
+    }
+  }
+
+  private flushLogsToElasticSearch() {
+    const oldStorage = this.docsStorage;
+    this.docsStorage = [];
+    
+    for (let logObject of oldStorage) {
+      this.elasticSearchRef.log(logObject, true);
+    }
+  }
+
+  public setRetry() {
+    setTimeout(() => {
+      this.flushLogsToElasticSearch();
+
+      if (this.docsStorage.length === 0) {
+        console.log('ElasticLog retry success!!!');
+        this.docsScheduled = false;
+      } else {
+        console.log('ElasticLog retry failed');
+        this.setRetry();
+      }
+    }, 5000);
+  }
+
+  public deferSend() {
+    if (!this.docsScheduled) {
+      console.log('Retry ElasticLog in 5 seconds!');
+      this.docsScheduled = true;
+      this.setRetry();
+    }
+  }
+}

ts/els.classes.fastpush.ts

@@ -0,0 +1,68 @@
+import { Client as ElasticClient } from '@elastic/elasticsearch';
+
+interface FastPushOptions {
+  deleteOldData?: boolean;     // Clear the index
+  deleteIndex?: boolean;       // Delete the entire index
+}
+
+export class FastPush {
+  private client: ElasticClient;
+
+  constructor(node: string, auth?: { username: string; password: string }) {
+    this.client = new ElasticClient({
+      node: node,
+      ...(auth && { auth: auth }),
+    });
+  }
+
+  async pushToIndex(indexName: string, docArray: any[], options?: FastPushOptions) {
+    if (docArray.length === 0) return;
+
+    const { body: indexExists } = await this.client.indices.exists({ index: indexName });
+
+    if (indexExists) {
+      if (options?.deleteIndex) {
+        await this.client.indices.delete({ index: indexName });
+      } else if (options?.deleteOldData) {
+        await this.client.deleteByQuery({
+          index: indexName,
+          body: {
+            query: {
+              match_all: {}
+            }
+          }
+        });
+      }
+    }
+
+    if (!indexExists || options?.deleteIndex) {
+      // Create index with mappings (for simplicity, we use dynamic mapping)
+      await this.client.indices.create({
+        index: indexName,
+        body: {
+          mappings: {
+            dynamic: "true"
+            // ... other specific mappings
+          },
+        },
+      });
+    }
+
+    // Bulk insert documents
+    const bulkBody = [];
+    for (const doc of docArray) {
+      bulkBody.push({
+        index: {
+          _index: indexName,
+        },
+      });
+      bulkBody.push(doc);
+    }
+
+    await this.client.bulk({ body: bulkBody });
+  }
+}
+
+// Usage example:
+// const fastPush = new FastPush('http://localhost:9200', { username: 'elastic', password: 'password' });
+// fastPush.pushToIndex('my_index', [{ name: 'John', age: 30 }, { name: 'Jane', age: 25 }], { deleteOldData: true });

ts/els.classes.kvstore.ts

@@ -0,0 +1,111 @@
+import * as plugins from './els.plugins.js';
+import { Client as ElasticClient } from '@elastic/elasticsearch';
+
+export interface IElasticKVStoreConstructorOptions {
+  index: string;
+  node: string;
+  auth?: {
+    username: string;
+    password: string;
+  };
+}
+
+export class ElasticKVStore {
+  public client: ElasticClient;
+  public index: string;
+  private readyDeferred: any;
+
+  constructor(options: IElasticKVStoreConstructorOptions) {
+    this.client = new ElasticClient({
+      node: options.node,
+      ...(options.auth && { auth: options.auth }),
+    });
+    this.index = options.index;
+    this.readyDeferred = plugins.smartpromise.defer();
+    this.setupIndex();
+  }
+
+  private async setupIndex() {
+    try {
+      const { body: indexExists } = await this.client.indices.exists({ index: this.index });
+
+      if (!indexExists) {
+        await this.client.indices.create({ 
+          index: this.index,
+          body: {
+            mappings: {
+              properties: {
+                key: {
+                  type: 'keyword'
+                },
+                value: {
+                  type: 'text'
+                }
+              }
+            }
+          }
+        });
+      }
+      this.readyDeferred.resolve();
+    } catch (err) {
+      this.readyDeferred.reject(err);
+    }
+  }
+
+  async set(key: string, value: string) {
+    await this.readyDeferred.promise;
+    await this.client.index({
+      index: this.index,
+      id: key,
+      body: {
+        key,
+        value
+      }
+    });
+  }
+
+  async get(key: string): Promise<string | null> {
+    await this.readyDeferred.promise;
+
+    try {
+      const response = await this.client.get({
+        index: this.index,
+        id: key
+      });
+      return response.body._source.value;
+    } catch (error) {
+      if (error.meta && error.meta.statusCode === 404) {
+        return null;
+      }
+      throw error;
+    }
+  }
+
+  async delete(key: string) {
+    await this.readyDeferred.promise;
+
+    try {
+      await this.client.delete({
+        index: this.index,
+        id: key
+      });
+    } catch (error) {
+      if (error.meta && error.meta.statusCode !== 404) {
+        throw error;
+      }
+    }
+  }
+
+  async clear() {
+    await this.readyDeferred.promise;
+
+    await this.client.deleteByQuery({
+      index: this.index,
+      body: {
+        query: {
+          match_all: {}
+        }
+      }
+    });
+  }
+}

ts/els.classes.smartlogdestination.ts

@@ -3,8 +3,8 @@ import { Client as ElasticClient } from '@elastic/elasticsearch';
 import type { ILogContext, ILogPackage, ILogDestination } from '@pushrocks/smartlog-interfaces';
 
 // other classes
-import { ElasticScheduler } from './elasticsearch.classes.elasticscheduler.js';
-import { ElasticIndex } from './elasticsearch.classes.elasticindex.js';
+import { ElasticScheduler } from './els.classes.elasticscheduler.js';
+import { ElasticIndex } from './els.classes.elasticindex.js';
 
 export interface IStandardLogParams {
   message: string;
@@ -21,7 +21,7 @@ export interface IElasticSearchConstructorOptions {
   };
 }
 
-export class ElasticSearch<T> {
+export class ElsSmartlogDestination<T> {
   public client: ElasticClient;
   public elasticScheduler = new ElasticScheduler(this);
   public elasticIndex: ElasticIndex = new ElasticIndex(this);
@@ -38,7 +38,7 @@ export class ElasticSearch<T> {
       node: optionsArg.node,
       ...(optionsArg.auth && { auth: optionsArg.auth }),
     });
-    this.indexPrefix = optionsArg.indexPrefix;
+    this.indexPrefix = `${optionsArg.indexPrefix}`;
     this.indexRetention = optionsArg.indexRetention;
   }
 
@@ -51,8 +51,31 @@ export class ElasticSearch<T> {
       return;
     }
 
-    const response = await this.elasticIndex.ensureIndex(this.indexPrefix, indexToUse);
-    console.log(response);
+    // Make sure the index is created with a mapping for dynamic JSON
+    const indexExists = await this.client.indices.exists({ index: indexToUse });
+    if (!indexExists.body) {
+      await this.client.indices.create({
+        index: indexToUse,
+        body: {
+          mappings: {
+            properties: {
+              '@timestamp': {
+                type: 'date',
+              },
+              logPackageArg: {
+                properties: {
+                  payload: {
+                    type: 'object',
+                    dynamic: true
+                  }
+                }
+              },
+            },
+          },
+        },
+      });
+    }
+
     this.client.index(
       {
         index: indexToUse,

ts/index.ts

@@ -1 +1,4 @@
-export * from './elasticsearch.classes.elasticsearch.js';
+export * from './els.classes.smartlogdestination.js';
+export * from './els.classes.fastpush.js';
+export * from './els.classes.elasticdoc.js';
+export * from './els.classes.kvstore.js';