triangle-utils 1.4.149 → 1.4.150

package/dist/src/UtilsS3.d.ts

@@ -8,7 +8,7 @@ export declare class UtilsS3 {
     private readonly s3;
     constructor(region: string);
     get_client(): S3;
-    get_s3_input(s3_id: string): S3Input | undefined;
+    static get_s3_input(s3_id: string): S3Input | undefined;
     get_headers(s3_id: string): Promise<{
         DeleteMarker?: boolean | undefined;
         AcceptRanges?: string | undefined;

package/dist/src/UtilsS3.js

@@ -10,7 +10,7 @@ export class UtilsS3 {
     get_client() {
         return this.s3;
     }
-    get_s3_input(s3_id) {
+    static get_s3_input(s3_id) {
         const [bucket, path] = (s3_id.match(/^s3\:\/\/([^\/]+)\/([^$]+)$/) || []).slice(1, 3);
         if (bucket === undefined || path === undefined) {
             return undefined;
@@ -18,7 +18,7 @@ export class UtilsS3 {
         return { Bucket: bucket, Key: path };
     }
     async get_headers(s3_id) {
-        const s3_input = this.get_s3_input(s3_id);
+        const s3_input = UtilsS3.get_s3_input(s3_id);
         if (s3_input === undefined) {
             return undefined;
         }
@@ -34,7 +34,7 @@ export class UtilsS3 {
         }
     }
     async get(s3_id, encoding = "utf-8") {
-        const s3_input = this.get_s3_input(s3_id);
+        const s3_input = UtilsS3.get_s3_input(s3_id);
         if (s3_input === undefined) {
             return undefined;
         }
@@ -55,7 +55,7 @@ export class UtilsS3 {
         }
     }
     async get_buffer(s3_id) {
-        const s3_input = this.get_s3_input(s3_id);
+        const s3_input = UtilsS3.get_s3_input(s3_id);
         if (s3_input === undefined) {
             return undefined;
         }
@@ -76,7 +76,7 @@ export class UtilsS3 {
         }
     }
     async get_stream(s3_id) {
-        const s3_input = this.get_s3_input(s3_id);
+        const s3_input = UtilsS3.get_s3_input(s3_id);
         if (s3_input === undefined) {
             return undefined;
         }
@@ -100,7 +100,7 @@ export class UtilsS3 {
     }
     async query_prefix(s3_id_prefix, options = {}) {
         const compile = options.compile !== undefined ? options.compile : false;
-        const s3_input = this.get_s3_input(s3_id_prefix);
+        const s3_input = UtilsS3.get_s3_input(s3_id_prefix);
         if (s3_input === undefined) {
             return [];
         }
@@ -140,7 +140,7 @@ export class UtilsS3 {
         if (headers !== undefined) {
             return undefined;
         }
-        const s3_input = this.get_s3_input(s3_id);
+        const s3_input = UtilsS3.get_s3_input(s3_id);
         if (s3_input === undefined) {
             return undefined;
         }
@@ -153,7 +153,7 @@ export class UtilsS3 {
     async upload(s3_id, content, options = {}) {
         const content_type_id = options.content_type_id !== undefined ? options.content_type_id : undefined;
         const part_size = options.part_size !== undefined ? options.part_size : undefined;
-        const s3_input = this.get_s3_input(s3_id);
+        const s3_input = UtilsS3.get_s3_input(s3_id);
         if (s3_input === undefined) {
             return undefined;
         }
@@ -169,7 +169,7 @@ export class UtilsS3 {
         await upload.done();
     }
     async delete(s3_id) {
-        const s3_input = this.get_s3_input(s3_id);
+        const s3_input = UtilsS3.get_s3_input(s3_id);
         if (s3_input === undefined) {
             return undefined;
         }
@@ -180,7 +180,7 @@ export class UtilsS3 {
         if (headers === undefined) {
             return undefined;
         }
-        const s3_input = this.get_s3_input(s3_id);
+        const s3_input = UtilsS3.get_s3_input(s3_id);
         if (s3_input === undefined) {
             return undefined;
         }
@@ -190,7 +190,7 @@ export class UtilsS3 {
     }
     async get_upload_url(s3_id, options = {}) {
         const content_type_id = options.content_type_id !== undefined ? options.content_type_id : undefined;
-        const s3_input = this.get_s3_input(s3_id);
+        const s3_input = UtilsS3.get_s3_input(s3_id);
         if (s3_input === undefined) {
             return undefined;
         }

package/dist/src/UtilsTextract.d.ts

@@ -0,0 +1,6 @@
+import { Block } from "@aws-sdk/client-textract";
+export declare class UtilsTextract {
+    private readonly textract;
+    constructor(region: string);
+    get_text(s3_id: string): Promise<Block[] | undefined>;
+}

package/dist/src/UtilsTextract.js

@@ -0,0 +1,45 @@
+import { Textract } from "@aws-sdk/client-textract";
+import { UtilsMisc } from "./UtilsMisc.js";
+import { UtilsS3 } from "./UtilsS3.js";
+export class UtilsTextract {
+    textract;
+    constructor(region) {
+        this.textract = new Textract({ region: region });
+    }
+    async get_text(s3_id) {
+        const s3_input = UtilsS3.get_s3_input(s3_id);
+        if (s3_input === undefined) {
+            return undefined;
+        }
+        const request = {
+            DocumentLocation: {
+                S3Object: {
+                    Bucket: s3_input.Bucket,
+                    Name: s3_input.Key
+                }
+            }
+        };
+        const job_id = await this.textract.startDocumentTextDetection(request)
+            .then(response => response.JobId);
+        if (job_id === undefined) {
+            return;
+        }
+        let token = undefined;
+        const blocks = [];
+        while (true) {
+            const result = await this.textract.getDocumentTextDetection({
+                JobId: job_id,
+                NextToken: token
+            });
+            if (result.JobStatus === "IN_PROGRESS") {
+                UtilsMisc.wait(1000);
+                continue;
+            }
+            blocks.push(...(result.Blocks || []));
+            if (result.NextToken === undefined) {
+                return blocks;
+            }
+            token = result.NextToken;
+        }
+    }
+}

package/dist/src/f.js

@@ -77,3 +77,7 @@ const utils = new TriangleUtils(config);
 // await utils.rds.connect()
 // console.log(await utils.rds.scan("web_articles", { is_verbose : true, is_forced : true, max_num_items : 100 }))
 // await utils.rds.disconnect()
+// await utils.twitter.get_user("1509572215251124226")
+//     .then(user => console.log(user))
+// await utils.bedrock.cohere_invoke("cohere.embed-v4", "yyyyyuuuuuurrrrr")
+//     .then(response => console.log(response))

package/dist/src/index.d.ts

@@ -13,12 +13,14 @@ import { UtilsTwitter } from "./UtilsTwitter.js";
 import { GlobalConfig } from "./types/GlobalConfig.js";
 import { UtilsRDS } from "./UtilsRDS.js";
 import { ApplicationConfig } from "./types/ApplicationConfig.js";
+import { UtilsTextract } from "./UtilsTextract.js";
 export declare class TriangleUtils extends UtilsMisc {
     readonly dynamodb: UtilsDynamoDB;
     readonly rds: UtilsRDS;
     readonly s3: UtilsS3;
     readonly s3vectors: UtilsS3Vectors;
     readonly bedrock: UtilsBedrock;
+    readonly textract: UtilsTextract;
     readonly cognito: UtilsCognito;
     readonly ses: UtilsSES;
     readonly bee: UtilsBee;

package/dist/src/index.js

@@ -11,12 +11,14 @@ import { UtilsXAI } from "./UtilsXAI.js";
 import { UtilsAnthropic } from "./UtilsAnthropic.js";
 import { UtilsTwitter } from "./UtilsTwitter.js";
 import { UtilsRDS } from "./UtilsRDS.js";
+import { UtilsTextract } from "./UtilsTextract.js";
 export class TriangleUtils extends UtilsMisc {
     dynamodb;
     rds;
     s3;
     s3vectors;
     bedrock;
+    textract;
     cognito;
     ses;
     bee;
@@ -31,6 +33,7 @@ export class TriangleUtils extends UtilsMisc {
         this.s3 = new UtilsS3(config.region);
         this.s3vectors = new UtilsS3Vectors(config.region);
         this.bedrock = new UtilsBedrock(config.region);
+        this.textract = new UtilsTextract(config.region);
         this.cognito = new UtilsCognito(config.region);
         this.ses = new UtilsSES(config.region);
         this.bee = new UtilsBee(config.scraping_bee_api_key);

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "triangle-utils",
-    "version": "1.4.149",
+    "version": "1.4.150",
     "main": "dist/src/index.js",
     "types": "dist/src/index.d.ts",
     "directories": {
@@ -26,6 +26,7 @@
         "@aws-sdk/client-s3vectors": "^3.953.0",
         "@aws-sdk/client-secrets-manager": "^3.965.0",
         "@aws-sdk/client-ses": "^3.1032.0",
+        "@aws-sdk/client-textract": "^3.1117.0",
         "@aws-sdk/lib-storage": "^3.1106.0",
         "@aws-sdk/s3-request-presigner": "^3.1106.0",
         "@types/jsdom": "^28.0.1",

package/src/UtilsS3.ts

@@ -20,7 +20,7 @@ export class UtilsS3 {
         return this.s3
     }
 
-    get_s3_input(s3_id : string) : S3Input | undefined {
+    static get_s3_input(s3_id : string) : S3Input | undefined {
         const [bucket, path] = (s3_id.match(/^s3\:\/\/([^\/]+)\/([^$]+)$/) || []).slice(1, 3)
         if (bucket === undefined || path === undefined) {
             return undefined
@@ -29,7 +29,7 @@ export class UtilsS3 {
     }
 
     async get_headers(s3_id : string) {
-        const s3_input = this.get_s3_input(s3_id)
+        const s3_input = UtilsS3.get_s3_input(s3_id)
         if (s3_input === undefined) {
             return undefined
         }
@@ -45,7 +45,7 @@ export class UtilsS3 {
     }
 
     async get(s3_id : string, encoding : string = "utf-8") : Promise<string | undefined> {
-        const s3_input = this.get_s3_input(s3_id)
+        const s3_input = UtilsS3.get_s3_input(s3_id)
         if (s3_input === undefined) {
             return undefined
         }
@@ -66,7 +66,7 @@ export class UtilsS3 {
     }
 
     async get_buffer(s3_id : string) : Promise<Buffer | undefined> {
-        const s3_input = this.get_s3_input(s3_id)
+        const s3_input = UtilsS3.get_s3_input(s3_id)
         if (s3_input === undefined) {
             return undefined
         }
@@ -87,7 +87,7 @@ export class UtilsS3 {
     }
 
     async get_stream(s3_id : string) : Promise<Readable | undefined> {
-        const s3_input = this.get_s3_input(s3_id)
+        const s3_input = UtilsS3.get_s3_input(s3_id)
         if (s3_input === undefined) {
             return undefined
         }
@@ -111,7 +111,7 @@ export class UtilsS3 {
 
     async query_prefix(s3_id_prefix : string, options : { compile? : boolean } = {}) {
         const compile = options.compile !== undefined ? options.compile : false
-        const s3_input = this.get_s3_input(s3_id_prefix)
+        const s3_input = UtilsS3.get_s3_input(s3_id_prefix)
         if (s3_input === undefined) {
             return []
         }
@@ -155,7 +155,7 @@ export class UtilsS3 {
         if (headers !== undefined) {
             return undefined
         }
-        const s3_input = this.get_s3_input(s3_id)
+        const s3_input = UtilsS3.get_s3_input(s3_id)
         if (s3_input === undefined) {
             return undefined
         }
@@ -172,7 +172,7 @@ export class UtilsS3 {
     } = {}) : Promise<void> {
         const content_type_id = options.content_type_id !== undefined ? options.content_type_id : undefined
         const part_size = options.part_size !== undefined ? options.part_size : undefined
-        const s3_input = this.get_s3_input(s3_id)
+        const s3_input = UtilsS3.get_s3_input(s3_id)
         if (s3_input === undefined) {
             return undefined
         }
@@ -189,7 +189,7 @@ export class UtilsS3 {
     }
 
     async delete(s3_id : string) : Promise<void>{
-        const s3_input = this.get_s3_input(s3_id)
+        const s3_input = UtilsS3.get_s3_input(s3_id)
         if (s3_input === undefined) {
             return undefined
         }
@@ -201,7 +201,7 @@ export class UtilsS3 {
         if (headers === undefined) {
             return undefined
         }
-        const s3_input = this.get_s3_input(s3_id)
+        const s3_input = UtilsS3.get_s3_input(s3_id)
         if (s3_input === undefined) {
             return undefined
         }
@@ -214,7 +214,7 @@ export class UtilsS3 {
         content_type_id? : string
     } = {}) : Promise<string | undefined> {
         const content_type_id = options.content_type_id !== undefined ? options.content_type_id : undefined
-        const s3_input = this.get_s3_input(s3_id)
+        const s3_input = UtilsS3.get_s3_input(s3_id)
         if (s3_input === undefined) {
             return undefined
         }

package/src/UtilsTextract.ts

@@ -0,0 +1,50 @@
+import { Block, GetDocumentAnalysisCommandOutput, StartDocumentTextDetectionRequest, Textract } from "@aws-sdk/client-textract"
+import { UtilsMisc } from "./UtilsMisc"
+import { UtilsS3 } from "./UtilsS3"
+
+export class UtilsTextract {
+
+    private readonly textract : Textract
+
+    constructor(region : string) {
+        this.textract = new Textract({ region : region })
+    }
+
+    async get_text(s3_id : string) {
+        const s3_input = UtilsS3.get_s3_input(s3_id)
+        if (s3_input === undefined) {
+            return undefined
+        }
+        const request : StartDocumentTextDetectionRequest = {
+            DocumentLocation : {
+                S3Object : {
+                    Bucket : s3_input.Bucket,
+                    Name : s3_input.Key
+                }
+            }
+        }
+        const job_id = await this.textract.startDocumentTextDetection(request)
+            .then(response => response.JobId)
+        if (job_id === undefined) {
+            return
+        }
+        let token : string | undefined = undefined
+        const blocks : Block[] = []
+        while (true) {
+            const result : GetDocumentAnalysisCommandOutput = await this.textract.getDocumentTextDetection({
+                JobId : job_id,
+                NextToken : token
+            })
+            if (result.JobStatus === "IN_PROGRESS") {
+                UtilsMisc.wait(1000)
+                continue
+            }
+            blocks.push(...(result.Blocks || []))
+            if (result.NextToken === undefined) {
+                return blocks
+            }
+            token = result.NextToken
+        }
+    }
+
+}

package/src/f.ts

@@ -95,4 +95,10 @@ const utils = new TriangleUtils(config)
 
 // await utils.rds.connect()
 // console.log(await utils.rds.scan("web_articles", { is_verbose : true, is_forced : true, max_num_items : 100 }))
-// await utils.rds.disconnect()
+// await utils.rds.disconnect()
+
+// await utils.twitter.get_user("1509572215251124226")
+//     .then(user => console.log(user))
+
+// await utils.bedrock.cohere_invoke("cohere.embed-v4", "yyyyyuuuuuurrrrr")
+//     .then(response => console.log(response))

package/src/index.ts

@@ -13,6 +13,7 @@ import { UtilsTwitter } from "./UtilsTwitter"
 import { GlobalConfig } from "./types/GlobalConfig"
 import { UtilsRDS } from "./UtilsRDS"
 import { ApplicationConfig } from "./types/ApplicationConfig"
+import { UtilsTextract } from "./UtilsTextract"
 
 
 
@@ -23,6 +24,7 @@ export class TriangleUtils extends UtilsMisc {
     readonly s3 : UtilsS3
     readonly s3vectors : UtilsS3Vectors
     readonly bedrock : UtilsBedrock
+    readonly textract : UtilsTextract
     readonly cognito : UtilsCognito
     readonly ses : UtilsSES
     readonly bee : UtilsBee
@@ -38,6 +40,7 @@ export class TriangleUtils extends UtilsMisc {
         this.s3 = new UtilsS3(config.region)
         this.s3vectors = new UtilsS3Vectors(config.region)
         this.bedrock = new UtilsBedrock(config.region)
+        this.textract = new UtilsTextract(config.region)
         this.cognito = new UtilsCognito(config.region)
         this.ses = new UtilsSES(config.region)
         this.bee = new UtilsBee(config.scraping_bee_api_key)