@smythos/sre 1.6.8 → 1.6.10

package/src/Components/DataSourceLookup.class.ts

@@ -1,161 +1,161 @@
-import Joi from 'joi';
-import { validateInteger } from '../utils';
-import { jsonrepair } from 'jsonrepair';
-import { TemplateString } from '@sre/helpers/TemplateString.helper';
-import { JSONContent } from '@sre/helpers/JsonContent.helper';
-import { ConnectorService } from '@sre/Core/ConnectorsService';
-import { AccessCandidate } from '@sre/Security/AccessControl/AccessCandidate.class';
-import { IAgent as Agent } from '@sre/types/Agent.types';
-import { Component } from './Component.class';
-
-// Note: LLMHelper renamed to LLMInference
-class LLMInference {
-    static async getInstance(model: string) {
-        throw new Error('Method not implemented.');
-    }
-}
-
-export class DataSourceLookup extends Component {
-    protected configSchema = Joi.object({
-        topK: Joi.alternatives([Joi.string(), Joi.number()]) // Value is now a number; keep string fallback for backward compatibility.
-
-            .custom(validateInteger({ min: 0 }), 'custom range validation')
-            .label('Result Count'),
-        model: Joi.string().valid('gpt-4o-mini', 'gpt-4', 'gpt-3.5-turbo', 'gpt-4', 'gpt-3.5-turbo-16k').optional(),
-        prompt: Joi.string().max(30000).allow('').label('Prompt').optional(),
-        postprocess: Joi.boolean().strict().optional(),
-        includeMetadata: Joi.boolean().strict().optional(),
-        namespace: Joi.string().allow('').max(80).messages({
-            // Need to reserve 30 characters for the prefixed unique id
-            'string.max': `The length of the 'namespace' name must be 50 characters or fewer.`,
-        }),
-        scoreThreshold: Joi.number().optional().label('Score Threshold'),
-        includeScore: Joi.boolean().optional().label('Include Score'),
-    });
-    constructor() {
-        super();
-    }
-    init() {}
-    async process(input, config, agent: Agent) {
-        await super.process(input, config, agent);
-        const componentId = config.id;
-        const component = agent.components[componentId];
-        const teamId = agent.teamId;
-        let debugOutput = agent.agentRuntime?.debug ? '== Data Source Lookup Log ==\n' : null;
-
-        const outputs = {};
-        for (let con of config.outputs) {
-            if (con.default) continue;
-            outputs[con.name] = '';
-        }
-
-        const namespace = config.data.namespace.split('_').slice(1).join('_') || config.data.namespace;
-        const model = config.data?.model || 'gpt-4o-mini';
-        const prompt = config.data?.prompt?.trim?.() || '';
-        const postprocess = config.data?.postprocess || false;
-        const includeMetadata = config.data?.includeMetadata || false;
-
-        const scoreThreshold = config.data?.scoreThreshold || 0.001; // Use low score (0.001) to return most results for backward compatibility
-        const includeScore = config.data?.includeScore || false;
-
-        const _input = typeof input.Query === 'string' ? input.Query : JSON.stringify(input.Query);
-
-        const topK = Math.max(config.data?.topK || 50, 50);
-
-        let vectorDbConnector = ConnectorService.getVectorDBConnector();
-        let existingNs = await vectorDbConnector.requester(AccessCandidate.team(teamId)).namespaceExists(namespace);
-
-        if (!existingNs) {
-            throw new Error(`Namespace ${namespace} does not exist`);
-        }
-
-        let results: string[] | { content: string; metadata: any; score?: number }[];
-        let _error;
-        try {
-            const response = await vectorDbConnector
-                .requester(AccessCandidate.team(teamId))
-                .search(namespace, _input, { topK, includeMetadata: true });
-
-            results = response.slice(0, config.data.topK).map((result) => ({
-                content: result.text,
-                metadata: result.metadata,
-                score: result.score, // use a very low score to return
-            }));
-
-            results = results.filter((result) => result.score >= scoreThreshold);
-
-            // Transform results based on inclusion flags
-            results = results.map((result) => {
-                const transformedResult: any = {
-                    content: result.content,
-                };
-
-                if (includeMetadata) {
-                    // legacy user-specific metadata key [result.metadata?.metadata]
-                    transformedResult.metadata = this.parseMetadata(result.metadata || result.metadata?.metadata);
-                }
-
-                if (includeScore) {
-                    transformedResult.score = result.score;
-                }
-
-                // If neither metadata nor score is included, return just the content string
-                return includeMetadata || includeScore ? transformedResult : result.content;
-            });
-
-            debugOutput += `[Results] \nLoaded ${results.length} results from namespace: ${namespace}\n\n`;
-        } catch (error) {
-            _error = error.toString();
-        }
-
-        //is there a post processing LLM?
-
-        //TODO : better handling of context window exceeding max length
-        if (postprocess && prompt) {
-            const promises: any = [];
-            for (let result of results) {
-                const _prompt = TemplateString(prompt.replace(/{{result}}/g, JSON.stringify(result))).parse(input).result;
-                const llmInference = await LLMInference.getInstance(model);
-                // const req = llmInference.prompt({ query: _prompt, params: { ...config, agentId: agent.id } }).catch((error) => ({ error: error }));
-                // promises.push(req);
-            }
-            results = await Promise.all(promises);
-            for (let i = 0; i < results.length; i++) {
-                if (typeof results[i] === 'string') {
-                    // results[i] = parseJson(results[i]);
-                    results[i] = JSONContent(results[i] as string).tryParse();
-                }
-            }
-        }
-
-        const totalLength = JSON.stringify(results).length;
-        debugOutput += `[Total Length] \n${totalLength}\n\n`;
-
-        return {
-            Results: results,
-            _error,
-            _debug: debugOutput,
-            //_debug: `Query: ${_input}. \nTotal Length = ${totalLength} \nResults: ${JSON.stringify(results)}`,
-        };
-    }
-
-    // private async checkIfTeamOwnsNamespace(teamId: string, namespaceId: string, token: string) {
-    //     try {
-    //         const res = await SmythAPIHelper.fromAuth({ token }).mwSysAPI.get(`/vectors/namespaces/${namespaceId}`);
-    //         if (res.data?.namespace?.teamId !== teamId) {
-    //             throw new Error(`Namespace does not exist`);
-    //         }
-    //         return true;
-    //     } catch (err) {
-    //         throw new Error(`Namespace does not exist`);
-    //     }
-    // }
-
-    private parseMetadata(metadata: any) {
-        try {
-            return JSON.parse(jsonrepair(metadata));
-        } catch (err) {
-            return metadata;
-        }
-    }
-}
+import Joi from 'joi';
+import { validateInteger } from '../utils';
+import { jsonrepair } from 'jsonrepair';
+import { TemplateString } from '@sre/helpers/TemplateString.helper';
+import { JSONContent } from '@sre/helpers/JsonContent.helper';
+import { ConnectorService } from '@sre/Core/ConnectorsService';
+import { AccessCandidate } from '@sre/Security/AccessControl/AccessCandidate.class';
+import { IAgent as Agent } from '@sre/types/Agent.types';
+import { Component } from './Component.class';
+
+// Note: LLMHelper renamed to LLMInference
+class LLMInference {
+    static async getInstance(model: string) {
+        throw new Error('Method not implemented.');
+    }
+}
+
+export class DataSourceLookup extends Component {
+    protected configSchema = Joi.object({
+        topK: Joi.alternatives([Joi.string(), Joi.number()]) // Value is now a number; keep string fallback for backward compatibility.
+
+            .custom(validateInteger({ min: 0 }), 'custom range validation')
+            .label('Result Count'),
+        model: Joi.string().valid('gpt-4o-mini', 'gpt-4', 'gpt-3.5-turbo', 'gpt-4', 'gpt-3.5-turbo-16k').optional(),
+        prompt: Joi.string().max(30000).allow('').label('Prompt').optional(),
+        postprocess: Joi.boolean().strict().optional(),
+        includeMetadata: Joi.boolean().strict().optional(),
+        namespace: Joi.string().allow('').max(80).messages({
+            // Need to reserve 30 characters for the prefixed unique id
+            'string.max': `The length of the 'namespace' name must be 50 characters or fewer.`,
+        }),
+        scoreThreshold: Joi.number().optional().label('Score Threshold'),
+        includeScore: Joi.boolean().optional().label('Include Score'),
+    });
+    constructor() {
+        super();
+    }
+    init() {}
+    async process(input, config, agent: Agent) {
+        await super.process(input, config, agent);
+        const componentId = config.id;
+        const component = agent.components[componentId];
+        const teamId = agent.teamId;
+        let debugOutput = agent.agentRuntime?.debug ? '== Data Source Lookup Log ==\n' : null;
+
+        const outputs = {};
+        for (let con of config.outputs) {
+            if (con.default) continue;
+            outputs[con.name] = '';
+        }
+
+        const namespace = config.data.namespace.split('_').slice(1).join('_') || config.data.namespace;
+        const model = config.data?.model || 'gpt-4o-mini';
+        const prompt = config.data?.prompt?.trim?.() || '';
+        const postprocess = config.data?.postprocess || false;
+        const includeMetadata = config.data?.includeMetadata || false;
+
+        const scoreThreshold = config.data?.scoreThreshold || 0.001; // Use low score (0.001) to return most results for backward compatibility
+        const includeScore = config.data?.includeScore || false;
+
+        const _input = typeof input.Query === 'string' ? input.Query : JSON.stringify(input.Query);
+
+        const topK = Math.max(config.data?.topK || 50, 50);
+
+        let vectorDbConnector = ConnectorService.getVectorDBConnector();
+        let existingNs = await vectorDbConnector.requester(AccessCandidate.team(teamId)).namespaceExists(namespace);
+
+        if (!existingNs) {
+            throw new Error(`Namespace ${namespace} does not exist`);
+        }
+
+        let results: string[] | { content: string; metadata: any; score?: number }[];
+        let _error;
+        try {
+            const response = await vectorDbConnector
+                .requester(AccessCandidate.team(teamId))
+                .search(namespace, _input, { topK, includeMetadata: true });
+
+            results = response.slice(0, config.data.topK).map((result) => ({
+                content: result.text,
+                metadata: result.metadata,
+                score: result.score, // use a very low score to return
+            }));
+
+            results = results.filter((result) => result.score >= scoreThreshold);
+
+            // Transform results based on inclusion flags
+            results = results.map((result) => {
+                const transformedResult: any = {
+                    content: result.content,
+                };
+
+                if (includeMetadata) {
+                    // legacy user-specific metadata key [result.metadata?.metadata]
+                    transformedResult.metadata = this.parseMetadata(result.metadata || result.metadata?.metadata);
+                }
+
+                if (includeScore) {
+                    transformedResult.score = result.score;
+                }
+
+                // If neither metadata nor score is included, return just the content string
+                return includeMetadata || includeScore ? transformedResult : result.content;
+            });
+
+            debugOutput += `[Results] \nLoaded ${results.length} results from namespace: ${namespace}\n\n`;
+        } catch (error) {
+            _error = error.toString();
+        }
+
+        //is there a post processing LLM?
+
+        //TODO : better handling of context window exceeding max length
+        if (postprocess && prompt) {
+            const promises: any = [];
+            for (let result of results) {
+                const _prompt = TemplateString(prompt.replace(/{{result}}/g, JSON.stringify(result))).parse(input).result;
+                const llmInference = await LLMInference.getInstance(model);
+                // const req = llmInference.prompt({ query: _prompt, params: { ...config, agentId: agent.id } }).catch((error) => ({ error: error }));
+                // promises.push(req);
+            }
+            results = await Promise.all(promises);
+            for (let i = 0; i < results.length; i++) {
+                if (typeof results[i] === 'string') {
+                    // results[i] = parseJson(results[i]);
+                    results[i] = JSONContent(results[i] as string).tryParse();
+                }
+            }
+        }
+
+        const totalLength = JSON.stringify(results).length;
+        debugOutput += `[Total Length] \n${totalLength}\n\n`;
+
+        return {
+            Results: results,
+            _error,
+            _debug: debugOutput,
+            //_debug: `Query: ${_input}. \nTotal Length = ${totalLength} \nResults: ${JSON.stringify(results)}`,
+        };
+    }
+
+    // private async checkIfTeamOwnsNamespace(teamId: string, namespaceId: string, token: string) {
+    //     try {
+    //         const res = await SmythAPIHelper.fromAuth({ token }).mwSysAPI.get(`/vectors/namespaces/${namespaceId}`);
+    //         if (res.data?.namespace?.teamId !== teamId) {
+    //             throw new Error(`Namespace does not exist`);
+    //         }
+    //         return true;
+    //     } catch (err) {
+    //         throw new Error(`Namespace does not exist`);
+    //     }
+    // }
+
+    private parseMetadata(metadata: any) {
+        try {
+            return JSON.parse(jsonrepair(metadata));
+        } catch (err) {
+            return metadata;
+        }
+    }
+}

package/src/Components/ECMASandbox.class.ts

@@ -1,72 +1,72 @@
-import { IAgent as Agent } from '@sre/types/Agent.types';
-import { Component } from './Component.class';
-import Joi from 'joi';
-import { ConnectorService } from '@sre/Core/ConnectorsService';
-import { CodeExecutionResult } from '@sre/ComputeManager/Code.service/CodeConnector';
-
-export class ECMASandbox extends Component {
-    protected configSchema = Joi.object({
-        code: Joi.string().max(500000).allow('').label('Code'),
-    });
-    constructor() {
-        super();
-    }
-    init() {}
-    async process(input, config, agent: Agent) {
-        await super.process(input, config, agent);
-
-        const logger = this.createComponentLogger(agent, config);
-        try {
-            let Output: any = {};
-            let _error = undefined;
-
-            let codeInputs = {};
-            for (let fieldName in input) {
-                const _type = typeof input[fieldName];
-                switch (_type) {
-                    case 'string':
-                        const b64encoded = Buffer.from(input[fieldName]).toString('base64');
-                        codeInputs[fieldName] = `___internal.b64decode('${b64encoded}')`;
-                        break;
-                    case 'number':
-                    case 'boolean':
-                        codeInputs[fieldName] = input[fieldName];
-                        break;
-                    default:
-                        codeInputs[fieldName] = input[fieldName];
-                        break;
-                }
-            }
-
-            //const inputVarsCode = this.generateInputVarCode(codeInputs);
-            //const code = inputVarsCode + '\n' + config.data.code;
-            const code = config.data.code;
-
-            logger.debug(`Running code: \n${code}\n`);
-
-            const ecmaCodeConnector = ConnectorService.getCodeConnector('ECMASandbox');
-
-            const executionResponse: CodeExecutionResult = await ecmaCodeConnector.agent(agent.id).execute(config.id, { code, inputs: input });
-            if (executionResponse.success) {
-                Output = executionResponse.output;
-            } else {
-                Output = undefined;
-                _error = executionResponse.errors;
-            }
-
-            return { Output, _error, _debug: logger.output };
-        } catch (err: any) {
-            const _error = err?.response?.data || err?.message || err.toString();
-            logger.error(`Error running code: \n${_error}\n`);
-            return { Output: undefined, _error, _debug: logger.output };
-        }
-    }
-
-    private generateInputVarCode(input: Record<string, any>) {
-        let input_vars = '';
-        for (const key in input) {
-            input_vars += `var ${key} = ${input[key]};\n`;
-        }
-        return input_vars;
-    }
-}
+import { IAgent as Agent } from '@sre/types/Agent.types';
+import { Component } from './Component.class';
+import Joi from 'joi';
+import { ConnectorService } from '@sre/Core/ConnectorsService';
+import { CodeExecutionResult } from '@sre/ComputeManager/Code.service/CodeConnector';
+
+export class ECMASandbox extends Component {
+    protected configSchema = Joi.object({
+        code: Joi.string().max(500000).allow('').label('Code'),
+    });
+    constructor() {
+        super();
+    }
+    init() {}
+    async process(input, config, agent: Agent) {
+        await super.process(input, config, agent);
+
+        const logger = this.createComponentLogger(agent, config);
+        try {
+            let Output: any = {};
+            let _error = undefined;
+
+            let codeInputs = {};
+            for (let fieldName in input) {
+                const _type = typeof input[fieldName];
+                switch (_type) {
+                    case 'string':
+                        const b64encoded = Buffer.from(input[fieldName]).toString('base64');
+                        codeInputs[fieldName] = `___internal.b64decode('${b64encoded}')`;
+                        break;
+                    case 'number':
+                    case 'boolean':
+                        codeInputs[fieldName] = input[fieldName];
+                        break;
+                    default:
+                        codeInputs[fieldName] = input[fieldName];
+                        break;
+                }
+            }
+
+            //const inputVarsCode = this.generateInputVarCode(codeInputs);
+            //const code = inputVarsCode + '\n' + config.data.code;
+            const code = config.data.code;
+
+            logger.debug(`Running code: \n${code}\n`);
+
+            const ecmaCodeConnector = ConnectorService.getCodeConnector('ECMASandbox');
+
+            const executionResponse: CodeExecutionResult = await ecmaCodeConnector.agent(agent.id).execute(config.id, { code, inputs: input });
+            if (executionResponse.success) {
+                Output = executionResponse.output;
+            } else {
+                Output = undefined;
+                _error = executionResponse.errors;
+            }
+
+            return { Output, _error, _debug: logger.output };
+        } catch (err: any) {
+            const _error = err?.response?.data || err?.message || err.toString();
+            logger.error(`Error running code: \n${_error}\n`);
+            return { Output: undefined, _error, _debug: logger.output };
+        }
+    }
+
+    private generateInputVarCode(input: Record<string, any>) {
+        let input_vars = '';
+        for (const key in input) {
+            input_vars += `var ${key} = ${input[key]};\n`;
+        }
+        return input_vars;
+    }
+}

package/src/Components/FEncDec.class.ts

@@ -1,29 +1,29 @@
-import { Component } from './Component.class';
-import { IAgent as Agent } from '@sre/types/Agent.types';
-
-export class FEncDec extends Component {
-    constructor() {
-        super();
-    }
-    init() {}
-    async process(input, config, agent: Agent) {
-        await super.process(input, config, agent);
-        const logger = this.createComponentLogger(agent, config);
-        try {
-            const _error = undefined;
-
-            const data = input.Data;
-            const action = config.data.action || 'Encode';
-            const encoding = config.data.encoding;
-            logger.debug(`${encoding} ${action} data`);
-
-            const Output = action == 'Encode' ? Buffer.from(data).toString(encoding) : Buffer.from(data, encoding).toString('utf8');
-
-            return { Output, _error, _debug: logger.output };
-        } catch (err: any) {
-            const _error = err?.response?.data || err?.message || err.toString();
-            logger.error(` Error processing data \n${_error}\n`);
-            return { hash: undefined, _error, _debug: logger.output };
-        }
-    }
-}
+import { Component } from './Component.class';
+import { IAgent as Agent } from '@sre/types/Agent.types';
+
+export class FEncDec extends Component {
+    constructor() {
+        super();
+    }
+    init() {}
+    async process(input, config, agent: Agent) {
+        await super.process(input, config, agent);
+        const logger = this.createComponentLogger(agent, config);
+        try {
+            const _error = undefined;
+
+            const data = input.Data;
+            const action = config.data.action || 'Encode';
+            const encoding = config.data.encoding;
+            logger.debug(`${encoding} ${action} data`);
+
+            const Output = action == 'Encode' ? Buffer.from(data).toString(encoding) : Buffer.from(data, encoding).toString('utf8');
+
+            return { Output, _error, _debug: logger.output };
+        } catch (err: any) {
+            const _error = err?.response?.data || err?.message || err.toString();
+            logger.error(` Error processing data \n${_error}\n`);
+            return { hash: undefined, _error, _debug: logger.output };
+        }
+    }
+}

package/src/Components/FHash.class.ts

@@ -1,33 +1,33 @@
-import { Component } from './Component.class';
-import { IAgent as Agent } from '@sre/types/Agent.types';
-import crypto from 'crypto';
-
-export class FHash extends Component {
-    constructor() {
-        super();
-    }
-    init() {}
-    async process(input, config, agent: Agent) {
-        await super.process(input, config, agent);
-        const logger = this.createComponentLogger(agent, config);
-        try {
-            const _error = undefined;
-
-            const data = input.Data;
-            const algorithm = config.data.algorithm;
-            const encoding = config.data.encoding;
-            logger.debug(` Generating hash using ${algorithm} algorithm and ${encoding} encoding`);
-
-            const hashAlgo = crypto.createHash(algorithm);
-            hashAlgo.update(data);
-
-            const Hash = hashAlgo.digest(encoding);
-            logger.debug(` Generated hash: ${Hash}`);
-            return { Hash, _error, _debug: logger.output };
-        } catch (err: any) {
-            const _error = err?.response?.data || err?.message || err.toString();
-            logger.error(` Error generating hash \n${_error}\n`);
-            return { hash: undefined, _error, _debug: logger.output };
-        }
-    }
-}
+import { Component } from './Component.class';
+import { IAgent as Agent } from '@sre/types/Agent.types';
+import crypto from 'crypto';
+
+export class FHash extends Component {
+    constructor() {
+        super();
+    }
+    init() {}
+    async process(input, config, agent: Agent) {
+        await super.process(input, config, agent);
+        const logger = this.createComponentLogger(agent, config);
+        try {
+            const _error = undefined;
+
+            const data = input.Data;
+            const algorithm = config.data.algorithm;
+            const encoding = config.data.encoding;
+            logger.debug(` Generating hash using ${algorithm} algorithm and ${encoding} encoding`);
+
+            const hashAlgo = crypto.createHash(algorithm);
+            hashAlgo.update(data);
+
+            const Hash = hashAlgo.digest(encoding);
+            logger.debug(` Generated hash: ${Hash}`);
+            return { Hash, _error, _debug: logger.output };
+        } catch (err: any) {
+            const _error = err?.response?.data || err?.message || err.toString();
+            logger.error(` Error generating hash \n${_error}\n`);
+            return { hash: undefined, _error, _debug: logger.output };
+        }
+    }
+}