- Updated the company service to include a new method for scheduling the refresh of cached AI recommendations after the AI pipeline execution.
- Introduced a new interface for managing cached recommendation refreshes, improving the separation of concerns within the service layer.
- Enhanced the worker initialization to include Redis client support, allowing for better management of recommendation caching.
- Added functionality to list company IDs with existing recommendation caches, ensuring efficient updates post-pipeline runs.
- Implemented unit tests to validate the new recommendation refresh logic and ensure proper handling of various scenarios.
This update significantly improves the handling of AI recommendations by integrating caching mechanisms with the AI pipeline, enhancing overall system performance and responsiveness.
- Renamed and refactored the AI pipeline auto run functionality to a daily run, enhancing clarity and purpose.
- Introduced a new `AIPipelineDailyWorker` to manage the daily execution of the AI pipeline, replacing the previous auto run implementation.
- Updated configuration fields and logging messages to reflect the change from auto to daily run, ensuring consistent terminology throughout the codebase.
- Removed the obsolete `ai_pipeline_auto.go` file to streamline the worker structure.
This update improves the maintainability and readability of the AI pipeline management by clearly distinguishing between auto and daily run functionalities.
- Introduced AIPipelineAutoWorker to manage the execution of the AI pipeline auto run, including startup catch-up and scheduled tasks.
- Enhanced WorkerConfig to include AIPipelineAutoEnabled and AIPipelineAutoInterval settings for better control over AI pipeline execution.
- Added logging for AI pipeline auto run status, including success and error handling, to improve observability.
- Updated daily job tracker to include AIPipelineAutoJobName for tracking AI pipeline job completions.
This update enhances the system's capability to automate AI pipeline executions, improving efficiency and reliability in processing AI tasks.
- Introduced a mutex to ensure only one TED scraper run executes at a time, preventing concurrent executions during startup and scheduled runs.
- Implemented a mechanism to check if today's TED scrape has already been completed during startup, logging appropriate messages for both completed and new runs.
- Added startup catch-up logic for tender translations and unprocessed notices, ensuring that any missed tasks are executed without blocking the application startup.
This update improves the reliability and efficiency of the TED scraper and worker processes, ensuring that all necessary tasks are completed after a server restart.
- Introduced a new `NotificationWorker` to promote due scheduled notifications from pending to sent, improving notification management.
- Added `NotificationInterval` configuration to schedule the notification delivery worker, with a default value for flexibility.
- Implemented `MarkDueScheduledAsSent` method in the notification repository to update the status of notifications based on their delivery time.
- Updated the notification service to process due scheduled notifications during relevant operations, ensuring timely delivery.
This update enhances the notification system by automating the delivery of scheduled notifications, improving user engagement and operational efficiency.
- Removed the DocumentSummarizationWorker and its related scheduling logic from the worker bootstrap.
- Updated the AI summarizer client initialization comment for clarity.
- Added a new error type for cases when tender documents have not been scraped yet, enhancing error handling in the tender service.
- Modified API documentation to reflect changes in AI summary retrieval logic, ensuring accurate descriptions of on-demand summarization behavior.
This update streamlines the AI summarization process by eliminating the document summarization worker, improving overall system efficiency and clarity in error handling.
- Updated the `ResolvedEstimatedValueAndCurrency` method to aggregate procurement lot values when the tender-level estimated value is not set, improving accuracy in value retrieval.
- Introduced the `AggregateProcurementLotEstimatedValue` function to sum estimated values from procurement lots and return the first found currency.
- Modified the `ToResponseWithLanguage` method to utilize the new estimated value resolution logic.
- Added unit tests for the new functionality, ensuring correct behavior for various scenarios in the `entity_test.go` and `budget_test.go` files.
This update improves the handling of estimated values in tenders, enhancing the overall reliability of the tender management system.
- Changed the default MinIO bucket name from "opplens-documents" to "opplens" across multiple configuration files.
- Introduced caching for dashboard statistics in the service layer to improve performance and reduce redundant data fetching.
- Implemented a mutex for thread-safe access to cached statistics, ensuring data integrity during concurrent requests.
This update streamlines the configuration for the AI summarizer and optimizes the dashboard service, enhancing overall system efficiency.
- Added `TranslationEnabled` and `TranslationInterval` fields to the worker configuration to manage automatic translation scheduling.
- Updated the worker initialization to log when the translation worker is disabled.
- Improved error handling in the AI summarizer client by introducing `APIStatusError` for better context on API failures, replacing direct error messages with structured error responses.
This update enhances the configurability of the worker and improves error reporting for AI service interactions, contributing to better maintainability and user experience.
- Added new routes and handlers for AI pipeline operations, including scraping documents, batch summarization, translation, and syncing with the Opplens AI service.
- Introduced request forms for handling tender references and batch operations.
- Enhanced the AI service with methods for triggering batch operations and managing pipeline runs.
- Updated Swagger documentation to reflect the new AI pipeline endpoints and their functionalities.
This update integrates comprehensive AI pipeline capabilities into the tender management system, improving operational efficiency and user experience.
- Removed the QueueConfig structure and related queue management files as they are no longer in use.
- Updated the worker initialization to reflect the removal of queue-related configurations.
- Cleaned up the bootstrap process by eliminating deprecated logging related to the queue system.
This update streamlines the worker's configuration and prepares the codebase for future enhancements without the legacy queue management components.
Paginate un-summarized tenders and mark rows missing notice/folder IDs as handled so the worker cannot spin forever. Return 400/404 for validation and not-found cases on AI summarize/analyze triggers instead of leaking internal errors as 500.
- Added a new endpoint to trigger on-demand agentic analysis for tenders via the AI service.
- Introduced `TriggerAIAnalyze` method in the `TenderHandler` to handle requests and responses for AI analysis.
- Updated the `tenderService` to include `TriggerAIAnalyze` method, which validates input and interacts with the AI summarizer client.
- Enhanced the `AIAnalyzeResponse` and `AIAnalyzeDocument` structures to support the new analysis feature.
- Refactored the `DocumentSummarizationWorker` and `TranslationWorker` to remove deprecated MinIO dependencies, streamlining the AI service interactions.
This update improves the functionality of the tender management system by allowing users to trigger AI analysis on-demand, enhancing the overall user experience and system capabilities.
- Updated `InitAISummarizerClient` to accept `mongoManager` for tracking translation success.
- Introduced new `Statistics` endpoint in the dashboard to fetch scraping and translation statistics.
- Enhanced `TranslationWorker` to utilize the new success counter for tracking successful translations.
- Added necessary data structures and query forms for statistics reporting.
This refactor improves the tracking of AI translation success and provides new insights through the dashboard statistics.
- Refactored the NoticeWorker to generate unique tender IDs using the PBL naming convention (SCDYYNNN format) and project names based on client and opportunity details.
- Introduced methods for generating tender IDs and project names, including logic for extracting client and opportunity names, ensuring consistency and clarity in naming.
- Updated the Tender entity to include a new ProjectName field, enhancing the data structure for better project identification.
- Added MongoDB indexing for the new project_name field to optimize query performance.
- Improved error handling and logging during the tender ID generation process, ensuring robustness in the worker's functionality.
- Updated the worker initialization process to include the new GLM SDK, enhancing the worker's capabilities for translation tasks.
- Modified the InitWorker function and NewNoticeWorker constructor to accept the GLM service, ensuring a cohesive integration.
- Implemented the GLM service initialization and logging for successful setup, improving maintainability and usability.
- Updated the NoticeWorker to utilize the GLM SDK for translating notice titles and descriptions, enhancing functionality and user experience.
- Removed the Ollama SDK from the worker initialization process, simplifying the worker's dependencies and enhancing maintainability.
- Updated the InitWorker function and NewNoticeWorker constructor to reflect the removal of the Ollama SDK, ensuring a cleaner and more focused implementation.
- Commented out the Ollama model listing logic for potential future use, maintaining clarity in the main function while reducing unnecessary complexity.
- Introduced a new build step in the Drone CI configuration for the worker service, enabling automated Docker image creation.
- Added a Dockerfile for the worker service, defining the build process and dependencies, ensuring a streamlined deployment for background tasks.
- Enhanced the overall CI/CD pipeline to support the new worker service, improving the project's build and deployment capabilities.
- Updated the ToTender method to accept a tender instance, allowing for direct population of tender fields from the notice entity.
- Simplified the tender creation and update process by checking if the tender ID is zero, streamlining the logic for handling tender entities.
- Removed the commented-out code related to AI translation, improving code clarity and maintainability.
- Enhanced error logging for tender creation and update failures, ensuring better visibility into potential issues during processing.
- Introduced AlertMail configuration in both scraper and worker bootstrap files, allowing for customizable email notifications.
- Updated the TED scraper to utilize the AlertMail configuration for sending completion notifications, improving flexibility in notification management.
- Enhanced error logging in the worker's main function to capture issues when listing Ollama models, ensuring better visibility into potential failures.
- Refactored notification sending logic to check for a valid AlertMail before dispatching emails, ensuring notifications are only sent when configured.
- Improved overall structure and readability of the bootstrap configuration files, aligning with best practices for maintainability.
- Introduced a new worker service with a dedicated main entry point for handling background tasks, including MongoDB connection management and notification service initialization.
- Added a bootstrap package to manage application configuration, logging, and service initialization for the worker.
- Implemented a NoticeWorker to process unprocessed notices and create corresponding tender entities, enhancing the integration of notice management within the tender system.
- Refactored the tender service to remove the Ollama SDK dependency, streamlining the service initialization in the main application.
- Enhanced the notice repository with a method to retrieve unprocessed notices, improving data handling capabilities.
Mazyar