mirror of
https://github.com/infiniflow/ragflow.git
synced 2026-06-29 23:41:12 +08:00
16d5b4fa02
### What problem does this PR solve? Closes #15673 — ports the Python `file2document_api.py` `convert()` endpoint to Go. | Method | Path | Handler | |--------|------|---------| | POST | `/api/v1/files/link-to-datasets` | `FileHandler.LinkToDatasets` | ### Type of change - [x] New Feature (non-breaking change which adds functionality) --- #### Implementation notes **Files changed:** ``` internal/service/file2document.go – new service (File2DocumentService) internal/dao/file2document.go – added Create method internal/handler/file.go – FileHandler gains file2DocumentService; LinkToDatasets HTTP handler internal/router/router.go – route registered ``` **Functional parity table:** | Concern | Go behaviour | |---------|-------------| | Required fields | `file_ids` and `kb_ids` both required; missing either → `CodeDataError` mirroring Python `@validate_request` | | File existence | `fileDAO.GetByIDs(fileIDs)` builds a set; any missing ID → `"File not found!"` | | KB existence | `kbDAO.GetByID(kbID)` per KB; missing → `"Can't find this dataset!"` | | Folder expansion | `getAllInnermostFileIDs` recursively calls `fileDAO.ListByParentID` — mirrors `FileService.get_all_innermost_file_ids` | | File permissions | `checkFileTeamPermission`: `file.TenantID == userID` OR user in tenant's team — mirrors `check_file_team_permission` | | KB permissions | `checkKBTeamPermission`: `kb.TenantID == userID` OR user in tenant's team — mirrors `check_kb_team_permission` | | Fire-and-forget | `go convertFiles(...)` goroutine after all validation passes — mirrors `loop.run_in_executor(None, _convert_files, …)` | | Conversion | `convertFiles`: for each file → delete existing mappings + hard-delete old documents → create new `Document` in each target KB → create `File2Document` mapping — mirrors Python `_convert_files` | | `getParser` | Extension-based lookup with fallback to `kb.ParserID` — mirrors `FileService.get_parser` | | Immediate return | `true` returned to caller as soon as goroutine is scheduled | --------- Co-authored-by: Yingfeng <yingfeng.zhang@gmail.com>