query-orchestration/api/docInitRunner/runner.go

package docinitrunner

import (
	"context"
	"database/sql"
	"errors"
	"log/slog"

	"queryorchestration/internal/database/repository"
	documentinit "queryorchestration/internal/document/init"
	"queryorchestration/internal/serviceconfig/objectstore"

	"github.com/google/uuid"
)

const Name = "docInitRunner"

type Services struct {
	Document *documentinit.Service
	Queries  *repository.Queries
}

type Runner struct {
	svc *Services
}

func New(svc *Services) Runner {
	return Runner{
		svc: svc,
	}
}

type Body struct {
	Bucket string `json:"bucket" validate:"required"`
	Key    string `json:"key" validate:"required"`
	Hash   string `json:"hash" validate:"required"`
}

func (s Runner) Process(ctx context.Context, body Body) bool {
	key, err := objectstore.ParseBucketKey(body.Key)
	if err != nil {
		slog.Error("unable to parse key", "key", body.Key, "error", err)
		return false
	}

	// Get filename and folder_id from documentUploads table
	var filename *string
	var folderID *uuid.UUID
	upload, err := s.svc.Queries.GetDocumentUploadByKey(ctx, &repository.GetDocumentUploadByKeyParams{
		Bucket: body.Bucket,
		Key:    body.Key,
	})
	if err != nil && !errors.Is(err, sql.ErrNoRows) {
		slog.Error("unable to get document upload", "bucket", body.Bucket, "key", body.Key, "error", err)
		return false
	}
	if err == nil {
		if upload.Filename != nil {
			filename = upload.Filename
		}
		folderID = upload.FolderID
	}

	id, err := s.svc.Document.Create(ctx, &documentinit.Create{
		Key:          key,
		Bucket:       body.Bucket,
		Hash:         body.Hash,
		Filename:     filename,
		OriginalPath: filename, // For batch uploads, filename contains the full original path
		FolderID:     folderID,
	})
	if err != nil {
		return false
	}
	slog.Debug("created document", "id", id, "filename", filename, "folder_id", folderID)

	return true
}