feat(office): Impled ability to download document from office and loa…

…d to elastic

.env

@@ -5,14 +5,14 @@ LOGGER_ENABLE_FILE_LOG=false
 DOC_NOTIFIER_SERVICE_ADDRESS=0.0.0.0:2893
 DOC_NOTIFIER_WATCHED_DIRS=./indexer/watcher
 
-OCR_SERVICE_ADDRESS=http://localhost:1231
-OCR_SERVICE_MODE=raw
+OCR_SERVICE_ADDRESS=http://192.168.0.59:8004
+OCR_SERVICE_MODE=assistant
 OCR_SERVICE_TIMEOUT=300
 
 DOC_SEARCHER_SERVICE_ADDRESS=http://localhost:2892
 DOC_SEARCHER_SERVICE_TIMEOUT=300
 
-TOKENIZER_SERVICE_ADDRESS=http://localhost:8001
+TOKENIZER_SERVICE_ADDRESS=http://192.168.0.59:8082
 TOKENIZER_SERVICE_MODE=none
 TOKENIZER_RETURN_CHUNKS=false
 TOKENIZER_CHUNK_SIZE=800
@@ -28,3 +28,5 @@ STORAGE_PORT=5432
 STORAGE_DB_NAME=documents
 STORAGE_ENABLE_SSL=disable
 STORAGE_LLM_ADDRESS=http://localhost:8081
+
+OFFICE_SERVICE_ADDRESS=http://localhost:8087/example

cmd/notifier/notifier.go

@@ -10,6 +10,7 @@ import (
 	"doc-notifier/cmd"
 	"doc-notifier/internal/logger"
 	"doc-notifier/internal/ocr"
+	"doc-notifier/internal/office"
 	"doc-notifier/internal/reader"
 	"doc-notifier/internal/searcher"
 	"doc-notifier/internal/server"
@@ -26,6 +27,7 @@ func main() {
 	}
 
 	readService := reader.New()
+	officeService := office.New(&serviceConfig.Office)
 	ocrService := ocr.New(&serviceConfig.Ocr)
 	searchService := searcher.New(&serviceConfig.Searcher)
 	tokenService := tokenizer.New(&serviceConfig.Tokenizer)
@@ -37,6 +39,7 @@ func main() {
 		searchService,
 		tokenService,
 		storeService,
+		officeService,
 	)
 
 	ctx, cancel := context.WithCancel(context.Background())

configs/config.toml

@@ -34,3 +34,6 @@ Port=5432
 DbName="documents"
 EnableSSL="disable"
 AddressLLM="http://localhost:8081"
+
+[office]
+Address="http://localhost:8087/example"

docs/docs.go

@@ -346,6 +346,48 @@ const docTemplate = `{
                 }
             }
         },
+        "/watcher/files/upload-from-office": {
+            "post": {
+                "description": "Upload files to analyse from onlyoffice",
+                "produces": [
+                    "application/json"
+                ],
+                "tags": [
+                    "files"
+                ],
+                "summary": "Upload files to analyse from onlyoffice",
+                "operationId": "files-upload-office",
+                "parameters": [
+                    {
+                        "type": "string",
+                        "description": "File name to download from office",
+                        "name": "fileName",
+                        "in": "query",
+                        "required": true
+                    }
+                ],
+                "responses": {
+                    "200": {
+                        "description": "Ok",
+                        "schema": {
+                            "$ref": "#/definitions/server.ResponseForm"
+                        }
+                    },
+                    "400": {
+                        "description": "Bad Request message",
+                        "schema": {
+                            "$ref": "#/definitions/server.BadRequestForm"
+                        }
+                    },
+                    "503": {
+                        "description": "Server does not available",
+                        "schema": {
+                            "$ref": "#/definitions/server.ServerErrorForm"
+                        }
+                    }
+                }
+            }
+        },
         "/watcher/folders/": {
             "get": {
                 "description": "Get watched directories list",

docs/swagger.json

@@ -335,6 +335,48 @@
                 }
             }
         },
+        "/watcher/files/upload-from-office": {
+            "post": {
+                "description": "Upload files to analyse from onlyoffice",
+                "produces": [
+                    "application/json"
+                ],
+                "tags": [
+                    "files"
+                ],
+                "summary": "Upload files to analyse from onlyoffice",
+                "operationId": "files-upload-office",
+                "parameters": [
+                    {
+                        "type": "string",
+                        "description": "File name to download from office",
+                        "name": "fileName",
+                        "in": "query",
+                        "required": true
+                    }
+                ],
+                "responses": {
+                    "200": {
+                        "description": "Ok",
+                        "schema": {
+                            "$ref": "#/definitions/server.ResponseForm"
+                        }
+                    },
+                    "400": {
+                        "description": "Bad Request message",
+                        "schema": {
+                            "$ref": "#/definitions/server.BadRequestForm"
+                        }
+                    },
+                    "503": {
+                        "description": "Server does not available",
+                        "schema": {
+                            "$ref": "#/definitions/server.ServerErrorForm"
+                        }
+                    }
+                }
+            }
+        },
         "/watcher/folders/": {
             "get": {
                 "description": "Get watched directories list",

docs/swagger.yaml

@@ -404,6 +404,34 @@ paths:
       summary: Upload files to analyse
       tags:
       - files
+  /watcher/files/upload-from-office:
+    post:
+      description: Upload files to analyse from onlyoffice
+      operationId: files-upload-office
+      parameters:
+      - description: File name to download from office
+        in: query
+        name: fileName
+        required: true
+        type: string
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: Ok
+          schema:
+            $ref: '#/definitions/server.ResponseForm'
+        "400":
+          description: Bad Request message
+          schema:
+            $ref: '#/definitions/server.BadRequestForm'
+        "503":
+          description: Server does not available
+          schema:
+            $ref: '#/definitions/server.ServerErrorForm'
+      summary: Upload files to analyse from onlyoffice
+      tags:
+      - files
   /watcher/folders/:
     get:
       description: Get watched directories list

internal/config/config.go

@@ -9,6 +9,7 @@ type Config struct {
 	Searcher  SearcherConfig
 	Tokenizer TokenizerConfig
 	Storage   StorageConfig
+	Office    OfficeConfig
 }
 
 type LoggerConfig struct {
@@ -53,3 +54,7 @@ type StorageConfig struct {
 	EnableSSL  string
 	AddressLLM string
 }
+
+type OfficeConfig struct {
+	Address string
+}

internal/config/from_env.go

@@ -90,13 +90,17 @@ func LoadEnv(enableDotenv bool) (*Config, error) {
 		AddressLLM: storageAddressLLM,
 	}
 
+	officeAddress := loadString("OFFICE_SERVICE_ADDRESS")
+	officeConfig := OfficeConfig{Address: officeAddress}
+
 	return &Config{
 		Logger:    loggerConfig,
 		Watcher:   watcherConfig,
 		Ocr:       ocrConfig,
 		Searcher:  searcherConfig,
 		Tokenizer: tokenizerConfig,
 		Storage:   storageConfig,
+		Office:    officeConfig,
 	}, nil
 }
 

internal/config/from_file.go

@@ -33,6 +33,8 @@ func FromFile(filePath string) (*Config, error) {
 	viperInstance.SetDefault("tokenizer.ChunkBySelf", false)
 	viperInstance.SetDefault("tokenizer.Timeout", 300)
 
+	viperInstance.SetDefault("office.Address", "http://localhost:8087")
+
 	if err := viperInstance.ReadInConfig(); err != nil {
 		confErr := fmt.Errorf("failed while reading config file %s: %w", filePath, err)
 		return config, confErr

internal/office/office.go

@@ -0,0 +1,37 @@
+package office
+
+import (
+	"fmt"
+	"log"
+	"os"
+
+	"doc-notifier/internal/config"
+	"doc-notifier/internal/sender"
+)
+
+type Service struct {
+	address string
+}
+
+func New(config *config.OfficeConfig) *Service {
+	return &Service{
+		address: config.Address,
+	}
+}
+
+func (s *Service) DownloadDocument(fileName string) error {
+	targetUrl := fmt.Sprintf("%s/download?fileName=%s", s.address, fileName)
+	fileData, err := sender.SendGETRequest(targetUrl)
+	if err != nil {
+		log.Println("failed to download file from office: ")
+		return err
+	}
+
+	filePath := fmt.Sprintf("./indexer/watcher/%s", fileName)
+	err = os.WriteFile(filePath, fileData, os.ModePerm)
+	if err != nil {
+		log.Println("failed to write file: ", err)
+	}
+
+	return err
+}

internal/server/files.go

@@ -19,6 +19,7 @@ func (s *Service) CreateFilesGroup() error {
 	//})
 
 	group.POST("/upload", s.UploadFilesToUnrecognized)
+	group.POST("/upload-from-office", s.UploadFileFromOnlyOffice)
 	//group.POST("/analyse", s.AnalyseFiles, timeoutMW)
 	group.GET("/artifacts", s.GetArtifactsTemplate)
 	group.POST("/analyse", s.AnalyseFiles)
@@ -67,6 +68,25 @@ func (s *Service) UploadFilesToUnrecognized(c echo.Context) error {
 	return c.JSON(200, uploadFiles)
 }
 
+// UploadFileFromOnlyOffice
+// @Summary Upload files to analyse from onlyoffice
+// @Description Upload files to analyse from onlyoffice
+// @ID files-upload-office
+// @Tags files
+// @Produce  json
+// @Param fileName query string true "File name to download from office"
+// @Success 200 {object} ResponseForm "Ok"
+// @Failure	400 {object} BadRequestForm "Bad Request message"
+// @Failure	503 {object} ServerErrorForm "Server does not available"
+// @Router /watcher/files/upload-from-office [post]
+func (s *Service) UploadFileFromOnlyOffice(c echo.Context) error {
+	fileName := c.QueryParam("fileName")
+	if err := s.watcher.Office.DownloadDocument(fileName); err != nil {
+		return c.JSON(400, createStatusResponse(400, err.Error()))
+	}
+	return c.JSON(200, createStatusResponse(200, "Done"))
+}
+
 // AnalyseFiles
 // @Summary Analyse uploaded files by ids
 // @Description Analyse uploaded files by ids

internal/watcher/watcher.go

@@ -2,6 +2,7 @@ package watcher
 
 import (
 	"context"
+	"doc-notifier/internal/office"
 	"errors"
 	"log"
 	"math"
@@ -37,6 +38,7 @@ type NotifyWatcher struct {
 	Searcher  *searcher.Service
 	Tokenizer *tokenizer.Service
 	Storage   *storage.Service
+	Office    *office.Service
 }
 
 func New(
@@ -46,6 +48,7 @@ func New(
 	searcherService *searcher.Service,
 	tokenService *tokenizer.Service,
 	storageService *storage.Service,
+	officeService *office.Service,
 ) *NotifyWatcher {
 	notifyWatcher, err := fsnotify.NewWatcher()
 	if err != nil {
@@ -69,6 +72,7 @@ func New(
 		Searcher:            searcherService,
 		Tokenizer:           tokenService,
 		Storage:             storageService,
+		Office:              officeService,
 	}
 }