add ocr

backend/Dockerfile

@@ -0,0 +1,25 @@
+FROM golang:1.24-alpine AS builder
+
+WORKDIR /app
+
+COPY go.mod go.sum ./
+RUN go mod download
+
+COPY . .
+
+RUN CGO_ENABLED=0 GOOS=linux go build -o /bin/server ./cmd/server
+
+FROM alpine:3.21
+
+RUN apk add --no-cache ca-certificates curl
+
+WORKDIR /app
+
+COPY --from=builder /bin/server .
+COPY internal/db/migrations ./internal/db/migrations
+
+RUN mkdir -p /app/uploads /data
+
+EXPOSE 8080
+
+CMD ["./server"]

backend/internal/handlers/files.go

@@ -13,6 +13,7 @@ import (
 
 	"github.com/gin-gonic/gin"
 	"github.com/vaultdrop/backend/internal/db"
+	"github.com/vaultdrop/backend/internal/ocr"
 	"github.com/vaultdrop/backend/internal/service"
 )
 
@@ -191,6 +192,8 @@ func (h *Handlers) UploadFiles(c *gin.Context) {
 		Id   string `json:"id"`
 	}
 
+	client := ocr.NewClient("http://localhost:9090")
+
 	filesStats := []FileStats{}
 
 	for _, file := range files {
@@ -260,6 +263,15 @@ func (h *Handlers) UploadFiles(c *gin.Context) {
 			Id:   dbFile.ID,
 		})
 
+		text, err := client.Recognize(fileByte)
+
+		if err != nil {
+			fmt.Println(err)
+			return
+		}
+
+		fmt.Println(text)
+
 	}
 
 	c.JSON(http.StatusOK, gin.H{

backend/internal/ocr/client.go

@@ -0,0 +1,91 @@
+package ocr
+
+import (
+	"bytes"
+	"encoding/base64"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"time"
+)
+
+type Client struct {
+	endpoint   string
+	httpClient *http.Client
+}
+
+func NewClient(endpoint string) *Client {
+	return &Client{
+		endpoint: endpoint,
+		httpClient: &http.Client{
+			Timeout: 120 * time.Second,
+		},
+	}
+}
+
+func (c *Client) Recognize(imageData []byte) ([]TextBlock, error) {
+	b64 := base64.StdEncoding.EncodeToString(imageData)
+
+	reqBody, err := json.Marshal(OCRRequest{Image: b64})
+	if err != nil {
+		return nil, fmt.Errorf("marshal request: %w", err)
+	}
+
+	resp, err := c.httpClient.Post(
+		c.endpoint+"/ocr",
+		"application/json",
+		bytes.NewReader(reqBody),
+	)
+	if err != nil {
+		return nil, fmt.Errorf("call paddleocr: %w", err)
+	}
+	defer resp.Body.Close()
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return nil, fmt.Errorf("read response: %w", err)
+	}
+
+	if resp.StatusCode != http.StatusOK {
+		return nil, fmt.Errorf("paddleocr returned %d: %s", resp.StatusCode, string(body))
+	}
+
+	var ocrResp OCRResponse
+	if err := json.Unmarshal(body, &ocrResp); err != nil {
+		return nil, fmt.Errorf("decode response: %w", err)
+	}
+
+	if ocrResp.ErrorCode != 0 {
+		return nil, fmt.Errorf("paddleocr error %d: %s", ocrResp.ErrorCode, ocrResp.Message)
+	}
+
+	return flattenResults(ocrResp.Result), nil
+}
+
+func (c *Client) HealthCheck() error {
+	resp, err := c.httpClient.Get(c.endpoint + "/health")
+	if err != nil {
+		return err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		return fmt.Errorf("health check failed: status %d", resp.StatusCode)
+	}
+	return nil
+}
+
+func flattenResults(result OCRResult) []TextBlock {
+	var blocks []TextBlock
+	for _, page := range result.OCRResults {
+		for i, text := range page.RecTexts {
+			score := 0.0
+			if i < len(page.RecScores) {
+				score = page.RecScores[i]
+			}
+			blocks = append(blocks, TextBlock{Text: text, Score: score})
+		}
+	}
+	return blocks
+}

backend/internal/ocr/types.go

@@ -0,0 +1,27 @@
+package ocr
+
+type OCRRequest struct {
+	Image string `json:"image"`
+}
+
+type OCRResponse struct {
+	ErrorCode int        `json:"errorCode"`
+	Result    OCRResult  `json:"result"`
+	Message   string     `json:"message,omitempty"`
+}
+
+type OCRResult struct {
+	OCRResults []OCRPageResult `json:"ocrResults"`
+}
+
+type OCRPageResult struct {
+	RecTexts  []string    `json:"rec_texts"`
+	RecScores []float64   `json:"rec_scores"`
+	RecBoxes  [][]int     `json:"rec_boxes"`
+	RecPolys  [][][]int   `json:"rec_polys"`
+}
+
+type TextBlock struct {
+	Text  string  `json:"text"`
+	Score float64 `json:"score"`
+}