Files
ragflow/internal/parser/parser/pdf_parser_paddleocr.go
Jin Hai add7b9486f Go: merge duplicate codes (#16783)
### Summary

1. merge heartbeat function.
2. introduce all environments

---------

Signed-off-by: Jin Hai <haijin.chn@gmail.com>
2026-07-10 11:58:32 +08:00

63 lines
1.7 KiB
Go

package parser
import (
"fmt"
"ragflow/internal/common"
"strings"
models "ragflow/internal/entity/models"
)
func parsePDFWithPaddleOCR(filename string, data []byte, parser *PDFParser) ParseResult {
if len(data) == 0 {
return emptyPDFResult(filename)
}
baseURL := strings.TrimSpace(parser.PaddleOCRBaseURL)
if baseURL == "" {
baseURL = strings.TrimSpace(common.GetEnv(common.EnvPaddleOCRBaseUrl))
}
if baseURL == "" {
baseURL = strings.TrimSpace(common.GetEnv(common.EnvPaddleOCRApiURL))
}
if baseURL == "" {
return ParseResult{Err: fmt.Errorf("parser: PaddleOCR requires paddleocr_base_url or PADDLEOCR_BASE_URL")}
}
apiKey := parser.PaddleOCRAPIKey
if strings.TrimSpace(apiKey) == "" {
apiKey = strings.TrimSpace(common.GetEnv(common.EnvPaddleOCRAccessToken))
}
algorithm := strings.TrimSpace(parser.PaddleOCRAlgorithm)
if algorithm == "" {
algorithm = strings.TrimSpace(common.GetEnv(common.EnvPaddleOCRAlgorithm))
}
if algorithm == "" {
algorithm = "PaddleOCR-VL"
}
driver := models.NewPaddleOCRLocalModel(
map[string]string{"default": baseURL},
models.URLSuffix{OCR: "layout-parsing"},
)
apiConfig := &models.APIConfig{
BaseURL: &baseURL,
}
if apiKey != "" {
apiConfig.ApiKey = &apiKey
}
resp, err := driver.OCRFile(&algorithm, data, &filename, apiConfig, &models.OCRConfig{
Algorithm: algorithm,
})
if err != nil {
return ParseResult{Err: fmt.Errorf("parser: PaddleOCR OCRFile: %w", err)}
}
if resp == nil || resp.Text == nil {
return ParseResult{Err: fmt.Errorf("parser: PaddleOCR returned empty text")}
}
pageCount := 1
if resp.Text != nil && strings.TrimSpace(*resp.Text) == "" {
pageCount = 0
}
return parseMinerUMarkdownResult(filename, *resp.Text, parser.OutputFormat, pageCount)
}