Extracting Text from Images

The useOCR hook from @reverbia/sdk/react provides Optical Character Recognition (OCR) capabilities. It extracts text from images and scanned documents, and can serve as a fallback when PDF text extraction is insufficient.

Hook Initialization


const { extractOCRContext, isProcessing, error } = useOCR();

Extracting Text from Images


const extractFromImages = useCallback(
  async (files: FileAttachment[]) => {
    const imageFiles = files.filter((f) => f.mediaType.startsWith("image/"));
 
    if (imageFiles.length === 0) {
      console.log("No image files to process");
      return null;
    }
 
    const context = await extractOCRContext(imageFiles);
 
    if (context && context.length > 0) {
      imageFiles.forEach((file) => {
        setExtractedTexts((prev) => [
          {
            filename: file.filename,
            content: context,
            timestamp: Date.now(),
          },
          ...prev,
        ]);
      });
 
      return context;
    }
 
    return null;
  },
  [extractOCRContext]
);