What causes "Could not find matching AF_initDataCallback" in SubtitleEdit/subtitleedit?

Check the console output of matches to see what callbacks Lens did return.

SubtitleEdit/subtitleedit · warning · Exception

Could not find matching AF_initDataCallback

Error message

Could not find matching AF_initDataCallback

What it means

Thrown by Core.GetAFData when the regex AF_initDataCallback({.*?}) finds no match whose value contains 'DetectedObject'. Lens embeds OCR results inside an AF_initDataCallback JS call tagged with DetectedObject; if none is present, either Lens found no text or the response structure changed. The matches are dumped to the console before throwing to aid debugging.

Source

Thrown at src/ui/Logic/Ocr/GoogleLens/Core.cs:324

            {
                if (DateTimeOffset.TryParse(subParts[1].Trim(), out var expires))
                {
                    cookie.Expires = expires.ToUnixTimeMilliseconds();
                }
            }
        }

        return cookie;
    }

    public static JObject? GetAFData(string text)
    {
        var matches = Regex.Matches(text, @"AF_initDataCallback\((\{.*?\})\)", RegexOptions.Singleline);
        var lensCallback = matches.FirstOrDefault(m => m.Value.Contains("DetectedObject"));
        if (lensCallback == null)
        {
            Console.WriteLine(matches);
            throw new Exception("Could not find matching AF_initDataCallback");
        }

        var json = lensCallback.Groups[1].Value;
        return JObject.Parse(json);
    }

    public static List<string> ParseResult(JObject afData, int[] imageDimensions)
    {
        var data = afData["data"];
        var fullTextPart = data?[3];
        var textSegments = new List<string>();
        var textRegions = new List<double[]>();

        try
        {
            // Method 1: get text segments and regions directly
            if (fullTextPart?[4]?[0]?[0] != null)
            {

View on GitHub (pinned to 17a9f07487)

Solutions

Check the console output of matches to see what callbacks Lens did return.
Treat 'no DetectedObject' as 'no text found' rather than a hard error for blank images.
Update the regex / marker string if Google renamed DetectedObject.
Verify the image actually contains text; retry with a higher-contrast crop.

Example fix

// before
var af = Core.GetAFData(text);

// after
var af = Core.GetAFData(text); // wrap:
try { var af = Core.GetAFData(text); }
catch (Exception ex) when (ex.Message.Contains("AF_initDataCallback")) { return new List<string>(); }

Defensive patterns

Strategy: try-catch

Type guard

static bool HasDetectedObjectCallback(string text) => Regex.IsMatch(text, @"AF_initDataCallback\(\{.*?DetectedObject.*?\}\)", RegexOptions.Singleline);

Try / catch

try { var af = Core.GetAFData(text); }
catch (Exception ex) when (ex.Message.Contains("AF_initDataCallback")) { return new List<string>(); // treat as no text found }

Prevention

Treat 'no DetectedObject' as an empty result for blank images rather than a fatal error.
Log the console matches dump to detect schema changes.
Update the regex/marker when Google renames the callback.
Confirm the image actually contains legible text.

When it happens

Trigger: A 200 Lens response with no detected objects (blank/no-text image), a response that is a captcha/interstitial, or a schema change where the DetectedObject marker was renamed. Also when the response is HTML error content.

Common situations: OCR'ing an image with no recognizable text; Google changed the callback key; the response is a consent/captcha page returned with status 200; regional response variants.

Related errors

AI-assisted analysis of SubtitleEdit/subtitleedit@17a9f07487 (2026-08-13). Data as JSON: /api/errors/f3dd3c5f35686c55. Report an issue: GitHub.