Henara.Scan.Client

Reads German medical act scans (Muster 13, Muster 56):

• Detects the page in a photo or scan, then crops and deskews it.
• Combines several pages into one PDF.
• Extracts every field of both sheets as typed objects.
• Extracts the front page of either form with Mistral OCR or the Henara (Azure) OCR, both in one shape, so a page read by either compares field by field.
• Cuts the form's stamp and signature boxes out as images, under the Vordruck's own field ids.
• Renders the filled form back as PDF or HTML.
• Reads arbitrary images against an annotation schema of your own.

Setup

dotnet add package Henara.Scan.Client

builder.Services.AddScanClient(builder.Configuration);

Binds from the configuration root, so environment variables alone are enough. Returns the IHttpClientBuilder:

builder.Services.AddScanClient(builder.Configuration)
    .AddHttpMessageHandler<MyTracingHandler>();

Configuration

Validated at host start, so a missing address or token fails there rather than on the first call.

Calling

One method per form. Every method returns Result<TValue, ScanError> and none throws: a rejected request, a failed call, a cancellation and an unexpected error all come back as the error side.

var message = (await scan.Crop(document, cancellationToken)).Match(
    page => Save(page.Content, page.ContentType),
    error => $"{error.Code}: {error.Message}");

if ((await scan.Crop(document, cancellationToken)).TryGetValue(out var page, out var error) == false)
    return error;

IsAutocropEnabled is off by default and the service never overrides it. Leave it off for images that were cropped already — warping twice degrades them.

Crop one page

An image no page could be detected in comes back as it was given.

var cropped = await scan.Crop(scannedFront, cancellationToken);

Combine pages into a PDF

One image per page, in page order, at least two.

var pdf = await scan.Combine([front, back], isAutocropEnabled: true, cancellationToken);

Read a Muster 13

One image reads that side alone against its own schema; two are read as one document spanning both sides.

var reading = await scan.ReadMuster13Images(
    new Muster13ImageReadRequest(
        [new SheetImage(SheetSide.Front, front), new SheetImage(SheetSide.Back, back)],
        IsRegionCropEnabled: true,
        new ReadOutputs(RenderTargets.Pdf | RenderTargets.Html, false, false),
        IsAutocropEnabled: true),
    cancellationToken);

Muster13Reading.Front and .Back carry every field of the sheet they name, and are Option.None for a sheet the read produced nothing for.

ReadMuster13Pdf takes a PDF of one page or two, and carries neither crop option. A PDF read returns no regions.

var reading = await scan.ReadMuster13Pdf(
    new PdfReadRequest(pdf, new ReadOutputs(RenderTargets.None, false, false)),
    cancellationToken);

Read a Muster 56

ReadMuster56Images and ReadMuster56Pdf, returning Muster56Reading. All three of this form's boxes sit on the back.

var reading = await scan.ReadMuster56Images(
    new Muster56ImageReadRequest(
        [new SheetImage(SheetSide.Front, front), new SheetImage(SheetSide.Back, back)],
        new ReadOutputs(RenderTargets.None, false, false),
        IsAutocropEnabled: true,
        IsRegionCropEnabled: true),
    cancellationToken);

Read a front page with either engine

The only two calls that take an engine, and the only ones that read with OcrEngine.Azure. Exactly one image.

var frontPage = await scan.ReadMuster13FrontPage(
    OcrEngine.Azure,
    new FrontPageReadRequest(
        front,
        new ReadOutputs(RenderTargets.None, false, false),
        IsRegionCropEnabled: true),
    cancellationToken);

ActScanFrontPage.Fields is the act-scan shape, so a Muster 56 read this way answers only with the fields the two forms share — not sport type, the units-and-months scope or the illness list. Use ReadMuster56Images for those.

ActScanFrontPage.MistralOnly is filled only after an OcrEngine.Mistral read: what Azure does not report at all, and the full values behind the single-valued fields — every ICD-10 code, every remedy, every treatment count.

Read against a schema of your own

The one call that answers with the annotation rather than typed fields; nothing is cropped or rendered. Several images are composed into one PDF and read as a single document.

var reading = await scan.ReadImagesWithSchemaWithMistral(
    new SchemaReadRequest(schemaJson, [page1, page2], IsAutocropEnabled: true),
    cancellationToken);

What comes back

The typed fields are always there. IsDocumentAnnotationRequested and IsRawJsonRequested add the engine's own annotation and response body; they are diagnostics. Render decides whether the filled form comes back as PDF, HTML or not at all.

A closed set on the paper is an enum: Copayment, RemedyField, TherapyChange, ApplicationType, SportType, RehabScope, FunctionalTrainingScope, TrainingForm, Illness, HeartGroup, PrescriptionKind, FollowUpReason, WeeklyFrequency, PayerRehabScope.

A field the read did not fill is an Option<T> carrying nothing, never null; a list it did not fill is empty.

if (front.PatientName.TryGetValue(out var name))
    Console.WriteLine(name);

var city = front.PatientCity.Or("unknown");

var age = front.PatientBirthDate.Match(born => Years(born), () => -1);

Or(fallback), Map, Then, HasValue, and ToString() for the value or an empty string.

Dates are DateOnly, whether the Vordruck prints TT.MM.JJJJ or the row of single boxes captioned T T M M J J; a two-digit year is read as this century. Muster56Back.ParticipatingSince is a YearMonth — its boxes are captioned M M J J and carry no day.

The three printed remedy rows of a Muster 13 are IReadOnlyList<PrescribedRemedy>. The supplementary remedy is not one of them: it has AdditionalRemedy and AdditionalRemedyUnits.

The receipt table on the back is IReadOnlyList<ReceiptRow>, its three columns already zipped by row, so a row left blank in one column still lines up with the others.

Regions

A stamp or signature box comes back as an image and is never transcribed: it may hold nothing, a stamp, a signature, a barcode-like graphic or any combination, and the signature usually sits on top of the stamp. IsRegionCropEnabled asks for them; every box the read looked at then comes back, whatever the box held.

The Id is the Vordruck's own field id, so a region binds straight to the field it belongs to; the receipt table appends the printed row number, 1-based.

HasInk says whether anything was written in the box. An unsigned prescription comes back with the region present and HasInk false — so a region that is missing entirely means only one thing, that the read was not asked to cut any.

var stamp = reading.Regions.FirstOrDefault(region => region.Id == "stempel-arzt");

if (stamp is { HasInk: true })
    Save(stamp.Content, stamp.ContentType);

Each crop reaches a little past the printed box, because a signature routinely overruns it and because autocrop and skew shift the rectangle across the paper. On the receipt table that means a row's crop can catch a stroke from the row above or below. Rect is where the crop was taken from, in 0 to 1 against the sheet.

Only image reads crop. ReadMuster13Pdf, ReadMuster56Pdf and ReadImagesWithSchemaWithMistral return no regions.

Errors

ScanError carries a ScanErrorCode, a message, and a ScanEngine — None when no upstream is named. Requests the client refuses on its own, such as an empty image or two images naming the same side, never reach the service.

Other languages

protos/scan.proto ships inside the package. The gRPC status carries the class of failure — FailedPrecondition is Unreadable — and the scan-error-code and scan-upstream trailers carry the exact code and engine.