DPMA Connect Plus Go Client

A Go client for the German Patent and Trade Mark Office (DPMA) Connect Plus REST API (DPMAregister web services), covering German patent, design, and trademark data over HTTP Basic authentication.

Overview

• Three services - patents and utility models, designs, and trademarks, selected with the dpma.ServicePatent, dpma.ServiceDesign, and dpma.ServiceTrademark constants.
• Expert search - SearchPatents, SearchDesigns, SearchTrademarks using DPMAregister expert search syntax, with optional client-side query validation.
• Register data - register info by number, register extracts, and searchable full text.
• Bulk and single downloads - weekly XML/PDF bulk archives, single publication PDFs, and design/trademark images.
• Streaming variants - every bulk and register-extract method has a *Stream variant that writes to an io.Writer without buffering the whole response.
• Typed errors - NotFoundError, DataNotAvailableError, APIError, and XMLParseError for errors.As handling.

Installation

go get github.com/patent-dev/dpma-connect-plus

Getting access

DPMAconnectPlus is a contractual data-supply service. The DPMA issues an HTTP Basic username and password after a signed standard agreement is in place; there is no self-service web portal. A one-time connection fee applies (200 EUR at time of writing); data retrieval itself is free.

1. Review the service and terms on the DPMAconnectPlus page.

2. Complete the standard agreement (Standardvertrag DPMAconnectPlus) including the purpose-of-use paragraph, and send two signed originals to Deutsches Patent- und Markenamt, Referat 2.1.2 - Kundenservice Datenabgabe, 80297 München, Germany.

3. The DPMA creates an account with a username and a password of your choice.

4. Export the credentials for the client and demo:

   export DPMA_CONNECT_PLUS_USERNAME=your-username
   export DPMA_CONNECT_PLUS_PASSWORD=your-password

Quick start

package main

import (
    "context"
    "fmt"
    "log"

    dpma "github.com/patent-dev/dpma-connect-plus"
)

func main() {
    client, err := dpma.NewClient(&dpma.Config{
        Username: "your-dpma-username",
        Password: "your-dpma-password",
    })
    if err != nil {
        log.Fatal(err)
    }

    ctx := context.Background()

    // Get service version
    version, err := client.GetVersion(ctx, dpma.ServicePatent)
    if err != nil {
        log.Fatal(err)
    }
    fmt.Printf("Version: %s\n", version)

    // Validate query before sending (optional but recommended)
    if err := dpma.ValidatePatentQuery("TI=Elektrofahrzeug"); err != nil {
        log.Fatal(err)
    }

    // Search patents (uses DPMAregister expert search syntax)
    results, err := client.SearchPatents(ctx, "TI=Elektrofahrzeug")
    if err != nil {
        log.Fatal(err)
    }
    fmt.Printf("Results: %d bytes\n", len(results))
}

Usage

All methods accept context.Context for timeout and cancellation support.

Configuration

config := &dpma.Config{
    Username: "your-username",                                    // Required
    Password: "your-password",                                    // Required
    BaseURL:  "https://dpmaconnect.dpma.de/dpmaws/rest-services", // Default
    Timeout:  20 * time.Minute,                                   // Default: 20 minutes
    HTTPClient: myCustomHTTPClient,                               // Optional: your own *http.Client
}

client, err := dpma.NewClient(config)

If HTTPClient is set, Timeout is ignored and the custom client's timeout applies instead.

Common operations

// Get service version
GetVersion(ctx, service string) (string, error)

Use the service constants: dpma.ServicePatent, dpma.ServiceDesign, dpma.ServiceTrademark.

Query validation

// Validate a query against service-specific field codes (returns nil or error)
ValidatePatentQuery(query string) error
ValidateDesignQuery(query string) error
ValidateTrademarkQuery(query string) error

For advanced usage (tokenization, field inspection), use the query sub-package directly:

import "github.com/patent-dev/dpma-connect-plus/query"

// Parse and validate a patent query
q, err := query.ParseQuery("TI=Elektrofahrzeug AND INH=Siemens", query.ServicePatent)
if err != nil {
    log.Fatal(err)
}
if err := q.Validate(); err != nil {
    log.Fatal(err) // e.g. unknown field, unmatched parentheses
}

// Inspect the query
fmt.Println(q.GetFields())    // ["TI", "INH"]
fmt.Println(q.HasField("TI")) // true

// Look up field definitions
f, ok := query.GetField("TI", query.ServicePatent)
fmt.Println(f.Description) // "title / designation"

// List all valid fields for a service
fields := query.GetValidFields(query.ServiceDesign)

The parser supports quoted values, comparison operators (=, >=, <=, >, <), parentheses and curly braces (procedure data), bracket/brace matching, and both English (AND, OR, NOT) and German (UND, ODER, NICHT) Boolean operators. Field validation is per service, based on the official DPMAregister field codes.

Patent service

// Search patents and utility models
SearchPatents(ctx, query string) ([]byte, error)

// Get patent info by registered number
GetPatentInfo(ctx, registeredNumber string) ([]byte, error)

// Get searchable full text for a document
GetSearchableFullText(ctx, documentID string) ([]byte, error)

// Download single patent publication PDF
GetPatentPublicationPDF(ctx, documentID string) ([]byte, error)

// Weekly bulk downloads (XML)
GetDisclosureDocumentsXML(ctx, year, week int) ([]byte, error)
GetPatentSpecificationsXML(ctx, year, week int) ([]byte, error)
GetUtilityModelsXML(ctx, year, week int) ([]byte, error)
GetPublicationDataXML(ctx, year, week int) ([]byte, error)
GetApplicantCitationsXML(ctx, year, week int) ([]byte, error)
GetEuropeanPatentSpecificationsXML(ctx, year, week int) ([]byte, error)

// Weekly bulk downloads (PDF)
GetDisclosureDocumentsPDF(ctx, year, week int) ([]byte, error)
GetPatentSpecificationsPDF(ctx, year, week int) ([]byte, error)
GetEuropeanPatentSpecificationsPDF(ctx, year, week int) ([]byte, error)
GetUtilityModelsPDF(ctx, year, week int) ([]byte, error)

// Register extract
GetPatentRegisterExtract(ctx, date time.Time, period string) ([]byte, error)

Design service

// Search designs
SearchDesigns(ctx, query string) ([]byte, error)

// Get design info by design number
GetDesignInfo(ctx, designNumber string) ([]byte, error)

// Get design image/thumbnail
GetDesignImage(ctx, designNumber, imageNumber string) ([]byte, error)
GetDesignThumbnail(ctx, designNumber, thumbnailNumber string) ([]byte, error)

// Weekly bulk downloads
GetDesignBibliographicDataXML(ctx, year, week int) ([]byte, error)
GetDesignImages(ctx, year, week int) ([]byte, error)

// Register extract
GetDesignRegisterExtract(ctx, date time.Time, period string) ([]byte, error)

Trademark service

// Search trademarks
SearchTrademarks(ctx, query string) ([]byte, error)

// Get trademark info by application number
GetTrademarkInfo(ctx, applicationNumber string) ([]byte, error)

// Get trademark image/thumbnail
GetTrademarkImage(ctx, applicationNumber string) ([]byte, error)
GetTrademarkThumbnail(ctx, applicationNumber string) ([]byte, error)

// Weekly bulk downloads
GetTrademarkBibDataApplied(ctx, year, week int) ([]byte, error)
GetTrademarkBibDataRegistered(ctx, year, week int) ([]byte, error)
GetTrademarkBibDataRejected(ctx, year, week int) ([]byte, error)

// Register extract
GetTrademarkRegisterExtract(ctx, date time.Time, period string) ([]byte, error)

Streaming downloads (memory-efficient)

Every bulk download and register extract method has a *Stream variant that writes to an io.Writer:

file, err := os.Create("patents_202445.zip")
if err != nil {
    log.Fatal(err)
}
defer file.Close()

err = client.GetPatentSpecificationsXMLStream(ctx, 2024, 45, file)
if err != nil {
    log.Fatal(err)
}

The full set of *Stream methods mirrors the buffered bulk and register-extract methods above (GetDisclosureDocumentsXMLStream, GetDesignImagesStream, GetTrademarkRegisterExtractStream, and so on).

Search query syntax

All search methods use DPMAregister expert search syntax. The format is FIELD=value with Boolean operators AND/UND, OR/ODER, NOT/NICHT (English and German are both accepted), comparison operators (=, >=, <=, >, <), wildcards (? any chars, ! one char, # one or no char), quoted values, and parentheses. Procedure-data fields are written inside curly braces, for example {VST=pub-offenlegungsschrift UND VSTT=05.01.2011}.

The query subpackage parses and validates queries against per-service field definitions (see Query validation). Each service supports a different set of field codes; the most common are listed below.

Common patent field codes:

Code	Description
TI	Bezeichnung/Titel (title / designation)
INH	Anmelder/Inhaber (applicant / proprietor)
IN	Erfinder (inventor)
IC	IPC-Klasse (IPC classification)
AKZ	Aktenzeichen (file number / publication number)
PN	Veröffentlichungsnummer (publication number)
AT	Anmeldetag (filing date)
PUB	Publikationstag (publication date)
ST	Status
AB	Abstract

Common design field codes:

Code	Description
TI	Bezeichnung/Titel (title / designation)
INH	Inhaber (proprietor)
ENTW	Entwerfer (designer)
ERZ	Erzeugnis(se) (product(s))
WKL	Warenklasse (Locarno class)
RN	Registernummer (registration number)
DNR	Designnummer (design number)
AT	Anmeldetag (filing date)
ET	Eintragungstag (registration date)

Common trademark field codes:

Code	Description
MARKE / md	Marke (trademark text)
INH	Anmelder/Inhaber (applicant / proprietor)
KL	Klasse(n) (Nice class(es))
BKL	Bildklasse(n) (Vienna / image class(es))
MF	Markenform (trademark form)
RN	Registernummer/Aktenzeichen (registration number / file number)
AT	Anmeldetag (filing date)
ST	Status

For full field references, see the DPMAregister help pages:

• Patent field codes
• Design field codes
• Trademark field codes

Date and week formatting

Publication weeks use YYYYWW format (6 digits); register extracts use YYYY-MM-DD:

pubWeek, err := dpma.FormatPublicationWeek(2024, 45) // "202445", nil
year, week, err := dpma.ParsePublicationWeek("202445") // 2024, 45, nil

date := time.Date(2024, 10, 23, 0, 0, 0, 0, time.UTC)
dateStr := dpma.FormatDate(date) // "2024-10-23"

Data availability

• Publication data is updated weekly: patents/utility models on Thursdays, designs/trademarks on Fridays.
• Register data is updated daily.

DataNotAvailableError is common for future publication weeks, very old weeks (before digital archiving), and weeks with no publications.

Error handling

The library returns typed errors you can match with errors.As:

// Resource not found (404)
var notFoundErr *dpma.NotFoundError
if errors.As(err, &notFoundErr) {
    fmt.Printf("not found: %s %s\n", notFoundErr.Resource, notFoundErr.ID)
}

// Data not available (common for old/future publication weeks)
var dataErr *dpma.DataNotAvailableError
if errors.As(err, &dataErr) {
    fmt.Println("data not available for the requested period")
}

// Generic API errors
var apiErr *dpma.APIError
if errors.As(err, &apiErr) {
    fmt.Printf("API error: %s (code: %s, HTTP %d)\n", apiErr.Message, apiErr.Code, apiErr.StatusCode)
}

// XML parsing failures (malformed response data)
var xmlErr *dpma.XMLParseError
if errors.As(err, &xmlErr) {
    fmt.Printf("failed to parse response in %s: %v\n", xmlErr.Operation, xmlErr.Unwrap())
}

Testing

make test              # unit tests (race), mock HTTP server
make test-integration  # integration tests (//go:build integration), needs credentials
make lint

Integration tests hit the live DPMA API and skip unless credentials are set:

export DPMA_CONNECT_PLUS_USERNAME=your-username
export DPMA_CONNECT_PLUS_PASSWORD=your-password
make test-integration

An interactive demo application is included under demo/; it reads the same environment variables and offers a menu for the patent, design, and trademark services.

Implementation status

Every endpoint of the DPMA Connect Plus API is implemented across the patent, design, and trademark services (see the Usage method lists). How much of that surface a given account can exercise depends on the permissions DPMA grants it.

A subset is covered by the integration test suite and verified against the live API, including GetVersion, the three Search* methods (and their *Parsed variants), GetPatentInfo / GetDesignInfo / GetTrademarkInfo, GetPatentRegisterExtract, GetPatentPublicationPDF, GetDisclosureDocumentsXML, and GetDesignBibliographicDataXML.

The remaining bulk endpoints (for example GetPublicationDataXML, GetApplicantCitationsXML, GetDesignImages, and the GetTrademarkBibData* methods) are implemented but require permissions the test account does not currently hold, so they are not yet exercised end to end.

Development

Regenerate the typed client when the OpenAPI spec (openapi.yaml) changes:

make generate

This re-applies any local OpenAPI fixes and runs go generate ./..., which drives oapi-codegen to rewrite generated/types_gen.go and generated/client_gen.go. Do not edit the files under generated/ by hand.

Other common targets:

make fmt     # gofmt
make lint    # golangci-lint
make tidy    # go mod tidy

Related projects

Part of the patent.dev open-source patent data ecosystem:

• epo-ops - EPO Open Patent Services client (bibliographic, full text, families, legal status, images)
• epo-bdds - EPO Bulk Data Distribution Service client (DOCDB, INPADOC, EP full text)
• uspto-odp - USPTO Open Data Portal client (patents, PTAB, TSDR, full text)

The bulk-file-loader uses these libraries for automated patent data downloads.

License

MIT - Funktionslust GmbH / patent.dev.