Barcode Hub
Enables barcode decoding from URLs, uploaded files, or raw bytes via HTTP and MCP, supporting formats like EAN13, QRCode, and DataMatrix.
README
Barcode Hub
Barcode Hub is a Python service for barcode recognition over HTTP and MCP. The
recognition backend is zxing-cpp.
The API response body is intentionally minimal:
{
"barcodes": [
{
"text": "4607084351323",
"data": "NDYwNzA4NDM1MTMyMw==",
"type": "EAN13",
"valid": "yes"
}
]
}
Build identity is emitted at runtime as an HTTP Server header on every
response, but it is not part of the OpenAPI or MCP contracts.
Endpoints
Single service port, default 8080:
GET /shows a small HTML server index with build info and resource linksGET /decode?url=...&types=EAN13,UPCAPOST /decodewith exactly onemultipart/form-dataimage filePUT /decodewith raw image bytes and animage/*Content-TypeGET /healthchecks thatzxingcppis importable and exposes the required APIGET /metricsexposes Prometheus metrics/mcpwhen MCP is enabled/docs,/redoc,/openapi.json
Run Locally
python3.12 -m venv .venv
. .venv/bin/activate
pip install -e '.[dev]'
barcode-hub
Decode from a URL:
curl 'http://localhost:8080/decode?url=https://example.com/image.jpg&types=EAN13,UPCA'
Decode an uploaded file:
curl -F 'file=@image.jpg;type=image/jpeg' 'http://localhost:8080/decode?types=EAN13'
Decode raw bytes:
curl -X PUT --data-binary @image.jpg -H 'Content-Type: image/jpeg' \
'http://localhost:8080/decode?types=EAN13'
Docker
Use the prebuilt image from GitHub Container Registry:
docker pull ghcr.io/yiivgeny/barcode-hub:latest
docker run --rm -p 8080:8080 ghcr.io/yiivgeny/barcode-hub:latest
Or build locally:
docker build -t barcode-hub:local .
docker run --rm -p 8080:8080 barcode-hub:local
Docker Compose builds the local image by default:
docker compose up --build
Configuration
Configuration is loaded from /etc/barcode-hub/config.yaml when it exists.
Environment variables override file values. Nested variables use __ and the
BARCODE_HUB_ prefix.
Example YAML:
app:
port: 8080
limits:
max_request_body_bytes: 16777216
max_file_bytes: 8388608
max_image_side_pixels: 4096
decode:
enabled_methods: ["GET", "POST", "PUT"]
default_formats: ["EAN13", "EAN8", "UPCA", "UPCE"]
allowed_formats: ["EAN13", "EAN8", "UPCA", "UPCE", "QRCode", "DataMatrix"]
request_timeout_seconds: 10
fetch:
timeout_seconds: 5
allowed_url_prefixes: ["https://*.example.com/", "data:*"]
media:
allowed_content_types: ["image/png", "image/jpeg", "image/webp"]
mcp:
enabled: true
logging:
format: human
level: INFO
Common environment variables:
BARCODE_HUB_LIMITS__MAX_REQUEST_BODY_BYTESBARCODE_HUB_LIMITS__MAX_FILE_BYTESBARCODE_HUB_LIMITS__MAX_IMAGE_SIDE_PIXELSBARCODE_HUB_DECODE__ENABLED_METHODSBARCODE_HUB_DECODE__REQUEST_TIMEOUT_SECONDSBARCODE_HUB_FETCH__TIMEOUT_SECONDSBARCODE_HUB_FETCH__ALLOWED_URL_PREFIXESBARCODE_HUB_MEDIA__ALLOWED_CONTENT_TYPESBARCODE_HUB_MCP__ENABLEDBARCODE_HUB_LOGGING__FORMAT
allowed_url_prefixes are matched by parsing URLs. For example,
https://*.example.com/ matches https://cdn.example.com/image.jpg, but does
not match https://test.tld/example.com/image.jpg.
Metrics
/metrics includes process/platform/gc metrics and service metrics:
barcode_hub_decode_requests_total{interaction,status}barcode_hub_decoded_barcodes_total{interaction,type,valid}barcode_hub_decode_request_duration_seconds{interaction,status}barcode_hub_recognition_duration_seconds{interaction,status}barcode_hub_resource_fetch_duration_seconds{interaction,status}barcode_hub_decode_input_bytes{interaction,source}barcode_hub_decode_image_max_side_pixels{interaction}barcode_hub_build_info
Interaction labels are get_url, post_multipart, put_binary, and mcp_url.
MCP
When enabled, /mcp exposes the decode_url tool:
decode_url(url: string, types?: list[BarcodeType]) -> DecodeResult
data:image/...;base64,... URLs are supported. MCP tool errors use MCP error
semantics and stable domain codes in the message, such as invalid_url,
disallowed_type, resource_timeout, and request_too_large.
Development
pip install -e '.[dev]'
python scripts/generate_contract.py --check
pytest
ruff check .
The shared response schema is spec/decode-result.schema.json. The OpenAPI
contract is spec/openapi.yaml. The MCP contract is spec/mcp.md.
Runtime /openapi.json is loaded from spec/openapi.yaml and then narrowed by
the active configuration: disabled /decode methods are removed, BarcodeType
is limited to decode.allowed_formats, and request media types reflect
media.allowed_content_types. The types query parameter default reflects
decode.default_formats. Swagger UI and ReDoc use that runtime schema.
DecodeResult, Barcode, BarcodeType, and BarcodeValidity are generated
from the Python models and format enum:
python scripts/generate_contract.py
The generator updates spec/decode-result.schema.json and only the marked
components.schemas block in spec/openapi.yaml. Paths, request/response
semantics, examples, and MCP documentation remain hand-authored.
License
Barcode Hub is MIT licensed. zxing-cpp is Apache-2.0 licensed; keep upstream
notices when redistributing images or derived distributions.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。