Skip to content

An API for storing, indexing and retrieving documentation

License

Notifications You must be signed in to change notification settings

deconst/content-service

Repository files navigation

Content Service

A content storage and retrieval service for deconst.

Build Status Docker Repository on Quay.io

Setup

To develop locally, you'll need to install:

Then, you can build and run the service with:

# Customize your environment settings.
cp environment.sample.sh environment.sh
${EDITOR} environment.sh
source environment.sh

docker-compose build
docker-compose up

Configuration

The content service is configured by passing environment variables to the Docker container. These are the available configuration options:

  • STORAGE: (default: "remote") Specify memory to use entirely in-memory storage, hybrid to use external MongoDB and Elasticsearch but an in-memory CDN/asset storage, or remote to use external storage services.
  • ADMIN_APIKEY: (required) An API key that can be used by administrators or other internal services to issue and revoke API keys.

Remote services

  • RACKSPACE_USERNAME: (required if STORAGE=remote) The username for your Rackspace account.
  • RACKSPACE_APIKEY: (required if STORAGE=remote) The API key for your Rackspace account.
  • RACKSPACE_REGION: (required if STORAGE=remote) The Rackspace region for the content service to use.
  • RACKSPACE_SERVICENET: (default: "false") If "true", connect to Cloud Files over ServiceNet rather than the public endpoint.
  • CONTENT_CONTAINER: (required if STORAGE=remote) Container name to use for the stored metadata envelopes.
  • ASSET_CONTAINER: (required if STORAGE=remote) Container name to use for published assets.
  • MONGODB_URL: (required if STORAGE=remote) MongoDB connection string, including any required authentication information.
  • MONGODB_PREFIX: (default: "") Prefix used to partition MongoDB collection names from other services using the same MongoDB database.
  • ELASTICSEARCH_HOST: (required if STORAGE=remote) Elasticsearch connection string, including any required authentication information.

Hybrid services

  • MONGODB_URL: (required if STORAGE=hybrid) MongoDB connection string, including any required authentication information.
  • MONGODB_PREFIX: (default: "") Prefix used to partition MongoDB collection names from other services using the same MongoDB database.
  • ELASTICSEARCH_HOST: (required if STORAGE=hybrid) Elasticsearch connection string, including any required authentication information.

Memory storage

  • MEMORY_ASSET_PREFIX: (default: "/__local_asset__/") Prefix used to construct URLs for assets present in memory storage.

Staging

  • STAGING_MODE: (default: "false") Act as a staging store, to stage many revisions of content simultaneously. When staging mode is active, proxied content IDs will have their first URL path segment, the revision ID, removed when making the upstream request.
  • PROXY_UPSTREAM: (required if STAGING_MODE=true) If a URL is specified, content not found in this content store will be requested from an upstream content store API. Named assets will be accumulated from the upstream content store and this service, with named assets from this service taking precedence.

Logging

  • CONTENT_LOG_LEVEL: (default: "info") Optional logging level, case-insensitive. The valid levels are TRACE, DEBUG, VERBOSE, INFO, WARN, and ERROR.
  • CONTENT_LOG_COLOR: (default: "false") Logging colorization. Set to "true" to enable colorful logs.
  • NODE_ENV: If set to "production", logs will be emitted as JSON objects.

Both Cloud Files containers will be created and configured on application launch if they do not already exist.

Deconst Dev Env in Kubernetes with Minikube

These instructions will create the resources necessary to run the deconst content service in a dev env in Kubernetes with Minikube.

  1. Run through Deconst Dev Env in Kubernetes with Minikube

  2. Open a new shell

  3. Customize your environment settings

    cp environment.sample.sh environment.sh
    ${EDITOR} environment.sh

    Edit the following environment variables. For additional env vars and a full description of each env var, see Configuration.

    • ADMIN_APIKEY=my-random-api-key
      • You can use the value of hexdump -v -e '1/1 "%.2x"' -n 128 /dev/random
    • DOCKER_IMAGE=kube-registry.kube-system.svc.cluster.local:31000/content-service:dev
      • If you want to use the production image instead, keep the default value and skip the next step
    • STORAGE=remote
    • RACKSPACE_USERNAME=my-rackspace-username
    • RACKSPACE_APIKEY=my-rackspace-api-key
    • RACKSPACE_REGION=IAD
    • RACKSPACE_SERVICENET=false
    • CONTENT_CONTAINER=deconst-${NODE_ENV}-content
    • ASSET_CONTAINER=deconst-${NODE_ENV}-asset
    • MONGODB_URL=mongodb://mongo.deconst.svc.cluster.local:27017/content
    • ELASTICSEARCH_HOST=http://elasticsearch-logging.kube-system.svc.cluster.local:9200/
    source environment.sh
  4. Build a development Docker image

    eval $(minikube docker-env)
    docker build --tag kube-registry.kube-system.svc.cluster.local:31000/content-service:dev .
    docker push kube-registry.kube-system.svc.cluster.local:31000/content-service:dev
  5. Create resources

    script/template kubernetes/deployment.yaml | kubectl apply -f -
  6. Watch and wait for resources

    watch kubectl get pods --namespace deconst
  7. Test that the content and staging content services are nominally working

    curl $(minikube service --url --namespace deconst content)/version/
    curl $(minikube service --url --namespace deconst staging-content)/version/
  8. Deploy the presenter service

  9. Delete resources

    kubectl delete deploy/content svc/content --namespace deconst
    kubectl delete deploy/staging-content svc/staging-content  --namespace deconst

API

Authorization

Endpoints that require authorization must be accompanied by a valid API key. Set the API key in an Authorization header with a scheme of "deconst".

PUT /content/https%3A%2F%2Fgithub.com%2Fsomeuser%2Fsomerepo%2Fsomeid
Authorization: deconst 12345

Or:

curl -H 'Authorization: deconst 12345' # ...

Valid API keys include:

The content service exposes the following API endpoints:

GET /version

Report the service name, version, and git commit.

Response

{
    "commit": "e2af254",
    "service": "content-service",
    "version": "1.0.0"
}

GET /content/[?prefix=:id_prefix&pageNumber=:num&perPage=:size]

List content IDs available within the content service, paginated. If :id_prefix is provided, list only content IDs beginning with that prefix.

The default page size is 100 content IDs.

Response: Successful

On success, the response includes an HTTP status of 200 and contains a JSON payload containing zero to many matching content IDs.

{
  "total": 123,
  "results": [
    {
      "contentID": "<matching content ID>",
      "url": "<API URL to fetch envelope contents>"
    }
  ]
}

PUT /content/:id

(Authorization required: any user)

Store and index content with a specific URL-encoded content ID.

Request

The request payload must be a valid Deconst metadata envelope in JSON form.

{
  "title": "Optional page title",
  "body": "<h1>page content</h1> <p>as raw HTML</p>"
}

Response: Successful

An HTTP status of 200 and an empty response body will be returned when content is accepted successfully.

DELETE /content/:id[?prefix=true]

(Authorization required: any user)

Delete content with a specific URL-encoded content ID. If '?prefix=true' is specified, delete all content with a content ID with the :id parameter as a prefix.

⚠️ Use caution when bulk-deleting content from production. Ensure that no other mapped content that you don't wish to delete shares a prefix with the :id you provide. ⚠️

Response: Successful

An HTTP status of 204 indicates that the content has been deleted successfully.

Response: Unsuccessful

An HTTP status of 404 will be returned if the content ID isn't recognized.

GET /content/:id

Access previously stored content by its URL-encoded content ID.

Response: Successful

{
  "envelope": {
    "body": "<h1>The content... </h1>"
  }
}

"envelope" will be the exact JSON document provided to PUT /content. "assets" will contain a set of site-wide layout asset variables.

Response: Unsuccessful

An HTTP status of 404 will be returned if the content ID isn't recognized.

POST /bulkcontent

(Authorization required: any user)

Bulk-upload content from a .tar.gz file.

Request

The POST body must be a .tar.gz file in the following format:

metadata/config.json  # Optional: config file
metadata/keep.json    # Optional: additional content IDs to keep
https%3A%2F%2Fgithub.com%2Fsomeorg%2Fsomerepo.json  # metadata envelopes with URL-encoded filenames
https%3A%2F%2Fgithub.com%2Fsomeorg%2Fsomerepo%2Fpageone.json
https%3A%2F%2Fgithub.com%2Fsomeorg%2Fsomerepo%2Fpagetwo.json

The config.json file, if present, should contain:

{
  "contentIDBase": "https://github.com/someorg/somerepo"
}

All envelopes that share the named content ID base which are not included in the tarball or mentioned in keep.json will be removed as part of this operation.

The keep.json file, if present, should contain:

{
  "keep": [
    "https://github.com/someorg/somerepo/oldpage1",
    "https://github.com/someorg/somerepo/oldpage2"
  ]
}

Response: Successful

An HTTP status of 200 is returned once all envelopes are accepted successfully. The response body reports counts of the actions taken:

{
  "accepted": 24,
  "failed": 0,
  "deleted": 12
}

Response: Unsuccessful

An HTTP status of 500 is returned if there are problems accepting one or more envelopes. The response body reports how many envelopes from the bundle failed:

{
  "accepted": 21,
  "failed": 3,
  "deleted": 12
}

Note that the successful envelopes were accepted and are live.

GET /checkcontent

Perform a bulk query to determine which envelopes need to be updated and which are unchanged.

Request

Attach a body to the GET request containing a JSON object mapping content IDs to SHA-256 checksums. Generate each checksums from a compact (whitespace-less) JSON representation with sorted object keys.

{
  "https://github.com/org/repo": "b1b6e4c544880769b42bdbf7f6338cba3db78cf734424af20e5c4d30251a984c",
  "https://github.com/org/repo/somepath": "0ff459af6d050d72d642eeb3a1ea5a8a93fd518993ac56711bd86a7e49b95191"
}

Response (successful)

The response will have a 200 status and a response body containing the queried content IDs and a boolean value of:

  • true if the envelope with that ID is already present and its checksum matches, or
  • false if the envelope is either missing entirely or its checksum differs.

The envelope that's queried for a fingerprint match is the envelope that would be rendered for a GET /content/:id request against this endpoint. In staging mode, this means that an envelope that's present in the local store with a different checksum will return false even if the upstream content service has that envelope with a matching checksum.

{
  "https://github.com/org/repo": true,
  "https://github.com/org/repo/somepath": false
}

Response (unsuccessful)

A status of 500 indicates that an internal storage error occurred. A status of 502 indicates that the content store is configured to proxy to an upstream service, but it couldn't be reached.

POST /assets[?named=true]

(Authorization required: any user)

Fingerprint and publish one or more static assets to a CDN-enabled Cloud Files container. Return the full URLs to the published assets.

Request

The request payload must be a multipart/form-data file upload containing the assets to upload. The content type of each file must be set appropriately.

If the query parameter named=true is provided, each asset's CDN URI will also be persisted in the layout asset map with a key derived from its form name and inserted in the assets object of every outgoing metadata envelope.

Response

{
  "file1.jpg": "https://assets.horse/url/for/file1-38be7d1b981f2fb6a4a0a052453f887373dc1fe8.jpg",
  "file2.css": "https://assets.horse/url/for/file2-d2da57e04b0818f7e3dd18da3b73c9b54a73cbe5.css"
}

GET /assets

Enumerate named assets from the content service.

Response

{
  "file1_jpg_url": "https://assets.horse/url/for/file1-38be7d1b981f2fb6a4a0a052453f887373dc1fe8.jpg",
  "file2_css_url": "https://assets.horse/url/for/file2-d2da57e04b0818f7e3dd18da3b73c9b54a73cbe5.css"
}

GET /assets/:filename

Fetch an asset directly by filename. Most useful when you're using STORAGE=memory; otherwise you should use the CDN URLs.

Successful Response

Return an HTTP status of 200, the correct content-type, and the asset body.

Unsuccessful Response

A response of 400 is returned if no asset with that filename exists.

POST /bulkasset

(Authorization required: any user)

Bulk-upload assets from a .tar.gz file.

Request

The POST body must be a .tar.gz file with a content-type of application/tar+gzip.

Successful Response

Returns an HTTP status of 200 and a map containing the tarball's path to each asset and its corresponding public CDN URL:

{
  "assets/0.png": "https://9b1a5535446999eb0355-a76b79aca86ab4305d3e9ef5c9a01022.ssl.cf5.rackcdn.com/0-ed56e65d28de88bbbfafc33e7c959b764e5c8aa35ad9874966dfda9d46b8b28e.png",
  "assets/100.png": "https://9b1a5535446999eb0355-a76b79aca86ab4305d3e9ef5c9a01022.ssl.cf5.rackcdn.com/100-6ae71f29cf579f37139540c32a42fe4078b24d95c5453acaf3be2b6909e175e4.png"
}

Unsuccessful Response

If the uploaded file is not a valid tarball, an HTTP status of 400 will be returned. If there are problems performing the batch upload an HTTP status of 500 will be returned.

GET /checkassets

Perform a bulk query to determine which assets need to be uploaded and which are already present.

Request

Attach a body to the GET request containing a JSON object mapping asset filenames to SHA-256 checksums.

{
  "header.jpg": "08facdf9cdab2065dd76d0c50c20d93141c1e2be8a1224782458b0f41ad04eee",
  "local/path/style.min.css": "5746528f57f4c571bcbcdd7334b4396277877488bff0207603d7fb829fa7f854"
}

Response (successful)

The response will have a 200 status and a response body mapping the queried paths to:

  • The known asset's public CDN URL if an asset with that filename and checksum is already present, or
  • null if no such asset exists.

In staging mode, the upstream content store will be queried for any assets that are not present locally.

{
  "header.jpg": "https://my.awesome.cdn/public/header-08facdf9cdab2065dd76d0c50c20d93141c1e2be8a1224782458b0f41ad04eee.jpg",
  "local/path/style.min.css": null
}

Response (unsuccessful)

A status of 500 indicates that an internal storage error occurred. A status of 502 indicates that the content store is configured to proxy to an upstream service, but it couldn't be reached.

POST /keys?named=:name

(Authorization required: admin only)

Issue a newly generated API key. The provided human-friendly name will be used to log actions performed with this key.

Response: Successful

{
  "apikey": "6aa856c50e5c8895020ef8d35..."
}

Response: Unsuccessful

An HTTP status code of 401 indicates that the request did not contain a valid administrator API key.

DELETE /keys/:apikey

(Authorization required: admin only)

Revoke a previously issued user API key.

Response: Successful

An HTTP status code of 204 indicates that the API key will no longer be valid.

Response: Unsuccessful

An HTTP status code of 401 indicates that the request did not contain a valid administrator API key. A 409 is generated if an administrator attempts to revoke their own key.

GET /search?q=:term&pageNumber=:num&perPage=:size&categories=deconst.horse

Perform a full-text search against all indexed documents.

q is a required parameter. pageNumber defaults to 1 if unspecified, and perPage defaults to 10. categories may be specified multiple times (or as categories[]), and if present at least once, will constrain search results to only those envelopes that contain at least one matching category.

Response: Successful

{
  "total": 100,
  "results": [
    {
      "contentID": "https://github.com/deconst/deconst-docs/one",
      "title": "First result title",
      "excerpt": "body excerpt with matching text <em>highlighted</em>"
    },
    {
      "contentID": "https://github.com/deconst/deconst-docs/two",
      "title": "Second result title",
      "excerpt": "first bit of the body if no body content matched"
    }
  ]
}

PUT /control

(Authorization required: any user)

Change the stored control repository SHA.

Request

The request payload must be a JSON object containing a 40-character git commit SHA.

{
  "sha": "f2fa39527c7b35b9960fd39e2eb77217db3ee517"
}

Response: Successful

A status code of 204 indicates that the new SHA has been accepted.

Response: Unsuccessful

An HTTP status code of 401 indicates that the request did not contain a valid API key. A status code of 400 indicates that the request body did not contain an appropriately formatted JSON document.

GET /control

Retrieve the most recently stored control repository SHA.

Response: Successful

If a control repository SHA has been stored:

{
  "sha": "f2fa39527c7b35b9960fd39e2eb77217db3ee517"
}

If no SHA has been stored yet:

{
  "sha": null
}

POST /reindex

(Authorization required: admin only)

Asynchronously reindex all content from the primary key-value content store in the full-text search store.

Response: Unsuccessful

An HTTP status code of 401 indicates that the request did not contain a valid administrator API key.

Response: Successful

A status code of 202 indicates the reindexing has begun. Watch the container's logs to see indexing progress.

About

An API for storing, indexing and retrieving documentation

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published