Public | Automated Build

Last pushed: a month ago
Short Description
Annotation Reader-Writer for Neo4j
Full Description

Annotations Reader/Writer for Neo4j (annotations-rw-neo4j)

A service and an API for reading/writing annotations into Neo4j.
If the consumer is enabled: the messages are consumed from the queue, they get written into Neo4j, and finally, if the producer is also enabled, they got forwarded into the next (PostAnnotations) queue.
The above flow can be initiated by the PUT endpoint as well. In this case, the service expects the annotations json to be supplied in the format that comes out of the annotations consumer.


  • go get -u
  • go get -u
  • cd $GOPATH/src/
  • govendor sync
  • go build .

Running locally

  • govendor test -v -race
  • go install
  • $GOPATH/bin/annotations-rw-neo4j [--help]

You have more options here:

  • run neo4j and kafka locally (by docker, or as native apps)
  • open a tunnel to one of your team clusters that your app can connect to
  • disable the functionality that requires kafka by setting the env var SHOULD_FORWARD_MESSAGES=false




Each annotation is added with a relationship according to the predicate property from the payload.
If that is empty: a default MENTIONS relationship will be added between the content and a concept.

This operation acts as a replace - all existing annotations are removed, and the new ones are created - for the specified annotations-lifecycle.
Supplying an empty list as the request body will remove all annotations for the content.

A successful PUT results in 201.

We run queries in batches. If a batch fails, all failing requests will get a 500 server error response.

Invalid json body input will result in a 400 bad request response.

NB: annotations don't have identifiers themselves currently - the id in the json is the id of the concept that is annotating the content.

See this doc for more details.


curl -XPUT -H "X-Request-Id: 123" -H "Content-Type: application/json" localhost:8080/content/3fa70485-3a57-3b9b-9449-774b001cd965/annotations/annotations-v1 --data

NB: Although provenances are supplied is a list, we don't expect to get more than one provenance: we will take the scores from that one
and apply them to the relationship that we are creating for that annotation.

If there is no provenance, or the provenance is incomplete (e.g. no agent role) we'll still
create the relationship, it just won't have score, agent and time properties.


This internal read should return what got written (i.e., this isn't the public annotations read API) - for the specified annotations-lifecycle.

If not found, you'll get a 404 response.

Empty fields are omitted from the response.
curl -H "X-Request-Id: 123" localhost:8080/content/3fa70485-3a57-3b9b-9449-774b001cd965/annotations/annotations-v1



Deletes all the annotations with the specified annotations-lifecycle.

Will return 204 if successful, 404 if not found

curl -XDELETE -H "X-Request-Id: 123" localhost:8080/3fa70485-3a57-3b9b-9449-774b001cd965/annotations/annotations-v1

NB: /content/{contentId}/annotations/mentions/{conceptId} also existed in the old annotations writer and was used to allow annotations to be removed in Spyglass (however it was not used because if the content is republished, we lose the fact an annotation was deleted). We have chosen not to replicate
that functionality in this app.


Docker Pull Command