v1.5.0

Added

• Adds CLI command to manage native queries with automatic type inference (#131)

Changed

• Updates MongoDB Rust driver from v2.8 to v3.1.0 (#124)

Fixed

• The connector previously used Cloudflare's DNS resolver. Now it uses the locally-configured DNS resolver. (#125)
• Fixed connector not picking up configuration changes when running locally using the ddn CLI workflow. (#133)

Managing native queries with the CLI

New in this release is a CLI plugin command to create, list, inspect, and delete
native queries. A big advantage of using the command versus writing native query
configurations by hand is that the command will type-check your query's
aggregation pipeline, and will write type declarations automatically.

This is a BETA feature - it is a work in progress, and will not work for all
cases. It is safe to experiment with since it is limited to managing native
query configuration files, and does not lock you into anything.

You can run the new command like this:

$ ddn connector plugin --connector app/connector/my_connector/connector.yaml -- native-query

To create a native query create a file with a .json extension that contains
the aggregation pipeline for you query. For example this pipeline in
title_word_frequency.json outputs frequency counts for words appearing in
movie titles in a given year:

[
  {
    "$match": {
      "year": "{{ year }}"
    }
  },
  { 
    "$replaceWith": {
      "title_words": { "$split": ["$title", " "] }
    }
  },
  { "$unwind": { "path": "$title_words" } },
  { 
    "$group": {
      "_id": "$title_words",
      "count": { "$count": {} }
    }
  }
]

In your supergraph directory run a command like this using the path to the pipeline file as an argument,

$ ddn connector plugin --connector app/connector/my_connector/connector.yaml -- native-query create title_word_frequency.json --collection movies

You should see output like this:

Wrote native query configuration to your-project/connector/native_queries/title_word_frequency.json

input collection: movies
representation: collection

v1.4.0

Added

• Adds _in and _nin operators (#122)

Changed

• BREAKING: If configuration.json cannot be parsed the connector will fail to start. This change also prohibits unknown keys in that file. These changes will help to prevent typos configuration being silently ignored. (#115)

Fixed

• Fixes for filtering by complex predicate that references variables, or field names that require escaping (#111)
• Escape names if necessary instead of failing when joining relationship on field names with special characters (#113)

_in and _nin

These operators compare document values for equality against a given set of
options. _in matches documents where one of the given values matches, _nin matches
documents where none of the given values matches. For example this query selects
movies that are rated either "G" or "TV-G":

query {
  movies(
    where: { rated: { _in: ["G", "TV-G"] } }
    order_by: { id: Asc }
    limit: 5
  ) {
    title
    rated
  }
}

v1.3.0

Added

Fixed

• Selecting nested fields with names that begin with a dollar sign (#108)
• Sorting by fields with names that begin with a dollar sign (#109)

Changed

v1.2.0

Added

• Extended JSON fields now support all comparison and aggregation functions (#99)
• Update to ndc-spec v0.1.6 which allows filtering by object values in array fields (#101)

Filtering by values in arrays

In this update you can filter by making comparisons to object values inside
arrays. For example consider a MongoDB database with these three documents:

{ "institution": "Black Mesa", "staff": [{ "name": "Freeman" }, { "name": "Calhoun" }] }
{ "institution": "Aperture Science", "staff": [{ "name": "GLaDOS" }, { "name": "Chell" }] }
{ "institution": "City 17", "staff": [{ "name": "Alyx" }, { "name": "Freeman" }, { "name": "Breen" }] }

You can now write a GraphQL query with a where clause that checks individual
entries in the staff arrays:

query {
  institutions(where: { staff: { name: { _eq: "Freeman" } } }) {
    institution
  }
}

Which produces the result:

{ "data": { "institutions": [
  { "institution": "Black Mesa" },
  { "institution": "City 17" } 
] } }

The filter selects documents where any element in the array passes the
condition. If you want to select only documents where every array element
passes then negate the comparison on array element values, and also negate the
entire predicate like this:

query EveryElementMustMatch {
  institutions(
    where: { _not: { staff: { name: { _neq: "Freeman" } } } }
  ) {
    institution
  }
}

Note: It is currently only possible to filter on arrays that contain
objects. Filtering on arrays that contain scalar values or nested arrays will
come later.

To configure DDN metadata to filter on array fields configure the
BooleanExpressionType for the containing document object type to use an
object boolean expression type for comparisons on the array field. The
GraphQL Engine will transparently distribute object comparisons over array
elements. For example the above example is configured with this boolean
expression type for documents:

---
kind: BooleanExpressionType
version: v1
definition:
  name: InstitutionComparisonExp
  operand:
    object:
      type: Institution
      comparableFields:
        - fieldName: id
          booleanExpressionType: ObjectIdComparisonExp
        - fieldName: institution
          booleanExpressionType: StringComparisonExp
        - fieldName: staff
          booleanExpressionType: InstitutionStaffComparisonExp
      comparableRelationships: []
  logicalOperators:
    enable: true
  isNull:
    enable: true
  graphql:
    typeName: InstitutionComparisonExp

InstitutionStaffComparisonExp is the boolean expression type for objects
inside the staff array. It looks like this:

---
kind: BooleanExpressionType
version: v1
definition:
  name: InstitutionStaffComparisonExp
  operand:
    object:
      type: InstitutionStaff
      comparableFields:
        - fieldName: name
          booleanExpressionType: StringComparisonExp
      comparableRelationships: []
  logicalOperators:
    enable: true
  isNull:
    enable: true
  graphql:
    typeName: InstitutionStaffComparisonExp

Fixed

Changed

v1.1.0

• Accept predicate arguments in native mutations and native queries (#92)
• Serialize aggregate results as simple JSON (instead of Extended JSON) for
  consistency with non-aggregate result serialization (#96)

v1.0.0

• Fix bug with operator lookup when filtering on nested fields (#82)
• Rework query plans for requests with variable sets to allow use of indexes (#83)
• Fix: error when requesting query plan if MongoDB is target of a remote join (#83)
• Fix: count aggregates return 0 instead of null if no rows match (#85)
• Breaking change: remote joins no longer work in MongoDB v5 (#83)
• Add configuration option to opt into "relaxed" mode for Extended JSON outputs (#84)

v0.1.0

• Support filtering and sorting by fields of related collections (#72)
• Support for root collection column references (#75)
• Fix for databases with field names that begin with a dollar sign, or that contain dots (#74)
• Implement column-to-column comparisons within the same collection (#74)
• Fix error tracking collection with no documents by skipping such collections during CLI introspection (#76)
• If a field contains both int and double values then the field type is inferred as double instead of ExtendedJSON (#77)
• Fix: schema generated with _id column nullable when introspecting schema via sampling (#78)
• Don't require _id field to have type ObjectId when generating primary uniqueness constraint (#79)

v0.0.6

• Enables logging events from the MongoDB driver by setting the RUST_LOG variable (#67)
  • To log all events set RUST_LOG=mongodb::command=debug,mongodb::connection=debug,mongodb::server_selection=debug,mongodb::topology=debug
• Relations with a single column mapping now use concise correlated subquery syntax in $lookup stage (#65)
• Add root configuration.json or configuration.yaml file to allow editing cli options. (#68)
• Update default sample size to 100. (#68)
• Add all_schema_nullable option defaulted to true. (#68)
• Change native_procedure to native_mutation along with code renaming (#70)
  • Note: native_procedures folder in configuration is now deprecated. It will continue to work for a few releases, but renaming your folder is all that is needed.

v0.0.5

• Fix incorrect order of results for query requests with more than 10 variable sets (#37)
• In the CLI update command, don't overwrite schema files that haven't changed (#49)
• In the CLI update command, if the database URI is not provided the error message now mentions the correct environment variable to use (MONGODB_DATABASE_URI) (#50)
• Update to latest NDC SDK (#51)
• Update rustls dependency to fix https://github.com/hasura/ndc-mongodb/security/dependabot/1 (#51)
• Serialize query and mutation response fields with known types using simple JSON instead of Extended JSON (#53) (#59)
• Add trace spans (#58)

v0.0.4

• Queries that attempt to compare a column to a column in the query root table, or a related table, will now fail instead of giving the incorrect result (#22)
• Fix bug in v2 to v3 conversion of query responses containing nested objects (PR #27)
• Fixed bug where use of aggregate functions in queries would fail (#26)
• Rename Any type to ExtendedJSON to make its representation clearer (#30)
• The collection primary key _id property now has a unique constraint generated in the NDC schema for it (#32)