CLI plugin

Cargo.toml

@@ -10,7 +10,8 @@ members = [
   "crates/dc-api",
   "crates/dc-api-types",
   "crates/dc-api-test-helpers",
-  "crates/ndc-test-helpers"
+  "crates/ndc-test-helpers",
+  "crates/cli"
 ]
 resolver = "2"
 

connector-definition/Makefile

@@ -0,0 +1,21 @@
+.DEFAULT_GOAL := build
+SHELL = /usr/bin/env bash
+
+.PHONY: build
+build: dist/connector-definition.tgz
+
+.PHONY: clean
+clean:
+	rm -rf dist
+
+dist dist/.hasura-connector:
+	mkdir dist
+	mkdir dist/.hasura-connector
+
+dist/.hasura-connector/connector-metadata.yaml: DOCKER_IMAGE ?= $(error The DOCKER_IMAGE variable must be defined)
+dist/.hasura-connector/connector-metadata.yaml: connector-metadata.yaml dist/.hasura-connector
+	cp -f connector-metadata.yaml dist/.hasura-connector/
+	yq -i '.packagingDefinition.dockerImage = "$(DOCKER_IMAGE)"' dist/.hasura-connector/connector-metadata.yaml
+
+dist/connector-definition.tgz: dist/.hasura-connector/connector-metadata.yaml
+	shopt -s dotglob && cd dist && tar -czvf connector-definition.tgz *

connector-definition/connector-metadata.yaml

@@ -0,0 +1,15 @@
+packagingDefinition:
+  type: PrebuiltDockerImage
+  dockerImage:
+supportedEnvironmentVariables:
+  - name: MONGODB_DATABASE_URI
+    description: The URI for the MongoDB database
+commands:
+  update: hasura-mongodb update
+cliPlugin:
+  name: hasura-mongodb
+  version: "0.0.1"
+dockerComposeWatch:
+  - path: ./
+    target: /etc/connector
+    action: sync+restart

crates/cli/Cargo.toml

@@ -0,0 +1,23 @@
+[package]
+name = "mongodb-cli-plugin"
+edition = "2021"
+version.workspace = true
+
+[[bin]]
+name = "hasura-mongodb"
+path = "./src/main.rs"
+
+[dependencies]
+configuration = { path = "../configuration" }
+mongodb-agent-common = { path = "../mongodb-agent-common" }
+mongodb = "2.8"
+mongodb-support = { path = "../mongodb-support" }
+
+anyhow = "1.0.80"
+clap = { version = "4.5.1", features = ["derive", "env"] }
+futures-util = "0.3.28"
+indexmap = { version = "1", features = ["serde"] } # must match the version that ndc-client uses
+serde = { version = "1.0", features = ["derive"] }
+serde_json = { version = "1.0.113", features = ["raw_value"] }
+thiserror = "1.0.57"
+tokio = { version = "1.36.0", features = ["full"] }

crates/cli/src/introspection.rs

@@ -0,0 +1,180 @@
+use configuration::{
+    metadata::{Collection, ObjectField, ObjectType, Type},
+    Metadata,
+};
+use futures_util::{StreamExt, TryStreamExt};
+use indexmap::IndexMap;
+use mongodb::bson::from_bson;
+use mongodb_agent_common::schema::{get_property_description, Property, ValidatorSchema};
+use mongodb_support::{BsonScalarType, BsonType};
+
+use mongodb_agent_common::interface_types::{MongoAgentError, MongoConfig};
+
+pub async fn get_metadata_from_validation_schema(
+    config: &MongoConfig,
+) -> Result<Metadata, MongoAgentError> {
+    let db = config.client.database(&config.database);
+    let collections_cursor = db.list_collections(None, None).await?;
+
+    let (object_types, collections) = collections_cursor
+        .into_stream()
+        .map(
+            |collection_spec| -> Result<(Vec<ObjectType>, Collection), MongoAgentError> {
+                let collection_spec_value = collection_spec?;
+                let name = &collection_spec_value.name;
+                let schema_bson_option = collection_spec_value
+                    .options
+                    .validator
+                    .as_ref()
+                    .and_then(|x| x.get("$jsonSchema"));
+
+                match schema_bson_option {
+                    Some(schema_bson) => {
+                        from_bson::<ValidatorSchema>(schema_bson.clone()).map_err(|err| {
+                            MongoAgentError::BadCollectionSchema(
+                                name.to_owned(),
+                                schema_bson.clone(),
+                                err,
+                            )
+                        })
+                    }
+                    None => Ok(ValidatorSchema {
+                        bson_type: BsonType::Object,
+                        description: None,
+                        required: Vec::new(),
+                        properties: IndexMap::new(),
+                    }),
+                }
+                .map(|validator_schema| make_collection(name, &validator_schema))
+            },
+        )
+        .try_collect::<(Vec<Vec<ObjectType>>, Vec<Collection>)>()
+        .await?;
+
+    Ok(Metadata {
+        collections,
+        object_types: object_types.concat(),
+    })
+}
+
+fn make_collection(
+    collection_name: &str,
+    validator_schema: &ValidatorSchema,
+) -> (Vec<ObjectType>, Collection) {
+    let properties = &validator_schema.properties;
+    let required_labels = &validator_schema.required;
+
+    let (mut object_type_defs, object_fields) = {
+        let type_prefix = format!("{collection_name}_");
+        let id_field = ObjectField {
+            name: "_id".to_string(),
+            description: Some("primary key _id".to_string()),
+            r#type: Type::Scalar(BsonScalarType::ObjectId),
+        };
+        let (object_type_defs, mut object_fields): (Vec<Vec<ObjectType>>, Vec<ObjectField>) =
+            properties
+                .iter()
+                .map(|prop| make_object_field(&type_prefix, required_labels, prop))
+                .unzip();
+        if !object_fields.iter().any(|info| info.name == "_id") {
+            // There should always be an _id field, so add it unless it was already specified in
+            // the validator.
+            object_fields.push(id_field);
+        }
+        (object_type_defs.concat(), object_fields)
+    };
+
+    let collection_type = ObjectType {
+        name: collection_name.to_string(),
+        description: Some(format!("Object type for collection {collection_name}")),
+        fields: object_fields,
+    };
+
+    object_type_defs.push(collection_type);
+
+    let collection_info = Collection {
+        name: collection_name.to_string(),
+        description: validator_schema.description.clone(),
+        r#type: collection_name.to_string(),
+    };
+
+    (object_type_defs, collection_info)
+}
+
+fn make_object_field(
+    type_prefix: &str,
+    required_labels: &[String],
+    (prop_name, prop_schema): (&String, &Property),
+) -> (Vec<ObjectType>, ObjectField) {
+    let description = get_property_description(prop_schema);
+
+    let object_type_name = format!("{type_prefix}{prop_name}");
+    let (collected_otds, field_type) = make_field_type(&object_type_name, prop_schema);
+
+    let object_field = ObjectField {
+        name: prop_name.clone(),
+        description,
+        r#type: maybe_nullable(field_type, !required_labels.contains(prop_name)),
+    };
+
+    (collected_otds, object_field)
+}
+
+fn maybe_nullable(
+    t: configuration::metadata::Type,
+    is_nullable: bool,
+) -> configuration::metadata::Type {
+    if is_nullable {
+        configuration::metadata::Type::Nullable(Box::new(t))
+    } else {
+        t
+    }
+}
+
+fn make_field_type(object_type_name: &str, prop_schema: &Property) -> (Vec<ObjectType>, Type) {
+    let mut collected_otds: Vec<ObjectType> = vec![];
+
+    match prop_schema {
+        Property::Object {
+            bson_type: _,
+            description: _,
+            required,
+            properties,
+        } => {
+            let type_prefix = format!("{object_type_name}_");
+            let (otds, otd_fields): (Vec<Vec<ObjectType>>, Vec<ObjectField>) = properties
+                .iter()
+                .map(|prop| make_object_field(&type_prefix, required, prop))
+                .unzip();
+
+            let object_type_definition = ObjectType {
+                name: object_type_name.to_string(),
+                description: Some("generated from MongoDB validation schema".to_string()),
+                fields: otd_fields,
+            };
+
+            collected_otds.append(&mut otds.concat());
+            collected_otds.push(object_type_definition);
+
+            (collected_otds, Type::Object(object_type_name.to_string()))
+        }
+        Property::Array {
+            bson_type: _,
+            description: _,
+            items,
+        } => {
+            let item_schemas = *items.clone();
+
+            let (mut otds, element_type) = make_field_type(object_type_name, &item_schemas);
+            let field_type = Type::ArrayOf(Box::new(element_type));
+
+            collected_otds.append(&mut otds);
+
+            (collected_otds, field_type)
+        }
+        Property::Scalar {
+            bson_type,
+            description: _,
+        } => (collected_otds, Type::Scalar(bson_type.to_owned())),
+    }
+}

crates/cli/src/lib.rs

@@ -0,0 +1,40 @@
+//! The interpretation of the commands that the CLI can handle.
+
+mod introspection;
+
+use std::path::PathBuf;
+
+use clap::Subcommand;
+
+use configuration::Configuration;
+use mongodb_agent_common::interface_types::MongoConfig;
+
+/// The command invoked by the user.
+#[derive(Debug, Clone, Subcommand)]
+pub enum Command {
+    /// Update the configuration by introspecting the database, using the configuration options.
+    Update,
+}
+
+pub struct Context {
+    pub path: PathBuf,
+    pub mongo_config: MongoConfig,
+}
+
+/// Run a command in a given directory.
+pub async fn run(command: Command, context: &Context) -> anyhow::Result<()> {
+    match command {
+        Command::Update => update(context).await?,
+    };
+    Ok(())
+}
+
+/// Update the configuration in the current directory by introspecting the database.
+async fn update(context: &Context) -> anyhow::Result<()> {
+    let metadata = introspection::get_metadata_from_validation_schema(&context.mongo_config).await?;
+    let configuration = Configuration { metadata };
+
+    configuration::write_directory(&context.path, &configuration).await?;
+
+    Ok(())
+}

crates/cli/src/main.rs

@@ -0,0 +1,54 @@
+//! The CLI application. This is used to configure a deployment of mongo-agent-v3.
+//!
+//! This is intended to be automatically downloaded and invoked via the Hasura CLI, as a plugin.
+//! It is unlikely that end-users will use it directly.
+
+use anyhow::anyhow;
+use std::env;
+use std::path::PathBuf;
+
+use clap::Parser;
+use mongodb_agent_common::state::{try_init_state_from_uri, DATABASE_URI_ENV_VAR};
+use mongodb_cli_plugin::{run, Command, Context};
+
+/// The command-line arguments.
+#[derive(Debug, Parser)]
+pub struct Args {
+    /// The path to the configuration. Defaults to the current directory.
+    #[arg(
+        long = "context-path",
+        env = "HASURA_PLUGIN_CONNECTOR_CONTEXT_PATH",
+        value_name = "DIRECTORY"
+    )]
+    pub context_path: Option<PathBuf>,
+
+    #[arg(
+        long = "connection-uri",
+        env = DATABASE_URI_ENV_VAR,
+        required = true,
+        value_name = "URI"
+    )]
+    pub connection_uri: String,
+
+    /// The command to invoke.
+    #[command(subcommand)]
+    pub subcommand: Command,
+}
+
+/// The application entrypoint. It pulls information from the environment and then calls the [run]
+/// function. The library remains unaware of the environment, so that we can more easily test it.
+#[tokio::main]
+pub async fn main() -> anyhow::Result<()> {
+    let args = Args::parse();
+    // Default the context path to the current directory.
+    let path = match args.context_path {
+        Some(path) => path,
+        None => env::current_dir()?,
+    };
+    let mongo_config = try_init_state_from_uri(&args.connection_uri)
+        .await
+        .map_err(|e| anyhow!("Error initializing MongoDB state {}", e))?;
+    let context = Context { path, mongo_config };
+    run(args.subcommand, &context).await?;
+    Ok(())
+}

crates/configuration/src/configuration.rs

@@ -1,11 +1,11 @@
 use std::{io, path::Path};
 
 use schemars::JsonSchema;
-use serde::Deserialize;
+use serde::{Deserialize, Serialize};
 
 use crate::{read_directory, Metadata};
 
-#[derive(Clone, Debug, Default, Deserialize, JsonSchema)]
+#[derive(Clone, Debug, Default, Serialize, Deserialize, JsonSchema)]
 #[serde(rename_all = "camelCase")]
 pub struct Configuration {
     pub metadata: Metadata,