6034126612

Committed 31 Aug 2023 07:03AM UTC coverage: 77.119%. First build

Build # 6034126612

Build Type

Pull #1945

github

Committed by

Jesse-Bakker

Commit Message

Write dozer.lock

Pull Request Pull Request #1945: Write dozer.lock

Run Details

86 of 86 new or added lines in 9 files covered. (100.0%)

49169 of 63757 relevant lines covered (77.12%)

69984.15 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

82.64

/dozer-cli/src/simple/executor.rs

use dozer_api::grpc::internal::internal_pipeline_server::LogEndpoint;
use dozer_cache::dozer_log::camino::Utf8Path;
use dozer_cache::dozer_log::home_dir::{BuildPath, HomeDir};
use dozer_cache::dozer_log::replication::Log;
use dozer_core::checkpoint::{CheckpointFactory, CheckpointFactoryOptions};
use dozer_core::processor_record::ProcessorRecordStore;
use dozer_tracing::LabelsAndProgress;
use dozer_types::models::api_endpoint::ApiEndpoint;
use dozer_types::models::flags::Flags;
use dozer_types::parking_lot::Mutex;
use tokio::runtime::Runtime;

use std::sync::{atomic::AtomicBool, Arc};

use dozer_types::models::source::Source;

use crate::pipeline::PipelineBuilder;
use crate::shutdown::ShutdownReceiver;
use dozer_core::executor::{DagExecutor, ExecutorOptions};

use dozer_types::models::connection::Connection;

use crate::errors::{BuildError, OrchestrationError};

use super::Contract;

pub struct Executor<'a> {
    connections: &'a [Connection],
    sources: &'a [Source],
    sql: Option<&'a str>,
    checkpoint_factory: Arc<CheckpointFactory>,
    /// `ApiEndpoint` and its log.
    endpoint_and_logs: Vec<(ApiEndpoint, LogEndpoint)>,
    labels: LabelsAndProgress,
}

impl<'a> Executor<'a> {
    // TODO: Refactor this to not require both `contract` and all of
    // connections, sources and sql
    #[allow(clippy::too_many_arguments)]
    pub async fn new(
        home_dir: &'a HomeDir,
        contract: &Contract,
        connections: &'a [Connection],
        sources: &'a [Source],
        sql: Option<&'a str>,
        api_endpoints: &'a [ApiEndpoint],
        checkpoint_factory_options: CheckpointFactoryOptions,
        labels: LabelsAndProgress,
    ) -> Result<Executor<'a>, OrchestrationError> {
        // Find the build path.
        let build_path = home_dir
            .find_latest_build_path()
            .map_err(|(path, error)| OrchestrationError::FileSystem(path.into(), error))?
            .ok_or(OrchestrationError::NoBuildFound)?;

        // Load pipeline checkpoint.
        let record_store = ProcessorRecordStore::new()?;
        let checkpoint_factory = CheckpointFactory::new(
            Arc::new(record_store),
            build_path.data_dir.to_string(),
            checkpoint_factory_options,
        )
        .await?
        .0;

        let mut endpoint_and_logs = vec![];
        for endpoint in api_endpoints {
            let log_endpoint =
                create_log_endpoint(contract, &build_path, &endpoint.name, &checkpoint_factory)
                    .await?;
            endpoint_and_logs.push((endpoint.clone(), log_endpoint));
        }

        Ok(Executor {
            connections,
            sources,
            sql,
            checkpoint_factory: Arc::new(checkpoint_factory),
            endpoint_and_logs,
            labels,
        })
    }

    pub fn endpoint_and_logs(&self) -> &[(ApiEndpoint, LogEndpoint)] {
        &self.endpoint_and_logs
    }

    pub async fn create_dag_executor(
        &self,
        runtime: &Arc<Runtime>,
        executor_options: ExecutorOptions,
        shutdown: ShutdownReceiver,
        flags: Flags,
    ) -> Result<DagExecutor, OrchestrationError> {
        let builder = PipelineBuilder::new(
            self.connections,
            self.sources,
            self.sql,
            self.endpoint_and_logs
                .iter()
                .map(|(endpoint, log)| (endpoint.clone(), Some(log.log.clone())))
                .collect(),
            self.labels.clone(),
            flags,
        );

        let dag = builder.build(runtime, shutdown).await?;
        let exec = DagExecutor::new(dag, self.checkpoint_factory.clone(), executor_options)?;

        Ok(exec)
    }
}

pub fn run_dag_executor(
    dag_executor: DagExecutor,
    running: Arc<AtomicBool>,
    labels: LabelsAndProgress,
) -> Result<(), OrchestrationError> {
    let join_handle = dag_executor.start(running, labels)?;
    join_handle
        .join()
        .map_err(OrchestrationError::ExecutionError)
}

async fn create_log_endpoint(
    contract: &Contract,
    build_path: &BuildPath,
    endpoint_name: &str,
    checkpoint_factory: &CheckpointFactory,
) -> Result<LogEndpoint, OrchestrationError> {
    let endpoint_path = build_path.get_endpoint_path(endpoint_name);

    let schema = contract
        .endpoints
        .get(endpoint_name)
        .ok_or_else(|| BuildError::MissingEndpoint(endpoint_name.to_owned()))?;
    let schema_string =
        dozer_types::serde_json::to_string(schema).map_err(BuildError::SerdeJson)?;

    let descriptor_bytes = tokio::fs::read(&build_path.descriptor_path)
        .await
        .map_err(|e| {
            OrchestrationError::FileSystem(build_path.descriptor_path.clone().into(), e)
        })?;

    let log_prefix = AsRef::<Utf8Path>::as_ref(checkpoint_factory.prefix())
        .join(&endpoint_path.log_dir_relative_to_data_dir);
    let log = Log::new(checkpoint_factory.storage(), log_prefix.into(), false).await?;
    let log = Arc::new(Mutex::new(log));

    Ok(LogEndpoint {
        build_id: build_path.id.clone(),
        schema_string,
        descriptor_bytes,
        log,
    })
}

1	use dozer_api::grpc::internal::internal_pipeline_server::LogEndpoint;
2	use dozer_cache::dozer_log::camino::Utf8Path;
3	use dozer_cache::dozer_log::home_dir::{BuildPath, HomeDir};
4	use dozer_cache::dozer_log::replication::Log;
5	use dozer_core::checkpoint::{CheckpointFactory, CheckpointFactoryOptions};
6	use dozer_core::processor_record::ProcessorRecordStore;
7	use dozer_tracing::LabelsAndProgress;
8	use dozer_types::models::api_endpoint::ApiEndpoint;
9	use dozer_types::models::flags::Flags;
10	use dozer_types::parking_lot::Mutex;
11	use tokio::runtime::Runtime;
12
13	use std::sync::{atomic::AtomicBool, Arc};
14
15	use dozer_types::models::source::Source;
16
17	use crate::pipeline::PipelineBuilder;
18	use crate::shutdown::ShutdownReceiver;
19	use dozer_core::executor::{DagExecutor, ExecutorOptions};
20
21	use dozer_types::models::connection::Connection;
22
23	use crate::errors::{BuildError, OrchestrationError};
24
25	use super::Contract;
26
27	pub struct Executor<'a> {
28	connections: &'a [Connection],
29	sources: &'a [Source],
30	sql: Option<&'a str>,
31	checkpoint_factory: Arc<CheckpointFactory>,
32	/// `ApiEndpoint` and its log.
33	endpoint_and_logs: Vec<(ApiEndpoint, LogEndpoint)>,
34	labels: LabelsAndProgress,
35	}
36
37	impl<'a> Executor<'a> {
38	// TODO: Refactor this to not require both `contract` and all of	×
39	// connections, sources and sql	×
40	#[allow(clippy::too_many_arguments)]	×
41	pub async fn new(	30✔
42	home_dir: &'a HomeDir,	30✔
43	contract: &Contract,	30✔
44	connections: &'a [Connection],	30✔
45	sources: &'a [Source],	30✔
46	sql: Option<&'a str>,	30✔
47	api_endpoints: &'a [ApiEndpoint],	30✔
48	checkpoint_factory_options: CheckpointFactoryOptions,	30✔
49	labels: LabelsAndProgress,	30✔
50	) -> Result<Executor<'a>, OrchestrationError> {	30✔
51	// Find the build path.	×
52	let build_path = home_dir	30✔
53	.find_latest_build_path()	30✔
54	.map_err(\|(path, error)\| OrchestrationError::FileSystem(path.into(), error))?	30✔
55	.ok_or(OrchestrationError::NoBuildFound)?;	30✔
56		×
57	// Load pipeline checkpoint.	×
58	let record_store = ProcessorRecordStore::new()?;	30✔
59	let checkpoint_factory = CheckpointFactory::new(	30✔
60	Arc::new(record_store),	30✔
61	build_path.data_dir.to_string(),	30✔
62	checkpoint_factory_options,	30✔
63	)	30✔
64	.await?	30✔
65	.0;	×
66		×
67	let mut endpoint_and_logs = vec![];	30✔
68	for endpoint in api_endpoints {	60✔
69	let log_endpoint =	30✔
70	create_log_endpoint(contract, &build_path, &endpoint.name, &checkpoint_factory)	30✔
71	.await?;	90✔
72	endpoint_and_logs.push((endpoint.clone(), log_endpoint));	30✔
73	}	×
74		×
75	Ok(Executor {	30✔
76	connections,	30✔
77	sources,	30✔
78	sql,	30✔
79	checkpoint_factory: Arc::new(checkpoint_factory),	30✔
80	endpoint_and_logs,	30✔
81	labels,	30✔
82	})	30✔
83	}	30✔
84		×
85	pub fn endpoint_and_logs(&self) -> &[(ApiEndpoint, LogEndpoint)] {	30✔
86	&self.endpoint_and_logs	30✔
87	}	30✔
88		×
89	pub async fn create_dag_executor(	30✔
90	&self,	30✔
91	runtime: &Arc<Runtime>,	30✔
92	executor_options: ExecutorOptions,	30✔
93	shutdown: ShutdownReceiver,	30✔
94	flags: Flags,	30✔
95	) -> Result<DagExecutor, OrchestrationError> {	30✔
96	let builder = PipelineBuilder::new(	30✔
97	self.connections,	30✔
98	self.sources,	30✔
99	self.sql,	30✔
100	self.endpoint_and_logs	30✔
101	.iter()	30✔
102	.map(\|(endpoint, log)\| (endpoint.clone(), Some(log.log.clone())))	30✔
103	.collect(),	30✔
104	self.labels.clone(),	30✔
105	flags,	30✔
106	);	30✔
107		×
108	let dag = builder.build(runtime, shutdown).await?;	60✔
109	let exec = DagExecutor::new(dag, self.checkpoint_factory.clone(), executor_options)?;	30✔
110		×
111	Ok(exec)	30✔
112	}	30✔
113	}	×
114		×
115	pub fn run_dag_executor(	30✔
116	dag_executor: DagExecutor,	30✔
117	running: Arc<AtomicBool>,	30✔
118	labels: LabelsAndProgress,	30✔
119	) -> Result<(), OrchestrationError> {	30✔
120	let join_handle = dag_executor.start(running, labels)?;	30✔
121	join_handle	30✔
122	.join()	30✔
123	.map_err(OrchestrationError::ExecutionError)	30✔
124	}	30✔
125		×
126	async fn create_log_endpoint(	30✔
127	contract: &Contract,	30✔
128	build_path: &BuildPath,	30✔
129	endpoint_name: &str,	30✔
130	checkpoint_factory: &CheckpointFactory,	30✔
131	) -> Result<LogEndpoint, OrchestrationError> {	30✔
132	let endpoint_path = build_path.get_endpoint_path(endpoint_name);	30✔
133		×
134	let schema = contract	30✔
135	.endpoints	30✔
136	.get(endpoint_name)	30✔
137	.ok_or_else(\|\| BuildError::MissingEndpoint(endpoint_name.to_owned()))?;	30✔
138	let schema_string =	30✔
139	dozer_types::serde_json::to_string(schema).map_err(BuildError::SerdeJson)?;	30✔
140		×
141	let descriptor_bytes = tokio::fs::read(&build_path.descriptor_path)	30✔
142	.await	30✔
143	.map_err(\|e\| {	30✔
144	OrchestrationError::FileSystem(build_path.descriptor_path.clone().into(), e)	×
145	})?;	30✔
146		×
147	let log_prefix = AsRef::<Utf8Path>::as_ref(checkpoint_factory.prefix())	30✔
148	.join(&endpoint_path.log_dir_relative_to_data_dir);	30✔
149	let log = Log::new(checkpoint_factory.storage(), log_prefix.into(), false).await?;	60✔
150	let log = Arc::new(Mutex::new(log));	30✔
151		30✔
152	Ok(LogEndpoint {	30✔
153	build_id: build_path.id.clone(),	30✔
154	schema_string,	30✔
155	descriptor_bytes,	30✔
156	log,	30✔
157	})	30✔
158	}	30✔

getdozer / dozer / 6034126612

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous