4075835066

Build Type

github

Committed by GitHub

Commit Message

Merge 39f3c7143 into 3223082a5

Pull Request Pull Request #790: refactor: Use `daggy` for the underlying data structure of `Dag`

Run Details

396 of 396 new or added lines in 11 files covered. (100.0%)

24551 of 36528 relevant lines covered (67.21%)

54898.93 hits per line

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

84.54

/dozer-sql/src/pipeline/tests/builder_test.rs

use dozer_core::dag::app::App;
use dozer_core::dag::appsource::{AppSource, AppSourceManager};
use dozer_core::dag::channels::SourceChannelForwarder;
use dozer_core::dag::dag::DEFAULT_PORT_HANDLE;
use dozer_core::dag::errors::ExecutionError;
use dozer_core::dag::executor::{DagExecutor, ExecutorOptions};
use dozer_core::dag::node::{
    OutputPortDef, OutputPortType, PortHandle, Sink, SinkFactory, Source, SourceFactory,
};
use dozer_core::dag::record_store::RecordReader;
use dozer_core::storage::lmdb_storage::{LmdbEnvironmentManager, SharedTransaction};
use dozer_types::log::debug;
use dozer_types::ordered_float::OrderedFloat;
use dozer_types::types::{
    Field, FieldDefinition, FieldType, Operation, Record, Schema, SourceDefinition,
};

use dozer_core::dag::epoch::Epoch;

use std::collections::HashMap;
use std::fs;

use std::sync::atomic::AtomicBool;
use std::sync::Arc;
use tempdir::TempDir;

use crate::pipeline::builder::{statement_to_pipeline, SchemaSQLContext};

/// Test Source
#[derive(Debug)]
pub struct TestSourceFactory {
    output_ports: Vec<PortHandle>,
}

impl TestSourceFactory {
    pub fn new(output_ports: Vec<PortHandle>) -> Self {
        Self { output_ports }
    }
}

impl SourceFactory<SchemaSQLContext> for TestSourceFactory {
    fn get_output_ports(&self) -> Result<Vec<OutputPortDef>, ExecutionError> {
        Ok(self
            .output_ports
            .iter()
            .map(|e| OutputPortDef::new(*e, OutputPortType::Stateless))
            .collect())
    }

    fn get_output_schema(
        &self,
        _port: &PortHandle,
    ) -> Result<(Schema, SchemaSQLContext), ExecutionError> {
        Ok((
            Schema::empty()
                .field(
                    FieldDefinition::new(
                        String::from("CustomerID"),
                        FieldType::Int,
                        false,
                        SourceDefinition::Dynamic,
                    ),
                    false,
                )
                .field(
                    FieldDefinition::new(
                        String::from("Country"),
                        FieldType::String,
                        false,
                        SourceDefinition::Dynamic,
                    ),
                    false,
                )
                .field(
                    FieldDefinition::new(
                        String::from("Spending"),
                        FieldType::Float,
                        false,
                        SourceDefinition::Dynamic,
                    ),
                    false,
                )
                .clone(),
            SchemaSQLContext::default(),
        ))
    }

    fn build(
        &self,
        _output_schemas: HashMap<PortHandle, Schema>,
    ) -> Result<Box<dyn Source>, ExecutionError> {
        Ok(Box::new(TestSource {}))
    }

    fn prepare(
        &self,
        _output_schemas: HashMap<PortHandle, (Schema, SchemaSQLContext)>,
    ) -> Result<(), ExecutionError> {
        Ok(())
    }
}

#[derive(Debug)]
pub struct TestSource {}

impl Source for TestSource {
    fn start(
        &self,
        fw: &mut dyn SourceChannelForwarder,
        _from_seq: Option<(u64, u64)>,
    ) -> Result<(), ExecutionError> {
        for n in 0..10000 {
            fw.send(
                n,
                0,
                Operation::Insert {
                    new: Record::new(
                        None,
                        vec![
                            Field::Int(0),
                            Field::String("Italy".to_string()),
                            Field::Float(OrderedFloat(5.5)),
                        ],
                        None,
                    ),
                },
                DEFAULT_PORT_HANDLE,
            )
            .unwrap();
        }
        Ok(())
    }
}

#[derive(Debug)]
pub struct TestSinkFactory {
    input_ports: Vec<PortHandle>,
}

impl TestSinkFactory {
    pub fn new(input_ports: Vec<PortHandle>) -> Self {
        Self { input_ports }
    }
}

impl SinkFactory<SchemaSQLContext> for TestSinkFactory {
    fn get_input_ports(&self) -> Vec<PortHandle> {
        self.input_ports.clone()
    }

    fn build(
        &self,
        _input_schemas: HashMap<PortHandle, Schema>,
    ) -> Result<Box<dyn Sink>, ExecutionError> {
        Ok(Box::new(TestSink {}))
    }

    fn prepare(
        &self,
        _input_schemas: HashMap<PortHandle, (Schema, SchemaSQLContext)>,
    ) -> Result<(), ExecutionError> {
        Ok(())
    }
}

#[derive(Debug)]
pub struct TestSink {}

impl Sink for TestSink {
    fn init(&mut self, _env: &mut LmdbEnvironmentManager) -> Result<(), ExecutionError> {
        debug!("SINK: Initialising TestSink");
        Ok(())
    }

    fn process(
        &mut self,
        _from_port: PortHandle,
        _op: Operation,
        _state: &SharedTransaction,
        _reader: &HashMap<PortHandle, Box<dyn RecordReader>>,
    ) -> Result<(), ExecutionError> {
        Ok(())
    }

    fn commit(&mut self, _epoch: &Epoch, _tx: &SharedTransaction) -> Result<(), ExecutionError> {
        Ok(())
    }
}

#[test]
fn test_pipeline_builder() {
    let (mut pipeline, (node, node_port)) = statement_to_pipeline(
        "SELECT COUNT(Spending), users.Country \
    FROM users \
    WHERE Spending >= 1",
    )
    .unwrap();

    let mut asm = AppSourceManager::new();
    asm.add(AppSource::new(
        "mem".to_string(),
        Arc::new(TestSourceFactory::new(vec![DEFAULT_PORT_HANDLE])),
        vec![("users".to_string(), DEFAULT_PORT_HANDLE)]
            .into_iter()
            .collect(),
    ))
    .unwrap();

    pipeline.add_sink(
        Arc::new(TestSinkFactory::new(vec![DEFAULT_PORT_HANDLE])),
        "sink",
    );
    pipeline
        .connect_nodes(&node, Some(node_port), "sink", Some(DEFAULT_PORT_HANDLE))
        .unwrap();

    let mut app = App::new(asm);
    app.add_pipeline(pipeline);

    let dag = app.get_dag().unwrap();

    let tmp_dir = TempDir::new("example").unwrap_or_else(|_e| panic!("Unable to create temp dir"));
    if tmp_dir.path().exists() {
        fs::remove_dir_all(tmp_dir.path()).unwrap_or_else(|_e| panic!("Unable to remove old dir"));
    }
    fs::create_dir(tmp_dir.path()).unwrap_or_else(|_e| panic!("Unable to create temp dir"));

    use std::time::Instant;
    let now = Instant::now();

    let tmp_dir = TempDir::new("test").unwrap();
    let mut executor = DagExecutor::new(
        &dag,
        tmp_dir.path(),
        ExecutorOptions::default(),
        Arc::new(AtomicBool::new(true)),
    )
    .unwrap();

    executor
        .start()
        .unwrap_or_else(|e| panic!("Unable to start the Executor: {e}"));
    assert!(executor.join().is_ok());

    let elapsed = now.elapsed();
    debug!("Elapsed: {:.2?}", elapsed);
}

1	use dozer_core::dag::app::App;
2	use dozer_core::dag::appsource::{AppSource, AppSourceManager};
3	use dozer_core::dag::channels::SourceChannelForwarder;
4	use dozer_core::dag::dag::DEFAULT_PORT_HANDLE;
5	use dozer_core::dag::errors::ExecutionError;
6	use dozer_core::dag::executor::{DagExecutor, ExecutorOptions};
7	use dozer_core::dag::node::{
8	OutputPortDef, OutputPortType, PortHandle, Sink, SinkFactory, Source, SourceFactory,
9	};
10	use dozer_core::dag::record_store::RecordReader;
11	use dozer_core::storage::lmdb_storage::{LmdbEnvironmentManager, SharedTransaction};
12	use dozer_types::log::debug;
13	use dozer_types::ordered_float::OrderedFloat;
14	use dozer_types::types::{
15	Field, FieldDefinition, FieldType, Operation, Record, Schema, SourceDefinition,
16	};
17
18	use dozer_core::dag::epoch::Epoch;
19
20	use std::collections::HashMap;
21	use std::fs;
22
23	use std::sync::atomic::AtomicBool;
24	use std::sync::Arc;
25	use tempdir::TempDir;
26
27	use crate::pipeline::builder::{statement_to_pipeline, SchemaSQLContext};
28
29	/// Test Source
30	#[derive(Debug)]	×
31	pub struct TestSourceFactory {
32	output_ports: Vec<PortHandle>,
33	}
34
35	impl TestSourceFactory {
36	pub fn new(output_ports: Vec<PortHandle>) -> Self {	1✔
37	Self { output_ports }	1✔
38	}	1✔
39	}
40
41	impl SourceFactory<SchemaSQLContext> for TestSourceFactory {
42	fn get_output_ports(&self) -> Result<Vec<OutputPortDef>, ExecutionError> {	3✔
43	Ok(self	3✔
44	.output_ports	3✔
45	.iter()	3✔
46	.map(\|e\| OutputPortDef::new(*e, OutputPortType::Stateless))	3✔
47	.collect())	3✔
48	}	3✔
49
50	fn get_output_schema(	1✔
51	&self,	1✔
52	_port: &PortHandle,	1✔
53	) -> Result<(Schema, SchemaSQLContext), ExecutionError> {	1✔
54	Ok((	1✔
55	Schema::empty()	1✔
56	.field(	1✔
57	FieldDefinition::new(	1✔
58	String::from("CustomerID"),	1✔
59	FieldType::Int,	1✔
60	false,	1✔
61	SourceDefinition::Dynamic,	1✔
62	),	1✔
63	false,	1✔
64	)	1✔
65	.field(	1✔
66	FieldDefinition::new(	1✔
67	String::from("Country"),	1✔
68	FieldType::String,	1✔
69	false,	1✔
70	SourceDefinition::Dynamic,	1✔
71	),	1✔
72	false,	1✔
73	)	1✔
74	.field(	1✔
75	FieldDefinition::new(	1✔
76	String::from("Spending"),	1✔
77	FieldType::Float,	1✔
78	false,	1✔
79	SourceDefinition::Dynamic,	1✔
80	),	1✔
81	false,	1✔
82	)	1✔
83	.clone(),	1✔
84	SchemaSQLContext::default(),	1✔
85	))	1✔
86	}	1✔
87
88	fn build(	1✔
89	&self,	1✔
90	_output_schemas: HashMap<PortHandle, Schema>,	1✔
91	) -> Result<Box<dyn Source>, ExecutionError> {	1✔
92	Ok(Box::new(TestSource {}))	1✔
93	}	1✔
94
95	fn prepare(	×
96	&self,	×
97	_output_schemas: HashMap<PortHandle, (Schema, SchemaSQLContext)>,	×
98	) -> Result<(), ExecutionError> {	×
99	Ok(())	×
100	}	×
101	}
102
103	#[derive(Debug)]	×
104	pub struct TestSource {}
105
106	impl Source for TestSource {
107	fn start(	1✔
108	&self,	1✔
109	fw: &mut dyn SourceChannelForwarder,	1✔
110	_from_seq: Option<(u64, u64)>,	1✔
111	) -> Result<(), ExecutionError> {	1✔
112	for n in 0..10000 {	10,001✔
113	fw.send(	10,000✔
114	n,	10,000✔
115	0,	10,000✔
116	Operation::Insert {	10,000✔
117	new: Record::new(	10,000✔
118	None,	10,000✔
119	vec![	10,000✔
120	Field::Int(0),	10,000✔
121	Field::String("Italy".to_string()),	10,000✔
122	Field::Float(OrderedFloat(5.5)),	10,000✔
123	],	10,000✔
124	None,	10,000✔
125	),	10,000✔
126	},	10,000✔
127	DEFAULT_PORT_HANDLE,	10,000✔
128	)	10,000✔
129	.unwrap();	10,000✔
130	}	10,000✔
131	Ok(())	1✔
132	}	1✔
133	}
134
135	#[derive(Debug)]	×
136	pub struct TestSinkFactory {
137	input_ports: Vec<PortHandle>,
138	}
139
140	impl TestSinkFactory {
141	pub fn new(input_ports: Vec<PortHandle>) -> Self {	1✔
142	Self { input_ports }	1✔
143	}	1✔
144	}
145
146	impl SinkFactory<SchemaSQLContext> for TestSinkFactory {
147	fn get_input_ports(&self) -> Vec<PortHandle> {	2✔
148	self.input_ports.clone()	2✔
149	}	2✔
150
151	fn build(	1✔
152	&self,	1✔
153	_input_schemas: HashMap<PortHandle, Schema>,	1✔
154	) -> Result<Box<dyn Sink>, ExecutionError> {	1✔
155	Ok(Box::new(TestSink {}))	1✔
156	}	1✔
157
158	fn prepare(	×
159	&self,	×
160	_input_schemas: HashMap<PortHandle, (Schema, SchemaSQLContext)>,	×
161	) -> Result<(), ExecutionError> {	×
162	Ok(())	×
163	}	×
164	}
165		×
166	#[derive(Debug)]	×
167	pub struct TestSink {}	×
168		×
169	impl Sink for TestSink {	×
170	fn init(&mut self, _env: &mut LmdbEnvironmentManager) -> Result<(), ExecutionError> {	1✔
171	debug!("SINK: Initialising TestSink");	1✔
172	Ok(())	1✔
173	}	1✔
174
175	fn process(	1✔
176	&mut self,	1✔
177	_from_port: PortHandle,	1✔
178	_op: Operation,	1✔
179	_state: &SharedTransaction,	1✔
180	_reader: &HashMap<PortHandle, Box<dyn RecordReader>>,	1✔
181	) -> Result<(), ExecutionError> {	1✔
182	Ok(())	1✔
183	}	1✔
184		×
185	fn commit(&mut self, _epoch: &Epoch, _tx: &SharedTransaction) -> Result<(), ExecutionError> {	1✔
186	Ok(())	1✔
187	}	1✔
188	}	×
189		×
190	#[test]	1✔
191	fn test_pipeline_builder() {	1✔
192	let (mut pipeline, (node, node_port)) = statement_to_pipeline(	1✔
193	"SELECT COUNT(Spending), users.Country \	1✔
194	FROM users \	1✔
195	WHERE Spending >= 1",	1✔
196	)	1✔
197	.unwrap();	1✔
198		1✔
199	let mut asm = AppSourceManager::new();	1✔
200	asm.add(AppSource::new(	1✔
201	"mem".to_string(),	1✔
202	Arc::new(TestSourceFactory::new(vec![DEFAULT_PORT_HANDLE])),	1✔
203	vec![("users".to_string(), DEFAULT_PORT_HANDLE)]	1✔
204	.into_iter()	1✔
205	.collect(),	1✔
206	))	1✔
207	.unwrap();	1✔
208		1✔
209	pipeline.add_sink(	1✔
210	Arc::new(TestSinkFactory::new(vec![DEFAULT_PORT_HANDLE])),	1✔
211	"sink",	1✔
212	);	1✔
213	pipeline	1✔
214	.connect_nodes(&node, Some(node_port), "sink", Some(DEFAULT_PORT_HANDLE))	1✔
215	.unwrap();	1✔
216		1✔
217	let mut app = App::new(asm);	1✔
218	app.add_pipeline(pipeline);	1✔
219		1✔
220	let dag = app.get_dag().unwrap();	1✔
221		1✔
222	let tmp_dir = TempDir::new("example").unwrap_or_else(\|_e\| panic!("Unable to create temp dir"));	1✔
223	if tmp_dir.path().exists() {	1✔
224	fs::remove_dir_all(tmp_dir.path()).unwrap_or_else(\|_e\| panic!("Unable to remove old dir"));	1✔
225	}	1✔
226	fs::create_dir(tmp_dir.path()).unwrap_or_else(\|_e\| panic!("Unable to create temp dir"));	1✔
227		1✔
228	use std::time::Instant;	1✔
229	let now = Instant::now();	1✔
230		1✔
231	let tmp_dir = TempDir::new("test").unwrap();	1✔
232	let mut executor = DagExecutor::new(	1✔
233	&dag,	1✔
234	tmp_dir.path(),	1✔
235	ExecutorOptions::default(),	1✔
236	Arc::new(AtomicBool::new(true)),	1✔
237	)	1✔
238	.unwrap();	1✔
239		1✔
240	executor	1✔
241	.start()	1✔
242	.unwrap_or_else(\|e\| panic!("Unable to start the Executor: {e}"));	1✔
243	assert!(executor.join().is_ok());	1✔
244		×
245	let elapsed = now.elapsed();	1✔
246	debug!("Elapsed: {:.2?}", elapsed);	1✔
247	}	1✔

getdozer / dozer / 4075835066

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