4392484403

Build Type

push

github

Committed by GitHub

Commit Message

feat: Asynchoronous indexing (#1206)

Run Details

270 of 270 new or added lines in 13 files covered. (100.0%)

28714 of 38777 relevant lines covered (74.05%)

89484.24 hits per line

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

81.34

/dozer-cache/src/cache/lmdb/utils.rs

use std::{
    fs,
    ops::Deref,
    path::{Path, PathBuf},
};

use crate::errors::CacheError;
use dozer_storage::{
    lmdb::EnvironmentFlags,
    lmdb_storage::{LmdbEnvironmentManager, LmdbEnvironmentOptions},
};
use tempdir::TempDir;

use super::cache::{CacheCommonOptions, CacheWriteOptions};

#[allow(clippy::type_complexity)]
pub fn init_env(
    common_options: &CacheCommonOptions,
    write_options: Option<CacheWriteOptions>,
) -> Result<(LmdbEnvironmentManager, (PathBuf, String), Option<TempDir>), CacheError> {
    if let Some(write_options) = write_options {
        create_env(common_options, write_options)
    } else {
        let (env, (base_path, name), temp_dir) = open_env(common_options)?;
        Ok((env, (base_path.to_path_buf(), name.to_string()), temp_dir))
    }
}

#[allow(clippy::type_complexity)]
fn create_env(
    common_options: &CacheCommonOptions,
    write_options: CacheWriteOptions,
) -> Result<(LmdbEnvironmentManager, (PathBuf, String), Option<TempDir>), CacheError> {
    let (base_path, name, temp_dir) = match &common_options.path {
        None => {
            let base_path =
                TempDir::new("dozer").map_err(|e| CacheError::Io("tempdir".into(), e))?;
            (
                base_path.path().to_path_buf(),
                "dozer-cache",
                Some(base_path),
            )
        }
        Some((base_path, name)) => {
            fs::create_dir_all(base_path).map_err(|e| CacheError::Io(base_path.clone(), e))?;
            (base_path.clone(), name.deref(), None)
        }
    };

    let options = LmdbEnvironmentOptions::new(
        common_options.max_db_size,
        common_options.max_readers,
        write_options.max_size,
        EnvironmentFlags::empty(),
    );

    Ok((
        LmdbEnvironmentManager::create(&base_path, name, options)?,
        (base_path, name.to_string()),
        temp_dir,
    ))
}

#[allow(clippy::type_complexity)]
fn open_env(
    options: &CacheCommonOptions,
) -> Result<(LmdbEnvironmentManager, (&Path, &str), Option<TempDir>), CacheError> {
    let (base_path, name) = options
        .path
        .as_ref()
        .ok_or(CacheError::PathNotInitialized)?;

    let env_options = LmdbEnvironmentOptions {
        max_dbs: options.max_db_size,
        max_readers: options.max_readers,
        flags: EnvironmentFlags::READ_ONLY,
        ..Default::default()
    };

    Ok((
        LmdbEnvironmentManager::create(base_path, name, env_options)?,
        (base_path, name),
        None,
    ))
}

#[cfg(test)]
mod tests {
    use dozer_storage::{
        lmdb::{Cursor, DatabaseFlags, RoCursor, Transaction, WriteFlags},
        lmdb_storage::CreateDatabase,
    };
    use dozer_types::types::Field;

    use super::*;

    fn cursor_dump(mut cursor: RoCursor) -> Vec<(&[u8], &[u8])> {
        cursor
            .iter_dup()
            .flatten()
            .collect::<dozer_storage::lmdb::Result<Vec<_>>>()
            .unwrap()
    }

    #[test]
    fn duplicate_test_nested() {
        let mut env = create_env(&Default::default(), Default::default())
            .unwrap()
            .0;

        let db = env
            .create_database(
                Some("test"),
                Some(DatabaseFlags::DUP_SORT | DatabaseFlags::INTEGER_KEY),
            )
            .unwrap();

        let txn = env.create_txn().unwrap();
        let mut master_txn = txn.write();
        let txn = master_txn.txn_mut();

        let mut c_txn = txn.begin_nested_txn().unwrap();

        let items: Vec<(i64, &[u8])> = vec![
            (1, b"a"),
            (2, b"a"),
            (3, b"a"),
            (1, b"b"),
            (2, b"b"),
            (3, b"b"),
            (1, b"c"),
            (2, b"c"),
            (3, b"c"),
            (1, b"e"),
            (2, b"e"),
            (3, b"e"),
        ];
        for (key, data) in &items {
            let key = [
                "idx".as_bytes().to_vec(),
                Field::Int(*key).encode(),
                key.to_be_bytes().to_vec(),
            ]
            .join("#".as_bytes());
            c_txn.put(db, &key, data, WriteFlags::empty()).unwrap();
        }
        c_txn.commit().unwrap();
        master_txn.commit_and_renew().unwrap();

        let rtxn = master_txn.txn();

        let cursor = rtxn.open_ro_cursor(db).unwrap();
        let vals = cursor_dump(cursor);
        assert_eq!(vals.len(), items.len(), "must have duplicate records");
    }
}

1	use std::{
2	fs,
3	ops::Deref,
4	path::{Path, PathBuf},
5	};
6
7	use crate::errors::CacheError;
8	use dozer_storage::{
9	lmdb::EnvironmentFlags,
10	lmdb_storage::{LmdbEnvironmentManager, LmdbEnvironmentOptions},
11	};
12	use tempdir::TempDir;
13
14	use super::cache::{CacheCommonOptions, CacheWriteOptions};
15
16	#[allow(clippy::type_complexity)]	×
17	pub fn init_env(	×
18	common_options: &CacheCommonOptions,
19	write_options: Option<CacheWriteOptions>,	×
20	) -> Result<(LmdbEnvironmentManager, (PathBuf, String), Option<TempDir>), CacheError> {	×
21	if let Some(write_options) = write_options {	2,588✔
22	create_env(common_options, write_options)	878✔
23	} else {
24	let (env, (base_path, name), temp_dir) = open_env(common_options)?;	1,710✔
25	Ok((env, (base_path.to_path_buf(), name.to_string()), temp_dir))	1,710✔
26	}	×
27	}	2,588✔
28		×
29	#[allow(clippy::type_complexity)]
30	fn create_env(	879✔
31	common_options: &CacheCommonOptions,	879✔
32	write_options: CacheWriteOptions,	879✔
33	) -> Result<(LmdbEnvironmentManager, (PathBuf, String), Option<TempDir>), CacheError> {	879✔
34	let (base_path, name, temp_dir) = match &common_options.path {	879✔
35	None => {	×
36	let base_path =	17✔
37	TempDir::new("dozer").map_err(\|e\| CacheError::Io("tempdir".into(), e))?;	17✔
38	(	17✔
39	base_path.path().to_path_buf(),	17✔
40	"dozer-cache",	17✔
41	Some(base_path),	17✔
42	)	17✔
43	}
44	Some((base_path, name)) => {	862✔
45	fs::create_dir_all(base_path).map_err(\|e\| CacheError::Io(base_path.clone(), e))?;	862✔
46	(base_path.clone(), name.deref(), None)	862✔
47	}	×
48	};	×
49		×
50	let options = LmdbEnvironmentOptions::new(	879✔
51	common_options.max_db_size,	879✔
52	common_options.max_readers,	879✔
53	write_options.max_size,	879✔
54	EnvironmentFlags::empty(),	879✔
55	);	879✔
56		879✔
57	Ok((	879✔
58	LmdbEnvironmentManager::create(&base_path, name, options)?,	879✔
59	(base_path, name.to_string()),	879✔
60	temp_dir,	879✔
61	))	×
62	}	879✔
63		×
64	#[allow(clippy::type_complexity)]	×
65	fn open_env(	1,710✔
66	options: &CacheCommonOptions,	1,710✔
67	) -> Result<(LmdbEnvironmentManager, (&Path, &str), Option<TempDir>), CacheError> {	1,710✔
68	let (base_path, name) = options	1,710✔
69	.path	1,710✔
70	.as_ref()	1,710✔
71	.ok_or(CacheError::PathNotInitialized)?;	1,710✔
72		×
73	let env_options = LmdbEnvironmentOptions {	1,710✔
74	max_dbs: options.max_db_size,	1,710✔
75	max_readers: options.max_readers,	1,710✔
76	flags: EnvironmentFlags::READ_ONLY,	1,710✔
77	..Default::default()	1,710✔
78	};	1,710✔
79		1,710✔
80	Ok((	1,710✔
81	LmdbEnvironmentManager::create(base_path, name, env_options)?,	1,710✔
82	(base_path, name),	1,710✔
83	None,	1,710✔
84	))
85	}	1,710✔
86		×
87	#[cfg(test)]	×
88	mod tests {	×
89	use dozer_storage::{	×
90	lmdb::{Cursor, DatabaseFlags, RoCursor, Transaction, WriteFlags},	×
91	lmdb_storage::CreateDatabase,	×
92	};	×
93	use dozer_types::types::Field;
94		×
95	use super::*;	×
96		×
97	fn cursor_dump(mut cursor: RoCursor) -> Vec<(&[u8], &[u8])> {	1✔
98	cursor	1✔
99	.iter_dup()	1✔
100	.flatten()	1✔
101	.collect::<dozer_storage::lmdb::Result<Vec<_>>>()	1✔
102	.unwrap()	1✔
103	}	1✔
104		×
105	#[test]	1✔
106	fn duplicate_test_nested() {	1✔
107	let mut env = create_env(&Default::default(), Default::default())	1✔
108	.unwrap()	1✔
109	.0;	1✔
110		1✔
111	let db = env	1✔
112	.create_database(	1✔
113	Some("test"),	1✔
114	Some(DatabaseFlags::DUP_SORT \| DatabaseFlags::INTEGER_KEY),	1✔
115	)	1✔
116	.unwrap();	1✔
117		1✔
118	let txn = env.create_txn().unwrap();	1✔
119	let mut master_txn = txn.write();	1✔
120	let txn = master_txn.txn_mut();	1✔
121		1✔
122	let mut c_txn = txn.begin_nested_txn().unwrap();	1✔
123		1✔
124	let items: Vec<(i64, &[u8])> = vec![	1✔
125	(1, b"a"),	1✔
126	(2, b"a"),	1✔
127	(3, b"a"),	1✔
128	(1, b"b"),	1✔
129	(2, b"b"),	1✔
130	(3, b"b"),	1✔
131	(1, b"c"),	1✔
132	(2, b"c"),	1✔
133	(3, b"c"),	1✔
134	(1, b"e"),	1✔
135	(2, b"e"),	1✔
136	(3, b"e"),	1✔
137	];	1✔
138	for (key, data) in &items {	13✔
139	let key = [	12✔
140	"idx".as_bytes().to_vec(),	12✔
141	Field::Int(*key).encode(),	12✔
142	key.to_be_bytes().to_vec(),	12✔
143	]	12✔
144	.join("#".as_bytes());	12✔
145	c_txn.put(db, &key, data, WriteFlags::empty()).unwrap();	12✔
146	}	12✔
147	c_txn.commit().unwrap();	1✔
148	master_txn.commit_and_renew().unwrap();	1✔
149		1✔
150	let rtxn = master_txn.txn();	1✔
151		1✔
152	let cursor = rtxn.open_ro_cursor(db).unwrap();	1✔
153	let vals = cursor_dump(cursor);	1✔
154	assert_eq!(vals.len(), items.len(), "must have duplicate records");	1✔
155	}	1✔
156	}

getdozer / dozer / 4392484403

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