Crate pinenut_log

An extremely high performance logging system for clients (iOS, Android, Desktop), written in Rust.

Compression

Pinenut supports streaming log compression, it uses the Zstandard (aka zstd), a high performance compression algorithm that has a good balance between compression rate and speed.

Encryption

Pinenut uses the AES 128 algorithm for symmetric encryption during logging. To prevent embedding the symmetric key directly into the code, Pinenut uses ECDH for key negotiation (RSA is not used because its key are too long). When initializing the Logger, there is no need to provide the symmetric encryption key, instead the ECDH public key should be passed.

Pinenut uses secp256r1 elliptic curve for ECDH. You can generate the secret and public keys for encryption yourself, or use Pinenut’s built-in command line tool: pinenut-cli.

Buffering

In order to minimize IO frequency, Pinenut buffers the log data before writing to the file. Client programs may exit unexpectedly (e.g., crash), Pinenut uses mmap as buffer support, so that if the program unexpectedly exits, the OS can still help to persist the buffered data. The next time the Logger is initialized, the buffered data is automatically read and written back to the log file.

In addition, Pinenut implements a double-buffering system to improve buffer read/write performance and prevent asynchronous IOs from affecting logging of the current thread.

Extraction

With Pinenut, we don’t need to retrieve all the log files in the directory to extract logs, it provides convenient extraction capabilities and supports extraction in time ranges with minute granularity.

Parsing

The content of Pinenut log files is a special binary sequence after encoding, compression and encryption, and we can parse the log files using the parsing capabilities provided by Pinenut.

Usage

Pinenut’s APIs are generally similar regardless of the language used.

Logger Initialization

Pinenut uses a Logger instance for logging. Before we initialize the Logger, we need to pass in the logger identifier and the path to the directory where the log files are stored to construct the Domain structure.

We can customize the Logger by explicitly specifying Config, see the API documentation for details.

let domain = Domain::new("MyApp".into(), "/path/to/dir".into());
let config = Config::new().key_str(Some("Public Key Base64")).compression_level(10);
let logger = Logger::new(domain, config);

Logging

Just construct the Record and call the log method.

Records can be constructed in Rust via the Builder pattern:

// Builds `Meta` & `Record`.
let meta = Meta::builder().level(Level::Info).build();
let record = Record::builder().meta(meta).content("Hello World").build();
logger.log(&record);

// Flushes any buffered records asynchronously.
logger.flush();

See the API documentation for details.

Extraction

Just call the extract method to extract the logs for the specified time range (with minute granularity) and write them to the destination file.

let domain = Domain::new("MyApp".into(), "/path/to/dir".into());
let now = chrono::Utc::now();
let range = now.sub(Duration::from_secs(1800))..=now;

if let Err(err) = pinenut_log::extract(domain, range, "/path/to/destination") {
    println!("Error: {err}");
}

Note: The content of the extracted file is still a binary sequence that has been encoded, compressed, and encrypted. We need to parse it to see the log text content that is easy to read.

Parsing

You can use the parse function for log parsing, and you can specify the format of the log parsed text. See the API documentation for details.

// Specifies the `DefaultFormater` as the log formatter.
if let Err(err) = pinenut_log::parse_to_file(&path, &output, secret_key, DefaultFormatter) {
    println!("Error: {err}");
}

Or use the built-in command line tool pinenut-cli:

$ pinenut-cli parse ./my_log.pine \
    --output ./plain.log          \
    --secret-key XXXXXXXXXXX

Keys Generation

Before initializing the Logger or parsing the logs, you need to have the public and secret keys ready (The public key is used to initialize the Logger and the secret key is used to parse the logs).

You can use pinenut-cli to generate this pair of keys:

$ pinenut-cli gen-keys

Re-exports

pub use compress::CompressionError;

pub use compress::DecompressionError;

pub use encrypt::DecryptionError;

pub use encrypt::EncryptionError;

pub use encrypt::EncryptionKey;

pub use encrypt::PublicKey;

pub use encrypt::SecretKey;

pub use encrypt::PUBLIC_KEY_LEN;

pub use codec::DecodingError;

pub use codec::EncodingError;

pub use chunk::Error as ChunkError;

pub use runloop::Error as RunloopError;

pub use record::*;

Modules

chunk

The Chunk structure.

codec

Encoding & Decoding.

compress

Compression & Decompression.

encrypt

Encryption & Decryption.

record

The Pinenut log record.

runloop

An event loop implementation.

Structs

Config

Configuration of a logger instance.

DefaultFormatter

The default formatter provides simple log formatting.

Domain

Represents the domain to which the logger belongs, the logs will be organized by domain.

Logger

The Pinenut logger.

Enums

ExtractionError

Errors that can be occurred during the log extraction process (extract).

LoggerError

The error type for Logger.

ParsingError

Errors that can be occurred during the log parsing process (parse).

TimeDimension

Represents the dimension of datetime, used for log rotation.

Constants

BUFFER_LEN

The default buffer length (320 KB) for Pinenut.

FILE_EXTENSION

The extension of the Pinenut log file.

FORMAT_VERSION

The current format version of the Pinenut log structure.

MMAP_BUFFER_EXTENSION

The extension of the Pinenut mmap buffer file.

PLAIN_FILE_EXTENSION

The extension of the Pinenut plain log file.

Traits

Format

Represents a formatter that formats log records into readable text.

Track

Represents a tracker used to track errors occurred from the logger operations.

Functions

extract

Extracts the logs for the specified time range and writes them to the destination file.

parse

Parses the compressed and encrypted binary log file into multiple log records and calls them back one by one.

parse_to_file

Parses the compressed and encrypted binary log file into readable text file.

Type Aliases

Tracker

Trait object type for Track.