[package]
edition = "2024"
name = "dataprof"
version = "0.8.0"
authors = ["Andrea Bozzo"]
build = false
exclude = [
"assets/animations/*",
"assets/images/*.png",
"examples/*",
".venv/*",
".mypy_cache/*",
"python/*",
".github/*",
]
autolib = false
autobins = false
autoexamples = false
autotests = false
autobenches = false
description = "High-performance data profiler with ISO 8000/25012 quality metrics for CSV, JSON/JSONL, and Parquet files"
homepage = "https://github.com/AndreaBozzo/dataprof"
documentation = "https://docs.rs/dataprof"
readme = "README.md"
keywords = [
"data",
"analysis",
"profiling",
"data-quality",
"parquet",
]
categories = ["data-structures"]
license = "MIT OR Apache-2.0"
repository = "https://github.com/AndreaBozzo/dataprof"
[package.metadata.docs.rs]
all-features = true
rustdoc-args = [
"--cfg",
"docsrs",
]
[features]
all-db = [
"postgres",
"mysql",
"sqlite",
]
arrow = [
"dataprof-core/arrow",
"dep:dataprof-parquet",
"dataprof-parquet/arrow",
"dataprof-engines/arrow",
]
async-streaming = [
"dep:tokio",
"dep:bytes",
"dep:reqwest",
"dataprof-core/async-streaming",
"dataprof-runtime/async-streaming",
"dataprof-engines/async-streaming",
"dataprof-partial/async-streaming",
]
database = [
"dep:dataprof-db",
"dataprof-core/database",
]
default = ["parquet"]
mysql = [
"database",
"dep:tokio",
"dataprof-db/mysql",
]
parquet = [
"arrow",
"dataprof-parquet/parquet",
"dataprof-engines/parquet",
"dataprof-partial/parquet",
]
parquet-async = [
"parquet",
"dep:reqwest",
"async-streaming",
"dataprof-parquet/parquet-async",
"dataprof-runtime/parquet-async",
"dataprof-engines/parquet-async",
]
postgres = [
"database",
"dep:tokio",
"dataprof-db/postgres",
]
sqlite = [
"database",
"dep:tokio",
"dataprof-db/sqlite",
]
[lib]
name = "dataprof"
crate-type = ["rlib"]
path = "crates/dataprof/src/lib.rs"
[[test]]
name = "async_streaming"
path = "tests/async_streaming.rs"
[[test]]
name = "boolean_columns"
path = "tests/boolean_columns.rs"
[[test]]
name = "cross_engine_consistency"
path = "tests/cross_engine_consistency.rs"
[[test]]
name = "database_integration"
path = "tests/database_integration.rs"
[[test]]
name = "public_api_facade"
path = "tests/public_api_facade.rs"
[[test]]
name = "semantic_hints"
path = "tests/semantic_hints.rs"
[[bench]]
name = "benchmarks"
path = "benches/benchmarks.rs"
harness = false
[dependencies.bytes]
version = "1.10"
optional = true
[dependencies.dataprof-core]
version = "0.8.0"
[dependencies.dataprof-csv]
version = "0.8.0"
[dependencies.dataprof-db]
version = "0.8.0"
optional = true
default-features = false
[dependencies.dataprof-engines]
version = "0.8.0"
default-features = false
[dependencies.dataprof-json]
version = "0.8.0"
[dependencies.dataprof-metrics]
version = "0.8.0"
[dependencies.dataprof-parquet]
version = "0.8.0"
optional = true
default-features = false
[dependencies.dataprof-partial]
version = "0.8.0"
default-features = false
[dependencies.dataprof-runtime]
version = "0.8.0"
[dependencies.reqwest]
version = "0.12"
features = [
"rustls-tls",
"stream",
]
optional = true
default-features = false
[dependencies.tokio]
version = "1.49"
features = ["full"]
optional = true
[dev-dependencies.anyhow]
version = "1.0"
[dev-dependencies.arrow]
version = "57.3.0"
features = ["ffi"]
[dev-dependencies.criterion]
version = "0.8"
features = ["html_reports"]
[dev-dependencies.env_logger]
version = "0.11"
[dev-dependencies.parquet]
version = "57.3.0"
features = ["arrow"]
[dev-dependencies.predicates]
version = "3.1"
[dev-dependencies.proptest]
version = "1.9"
[dev-dependencies.serde_json]
version = "1.0"
[dev-dependencies.serial_test]
version = "3.4.0"
[dev-dependencies.sqlx]
version = "0.8.1"
features = [
"runtime-tokio-rustls",
"chrono",
"uuid",
"sqlite",
]
default-features = false
[dev-dependencies.tempfile]
version = "3.24"
[lints.clippy]
all = "deny"
[lints.rust]
unsafe_code = "warn"
[profile.ci]
opt-level = 0
codegen-units = 1
debug = 2
incremental = false
inherits = "test"
[profile.dev]
opt-level = 0
codegen-units = 256
debug = 2
split-debuginfo = "packed"
incremental = true
[profile.dev.package."*"]
opt-level = 1
[profile.release]
lto = true
codegen-units = 1
panic = "abort"
strip = true
[profile.release-lto]
inherits = "release"
[profile.test]
opt-level = 1
codegen-units = 256
debug = 2
incremental = true
[profile.test.package."*"]
opt-level = 1