[package]
edition = "2024"
rust-version = "1.96"
name = "dataprof"
version = "0.13.1"
authors = ["Andrea Bozzo"]
build = false
exclude = [
"fuzz/*",
"assets/animations/*",
"assets/images/*.png",
"examples/test_data/*",
"examples/test_datasets/*",
".venv/*",
".mypy_cache/*",
"python/*",
".github/*",
"site/*",
]
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://andreabozzo.github.io/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-core/parquet",
"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"
[[example]]
name = "before_after_cleaning"
path = "examples/before_after_cleaning.rs"
[[example]]
name = "etl_quality_gate"
path = "examples/etl_quality_gate.rs"
[[example]]
name = "generate_profile_schema"
path = "examples/generate_profile_schema.rs"
[[example]]
name = "messy_csv_inspection"
path = "examples/messy_csv_inspection.rs"
[[test]]
name = "accuracy_range_names"
path = "tests/accuracy_range_names.rs"
[[test]]
name = "accuracy_rule_units"
path = "tests/accuracy_rule_units.rs"
[[test]]
name = "all_null_columns"
path = "tests/all_null_columns.rs"
[[test]]
name = "analysis_option_parity"
path = "tests/analysis_option_parity.rs"
[[test]]
name = "arrow_map_duplicate_rows"
path = "tests/arrow_map_duplicate_rows.rs"
[[test]]
name = "async_header_gap"
path = "tests/async_header_gap.rs"
[[test]]
name = "async_json_late_keys"
path = "tests/async_json_late_keys.rs"
[[test]]
name = "async_streaming"
path = "tests/async_streaming.rs"
[[test]]
name = "baseline_gate"
path = "tests/baseline_gate.rs"
[[test]]
name = "benchmark_fixtures"
path = "tests/benchmark_fixtures.rs"
[[test]]
name = "binary_input"
path = "tests/binary_input.rs"
[[test]]
name = "boolean_columns"
path = "tests/boolean_columns.rs"
[[test]]
name = "column_order"
path = "tests/column_order.rs"
[[test]]
name = "column_selection_record"
path = "tests/column_selection_record.rs"
[[test]]
name = "complete_records_ratio"
path = "tests/complete_records_ratio.rs"
[[test]]
name = "cross_engine_consistency"
path = "tests/cross_engine_consistency.rs"
[[test]]
name = "csv_bom_chunks"
path = "tests/csv_bom_chunks.rs"
[[test]]
name = "csv_delimiter_header"
path = "tests/csv_delimiter_header.rs"
[[test]]
name = "csv_flexible_delimiter"
path = "tests/csv_flexible_delimiter.rs"
[[test]]
name = "csv_without_records"
path = "tests/csv_without_records.rs"
[[test]]
name = "database_integration"
path = "tests/database_integration.rs"
[[test]]
name = "database_option_parity"
path = "tests/database_option_parity.rs"
[[test]]
name = "date_column_names"
path = "tests/date_column_names.rs"
[[test]]
name = "date_share_every_value"
path = "tests/date_share_every_value.rs"
[[test]]
name = "decimal_columns"
path = "tests/decimal_columns.rs"
[[test]]
name = "digit_codes"
path = "tests/digit_codes.rs"
[[test]]
name = "encoding_sample_boundary"
path = "tests/encoding_sample_boundary.rs"
[[test]]
name = "error_contract"
path = "tests/error_contract.rs"
[[test]]
name = "exact_set_budget"
path = "tests/exact_set_budget.rs"
[[test]]
name = "execution_controls"
path = "tests/execution_controls.rs"
[[test]]
name = "findings_parity"
path = "tests/findings_parity.rs"
[[test]]
name = "json_array_fault_drain"
path = "tests/json_array_fault_drain.rs"
[[test]]
name = "json_fieldless_rows"
path = "tests/json_fieldless_rows.rs"
[[test]]
name = "json_wide_integers"
path = "tests/json_wide_integers.rs"
[[test]]
name = "key_column_source"
path = "tests/key_column_source.rs"
[[test]]
name = "key_uniqueness_units"
path = "tests/key_uniqueness_units.rs"
[[test]]
name = "large_spread_variance"
path = "tests/large_spread_variance.rs"
[[test]]
name = "locale_numbers"
path = "tests/locale_numbers.rs"
[[test]]
name = "locale_tags"
path = "tests/locale_tags.rs"
[[test]]
name = "memory_row_limits"
path = "tests/memory_row_limits.rs"
[[test]]
name = "mixed_date_formats"
path = "tests/mixed_date_formats.rs"
[[test]]
name = "nested_columns"
path = "tests/nested_columns.rs"
[[test]]
name = "nested_duplicate_rows"
path = "tests/nested_duplicate_rows.rs"
[[test]]
name = "null_markers"
path = "tests/null_markers.rs"
[[test]]
name = "numeric_stability"
path = "tests/numeric_stability.rs"
[[test]]
name = "oracle"
path = "tests/oracle.rs"
[[test]]
name = "outlier_component_units"
path = "tests/outlier_component_units.rs"
[[test]]
name = "parquet_encoding_parity"
path = "tests/parquet_encoding_parity.rs"
[[test]]
name = "partial_io_causes"
path = "tests/partial_io_causes.rs"
[[test]]
name = "partial_schema_parity"
path = "tests/partial_schema_parity.rs"
[[test]]
name = "pattern_detection"
path = "tests/pattern_detection.rs"
[[test]]
name = "pattern_locale_record"
path = "tests/pattern_locale_record.rs"
[[test]]
name = "profile_report_schema"
path = "tests/profile_report_schema.rs"
[[test]]
name = "progress_routes"
path = "tests/progress_routes.rs"
[[test]]
name = "public_api_facade"
path = "tests/public_api_facade.rs"
[[test]]
name = "quality_gate_parity"
path = "tests/quality_gate_parity.rs"
[[test]]
name = "quick_quality_check"
path = "tests/quick_quality_check.rs"
[[test]]
name = "ragged_rows"
path = "tests/ragged_rows.rs"
[[test]]
name = "raw_date_forms"
path = "tests/raw_date_forms.rs"
[[test]]
name = "remote_parquet_versions"
path = "tests/remote_parquet_versions.rs"
[[test]]
name = "report_delta"
path = "tests/report_delta.rs"
[[test]]
name = "report_generator"
path = "tests/report_generator.rs"
[[test]]
name = "report_rounding_parity"
path = "tests/report_rounding_parity.rs"
[[test]]
name = "rfc3339_timestamps"
path = "tests/rfc3339_timestamps.rs"
[[test]]
name = "rounding_parity"
path = "tests/rounding_parity.rs"
[[test]]
name = "row_estimation"
path = "tests/row_estimation.rs"
[[test]]
name = "sampling_strategies"
path = "tests/sampling_strategies.rs"
[[test]]
name = "schema_rules"
path = "tests/schema_rules.rs"
[[test]]
name = "score_weights_recorded"
path = "tests/score_weights_recorded.rs"
[[test]]
name = "semantic_hint_duplicates"
path = "tests/semantic_hint_duplicates.rs"
[[test]]
name = "semantic_hints"
path = "tests/semantic_hints.rs"
[[test]]
name = "slash_date_order"
path = "tests/slash_date_order.rs"
[[test]]
name = "small_spread_variance"
path = "tests/small_spread_variance.rs"
[[test]]
name = "statistics_estimators"
path = "tests/statistics_estimators.rs"
[[test]]
name = "stop_condition_composition"
path = "tests/stop_condition_composition.rs"
[[test]]
name = "streaming_allocations"
path = "tests/streaming_allocations.rs"
[[test]]
name = "temporal_pairs"
path = "tests/temporal_pairs.rs"
[[test]]
name = "timeliness_age_cutoff"
path = "tests/timeliness_age_cutoff.rs"
[[test]]
name = "timeliness_declared_pairs"
path = "tests/timeliness_declared_pairs.rs"
[[test]]
name = "timeliness_role_words"
path = "tests/timeliness_role_words.rs"
[[test]]
name = "uniqueness_bounds_past_a_million"
path = "tests/uniqueness_bounds_past_a_million.rs"
[[test]]
name = "unknown_component_coverage"
path = "tests/unknown_component_coverage.rs"
[[test]]
name = "unlisted_component_coverage"
path = "tests/unlisted_component_coverage.rs"
[[test]]
name = "unterminated_quote"
path = "tests/unterminated_quote.rs"
[[test]]
name = "us_phone_pattern"
path = "tests/us_phone_pattern.rs"
[[test]]
name = "wide_integer_stats"
path = "tests/wide_integer_stats.rs"
[[bench]]
name = "benchmarks"
path = "benches/benchmarks.rs"
harness = false
[dependencies.bytes]
version = "1.10"
optional = true
[dependencies.dataprof-core]
version = "0.13.1"
[dependencies.dataprof-csv]
version = "0.13.1"
[dependencies.dataprof-db]
version = "0.13.1"
optional = true
[dependencies.dataprof-engines]
version = "0.13.1"
[dependencies.dataprof-json]
version = "0.13.1"
[dependencies.dataprof-metrics]
version = "0.13.1"
[dependencies.dataprof-parquet]
version = "0.13.1"
optional = true
[dependencies.dataprof-partial]
version = "0.13.1"
[dependencies.dataprof-runtime]
version = "0.13.1"
[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 = "60.0.0"
features = [
"ffi",
"chrono-tz",
]
[dev-dependencies.chrono]
version = "0.4"
features = ["serde"]
[dev-dependencies.criterion]
version = "0.8"
features = ["html_reports"]
[dev-dependencies.csv]
version = "1.4"
[dev-dependencies.env_logger]
version = "0.11"
[dev-dependencies.jsonschema]
version = "0.60.0"
default-features = false
[dev-dependencies.parquet]
version = "60.0.0"
features = ["arrow"]
[dev-dependencies.predicates]
version = "3.1"
[dev-dependencies.proptest]
version = "1.9"
[dev-dependencies.serde]
version = "1.0"
features = ["derive"]
[dev-dependencies.serde_json]
version = "1.0.149"
features = [
"preserve_order",
"float_roundtrip",
]
[dev-dependencies.serial_test]
version = "4.0.1"
[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
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