triplexanalytics 0.2.2__tar.gz → 0.4.0__tar.gz

{triplexanalytics-0.2.2 → triplexanalytics-0.4.0}/Cargo.lock

@@ -1389,6 +1389,7 @@ version = "0.0.0"
 dependencies = [
  "serde_json",
  "triplex-anonymizer",
+ "triplex-datasetcard",
  "triplex-driftguard",
  "triplex-modelcard",
  "triplex-schemacast",
@@ -1425,6 +1426,8 @@ dependencies = [
  "serde",
  "serde_json",
  "triplex-core",
+ "triplex-io",
+ "triplex-schemacast",
 ]
 
 [[package]]

{triplexanalytics-0.2.2 → triplexanalytics-0.4.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: triplexanalytics
-Version: 0.2.2
+Version: 0.4.0
 Classifier: Development Status :: 3 - Alpha
 Classifier: Intended Audience :: Developers
 Classifier: Intended Audience :: Science/Research
@@ -30,7 +30,14 @@ Triplex Analytics in plain Python: infer and check schema contracts, anonymize a
 synthetic data. No notebook, no browser, no server. The engine is the same native code the
 `triplex` command line runs.
 
-Not published yet. Install from a source checkout (needs the Rust toolchain):
+```sh
+pip install triplexanalytics
+```
+
+The distribution is called `triplexanalytics`; the import is `import triplex`. Wheels cover Linux
+(x86_64 and aarch64), macOS and Windows, for Python 3.9 and later.
+
+To build from a source checkout instead (needs the Rust toolchain):
 
 ```sh
 cd packages/triplex-python

{triplexanalytics-0.2.2/packages/triplex-python → triplexanalytics-0.4.0}/README.md

@@ -4,7 +4,14 @@ Triplex Analytics in plain Python: infer and check schema contracts, anonymize a
 synthetic data. No notebook, no browser, no server. The engine is the same native code the
 `triplex` command line runs.
 
-Not published yet. Install from a source checkout (needs the Rust toolchain):
+```sh
+pip install triplexanalytics
+```
+
+The distribution is called `triplexanalytics`; the import is `import triplex`. Wheels cover Linux
+(x86_64 and aarch64), macOS and Windows, for Python 3.9 and later.
+
+To build from a source checkout instead (needs the Rust toolchain):
 
 ```sh
 cd packages/triplex-python

{triplexanalytics-0.2.2 → triplexanalytics-0.4.0}/crates/triplex-anonymizer/src/bucket.rs

@@ -1,11 +1,14 @@
 //! Numeric generalization/bucketing — the Tier 1 k-anonymity
-//! mechanism. The named risk models a
+//! mechanism. The bucketing itself (`bucket_value`, `bucket_numeric`) lives in
+//! `triplex-datasetcard`, so masking and risk scoring generalize a number the
+//! same way; this module wraps it as a masking mechanism. The named risk models a
 //! bucketed column can be checked against (`RiskModel`/`evaluate_risk_model`
 //! and friends) live in `triplex-datasetcard`,
 //! since scoring risk needs to run independently of any recipe, before a
 //! masking mechanism is even chosen.
 
 use crate::mechanism::Mechanism;
+use triplex_datasetcard::bucket_value;
 
 pub(crate) struct Bucketer {
     width: u64,
@@ -25,37 +28,6 @@ impl Mechanism for Bucketer {
     }
 }
 
-/// Replaces a numeric value with the `"lo-hi"` range it falls in, e.g.
-/// width 10 turns `"23"` into `"20-29"`. A value that isn't a whole number
-/// has nothing to generalize and passes through unchanged.
-fn bucket_number(width: u64, n: i64) -> String {
-    let width = width.max(1) as i64;
-    let lo = n.div_euclid(width) * width;
-    let hi = lo + width - 1;
-    format!("{lo}-{hi}")
-}
-
-pub fn bucket_value(width: u64, value: &str) -> String {
-    let Ok(n) = value.parse::<i64>() else {
-        return value.to_string();
-    };
-    bucket_number(width, n)
-}
-
-/// Same bucketing as `bucket_value`, for a caller that already has a typed
-/// numeric value instead of a string to parse (e.g. `apps/triplex-web`'s
-/// risk-scoring wasm exports, which take a typed-array of `f64` straight
-/// from the browser instead of a JSON string of stringified cells).
-/// `NaN` represents a
-/// missing/non-numeric cell, passed through as an empty string, the same
-/// convention `bucket_value` uses for a value that fails to parse.
-pub fn bucket_numeric(width: u64, value: f64) -> String {
-    if value.is_nan() {
-        return String::new();
-    }
-    bucket_number(width, value as i64)
-}
-
 /// Generalizes a string value by keeping only its first `width` characters
 /// (Unicode scalar values, not bytes, so a multi-byte character is never
 /// split mid-codepoint) — the "automatic" string generalization ARX ships
@@ -82,36 +54,6 @@ pub fn generalize_string_prefix(width: usize, value: &str) -> String {
 mod tests {
     use super::*;
 
-    #[test]
-    fn buckets_a_number_into_its_width_sized_range() {
-        assert_eq!(bucket_value(10, "23"), "20-29");
-        assert_eq!(bucket_value(10, "20"), "20-29");
-        assert_eq!(bucket_value(10, "29"), "20-29");
-        assert_eq!(bucket_value(10, "30"), "30-39");
-    }
-
-    #[test]
-    fn negative_numbers_bucket_toward_negative_infinity_not_toward_zero() {
-        // div_euclid, not integer truncation: -1 must land in "-10--1", not "0-9".
-        assert_eq!(bucket_value(10, "-1"), "-10--1");
-    }
-
-    #[test]
-    fn a_non_numeric_value_passes_through_unchanged() {
-        assert_eq!(bucket_value(10, "n/a"), "n/a");
-    }
-
-    #[test]
-    fn bucket_numeric_matches_bucket_value_for_the_same_number() {
-        assert_eq!(bucket_numeric(10, 23.0), bucket_value(10, "23"));
-        assert_eq!(bucket_numeric(10, -1.0), bucket_value(10, "-1"));
-    }
-
-    #[test]
-    fn bucket_numeric_treats_nan_as_the_missing_value_case() {
-        assert_eq!(bucket_numeric(10, f64::NAN), "");
-    }
-
     #[test]
     fn generalize_string_prefix_keeps_only_the_leading_characters() {
         assert_eq!(generalize_string_prefix(4, "3512AD"), "3512");

{triplexanalytics-0.2.2 → triplexanalytics-0.4.0}/crates/triplex-anonymizer/src/lib.rs

@@ -14,14 +14,16 @@ mod recipe;
 pub mod synthesize;
 mod tokenize;
 
-pub use bucket::{bucket_numeric, bucket_value, generalize_string_prefix};
+pub use bucket::generalize_string_prefix;
 pub use fpe::fpe_mask;
 pub use mapping::{decrypt_mapping, encrypt_mapping, EncryptedMapping, MappingEntry};
 pub use partial_mask::partial_mask;
 pub use recipe::{
-    load_recipe, verify_recipe_well_formed, AnonymizeColumn, AnonymizeMethod, AnonymizeRecipe,
+    load_recipe, recipe_from_yaml, verify_recipe_well_formed, AnonymizeColumn, AnonymizeMethod,
+    AnonymizeRecipe, RECIPE_VERSION,
 };
 pub use tokenize::hmac_token;
+pub use triplex_datasetcard::{bucket_numeric, bucket_value};
 // Risk-model scoring moved to `triplex-datasetcard` — kept
 // re-exported here so nothing that already does
 // `use triplex_anonymizer::{RiskModel, ...}` breaks.
@@ -169,7 +171,12 @@ fn resolve_columns<'a>(
             drop_indices.insert(header_index);
             continue;
         }
-        let salt = salt_for(declared)?;
+        // Only a keyed method needs a salt; the others give the same output for anyone.
+        let salt = if declared.method.is_keyed() {
+            salt_for(declared)?
+        } else {
+            Vec::new()
+        };
         resolved.push(ResolvedColumn {
             header_index,
             declared,
@@ -183,6 +190,12 @@ fn resolve_columns<'a>(
 /// itself is never a CLI argument, so it never lands in shell history or a
 /// process listing.
 fn salt_from_env(declared: &AnonymizeColumn) -> Result<Vec<u8>> {
+    if declared.salt_env.is_empty() {
+        return Err(Error::Validation(format!(
+            "column {:?} uses {:?}, which needs salt_env: the environment variable holding its salt",
+            declared.name, declared.method
+        )));
+    }
     std::env::var(&declared.salt_env)
         .map(String::into_bytes)
         .map_err(|_| Error::MissingEnvVar(declared.salt_env.clone()))