primitives transform strategy

Transform strategy

Active contributors: Douwe de Vries

Purpose

Transform strategy is the selected method for changing or preserving values in selected columns. It combines user intent, detected data type, and in-run state so repeated source values stay consistent while output remains local-first.

Directory layout

Path	Role
crates/csv-anonymizer-core/src/types.rs	Defines AnonymizationStrategy, ColumnControl, TransformContext, TransformReport, and process result types.
crates/csv-anonymizer-core/src/strategies/mod.rs	Dispatches strategies and detected data types to concrete transformation functions.
crates/csv-anonymizer-core/src/strategies/state.rs	Tracks pseudonym maps, opaque token maps, Smart replacements, reuse counts, collisions, and fallbacks.
crates/csv-anonymizer-core/src/strategies/structured.rs	Handles email, UUID, timestamp, phone, generic string, and opaque token transformations.
crates/csv-anonymizer-core/src/strategies/numeric.rs	Handles numeric ID and numeric value transformations while preserving width and decimal shape.
crates/csv-anonymizer-core/src/strategies/names.rs	Handles first name, last name, and full name replacement from local name pools.
crates/csv-anonymizer-core/src/strategies/redaction.rs	Selects typed redaction placeholders from data type or privacy evidence.

Key abstractions

• AnonymizationStrategy supports auto, pseudonymize, tokenize, localAi, mask, redact, and passThrough.
• ColumnControl carries frontend overrides for type and strategy.
• TransformContext identifies column name, column index, row index, and empty-value format during transformation.
• TransformState preserves per-run mapping consistency and accumulates TransformReport counters.
• TransformReport feeds Privacy report counters for reuse, tokens, Smart replacement values, rejections, and fallbacks.
• STRUCTURED_SCALAR_REDACTION_WARNING explains cases where redaction placeholders can change JSON or YAML scalar types.

How it works

graph TD
    Row[Input row] --> Selected{Column selected?}
    Selected -->|no| Original[Original value]
    Selected -->|yes| Strategy{Strategy}
    Strategy -->|passThrough| Original
    Strategy -->|mask| Mask[Mask characters]
    Strategy -->|redact| Placeholder[Typed placeholder]
    Strategy -->|tokenize| Token[Opaque token]
    Strategy -->|localAi| Smart{Map hit?}
    Smart -->|yes| Replacement[Smart replacement]
    Smart -->|no| Fallback[Rule-based fallback]
    Strategy -->|auto or pseudonymize| ByType[Detected-type transformer]
    Fallback --> ByType

Loading

transform_row_with_state walks each cell and only transforms selected columns. Empty values pass through. Explicit strategies run first. localAi uses TransformState::smart_replacement; a missing replacement records a fallback and then continues through rule-based pseudonymization. auto and pseudonymize dispatch by detected data type, while Boolean, currency, percentage, country code, and enum currently pass through.

Integration points

• Column metadata supplies selected state, detected type, empty format, and strategy.
• Smart replacement map provides accepted replacements for the localAi strategy.
• Privacy report consumes TransformReport and selected metadata.
• csv-anonymizer-core uses strategies in CSV streaming, preview generation, direct input, and quick generation.
• Frontend workflow state lets users change strategy through ColumnControl.

Entry points for modification

• Add a strategy enum value in crates/csv-anonymizer-core/src/types.rs and mirror it in frontend/src/types.ts.
• Change strategy dispatch in crates/csv-anonymizer-core/src/strategies/mod.rs.
• Change per-run reuse, report counters, token generation, or Smart replacement fallback counts in crates/csv-anonymizer-core/src/strategies/state.rs.
• Change structured value behavior in crates/csv-anonymizer-core/src/strategies/structured.rs.
• Change numeric shape preservation in crates/csv-anonymizer-core/src/strategies/numeric.rs.
• Change local name pools or name-token handling in crates/csv-anonymizer-core/src/strategies/names.rs.
• Change redaction placeholder selection in crates/csv-anonymizer-core/src/strategies/redaction.rs.

Key source files

• crates/csv-anonymizer-core/src/types.rs
• crates/csv-anonymizer-core/src/strategies/mod.rs
• crates/csv-anonymizer-core/src/strategies/state.rs
• crates/csv-anonymizer-core/src/strategies/structured.rs
• crates/csv-anonymizer-core/src/strategies/numeric.rs
• crates/csv-anonymizer-core/src/strategies/names.rs
• crates/csv-anonymizer-core/src/strategies/redaction.rs