Crate fp_expr 

Expression engine for frankenpandas — implements pandas DataFrame.eval() / DataFrame.query() semantics for Rust callers and provides the underlying Expr AST + parser / evaluator that the rest of the workspace can build on.

Why

Pandas users write things like:

df.query("price > 100 and category == 'A'")
df.eval("margin = revenue - cost")

This crate gives Rust callers the equivalent: df.query(...), df.eval(...), with a parsed Expr AST you can build programmatically or via the string parser.

Public surface

• Expr: the expression AST. Built either by hand (Expr::Series, Expr::Add, Expr::Sub, Expr::Local, etc.) or by parsing a pandas-style string with parse_expr.
• SeriesRef: a typed column-name reference. Used by Expr variants and by callers building expressions programmatically.
• EvalContext: the evaluation environment — bindings for local variables (the @local_var syntax in pandas eval) plus the RuntimePolicy / EvidenceLedger from fp-runtime for decision recording.
• ExprError: failure modes (parse error, unknown column, type mismatch, division by zero, …).

Evaluation entry points

Direct AST eval:

• evaluate: evaluate an Expr against a Series-like input.
• evaluate_on_dataframe: bind columns from a DataFrame as identifiers and evaluate.
• evaluate_on_dataframe_with_locals: same, plus an external @local binding map.
• filter_dataframe_on_expr / filter_dataframe_on_expr_with_locals: shortcut for df[df.eval(expr)] — the boolean-mask filter pattern.

String entry points (parse-then-eval):

• eval_str / eval_str_with_locals: pandas df.eval(string) — returns a new Series / DataFrame column.
• query_str / query_str_with_locals: pandas df.query(string) — returns the row-filtered DataFrame.
• parse_expr: the standalone parser if you only want the AST.

DataFrame extension trait

DataFrameExprExt adds df.eval(expr) / df.query(expr) method-style entry points on DataFrame so users can call them fluently after use fp_expr::DataFrameExprExt;.

Incremental views

MaterializedView + Delta: the foundation for incremental eval-derived columns. A MaterializedView caches the last result and the input fingerprint; on next call, only re-evaluates when inputs change. Delta records what changed for downstream consumers.

Cross-crate relationships

• fp-types (Scalar), fp-columnar (ComparisonOp), fp-index (Index), fp-frame (Series, DataFrame, FrameError) are all consumed.
• fp-runtime (RuntimePolicy, EvidenceLedger) provides the optional decision-policy hook threaded through EvalContext.

Structs

Delta

A delta represents new rows appended to a base series.

EvalContext

MaterializedView

Cached result of a previous full evaluation, used as base for incremental updates.

SeriesRef

Enums

BetweenInclusive

Expr

ExprDuplicateKeep

ExprError

Traits

DataFrameExprExt

Extension trait that adds eval() and query() methods directly to DataFrame.

Functions

eval_str

Evaluate a string expression against a DataFrame and return a Series.

eval_str_with_locals

evaluate

evaluate_on_dataframe

evaluate_on_dataframe_with_locals

filter_dataframe_on_expr

filter_dataframe_on_expr_with_locals

parse_expr

Parse a string expression into an Expr AST.

query_str

Filter a DataFrame using a string expression.

query_str_with_locals