4324181106

Build Type

push

github

Committed by GitHub

Commit Message

chore: Remove some unnecessary `clone`s (#1131)

Run Details

81 of 81 new or added lines in 6 files covered. (100.0%)

28161 of 36907 relevant lines covered (76.3%)

27502.29 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

71.39

/dozer-sql/src/pipeline/expression/builder.rs

use dozer_types::{
    ordered_float::OrderedFloat,
    types::{Field, FieldDefinition, Schema, SourceDefinition},
};
use sqlparser::ast::{
    BinaryOperator as SqlBinaryOperator, DataType, Expr as SqlExpr, Expr, Function, FunctionArg,
    FunctionArgExpr, Ident, TrimWhereField, UnaryOperator as SqlUnaryOperator, Value as SqlValue,
};

use crate::pipeline::errors::PipelineError::{
    InvalidArgument, InvalidExpression, InvalidNestedAggregationFunction, InvalidOperator,
    InvalidValue,
};
use crate::pipeline::errors::{PipelineError, SqlError};
use crate::pipeline::expression::aggregate::AggregateFunctionType;
use crate::pipeline::expression::datetime::DateTimeFunctionType;

use crate::pipeline::expression::execution::Expression;
use crate::pipeline::expression::execution::Expression::{
    DateTimeFunction, GeoFunction, ScalarFunction,
};
use crate::pipeline::expression::geo::common::GeoFunctionType;
use crate::pipeline::expression::operator::{BinaryOperatorType, UnaryOperatorType};
use crate::pipeline::expression::scalar::common::ScalarFunctionType;
use crate::pipeline::expression::scalar::string::TrimType;

use super::cast::CastOperatorType;

#[derive(Clone, PartialEq, Debug)]
pub struct ExpressionBuilder {
    // Must be an aggregation function
    pub aggregations: Vec<Expression>,
    pub offset: usize,
}

impl ExpressionBuilder {
    pub fn new(offset: usize) -> Self {
        Self {
            aggregations: Vec::new(),
            offset,
        }
    }

    pub fn from(offset: usize, aggregations: Vec<Expression>) -> Self {
        Self {
            aggregations,
            offset,
        }
    }

    pub fn build(
        &mut self,
        parse_aggregations: bool,
        sql_expression: &SqlExpr,
        schema: &Schema,
    ) -> Result<Expression, PipelineError> {
        self.parse_sql_expression(parse_aggregations, sql_expression, schema)
    }

    pub(crate) fn parse_sql_expression(
        &mut self,
        parse_aggregations: bool,
        expression: &SqlExpr,
        schema: &Schema,
    ) -> Result<Expression, PipelineError> {
        match expression {
            SqlExpr::Trim {
                expr,
                trim_where,
                trim_what,
            } => self.parse_sql_trim_function(
                parse_aggregations,
                expr,
                trim_where,
                trim_what,
                schema,
            ),
            SqlExpr::Identifier(ident) => Self::parse_sql_column(&[ident.clone()], schema),
            SqlExpr::CompoundIdentifier(ident) => Self::parse_sql_column(ident, schema),
            SqlExpr::Value(SqlValue::Number(n, _)) => Self::parse_sql_number(n),
            SqlExpr::Value(SqlValue::Null) => Ok(Expression::Literal(Field::Null)),
            SqlExpr::Value(SqlValue::SingleQuotedString(s) | SqlValue::DoubleQuotedString(s)) => {
                Self::parse_sql_string(s)
            }
            SqlExpr::UnaryOp { expr, op } => {
                self.parse_sql_unary_op(parse_aggregations, op, expr, schema)
            }
            SqlExpr::BinaryOp { left, op, right } => {
                self.parse_sql_binary_op(parse_aggregations, left, op, right, schema)
            }
            SqlExpr::Nested(expr) => self.parse_sql_expression(parse_aggregations, expr, schema),
            SqlExpr::Function(sql_function) => {
                self.parse_sql_function(parse_aggregations, sql_function, schema)
            }
            SqlExpr::Like {
                negated,
                expr,
                pattern,
                escape_char,
            } => self.parse_sql_like_operator(
                parse_aggregations,
                negated,
                expr,
                pattern,
                escape_char,
                schema,
            ),
            SqlExpr::Cast { expr, data_type } => {
                self.parse_sql_cast_operator(parse_aggregations, expr, data_type, schema)
            }
            _ => Err(InvalidExpression(format!("{expression:?}"))),
        }
    }

    fn parse_sql_column(ident: &[Ident], schema: &Schema) -> Result<Expression, PipelineError> {
        let (src_field, src_table_or_alias, src_connection) = match ident.len() {
            1 => (&ident[0].value, None, None),
            2 => (&ident[1].value, Some(&ident[0].value), None),
            3 => (
                &ident[2].value,
                Some(&ident[1].value),
                Some(&ident[0].value),
            ),
            _ => {
                return Err(PipelineError::SqlError(SqlError::InvalidColumn(
                    ident
                        .iter()
                        .fold(String::new(), |a, b| a + "." + b.value.as_str()),
                )));
            }
        };

        let matching_by_field: Vec<(usize, &FieldDefinition)> = schema
            .fields
            .iter()
            .enumerate()
            .filter(|(_idx, f)| &f.name == src_field)
            .collect();

        match matching_by_field.len() {
            1 => Ok(Expression::Column {
                index: matching_by_field[0].0,
            }),
            _ => match src_table_or_alias {
                None => Err(PipelineError::SqlError(SqlError::InvalidColumn(
                    ident
                        .iter()
                        .fold(String::new(), |a, b| a + "." + b.value.as_str()),
                ))),
                Some(src_table_or_alias) => {
                    let matching_by_table_or_alias: Vec<(usize, &FieldDefinition)> =
                        matching_by_field
                            .into_iter()
                            .filter(|(_idx, field)| match &field.source {
                                SourceDefinition::Alias { name } => name == src_table_or_alias,
                                SourceDefinition::Table {
                                    name,
                                    connection: _,
                                } => name == src_table_or_alias,
                                _ => false,
                            })
                            .collect();

                    match matching_by_table_or_alias.len() {
                        1 => Ok(Expression::Column {
                            index: matching_by_table_or_alias[0].0,
                        }),
                        _ => match src_connection {
                            None => Err(PipelineError::SqlError(SqlError::InvalidColumn(
                                ident
                                    .iter()
                                    .fold(String::new(), |a, b| a + "." + b.value.as_str()),
                            ))),
                            Some(src_connection) => {
                                let matching_by_connection: Vec<(usize, &FieldDefinition)> =
                                    matching_by_table_or_alias
                                        .into_iter()
                                        .filter(|(_idx, field)| match &field.source {
                                            SourceDefinition::Table {
                                                name: _,
                                                connection,
                                            } => connection == src_connection,
                                            _ => false,
                                        })
                                        .collect();

                                match matching_by_connection.len() {
                                    1 => Ok(Expression::Column {
                                        index: matching_by_connection[0].0,
                                    }),
                                    _ => Err(PipelineError::SqlError(SqlError::InvalidColumn(
                                        ident
                                            .iter()
                                            .fold(String::new(), |a, b| a + "." + b.value.as_str()),
                                    ))),
                                }
                            }
                        },
                    }
                }
            },
        }
    }

    fn parse_sql_trim_function(
        &mut self,
        parse_aggregations: bool,
        expr: &Expr,
        trim_where: &Option<TrimWhereField>,
        trim_what: &Option<Box<Expr>>,
        schema: &Schema,
    ) -> Result<Expression, PipelineError> {
        let arg = Box::new(self.parse_sql_expression(parse_aggregations, expr, schema)?);
        let what = match trim_what {
            Some(e) => Some(Box::new(self.parse_sql_expression(
                parse_aggregations,
                e,
                schema,
            )?)),
            _ => None,
        };
        let typ = trim_where.as_ref().map(|e| match e {
            TrimWhereField::Both => TrimType::Both,
            TrimWhereField::Leading => TrimType::Leading,
            TrimWhereField::Trailing => TrimType::Trailing,
        });
        Ok(Expression::Trim { arg, what, typ })
    }

    fn parse_sql_function(
        &mut self,
        parse_aggregations: bool,
        sql_function: &Function,
        schema: &Schema,
    ) -> Result<Expression, PipelineError> {
        let function_name = sql_function.name.to_string().to_lowercase();

        #[cfg(feature = "python")]
        if function_name.starts_with("py_") {
            // The function is from python udf.
            let udf_name = function_name.strip_prefix("py_").unwrap();
            return self.parse_python_udf(udf_name, sql_function, schema);
        }

        match (
            AggregateFunctionType::new(function_name.as_str()),
            parse_aggregations,
        ) {
            (Ok(aggr), true) => {
                let mut arg_expr: Vec<Expression> = Vec::new();
                for arg in &sql_function.args {
                    let aggregation = self.parse_sql_function_arg(true, arg, schema)?;
                    arg_expr.push(aggregation);
                }
                let measure = Expression::AggregateFunction {
                    fun: aggr,
                    args: arg_expr,
                };
                let index = match self
                    .aggregations
                    .iter()
                    .enumerate()
                    .find(|e| e.1 == &measure)
                {
                    Some((index, _existing)) => index,
                    _ => {
                        self.aggregations.push(measure);
                        self.aggregations.len() - 1
                    }
                };
                Ok(Expression::Column {
                    index: self.offset + index,
                })
            }
            (Ok(_agg), false) => Err(InvalidNestedAggregationFunction(function_name)),
            (Err(_), _) => {
                let mut function_args: Vec<Expression> = Vec::new();
                for arg in &sql_function.args {
                    function_args.push(self.parse_sql_function_arg(
                        parse_aggregations,
                        arg,
                        schema,
                    )?);
                }

                match ScalarFunctionType::new(function_name.as_str()) {
                    Ok(sft) => Ok(ScalarFunction {
                        fun: sft,
                        args: function_args.clone(),
                    }),
                    Err(_d) => match GeoFunctionType::new(function_name.as_str()) {
                        Ok(gft) => Ok(GeoFunction {
                            fun: gft,
                            args: function_args.clone(),
                        }),
                        Err(_e) => match DateTimeFunctionType::new(function_name.as_str()) {
                            Ok(dft) => {
                                let arg = function_args
                                    .first()
                                    .ok_or(InvalidArgument(function_name))
                                    .unwrap();
                                Ok(DateTimeFunction {
                                    fun: dft,
                                    arg: Box::new(arg.clone()),
                                })
                            }
                            Err(_err) => Err(InvalidNestedAggregationFunction(function_name)),
                        },
                    },
                }
            }
        }
    }

    fn parse_sql_function_arg(
        &mut self,
        parse_aggregations: bool,
        argument: &FunctionArg,
        schema: &Schema,
    ) -> Result<Expression, PipelineError> {
        match argument {
            FunctionArg::Named {
                name: _,
                arg: FunctionArgExpr::Expr(arg),
            } => self.parse_sql_expression(parse_aggregations, arg, schema),
            FunctionArg::Named {
                name: _,
                arg: FunctionArgExpr::Wildcard,
            } => Err(InvalidArgument(format!("{argument:?}"))),
            FunctionArg::Unnamed(FunctionArgExpr::Expr(arg)) => {
                self.parse_sql_expression(parse_aggregations, arg, schema)
            }
            FunctionArg::Unnamed(FunctionArgExpr::Wildcard) => {
                Err(InvalidArgument(format!("{argument:?}")))
            }
            FunctionArg::Named {
                name: _,
                arg: FunctionArgExpr::QualifiedWildcard(_),
            } => Err(InvalidArgument(format!("{argument:?}"))),
            FunctionArg::Unnamed(FunctionArgExpr::QualifiedWildcard(_)) => {
                Err(InvalidArgument(format!("{argument:?}")))
            }
        }
    }

    fn parse_sql_unary_op(
        &mut self,
        parse_aggregations: bool,
        op: &SqlUnaryOperator,
        expr: &SqlExpr,
        schema: &Schema,
    ) -> Result<Expression, PipelineError> {
        let arg = Box::new(self.parse_sql_expression(parse_aggregations, expr, schema)?);
        let operator = match op {
            SqlUnaryOperator::Not => UnaryOperatorType::Not,
            SqlUnaryOperator::Plus => UnaryOperatorType::Plus,
            SqlUnaryOperator::Minus => UnaryOperatorType::Minus,
            _ => return Err(InvalidOperator(format!("{op:?}"))),
        };

        Ok(Expression::UnaryOperator { operator, arg })
    }

    fn parse_sql_binary_op(
        &mut self,
        parse_aggregations: bool,
        left: &SqlExpr,
        op: &SqlBinaryOperator,
        right: &SqlExpr,
        schema: &Schema,
    ) -> Result<Expression, PipelineError> {
        let left_op = self.parse_sql_expression(parse_aggregations, left, schema)?;
        let right_op = self.parse_sql_expression(parse_aggregations, right, schema)?;

        let operator = match op {
            SqlBinaryOperator::Gt => BinaryOperatorType::Gt,
            SqlBinaryOperator::GtEq => BinaryOperatorType::Gte,
            SqlBinaryOperator::Lt => BinaryOperatorType::Lt,
            SqlBinaryOperator::LtEq => BinaryOperatorType::Lte,
            SqlBinaryOperator::Eq => BinaryOperatorType::Eq,
            SqlBinaryOperator::NotEq => BinaryOperatorType::Ne,
            SqlBinaryOperator::Plus => BinaryOperatorType::Add,
            SqlBinaryOperator::Minus => BinaryOperatorType::Sub,
            SqlBinaryOperator::Multiply => BinaryOperatorType::Mul,
            SqlBinaryOperator::Divide => BinaryOperatorType::Div,
            SqlBinaryOperator::Modulo => BinaryOperatorType::Mod,
            SqlBinaryOperator::And => BinaryOperatorType::And,
            SqlBinaryOperator::Or => BinaryOperatorType::Or,
            _ => return Err(InvalidOperator(format!("{op:?}"))),
        };

        Ok(Expression::BinaryOperator {
            left: Box::new(left_op),
            operator,
            right: Box::new(right_op),
        })
    }

    fn parse_sql_number(n: &str) -> Result<Expression, PipelineError> {
        match n.parse::<i64>() {
            Ok(n) => Ok(Expression::Literal(Field::Int(n))),
            Err(_) => match n.parse::<f64>() {
                Ok(f) => Ok(Expression::Literal(Field::Float(OrderedFloat(f)))),
                Err(_) => Err(InvalidValue(n.to_string())),
            },
        }
    }

    fn parse_sql_like_operator(
        &mut self,
        parse_aggregations: bool,
        negated: &bool,
        expr: &Expr,
        pattern: &Expr,
        escape_char: &Option<char>,
        schema: &Schema,
    ) -> Result<Expression, PipelineError> {
        let arg = self.parse_sql_expression(parse_aggregations, expr, schema)?;
        let pattern = self.parse_sql_expression(parse_aggregations, pattern, schema)?;
        let like_expression = Expression::Like {
            arg: Box::new(arg),
            pattern: Box::new(pattern),
            escape: *escape_char,
        };
        if *negated {
            Ok(Expression::UnaryOperator {
                operator: UnaryOperatorType::Not,
                arg: Box::new(like_expression),
            })
        } else {
            Ok(like_expression)
        }
    }

    fn parse_sql_cast_operator(
        &mut self,
        parse_aggregations: bool,
        expr: &Expr,
        data_type: &DataType,
        schema: &Schema,
    ) -> Result<Expression, PipelineError> {
        let expression = self.parse_sql_expression(parse_aggregations, expr, schema)?;
        let cast_to = match data_type {
            DataType::Decimal(_) => CastOperatorType::Decimal,
            DataType::Binary(_) => CastOperatorType::Binary,
            DataType::Float(_) => CastOperatorType::Float,
            DataType::Int(_) => CastOperatorType::Int,
            DataType::Integer(_) => CastOperatorType::Int,
            DataType::UnsignedInt(_) => CastOperatorType::UInt,
            DataType::UnsignedInteger(_) => CastOperatorType::UInt,
            DataType::Boolean => CastOperatorType::Boolean,
            DataType::Date => CastOperatorType::Date,
            DataType::Timestamp(..) => CastOperatorType::Timestamp,
            DataType::Text => CastOperatorType::Text,
            DataType::String => CastOperatorType::String,
            DataType::Custom(name, ..) => {
                if name.to_string().to_lowercase() == "bson" {
                    CastOperatorType::Bson
                } else {
                    Err(PipelineError::InvalidFunction(format!(
                        "Unsupported Cast type {name}"
                    )))?
                }
            }
            _ => Err(PipelineError::InvalidFunction(format!(
                "Unsupported Cast type {data_type}"
            )))?,
        };
        Ok(Expression::Cast {
            arg: Box::new(expression),
            typ: cast_to,
        })
    }

    fn parse_sql_string(s: &str) -> Result<Expression, PipelineError> {
        Ok(Expression::Literal(Field::String(s.to_owned())))
    }

    pub fn fullname_from_ident(ident: &[Ident]) -> String {
        let mut ident_tokens = vec![];
        for token in ident.iter() {
            ident_tokens.push(token.value.clone());
        }
        ident_tokens.join(".")
    }

    pub(crate) fn normalize_ident(id: &Ident) -> String {
        match id.quote_style {
            Some(_) => id.value.clone(),
            None => id.value.clone(),
        }
    }

    #[cfg(feature = "python")]
    fn parse_python_udf(
        &mut self,
        name: &str,
        function: &Function,
        schema: &Schema,
    ) -> Result<Expression, PipelineError> {
        // First, get python function define by name.
        // Then, transfer python function to Expression::PythonUDF

        use dozer_types::types::FieldType;
        use PipelineError::InvalidQuery;

        let args = function
            .args
            .iter()
            .map(|argument| self.parse_sql_function_arg(false, argument, schema))
            .collect::<Result<Vec<_>, PipelineError>>()?;

        let last_arg = args
            .last()
            .ok_or_else(|| InvalidQuery("Can't get python udf return type".to_string()))?;

        let return_type = match last_arg {
            Expression::Literal(Field::String(s)) => {
                FieldType::try_from(s.as_str()).map_err(|e| InvalidQuery(format!("Failed to parse Python UDF return type: {e}")))?
            }
            _ => return Err(InvalidArgument("The last arg for python udf should be a string literal, which represents return type".to_string())),
        };

        Ok(Expression::PythonUDF {
            name: name.to_string(),
            args,
            return_type,
        })
    }
}

#[derive(Debug, Clone, Hash, PartialEq, Eq)]
pub struct NameOrAlias(pub String, pub Option<String>);

pub fn extend_schema_source_def(schema: &Schema, name: &NameOrAlias) -> Schema {
    let mut output_schema = schema.clone();
    let mut fields = vec![];
    for mut field in schema.clone().fields.into_iter() {
        if let Some(alias) = &name.1 {
            field.source = SourceDefinition::Alias {
                name: alias.to_string(),
            };
        }

        fields.push(field);
    }
    output_schema.fields = fields;

    output_schema
}

1	use dozer_types::{
2	ordered_float::OrderedFloat,
3	types::{Field, FieldDefinition, Schema, SourceDefinition},
4	};
5	use sqlparser::ast::{
6	BinaryOperator as SqlBinaryOperator, DataType, Expr as SqlExpr, Expr, Function, FunctionArg,
7	FunctionArgExpr, Ident, TrimWhereField, UnaryOperator as SqlUnaryOperator, Value as SqlValue,
8	};
9
10	use crate::pipeline::errors::PipelineError::{
11	InvalidArgument, InvalidExpression, InvalidNestedAggregationFunction, InvalidOperator,
12	InvalidValue,
13	};
14	use crate::pipeline::errors::{PipelineError, SqlError};
15	use crate::pipeline::expression::aggregate::AggregateFunctionType;
16	use crate::pipeline::expression::datetime::DateTimeFunctionType;
17
18	use crate::pipeline::expression::execution::Expression;
19	use crate::pipeline::expression::execution::Expression::{
20	DateTimeFunction, GeoFunction, ScalarFunction,
21	};
22	use crate::pipeline::expression::geo::common::GeoFunctionType;
23	use crate::pipeline::expression::operator::{BinaryOperatorType, UnaryOperatorType};
24	use crate::pipeline::expression::scalar::common::ScalarFunctionType;
25	use crate::pipeline::expression::scalar::string::TrimType;
26
27	use super::cast::CastOperatorType;
28
29	#[derive(Clone, PartialEq, Debug)]	9✔
30	pub struct ExpressionBuilder {
31	// Must be an aggregation function
32	pub aggregations: Vec<Expression>,
33	pub offset: usize,
34	}
35
36	impl ExpressionBuilder {
37	pub fn new(offset: usize) -> Self {	2,969✔
38	Self {	2,969✔
39	aggregations: Vec::new(),	2,969✔
40	offset,	2,969✔
41	}	2,969✔
42	}	2,969✔
43
44	pub fn from(offset: usize, aggregations: Vec<Expression>) -> Self {	1✔
45	Self {	1✔
46	aggregations,	1✔
47	offset,	1✔
48	}	1✔
49	}	1✔
50
51	pub fn build(	2,848✔
52	&mut self,	2,848✔
53	parse_aggregations: bool,	2,848✔
54	sql_expression: &SqlExpr,	2,848✔
55	schema: &Schema,	2,848✔
56	) -> Result<Expression, PipelineError> {	2,848✔
57	self.parse_sql_expression(parse_aggregations, sql_expression, schema)	2,848✔
58	}	2,848✔
59
60	pub(crate) fn parse_sql_expression(
61	&mut self,
62	parse_aggregations: bool,
63	expression: &SqlExpr,
64	schema: &Schema,
65	) -> Result<Expression, PipelineError> {
66	match expression {	158✔
67	SqlExpr::Trim {
68	expr,	44✔
69	trim_where,	44✔
70	trim_what,	44✔
71	} => self.parse_sql_trim_function(	44✔
72	parse_aggregations,	44✔
73	expr,	44✔
74	trim_where,	44✔
75	trim_what,	44✔
76	schema,	44✔
77	),	44✔
78	SqlExpr::Identifier(ident) => Self::parse_sql_column(&[ident.clone()], schema),	2,720✔
79	SqlExpr::CompoundIdentifier(ident) => Self::parse_sql_column(ident, schema),	335✔
80	SqlExpr::Value(SqlValue::Number(n, _)) => Self::parse_sql_number(n),	80✔
81	SqlExpr::Value(SqlValue::Null) => Ok(Expression::Literal(Field::Null)),	×
82	SqlExpr::Value(SqlValue::SingleQuotedString(s) \| SqlValue::DoubleQuotedString(s)) => {	78✔
83	Self::parse_sql_string(s)	78✔
84	}
85	SqlExpr::UnaryOp { expr, op } => {	×
86	self.parse_sql_unary_op(parse_aggregations, op, expr, schema)	×
87	}
88	SqlExpr::BinaryOp { left, op, right } => {	138✔
89	self.parse_sql_binary_op(parse_aggregations, left, op, right, schema)	138✔
90	}
91	SqlExpr::Nested(expr) => self.parse_sql_expression(parse_aggregations, expr, schema),	21✔
92	SqlExpr::Function(sql_function) => {	333✔
93	self.parse_sql_function(parse_aggregations, sql_function, schema)	333✔
94	}
95	SqlExpr::Like {
96	negated,	×
97	expr,	×
98	pattern,	×
99	escape_char,	×
100	} => self.parse_sql_like_operator(	×
101	parse_aggregations,	×
102	negated,	×
103	expr,	×
104	pattern,	×
105	escape_char,	×
106	schema,	×
107	),	×
108	SqlExpr::Cast { expr, data_type } => {	65✔
109	self.parse_sql_cast_operator(parse_aggregations, expr, data_type, schema)	65✔
110	}
111	_ => Err(InvalidExpression(format!("{expression:?}"))),	×
112	}
113	}	3,814✔
114
115	fn parse_sql_column(ident: &[Ident], schema: &Schema) -> Result<Expression, PipelineError> {	3,056✔
116	let (src_field, src_table_or_alias, src_connection) = match ident.len() {	3,056✔
117	1 => (&ident[0].value, None, None),	2,721✔
118	2 => (&ident[1].value, Some(&ident[0].value), None),	334✔
119	3 => (	1✔
120	&ident[2].value,	1✔
121	Some(&ident[1].value),	1✔
122	Some(&ident[0].value),	1✔
123	),	1✔
124	_ => {
125	return Err(PipelineError::SqlError(SqlError::InvalidColumn(	×
126	ident	×
127	.iter()	×
128	.fold(String::new(), \|a, b\| a + "." + b.value.as_str()),	×
129	)));	×
130	}
131	};
132
133	let matching_by_field: Vec<(usize, &FieldDefinition)> = schema	3,056✔
134	.fields	3,056✔
135	.iter()	3,056✔
136	.enumerate()	3,056✔
137	.filter(\|(_idx, f)\| &f.name == src_field)	13,018✔
138	.collect();	3,056✔
139		3,056✔
140	match matching_by_field.len() {	3,056✔
141	1 => Ok(Expression::Column {	2,900✔
142	index: matching_by_field[0].0,	2,900✔
143	}),	2,900✔
144	_ => match src_table_or_alias {	156✔
145	None => Err(PipelineError::SqlError(SqlError::InvalidColumn(	×
146	ident	×
147	.iter()	×
148	.fold(String::new(), \|a, b\| a + "." + b.value.as_str()),	×
149	))),	×
150	Some(src_table_or_alias) => {	156✔
151	let matching_by_table_or_alias: Vec<(usize, &FieldDefinition)> =	156✔
152	matching_by_field	156✔
153	.into_iter()	156✔
154	.filter(\|(_idx, field)\| match &field.source {	301✔
155	SourceDefinition::Alias { name } => name == src_table_or_alias,	240✔
156	SourceDefinition::Table {
157	name,	60✔
158	connection: _,	60✔
159	} => name == src_table_or_alias,	60✔
160	_ => false,	×
161	})	301✔
162	.collect();	156✔
163		156✔
164	match matching_by_table_or_alias.len() {	156✔
165	1 => Ok(Expression::Column {	150✔
166	index: matching_by_table_or_alias[0].0,	150✔
167	}),	150✔
168	_ => match src_connection {	6✔
169	None => Err(PipelineError::SqlError(SqlError::InvalidColumn(	×
170	ident	×
171	.iter()	×
172	.fold(String::new(), \|a, b\| a + "." + b.value.as_str()),	×
173	))),	×
174	Some(src_connection) => {	6✔
175	let matching_by_connection: Vec<(usize, &FieldDefinition)> =	6✔
176	matching_by_table_or_alias	6✔
177	.into_iter()	6✔
178	.filter(\|(_idx, field)\| match &field.source {	6✔
179	SourceDefinition::Table {
180	name: _,
181	connection,	×
182	} => connection == src_connection,	×
183	_ => false,	×
184	})	6✔
185	.collect();	6✔
186		6✔
187	match matching_by_connection.len() {	6✔
188	1 => Ok(Expression::Column {	×
189	index: matching_by_connection[0].0,	×
190	}),	×
191	_ => Err(PipelineError::SqlError(SqlError::InvalidColumn(	×
192	ident	×
193	.iter()	×
194	.fold(String::new(), \|a, b\| a + "." + b.value.as_str()),	×
195	))),	×
196	}
197	}
198	},
199	}
200	}
201	},
202	}
203	}	3,050✔
204
205	fn parse_sql_trim_function(	44✔
206	&mut self,	44✔
207	parse_aggregations: bool,	44✔
208	expr: &Expr,	44✔
209	trim_where: &Option<TrimWhereField>,	44✔
210	trim_what: &Option<Box<Expr>>,	44✔
211	schema: &Schema,	44✔
212	) -> Result<Expression, PipelineError> {	44✔
213	let arg = Box::new(self.parse_sql_expression(parse_aggregations, expr, schema)?);	44✔
214	let what = match trim_what {	44✔
215	Some(e) => Some(Box::new(self.parse_sql_expression(	8✔
216	parse_aggregations,	8✔
217	e,	8✔
218	schema,	8✔
219	)?)),	8✔
220	_ => None,	36✔
221	};
222	let typ = trim_where.as_ref().map(\|e\| match e {	44✔
223	TrimWhereField::Both => TrimType::Both,	2✔
224	TrimWhereField::Leading => TrimType::Leading,	2✔
225	TrimWhereField::Trailing => TrimType::Trailing,	2✔
226	});	44✔
227	Ok(Expression::Trim { arg, what, typ })	44✔
228	}	44✔
229
230	fn parse_sql_function(	333✔
231	&mut self,	333✔
232	parse_aggregations: bool,	333✔
233	sql_function: &Function,	333✔
234	schema: &Schema,	333✔
235	) -> Result<Expression, PipelineError> {	333✔
236	let function_name = sql_function.name.to_string().to_lowercase();	333✔
237		333✔
238	#[cfg(feature = "python")]	333✔
239	if function_name.starts_with("py_") {	333✔
240	// The function is from python udf.
241	let udf_name = function_name.strip_prefix("py_").unwrap();	30✔
242	return self.parse_python_udf(udf_name, sql_function, schema);	30✔
243	}	303✔
244		303✔
245	match (	303✔
246	AggregateFunctionType::new(function_name.as_str()),	303✔
247	parse_aggregations,	303✔
248	) {	303✔
249	(Ok(aggr), true) => {	252✔
250	let mut arg_expr: Vec<Expression> = Vec::new();	252✔
251	for arg in &sql_function.args {	507✔
252	let aggregation = self.parse_sql_function_arg(true, arg, schema)?;	255✔
253	arg_expr.push(aggregation);	255✔
254	}
255	let measure = Expression::AggregateFunction {	255✔
256	fun: aggr,	255✔
257	args: arg_expr,	255✔
258	};	255✔
259	let index = match self	255✔
260	.aggregations	255✔
261	.iter()	255✔
262	.enumerate()	255✔
263	.find(\|e\| e.1 == &measure)	255✔
264	{
265	Some((index, _existing)) => index,	2✔
266	_ => {
267	self.aggregations.push(measure);	250✔
268	self.aggregations.len() - 1	250✔
269	}
270	};
271	Ok(Expression::Column {	252✔
272	index: self.offset + index,	252✔
273	})	252✔
274	}
275	(Ok(_agg), false) => Err(InvalidNestedAggregationFunction(function_name)),	×
276	(Err(_), _) => {
277	let mut function_args: Vec<Expression> = Vec::new();	52✔
278	for arg in &sql_function.args {	151✔
279	function_args.push(self.parse_sql_function_arg(	99✔
280	parse_aggregations,	99✔
281	arg,	99✔
282	schema,	99✔
283	)?);	99✔
284	}
285
286	match ScalarFunctionType::new(function_name.as_str()) {	52✔
287	Ok(sft) => Ok(ScalarFunction {	34✔
288	fun: sft,	34✔
289	args: function_args.clone(),	34✔
290	}),	34✔
291	Err(_d) => match GeoFunctionType::new(function_name.as_str()) {	18✔
292	Ok(gft) => Ok(GeoFunction {	14✔
293	fun: gft,	14✔
294	args: function_args.clone(),	14✔
295	}),	14✔
296	Err(_e) => match DateTimeFunctionType::new(function_name.as_str()) {	4✔
297	Ok(dft) => {	4✔
298	let arg = function_args	4✔
299	.first()	4✔
300	.ok_or(InvalidArgument(function_name))	4✔
301	.unwrap();	4✔
302	Ok(DateTimeFunction {	4✔
303	fun: dft,	4✔
304	arg: Box::new(arg.clone()),	4✔
305	})	4✔
306	}
307	Err(_err) => Err(InvalidNestedAggregationFunction(function_name)),	×
308	},
309	},
310	}
311	}
312	}
313	}	334✔
314
315	fn parse_sql_function_arg(
316	&mut self,
317	parse_aggregations: bool,
318	argument: &FunctionArg,
319	schema: &Schema,
320	) -> Result<Expression, PipelineError> {
321	match argument {	426✔
322	FunctionArg::Named {
323	name: _,
324	arg: FunctionArgExpr::Expr(arg),	×
325	} => self.parse_sql_expression(parse_aggregations, arg, schema),	×
326	FunctionArg::Named {
327	name: _,
328	arg: FunctionArgExpr::Wildcard,
329	} => Err(InvalidArgument(format!("{argument:?}"))),	×
330	FunctionArg::Unnamed(FunctionArgExpr::Expr(arg)) => {	426✔
331	self.parse_sql_expression(parse_aggregations, arg, schema)	426✔
332	}
333	FunctionArg::Unnamed(FunctionArgExpr::Wildcard) => {
334	Err(InvalidArgument(format!("{argument:?}")))	×
335	}
336	FunctionArg::Named {
337	name: _,
338	arg: FunctionArgExpr::QualifiedWildcard(_),
339	} => Err(InvalidArgument(format!("{argument:?}"))),	×
340	FunctionArg::Unnamed(FunctionArgExpr::QualifiedWildcard(_)) => {
341	Err(InvalidArgument(format!("{argument:?}")))	×
342	}
343	}
344	}	426✔
345
346	fn parse_sql_unary_op(	×
347	&mut self,	×
348	parse_aggregations: bool,	×
349	op: &SqlUnaryOperator,	×
350	expr: &SqlExpr,	×
351	schema: &Schema,	×
352	) -> Result<Expression, PipelineError> {	×
353	let arg = Box::new(self.parse_sql_expression(parse_aggregations, expr, schema)?);	×
354	let operator = match op {	×
355	SqlUnaryOperator::Not => UnaryOperatorType::Not,	×
356	SqlUnaryOperator::Plus => UnaryOperatorType::Plus,	×
357	SqlUnaryOperator::Minus => UnaryOperatorType::Minus,	×
358	_ => return Err(InvalidOperator(format!("{op:?}"))),	×
359	};
360
361	Ok(Expression::UnaryOperator { operator, arg })	×
362	}	×
363
364	fn parse_sql_binary_op(	138✔
365	&mut self,	138✔
366	parse_aggregations: bool,	138✔
367	left: &SqlExpr,	138✔
368	op: &SqlBinaryOperator,	138✔
369	right: &SqlExpr,	138✔
370	schema: &Schema,	138✔
371	) -> Result<Expression, PipelineError> {	138✔
372	let left_op = self.parse_sql_expression(parse_aggregations, left, schema)?;	138✔
373	let right_op = self.parse_sql_expression(parse_aggregations, right, schema)?;	138✔
374
375	let operator = match op {	138✔
376	SqlBinaryOperator::Gt => BinaryOperatorType::Gt,	31✔
377	SqlBinaryOperator::GtEq => BinaryOperatorType::Gte,	1✔
378	SqlBinaryOperator::Lt => BinaryOperatorType::Lt,	20✔
379	SqlBinaryOperator::LtEq => BinaryOperatorType::Lte,	20✔
380	SqlBinaryOperator::Eq => BinaryOperatorType::Eq,	30✔
381	SqlBinaryOperator::NotEq => BinaryOperatorType::Ne,	×
382	SqlBinaryOperator::Plus => BinaryOperatorType::Add,	4✔
383	SqlBinaryOperator::Minus => BinaryOperatorType::Sub,	2✔
384	SqlBinaryOperator::Multiply => BinaryOperatorType::Mul,	×
385	SqlBinaryOperator::Divide => BinaryOperatorType::Div,	×
386	SqlBinaryOperator::Modulo => BinaryOperatorType::Mod,	×
387	SqlBinaryOperator::And => BinaryOperatorType::And,	20✔
388	SqlBinaryOperator::Or => BinaryOperatorType::Or,	10✔
389	_ => return Err(InvalidOperator(format!("{op:?}"))),	×
390	};
391
392	Ok(Expression::BinaryOperator {	138✔
393	left: Box::new(left_op),	138✔
394	operator,	138✔
395	right: Box::new(right_op),	138✔
396	})	138✔
397	}	138✔
398
399	fn parse_sql_number(n: &str) -> Result<Expression, PipelineError> {	80✔
400	match n.parse::<i64>() {	80✔
401	Ok(n) => Ok(Expression::Literal(Field::Int(n))),	80✔
402	Err(_) => match n.parse::<f64>() {	×
403	Ok(f) => Ok(Expression::Literal(Field::Float(OrderedFloat(f)))),	×
404	Err(_) => Err(InvalidValue(n.to_string())),	×
405	},
406	}
407	}	80✔
408
409	fn parse_sql_like_operator(	×
410	&mut self,	×
411	parse_aggregations: bool,	×
412	negated: &bool,	×
413	expr: &Expr,	×
414	pattern: &Expr,	×
415	escape_char: &Option<char>,	×
416	schema: &Schema,	×
417	) -> Result<Expression, PipelineError> {	×
418	let arg = self.parse_sql_expression(parse_aggregations, expr, schema)?;	×
419	let pattern = self.parse_sql_expression(parse_aggregations, pattern, schema)?;	×
420	let like_expression = Expression::Like {	×
421	arg: Box::new(arg),	×
422	pattern: Box::new(pattern),	×
423	escape: *escape_char,	×
424	};	×
425	if *negated {	×
426	Ok(Expression::UnaryOperator {	×
427	operator: UnaryOperatorType::Not,	×
428	arg: Box::new(like_expression),	×
429	})	×
430	} else {
431	Ok(like_expression)	×
432	}
433	}	×
434
435	fn parse_sql_cast_operator(	65✔
436	&mut self,	65✔
437	parse_aggregations: bool,	65✔
438	expr: &Expr,	65✔
439	data_type: &DataType,	65✔
440	schema: &Schema,	65✔
441	) -> Result<Expression, PipelineError> {	65✔
442	let expression = self.parse_sql_expression(parse_aggregations, expr, schema)?;	65✔
443	let cast_to = match data_type {	65✔
444	DataType::Decimal(_) => CastOperatorType::Decimal,	×
445	DataType::Binary(_) => CastOperatorType::Binary,	×
446	DataType::Float(_) => CastOperatorType::Float,	10✔
447	DataType::Int(_) => CastOperatorType::Int,	6✔
448	DataType::Integer(_) => CastOperatorType::Int,	×
449	DataType::UnsignedInt(_) => CastOperatorType::UInt,	×
450	DataType::UnsignedInteger(_) => CastOperatorType::UInt,	×
451	DataType::Boolean => CastOperatorType::Boolean,	12✔
452	DataType::Date => CastOperatorType::Date,	×
453	DataType::Timestamp(..) => CastOperatorType::Timestamp,	×
454	DataType::Text => CastOperatorType::Text,	18✔
455	DataType::String => CastOperatorType::String,	19✔
456	DataType::Custom(name, ..) => {	×
457	if name.to_string().to_lowercase() == "bson" {	×
458	CastOperatorType::Bson	×
459	} else {
460	Err(PipelineError::InvalidFunction(format!(	×
461	"Unsupported Cast type {name}"	×
462	)))?	×
463	}
464	}
465	_ => Err(PipelineError::InvalidFunction(format!(	×
466	"Unsupported Cast type {data_type}"	×
467	)))?,	×
468	};
469	Ok(Expression::Cast {	65✔
470	arg: Box::new(expression),	65✔
471	typ: cast_to,	65✔
472	})	65✔
473	}	65✔
474
475	fn parse_sql_string(s: &str) -> Result<Expression, PipelineError> {	78✔
476	Ok(Expression::Literal(Field::String(s.to_owned())))	78✔
477	}	78✔
478
479	pub fn fullname_from_ident(ident: &[Ident]) -> String {	110✔
480	let mut ident_tokens = vec![];	110✔
481	for token in ident.iter() {	110✔
482	ident_tokens.push(token.value.clone());	110✔
483	}	110✔
484	ident_tokens.join(".")	110✔
485	}	110✔
486
487	pub(crate) fn normalize_ident(id: &Ident) -> String {	391✔
488	match id.quote_style {	391✔
489	Some(_) => id.value.clone(),	×
490	None => id.value.clone(),	391✔
491	}
492	}	391✔
493
494	#[cfg(feature = "python")]
495	fn parse_python_udf(	30✔
496	&mut self,	30✔
497	name: &str,	30✔
498	function: &Function,	30✔
499	schema: &Schema,	30✔
500	) -> Result<Expression, PipelineError> {	30✔
501	// First, get python function define by name.
502	// Then, transfer python function to Expression::PythonUDF
503
504	use dozer_types::types::FieldType;
505	use PipelineError::InvalidQuery;
506
507	let args = function	30✔
508	.args	30✔
509	.iter()	30✔
510	.map(\|argument\| self.parse_sql_function_arg(false, argument, schema))	75✔
511	.collect::<Result<Vec<_>, PipelineError>>()?;	30✔
512
513	let last_arg = args	30✔
514	.last()	30✔
515	.ok_or_else(\|\| InvalidQuery("Can't get python udf return type".to_string()))?;	30✔
516
517	let return_type = match last_arg {	30✔
518	Expression::Literal(Field::String(s)) => {	30✔
519	FieldType::try_from(s.as_str()).map_err(\|e\| InvalidQuery(format!("Failed to parse Python UDF return type: {e}")))?	30✔
520	}
521	_ => return Err(InvalidArgument("The last arg for python udf should be a string literal, which represents return type".to_string())),	×
522	};
523
524	Ok(Expression::PythonUDF {	30✔
525	name: name.to_string(),	30✔
526	args,	30✔
527	return_type,	30✔
528	})	30✔
529	}	30✔
530	}
531
532	#[derive(Debug, Clone, Hash, PartialEq, Eq)]	5,304✔
533	pub struct NameOrAlias(pub String, pub Option<String>);
534
535	pub fn extend_schema_source_def(schema: &Schema, name: &NameOrAlias) -> Schema {	1,169✔
536	let mut output_schema = schema.clone();	1,169✔
537	let mut fields = vec![];	1,169✔
538	for mut field in schema.clone().fields.into_iter() {	4,500✔
539	if let Some(alias) = &name.1 {	4,500✔
540	field.source = SourceDefinition::Alias {	1,290✔
541	name: alias.to_string(),	1,290✔
542	};	1,290✔
543	}	3,210✔
544
545	fields.push(field);	4,500✔
546	}
547	output_schema.fields = fields;	1,169✔
548		1,169✔
549	output_schema	1,169✔
550	}	1,169✔

getdozer / dozer / 4324181106

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous