4423198119

Build # 4423198119

Build Type

push

github

Committed by GitHub

Commit Message

Bump geo from 0.23.1 to 0.24.0 (#1207)

Run Details

28601 of 39109 relevant lines covered (73.13%)

51246.35 hits per line

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

70.7

/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::{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)
            }
            SqlExpr::Extract { field, expr } => {
                self.parse_sql_extract_operator(parse_aggregations, field, expr, 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()
                        .map(|e| e.value.as_str())
                        .collect::<Vec<&str>>()
                        .join("."),
                )));
            }
        };

        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()
                        .map(|e| e.value.as_str())
                        .collect::<Vec<&str>>()
                        .join("."),
                ))),
                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()
                                            .map(|e| e.value.as_str())
                                            .collect::<Vec<&str>>()
                                            .join("."),
                                    ))),
                                }
                            }
                        },
                    }
                }
            },
        }
    }

    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(_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_extract_operator(
        &mut self,
        parse_aggregations: bool,
        field: &sqlparser::ast::DateTimeField,
        expr: &Expr,
        schema: &Schema,
    ) -> Result<Expression, PipelineError> {
        let right = self.parse_sql_expression(parse_aggregations, expr, schema)?;
        Ok(Expression::DateTimeFunction {
            fun: DateTimeFunctionType::Extract { field: *field },
            arg: Box::new(right),
        })
    }

    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::{GeoFunction, ScalarFunction};
20	use crate::pipeline::expression::geo::common::GeoFunctionType;
21	use crate::pipeline::expression::operator::{BinaryOperatorType, UnaryOperatorType};
22	use crate::pipeline::expression::scalar::common::ScalarFunctionType;
23	use crate::pipeline::expression::scalar::string::TrimType;
24
25	use super::cast::CastOperatorType;
26
27	#[derive(Clone, PartialEq, Debug)]	9✔
28	pub struct ExpressionBuilder {
29	// Must be an aggregation function
30	pub aggregations: Vec<Expression>,
31	pub offset: usize,
32	}
33
34	impl ExpressionBuilder {
35	pub fn new(offset: usize) -> Self {	3,161✔
36	Self {	3,161✔
37	aggregations: Vec::new(),	3,161✔
38	offset,	3,161✔
39	}	3,161✔
40	}	3,161✔
41
42	pub fn from(offset: usize, aggregations: Vec<Expression>) -> Self {	6✔
43	Self {	6✔
44	aggregations,	6✔
45	offset,	6✔
46	}	6✔
47	}	6✔
48
49	pub fn build(	2,984✔
50	&mut self,	2,984✔
51	parse_aggregations: bool,	2,984✔
52	sql_expression: &SqlExpr,	2,984✔
53	schema: &Schema,	2,984✔
54	) -> Result<Expression, PipelineError> {	2,984✔
55	self.parse_sql_expression(parse_aggregations, sql_expression, schema)	2,984✔
56	}	2,984✔
57
58	pub(crate) fn parse_sql_expression(
59	&mut self,
60	parse_aggregations: bool,
61	expression: &SqlExpr,
62	schema: &Schema,
63	) -> Result<Expression, PipelineError> {
64	match expression {	165✔
65	SqlExpr::Trim {
66	expr,	44✔
67	trim_where,	44✔
68	trim_what,	44✔
69	} => self.parse_sql_trim_function(	44✔
70	parse_aggregations,	44✔
71	expr,	44✔
72	trim_where,	44✔
73	trim_what,	44✔
74	schema,	44✔
75	),	44✔
76	SqlExpr::Identifier(ident) => Self::parse_sql_column(&[ident.clone()], schema),	2,862✔
77	SqlExpr::CompoundIdentifier(ident) => Self::parse_sql_column(ident, schema),	385✔
78	SqlExpr::Value(SqlValue::Number(n, _)) => Self::parse_sql_number(n),	87✔
79	SqlExpr::Value(SqlValue::Null) => Ok(Expression::Literal(Field::Null)),	×
80	SqlExpr::Value(SqlValue::SingleQuotedString(s) \| SqlValue::DoubleQuotedString(s)) => {	78✔
81	Self::parse_sql_string(s)	78✔
82	}
83	SqlExpr::UnaryOp { expr, op } => {	×
84	self.parse_sql_unary_op(parse_aggregations, op, expr, schema)	×
85	}
86	SqlExpr::BinaryOp { left, op, right } => {	147✔
87	self.parse_sql_binary_op(parse_aggregations, left, op, right, schema)	147✔
88	}
89	SqlExpr::Nested(expr) => self.parse_sql_expression(parse_aggregations, expr, schema),	21✔
90	SqlExpr::Function(sql_function) => {	346✔
91	self.parse_sql_function(parse_aggregations, sql_function, schema)	346✔
92	}
93	SqlExpr::Like {
94	negated,	×
95	expr,	×
96	pattern,	×
97	escape_char,	×
98	} => self.parse_sql_like_operator(	×
99	parse_aggregations,	×
100	negated,	×
101	expr,	×
102	pattern,	×
103	escape_char,	×
104	schema,	×
105	),	×
106	SqlExpr::Cast { expr, data_type } => {	65✔
107	self.parse_sql_cast_operator(parse_aggregations, expr, data_type, schema)	65✔
108	}
109	SqlExpr::Extract { field, expr } => {	64✔
110	self.parse_sql_extract_operator(parse_aggregations, field, expr, schema)	64✔
111	}
112	_ => Err(InvalidExpression(format!("{expression:?}"))),	×
113	}
114	}	4,099✔
115
116	fn parse_sql_column(ident: &[Ident], schema: &Schema) -> Result<Expression, PipelineError> {	3,249✔
117	let (src_field, src_table_or_alias, src_connection) = match ident.len() {	3,249✔
118	1 => (&ident[0].value, None, None),	2,864✔
119	2 => (&ident[1].value, Some(&ident[0].value), None),	384✔
120	3 => (	1✔
121	&ident[2].value,	1✔
122	Some(&ident[1].value),	1✔
123	Some(&ident[0].value),	1✔
124	),	1✔
125	_ => {
126	return Err(PipelineError::SqlError(SqlError::InvalidColumn(	×
127	ident	×
128	.iter()	×
129	.map(\|e\| e.value.as_str())	×
130	.collect::<Vec<&str>>()	×
131	.join("."),	×
132	)));	×
133	}
134	};
135
136	let matching_by_field: Vec<(usize, &FieldDefinition)> = schema	3,249✔
137	.fields	3,249✔
138	.iter()	3,249✔
139	.enumerate()	3,249✔
140	.filter(\|(_idx, f)\| &f.name == src_field)	16,369✔
141	.collect();	3,249✔
142		3,249✔
143	match matching_by_field.len() {	3,249✔
144	1 => Ok(Expression::Column {	3,044✔
145	index: matching_by_field[0].0,	3,044✔
146	}),	3,044✔
147	_ => match src_table_or_alias {	205✔
148	None => Err(PipelineError::SqlError(SqlError::InvalidColumn(	×
149	ident	×
150	.iter()	×
151	.map(\|e\| e.value.as_str())	×
152	.collect::<Vec<&str>>()	×
153	.join("."),	×
154	))),	×
155	Some(src_table_or_alias) => {	205✔
156	let matching_by_table_or_alias: Vec<(usize, &FieldDefinition)> =	205✔
157	matching_by_field	205✔
158	.into_iter()	205✔
159	.filter(\|(_idx, field)\| match &field.source {	400✔
160	SourceDefinition::Alias { name } => name == src_table_or_alias,	340✔
161	SourceDefinition::Table {
162	name,	60✔
163	connection: _,	60✔
164	} => name == src_table_or_alias,	60✔
165	_ => false,	×
166	})	400✔
167	.collect();	205✔
168		205✔
169	match matching_by_table_or_alias.len() {	205✔
170	1 => Ok(Expression::Column {	200✔
171	index: matching_by_table_or_alias[0].0,	200✔
172	}),	200✔
173	_ => match src_connection {	5✔
174	None => Err(PipelineError::SqlError(SqlError::InvalidColumn(	×
175	ident	×
176	.iter()	×
177	.fold(String::new(), \|a, b\| a + "." + b.value.as_str()),	×
178	))),	×
179	Some(src_connection) => {	5✔
180	let matching_by_connection: Vec<(usize, &FieldDefinition)> =	5✔
181	matching_by_table_or_alias	5✔
182	.into_iter()	5✔
183	.filter(\|(_idx, field)\| match &field.source {	5✔
184	SourceDefinition::Table {
185	name: _,
186	connection,	×
187	} => connection == src_connection,	×
188	_ => false,	×
189	})	5✔
190	.collect();	5✔
191		5✔
192	match matching_by_connection.len() {	5✔
193	1 => Ok(Expression::Column {	×
194	index: matching_by_connection[0].0,	×
195	}),	×
196	_ => Err(PipelineError::SqlError(SqlError::InvalidColumn(	×
197	ident	×
198	.iter()	×
199	.map(\|e\| e.value.as_str())	×
200	.collect::<Vec<&str>>()	×
201	.join("."),	×
202	))),	×
203	}
204	}
205	},
206	}
207	}
208	},
209	}
210	}	3,244✔
211
212	fn parse_sql_trim_function(	44✔
213	&mut self,	44✔
214	parse_aggregations: bool,	44✔
215	expr: &Expr,	44✔
216	trim_where: &Option<TrimWhereField>,	44✔
217	trim_what: &Option<Box<Expr>>,	44✔
218	schema: &Schema,	44✔
219	) -> Result<Expression, PipelineError> {	44✔
220	let arg = Box::new(self.parse_sql_expression(parse_aggregations, expr, schema)?);	44✔
221	let what = match trim_what {	44✔
222	Some(e) => Some(Box::new(self.parse_sql_expression(	8✔
223	parse_aggregations,	8✔
224	e,	8✔
225	schema,	8✔
226	)?)),	8✔
227	_ => None,	36✔
228	};
229	let typ = trim_where.as_ref().map(\|e\| match e {	44✔
230	TrimWhereField::Both => TrimType::Both,	2✔
231	TrimWhereField::Leading => TrimType::Leading,	2✔
232	TrimWhereField::Trailing => TrimType::Trailing,	2✔
233	});	44✔
234	Ok(Expression::Trim { arg, what, typ })	44✔
235	}	44✔
236
237	fn parse_sql_function(	345✔
238	&mut self,	345✔
239	parse_aggregations: bool,	345✔
240	sql_function: &Function,	345✔
241	schema: &Schema,	345✔
242	) -> Result<Expression, PipelineError> {	345✔
243	let function_name = sql_function.name.to_string().to_lowercase();	345✔
244		345✔
245	#[cfg(feature = "python")]	345✔
246	if function_name.starts_with("py_") {	345✔
247	// The function is from python udf.
248	let udf_name = function_name.strip_prefix("py_").unwrap();	30✔
249	return self.parse_python_udf(udf_name, sql_function, schema);	30✔
250	}	315✔
251		315✔
252	match (	315✔
253	AggregateFunctionType::new(function_name.as_str()),	315✔
254	parse_aggregations,	315✔
255	) {	315✔
256	(Ok(aggr), true) => {	267✔
257	let mut arg_expr: Vec<Expression> = Vec::new();	267✔
258	for arg in &sql_function.args {	537✔
259	let aggregation = self.parse_sql_function_arg(true, arg, schema)?;	270✔
260	arg_expr.push(aggregation);	270✔
261	}
262	let measure = Expression::AggregateFunction {	270✔
263	fun: aggr,	270✔
264	args: arg_expr,	270✔
265	};	270✔
266	let index = match self	270✔
267	.aggregations	270✔
268	.iter()	270✔
269	.enumerate()	270✔
270	.find(\|e\| e.1 == &measure)	270✔
271	{
272	Some((index, _existing)) => index,	9✔
273	_ => {
274	self.aggregations.push(measure);	257✔
275	self.aggregations.len() - 1	257✔
276	}
277	};
278	Ok(Expression::Column {	266✔
279	index: self.offset + index,	266✔
280	})	266✔
281	}
282	(Ok(_agg), false) => Err(InvalidNestedAggregationFunction(function_name)),	×
283	(Err(_), _) => {
284	let mut function_args: Vec<Expression> = Vec::new();	48✔
285	for arg in &sql_function.args {	142✔
286	function_args.push(self.parse_sql_function_arg(	94✔
287	parse_aggregations,	94✔
288	arg,	94✔
289	schema,	94✔
290	)?);	94✔
291	}
292
293	match ScalarFunctionType::new(function_name.as_str()) {	48✔
294	Ok(sft) => Ok(ScalarFunction {	34✔
295	fun: sft,	34✔
296	args: function_args.clone(),	34✔
297	}),	34✔
298	Err(_d) => match GeoFunctionType::new(function_name.as_str()) {	14✔
299	Ok(gft) => Ok(GeoFunction {	14✔
300	fun: gft,	14✔
301	args: function_args.clone(),	14✔
302	}),	14✔
303	Err(_err) => Err(InvalidNestedAggregationFunction(function_name)),	×
304	},
305	}
306	}
307	}
308	}	344✔
309
310	fn parse_sql_function_arg(
311	&mut self,
312	parse_aggregations: bool,
313	argument: &FunctionArg,
314	schema: &Schema,
315	) -> Result<Expression, PipelineError> {
316	match argument {	435✔
317	FunctionArg::Named {
318	name: _,
319	arg: FunctionArgExpr::Expr(arg),	×
320	} => self.parse_sql_expression(parse_aggregations, arg, schema),	×
321	FunctionArg::Named {
322	name: _,
323	arg: FunctionArgExpr::Wildcard,
324	} => Err(InvalidArgument(format!("{argument:?}"))),	×
325	FunctionArg::Unnamed(FunctionArgExpr::Expr(arg)) => {	436✔
326	self.parse_sql_expression(parse_aggregations, arg, schema)	436✔
327	}
328	FunctionArg::Unnamed(FunctionArgExpr::Wildcard) => {
329	Err(InvalidArgument(format!("{argument:?}")))	×
330	}
331	FunctionArg::Named {
332	name: _,
333	arg: FunctionArgExpr::QualifiedWildcard(_),
334	} => Err(InvalidArgument(format!("{argument:?}"))),	×
335	FunctionArg::Unnamed(FunctionArgExpr::QualifiedWildcard(_)) => {
336	Err(InvalidArgument(format!("{argument:?}")))	×
337	}
338	}
339	}	436✔
340
341	fn parse_sql_unary_op(	×
342	&mut self,	×
343	parse_aggregations: bool,	×
344	op: &SqlUnaryOperator,	×
345	expr: &SqlExpr,	×
346	schema: &Schema,	×
347	) -> Result<Expression, PipelineError> {	×
348	let arg = Box::new(self.parse_sql_expression(parse_aggregations, expr, schema)?);	×
349	let operator = match op {	×
350	SqlUnaryOperator::Not => UnaryOperatorType::Not,	×
351	SqlUnaryOperator::Plus => UnaryOperatorType::Plus,	×
352	SqlUnaryOperator::Minus => UnaryOperatorType::Minus,	×
353	_ => return Err(InvalidOperator(format!("{op:?}"))),	×
354	};
355
356	Ok(Expression::UnaryOperator { operator, arg })	×
357	}	×
358
359	fn parse_sql_binary_op(	147✔
360	&mut self,	147✔
361	parse_aggregations: bool,	147✔
362	left: &SqlExpr,	147✔
363	op: &SqlBinaryOperator,	147✔
364	right: &SqlExpr,	147✔
365	schema: &Schema,	147✔
366	) -> Result<Expression, PipelineError> {	147✔
367	let left_op = self.parse_sql_expression(parse_aggregations, left, schema)?;	147✔
368	let right_op = self.parse_sql_expression(parse_aggregations, right, schema)?;	147✔
369
370	let operator = match op {	147✔
371	SqlBinaryOperator::Gt => BinaryOperatorType::Gt,	36✔
372	SqlBinaryOperator::GtEq => BinaryOperatorType::Gte,	1✔
373	SqlBinaryOperator::Lt => BinaryOperatorType::Lt,	22✔
374	SqlBinaryOperator::LtEq => BinaryOperatorType::Lte,	20✔
375	SqlBinaryOperator::Eq => BinaryOperatorType::Eq,	30✔
376	SqlBinaryOperator::NotEq => BinaryOperatorType::Ne,	×
377	SqlBinaryOperator::Plus => BinaryOperatorType::Add,	4✔
378	SqlBinaryOperator::Minus => BinaryOperatorType::Sub,	2✔
379	SqlBinaryOperator::Multiply => BinaryOperatorType::Mul,	×
380	SqlBinaryOperator::Divide => BinaryOperatorType::Div,	×
381	SqlBinaryOperator::Modulo => BinaryOperatorType::Mod,	×
382	SqlBinaryOperator::And => BinaryOperatorType::And,	22✔
383	SqlBinaryOperator::Or => BinaryOperatorType::Or,	10✔
384	_ => return Err(InvalidOperator(format!("{op:?}"))),	×
385	};
386
387	Ok(Expression::BinaryOperator {	147✔
388	left: Box::new(left_op),	147✔
389	operator,	147✔
390	right: Box::new(right_op),	147✔
391	})	147✔
392	}	147✔
393
394	fn parse_sql_number(n: &str) -> Result<Expression, PipelineError> {	87✔
395	match n.parse::<i64>() {	87✔
396	Ok(n) => Ok(Expression::Literal(Field::Int(n))),	87✔
397	Err(_) => match n.parse::<f64>() {	×
398	Ok(f) => Ok(Expression::Literal(Field::Float(OrderedFloat(f)))),	×
399	Err(_) => Err(InvalidValue(n.to_string())),	×
400	},
401	}
402	}	87✔
403
404	fn parse_sql_like_operator(	×
405	&mut self,	×
406	parse_aggregations: bool,	×
407	negated: &bool,	×
408	expr: &Expr,	×
409	pattern: &Expr,	×
410	escape_char: &Option<char>,	×
411	schema: &Schema,	×
412	) -> Result<Expression, PipelineError> {	×
413	let arg = self.parse_sql_expression(parse_aggregations, expr, schema)?;	×
414	let pattern = self.parse_sql_expression(parse_aggregations, pattern, schema)?;	×
415	let like_expression = Expression::Like {	×
416	arg: Box::new(arg),	×
417	pattern: Box::new(pattern),	×
418	escape: *escape_char,	×
419	};	×
420	if *negated {	×
421	Ok(Expression::UnaryOperator {	×
422	operator: UnaryOperatorType::Not,	×
423	arg: Box::new(like_expression),	×
424	})	×
425	} else {
426	Ok(like_expression)	×
427	}
428	}	×
429
430	fn parse_sql_extract_operator(	64✔
431	&mut self,	64✔
432	parse_aggregations: bool,	64✔
433	field: &sqlparser::ast::DateTimeField,	64✔
434	expr: &Expr,	64✔
435	schema: &Schema,	64✔
436	) -> Result<Expression, PipelineError> {	64✔
437	let right = self.parse_sql_expression(parse_aggregations, expr, schema)?;	64✔
438	Ok(Expression::DateTimeFunction {	64✔
439	fun: DateTimeFunctionType::Extract { field: *field },	64✔
440	arg: Box::new(right),	64✔
441	})	64✔
442	}	64✔
443
444	fn parse_sql_cast_operator(	65✔
445	&mut self,	65✔
446	parse_aggregations: bool,	65✔
447	expr: &Expr,	65✔
448	data_type: &DataType,	65✔
449	schema: &Schema,	65✔
450	) -> Result<Expression, PipelineError> {	65✔
451	let expression = self.parse_sql_expression(parse_aggregations, expr, schema)?;	65✔
452	let cast_to = match data_type {	65✔
453	DataType::Decimal(_) => CastOperatorType::Decimal,	×
454	DataType::Binary(_) => CastOperatorType::Binary,	×
455	DataType::Float(_) => CastOperatorType::Float,	10✔
456	DataType::Int(_) => CastOperatorType::Int,	6✔
457	DataType::Integer(_) => CastOperatorType::Int,	×
458	DataType::UnsignedInt(_) => CastOperatorType::UInt,	×
459	DataType::UnsignedInteger(_) => CastOperatorType::UInt,	×
460	DataType::Boolean => CastOperatorType::Boolean,	12✔
461	DataType::Date => CastOperatorType::Date,	×
462	DataType::Timestamp(..) => CastOperatorType::Timestamp,	×
463	DataType::Text => CastOperatorType::Text,	18✔
464	DataType::String => CastOperatorType::String,	19✔
465	DataType::Custom(name, ..) => {	×
466	if name.to_string().to_lowercase() == "bson" {	×
467	CastOperatorType::Bson	×
468	} else {
469	Err(PipelineError::InvalidFunction(format!(	×
470	"Unsupported Cast type {name}"	×
471	)))?	×
472	}
473	}
474	_ => Err(PipelineError::InvalidFunction(format!(	×
475	"Unsupported Cast type {data_type}"	×
476	)))?,	×
477	};
478	Ok(Expression::Cast {	65✔
479	arg: Box::new(expression),	65✔
480	typ: cast_to,	65✔
481	})	65✔
482	}	65✔
483
484	fn parse_sql_string(s: &str) -> Result<Expression, PipelineError> {	78✔
485	Ok(Expression::Literal(Field::String(s.to_owned())))	78✔
486	}	78✔
487
488	pub fn fullname_from_ident(ident: &[Ident]) -> String {	130✔
489	let mut ident_tokens = vec![];	130✔
490	for token in ident.iter() {	130✔
491	ident_tokens.push(token.value.clone());	130✔
492	}	130✔
493	ident_tokens.join(".")	130✔
494	}	130✔
495
496	pub(crate) fn normalize_ident(id: &Ident) -> String {	842✔
497	match id.quote_style {	842✔
498	Some(_) => id.value.clone(),	×
499	None => id.value.clone(),	842✔
500	}
501	}	842✔
502
503	#[cfg(feature = "python")]
504	fn parse_python_udf(	30✔
505	&mut self,	30✔
506	name: &str,	30✔
507	function: &Function,	30✔
508	schema: &Schema,	30✔
509	) -> Result<Expression, PipelineError> {	30✔
510	// First, get python function define by name.
511	// Then, transfer python function to Expression::PythonUDF
512
513	use dozer_types::types::FieldType;
514	use PipelineError::InvalidQuery;
515
516	let args = function	30✔
517	.args	30✔
518	.iter()	30✔
519	.map(\|argument\| self.parse_sql_function_arg(false, argument, schema))	75✔
520	.collect::<Result<Vec<_>, PipelineError>>()?;	30✔
521
522	let last_arg = args	30✔
523	.last()	30✔
524	.ok_or_else(\|\| InvalidQuery("Can't get python udf return type".to_string()))?;	30✔
525
526	let return_type = match last_arg {	30✔
527	Expression::Literal(Field::String(s)) => {	30✔
528	FieldType::try_from(s.as_str()).map_err(\|e\| InvalidQuery(format!("Failed to parse Python UDF return type: {e}")))?	30✔
529	}
530	_ => return Err(InvalidArgument("The last arg for python udf should be a string literal, which represents return type".to_string())),	×
531	};
532
533	Ok(Expression::PythonUDF {	30✔
534	name: name.to_string(),	30✔
535	args,	30✔
536	return_type,	30✔
537	})	30✔
538	}	30✔
539	}
540
541	#[derive(Debug, Clone, Hash, PartialEq, Eq)]	5,346✔
542	pub struct NameOrAlias(pub String, pub Option<String>);
543
544	pub fn extend_schema_source_def(schema: &Schema, name: &NameOrAlias) -> Schema {	1,244✔
545	let mut output_schema = schema.clone();	1,244✔
546	let mut fields = vec![];	1,244✔
547	for mut field in schema.clone().fields.into_iter() {	5,300✔
548	if let Some(alias) = &name.1 {	5,300✔
549	field.source = SourceDefinition::Alias {	1,490✔
550	name: alias.to_string(),	1,490✔
551	};	1,490✔
552	}	3,810✔
553
554	fields.push(field);	5,300✔
555	}
556	output_schema.fields = fields;	1,244✔
557		1,244✔
558	output_schema	1,244✔
559	}	1,244✔

getdozer / dozer / 4423198119

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