Merge branch 'main' into server-work

minisql/src/schema.rs

@@ -12,8 +12,8 @@ use std::collections::HashMap;
 pub struct TableSchema {
     table_name: TableName, // used for descriptive errors
     primary_key: ColumnPosition,
-    pub column_name_position_mapping: BiMap<ColumnName, ColumnPosition>,
+    column_name_position_mapping: BiMap<ColumnName, ColumnPosition>,
-    pub types: Vec<DbType>,
+    types: Vec<DbType>,
 }
 
 pub type TableName = String;
@@ -36,6 +36,24 @@ impl TableSchema {
         self.types[column_position]
     }
 
+    pub fn get_columns(&self) -> Vec<&ColumnName> {
+        self.column_name_position_mapping.iter().map(|(name, _)| name).collect()
+    }
+
+    pub fn does_column_exist(&self, column_name: &ColumnName) -> bool {
+        self.column_name_position_mapping.contains_left(column_name)
+    }
+
+    pub fn get_column_position(&self, column_name: &ColumnName) -> Option<ColumnPosition> {
+        self.column_name_position_mapping.get_by_left(column_name).copied()
+    }
+
+    pub fn get_type_at(&self, column_name: &ColumnName) -> Option<DbType> {
+        let position = self.get_column_position(column_name)?;
+        self.types.get(position).copied()
+    }
+
+    // TODO: Get rid of this after validation is merged
     fn get_column(&self, column_name: &ColumnName) -> DbResult<(DbType, ColumnPosition)> {
         match self.column_name_position_mapping.get_by_left(column_name) {
             Some(column_position) => match self.types.get(*column_position) {
@@ -52,6 +70,7 @@ impl TableSchema {
         }
     }
 
+    // TODO: Get rid of this after validation is merged
     pub fn column_position_from_column_name(
         &self,
         column_name: &ColumnName,

parser/src/core.rs

@@ -1,12 +1,15 @@
 use minisql::{operation::Operation, schema::TableSchema};
 use nom::{branch::alt, multi::many0, IResult};
+use thiserror::Error;
 
-use crate::{create::parse_create, delete::parse_delete, index::parse_create_index, insert::parse_insert, select::parse_select, validation::{validate_operation, ValidationError}};
+use crate::{parsing::{create::parse_create, delete::parse_delete, index::parse_create_index, insert::parse_insert, select::parse_select}, validation::{validate_operation, ValidationError}};
 
-#[derive(Debug)]
+#[derive(Debug, Error)]
 pub enum Error {
+    #[error("parsing error: {0}")]
     ParsingError(String),
-    ValidationError(ValidationError)
+    #[error("validation error: {0}")]
+    ValidationError(#[from] ValidationError)
 }
 
 pub fn parse_statement<'a>(input: &'a str) -> IResult<&str, Operation> {
@@ -31,7 +34,7 @@ pub fn parse_and_validate(query: String, db_metadata: &Vec<(String, &TableSchema
             Error::ParsingError(err.to_string())
         })?;
 
-    validate_operation(&op, db_metadata).map_err(|err| Error::ValidationError(err))?;
+    validate_operation(&op, db_metadata)?;
     Ok(op)
 }
 

parser/src/lib.rs

@@ -1,11 +1,5 @@
 
-mod literal;
+mod parsing;
-mod select;
-mod common;
-mod create;
-mod insert;
-mod delete;
-mod index;
 mod validation;
 mod core;
 

parser/src/parsing/common.rs

@@ -8,7 +8,7 @@ use nom::{
 };
 use minisql::{operation::Condition, type_system::DbType};
 
-use crate::literal::parse_db_value;
+use super::literal::parse_db_value;
 
 pub fn parse_table_name(input: &str) -> IResult<&str, &str> {
     alt((
@@ -32,12 +32,12 @@ pub fn parse_column_name(input: &str) -> IResult<&str, String> {
 }
 
 pub fn parse_db_type(input: &str) -> IResult<&str, DbType> {
-    let (input, type_name) = alt((tag("STRING"), tag("INT"), tag("Float"), tag("UUID")))(input)?;
+    let (input, type_name) = alt((tag("STRING"), tag("INT"), tag("NUMBER"), tag("UUID")))(input)?;
     let db_type = match type_name {
         "STRING" => DbType::String,
         "INT" => DbType::Int,
         "UUID" => DbType::Uuid,
-        "Float" => DbType::Number,
+        "NUMBER" => DbType::Number,
         _ => return Err(nom::Err::Failure(make_error(input, nom::error::ErrorKind::IsNot)))
     };
     Ok((input, db_type))
@@ -68,7 +68,7 @@ fn parse_equality(input: &str) -> IResult<&str, Condition> {
 #[cfg(test)]
 mod tests {
     use minisql::{operation::Condition, type_system::DbType};
-    use crate::common::{parse_db_type, parse_equality};
+    use crate::parsing::common::{parse_db_type, parse_equality};
     
     #[test]
     fn test_parse_equality() {

parser/src/parsing/create.rs

@@ -7,7 +7,7 @@ use nom::{
     IResult, combinator::opt,
 };
 
-use crate::common::{parse_table_name, parse_identifier, parse_db_type};
+use super::common::{parse_table_name, parse_identifier, parse_db_type};
 
 pub fn parse_create(input: &str) -> IResult<&str, Operation> {
     let (input, _) = tag("CREATE")(input)?;
@@ -69,7 +69,7 @@ pub fn parse_column_definition(input: &str) -> IResult<&str, (ColumnName, DbType
 #[cfg(test)]
 mod tests {
     use minisql::operation::Operation;
-    use crate::create::parse_create;
+    use crate::parsing::create::parse_create;
 
     #[test]
     fn test_parse_create_no_spaces() {

parser/src/parsing/delete.rs

@@ -5,7 +5,7 @@ use nom::{
     IResult,
 };
 
-use crate::common::{parse_table_name, parse_condition};
+use super::common::{parse_table_name, parse_condition};
 
 pub fn parse_delete(input: &str) -> IResult<&str, Operation> {
     let (input, _) = tag("DELETE")(input)?;
@@ -26,7 +26,7 @@ pub fn parse_delete(input: &str) -> IResult<&str, Operation> {
 #[cfg(test)]
 mod tests {
     use minisql::operation::Operation;
-    use crate::delete::parse_delete;
+    use crate::parsing::delete::parse_delete;
 
     #[test]
     fn test_parse_delete() {

parser/src/parsing/index.rs

@@ -5,7 +5,7 @@ use nom::{
     IResult, combinator::opt,
 };
 
-use crate::common::{parse_identifier, parse_table_name};
+use super::common::{parse_identifier, parse_table_name};
 
 pub fn parse_create_index(input: &str) -> IResult<&str, Operation> {
     let (input, _) = tag("CREATE")(input)?;
@@ -39,7 +39,7 @@ pub fn parse_create_index(input: &str) -> IResult<&str, Operation> {
 #[cfg(test)]
 mod tests {
     use minisql::operation::Operation;
-    use crate::index::parse_create_index;
+    use crate::parsing::index::parse_create_index;
 
         
     #[test]

parser/src/parsing/insert.rs

@@ -1,4 +1,4 @@
-use crate::{literal::parse_db_value, common::{parse_table_name, parse_identifier}};
+use super::{literal::parse_db_value, common::{parse_table_name, parse_identifier}};
 use minisql::{operation::Operation, type_system::Value};
 use nom::{
     bytes::complete::tag,

parser/src/parsing/literal.rs

@@ -100,17 +100,17 @@ pub fn parse_string(input: &str) -> IResult<&str, Value> {
     Ok((input, Value::Indexable(IndexableValue::String(value))))
 }
 
-fn parse_uuid(input: &str) -> IResult<&str, Value> {
+pub fn parse_uuid(input: &str) -> IResult<&str, Value> {
-    // TODO: make it actually uuid
+    let (input, value) = pair(char('u'), u64)(input).map(|(input, (_, v))| {
-    u64(input).map(|(input, v)| {
         (input, Value::Indexable(IndexableValue::Uuid(v)))
-    })
+    })?;
+    Ok((input, value))
 }
 
 #[cfg(test)]
 mod tests {
     use minisql::type_system::{IndexableValue, Value};
-    use crate::literal::{parse_db_value, parse_string};
+    use crate::parsing::literal::{parse_db_value, parse_string, parse_uuid};
 
 
     #[test]
@@ -161,4 +161,9 @@ mod tests {
     fn test_parse_int() {
         assert_eq!(parse_db_value("5134616"), Ok(("", Value::Indexable(IndexableValue::Int(5134616)))));
     }
+
+    #[test]
+    fn test_parse_uuid() {
+        assert_eq!(parse_uuid("u131515"), Ok(("", Value::Indexable(IndexableValue::Uuid(131515)))))
+    }
 }

parser/src/parsing/mod.rs

@@ -0,0 +1,7 @@
+pub(crate) mod literal;
+pub(crate) mod select;
+pub(crate) mod common;
+pub(crate) mod create;
+pub(crate) mod insert;
+pub(crate) mod delete;
+pub(crate) mod index;

parser/src/parsing/select.rs

@@ -1,4 +1,4 @@
-use crate::common::{parse_table_name, parse_column_name, parse_condition};
+use super::common::{parse_table_name, parse_column_name, parse_condition};
 use minisql::operation::{ColumnSelection, Operation};
 use nom::{
     branch::alt,
@@ -45,7 +45,7 @@ pub fn try_parse_column_selection(input: &str) -> IResult<&str, ColumnSelection>
 #[cfg(test)]
 mod tests {
     use minisql::operation::{ColumnSelection, Operation};
-    use crate::{common::{parse_column_name, parse_table_name}, select::parse_select};
+    use crate::parsing::{common::{parse_column_name, parse_table_name}, select::parse_select};
 
         
     #[test]

parser/src/validation.rs

@@ -1,98 +1,111 @@
 
 use std::collections::HashSet;
+use thiserror::Error;
 
-use minisql::{operation::{ColumnSelection, Condition, InsertionValues, Operation}, schema::TableSchema, type_system::{DbType, IndexableValue, Value}};
+use minisql::{operation::{ColumnSelection, Condition, InsertionValues, Operation}, schema::{TableSchema, ColumnName, TableName}, type_system::DbType};
 
 
-#[derive(Debug)]
+#[derive(Debug, Error)]
 pub enum ValidationError {
-    TableDoesNotExist(String),
+    #[error("table {0} does not exist")]
-    TableExists(String),
+    TableDoesNotExist(TableName),
-    ColumnDoesNotExist(String),
+    #[error("table {0} already exists")]
-    BadColumnPosition(usize),
+    TableAlreadyExists(TableName),
-    DuplicateColumn(String),
+    #[error("columns {0:?} do not exist")]
-    TypeMismatch,
+    ColumnsDoNotExist(Vec<ColumnName>),
-    ValueForRequiredColumnIsMissing(String)
+    #[error("duplicate column {0}")]
+    DuplicateColumn(ColumnName),
+    #[error("type mismatch at column `{column_name:?}` (expected {expected_type:?}, found {received_type:?})")]
+    TypeMismatch {
+        column_name: ColumnName,
+        received_type: DbType,
+        expected_type: DbType,
+    },
+    #[error("values for required columns {0:?} are missing")]
+    RequiredColumnsAreMissing(Vec<ColumnName>)
 }
 
-pub fn type_of(value: &Value) -> DbType {
+pub type DbSchema<'a> = Vec<(TableName, &'a TableSchema)>;
-    match value {
-        Value::Indexable(IndexableValue::Int(_)) => DbType::Int,
-        Value::Indexable(IndexableValue::String(_)) => DbType::String,
-        Value::Number(_) => DbType::Number,
-        Value::Indexable(IndexableValue::Uuid(_)) => DbType::Uuid
-    }
-}
 
 /// Validates the operation based on db_metadata
-pub fn validate_operation(operation: &Operation, db_metadata: &Vec<(String, &TableSchema)>) -> Result<(), ValidationError> {
+pub fn validate_operation(operation: &Operation, db_schema: &DbSchema) -> Result<(), ValidationError> {
     match operation {
         Operation::Select(table_name, column_selection, condition) => {
-            validate_select(table_name, column_selection, condition, db_metadata)?;
+            validate_select(table_name, column_selection, condition, db_schema)?;
         },
         Operation::Insert(table_name, insertion_values) => {
-            validate_insert(&table_name, insertion_values, db_metadata)?;
+            validate_insert(&table_name, insertion_values, db_schema)?;
         },
         Operation::Delete(table_name, condition) => {
-            validate_delete(table_name, condition, db_metadata)?;
+            validate_delete(table_name, condition, db_schema)?;
         },
         // Operation::Update(table_name, insertion_values, condition) => {
         //     validate_update(table_name, insertion_values, db_metadata)?;
         // },
         Operation::CreateTable(table_name, schema) => {
-            validate_create(table_name, schema, db_metadata)?;
+            validate_create(table_name, schema, db_schema)?;
         },
         Operation::CreateIndex(table_name, column_name) => {
-            validate_create_index(table_name, column_name, db_metadata)?;
+            validate_create_index(table_name, column_name, db_schema)?;
         },
         // Operation::DropTable(table_name) => {
-        //     validate_drop(table_name, db_metadata)?;
+        //     validate_drop(table_name, db_schema)?;
         // }
     }
     Ok(())
 }
 
+fn validate_table_exists<'a>(db_schema: &DbSchema<'a>, table_name: &'a TableName) -> Result<&'a TableSchema, ValidationError> {
+    db_schema.iter().find(|(tname, _)| table_name.eq(tname))
+        .ok_or(ValidationError::TableDoesNotExist(table_name.to_string()))
+        .map(|(_, table_schema)| table_schema).copied()
+}
+
+
 // pub fn validate_drop(table_name: &str, db_metadata: &Vec<(String, TableSchema)>) -> Result<(), ValidationError> {
 //     db_metadata.iter().find(|(tname, _)| table_name.eq(tname))
 //         .ok_or(ValidationError::TableDoesNotExist(table_name.to_string()))?;
 //     Ok(())
 // }
 
-pub fn validate_create(table_name: &str, schema: &TableSchema, db_metadata: &Vec<(String, &TableSchema)>) -> Result<(), ValidationError> {
+pub fn validate_create(table_name: &TableName, schema: &TableSchema, db_schema: &DbSchema) -> Result<(), ValidationError> {
-    if db_metadata.iter().find(|(tname, _)| table_name.eq(tname)).is_some() {
+    if let Some(_) = get_table_schema(db_schema, table_name) {
-        return Err(ValidationError::TableExists(table_name.to_string()));
+        return Err(ValidationError::TableAlreadyExists(table_name.to_string()));
-    }
-    let mut column_names = HashSet::new();
-    for (name, _) in &schema.column_name_position_mapping {
-        if column_names.contains(name) {
-            return Err(ValidationError::DuplicateColumn(name.clone()));
-        } else {
-            column_names.insert(name.clone());
-        }
     }
 
+    find_first_duplicate(&schema.get_columns())
+        .map_or_else(
+            || Ok(()),
+            |duplicate_column| Err(ValidationError::DuplicateColumn(duplicate_column.to_string()))
+        )?;
+
     // TODO: Ensure it has a primary key??
     Ok(())
 }
 
-pub fn validate_select(table_name: &str, column_selection: &ColumnSelection, condition: &Option<Condition>, db_metadata: &Vec<(String, &TableSchema)>) -> Result<(), ValidationError> {
+pub fn validate_select(table_name: &TableName, column_selection: &ColumnSelection, condition: &Option<Condition>, db_schema: &Vec<(TableName, &TableSchema)>) -> Result<(), ValidationError> {
-    let (_, schema) = db_metadata.iter().find(|(tname, _)| table_name.eq(tname))
+    let schema = validate_table_exists(db_schema, table_name)?;
-        .ok_or(ValidationError::TableDoesNotExist(table_name.to_string()))?;
     match column_selection {
         ColumnSelection::Columns(columns) => {
-            columns.iter().find(|c| {
+            let non_existant_columns: Vec<String> =
-                !schema.column_name_position_mapping.contains_left(*c)
+                columns.iter().filter_map(|column|
-            }).map_or_else(||Ok(()), |c| Err(ValidationError::ColumnDoesNotExist(c.to_string())))?;
+                    if schema.does_column_exist(&column) {
+                        Some(column.clone())
+                    } else {
+                        None
+                    }).collect();
+            if non_existant_columns.len() > 0 {
+                Err(ValidationError::ColumnsDoNotExist(non_existant_columns))
+            } else {
+                validate_condition(condition, schema)
             }
-        _ => {}
         }
-    validate_condition(condition, schema)?;
+        ColumnSelection::All => Ok(())
-    Ok(())
+    }
 }
 
 // pub fn validate_update(table_name: &str, insertion_values: &InsertionValues, db_metadata: &Vec<(String, TableSchema)>) -> Result<(), ValidationError> {
-//     let (_, schema) = db_metadata.iter().find(|(tname, _)| table_name.eq(tname))
+//     let schema = validate_table_exists(db_schema, table_name)?;
-//         .ok_or(ValidationError::TableDoesNotExist(table_name.to_string()))?;
 //     let mut column_names = HashSet::new();
 //     // Find duplicate columns
 //     for (name, _) in insertion_values {
@@ -111,7 +124,7 @@ pub fn validate_select(table_name: &str, column_selection: &ColumnSelection, con
 //     if let Some((name, _, _)) = column_value_type.iter().find(|(_, _, t)| {
 //         t.is_none()
 //     }) {
-//         return Err(ValidationError::ColumnDoesNotExist((*name).clone()));
+//         return Err(ValidationError::ColumnsDoNotExist(vec![(*name).clone())]);
 //     }
 
 //     // Check types
@@ -128,63 +141,56 @@ pub fn validate_select(table_name: &str, column_selection: &ColumnSelection, con
 //     Ok(())
 // }
 
-pub fn validate_insert(table_name: &str, insertion_values: &InsertionValues, db_metadata: &Vec<(String, &TableSchema)>) -> Result<(), ValidationError> {
+pub fn validate_insert(table_name: &TableName, insertion_values: &InsertionValues, db_schema: &DbSchema) -> Result<(), ValidationError> {
-    let (_, schema) = db_metadata.iter().find(|(tname, _)| table_name.eq(tname))
+    let schema = validate_table_exists(db_schema, table_name)?;
-        .ok_or(ValidationError::TableDoesNotExist(table_name.to_string()))?;
+
-    let inserted_columns: HashSet<String> = HashSet::from_iter(insertion_values.iter().map(|(name, _)| name.clone()));
+    // Check for duplicate columns in insertion_values.
-    // TODO: primary key is not required
+    let columns_in_query_vec: Vec<&ColumnName> = insertion_values.iter().map(|(column_name, _)| column_name).collect();
-    for (column_name, _) in &schema.column_name_position_mapping {
+    find_first_duplicate(&columns_in_query_vec)
-        if !inserted_columns.contains(column_name) {
+        .map_or_else(
-            return Err(ValidationError::ValueForRequiredColumnIsMissing(column_name.clone()))
+            || Ok(()),
+            |duplicate_column| Err(ValidationError::DuplicateColumn(duplicate_column.to_string()))
+        )?;
+
+    // Check that the set of columns in the insertion_values is the same as the set of required columns of the table.
+    let columns_in_query: HashSet<&ColumnName> = HashSet::from_iter(columns_in_query_vec);
+    let columns_in_schema: HashSet<&ColumnName> = HashSet::from_iter(schema.get_columns());
+    let non_existant_columns = Vec::from_iter(columns_in_query.difference(&columns_in_schema));
+    if non_existant_columns.len() > 0 {
+        return Err(ValidationError::ColumnsDoNotExist(non_existant_columns.iter().map(|str| str.to_string()).collect()));
     }
-    }
+    let missing_required_columns = Vec::from_iter(columns_in_schema.difference(&columns_in_query));
-    // Ensure columns exist in schema
+    if missing_required_columns.len() > 0 {
-    let column_value_type: Vec<_> = insertion_values.iter().map(|(column, value)| {
+        return Err(ValidationError::RequiredColumnsAreMissing(missing_required_columns.iter().map(|str| str.to_string()).collect()));
-        (column, value, schema.column_name_position_mapping.iter().find(|(name, _) | {
-            (*name).eq(column)
-        }).map(|(_, t)| schema.types.get(*t as usize)))
-    }).collect();
-    if let Some((name, _, _)) = column_value_type.iter().find(|(_, _, t)| {
-        match t {
-            Some(Some(_)) => false,
-            _ => true
-        }
-    }) {
-        return Err(ValidationError::ColumnDoesNotExist((*name).clone()));
     }
 
     // Check types
-    if let Some((_, _, _)) = column_value_type.iter().find(|(_, value, t)| {
+    for (column_name, value) in insertion_values {
-        if let Some(Some(t)) = t {
+        let expected_type = schema.get_type_at(column_name).ok_or(ValidationError::ColumnsDoNotExist(vec![column_name.to_string()]))?; // By the previous validation steps this is never gonna trigger an error.
-            !type_of(value).eq(t)
+        let value_type = value.to_type();
-        } else {
+        if value_type != expected_type {
-            false
+            return Err(ValidationError::TypeMismatch { column_name: column_name.to_string(), received_type: value_type, expected_type });
         }
-    }) {
-        // TODO: Add column name information
-        return Err(ValidationError::TypeMismatch);
     }
+
     Ok(())
 }
 
-pub fn validate_delete(table_name: &str, condition: &Option<Condition>, db_metadata: &Vec<(String, &TableSchema)>) -> Result<(), ValidationError> {
+pub fn validate_delete(table_name: &TableName, condition: &Option<Condition>, db_schema: &DbSchema) -> Result<(), ValidationError> {
-    let (_, schema) = db_metadata.iter().find(|(tname, _)| table_name.eq(tname))
+    let schema = validate_table_exists(db_schema, table_name)?;
-        .ok_or(ValidationError::TableDoesNotExist(table_name.to_string()))?;
     validate_condition(condition, schema)?;
     Ok(())
 }
 
 fn validate_condition(condition: &Option<Condition>, schema: &TableSchema) -> Result<(), ValidationError> {
     match condition {
-        Some(c) => {
+        Some(condition) => {
-            match c {
+            match condition {
-                Condition::Eq(left, right) => {
+                Condition::Eq(column_name, value) => {
-                    let position = schema.column_name_position_mapping.get_by_left(left)
+                    let expected_type: DbType = schema.get_type_at(column_name).ok_or(ValidationError::ColumnsDoNotExist(vec![column_name.to_string()]))?;
-                        .ok_or(ValidationError::ColumnDoesNotExist(left.clone()))?;
+                    let value_type: DbType = value.to_type();
-                    let column_type = schema.types.get(*position as usize)
+                    if !expected_type.eq(&value_type) {
-                        .ok_or(ValidationError::BadColumnPosition(*position))?;
+                        return Err(ValidationError::TypeMismatch { column_name: column_name.to_string(), received_type: value_type, expected_type });
-                    if !column_type.eq(&type_of(right)) {
-                        return Err(ValidationError::TypeMismatch);
                     }
                 }
             }
@@ -194,13 +200,31 @@ fn validate_condition(condition: &Option<Condition>, schema: &TableSchema) -> Re
     Ok(())
 }
 
-fn validate_create_index(table_name: &str, column_name: &str, db_metadata: &Vec<(String, &TableSchema)>) -> Result<(), ValidationError> {
+fn validate_create_index(table_name: &TableName, column_name: &ColumnName, db_schema: &DbSchema) -> Result<(), ValidationError> {
-    // Ensure table exists
+    let schema = validate_table_exists(db_schema, table_name)?;
-    let (_, schema) = db_metadata.iter().find(|(tname, _)| table_name.eq(tname))
+    if schema.does_column_exist(column_name) {
-        .ok_or(ValidationError::TableDoesNotExist(table_name.to_string()))?;
-    // Ensure column exists
-    if !schema.column_name_position_mapping.contains_left(column_name) {
-        return Err(ValidationError::ColumnDoesNotExist(column_name.to_string()));
-    }
         Ok(())
+    } else {
+        Err(ValidationError::ColumnsDoNotExist(vec![column_name.to_string()]))
+    }
+}
+
+// ===Helpers===
+fn find_first_duplicate<A>(xs: &[A]) -> Option<&A> 
+where A: Eq + std::hash::Hash
+{
+    let mut already_seen_elements: HashSet<&A> = HashSet::new();
+    for x in xs {
+        if already_seen_elements.contains(x) {
+            return Some(x);
+        } else {
+            already_seen_elements.insert(&x);
+        }
+    }
+    None
+}
+
+fn get_table_schema<'a>(db_schema: &DbSchema<'a>, table_name: &'a TableName) -> Option<&'a TableSchema> {
+    let (_, table_schema) = db_schema.iter().find(|(tname, _)| table_name.eq(tname))?;
+    Some(table_schema)
 }