Compare commits
6 commits
2dead7de0a
...
33c4edf91d
| Author | SHA1 | Date | |
|---|---|---|---|
| 33c4edf91d | |||
| 64d93e9a27 | |||
| 567aa31c07 | |||
| e78511f692 | |||
| 28cb288eaf | |||
| 315703d46b |
18 changed files with 647 additions and 80 deletions
15
grammar.ebnf
15
grammar.ebnf
|
|
@ -1,15 +1,26 @@
|
||||||
|
/* token is first stage of parsing */
|
||||||
token ::= insert
|
token ::= insert
|
||||||
| select
|
| select
|
||||||
| meta-command
|
| meta-command
|
||||||
| int
|
| int
|
||||||
| string
|
| string
|
||||||
|
| semicolon
|
||||||
| end-of-file
|
| end-of-file
|
||||||
|
|
||||||
|
/* command is second stage of parsing */
|
||||||
|
command ::= cmd-insert semicolon
|
||||||
|
| cmd-select semicolon
|
||||||
|
cmd-insert ::= insert int string string
|
||||||
|
cmd-select ::= select
|
||||||
|
|
||||||
insert ::= "insert"
|
insert ::= "insert"
|
||||||
select ::= "select"
|
select ::= "select"
|
||||||
|
semicolon ::= ";"
|
||||||
|
|
||||||
meta-command ::= "." "exit"
|
meta-command ::= meta-command-verb end-of-file
|
||||||
| "about"
|
meta-command-verb ::= ".exit"
|
||||||
|
| ".about"
|
||||||
|
| ".version"
|
||||||
|
|
||||||
int ::= sign? digit+
|
int ::= sign? digit+
|
||||||
sign ::= "+"
|
sign ::= "+"
|
||||||
|
|
|
||||||
50
notes.org
50
notes.org
|
|
@ -226,17 +226,53 @@ i will use rustyline, since it seems like the most feature-complete
|
||||||
|
|
||||||
* DONE write a proper grammar
|
* DONE write a proper grammar
|
||||||
|
|
||||||
* TODO .about meta-command
|
* DONE .about meta-command
|
||||||
* TODO .version meta-command
|
* DONE .version meta-command
|
||||||
* TODO .license meta-command
|
* TODO .license meta-command
|
||||||
* TODO .help meta-command
|
* TODO .help meta-command
|
||||||
|
|
||||||
* TODO parse insert statements in the form
|
* DONE parse insert statements in the form
|
||||||
insert <id:int> <username:string> <email:string>
|
insert <id:int> <username:string> <email:string>
|
||||||
** TODO Row struct
|
** DONE parse row insert
|
||||||
** TODO parse row insert
|
* DONE separate statements with semicolons
|
||||||
** TODO serialize/deserialize row to/from raw bytes
|
* DONE this error message could be better
|
||||||
*** TODO look for best practices for creating binary formats
|
#+begin example
|
||||||
|
Error: unexpected token
|
||||||
|
╭─[ <stdin>:1:24 ]
|
||||||
|
│
|
||||||
|
1 │ insert 0 "user" "email"
|
||||||
|
│ │
|
||||||
|
│ ╰─ found end of file ""
|
||||||
|
│
|
||||||
|
│ Note: expected token type to be one of ["semicolon"]
|
||||||
|
───╯
|
||||||
|
#+end example
|
||||||
|
** plan
|
||||||
|
1. Create an example mapping system
|
||||||
|
- Define a mapping of token types to example values
|
||||||
|
- Example: "integer" → "42", "string" → "example", "semicolon" → ";"
|
||||||
|
2. Enhance CommandParseError
|
||||||
|
- Add a method to generate user-friendly error messages
|
||||||
|
- Include both the expected token type and concrete examples
|
||||||
|
3. Implementation approach
|
||||||
|
- Create a static lookup table or function that returns examples
|
||||||
|
- Extend existing error handling to include examples in messages
|
||||||
|
- Make sure the examples follow SQL syntax conventions
|
||||||
|
4. Error display refinement
|
||||||
|
- Update error_display.rs to include these examples
|
||||||
|
- Format error messages to show both what was expected and example syntax
|
||||||
|
5. Testing
|
||||||
|
- Add tests that verify the error messages include helpful examples
|
||||||
|
- Ensure examples are contextually appropriate
|
||||||
|
|
||||||
|
This will make errors like "expected semicolon" more helpful by showing "expected semicolon (example: ;)".
|
||||||
|
* DONE correct all instances of <unknown> in locations
|
||||||
|
* DONE meta-commands must be followed by end-of-file
|
||||||
|
* TODO project code documentation
|
||||||
|
* TODO project usage documentation
|
||||||
|
* DONE in case of parse error, skip until next semicolon to better recover
|
||||||
|
* TODO serialize/deserialize row to/from raw bytes
|
||||||
|
** TODO look for best practices for creating binary formats
|
||||||
|
|
||||||
* WAIT cli tests using insta-cmd
|
* WAIT cli tests using insta-cmd
|
||||||
https://insta.rs/docs/cmd/
|
https://insta.rs/docs/cmd/
|
||||||
|
|
|
||||||
|
|
@ -13,12 +13,18 @@ pub fn startup_msg() -> String {
|
||||||
)
|
)
|
||||||
}
|
}
|
||||||
|
|
||||||
pub fn about_msg() -> String {
|
pub fn version_msg() -> String {
|
||||||
let name = env!("CARGO_PKG_NAME");
|
let name = env!("CARGO_PKG_NAME");
|
||||||
let version = env!("CARGO_PKG_VERSION");
|
let version = env!("CARGO_PKG_VERSION");
|
||||||
|
|
||||||
|
format!("{name} v{version}")
|
||||||
|
}
|
||||||
|
|
||||||
|
pub fn about_msg() -> String {
|
||||||
|
let version = version_msg();
|
||||||
|
|
||||||
format!(
|
format!(
|
||||||
"{name} v{version} -- A database engine\n\
|
"{version} -- A database engine\n\
|
||||||
Note: This is experimental software. No maintenance is intendend."
|
Note: This is experimental software. No maintenance is intendend."
|
||||||
)
|
)
|
||||||
}
|
}
|
||||||
|
|
|
||||||
|
|
@ -1,6 +1,6 @@
|
||||||
use crate::meta_commands::{MetaCommand, MetaCommandExecuteResult};
|
use crate::meta_commands::{MetaCommand, MetaCommandExecuteResult};
|
||||||
use crate::statements::{Statement, StatementExecuteResult};
|
use crate::statements::{Statement, StatementExecuteResult};
|
||||||
use crate::tokens::{ScanError, Token};
|
use crate::tokens::{Location, ScanError, Token};
|
||||||
|
|
||||||
#[derive(Debug)]
|
#[derive(Debug)]
|
||||||
pub enum Command {
|
pub enum Command {
|
||||||
|
|
@ -47,10 +47,48 @@ impl Command {
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
|
||||||
|
pub enum ExpectedToken {
|
||||||
|
Integer,
|
||||||
|
String,
|
||||||
|
Semicolon,
|
||||||
|
Statement,
|
||||||
|
MetaCommand,
|
||||||
|
EndOfFile,
|
||||||
|
}
|
||||||
|
|
||||||
|
impl ExpectedToken {
|
||||||
|
/// Returns an example value for this token type
|
||||||
|
pub fn example(&self) -> &'static str {
|
||||||
|
match self {
|
||||||
|
ExpectedToken::Integer => "42",
|
||||||
|
ExpectedToken::String => "\"example\"",
|
||||||
|
ExpectedToken::Semicolon => ";",
|
||||||
|
ExpectedToken::Statement => "select",
|
||||||
|
ExpectedToken::MetaCommand => ".exit",
|
||||||
|
ExpectedToken::EndOfFile => "<end of input>",
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
impl std::fmt::Display for ExpectedToken {
|
||||||
|
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
|
||||||
|
match self {
|
||||||
|
ExpectedToken::Integer => write!(f, "integer"),
|
||||||
|
ExpectedToken::String => write!(f, "string"),
|
||||||
|
ExpectedToken::Semicolon => write!(f, "semicolon"),
|
||||||
|
ExpectedToken::Statement => write!(f, "statement"),
|
||||||
|
ExpectedToken::MetaCommand => write!(f, "meta command"),
|
||||||
|
ExpectedToken::EndOfFile => write!(f, "end of file"),
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
#[derive(Debug)]
|
#[derive(Debug)]
|
||||||
pub enum CommandParseError {
|
pub enum CommandParseError {
|
||||||
Scan(ScanError),
|
Scan(ScanError),
|
||||||
UnexpectedToken(Token, &'static [&'static str]),
|
UnexpectedToken(Token, &'static [ExpectedToken]),
|
||||||
|
UnexpectedEndOfFile(Location, &'static [ExpectedToken]),
|
||||||
}
|
}
|
||||||
|
|
||||||
impl From<MetaCommand> for Command {
|
impl From<MetaCommand> for Command {
|
||||||
|
|
@ -80,7 +118,12 @@ mod tests {
|
||||||
|
|
||||||
#[test]
|
#[test]
|
||||||
fn test_execute_insert_statement() {
|
fn test_execute_insert_statement() {
|
||||||
let statement: Command = Statement::Insert.into();
|
let statement: Command = Statement::Insert {
|
||||||
|
id: 45,
|
||||||
|
username: String::from("user"),
|
||||||
|
email: String::from("user@example.org"),
|
||||||
|
}
|
||||||
|
.into();
|
||||||
let result = statement.execute().display();
|
let result = statement.execute().display();
|
||||||
assert_snapshot!(result);
|
assert_snapshot!(result);
|
||||||
}
|
}
|
||||||
|
|
|
||||||
|
|
@ -1,4 +1,7 @@
|
||||||
use crate::{command::CommandParseError, tokens::ScanError};
|
use crate::{
|
||||||
|
command::{CommandParseError, ExpectedToken},
|
||||||
|
tokens::ScanError,
|
||||||
|
};
|
||||||
use ariadne::{Color, Label, Report, ReportKind, Source};
|
use ariadne::{Color, Label, Report, ReportKind, Source};
|
||||||
|
|
||||||
pub trait OSDBError {
|
pub trait OSDBError {
|
||||||
|
|
@ -11,24 +14,79 @@ impl OSDBError for CommandParseError {
|
||||||
CommandParseError::Scan(x) => {
|
CommandParseError::Scan(x) => {
|
||||||
x.display(file, input);
|
x.display(file, input);
|
||||||
}
|
}
|
||||||
CommandParseError::UnexpectedToken(token, items) => {
|
CommandParseError::UnexpectedToken(token, expected_tokens) => {
|
||||||
let location = (file, Into::<std::ops::Range<usize>>::into(&token.location));
|
let location = (file, Into::<std::ops::Range<usize>>::into(&token.location));
|
||||||
Report::build(ReportKind::Error, location.clone())
|
|
||||||
|
let mut report = Report::build(ReportKind::Error, location.clone())
|
||||||
.with_message("unexpected token")
|
.with_message("unexpected token")
|
||||||
.with_label(
|
.with_label(
|
||||||
Label::new(location.clone())
|
Label::new(location.clone())
|
||||||
.with_color(Color::Red)
|
.with_color(Color::Red)
|
||||||
.with_message(format!("found {token}")),
|
.with_message(format!("found {token}")),
|
||||||
)
|
);
|
||||||
.with_note(format!("expected token type to be one of {items:?}"))
|
|
||||||
.finish()
|
report = add_expected_tokens_to_report(report, expected_tokens);
|
||||||
.eprint((file, Source::from(input)))
|
|
||||||
.unwrap()
|
report.finish().eprint((file, Source::from(input))).unwrap()
|
||||||
|
}
|
||||||
|
CommandParseError::UnexpectedEndOfFile(location, expected_tokens) => {
|
||||||
|
let location = (file, Into::<std::ops::Range<usize>>::into(location));
|
||||||
|
let report = Report::build(ReportKind::Error, location)
|
||||||
|
.with_message("unexpected end of file");
|
||||||
|
let report = add_expected_tokens_to_report(report, expected_tokens);
|
||||||
|
report.finish().eprint((file, Source::from(input))).unwrap()
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
type OSDBReport<'a> = ariadne::ReportBuilder<'a, (&'a str, std::ops::Range<usize>)>;
|
||||||
|
|
||||||
|
fn add_expected_tokens_to_report<'a>(
|
||||||
|
mut report: OSDBReport<'a>,
|
||||||
|
expected_tokens: &'static [ExpectedToken],
|
||||||
|
) -> OSDBReport<'a> {
|
||||||
|
// If we have expected tokens, show an example for the first one
|
||||||
|
if let Some(help) = expected_token_example_msg(expected_tokens) {
|
||||||
|
report = report.with_help(help)
|
||||||
|
}
|
||||||
|
|
||||||
|
// If we have at least one expected token, show a message showing what type was expected
|
||||||
|
if let Some(note) = expected_token_msg(expected_tokens) {
|
||||||
|
report = report.with_note(note);
|
||||||
|
}
|
||||||
|
report
|
||||||
|
}
|
||||||
|
|
||||||
|
fn expected_token_msg(expected_tokens: &'static [ExpectedToken]) -> Option<String> {
|
||||||
|
if !expected_tokens.is_empty() {
|
||||||
|
// Add a note with all expected types
|
||||||
|
let expected_types: Vec<_> = expected_tokens.iter().map(|t| format!("{}", t)).collect();
|
||||||
|
|
||||||
|
// Use singular form when there's only one expected token type
|
||||||
|
Some(match expected_types.as_slice() {
|
||||||
|
[single_type] => {
|
||||||
|
format!("expected: {}", single_type)
|
||||||
|
}
|
||||||
|
_ => {
|
||||||
|
format!("expected one of: {}", expected_types.join(", "))
|
||||||
|
}
|
||||||
|
})
|
||||||
|
} else {
|
||||||
|
// there are no expected tokens
|
||||||
|
None
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
fn expected_token_example_msg(expected_tokens: &'static [ExpectedToken]) -> Option<String> {
|
||||||
|
if let Some(first_expected) = expected_tokens.first() {
|
||||||
|
let example = first_expected.example();
|
||||||
|
Some(format!("try a token of the expected type: {example}"))
|
||||||
|
} else {
|
||||||
|
None
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
impl OSDBError for ScanError {
|
impl OSDBError for ScanError {
|
||||||
fn display(&self, file: &str, input: &str) {
|
fn display(&self, file: &str, input: &str) {
|
||||||
let location = (file, Into::<std::ops::Range<usize>>::into(&self.location));
|
let location = (file, Into::<std::ops::Range<usize>>::into(&self.location));
|
||||||
|
|
|
||||||
|
|
@ -1,9 +1,10 @@
|
||||||
use crate::branding;
|
use crate::branding;
|
||||||
|
|
||||||
#[derive(Debug, Eq, PartialEq)]
|
#[derive(Debug, Eq, PartialEq, Clone)]
|
||||||
pub enum MetaCommand {
|
pub enum MetaCommand {
|
||||||
Exit,
|
Exit,
|
||||||
About,
|
About,
|
||||||
|
Version,
|
||||||
}
|
}
|
||||||
|
|
||||||
impl std::fmt::Display for MetaCommand {
|
impl std::fmt::Display for MetaCommand {
|
||||||
|
|
@ -11,6 +12,7 @@ impl std::fmt::Display for MetaCommand {
|
||||||
match self {
|
match self {
|
||||||
MetaCommand::Exit => write!(f, "exit"),
|
MetaCommand::Exit => write!(f, "exit"),
|
||||||
MetaCommand::About => write!(f, "about"),
|
MetaCommand::About => write!(f, "about"),
|
||||||
|
MetaCommand::Version => write!(f, "version"),
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
@ -27,6 +29,10 @@ impl MetaCommand {
|
||||||
print!("{}", branding::about_msg());
|
print!("{}", branding::about_msg());
|
||||||
MetaCommandExecuteResult { should_exit: false }
|
MetaCommandExecuteResult { should_exit: false }
|
||||||
}
|
}
|
||||||
|
MetaCommand::Version => {
|
||||||
|
print!("{}", branding::version_msg());
|
||||||
|
MetaCommandExecuteResult { should_exit: false }
|
||||||
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
|
||||||
271
src/parser.rs
271
src/parser.rs
|
|
@ -1,38 +1,241 @@
|
||||||
use std::collections::VecDeque;
|
use std::collections::VecDeque;
|
||||||
|
|
||||||
use crate::{
|
use crate::{
|
||||||
command::{Command, CommandParseError},
|
command::{Command, CommandParseError, ExpectedToken},
|
||||||
statements::Statement,
|
statements::Statement,
|
||||||
tokens::tokenize,
|
tokens::{Location, Token, TokenData, tokenize},
|
||||||
};
|
};
|
||||||
|
|
||||||
|
// Helper function to skip tokens until reaching a semicolon or end of file
|
||||||
|
// This helps with error recovery when a statement has a syntax error
|
||||||
|
fn skip_to_next_statement(tokens: &mut VecDeque<Token>) {
|
||||||
|
while let Some(token) = tokens.front() {
|
||||||
|
match token.data {
|
||||||
|
TokenData::Semicolon | TokenData::EndOfFile => break,
|
||||||
|
_ => {
|
||||||
|
tokens.pop_front();
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
// Consume the semicolon if that's what we stopped at
|
||||||
|
if tokens
|
||||||
|
.front()
|
||||||
|
.is_some_and(|t| matches!(t.data, TokenData::Semicolon))
|
||||||
|
{
|
||||||
|
tokens.pop_front();
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
// Helper function to check for a semicolon after a statement
|
||||||
|
fn expect_semicolon(tokens: &mut VecDeque<crate::tokens::Token>) -> Result<(), CommandParseError> {
|
||||||
|
if let Some(next_token) = tokens.front() {
|
||||||
|
match next_token.data {
|
||||||
|
TokenData::Semicolon => {
|
||||||
|
tokens.pop_front(); // Consume the semicolon
|
||||||
|
Ok(())
|
||||||
|
}
|
||||||
|
_ => Err(CommandParseError::UnexpectedToken(
|
||||||
|
next_token.clone(),
|
||||||
|
&[ExpectedToken::Semicolon],
|
||||||
|
)),
|
||||||
|
}
|
||||||
|
} else {
|
||||||
|
// Even at the end of input, we need a semicolon
|
||||||
|
Err(CommandParseError::UnexpectedToken(
|
||||||
|
Token {
|
||||||
|
location: tokens
|
||||||
|
.back()
|
||||||
|
.map_or_else(Location::default, |t| t.location.clone()),
|
||||||
|
data: TokenData::EndOfFile,
|
||||||
|
lexeme: String::new(),
|
||||||
|
},
|
||||||
|
&[ExpectedToken::Semicolon],
|
||||||
|
))
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
fn parse_select_command(
|
||||||
|
tokens: &mut VecDeque<crate::tokens::Token>,
|
||||||
|
) -> Result<Command, CommandParseError> {
|
||||||
|
// Parse the select command (currently doesn't require additional tokens)
|
||||||
|
let cmd = Command::Statement(Statement::Select);
|
||||||
|
|
||||||
|
// Check for semicolon after select command
|
||||||
|
expect_semicolon(tokens)?;
|
||||||
|
|
||||||
|
Ok(cmd)
|
||||||
|
}
|
||||||
|
|
||||||
pub fn parse(file: String, input: String) -> Result<Vec<Command>, Vec<CommandParseError>> {
|
pub fn parse(file: String, input: String) -> Result<Vec<Command>, Vec<CommandParseError>> {
|
||||||
let mut tokens: VecDeque<_> = tokenize(input, file)
|
let mut tokens: VecDeque<_> = tokenize(input, file)
|
||||||
.map_err(|x| x.into_iter().map(|x| x.into()).collect::<Vec<_>>())?
|
.map_err(|x| x.into_iter().map(|x| x.into()).collect::<Vec<_>>())?
|
||||||
.into();
|
.into();
|
||||||
let mut cmds = Vec::new();
|
let mut cmds = Vec::new();
|
||||||
let mut errs = Vec::new();
|
let mut errs = Vec::new();
|
||||||
|
|
||||||
while let Some(token) = tokens.pop_front() {
|
while let Some(token) = tokens.pop_front() {
|
||||||
match token.data {
|
match token.data {
|
||||||
crate::tokens::TokenData::Insert => cmds.push(Command::Statement(Statement::Insert)),
|
TokenData::Insert => match parse_insert_command(&mut tokens) {
|
||||||
crate::tokens::TokenData::Select => cmds.push(Command::Statement(Statement::Select)),
|
Ok(cmd) => cmds.push(cmd),
|
||||||
crate::tokens::TokenData::MetaCommand(meta_command) => {
|
Err(err) => {
|
||||||
cmds.push(Command::MetaCommand(meta_command))
|
errs.push(err);
|
||||||
|
skip_to_next_statement(&mut tokens); // Skip to next statement for error recovery
|
||||||
}
|
}
|
||||||
crate::tokens::TokenData::Int(_) => errs.push(CommandParseError::UnexpectedToken(
|
},
|
||||||
|
TokenData::Select => match parse_select_command(&mut tokens) {
|
||||||
|
Ok(cmd) => cmds.push(cmd),
|
||||||
|
Err(err) => {
|
||||||
|
errs.push(err);
|
||||||
|
skip_to_next_statement(&mut tokens); // Skip to next statement for error recovery
|
||||||
|
}
|
||||||
|
},
|
||||||
|
TokenData::MetaCommand(meta_command) => {
|
||||||
|
match parse_meta_command(meta_command, &mut tokens) {
|
||||||
|
Ok(cmd) => cmds.push(cmd),
|
||||||
|
Err(err) => {
|
||||||
|
errs.push(err);
|
||||||
|
skip_to_next_statement(&mut tokens); // Skip to next statement for error recovery
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
TokenData::Semicolon => {
|
||||||
|
// Empty statement (just a semicolon) - ignore it
|
||||||
|
}
|
||||||
|
TokenData::Int(_) => {
|
||||||
|
errs.push(CommandParseError::UnexpectedToken(
|
||||||
token,
|
token,
|
||||||
&["statement", "meta command", "eof"],
|
&[
|
||||||
)),
|
ExpectedToken::Statement,
|
||||||
crate::tokens::TokenData::String(_) => errs.push(CommandParseError::UnexpectedToken(
|
ExpectedToken::MetaCommand,
|
||||||
|
ExpectedToken::EndOfFile,
|
||||||
|
],
|
||||||
|
));
|
||||||
|
skip_to_next_statement(&mut tokens);
|
||||||
|
}
|
||||||
|
TokenData::String(_) => {
|
||||||
|
errs.push(CommandParseError::UnexpectedToken(
|
||||||
token,
|
token,
|
||||||
&["statement", "meta command", "eof"],
|
&[
|
||||||
)),
|
ExpectedToken::Statement,
|
||||||
crate::tokens::TokenData::EndOfFile => (),
|
ExpectedToken::MetaCommand,
|
||||||
|
ExpectedToken::EndOfFile,
|
||||||
|
],
|
||||||
|
));
|
||||||
|
skip_to_next_statement(&mut tokens);
|
||||||
|
}
|
||||||
|
TokenData::EndOfFile => (), // End of parsing
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
if errs.is_empty() { Ok(cmds) } else { Err(errs) }
|
if errs.is_empty() { Ok(cmds) } else { Err(errs) }
|
||||||
}
|
}
|
||||||
|
|
||||||
|
fn parse_meta_command(
|
||||||
|
meta_command: crate::meta_commands::MetaCommand,
|
||||||
|
tokens: &mut VecDeque<Token>,
|
||||||
|
) -> Result<Command, CommandParseError> {
|
||||||
|
if let Some(token) = tokens.pop_front() {
|
||||||
|
if matches!(token.data, TokenData::EndOfFile) {
|
||||||
|
Ok(Command::MetaCommand(meta_command))
|
||||||
|
} else {
|
||||||
|
Err(CommandParseError::UnexpectedToken(
|
||||||
|
token,
|
||||||
|
&[ExpectedToken::EndOfFile],
|
||||||
|
))
|
||||||
|
}
|
||||||
|
} else {
|
||||||
|
Ok(Command::MetaCommand(meta_command))
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
fn parse_insert_command(
|
||||||
|
tokens: &mut VecDeque<crate::tokens::Token>,
|
||||||
|
) -> Result<Command, CommandParseError> {
|
||||||
|
// According to grammar.ebnf, insert command should be: insert int string string semicolon
|
||||||
|
|
||||||
|
// Parse the id (integer)
|
||||||
|
let id_token = tokens.pop_front().ok_or_else(|| {
|
||||||
|
CommandParseError::UnexpectedToken(
|
||||||
|
Token {
|
||||||
|
location: tokens
|
||||||
|
.back()
|
||||||
|
.map_or_else(Location::default, |t| t.location.clone()),
|
||||||
|
data: TokenData::EndOfFile,
|
||||||
|
lexeme: String::new(),
|
||||||
|
},
|
||||||
|
&[ExpectedToken::Integer],
|
||||||
|
)
|
||||||
|
})?;
|
||||||
|
|
||||||
|
let id = match id_token.data {
|
||||||
|
TokenData::Int(id) => id,
|
||||||
|
_ => {
|
||||||
|
return Err(CommandParseError::UnexpectedToken(
|
||||||
|
id_token,
|
||||||
|
&[ExpectedToken::Integer],
|
||||||
|
));
|
||||||
|
}
|
||||||
|
};
|
||||||
|
|
||||||
|
// Parse the username (string)
|
||||||
|
let username_token = tokens.pop_front().ok_or_else(|| {
|
||||||
|
CommandParseError::UnexpectedToken(
|
||||||
|
Token {
|
||||||
|
location: tokens
|
||||||
|
.back()
|
||||||
|
.map_or_else(Location::default, |t| t.location.clone()),
|
||||||
|
data: TokenData::EndOfFile,
|
||||||
|
lexeme: String::new(),
|
||||||
|
},
|
||||||
|
&[ExpectedToken::String],
|
||||||
|
)
|
||||||
|
})?;
|
||||||
|
|
||||||
|
let username = match username_token.data {
|
||||||
|
TokenData::String(username) => username,
|
||||||
|
_ => {
|
||||||
|
return Err(CommandParseError::UnexpectedToken(
|
||||||
|
username_token,
|
||||||
|
&[ExpectedToken::String],
|
||||||
|
));
|
||||||
|
}
|
||||||
|
};
|
||||||
|
|
||||||
|
// Parse the email (string)
|
||||||
|
let email_token = tokens.pop_front().ok_or_else(|| {
|
||||||
|
CommandParseError::UnexpectedToken(
|
||||||
|
Token {
|
||||||
|
location: tokens
|
||||||
|
.back()
|
||||||
|
.map_or_else(Location::default, |t| t.location.clone()),
|
||||||
|
data: TokenData::EndOfFile,
|
||||||
|
lexeme: String::new(),
|
||||||
|
},
|
||||||
|
&[ExpectedToken::String],
|
||||||
|
)
|
||||||
|
})?;
|
||||||
|
|
||||||
|
let email = match email_token.data {
|
||||||
|
TokenData::String(email) => email,
|
||||||
|
_ => {
|
||||||
|
return Err(CommandParseError::UnexpectedToken(
|
||||||
|
email_token,
|
||||||
|
&[ExpectedToken::String],
|
||||||
|
));
|
||||||
|
}
|
||||||
|
};
|
||||||
|
|
||||||
|
// Check for semicolon after the insert command
|
||||||
|
expect_semicolon(tokens)?;
|
||||||
|
|
||||||
|
Ok(Command::Statement(Statement::Insert {
|
||||||
|
id,
|
||||||
|
username,
|
||||||
|
email,
|
||||||
|
}))
|
||||||
|
}
|
||||||
|
|
||||||
#[cfg(test)]
|
#[cfg(test)]
|
||||||
mod tests {
|
mod tests {
|
||||||
use super::*;
|
use super::*;
|
||||||
|
|
@ -42,10 +245,32 @@ mod tests {
|
||||||
fn test_parse_single_correct() {
|
fn test_parse_single_correct() {
|
||||||
let file = String::from("<stdin>");
|
let file = String::from("<stdin>");
|
||||||
assert_debug_snapshot!(parse(file.clone(), String::from(".exit")));
|
assert_debug_snapshot!(parse(file.clone(), String::from(".exit")));
|
||||||
|
assert_debug_snapshot!(parse(file.clone(), String::from("select;")));
|
||||||
|
assert_debug_snapshot!(parse(file.clone(), String::from("sElEcT;")));
|
||||||
|
}
|
||||||
|
|
||||||
|
#[test]
|
||||||
|
fn test_parse_insert_command() {
|
||||||
|
let file = String::from("<stdin>");
|
||||||
|
assert_debug_snapshot!(parse(
|
||||||
|
file.clone(),
|
||||||
|
String::from(r#"insert 1 "username" "email@example.com";"#)
|
||||||
|
));
|
||||||
|
assert_debug_snapshot!(parse(
|
||||||
|
file.clone(),
|
||||||
|
String::from(r#"insert "not_an_id" "username" "email@example.com";"#)
|
||||||
|
));
|
||||||
|
assert_debug_snapshot!(parse(file.clone(), String::from(r#"insert 1 "username";"#)));
|
||||||
|
}
|
||||||
|
|
||||||
|
#[test]
|
||||||
|
fn test_parse_missing_semicolon() {
|
||||||
|
let file = String::from("<stdin>");
|
||||||
assert_debug_snapshot!(parse(file.clone(), String::from("select")));
|
assert_debug_snapshot!(parse(file.clone(), String::from("select")));
|
||||||
assert_debug_snapshot!(parse(file.clone(), String::from("sElEcT")));
|
assert_debug_snapshot!(parse(
|
||||||
assert_debug_snapshot!(parse(file.clone(), String::from("INSERT")));
|
file.clone(),
|
||||||
assert_debug_snapshot!(parse(file.clone(), String::from("InSErT")));
|
String::from(r#"insert 1 "username" "email@example.com""#)
|
||||||
|
));
|
||||||
}
|
}
|
||||||
|
|
||||||
#[test]
|
#[test]
|
||||||
|
|
@ -60,10 +285,22 @@ mod tests {
|
||||||
|
|
||||||
#[test]
|
#[test]
|
||||||
fn test_parse_multiple_correct() {
|
fn test_parse_multiple_correct() {
|
||||||
|
let file = String::from("<stdin>");
|
||||||
|
assert_debug_snapshot!(parse(file.clone(), String::from("select; select; select;")));
|
||||||
|
}
|
||||||
|
|
||||||
|
#[test]
|
||||||
|
fn test_meta_command_require_eof() {
|
||||||
|
let file = String::from("<stdin>");
|
||||||
|
assert_debug_snapshot!(parse(file.clone(), String::from(".exit select; select;")));
|
||||||
|
}
|
||||||
|
|
||||||
|
#[test]
|
||||||
|
fn test_parse_multiple_statements_with_insert() {
|
||||||
let file = String::from("<stdin>");
|
let file = String::from("<stdin>");
|
||||||
assert_debug_snapshot!(parse(
|
assert_debug_snapshot!(parse(
|
||||||
file.clone(),
|
file.clone(),
|
||||||
String::from(".exit select select insert select")
|
String::from(r#"select; insert 1 "user" "email@test.com"; select;"#)
|
||||||
));
|
));
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
|
||||||
|
|
@ -2,4 +2,4 @@
|
||||||
source: src/command.rs
|
source: src/command.rs
|
||||||
expression: result
|
expression: result
|
||||||
---
|
---
|
||||||
insert
|
insert 45 "user" "user@example.org"
|
||||||
|
|
|
||||||
|
|
@ -0,0 +1,22 @@
|
||||||
|
---
|
||||||
|
source: src/parser.rs
|
||||||
|
expression: "parse(file.clone(), String::from(\".exit select; select;\"))"
|
||||||
|
---
|
||||||
|
Err(
|
||||||
|
[
|
||||||
|
UnexpectedToken(
|
||||||
|
Token {
|
||||||
|
location: Location {
|
||||||
|
file: "<stdin>",
|
||||||
|
offset: 6,
|
||||||
|
length: 6,
|
||||||
|
},
|
||||||
|
data: Select,
|
||||||
|
lexeme: "select",
|
||||||
|
},
|
||||||
|
[
|
||||||
|
EndOfFile,
|
||||||
|
],
|
||||||
|
),
|
||||||
|
],
|
||||||
|
)
|
||||||
|
|
@ -0,0 +1,24 @@
|
||||||
|
---
|
||||||
|
source: src/parser.rs
|
||||||
|
expression: "parse(file.clone(),\nString::from(r#\"insert \"not_an_id\" \"username\" \"email@example.com\";\"#))"
|
||||||
|
---
|
||||||
|
Err(
|
||||||
|
[
|
||||||
|
UnexpectedToken(
|
||||||
|
Token {
|
||||||
|
location: Location {
|
||||||
|
file: "<stdin>",
|
||||||
|
offset: 7,
|
||||||
|
length: 11,
|
||||||
|
},
|
||||||
|
data: String(
|
||||||
|
"not_an_id",
|
||||||
|
),
|
||||||
|
lexeme: "\"not_an_id\"",
|
||||||
|
},
|
||||||
|
[
|
||||||
|
Integer,
|
||||||
|
],
|
||||||
|
),
|
||||||
|
],
|
||||||
|
)
|
||||||
|
|
@ -0,0 +1,22 @@
|
||||||
|
---
|
||||||
|
source: src/parser.rs
|
||||||
|
expression: "parse(file.clone(), String::from(r#\"insert 1 \"username\";\"#))"
|
||||||
|
---
|
||||||
|
Err(
|
||||||
|
[
|
||||||
|
UnexpectedToken(
|
||||||
|
Token {
|
||||||
|
location: Location {
|
||||||
|
file: "<stdin>",
|
||||||
|
offset: 19,
|
||||||
|
length: 1,
|
||||||
|
},
|
||||||
|
data: Semicolon,
|
||||||
|
lexeme: ";",
|
||||||
|
},
|
||||||
|
[
|
||||||
|
String,
|
||||||
|
],
|
||||||
|
),
|
||||||
|
],
|
||||||
|
)
|
||||||
15
src/snapshots/osdb__parser__tests__parse_insert_command.snap
Normal file
15
src/snapshots/osdb__parser__tests__parse_insert_command.snap
Normal file
|
|
@ -0,0 +1,15 @@
|
||||||
|
---
|
||||||
|
source: src/parser.rs
|
||||||
|
expression: "parse(file.clone(),\nString::from(r#\"insert 1 \"username\" \"email@example.com\"\"#))"
|
||||||
|
---
|
||||||
|
Ok(
|
||||||
|
[
|
||||||
|
Statement(
|
||||||
|
Insert {
|
||||||
|
id: 1,
|
||||||
|
username: "username",
|
||||||
|
email: "email@example.com",
|
||||||
|
},
|
||||||
|
),
|
||||||
|
],
|
||||||
|
)
|
||||||
|
|
@ -0,0 +1,22 @@
|
||||||
|
---
|
||||||
|
source: src/parser.rs
|
||||||
|
expression: "parse(file.clone(),\nString::from(r#\"insert 1 \"username\" \"email@example.com\"\"#))"
|
||||||
|
---
|
||||||
|
Err(
|
||||||
|
[
|
||||||
|
UnexpectedToken(
|
||||||
|
Token {
|
||||||
|
location: Location {
|
||||||
|
file: "<stdin>",
|
||||||
|
offset: 39,
|
||||||
|
length: 0,
|
||||||
|
},
|
||||||
|
data: EndOfFile,
|
||||||
|
lexeme: "",
|
||||||
|
},
|
||||||
|
[
|
||||||
|
Semicolon,
|
||||||
|
],
|
||||||
|
),
|
||||||
|
],
|
||||||
|
)
|
||||||
|
|
@ -0,0 +1,22 @@
|
||||||
|
---
|
||||||
|
source: src/parser.rs
|
||||||
|
expression: "parse(file.clone(), String::from(\"select\"))"
|
||||||
|
---
|
||||||
|
Err(
|
||||||
|
[
|
||||||
|
UnexpectedToken(
|
||||||
|
Token {
|
||||||
|
location: Location {
|
||||||
|
file: "<stdin>",
|
||||||
|
offset: 6,
|
||||||
|
length: 0,
|
||||||
|
},
|
||||||
|
data: EndOfFile,
|
||||||
|
lexeme: "",
|
||||||
|
},
|
||||||
|
[
|
||||||
|
Semicolon,
|
||||||
|
],
|
||||||
|
),
|
||||||
|
],
|
||||||
|
)
|
||||||
|
|
@ -1,21 +1,15 @@
|
||||||
---
|
---
|
||||||
source: src/parser.rs
|
source: src/parser.rs
|
||||||
expression: "parse(file.clone(), String::from(\".exit select select insert select\"))"
|
expression: "parse(file.clone(), String::from(\"select; select; select;\"))"
|
||||||
---
|
---
|
||||||
Ok(
|
Ok(
|
||||||
[
|
[
|
||||||
MetaCommand(
|
|
||||||
Exit,
|
|
||||||
),
|
|
||||||
Statement(
|
Statement(
|
||||||
Select,
|
Select,
|
||||||
),
|
),
|
||||||
Statement(
|
Statement(
|
||||||
Select,
|
Select,
|
||||||
),
|
),
|
||||||
Statement(
|
|
||||||
Insert,
|
|
||||||
),
|
|
||||||
Statement(
|
Statement(
|
||||||
Select,
|
Select,
|
||||||
),
|
),
|
||||||
|
|
|
||||||
|
|
@ -0,0 +1,21 @@
|
||||||
|
---
|
||||||
|
source: src/parser.rs
|
||||||
|
expression: "parse(file.clone(),\nString::from(r#\"select; insert 1 \"user\" \"email@test.com\"; select;\"#))"
|
||||||
|
---
|
||||||
|
Ok(
|
||||||
|
[
|
||||||
|
Statement(
|
||||||
|
Select,
|
||||||
|
),
|
||||||
|
Statement(
|
||||||
|
Insert {
|
||||||
|
id: 1,
|
||||||
|
username: "user",
|
||||||
|
email: "email@test.com",
|
||||||
|
},
|
||||||
|
),
|
||||||
|
Statement(
|
||||||
|
Select,
|
||||||
|
),
|
||||||
|
],
|
||||||
|
)
|
||||||
|
|
@ -1,6 +1,10 @@
|
||||||
#[derive(Debug)]
|
#[derive(Debug)]
|
||||||
pub enum Statement {
|
pub enum Statement {
|
||||||
Insert,
|
Insert {
|
||||||
|
id: i64,
|
||||||
|
username: String,
|
||||||
|
email: String,
|
||||||
|
},
|
||||||
Select,
|
Select,
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
@ -11,8 +15,12 @@ pub struct StatementExecuteResult {
|
||||||
impl Statement {
|
impl Statement {
|
||||||
pub fn execute(&self) -> StatementExecuteResult {
|
pub fn execute(&self) -> StatementExecuteResult {
|
||||||
match self {
|
match self {
|
||||||
Statement::Insert => StatementExecuteResult {
|
Statement::Insert {
|
||||||
msg: String::from("insert"),
|
id,
|
||||||
|
username,
|
||||||
|
email,
|
||||||
|
} => StatementExecuteResult {
|
||||||
|
msg: format!("insert {id:?} {username:?} {email:?}"),
|
||||||
},
|
},
|
||||||
Statement::Select => StatementExecuteResult {
|
Statement::Select => StatementExecuteResult {
|
||||||
msg: String::from("select"),
|
msg: String::from("select"),
|
||||||
|
|
|
||||||
|
|
@ -1,6 +1,6 @@
|
||||||
use crate::meta_commands::MetaCommand;
|
use crate::meta_commands::MetaCommand;
|
||||||
|
|
||||||
#[derive(Debug, Eq, PartialEq)]
|
#[derive(Debug, Eq, PartialEq, Clone)]
|
||||||
pub enum TokenData {
|
pub enum TokenData {
|
||||||
Insert,
|
Insert,
|
||||||
Select,
|
Select,
|
||||||
|
|
@ -8,9 +8,10 @@ pub enum TokenData {
|
||||||
EndOfFile,
|
EndOfFile,
|
||||||
Int(i64),
|
Int(i64),
|
||||||
String(String),
|
String(String),
|
||||||
|
Semicolon,
|
||||||
}
|
}
|
||||||
|
|
||||||
#[derive(Debug, Eq, PartialEq)]
|
#[derive(Debug, Eq, PartialEq, Clone)]
|
||||||
pub struct Location {
|
pub struct Location {
|
||||||
/// file name
|
/// file name
|
||||||
pub file: String,
|
pub file: String,
|
||||||
|
|
@ -29,6 +30,12 @@ impl From<&Location> for std::ops::Range<usize> {
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
impl Default for Location {
|
||||||
|
fn default() -> Self {
|
||||||
|
Self::new(String::from("<unknown>"), 0, 0)
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
impl Location {
|
impl Location {
|
||||||
/// ```
|
/// ```
|
||||||
/// use osdb::tokens::Location;
|
/// use osdb::tokens::Location;
|
||||||
|
|
@ -46,7 +53,7 @@ impl Location {
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
#[derive(Debug, Eq, PartialEq)]
|
#[derive(Debug, Eq, PartialEq, Clone)]
|
||||||
pub struct Token {
|
pub struct Token {
|
||||||
/// Where in the input was this token found?
|
/// Where in the input was this token found?
|
||||||
pub location: Location,
|
pub location: Location,
|
||||||
|
|
@ -65,6 +72,7 @@ impl std::fmt::Display for Token {
|
||||||
TokenData::EndOfFile => write!(f, "end of file"),
|
TokenData::EndOfFile => write!(f, "end of file"),
|
||||||
TokenData::Int(x) => write!(f, "integer {x}"),
|
TokenData::Int(x) => write!(f, "integer {x}"),
|
||||||
TokenData::String(x) => write!(f, "string {x:?}"),
|
TokenData::String(x) => write!(f, "string {x:?}"),
|
||||||
|
TokenData::Semicolon => write!(f, "semicolon"),
|
||||||
}?;
|
}?;
|
||||||
let lexeme = &self.lexeme;
|
let lexeme = &self.lexeme;
|
||||||
write!(f, " {lexeme:?}")
|
write!(f, " {lexeme:?}")
|
||||||
|
|
@ -161,6 +169,7 @@ impl Tokenizer {
|
||||||
match word.to_lowercase().as_str() {
|
match word.to_lowercase().as_str() {
|
||||||
".exit" => Some(TokenData::MetaCommand(MetaCommand::Exit)),
|
".exit" => Some(TokenData::MetaCommand(MetaCommand::Exit)),
|
||||||
".about" => Some(TokenData::MetaCommand(MetaCommand::About)),
|
".about" => Some(TokenData::MetaCommand(MetaCommand::About)),
|
||||||
|
".version" => Some(TokenData::MetaCommand(MetaCommand::Version)),
|
||||||
_ => None,
|
_ => None,
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
@ -306,6 +315,15 @@ impl Tokenizer {
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
fn scan_semicolon(&mut self) -> Result<Token, ScanError> {
|
||||||
|
self.advance();
|
||||||
|
Ok(Token {
|
||||||
|
location: self.previous_location(1),
|
||||||
|
data: TokenData::Semicolon,
|
||||||
|
lexeme: String::from(";"),
|
||||||
|
})
|
||||||
|
}
|
||||||
|
|
||||||
fn scan_token(&mut self) -> Result<Option<Token>, ScanError> {
|
fn scan_token(&mut self) -> Result<Option<Token>, ScanError> {
|
||||||
loop {
|
loop {
|
||||||
if let Some(c) = self.peek() {
|
if let Some(c) = self.peek() {
|
||||||
|
|
@ -317,6 +335,8 @@ impl Tokenizer {
|
||||||
return self.scan_integer().map(Some);
|
return self.scan_integer().map(Some);
|
||||||
} else if c == '"' {
|
} else if c == '"' {
|
||||||
return self.scan_string().map(Some);
|
return self.scan_string().map(Some);
|
||||||
|
} else if c == ';' {
|
||||||
|
return self.scan_semicolon().map(Some);
|
||||||
} else if c.is_whitespace() {
|
} else if c.is_whitespace() {
|
||||||
self.advance();
|
self.advance();
|
||||||
} else {
|
} else {
|
||||||
|
|
|
||||||
Loading…
Add table
Add a link
Reference in a new issue