added some basic validation

dev/src/idl/mod.rs

@@ -1,66 +1,72 @@
-pub mod protocol;
 mod parser;
+mod types;
+mod protocol;
 
+use crate::idl::{parser::parse, types::get_protocols};
 use std::io::Read;
 
-use self::parser::parse;
-
-use logos::Logos;
+use logos::{Lexer, Logos, Skip};
 
 #[derive(Logos, Debug, PartialEq, Clone)]
-#[logos(skip r"[ \t\n\f]+")] // Ignore this regex pattern between tokens
+#[logos(extras = (usize, usize))]
 enum Token {
-    #[regex("//[^\n]*\n", logos::skip)]
+    #[regex("//[^\n]*", logos::skip)]
+    Comment,
+
+    #[regex(r"[ \t\f]+", logos::skip)]
     Ignored,
 
-    // Tokens can be literal strings, of any length.
-    #[token("protocol")]
-    Protocol,
+    #[regex(r"\n", newline_callback)]
+    Newline,
 
     // Tokens can be literal strings, of any length.
-    #[token("type")]
-    Type,
-    #[token("fn", priority = 5)]
-    Fn,
+    #[token("protocol", get_token_position)]
+    Protocol((usize,usize)),
 
-    #[token("enum")]
-    Enum,
+    // Tokens can be literal strings, of any length.
+    #[token("type", get_token_position)]
+    Type((usize, usize)),
 
-    #[token("struct")]
-    Struct,
+    #[token("fn", priority = 5, callback = get_token_position)]
+    Fn((usize, usize)),
 
-    #[token("{")]
-    LBrace,
+    #[token("enum", get_token_position)]
+    Enum((usize, usize)),
 
-    #[token("}")]
-    RBrace,
+    #[token("struct", get_token_position)]
+    Struct((usize, usize)),
 
-    #[token("(")]
-    LParen,
+    #[token("{", get_token_position)]
+    LBrace((usize, usize)),
 
-    #[token(")")]
-    RParen,
+    #[token("}", get_token_position)]
+    RBrace((usize, usize)),
 
-    #[token(":")]
-    Colon,
+    #[token("(", get_token_position)]
+    LParen((usize, usize)),
 
-    #[token(";")]
-    SemiColon,
+    #[token(")", get_token_position)]
+    RParen((usize, usize)),
 
-    #[token(",")]
-    Comma,
+    #[token(":", get_token_position)]
+    Colon((usize, usize)),
 
-    #[token("=")]
-    Equal,
+    #[token(";", get_token_position)]
+    SemiColon((usize, usize)),
 
-    #[token("->")]
-    RArrow,
+    #[token(",", get_token_position)] Comma((usize, usize)),
 
-    #[regex("[a-zA-Z_][a-zA-Z_1234567890]+", |lex|{lex.slice().to_string()})]
-    Identifier(String),
+    #[token("=", get_token_position)]
+    Equal((usize, usize)),
 
-    #[regex("[1234567890]+", |lex|{lex.slice().parse::<u64>().unwrap()})]
-    Number(u64),
+    #[token("->", get_token_position)]
+    RArrow((usize, usize)),
+
+    #[regex("[a-zA-Z_][a-zA-Z_1234567890]+", |lex|{let text = lex.slice().to_string(); let (line, col) = get_token_position(lex); (line, col, text)})]
+    Identifier((usize, usize, String)),
+
+    #[regex("[1234567890]+", |lex|{let num = lex.slice().parse::<u64>().unwrap(); let (line, col) = get_token_position(lex); (line, col, num) })]
+    Number((usize, usize, u64)),
 
     #[regex(r"@[a-zA-Z_]+", /*|lex|{lex.slice().to_string()}*/ logos::skip)]
     Decorator,
@@ -76,14 +82,16 @@ pub fn build_idl(name: String) {
     for x in lex {
         match x {
             Ok(token) => {
-                println!("{:?}", token);
                 tokens.push(token);
             }
             Err(err) => println!("{:?}", err),
         }
     }
 
-    println!("{:#?}", parse(tokens));
+    let protocols = get_protocols(parse(tokens));
+    let data : Vec<u8> = vec![1, 5, 12, 12, 12, 12, 3, 28, 8, 28];
+    println!("{:#?}", &protocols);
+    protocols.validate("Foo", "bar" , data).unwrap();
 }
 
 fn open_protocol(name: String) -> String {
@@ -93,3 +101,17 @@ fn open_protocol(name: String) -> String {
     file.read_to_string(&mut contents).unwrap();
     contents
 }
+
+fn get_token_position(lex : &mut Lexer<Token>) -> (usize, usize) {
+    let line = lex.extras.0;
+    let column = lex.span().start - lex.extras.1;
+
+    (line, column + 1)
+}
+
+/// Update the line count and the char index.
+fn newline_callback(lex: &mut Lexer<Token>) -> Skip {
+    lex.extras.0 += 1;
+    lex.extras.1 = lex.span().end;
+    Skip
+}

dev/src/idl/parser.rs

@@ -2,67 +2,70 @@ use std::{iter::Peekable, slice::Iter};
 use super::Token;
 
 #[derive(Debug, Clone, PartialEq)]
-pub struct AST(Vec<Declaration>);
+pub struct AST(pub Vec<Declaration>);
 
 #[derive(Debug, Clone, PartialEq)]
-enum Declaration{
+pub enum Declaration{
     EnumDeclaration(EnumDeclaration),
     StructDeclaration(StructDeclaration),
     TypeDeclaration(TypeDeclaration),
     ProtocolDeclaration(ProtocolDeclaration),
 }
+
 #[derive(Debug, Clone, PartialEq)]
-struct StructDeclaration {
-    name : String,
-    members : Vec<StructMember>,
+pub struct StructDeclaration {
+    pub name : String,
+    pub members : Vec<StructMember>,
 }
 #[derive(Debug, Clone, PartialEq)]
-struct TypeDeclaration {
-    name : String,
-    type_name : String,
+pub struct TypeDeclaration {
+    pub name : String,
+    pub type_name : String,
 }
 #[derive(Debug, Clone, PartialEq)]
-struct StructMember {
-    name : String,
-    type_name : String,
+pub struct StructMember {
+    pub name : String,
+    pub type_name : String,
 }
 
 #[derive(Debug, Clone, PartialEq)]
-struct EnumDeclaration {
-    name : String,
-    members : Vec<EnumMember>,
+pub struct EnumDeclaration {
+    pub name : String,
+    pub members : Vec<EnumMember>,
 }
 #[derive(Debug, Clone, PartialEq)]
-struct EnumMember {
-    name : String,
-    number: u64,
+pub struct EnumMember {
+    pub name : String,
+    pub number: u64,
 }
 
 #[derive(Debug, Clone, PartialEq)]
-struct ProtocolDeclaration{
-    name : String,
-    interface : Vec<FuncDeclaration>,
+pub struct ProtocolDeclaration{
+    pub name : String,
+    pub interface : Vec<FuncDeclaration>,
 }
 
 #[derive(Debug, Clone, PartialEq)]
-struct FuncDeclaration {
-    name : String,
-    arg_list : Vec<String>,
-    return_type : String,
+pub struct FuncDeclaration {
+    pub name : String,
+    pub arg_list : Vec<String>,
+    pub return_type : String,
 }
 
-
-/// Consume's a token that's expected. If the token that's consumed is not
-/// the given expected token then panic
-fn consume(tokens : &mut Peekable<Iter<'_, Token>>, token : Token) {
-    let a = tokens.next();
+macro_rules! consume {
+    ($tokens:ident, $token_type:path) => { 
+        let a = $tokens.next();
         match a {
-        None => panic!("Expected {:?}, Got End Of File", token),
-        Some(a) => if *a != token {
-            panic!("Expected {:?}, Got {:?}", token, *a);
+            None => panic!("Expected {}, Got End Of Tokens", stringify!($token_type)),
+            Some(t) => {
+                if matches!(t, $token_type(_)){}
+                else {
+                    panic!("Expected {:?}, Got {:?}", stringify!($token_type), t);
                 }
             }
         }
+    };
+}
 
 pub fn parse(tokens : Vec<Token>) -> AST{
     let mut tokens_iter = tokens.iter().peekable();
@@ -84,29 +87,29 @@ fn declaration(tokens : &mut Peekable<Iter<'_, Token>>) -> Option<Declaration> {
     match tokens.peek(){
         None => None,
         Some(tok) => match tok {
-            Token::Enum => Some(enum_decl(tokens)),
-            Token::Struct => Some(struct_decl(tokens)),
-            Token::Type => Some(type_declaration(tokens)),
-            Token::Protocol => Some(protocol_declaration(tokens)),
+            Token::Enum(_)     => Some(enum_decl(tokens)),
+            Token::Struct(_)   => Some(struct_decl(tokens)),
+            Token::Type(_)     => Some(type_declaration(tokens)),
+            Token::Protocol(_) => Some(protocol_declaration(tokens)),
             _ => None,
         }
     }
 }
 
 fn enum_decl(tokens : &mut Peekable<Iter<'_, Token>>) -> Declaration {
-    consume(tokens, Token::Enum);
+    consume!(tokens, Token::Enum);
     let name = identifier(tokens).expect("Expected Identifier after `enum`");
-    consume(tokens, Token::LBrace);
+    consume!(tokens, Token::LBrace);
     let mut members = Vec::new();
 
     match tokens.peek().expect("Unexpected EOF after LBrace") {
-        Token::RBrace => {}, // skip checking for enum_members if empty
+        Token::RBrace(_) => {}, // skip checking for enum_members if empty
         _ => {
             enum_members(tokens, &mut members);
         },
     }
 
-    consume(tokens, Token::RBrace);
+    consume!(tokens, Token::RBrace);
 
     Declaration::EnumDeclaration(EnumDeclaration{name, members})
 }
@@ -115,8 +118,8 @@ fn enum_members(tokens : &mut Peekable<Iter<'_, Token>>, members: &mut Vec<EnumM
     members.push(enum_member(tokens).unwrap());
     loop {
         match tokens.peek().expect("Unexpected EOF inside enum declaration") {
-            Token::Comma => {
-                consume(tokens, Token::Comma);
+            Token::Comma(_) => {
+                consume!(tokens, Token::Comma);
                 if let Some(member) = enum_member(tokens) {
                     members.push(member);
                 } else {
@@ -131,7 +134,7 @@ fn enum_members(tokens : &mut Peekable<Iter<'_, Token>>, members: &mut Vec<EnumM
 fn enum_member(tokens : &mut Peekable<Iter<'_, Token>>) -> Option<EnumMember> {
     let name = identifier(tokens);
     if let Some(name) = name {
-        consume(tokens, Token::Equal);
+        consume!(tokens, Token::Equal);
         let number = parse_number(tokens).expect("Expected Number after `=`");
         Some(EnumMember{name, number})
     } else {
@@ -140,19 +143,19 @@ fn enum_member(tokens : &mut Peekable<Iter<'_, Token>>) -> Option<EnumMember> {
 }
  
 fn struct_decl(tokens : &mut Peekable<Iter<'_, Token>>) -> Declaration {
-    consume(tokens, Token::Struct);
+    consume!(tokens, Token::Struct);
     let name = identifier(tokens).expect("Expected Identifier after `struct`");
-    consume(tokens, Token::LBrace);
+    consume!(tokens, Token::LBrace);
     let mut members = Vec::new();
 
     match tokens.peek().expect("Unexpected EOF after LBrace") {
-        Token::RBrace => {}, // skip checking for struct_members if empty
+        Token::RBrace(_) => {}, // skip checking for struct_members if empty
         _ => {
             struct_members(tokens, &mut members);
         },
     }
 
-    consume(tokens, Token::RBrace);
+    consume!(tokens, Token::RBrace);
 
     Declaration::StructDeclaration(StructDeclaration{name, members})
 }
@@ -161,8 +164,8 @@ fn struct_members(tokens : &mut Peekable<Iter<'_, Token>>, members: &mut Vec<Str
     members.push(struct_member(tokens).unwrap());
     loop {
         match tokens.peek().expect("Unexpected EOF inside struct declaration") {
-            Token::Comma => {
-                consume(tokens, Token::Comma);
+            Token::Comma(_) => {
+                consume!(tokens, Token::Comma);
                 if let Some(member) = struct_member(tokens) {
                     members.push(member);
                 } else {
@@ -176,23 +179,23 @@ fn struct_members(tokens : &mut Peekable<Iter<'_, Token>>, members: &mut Vec<Str
 fn struct_member(tokens : &mut Peekable<Iter<'_, Token>>) -> Option<StructMember> {
     let name = identifier(tokens);
     if let Some(name) = name {
-        consume(tokens, Token::Colon);
+        consume!(tokens, Token::Colon);
         let type_name = identifier(tokens).expect("Expected Type after Colon");
         Some(StructMember{name, type_name})
     } else {
         None
     }
-}
 
+}
 fn type_declaration(tokens : &mut Peekable<Iter<'_, Token>>) -> Declaration {
-    consume(tokens, Token::Type);
+    consume!(tokens, Token::Type);
     let name = identifier(tokens).expect("Expected Identifier after `type`");
     let type_name = identifier(tokens).expect("Expected type after Identifier");
     Declaration::TypeDeclaration(TypeDeclaration{name, type_name})
 }
 fn identifier(tokens : &mut Peekable<Iter<'_, Token>>) -> Option<String> {
     let result = tokens.peek().map_or(None, |x| match x {
-        Token::Identifier(s) => {
+        Token::Identifier((_, _, s)) => {
             Some(s.to_string())
             },
         _ => None
@@ -206,7 +209,7 @@ fn identifier(tokens : &mut Peekable<Iter<'_, Token>>) -> Option<String> {
 }
 fn parse_number(tokens : &mut Peekable<Iter<'_, Token>>) -> Option<u64> {
     let result = tokens.peek().map_or(None, |x| match x {
-        Token::Number(s) => Some(*s),
+        Token::Number((_, _, s)) => Some(*s),
         _ => None
     });
     if let Some(_) = result {
@@ -216,12 +219,12 @@ fn parse_number(tokens : &mut Peekable<Iter<'_, Token>>) -> Option<u64> {
 }
 
 fn protocol_declaration(tokens : &mut Peekable<Iter<'_, Token>>) -> Declaration {
-    consume(tokens, Token::Protocol);
+    consume!(tokens, Token::Protocol);
     let name = identifier(tokens).expect("Expected Identifier after `protocol`");
-    consume(tokens, Token::LBrace);
+    consume!(tokens, Token::LBrace);
     let mut interface = Vec::new();
     match tokens.peek().expect("Unexpected EOF after LBrace") {
-        Token::RBrace => {},
+        Token::RBrace(_) => {},
         _ => {
             functions(tokens, &mut interface);
         },
@@ -239,13 +242,13 @@ fn functions(tokens : &mut Peekable<Iter<'_, Token>>, interface : &mut Vec<FuncD
 }
 
 fn function(tokens : &mut Peekable<Iter<'_, Token>>) -> Option<FuncDeclaration>{
-    if let Some(Token::Fn) = tokens.peek() {
-        consume(tokens, Token::Fn);
+    if let Some(Token::Fn(_)) = tokens.peek() {
+        consume!(tokens, Token::Fn);
         let name = identifier(tokens).expect("Expected Identifier after `fn`");
-        consume(tokens, Token::LParen);
+        consume!(tokens, Token::LParen);
         let arg_list = arg_list(tokens);
-        consume(tokens, Token::RParen);
-        consume(tokens, Token::RArrow);
+        consume!(tokens, Token::RParen);
+        consume!(tokens, Token::RArrow);
         let return_type = identifier(tokens).expect("Expected return type after `->");
         Some(FuncDeclaration{name, arg_list, return_type})
     } else {
@@ -260,8 +263,8 @@ fn arg_list(tokens : &mut Peekable<Iter<'_, Token>>) -> Vec<String> {
             None => break,
             Some(i) =>{
                 result.push(i);
-                if **tokens.peek().expect("Unexpected EOF in argument list") == Token::Comma{
-                    consume(tokens, Token::Comma);
+                if let Token::Comma(_) = **tokens.peek().expect("Unexpected EOF in argument list"){
+                    consume!(tokens, Token::Comma);
                     match tokens.peek().expect("Unexpected EOF in argument list") {
                         Token::Identifier(_) => {},
                         _ => panic!("Unexpected symbol after Comma in argument list"),

dev/src/idl/protocol.rs

@@ -1,17 +1,167 @@
-pub enum ProtocolTypes {
-    Byte,
+use std::collections::HashMap;
+use crate::idl::types::Type;
+
+#[derive(Debug, Clone, PartialEq)]
+pub struct Function{
+    pub arguments : Vec<String>,
+}
+#[derive(Debug, Clone, PartialEq)]
+pub struct Protocol{
+    interface : HashMap<String, Function>,
 }
 
-pub struct Protocol {}
+#[derive(Debug, Clone, PartialEq)]
+pub struct Protocols {
+    protocols : HashMap<String, Protocol>,
+    symbol_table : HashMap<String, Type>,
+}
+
+#[derive(Debug, Clone, PartialEq)]
+pub enum ValidationError{
+    IncorrectVersion,
+    InvalidHeader,
+    FunctionDoesNotExist,
+    ProtocolDoesNotExist,
+    InvalidSize,
+    InvalidArgument,
+    NonExistentType(String),
+}
+
+impl Protocols {
+    pub fn new(symbol_table: HashMap<String, Type>) -> Self {
+        let protocols = HashMap::new();
+        Self { protocols, symbol_table }
+    }
+    
+    pub fn add_protocol(&mut self, name : String, interface : HashMap<String, Function>) {
+        self.protocols.insert(name, Protocol::new(interface));
+    }
+
+    pub fn validate(&self, protocol_name : &str, function_name : &str, data : Vec<u8>) -> Result<(), ValidationError>{
+        match self.protocols.get(protocol_name) {
+            Some(s) => s.validate(function_name, data, &self.symbol_table),
+            None => Err(ValidationError::ProtocolDoesNotExist),
+        }
+    }
+
+}
+
+
+
 impl Protocol {
-    pub fn is_empty(&self) -> bool {
-        true
+    pub fn new(interface: HashMap<String, Function>) -> Self {
+        Self {interface}
+    }
+    fn validate(&self, function_name : &str, data : Vec<u8>, symbols : &HashMap<String, Type>) -> Result<(), ValidationError> {
+        match self.interface.get(function_name){
+            Some(s) => s.validate(data, symbols),
+            None => Err(ValidationError::FunctionDoesNotExist),
+        }
+    }
 }
 
-    pub fn validate_data(&self, data: Vec<u8>) -> bool {
-        if !data.is_empty() && self.is_empty() {
-            return false;
+impl Function {
+    fn validate(&self, data : Vec<u8>, symbols : &HashMap<String, Type>) -> Result<(), ValidationError> {
+        let mut types = Vec::new();
+        for arg in self.arguments.iter() {
+            let type_value = symbols.get(arg);
+            if let Some(type_value) = type_value {
+                types.push(type_value);
             }
-        true
+            else{
+                return Err(ValidationError::NonExistentType(arg.to_string()));
             }
         }
+        let mut data = data.iter();
+        if let Some(ver) = data.next() {
+            if *ver == 1 {
+                // We got the correct version number
+                // Now to parse individual argumebts
+                let mut types = types.iter().peekable();
+                loop {
+                    let type_byte = data.next();
+                    if let Some(type_byte) = type_byte {
+                        let type_value = types.next();
+                        if type_value.is_none() {
+                            return Err(ValidationError::InvalidSize);
+                        }
+                        let type_value = type_value.unwrap();
+                        let data_type = match type_byte {
+                            0 => Some(Type::U64),
+                            1 => Some(Type::U32),
+                            2 => Some(Type::U16),
+                            3 => Some(Type::U8),
+                            4 => Some(Type::I64),
+                            5 => Some(Type::I32),
+                            6 => Some(Type::I16),
+                            7 => Some(Type::I8),
+                            8 => Some(Type::Bool),
+                            9 => Some(Type::F32),
+                            10 => Some(Type::F64),
+                            11 => Some(Type::Str),
+                            _ => None,
+                        };
+                        if data_type.is_none() || data_type.as_ref().unwrap() != *type_value {
+                            println!("{:#?}", *type_value);
+                            return Err(ValidationError::InvalidArgument);
+                        }
+
+                        match data_type.unwrap(){
+                            Type::U64 | Type::I64 | Type::F64 => {
+                                data.next();
+                                data.next();
+                                data.next();
+                                data.next();
+                                data.next();
+                                data.next();
+                                data.next();
+                                data.next();
+                            },
+                            Type::U32 | Type::I32 | Type::F32 => {
+                                data.next();
+                                data.next();
+                                data.next();
+                                data.next();
+                            },
+                            Type::U16 | Type::I16 => {
+                                data.next();
+                                data.next();
+                            },
+                            Type::U8  | Type::I8  | Type::Bool => {
+                                data.next();
+                            },
+                            Type::Str => todo!(),
+                            _ => panic!("Should not be possible"),
+                        }
+                    } else if types.peek().is_none() {
+                        return Ok(());
+                    }
+                    break;
+                }
+            }
+            else {
+                return Err(ValidationError::IncorrectVersion);
+            }
+        } else {
+            return Err(ValidationError::InvalidHeader);
+        }
+        Ok(())
+    }
+}
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+

dev/src/idl/types.rs

@@ -0,0 +1,98 @@
+use std::collections::HashMap;
+
+use crate::idl::parser::AST;
+
+use super::protocol::{Function, Protocols};
+
+#[derive(Debug, Clone, PartialEq)]
+pub enum Type {
+    U64,
+    U32,
+    U16,
+    U8,
+
+    I64,
+    I32,
+    I16,
+    I8,
+
+    Bool,
+
+    F32,
+    F64,
+    Str,
+
+    Struct(StructType),
+    Enum(EnumType),
+    Alias(String),
+}
+
+#[derive(Debug, Clone, PartialEq)]
+pub struct StructType {
+    members : HashMap<String, String>
+}
+
+#[derive(Debug, Clone, PartialEq)]
+pub struct EnumType {
+    members : HashMap<String, u8>
+}
+
+
+fn add_builtin_types(symbol_table : &mut HashMap<String, Type>) {
+    symbol_table.insert("u8".to_string(), Type::U8);
+    symbol_table.insert("u16".to_string(), Type::U16);
+    symbol_table.insert("u32".to_string(), Type::U32);
+    symbol_table.insert("u64".to_string(), Type::U64);
+    symbol_table.insert("i8".to_string(), Type::I8);
+    symbol_table.insert("i16".to_string(), Type::I16);
+    symbol_table.insert("i32".to_string(), Type::I32);
+    symbol_table.insert("i64".to_string(), Type::I64);
+    symbol_table.insert("bool".to_string(), Type::Bool);
+    symbol_table.insert("f32".to_string(), Type::F32);
+    symbol_table.insert("f64".to_string(), Type::F64);
+}
+
+pub fn get_protocols(ast : AST) -> Protocols{
+    let mut symbol_table : HashMap<String, Type> = HashMap::new();
+    let declarations = ast.0;
+    add_builtin_types(&mut symbol_table);
+    // First Pass
+    // We just populate the symbol table here
+    for decl in declarations.iter() {
+        match decl{
+            super::parser::Declaration::EnumDeclaration(e) => {
+                let mut members = HashMap::new();
+                for m in e.members.iter(){
+                    members.insert(m.name.to_string(), m.number as u8);
+                }
+                symbol_table.insert(e.name.to_string(), Type::Enum(EnumType{members}));
+            },
+            super::parser::Declaration::StructDeclaration(s) => {
+                let mut members = HashMap::new();
+                for m in s.members.iter() {
+                    members.insert(m.name.to_string(), m.type_name.to_string());
+                }
+                symbol_table.insert(s.name.to_string(), Type::Struct(StructType{members}));
+            },
+            super::parser::Declaration::TypeDeclaration(t) => {
+                symbol_table.insert(t.name.to_string(), Type::Alias(t.type_name.to_string()));
+            },
+            super::parser::Declaration::ProtocolDeclaration(_) => {},
+        }
+    }
+
+    let mut protocols = Protocols::new(symbol_table);
+    for decl in declarations.iter(){
+        match decl {
+            super::parser::Declaration::ProtocolDeclaration(p) => {
+                let mut funcs : HashMap<String, Function> = HashMap::new();
+                for i in p.interface.iter(){
+                    funcs.insert(i.name.to_string(), Function{arguments : i.arg_list.clone()});
+                }
+                protocols.add_protocol(p.name.to_string(), funcs);
+            },
+            _ => {}
+        }
+    }
+    protocols
+}

sysdata/idl/test/src/protocol.aldi

@@ -0,0 +1,3 @@
+protocol Foo{
+	fn bar(i32, u8, bool) -> void;	
+}