feat(assembler): Implement lexer

2 files changed, 101 insertions, 1 deletions

diff --git a/assembler/src/lexer/mod.rs b/assembler/src/lexer/mod.rs
new file mode 100644
index 0000000..617aca1
--- /dev/null
+++ b/assembler/src/lexer/mod.rs
@@ -0,0 +1,85 @@
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub(crate) enum Token {
+    Label(String),
+    Operator(String),
+    Symbol(String),
+    Number(i64),
+    String(String),
+}
+
+pub fn split(mut line: String) -> Vec<Token> {
+    line.push('\n');
+
+    let mut res = Vec::new();
+    let mut buffer = String::new();
+
+    let mut has_label = false;
+    let mut string_started = false;
+    let mut skip_next = false;
+
+    for c in line.chars() {
+        buffer.push(c);
+        if skip_next || (string_started && c != '"') {
+            skip_next = false;
+            continue;
+        }
+
+        match c {
+            ':' => if ! has_label {
+                    buffer.pop();
+                    res.push(Token::Label(buffer.clone()));
+                    buffer.clear();
+                    has_label = true;
+                }
+                else {
+                    panic!("Found double labels!");
+                },
+            ' ' | '\t' => {
+                buffer.pop();
+                if buffer.is_empty() { continue; }
+
+                if (has_label && (res.len() == 1)) ||
+                   (!has_label && (res.is_empty())) {
+                    res.push(Token::Operator(buffer.clone()));
+                    buffer.clear();
+                }
+            },
+            '"' => {
+                buffer.pop();
+                if string_started {
+                    res.push(Token::String(buffer.clone()));
+                    buffer.clear();
+                }
+                string_started = !string_started;
+            },
+            ',' | '\n' => {
+                buffer.pop();
+                if buffer.is_empty() { continue; }
+
+                if let Some(num) = buffer.parse::<i64>().ok() {
+                    res.push(Token::Number(num));
+                }
+                else {
+                    res.push(Token::Symbol(buffer.clone()));
+                }
+                buffer.clear();
+            },
+            '\\' => skip_next = true,
+            ';' => break,
+            '=' => if buffer.is_empty() {
+                    panic!("Cannot begin the line with an =")
+                }
+                else if let Token::Operator(name) = &res[0] {
+                    res[0] = Token::Symbol(name.clone());
+                    buffer.clear();
+                }
+                else {
+                    panic!("Before an = you may only have a symbol name!");
+                },
+            'A'..='Z' | 'a'..='z' | '0'..='9' | '_' | '-' | '.' | '$' | '?'
+                => {},
+            _ => panic!("Found invalid character '{}'", c),
+        }
+    }
+    res
+}
diff --git a/assembler/src/main.rs b/assembler/src/main.rs
index e7a11a9..0cda74e 100644
--- a/assembler/src/main.rs
+++ b/assembler/src/main.rs
@@ -1,3 +1,18 @@
+mod lexer;
+
+use std::fs::File;
+use std::io::{BufRead, BufReader};
+
 fn main() {
-    println!("Hello, world!");
+    let file_path = "./example.s";
+
+    let file = File::open(file_path).unwrap();
+    let lines = BufReader::new(file).lines();
+
+    let mut tokens = Vec::new();
+    for line in lines.map_while(Result::ok) {
+        tokens.extend(lexer::split(line));
+    }
+
+    println!("{:#?}", tokens);
 }