lexer: recognise all tokens in laravel/framework

meta/run_on_files.sh

@@ -6,7 +6,7 @@ dir=$(realpath $1)
 
 for file in $(find $dir -name "*.php")
 do
-    cargo run -- $file --lexer
+    cargo run -q -- $file --lexer
 
     if [ $? -ne 0 ]
     then

phpast/src/main.rs

@@ -8,29 +8,45 @@ use trunk_parser::Parser;
 #[structopt(name = "phpast", about = "Generate an abstract syntax tree from a PHP file.")]
 struct Args {
     #[structopt(parse(from_os_str), help = "The input file to use.")]
-    file: PathBuf,
+    file: Option<PathBuf>,
 
     #[structopt(short, long, help = "Output the abstract syntax tree as JSON.")]
     json: bool,
 
     #[structopt(short, long, help = "Only execute the lexer on the source file.")]
     lexer: bool,
+
+    #[structopt(short, long, help = "Provide a string to execute.")]
+    run: Option<String>,
+
+    #[structopt(short, long, help = "Dump tokens.")]
+    dump_tokens: bool,
 }
 
 fn main() {
     let args = Args::from_args();
     
-    let input = match std::fs::read_to_string(args.file) {
+    let input = if args.file.is_some() {
+        match std::fs::read_to_string(args.file.unwrap()) {
             Ok(contents) => contents,
             Err(e) => {
                 eprintln!("{}", e);
                 exit(1);
             },
+        }
+    } else if args.run.is_some() {
+        args.run.unwrap()
+    } else {
+        panic!("boo!");
     };
 
     let mut lexer = Lexer::new(None);
     let tokens = lexer.tokenize(&input[..]).unwrap();
 
+    if args.dump_tokens {
+        dbg!(&tokens);
+    }
+
     if args.lexer {
         return;
     }

trunk_lexer/src/lexer.rs

@@ -156,6 +156,23 @@ impl Lexer {
         let char = it.next().unwrap();
 
         let kind = match char {
+            '!' => {
+                self.col += 1;
+                TokenKind::Bang
+            },
+            '&' => {
+                self.col += 1;
+
+                if let Some('&') = it.peek() {
+                    self.col += 1;
+
+                    it.next();
+
+                    TokenKind::BooleanAnd
+                } else {
+                    TokenKind::BitAnd
+                }
+            },
             '?' => {
                 // This is a close tag, we can enter "Initial" mode again.
                 if let Some('>') = it.peek() {
@@ -167,7 +184,7 @@ impl Lexer {
 
                     TokenKind::CloseTag
                 } else {
-                    todo!();
+                    TokenKind::Question
                 }
             },
             '=' => {
@@ -229,6 +246,8 @@ impl Lexer {
                         self.col += 1;
                     }
 
+                    escaping = false;
+
                     buffer.push(*n);
                     it.next();
                 }
@@ -329,7 +348,7 @@ impl Lexer {
                     }
                 }
 
-                if (is_float) {
+                if is_float {
                     TokenKind::Float(buffer.parse().unwrap())
                 } else {
                     TokenKind::Int(buffer.parse().unwrap())
@@ -465,7 +484,29 @@ impl Lexer {
             },
             '<' => {
                 self.col += 1;
+
+                if let Some('=') = it.peek() {
+                    it.next();
+
+                    self.col += 1;
+
+                    TokenKind::LessThanEquals
+                } else {
                     TokenKind::LessThan
+                }
+            },
+            '>' => {
+                self.col += 1;
+
+                if let Some('=') = it.peek() {
+                    it.next();
+
+                    self.col += 1;
+
+                    TokenKind::GreaterThanEquals
+                } else {
+                    TokenKind::GreaterThan
+                }
             },
             ',' => {
                 self.col += 1;

trunk_lexer/src/token.rs

@@ -102,6 +102,11 @@ pub enum TokenKind {
     FullyQualifiedIdentifier(String),
     QualifiedIdentifier(String),
     Colon,
+    Caret,
+    Question,
+    Bang,
+    And,
+    BitAnd,
 }
 
 #[derive(Debug, Clone)]