From 758f551db0e231fd29a9aeaa0b93ffad5c795fcc Mon Sep 17 00:00:00 2001
From: Araozu <fernando@araozu.dev>
Date: Sat, 16 Nov 2024 17:08:07 -0500
Subject: [PATCH] feat: structure for more number lexers

---
 src/01_lexic/number.zig | 91 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 87 insertions(+), 4 deletions(-)

diff --git a/src/01_lexic/number.zig b/src/01_lexic/number.zig
index 7885b6f..f36a849 100644
--- a/src/01_lexic/number.zig
+++ b/src/01_lexic/number.zig
@@ -7,7 +7,55 @@ const TokenType = token.TokenType;
 
 const is_decimal_digit = utils.is_decimal_digit;
 
-fn integer(input: []const u8, cap: usize, start: usize) !?Token {
+const LexReturn = struct { Token, usize };
+
+/// Attempts to lex a number, as per the language grammar.
+///
+/// A number is either an Int or a Float.
+/// No number can have a leading zero. That is an error to
+/// avoid confussion with PHP literal octals.
+fn lex(input: []const u8, cap: usize, start: usize) !?LexReturn {
+    const first_char = input[start];
+
+    // Attempt to lex a hex, octal or binary number
+    if (first_char == '0' and cap > start + 1) {
+        const second_char = input[start + 1];
+        switch (second_char) {
+            'x', 'X' => return hex(),
+            'o', 'O' => return octal(),
+            'b', 'B' => return binary(),
+        }
+
+        // Leading zero found. Throw an error.
+        // TODO: throw an error :c
+    }
+
+    // Attempt to lex an integer.
+    // Floating point numbers are lexed through the int lexer
+    return integer(input, cap, start);
+}
+
+fn hex() !?LexReturn {
+    return null;
+}
+
+fn octal() !?LexReturn {
+    return null;
+}
+
+fn binary() !?LexReturn {
+    return null;
+}
+
+/// Attempts to lex an integer number.
+///
+/// This function fails if the first digit it encounters is a `0`,
+/// this is because it could cause confusion with PHP literal integers,
+/// where a number that starts with a `0` is octal, not decimal.
+///
+/// For this reason, this function should be called after the lexers
+/// for hex, octal and binary have been called.
+fn integer(input: []const u8, cap: usize, start: usize) !?LexReturn {
     const first_char = input[start];
     if (!is_decimal_digit(first_char)) {
         return null;
@@ -18,16 +66,51 @@ fn integer(input: []const u8, cap: usize, start: usize) !?Token {
         last_pos += 1;
     }
 
-    return Token.init(input[start..last_pos], TokenType.Int, start);
+    return .{
+        Token.init(input[start..last_pos], TokenType.Int, start),
+        last_pos,
+    };
 }
 
-test "number lexer" {
+test "int lexer 1" {
     const input = "322   ";
     const result = try integer(input, input.len, 0);
 
-    if (result) |r| {
+    if (result) |tuple| {
+        const r = tuple[0];
         try std.testing.expectEqualDeep("322", r.value);
     } else {
         try std.testing.expect(false);
     }
 }
+
+test "int lexer 2" {
+    const input = "   644   ";
+    const result = try integer(input, input.len, 3);
+
+    if (result) |tuple| {
+        const r = tuple[0];
+        try std.testing.expectEqualDeep("644", r.value);
+    } else {
+        try std.testing.expect(false);
+    }
+}
+
+test "int lexer 3" {
+    const input = "4";
+    const result = try integer(input, input.len, 0);
+
+    if (result) |tuple| {
+        const r = tuple[0];
+        try std.testing.expectEqualDeep("4", r.value);
+    } else {
+        try std.testing.expect(false);
+    }
+}
+
+test "should return null if not an integer" {
+    const input = "prosor prosor";
+    const result = try integer(input, input.len, 0);
+
+    try std.testing.expect(result == null);
+}