From 71d617928a6038fbd1784c907f3e91facfc9224c Mon Sep 17 00:00:00 2001
From: Araozu <fernando@araozu.dev>
Date: Wed, 27 Nov 2024 20:07:01 -0500
Subject: [PATCH] feat: lex escape characters inside strings

---
 src/01_lexic/string.zig | 90 ++++++++++++++++++++++++++++++++++++-----
 1 file changed, 79 insertions(+), 11 deletions(-)

diff --git a/src/01_lexic/string.zig b/src/01_lexic/string.zig
index 1d673f3..b5ae7d1 100644
--- a/src/01_lexic/string.zig
+++ b/src/01_lexic/string.zig
@@ -19,21 +19,43 @@ pub fn lex(input: []const u8, start: usize) LexError!?LexReturn {
 
     // lex everything but quote and newline
     // TODO: escape characters
+
     var current_pos = start + 1;
-    while (current_pos < cap and input[current_pos] != '"' and input[current_pos] != '\n') {
+
+    while (current_pos < cap) {
+        const next_char = input[current_pos];
+        // string is finished, return it
+        if (next_char == '"') {
+            return .{
+                Token.init(input[start .. current_pos + 1], TokenType.String, start),
+                current_pos + 1,
+            };
+        }
+        // new line, return error
+        else if (next_char == '\n') {
+            return LexError.IncompleteString;
+        }
+        // lex escape characters
+        else if (next_char == '\\') {
+            // if next char is EOF, return error
+            if (current_pos + 1 == cap) {
+                return LexError.IncompleteString;
+            }
+            // if next char is newline, return error
+            else if (input[current_pos + 1] == '\n') {
+                return LexError.IncompleteString;
+            }
+            // here just consume whatever char is after
+            // TODO: if next char is not an escape char, return warning?
+            current_pos += 2;
+            continue;
+        }
+
         current_pos += 1;
     }
 
-    // expect ending quote
-    if (current_pos == cap or input[current_pos] == '\n') {
-        // Error: EOF before ending the string
-        return LexError.IncompleteString;
-    } else {
-        return .{
-            Token.init(input[start .. current_pos + 1], TokenType.String, start),
-            current_pos + 1,
-        };
-    }
+    // this could only reach when EOF is hit, return error
+    return LexError.IncompleteString;
 }
 
 test "should lex empty string" {
@@ -100,3 +122,49 @@ test "should fail on newline before closing string" {
 
     try std.testing.expect(false);
 }
+
+test "should lex string with escape character 1" {
+    const input = "\"test\\\"string\"";
+    const output = try lex(input, 0);
+
+    if (output) |tuple| {
+        const t = tuple[0];
+        try std.testing.expectEqualDeep("\"test\\\"string\"", t.value);
+        try std.testing.expectEqual(14, tuple[1]);
+    } else {
+        try std.testing.expect(false);
+    }
+}
+
+test "should lex string with escape character 2" {
+    const input = "\"test\\\\string\"";
+    const output = try lex(input, 0);
+
+    if (output) |tuple| {
+        const t = tuple[0];
+        try std.testing.expectEqualDeep("\"test\\\\string\"", t.value);
+        try std.testing.expectEqual(14, tuple[1]);
+    } else {
+        try std.testing.expect(false);
+    }
+}
+
+test "should fail on EOF after backslash" {
+    const input = "\"hello \\";
+    _ = lex(input, 0) catch |err| {
+        try std.testing.expectEqual(LexError.IncompleteString, err);
+        return;
+    };
+
+    try std.testing.expect(false);
+}
+
+test "should fail on newline after backslash" {
+    const input = "\"hello \\\n";
+    _ = lex(input, 0) catch |err| {
+        try std.testing.expectEqual(LexError.IncompleteString, err);
+        return;
+    };
+
+    try std.testing.expect(false);
+}