[move-compiler-v2] Print a better error message for a string without …

…b or x (#14687)

Fixes #14364.

third_party/move/move-compiler-v2/tests/folding/constants_hexstring.exp

@@ -0,0 +1,13 @@
+
+Diagnostics:
+error: invalid hex string
+  ┌─ tests/folding/constants_hexstring.move:6:29
+  │
+6 │     const C12: vector<u8> = x"hello";
+  │                             ^^^^^^^^ Odd number of characters in hex string. Expected 2 hexadecimal digits for each byte
+
+error: invalid hex string
+  ┌─ tests/folding/constants_hexstring.move:7:31
+  │
+7 │     const C13: vector<u8> = x"hello!";
+  │                               ^ Invalid hexadecimal character: 'h'

third_party/move/move-compiler-v2/tests/folding/constants_hexstring.move

@@ -0,0 +1,9 @@
+address 0x42 {
+module M {
+    fun u(): u64 { 0 }
+
+    const C11: vector<u8> = x"deadbeef";
+    const C12: vector<u8> = x"hello";
+    const C13: vector<u8> = x"hello!";
+}
+}

third_party/move/move-compiler-v2/tests/folding/constants_quoted_string.exp

@@ -0,0 +1,7 @@
+
+Diagnostics:
+error: invalid byte string
+  ┌─ tests/folding/constants_quoted_string.move:5:29
+  │
+5 │     const C12: vector<u8> = "<empty>";
+  │                             ^ String literal must begin with b" (for a byte string) or x" (for a hex string)

third_party/move/move-compiler-v2/tests/folding/constants_quoted_string.move

@@ -0,0 +1,8 @@
+address 0x42 {
+module M {
+    fun u(): u64 { 0 }
+
+    const C12: vector<u8> = "<empty>";
+    const C13: vector<u8> = "foo;
+}
+}

third_party/move/move-compiler-v2/tests/folding/constants_simple.move

@@ -11,6 +11,9 @@ module M {
     const C7: u256 = 4 / 3 + 4 - 1 << 143;
     const C8: u16 = 123;
     const C9: u32 = (453 as u32);
-
+    const C10: vector<u8> = b"<empty>";
+    const C11: vector<u8> = x"deadbeef";
+    const C12: vector<u8> = b"\"foo\"";
+    const C13: vector<u8> = b"\"\x48";
 }
 }

third_party/move/move-compiler-v2/tests/folding/constants_single_quote.exp

@@ -0,0 +1,7 @@
+
+Diagnostics:
+error: invalid character
+  ┌─ tests/folding/constants_single_quote.move:4:29
+  │
+4 │     const C13: vector<u8> = "foo
+  │                             ^ Invalid character: '"'; string literal must begin with `b"` and closing quote `"` must appear on same line

third_party/move/move-compiler-v2/tests/folding/constants_single_quote.move

@@ -0,0 +1,10 @@
+address 0x42 {
+module M {
+    fun u(): u64 { 0 }
+    const C13: vector<u8> = "foo
+    ";
+    const C14: vector<u8> = "bar";
+    const C15: vector<u8> = b"baz";
+    const C16: vector<u8> = x"quux";
+}
+}

third_party/move/move-compiler-v2/tests/folding/constants_string_non_utf8.exp

@@ -0,0 +1,7 @@
+
+Diagnostics:
+error: invalid character
+  ┌─ tests/folding/constants_string_non_utf8.move:5:31
+  │
+5 │     const C12: vector<u8> = b"ÿ";
+  │                               ^ Invalid character 'ÿ' found when reading file. Only ASCII printable characters, tabs (\t), lf (\n) and crlf (\r\n) are permitted.

third_party/move/move-compiler-v2/tests/folding/constants_string_non_utf8.move

@@ -0,0 +1,7 @@
+address 0x42 {
+module M {
+    fun u(): u64 { 0 }
+
+    const C12: vector<u8> = b"ÿ";
+}
+}

third_party/move/move-compiler/src/parser/lexer.rs

@@ -478,6 +478,25 @@ fn find_token(
                 (get_name_token(&text[..len]), len)
             }
         },
+        '"' => {
+            let line = &text.lines().next().unwrap()[1..];
+            match get_string_len(line) {
+                Some(_last_quote) => {
+                    let loc = make_loc(file_hash, start_offset, start_offset);
+                    return Err(Box::new(diag!(
+                        Syntax::InvalidByteString,
+                        (loc, "String literal must begin with b\" (for a byte string) or x\" (for a hex string)")
+                    )));
+                },
+                None => {
+                    let loc = make_loc(file_hash, start_offset, start_offset);
+                    return Err(Box::new(diag!(
+                        Syntax::InvalidCharacter,
+                        (loc, format!("Invalid character: '{}'; string literal must begin with `b\"` and closing quote `\"` must appear on same line", c))
+                    )));
+                },
+            }
+        },
         '&' => {
             if text.starts_with("&mut ") {
                 (Tok::AmpMut, 5)

third_party/move/move-compiler/tests/move_check/parser/constants_hexstring.exp

@@ -0,0 +1,24 @@
+error[E01008]: invalid hex string
+  ┌─ tests/move_check/parser/constants_hexstring.move:6:29
+  │
+6 │     const C12: vector<u8> = x"hello";
+  │                             ^^^^^^^^ Odd number of characters in hex string. Expected 2 hexadecimal digits for each byte
+
+error[E08001]: cannot compute constant value
+  ┌─ tests/move_check/parser/constants_hexstring.move:6:29
+  │
+6 │     const C12: vector<u8> = x"hello";
+  │                             ^^^^^^^^ Invalid expression in 'const'. This expression could not be evaluated to a value
+
+error[E08001]: cannot compute constant value
+  ┌─ tests/move_check/parser/constants_hexstring.move:7:29
+  │
+7 │     const C13: vector<u8> = x"hello!";
+  │                             ^^^^^^^^^ Invalid expression in 'const'. This expression could not be evaluated to a value
+
+error[E01008]: invalid hex string
+  ┌─ tests/move_check/parser/constants_hexstring.move:7:31
+  │
+7 │     const C13: vector<u8> = x"hello!";
+  │                               ^ Invalid hexadecimal character: 'h'
+

third_party/move/move-compiler/tests/move_check/parser/constants_hexstring.move

@@ -0,0 +1,9 @@
+address 0x42 {
+module M {
+    fun u(): u64 { 0 }
+
+    const C11: vector<u8> = x"deadbeef";
+    const C12: vector<u8> = x"hello";
+    const C13: vector<u8> = x"hello!";
+}
+}

third_party/move/move-compiler/tests/move_check/parser/constants_quoted_string.exp

@@ -0,0 +1,6 @@
+error[E01007]: invalid byte string
+  ┌─ tests/move_check/parser/constants_quoted_string.move:5:29
+  │
+5 │     const C12: vector<u8> = "<empty>";
+  │                             ^ String literal must begin with b" (for a byte string) or x" (for a hex string)
+

third_party/move/move-compiler/tests/move_check/parser/constants_quoted_string.move

@@ -0,0 +1,8 @@
+address 0x42 {
+module M {
+    fun u(): u64 { 0 }
+
+    const C12: vector<u8> = "<empty>";
+    const C13: vector<u8> = "foo;
+}
+}

third_party/move/move-compiler/tests/move_check/parser/constants_simple.move

@@ -11,6 +11,9 @@ module M {
     const C7: u256 = 4 / 3 + 4 - 1 << 143;
     const C8: u16 = 123;
     const C9: u32 = (453 as u32);
-
+    const C10: vector<u8> = b"<empty>";
+    const C11: vector<u8> = x"deadbeef";
+    const C12: vector<u8> = b"\"foo\"";
+    const C13: vector<u8> = b"\"\x48";
 }
 }

third_party/move/move-compiler/tests/move_check/parser/constants_single_quote.exp

@@ -0,0 +1,6 @@
+error[E01001]: invalid character
+  ┌─ tests/move_check/parser/constants_single_quote.move:4:29
+  │
+4 │     const C13: vector<u8> = "foo
+  │                             ^ Invalid character: '"'; string literal must begin with `b"` and closing quote `"` must appear on same line
+

third_party/move/move-compiler/tests/move_check/parser/constants_single_quote.move

@@ -0,0 +1,10 @@
+address 0x42 {
+module M {
+    fun u(): u64 { 0 }
+    const C13: vector<u8> = "foo
+    ";
+    const C14: vector<u8> = "bar";
+    const C15: vector<u8> = b"baz";
+    const C16: vector<u8> = x"quux";
+}
+}

third_party/move/move-compiler/tests/move_check/parser/constants_string_non_utf8.exp

@@ -0,0 +1,6 @@
+error[E01001]: invalid character
+  ┌─ tests/move_check/parser/constants_string_non_utf8.move:5:31
+  │
+5 │     const C12: vector<u8> = b"ÿ";
+  │                               ^ Invalid character 'ÿ' found when reading file. Only ASCII printable characters, tabs (\t), lf (\n) and crlf (\r\n) are permitted.
+

third_party/move/move-compiler/tests/move_check/parser/constants_string_non_utf8.move

@@ -0,0 +1,7 @@
+address 0x42 {
+module M {
+    fun u(): u64 { 0 }
+
+    const C12: vector<u8> = b"ÿ";
+}
+}