fix(parser): fix lexing escaped and unicode characters in block strings

fixes #341 fixes #342
apollographql · Nov 10, 2022 · c6e1339 · c6e1339
1 parent c5e2f95
commit c6e1339
Show file tree

Hide file tree

Showing 2 changed files with 42 additions and 18 deletions.
diff --git a/crates/apollo-parser/src/lexer/mod.rs b/crates/apollo-parser/src/lexer/mod.rs
@@ -178,7 +178,6 @@ impl Cursor<'_> {
                     } else if is_line_terminator(c) {
                         self.add_err(Error::new("unexpected line terminator", c.to_string()));
                     }
-                    was_backslash = c == '\\';
                 }
 
                 if !buf.ends_with('"') {
@@ -214,19 +213,27 @@ impl Cursor<'_> {
 
             while !self.is_eof() {
                 let c = self.bump().unwrap();
-                if c == '"' {
-                    buf.push(c);
+                let was_backslash = c == '\\';
+
+                if was_backslash && !is_escaped_char(c) && c != 'u' {
+                    self.add_err(Error::new("unexpected escaped character", c.to_string()));
+                }
+
+                buf.push(c);
+
+                if was_backslash {
+                    while self.first() == '"' {
+                        buf.push(self.first());
+                        self.bump();
+                    }
+                } else if c == '"' {
                     if ('"', '"') == (self.first(), self.second()) {
                         buf.push(self.first());
                         buf.push(self.second());
                         self.bump();
                         self.bump();
                         break;
                     }
-                } else if is_source_char(c) {
-                    buf.push(c);
-                } else {
-                    break;
                 }
             }
         }
@@ -437,9 +444,9 @@ fn is_escaped_char(c: char) -> bool {
 
 // SourceCharacter
 //     /[\u0009\u000A\u000D\u0020-\uFFFF]/
-fn is_source_char(c: char) -> bool {
-    matches!(c, '\t' | '\r' | '\n' | '\u{0020}'..='\u{FFFF}')
-}
+// fn is_source_char(c: char) -> bool {
+//     matches!(c, '\t' | '\r' | '\n' | '\u{0020}'..='\u{FFFF}')
+// }
 
 #[cfg(test)]
 mod test {

diff --git a/crates/apollo-parser/test_data/lexer/ok/0008_block_string.txt b/crates/apollo-parser/test_data/lexer/ok/0008_block_string.txt
@@ -5,11 +5,28 @@ WHITESPACE@47:48 " "
 NAME@48:54 "Filter"
 WHITESPACE@54:55 " "
 L_CURLY@55:56 "{"
-WHITESPACE@56:61 "\n    "
-NAME@61:66 "title"
-COLON@66:67 ":"
-WHITESPACE@67:68 " "
-NAME@68:74 "String"
-WHITESPACE@74:75 "\n"
-R_CURLY@75:76 "}"
-EOF@76:76
+WHITESPACE@56:59 "\n  "
+STRING_VALUE@59:93 "\"\"\"unicode in block string 🤷\"\"\""
+WHITESPACE@93:98 "\n    "
+NAME@98:103 "title"
+COLON@103:104 ":"
+WHITESPACE@104:105 " "
+NAME@105:111 "String"
+WHITESPACE@111:112 "\n"
+R_CURLY@112:113 "}"
+WHITESPACE@113:115 "\n\n"
+STRING_VALUE@115:136 "\"\"\"\n\\\"\"\" a/b \\\"\"\"\n\"\"\""
+WHITESPACE@136:137 "\n"
+NAME@137:142 "input"
+WHITESPACE@142:143 " "
+NAME@143:149 "Filter"
+WHITESPACE@149:150 " "
+L_CURLY@150:151 "{"
+WHITESPACE@151:156 "\n    "
+NAME@156:161 "title"
+COLON@161:162 ":"
+WHITESPACE@162:163 " "
+NAME@163:169 "String"
+WHITESPACE@169:170 "\n"
+R_CURLY@170:171 "}"
+EOF@171:171