From df58de52f46534ecc301acf08549c99bb6237e41 Mon Sep 17 00:00:00 2001
From: Richard Newman <rnewman@twinql.com>
Date: Thu, 15 Mar 2018 07:13:27 -0700
Subject: [PATCH] Correctly parse and unescape quotes etc. inside EDN strings.
 (#434) (#589)

---
 edn/src/edn.rustpeg | 20 +++++++++++++-------
 edn/tests/tests.rs  |  8 ++++++++
 2 files changed, 21 insertions(+), 7 deletions(-)

diff --git a/edn/src/edn.rustpeg b/edn/src/edn.rustpeg
index 3e0e8674..729d8176 100644
--- a/edn/src/edn.rustpeg
+++ b/edn/src/edn.rustpeg
@@ -69,14 +69,20 @@ pub float -> SpannedValue = f:$(sign? digit+ ("." digit+)? ([eE] sign? digit+)?)
 
 number -> SpannedValue = ( bigint / basedinteger / hexinteger / octalinteger / integer / float )
 
-// TODO: \newline, \return, \space and \tab
-special_char = quote / tab
-quote = "\\\""
-tab = "\\tab"
-char = [^"] / special_char
+// TODO: standalone characters: \<char>, \newline, \return, \space and \tab.
 
-pub text -> SpannedValue = "\"" t:$( char* ) "\""
-    {  SpannedValue::Text(t.to_string()) }
+string_special_char -> &'input str = "\\" $([\\"ntr])
+string_normal_chars -> &'input str = $([^"\\]+)
+
+// This is what we need to do in order to unescape. We can't just match the entire string slice:
+// we get a Vec<&str> from rust-peg, where some of the parts might be unescaped special characters,
+// and we join it together to form an output string.
+// E.g., input = r#"\"foo\\\\bar\""#
+//      output = [quote, "foo", backslash, "bar", quote]
+//      result = r#""foo\\bar""#
+// For the typical case, string_normal_chars will match multiple, leading to a single-element vec.
+pub text -> SpannedValue = "\"" t:((string_special_char / string_normal_chars)*) "\""
+    {  SpannedValue::Text(t.join(&"").to_string()) }
 
 // RFC 3339 timestamps. #inst "1985-04-12T23:20:50.52Z"
 // We accept an arbitrary depth of decimals.
diff --git a/edn/tests/tests.rs b/edn/tests/tests.rs
index b2e37597..48692467 100644
--- a/edn/tests/tests.rs
+++ b/edn/tests/tests.rs
@@ -327,6 +327,14 @@ fn test_text() {
 
     assert!(text("\"").is_err());
     assert!(text("nil").is_err());
+
+    let raw_edn = r#""This string contains a \" and a \\""#;
+    let raw_string = r#"This string contains a " and a \"#;
+    assert_eq!(parse::value(raw_edn).unwrap(),
+               ValueAndSpan {
+                   inner: SpannedValue::Text(raw_string.to_string()),
+                   span: Span(0, raw_edn.len() as u32)
+               });
 }
 
 #[test]