From 4a45ed7e1f80a7d3e4032e08eb499dab3412453f Mon Sep 17 00:00:00 2001
From: Nadrieril
Date: Sat, 23 Mar 2019 01:09:05 +0100
Subject: Parse unicode escapes in string

---
 dhall_core/src/parser.rs  | 44 ++++++++++++++++++++++++--------------------
 dhall_core/src/printer.rs | 20 +++++++++++++-------
 dhall_core/src/text.rs    | 20 ++++++++++----------
 3 files changed, 47 insertions(+), 37 deletions(-)

(limited to 'dhall_core/src')

diff --git a/dhall_core/src/parser.rs b/dhall_core/src/parser.rs
index 8908c78..9fd3213 100644
--- a/dhall_core/src/parser.rs
+++ b/dhall_core/src/parser.rs
@@ -13,7 +13,7 @@ use crate::*;
 // are here and hopefully you can figure out how they work.
 
 type ParsedText = InterpolatedText<X, Import>;
-type ParsedTextContents<'a> = InterpolatedTextContents<'a, X, Import>;
+type ParsedTextContents = InterpolatedTextContents<X, Import>;
 
 pub type ParseError = pest::error::Error<Rule>;
 
@@ -270,7 +270,7 @@ make_parser! {
         }
     ));
 
-    rule!(double_quote_chunk<ParsedTextContents<'a>>; children!(
+    rule!(double_quote_chunk<ParsedTextContents>; children!(
         [interpolation(e)] => {
             InterpolatedTextContents::Expr(e)
         },
@@ -278,24 +278,28 @@ make_parser! {
             InterpolatedTextContents::Text(s)
         },
         [double_quote_char(s)] => {
-            InterpolatedTextContents::Text(s)
+            InterpolatedTextContents::Text(s.to_owned())
         },
     ));
-    rule!(double_quote_escaped<&'a str>;
-        // TODO: parse all escapes
+    rule!(double_quote_escaped<String>;
         captured_str!(s) => {
             match s {
-                "\"" => "\"",
-                "$" => "$",
-                "\\" => "\\",
-                "/" => "/",
-                // "b" => "\b",
-                // "f" => "\f",
-                "n" => "\n",
-                "r" => "\r",
-                "t" => "\t",
-                // "uXXXX"
-                _ => unimplemented!(),
+                "\"" => "\"".to_owned(),
+                "$" => "$".to_owned(),
+                "\\" => "\\".to_owned(),
+                "/" => "/".to_owned(),
+                "b" => "\u{0008}".to_owned(),
+                "f" => "\u{000C}".to_owned(),
+                "n" => "\n".to_owned(),
+                "r" => "\r".to_owned(),
+                "t" => "\t".to_owned(),
+                _ => {
+                    // "uXXXX"
+                    use std::convert::TryFrom;
+                    let c = u16::from_str_radix(&s[1..5], 16).unwrap();
+                    let c = char::try_from(c as u32).unwrap();
+                    std::iter::once(c).collect()
+                }
             }
         }
     );
@@ -323,22 +327,22 @@ make_parser! {
         [expression(e)] => e
     ));
 
-    rule!(single_quote_continue<Vec<ParsedTextContents<'a>>>; children!(
+    rule!(single_quote_continue<Vec<ParsedTextContents>>; children!(
         [interpolation(c), single_quote_continue(rest)] => {
             let mut rest = rest;
             rest.push(InterpolatedTextContents::Expr(c)); rest
         },
         [escaped_quote_pair(c), single_quote_continue(rest)] => {
             let mut rest = rest;
-            rest.push(InterpolatedTextContents::Text(c)); rest
+            rest.push(InterpolatedTextContents::Text(c.to_owned())); rest
         },
         [escaped_interpolation(c), single_quote_continue(rest)] => {
             let mut rest = rest;
-            rest.push(InterpolatedTextContents::Text(c)); rest
+            rest.push(InterpolatedTextContents::Text(c.to_owned())); rest
         },
         [single_quote_char(c), single_quote_continue(rest)] => {
             let mut rest = rest;
-            rest.push(InterpolatedTextContents::Text(c)); rest
+            rest.push(InterpolatedTextContents::Text(c.to_owned())); rest
         },
         [] => {
             vec![]
diff --git a/dhall_core/src/printer.rs b/dhall_core/src/printer.rs
index 508c1c8..5ecf5ce 100644
--- a/dhall_core/src/printer.rs
+++ b/dhall_core/src/printer.rs
@@ -206,13 +206,19 @@ impl<S, A: Display> Expr<S, A> {
                 for x in a.iter() {
                     match x {
                         InterpolatedTextContents::Text(a) => {
-                            // TODO Format all escapes properly
-                            f.write_str(
-                                &a.replace("\n", "\\n")
-                                    .replace("\t", "\\t")
-                                    .replace("\r", "\\r")
-                                    .replace("\"", "\\\""),
-                            )?;
+                            for c in a.chars() {
+                                match c {
+                                    '\\' => f.write_str("\\\\"),
+                                    '"' => f.write_str("\\\""),
+                                    '$' => f.write_str("\\$"),
+                                    '\u{0008}' => f.write_str("\\b"),
+                                    '\u{000C}' => f.write_str("\\f"),
+                                    '\n' => f.write_str("\\n"),
+                                    '\r' => f.write_str("\\r"),
+                                    '\t' => f.write_str("\\t"),
+                                    c => write!(f, "{}", c),
+                                }?;
+                            }
                         }
                         InterpolatedTextContents::Expr(e) => {
                             f.write_str("${ ")?;
diff --git a/dhall_core/src/text.rs b/dhall_core/src/text.rs
index eb0d39c..2a468d2 100644
--- a/dhall_core/src/text.rs
+++ b/dhall_core/src/text.rs
@@ -30,8 +30,8 @@ impl<N, E> From<String> for InterpolatedText<N, E> {
 }
 
 #[derive(Debug, Clone)]
-pub enum InterpolatedTextContents<'a, Note, Embed> {
-    Text(&'a str),
+pub enum InterpolatedTextContents<Note, Embed> {
+    Text(String),
     Expr(SubExpr<Note, Embed>),
 }
 
@@ -46,35 +46,35 @@ impl<N, E> InterpolatedText<N, E> {
         }
     }
 
-    pub fn iter(&self) -> impl Iterator<Item = InterpolatedTextContents<N, E>> {
+    pub fn iter<'a>(
+        &'a self,
+    ) -> impl Iterator<Item = InterpolatedTextContents<N, E>> + 'a {
         use std::iter::once;
-        once(InterpolatedTextContents::Text(self.head.as_ref())).chain(
+        once(InterpolatedTextContents::Text(self.head.clone())).chain(
             self.tail.iter().flat_map(|(e, s)| {
                 once(InterpolatedTextContents::Expr(Rc::clone(e)))
-                    .chain(once(InterpolatedTextContents::Text(s)))
+                    .chain(once(InterpolatedTextContents::Text(s.clone())))
             }),
         )
     }
 }
 
-impl<'a, N: 'a, E: 'a> FromIterator<InterpolatedTextContents<'a, N, E>>
+impl<'a, N: 'a, E: 'a> FromIterator<InterpolatedTextContents<N, E>>
     for InterpolatedText<N, E>
 {
     fn from_iter<T>(iter: T) -> Self
     where
-        T: IntoIterator<Item = InterpolatedTextContents<'a, N, E>>,
+        T: IntoIterator<Item = InterpolatedTextContents<N, E>>,
     {
         let mut res = InterpolatedText {
             head: "".to_owned(),
             tail: vec![],
         };
-        // let mut empty_string = "".to_owned();
         let mut crnt_str = &mut res.head;
         for x in iter.into_iter() {
             match x {
-                InterpolatedTextContents::Text(s) => crnt_str.push_str(s),
+                InterpolatedTextContents::Text(s) => crnt_str.push_str(&s),
                 InterpolatedTextContents::Expr(e) => {
-                    // crnt_str = &mut empty_string;
                     res.tail.push((e.clone(), "".to_owned()));
                     crnt_str = &mut res.tail.last_mut().unwrap().1;
                 }
-- 
cgit v1.2.3