Parse RFC3986 URLs

5 files changed, 27 insertions, 28 deletions

diff --git a/dhall/build.rs b/dhall/build.rs
index e6ed0da..c81b302 100644
--- a/dhall/build.rs
+++ b/dhall/build.rs
@@ -163,11 +163,6 @@ fn generate_tests() -> std::io::Result<()> {
                 false
                     // Pretty sure the test is incorrect
                     || path == "unit/import/urls/quotedPathFakeUrlEncode"
-                    // TODO: RFC3986 URLs
-                    || path == "unit/import/urls/emptyPath0"
-                    || path == "unit/import/urls/emptyPath1"
-                    || path == "unit/import/urls/emptyPathSegment"
-                    || path == "usingToMap"
             }),
             input_type: FileType::Text,
             output_type: Some(FileType::Binary),
@@ -186,14 +181,7 @@ fn generate_tests() -> std::io::Result<()> {
             directory: "parser/success/",
             variant: "Printer",
             too_slow_path: Box::new(|path: &str| path == "largeExpression"),
-            exclude_path: Box::new(|path: &str| {
-                false
-                    // TODO: RFC3986 URLs
-                    || path == "unit/import/urls/emptyPath0"
-                    || path == "unit/import/urls/emptyPath1"
-                    || path == "unit/import/urls/emptyPathSegment"
-                    || path == "usingToMap"
-            }),
+            exclude_path: Box::new(|_path: &str| false),
             input_type: FileType::Text,
             output_type: Some(FileType::Binary),
         },
@@ -210,11 +198,6 @@ fn generate_tests() -> std::io::Result<()> {
                     || path == "double"
                     || path == "unit/DoubleLitExponentNoDot"
                     || path == "unit/DoubleLitSecretelyInt"
-                    // TODO: RFC3986 URLs
-                    || path == "unit/import/urls/emptyPath0"
-                    || path == "unit/import/urls/emptyPath1"
-                    || path == "unit/import/urls/emptyPathSegment"
-                    || path == "usingToMap"
             }),
             input_type: FileType::Text,
             output_type: Some(FileType::Binary),
@@ -373,10 +356,6 @@ fn convert_abnf_to_pest() -> std::io::Result<()> {
     let mut file = File::create(grammar_path)?;
     writeln!(&mut file, "// AUTO-GENERATED FILE. See build.rs.")?;
 
-    // TODO: this is a cheat; properly support RFC3986 URLs instead
-    rules.remove("url_path");
-    writeln!(&mut file, "url_path = _{{ path }}")?;
-
     // Work around some greediness issue in the grammar.
     rules.remove("missing");
     writeln!(
diff --git a/dhall/src/syntax/text/dhall.pest.visibility b/dhall/src/syntax/text/dhall.pest.visibility
index 2a89076..b2114ce 100644
--- a/dhall/src/syntax/text/dhall.pest.visibility
+++ b/dhall/src/syntax/text/dhall.pest.visibility
@@ -120,6 +120,7 @@ home_path
 absolute_path
 scheme
 http_raw
+url_path
 authority
 # userinfo
 # host
@@ -133,6 +134,7 @@ authority
 # dec_octet
 # domain
 # domainlabel
+segment
 # pchar
 query
 # pct_encoded
diff --git a/dhall/src/syntax/text/parser.rs b/dhall/src/syntax/text/parser.rs
index 5686300..03211c7 100644
--- a/dhall/src/syntax/text/parser.rs
+++ b/dhall/src/syntax/text/parser.rs
@@ -551,14 +551,14 @@ impl DhallParser {
 
     fn http_raw(input: ParseInput) -> ParseResult<URL<Expr>> {
         Ok(match_nodes!(input.into_children();
-            [scheme(sch), authority(auth), path(p)] => URL {
+            [scheme(sch), authority(auth), url_path(p)] => URL {
                 scheme: sch,
                 authority: auth,
                 path: p,
                 query: None,
                 headers: None,
             },
-            [scheme(sch), authority(auth), path(p), query(q)] => URL {
+            [scheme(sch), authority(auth), url_path(p), query(q)] => URL {
                 scheme: sch,
                 authority: auth,
                 path: p,
@@ -568,10 +568,28 @@ impl DhallParser {
         ))
     }
 
+    fn url_path(input: ParseInput) -> ParseResult<FilePath> {
+        Ok(match_nodes!(input.into_children();
+            [path_component(components)..] => {
+                let mut file_path: Vec<_> = components.collect();
+                // An empty path normalizes to "/"
+                if file_path.is_empty() {
+                    file_path = vec!["".to_owned()];
+                }
+                FilePath { file_path }
+            }
+        ))
+    }
+
     fn authority(input: ParseInput) -> ParseResult<String> {
         Ok(input.as_str().to_owned())
     }
 
+    #[alias(path_component)]
+    fn segment(input: ParseInput) -> ParseResult<String> {
+        Ok(input.as_str().to_string())
+    }
+
     fn query(input: ParseInput) -> ParseResult<String> {
         Ok(input.as_str().to_owned())
     }
diff --git a/dhall/tests/parser/failure/spacing/MergeNoSpace2.txt b/dhall/tests/parser/failure/spacing/MergeNoSpace2.txt
index 96d937b..77314e0 100644
--- a/dhall/tests/parser/failure/spacing/MergeNoSpace2.txt
+++ b/dhall/tests/parser/failure/spacing/MergeNoSpace2.txt
@@ -3,4 +3,4 @@
 1 | merge x(y)␊
   |       ^---
   |
-  = expected missing, double_quote_literal, single_quote_literal, if_, merge, non_empty_list_literal, NaN, Some_, toMap, assert, forall, numeric_double_literal, minus_infinity_literal, plus_infinity_literal, natural_literal, integer_literal, or import_hashed
+  = expected missing, non_empty_list_literal, double_quote_literal, single_quote_literal, if_, merge, NaN, Some_, toMap, assert, forall, numeric_double_literal, minus_infinity_literal, plus_infinity_literal, natural_literal, integer_literal, or import_hashed
diff --git a/dhall/tests/parser/failure/unit/UsingToMap.txt b/dhall/tests/parser/failure/unit/UsingToMap.txt
index f6ab2a5..8b1a8aa 100644
--- a/dhall/tests/parser/failure/unit/UsingToMap.txt
+++ b/dhall/tests/parser/failure/unit/UsingToMap.txt
@@ -1,6 +1,6 @@
- --> 8:20
+ --> 8:27
   |
 8 | https://example.com using toMap { Foo = "Bar" }␊
-  |                    ^---
+  |                           ^---
   |
-  = expected path
+  = expected import_hashed or primitive_expression