Optimizes LN parsing for Turtle and SPARQL queries

lib/src/rio/turtle/mod.rs

@@ -73,9 +73,30 @@ mod grammar {
         );
     }
 
-    pub fn unescape_echars(input: &str) -> Cow<str> {
+    fn unescape_echars(input: &str) -> Cow<str> {
         unescape_characters(input, &UNESCAPE_CHARACTERS, &UNESCAPE_REPLACEMENT)
     }
+
+    const UNESCAPE_PN_CHARACTERS: [u8; 20] = [
+        b'_', b'~', b'.', b'-', b'!', b'$', b'&', b'\'', b'(', b')', b'*', b'+', b',', b';', b'=',
+        b'/', b'?', b'#', b'@', b'%',
+    ];
+    lazy_static! {
+        static ref UNESCAPE_PN_REPLACEMENT: StaticSliceMap<char, char> = StaticSliceMap::new(
+            &[
+                '_', '~', '.', '-', '!', '$', '&', '\'', '(', ')', '*', '+', ',', ';', '=', '/',
+                '?', '#', '@', '%'
+            ],
+            &[
+                '_', '~', '.', '-', '!', '$', '&', '\'', '(', ')', '*', '+', ',', ';', '=', '/',
+                '?', '#', '@', '%'
+            ]
+        );
+    }
+
+    pub fn unescape_pn_local(input: &str) -> Cow<str> {
+        unescape_characters(input, &UNESCAPE_PN_CHARACTERS, &UNESCAPE_PN_REPLACEMENT)
+    }
 }
 
 pub use self::grammar::read_turtle;

lib/src/rio/turtle/turtle_grammar.rustpeg

@@ -3,7 +3,6 @@
 use std::char;
 use model::vocab::rdf;
 use model::vocab::xsd;
-use std::iter;
 use std::str::FromStr;
 use rio::utils::unescape_unicode_codepoints;
 
@@ -170,9 +169,9 @@ PNAME_NS -> &'input str = ns:$(PN_PREFIX? ":") {
 }
 
 //[140s]
-PNAME_LN -> Url = ns:$(PNAME_NS) local:PN_LOCAL {?
+PNAME_LN -> Url = ns:$(PNAME_NS) local:$(PN_LOCAL) {?
     match state.namespaces.get(ns) {
-        Some(ns) => match Url::parse(&(ns.to_string() + &local)) {
+        Some(ns) => match Url::parse(&(ns.to_string() + &unescape_pn_local(local))) {
             Ok(url) => Ok(url),
             Err(error) => Err("IRI parsing failed")
         },
@@ -249,21 +248,10 @@ PN_CHARS -> () = [\-0-9\u{00B7}\u{0300}-\u{036F}\u{203F}-\u{2040}] / PN_CHARS_U
 PN_PREFIX -> () = PN_CHARS_BASE PN_CHARS* ("."+ PN_CHARS+)*
 
 //[168s]
-PN_LOCAL -> String = f:PN_LOCAL_first c:(PN_LOCAL_next*) e:(PN_LOCAL_next_dot*) {
+PN_LOCAL -> () = (PN_CHARS_U / ':' / [0-9] / PLX) (PN_CHARS / ':' / PLX)* ('.'+ (PN_CHARS / ':' / PLX)+)?
-    f.to_string() + &c.concat() + &e.concat()
-}
-PN_LOCAL_first -> String =
-    c:$(":" / [0-9] / PN_CHARS_U) { c.into() } /
-    s:PLX { s }
-PN_LOCAL_next -> String =
-    c:$(":" / PN_CHARS) { c.into() } /
-    s:PLX { s }
-PN_LOCAL_next_dot -> String = d:$("."+) f:PN_LOCAL_next { d.to_string() + &f}
 
 //[169s]
-PLX -> String =
+PLX -> () = PERCENT / PN_LOCAL_ESC
-    p:$(PERCENT) { p.into() } /
-    e:PN_LOCAL_ESC { iter::once(e).collect() }
 
 //[170s]
 PERCENT -> () = "%" HEX HEX
@@ -272,7 +260,7 @@ PERCENT -> () = "%" HEX HEX
 HEX -> () = ([0-9A-Fa-f])
 
 //[172s]
-PN_LOCAL_ESC -> char = "\\" c:$([_~\.\-!$&'()*+,;=/?#@%]) { c.chars().next().unwrap() }
+PN_LOCAL_ESC -> () = "\\" [_~\.\-!$&'()*+,;=/?#@%]
 
 
 //space

lib/src/sparql/parser.rs

@@ -329,6 +329,27 @@ mod grammar {
         unescape_characters(input, &UNESCAPE_CHARACTERS, &UNESCAPE_REPLACEMENT)
     }
 
+    const UNESCAPE_PN_CHARACTERS: [u8; 20] = [
+        b'_', b'~', b'.', b'-', b'!', b'$', b'&', b'\'', b'(', b')', b'*', b'+', b',', b';', b'=',
+        b'/', b'?', b'#', b'@', b'%',
+    ];
+    lazy_static! {
+        static ref UNESCAPE_PN_REPLACEMENT: StaticSliceMap<char, char> = StaticSliceMap::new(
+            &[
+                '_', '~', '.', '-', '!', '$', '&', '\'', '(', ')', '*', '+', ',', ';', '=', '/',
+                '?', '#', '@', '%'
+            ],
+            &[
+                '_', '~', '.', '-', '!', '$', '&', '\'', '(', ')', '*', '+', ',', ';', '=', '/',
+                '?', '#', '@', '%'
+            ]
+        );
+    }
+
+    pub fn unescape_pn_local(input: &str) -> Cow<str> {
+        unescape_characters(input, &UNESCAPE_PN_CHARACTERS, &UNESCAPE_PN_REPLACEMENT)
+    }
+
     include!(concat!(env!("OUT_DIR"), "/sparql_grammar.rs"));
 
     pub fn read_sparql_query<'a, R: Read + 'a>(

lib/src/sparql/sparql_grammar.rustpeg

@@ -3,7 +3,6 @@
 use std::char;
 use model::vocab::rdf;
 use model::vocab::xsd;
-use std::iter;
 use std::str::FromStr;
 
 #![arguments(state: &mut ParserState)]
@@ -933,8 +932,8 @@ PNAME_NS -> &'input str = ns:$(PN_PREFIX? ":") {
 }
 
 //[141]
-PNAME_LN -> String = ns:$(PNAME_NS) local:PN_LOCAL {?
+PNAME_LN -> String = ns:$(PNAME_NS) local:$(PN_LOCAL) {?
-    state.namespaces.get(ns).map(|v| v.clone() + &local).ok_or("Prefix not found")
+    state.namespaces.get(ns).map(|v| v.clone() + &unescape_pn_local(local)).ok_or("Prefix not found")
 }
 
 //[142]
@@ -1038,21 +1037,10 @@ PN_CHARS -> () = [\-0-9\u{00B7}\u{0300}-\u{036F}\u{203F}-\u{2040}] / PN_CHARS_U
 PN_PREFIX -> () = PN_CHARS_BASE PN_CHARS* ("."+ PN_CHARS+)*
 
 //[169]
-PN_LOCAL -> String = f:PN_LOCAL_first c:(PN_LOCAL_next*) e:(PN_LOCAL_next_dot*) {
+PN_LOCAL -> () = (PN_CHARS_U / ':' / [0-9] / PLX) (PN_CHARS / ':' / PLX)* ('.'+ (PN_CHARS / ':' / PLX)+)?
-    f.to_string() + &c.concat() + &e.concat()
-}
-PN_LOCAL_first -> String =
-    c:$(":" / [0-9] / PN_CHARS_U) { c.into() } /
-    PLX
-PN_LOCAL_next -> String =
-    c:$(":" / PN_CHARS) { c.into() } /
-    PLX
-PN_LOCAL_next_dot -> String = d:$('.'+) f:PN_LOCAL_next* { d.to_string() + &f.concat()}
 
 //[170]
-PLX -> String =
+PLX -> () = PERCENT / PN_LOCAL_ESC
-    p:$(PERCENT) { p.into() } /
-    e:PN_LOCAL_ESC { iter::once(e).collect() }
 
 //[171]
 PERCENT -> () = "%" HEX HEX
@@ -1061,7 +1049,7 @@ PERCENT -> () = "%" HEX HEX
 HEX -> () = ([0-9A-Fa-f])
 
 //[173]
-PN_LOCAL_ESC -> char = "\\" c:$([_~\.\-!$&'()*+,;=/?#@%]) { c.chars().next().unwrap() } //TODO: added '/' to make tests pass but is it valid?
+PN_LOCAL_ESC -> () = "\\" [_~\.\-!$&'()*+,;=/?#@%] //TODO: added '/' to make tests pass but is it valid?
 
 //space
 _ = #quiet<([ \t\n\r] / comment)*>