only check if has scheme to determine if it's a relative uri (faster)

nbittich · Dec 1, 2024 · e6da8bc · e6da8bc
1 parent f829121
commit e6da8bc
Show file tree

Hide file tree

Showing 3 changed files with 23 additions and 11 deletions.
diff --git a/src/iri.rs b/src/iri.rs
@@ -79,7 +79,7 @@ use nom::{
     combinator::complete,
     error::{ParseError, VerboseError},
 };
-use parser::{parse_absolute_iri, parse_iri, parse_iri_reference};
+use parser::{parse_absolute_iri, parse_iri, parse_iri_reference, parse_scheme};
 
 use crate::prelude::alt;
 
@@ -111,6 +111,12 @@ impl IRI<'_> {
     pub fn is_relative(&self) -> bool {
         matches!(self, IRI::Reference(_))
     }
+    pub fn has_scheme(s: &str) -> bool {
+        match parse_scheme(s) {
+            Ok((_, scheme)) => !scheme.is_empty(),
+            Err(_) => false,
+        }
+    }
 }
 mod ip {
     use nom::{
@@ -231,7 +237,7 @@ mod parser {
     pub(super) fn parse_iri(s: &str) -> ParserResult<IRI> {
         map(
             tuple((
-                terminated(parse_scheme, tag(":")),
+                parse_scheme,
                 parse_i_hier_part,
                 preceded(opt(tag("?")), parse_i_query),
                 preceded(opt(tag("#")), parse_i_fragment),
@@ -248,7 +254,7 @@ mod parser {
         map(
             tuple((
                 parse_scheme,
-                preceded(tag(":"), parse_i_hier_part),
+                parse_i_hier_part,
                 preceded(opt(tag("?")), parse_i_query),
             )),
             |(scheme, hier_part, query)| IRI::Absolute {
@@ -383,10 +389,13 @@ mod parser {
             tag("@"),
         ))(s)
     }
-    fn parse_scheme(s: &str) -> ParserResult<&str> {
-        verify(
-            take_while1(|c: char| c.is_alphanumeric() || c == '.' || c == '-' || c == '+'),
-            |scheme: &str| scheme.starts_with(|c: char| c.is_alphabetic()),
+    pub(super) fn parse_scheme(s: &str) -> ParserResult<&str> {
+        terminated(
+            verify(
+                take_while1(|c: char| c.is_alphanumeric() || c == '.' || c == '-' || c == '+'),
+                |scheme: &str| scheme.starts_with(|c: char| c.is_alphabetic()),
+            ),
+            tag(":"),
         )(s)
     }
     fn parse_userinfo(s: &str) -> ParserResult<&str> {

diff --git a/src/tests/mod.rs b/src/tests/mod.rs
@@ -21,6 +21,7 @@ fn cmp_input_file(
         &mut input_buf,
     )
     .unwrap();
+    println!("len: {}", input.len());
 
     if output_json {
         let f = File::open(

diff --git a/src/turtle/turtle_doc.rs b/src/turtle/turtle_doc.rs
@@ -408,10 +408,12 @@ impl<'a> TurtleDoc<'a> {
     ) -> Result<Node<'x>, TurtleDocError> {
         match s {
             TurtleValue::Iri(Iri::Enclosed(iri)) => {
-                let iri_rfc3987 = IRI::try_from(iri).map_err(|e| TurtleDocError {
-                    message: e.to_string(),
-                })?;
-                if iri_rfc3987.is_relative() {
+                // FIXME this is better but slow.
+                // Call if iri.is_relative instead of has_scheme
+                // let iri_rfc3987 = IRI::try_from(iri).map_err(|e| TurtleDocError {
+                //     message: e.to_string(),
+                // })?;
+                if !IRI::has_scheme(iri) {
                     if let Some(base) = base {
                         let iri = (*base).to_owned() + iri;
                         return Ok(Node::Iri(Cow::Owned(iri.to_string())));