From 768a5fcd01da5eaa398a415676a4c82f8052eae3 Mon Sep 17 00:00:00 2001 From: Nordine Bittich Date: Sat, 27 Jan 2024 11:55:37 +0100 Subject: [PATCH] more tests, add configurable well known prefix --- Cargo.lock | 29 +---- Cargo.toml | 2 +- README.md | 59 +++++++-- index.html | 92 +++++++++++--- lib-rdfa/examples/earl_html5/example0035.html | 15 +++ lib-rdfa/examples/earl_html5/example0035.ttl | 1 + lib-rdfa/examples/earl_html5/example0037.html | 18 +++ lib-rdfa/examples/earl_html5/example0037.ttl | 5 + lib-rdfa/examples/earl_html5/example0039.html | 15 +++ lib-rdfa/examples/earl_html5/example0039.ttl | 1 + lib-rdfa/src/lib.rs | 13 +- lib-rdfa/src/structs.rs | 116 +++++++++++------- lib-rdfa/src/tests/earl_html5.rs | 3 + lib-rdfa/src/tests/mod.rs | 6 +- rdfa-wasm/src/lib.rs | 12 +- 15 files changed, 277 insertions(+), 110 deletions(-) create mode 100644 lib-rdfa/examples/earl_html5/example0035.html create mode 100644 lib-rdfa/examples/earl_html5/example0035.ttl create mode 100644 lib-rdfa/examples/earl_html5/example0037.html create mode 100644 lib-rdfa/examples/earl_html5/example0037.ttl create mode 100644 lib-rdfa/examples/earl_html5/example0039.html create mode 100644 lib-rdfa/examples/earl_html5/example0039.ttl diff --git a/Cargo.lock b/Cargo.lock index e54565b..8b7d6fc 100644 --- a/Cargo.lock +++ b/Cargo.lock @@ -388,7 +388,7 @@ version = "0.1.2" dependencies = [ "ego-tree", "env_logger", - "itertools 0.12.0", + "itertools", "lazy_static", "log", "regex", @@ -400,22 +400,6 @@ dependencies = [ "uuid", ] -[[package]] -name = "graph-rdfa-processor" -version = "0.1.2" -source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "a40adbf551dc0d0f9f246290b93b00a4745e08c3cc4f42025e41cd8d923d5187" -dependencies = [ - "ego-tree", - "itertools 0.11.0", - "lazy_static", - "log", - "regex", - "scraper", - "url", - "uuid", -] - [[package]] name = "hashbrown" version = "0.14.3" @@ -475,15 +459,6 @@ dependencies = [ "unicode-normalization", ] -[[package]] -name = "itertools" -version = "0.11.0" -source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "b1c173a5686ce8bfa551b3563d0c2170bf24ca44da99c7ca4bfdab5418c3fe57" -dependencies = [ - "either", -] - [[package]] name = "itertools" version = "0.12.0" @@ -785,7 +760,7 @@ name = "rdfa-wasm" version = "0.1.2" dependencies = [ "getrandom", - "graph-rdfa-processor 0.1.2 (registry+https://github.com/rust-lang/crates.io-index)", + "graph-rdfa-processor", "wasm-bindgen", ] diff --git a/Cargo.toml b/Cargo.toml index 8a72bf4..2958e71 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -28,7 +28,7 @@ test-case = "3.3.1" env_logger = "0.11.1" tortank = "0.20.0" wasm-bindgen = "0.2.90" -graph-rdfa-processor = "0.1.2" +graph-rdfa-processor = { path = "./lib-rdfa" } [profile.release] opt-level = 'z' # Optimize for size. diff --git a/README.md b/README.md index 6e4143a..f377b31 100644 --- a/README.md +++ b/README.md @@ -1,23 +1,58 @@ -# graph-rdfa-processor +# RDFa processor library (WIP) -### RDFa processor library (WIP) +Rust and wasm library to extract rdf statements (in n-triples format) from an html file +based on rdfa annotations. -- covers [RDFa 1.1 Primer - Third Edition](https://www.w3.org/TR/rdfa-primer/). -- used [RDFa/Play](https://rdfa.info/play/) for comparing. -- WIP: +## Examples + +```rust +use graph_rdfa_processor::RdfaGraph; + let html = r#" +
+

Dan Brickley

+
+ "#; + let base = "http://example.com" + let well_known_prefix = Some("http://example.org/.well_known"); + + RdfaGraph::parse_str(html, base, well_known_prefix).unwrap() + +``` + +```js + + +``` + +- covers: + + - [RDFa 1.1 Primer - Third Edition](https://www.w3.org/TR/rdfa-primer/) - [RDFa Core](https://www.w3.org/TR/rdfa-core/) - [Earl-Reports](https://rdfa.info/earl-reports/#RDFa-rdfa1.1-tests-for-html5) +- used [RDFa/Play](https://rdfa.info/play/) for comparing. - [Demo](https://nbittich.github.io/graph-rdfa-processor/) - ### WIP - - The plan is to make as much tests as possible, and fix the bugs along the way. +### WIP - Once enough tests have been implemented, a full refactoring will be needed before - doing anything else. +The plan is to make as much tests as possible, and fix the bugs along the way. - ### KNOWN ISSUES +### KNOWN ISSUES - - White space not preserved. See test 0329. +- White space not preserved. See test 0329. diff --git a/index.html b/index.html index 8a93622..528f1fd 100644 --- a/index.html +++ b/index.html @@ -17,28 +17,64 @@ form.classList.remove("d-none"); const text_area = document.querySelector("#html"); - text_area.value = ` - - - Test 0083 - - -
-

Ivan Herman

-

mailto:ivan@w3.org

-

Dan Brickley

-
- - -`; + text_area.value = ` + + + + Test 0083 + + +
+

Ivan Herman

+

mailto:ivan@w3.org

+

Dan Brickley

+
+ + + `; form.addEventListener("submit", (e) => { e.preventDefault(); const data = new FormData(e.target); - let res = html_to_rdfa(data.get("html"), data.get("base")); + let res = html_to_rdfa( + data.get("html") || "", + data.get("base") || "", + data.get("wellKnownPrefix") || "", + ); const out = document.querySelector("pre"); out.innerText = res; }); + const issueLink = document.querySelector("#issueLink"); + issueLink.onclick = (e) => { + e.preventDefault(); + const a = document.createElement("a"); + let params = new URLSearchParams(); + const data = new FormData(form); + params.append("title", "RDFa processing bug"); + + params.append( + "body", + ` +### Base: + + \`${data.get("base") || ""}\` + +### Well Known Prefix: + + \`${data.get("base") || ""}\` + +### Html: + + \`\`\`html + ${data.get("html") || ""} + \`\`\` + + `, + ); + a.href = `https://github.com/nbittich/graph-rdfa-processor/issues/new?${params.toString()}`; + a.target = "_blank"; + a.click(); + }; } run(); @@ -48,11 +84,11 @@

RDFa Playground

This is a poc of an rdfa library written in rust. If you find any bug, - please open an issue. + please open an issue.

-
-
+
+
@@ -63,6 +99,20 @@

RDFa Playground

value="http://example.com" />
+
+
+ +
+
+ +
+ Well known prefix for + Skolemisation +
+
+
+
@@ -72,10 +122,12 @@

RDFa Playground

aria-describedby="Html content">
- +
+ +
-
+ 

     

   

diff --git a/lib-rdfa/examples/earl_html5/example0035.html b/lib-rdfa/examples/earl_html5/example0035.html
new file mode 100644
index 0000000..553d5f4
--- /dev/null
+++ b/lib-rdfa/examples/earl_html5/example0035.html
@@ -0,0 +1,15 @@
+
+
+  
+		Test 0035    
+  
+  
+  	

+      A photo depicting Michael  	
+  	

+  
+
\ No newline at end of file
diff --git a/lib-rdfa/examples/earl_html5/example0035.ttl b/lib-rdfa/examples/earl_html5/example0035.ttl
new file mode 100644
index 0000000..505a166
--- /dev/null
+++ b/lib-rdfa/examples/earl_html5/example0035.ttl
@@ -0,0 +1 @@
+  .
diff --git a/lib-rdfa/examples/earl_html5/example0037.html b/lib-rdfa/examples/earl_html5/example0037.html
new file mode 100644
index 0000000..4f732cf
--- /dev/null
+++ b/lib-rdfa/examples/earl_html5/example0037.html
@@ -0,0 +1,18 @@
+
+
+
+
+  
+		Test 0037    
+  
+  
+  	

+      A photo depicting Michael  	
+  	

+  
+
\ No newline at end of file
diff --git a/lib-rdfa/examples/earl_html5/example0037.ttl b/lib-rdfa/examples/earl_html5/example0037.ttl
new file mode 100644
index 0000000..e7e4a0e
--- /dev/null
+++ b/lib-rdfa/examples/earl_html5/example0037.ttl
@@ -0,0 +1,5 @@
+
+
+  .
+
+
diff --git a/lib-rdfa/examples/earl_html5/example0039.html b/lib-rdfa/examples/earl_html5/example0039.html
new file mode 100644
index 0000000..9138b79
--- /dev/null
+++ b/lib-rdfa/examples/earl_html5/example0039.html
@@ -0,0 +1,15 @@
+
+
+  
+		Test 0039    
+  
+  
+  	

+      A photo depicting Michael  	
+  	

+  
+
\ No newline at end of file
diff --git a/lib-rdfa/examples/earl_html5/example0039.ttl b/lib-rdfa/examples/earl_html5/example0039.ttl
new file mode 100644
index 0000000..25a4624
--- /dev/null
+++ b/lib-rdfa/examples/earl_html5/example0039.ttl
@@ -0,0 +1 @@
+  .
diff --git a/lib-rdfa/src/lib.rs b/lib-rdfa/src/lib.rs
index a93dbe4..645f6d1 100644
--- a/lib-rdfa/src/lib.rs
+++ b/lib-rdfa/src/lib.rs
@@ -27,19 +27,28 @@ impl<'a> RdfaGraph<'a> {
         initial_context: Context<'a>,
     ) -> Result, Box> {
         let mut triples = vec![];
+        let well_known_prefix = initial_context.well_known_prefix.clone();
         traverse_element(input, None, initial_context, &mut triples, &mut vec![])?;
 
         triples = copy_pattern(triples)?;
 
-        Ok(RdfaGraph(triples.into_iter().collect()))
+        Ok(RdfaGraph {
+            statements: triples.into_iter().collect(),
+            well_known_prefix,
+        })
     }
     // temporary thing
-    pub fn parse_str(html: &'a str, base: &'a str) -> Result> {
+    pub fn parse_str(
+        html: &'a str,
+        base: &'a str,
+        well_known_prefix: Option<&'a str>,
+    ) -> Result> {
         let document = scraper::Html::parse_document(html);
         let root = document.root_element();
 
         let root_ctx = Context {
             base,
+            well_known_prefix,
             ..Default::default()
         };
         RdfaGraph::parse(&root, root_ctx).map(|g| g.to_string())
diff --git a/lib-rdfa/src/structs.rs b/lib-rdfa/src/structs.rs
index 14b5b68..61528f8 100644
--- a/lib-rdfa/src/structs.rs
+++ b/lib-rdfa/src/structs.rs
@@ -19,11 +19,15 @@ macro_rules! iri {
 }
 
 #[derive(Debug)]
-pub struct RdfaGraph<'a>(pub HashSet>);
+pub struct RdfaGraph<'a> {
+    pub well_known_prefix: Option<&'a str>,
+    pub statements: HashSet>,
+}
 
 #[derive(Debug, Default, Clone)]
 pub struct Context<'a> {
     pub base: &'a str,
+    pub well_known_prefix: Option<&'a str>,
     pub vocab: Option<&'a str>,
     pub lang: Option<&'a str>,
     pub in_rel: Option>>,
@@ -63,6 +67,22 @@ pub struct Statement<'a> {
     pub object: Node<'a>,
 }
 
+impl Statement<'_> {
+    fn as_ntriple_string(&self, well_known_prefix: Option<&str>) -> String {
+        let Statement {
+            subject,
+            predicate,
+            object,
+        } = self;
+        format!(
+            r#"{} {} {}."#,
+            subject.as_ntriple_string(well_known_prefix),
+            predicate.as_ntriple_string(well_known_prefix),
+            object.as_ntriple_string(well_known_prefix)
+        )
+    }
+}
+
 impl Node<'_> {
     pub fn is_empty(&self) -> bool {
         match self {
@@ -78,31 +98,10 @@ impl Node<'_> {
             Node::RefBNode((s, _)) => s.is_empty(),
         }
     }
-}
-
-impl PartialEq for Node<'_> {
-    fn eq(&self, other: &Self) -> bool {
-        match (self, other) {
-            (Self::Iri(l0), Self::Iri(r0)) => l0 == r0,
-            (Self::Iri(l0), Self::TermIri(r0)) => l0 == r0,
-            (Self::TermIri(l0), Self::TermIri(r0)) => l0 == r0,
-            (Self::TermIri(l0), Self::Iri(r0)) => l0 == r0,
-            (Self::Literal(l0), Self::Literal(r0)) => l0 == r0,
-            (Self::Ref(l0), Self::Ref(r0)) => l0 == r0,
-            (Self::Ref(l0), rhs) => l0.as_ref() == rhs,
-            (lhs, Self::Ref(r0)) => lhs == r0.as_ref(),
-            (Self::BNode(l0), Self::BNode(r0)) => l0 == r0,
-            (Self::RefBNode(l0), Self::RefBNode(r0)) => l0 == r0,
-            _ => false,
-        }
-    }
-}
-
-impl Display for Node<'_> {
-    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+    fn as_ntriple_string(&self, well_known_prefix: Option<&str>) -> String {
         match self {
-            Node::Iri(iri) | Node::TermIri(iri) => f.write_str(&format!("<{}>", iri)),
-            Node::Ref(iri) => f.write_str(&format!("{}", iri)),
+            Node::Iri(iri) | Node::TermIri(iri) => format!("<{}>", iri),
+            Node::Ref(iri) => format!("{}", iri.as_ntriple_string(well_known_prefix)),
             Node::Literal(Literal {
                 datatype,
                 lang,
@@ -122,51 +121,82 @@ impl Display for Node<'_> {
                     .as_ref()
                     .filter(|dt| dt.as_ref() != &*NODE_RDF_XSD_STRING)
                 {
-                    s.push_str(&format!(r#"^^{datatype}"#));
+                    s.push_str(&format!(
+                        r#"^^{}"#,
+                        datatype.as_ntriple_string(well_known_prefix)
+                    ));
                 } else if let Some(lang) = lang {
                     s.push_str(&format!(r#"@{lang}"#));
                 }
-                f.write_str(&s)
+                s
             }
             Node::BNode(id) => {
                 // todo maybe this should use the base?
-                f.write_str(&format!("<{}{}>", DEFAULT_WELL_KNOWN_PREFIX, id))
+                format!(
+                    "<{}{}>",
+                    well_known_prefix.unwrap_or(DEFAULT_WELL_KNOWN_PREFIX),
+                    id
+                )
             }
             Node::RefBNode((id, uuid)) => {
                 if let Ok(v) = id.parse::() {
                     if v <= BNODE_ID_GENERATOR.load(std::sync::atomic::Ordering::SeqCst) {
-                        f.write_str(&format!("<{}{}>", DEFAULT_WELL_KNOWN_PREFIX, uuid))
+                        format!(
+                            "<{}{}>",
+                            well_known_prefix.unwrap_or(DEFAULT_WELL_KNOWN_PREFIX),
+                            uuid
+                        )
                     } else {
-                        f.write_str(&format!("<{}{}>", DEFAULT_WELL_KNOWN_PREFIX, id))
+                        format!(
+                            "<{}{}>",
+                            well_known_prefix.unwrap_or(DEFAULT_WELL_KNOWN_PREFIX),
+                            id
+                        )
                     }
                 } else if id.is_empty() {
-                    f.write_str(&format!("<{}{}>", DEFAULT_WELL_KNOWN_PREFIX, uuid))
+                    format!(
+                        "<{}{}>",
+                        well_known_prefix.unwrap_or(DEFAULT_WELL_KNOWN_PREFIX),
+                        uuid
+                    )
                 } else {
-                    f.write_str(&format!("<{}{}>", DEFAULT_WELL_KNOWN_PREFIX, id))
+                    format!(
+                        "<{}{}>",
+                        well_known_prefix.unwrap_or(DEFAULT_WELL_KNOWN_PREFIX),
+                        id
+                    )
                 }
             }
         }
     }
 }
 
-impl Display for Statement<'_> {
-    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
-        let Statement {
-            subject,
-            predicate,
-            object,
-        } = self;
-        f.write_str(&format!(r#"{subject} {predicate} {object}."#))
+impl PartialEq for Node<'_> {
+    fn eq(&self, other: &Self) -> bool {
+        match (self, other) {
+            (Self::Iri(l0), Self::Iri(r0)) => l0 == r0,
+            (Self::Iri(l0), Self::TermIri(r0)) => l0 == r0,
+            (Self::TermIri(l0), Self::TermIri(r0)) => l0 == r0,
+            (Self::TermIri(l0), Self::Iri(r0)) => l0 == r0,
+            (Self::Literal(l0), Self::Literal(r0)) => l0 == r0,
+            (Self::Ref(l0), Self::Ref(r0)) => l0 == r0,
+            (Self::Ref(l0), rhs) => l0.as_ref() == rhs,
+            (lhs, Self::Ref(r0)) => lhs == r0.as_ref(),
+            (Self::BNode(l0), Self::BNode(r0)) => l0 == r0,
+            (Self::RefBNode(l0), Self::RefBNode(r0)) => l0 == r0,
+            _ => false,
+        }
     }
 }
+
 impl Display for RdfaGraph<'_> {
     fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
         f.write_str(
             &self
-                .0
+                .statements
                 .iter()
-                .map(Statement::to_string)
-                .collect::>()
+                .map(|s| s.as_ntriple_string(self.well_known_prefix))
+                .collect::>()
                 .join("\n"),
         )
     }
diff --git a/lib-rdfa/src/tests/earl_html5.rs b/lib-rdfa/src/tests/earl_html5.rs
index 330fab4..e6b103b 100644
--- a/lib-rdfa/src/tests/earl_html5.rs
+++ b/lib-rdfa/src/tests/earl_html5.rs
@@ -28,8 +28,11 @@ const INPUT_OUTPUT_DIR: &str = "examples/earl_html5";
 #[test_case("example0032"  ; "@resource overrides @href                                                             : earl_reports_html5_0032 ")]
 #[test_case("example0033"  ; "simple chaining test with bNode                                                       : earl_reports_html5_0033 ")]
 #[test_case("example0034"  ; "simple img[@src] test                                                                 : earl_reports_html5_0034 ")]
+#[test_case("example0035"  ; "@src/@href test                                                                       : earl_reports_html5_0035 ")]
 #[test_case("example0036"  ; "@src/@resource test                                                                   : earl_reports_html5_0036 ")]
+#[test_case("example0037"  ; "@src/@resource test                                                                   : earl_reports_html5_0037 ")]
 #[test_case("example0038"  ; "@rev - img[@src] test                                                                 : earl_reports_html5_0038 ")]
+#[test_case("example0039"  ; "@rev - img[@src] test                                                                 : earl_reports_html5_0039 ")]
 #[test_case("example0048"  ; "@typeof with @about and @rel present, no @resource                                    : earl_reports_html5_0048 ")]
 #[test_case("example0049"  ; "@typeof with @about, no @rel or @resource                                             : earl_reports_html5_0049 ")]
 #[test_case("example0050"  ; "@typeof without anything else                                                         : earl_reports_html5_0050 ")]
diff --git a/lib-rdfa/src/tests/mod.rs b/lib-rdfa/src/tests/mod.rs
index 39494e5..09ecd63 100644
--- a/lib-rdfa/src/tests/mod.rs
+++ b/lib-rdfa/src/tests/mod.rs
@@ -1,4 +1,4 @@
-use std::path::PathBuf;
+use std::{ops::Add, path::PathBuf};
 
 use scraper::Html;
 use tortank::turtle::turtle_doc::TurtleDoc;
@@ -60,8 +60,8 @@ fn cmp_files(test_name: &str, input_output_dir: &str, base: &str) {
         Some(constants::DEFAULT_WELL_KNOWN_PREFIX.to_string()),
     ))
     .unwrap();
-    let diff = ttl.difference(&graph).unwrap();
-    //  diff = diff.add(graph.difference(&ttl).unwrap());
+    let mut diff = ttl.difference(&graph).unwrap();
+    diff = diff.add(graph.difference(&ttl).unwrap());
     if !diff.is_empty() {
         println!("============ Difference ============");
         println!("{diff}");
diff --git a/rdfa-wasm/src/lib.rs b/rdfa-wasm/src/lib.rs
index 3be6380..de46ef8 100644
--- a/rdfa-wasm/src/lib.rs
+++ b/rdfa-wasm/src/lib.rs
@@ -2,6 +2,14 @@ use graph_rdfa_processor::RdfaGraph;
 use wasm_bindgen::prelude::*;
 
 #[wasm_bindgen]
-pub fn html_to_rdfa(html: &str, base: &str) -> String {
-    RdfaGraph::parse_str(html, base).unwrap()
+pub fn html_to_rdfa(html: &str, base: &str, well_known_prefix: &str) -> String {
+    let wkp = {
+        let wkp = well_known_prefix.trim();
+        if wkp.is_empty() {
+            None
+        } else {
+            Some(wkp)
+        }
+    };
+    RdfaGraph::parse_str(html, base, wkp).unwrap()
 }