Import code, add CLI

Author: Porges

Cargo.lock

@@ -0,0 +1,12 @@
+[workspace]
+resolver = "2"
+members = [ "html2rdf" , "html2rdf-cli"]
+
+[workspace.package]
+authors = ["George Pollard <porges@porg.es>"]
+edition = "2024"
+keywords = ["html", "rdf", "rdfa"]
+license = "Apache-2.0"
+publish = false
+readme = "README.md"
+repository = "https://github.com/Porges/html2rdf"

Cargo.toml

@@ -0,0 +1,18 @@
+[advisories]
+ignore = [ ]
+
+[licenses]
+allow = [
+    "Apache-2.0",
+    "BSD-3-Clause",
+    "ISC",
+    "MIT",
+    "MPL-2.0",
+    "Unicode-3.0",
+]
+
+[licenses.private]
+ignore = true
+
+[bans]
+multiple-versions = "warn"

deny.toml

@@ -0,0 +1,23 @@
+[package]
+name = "html2rdf-cli"
+version = "0.1.0"
+authors.workspace = true
+edition.workspace = true
+keywords.workspace = true
+license.workspace = true
+publish.workspace = true
+readme.workspace = true
+repository.workspace = true
+
+[[bin]]
+name = "html2rdf"
+path = "src/main.rs"
+
+[dependencies]
+clap = { version = "4.5.38", features = ["derive", "wrap_help"] }
+html2rdf = { path = "../html2rdf" }
+oxiri = "0.2.11"
+oxrdf = "0.2.4"
+oxttl = "0.1.8"
+reqwest = { version = "0.12.15", features = ["blocking"] }
+url = "2.5.4"

html2rdf-cli/Cargo.toml

@@ -0,0 +1,69 @@
+use std::process::ExitCode;
+
+use clap::Parser;
+
+#[derive(Parser)]
+#[command(version, about)]
+struct Args {
+    #[arg(value_name = "URL")]
+    target: url::Url,
+}
+
+fn main() -> Result<ExitCode, Box<dyn std::error::Error>> {
+    let args = Args::parse();
+    let client = reqwest::blocking::Client::new();
+    let base = args.target.to_string();
+    let base_iri = oxiri::Iri::parse(base.clone())?;
+    let response = client.get(args.target).send()?.error_for_status()?;
+    let content_type = response
+        .headers()
+        .get(reqwest::header::CONTENT_TYPE)
+        .and_then(|v| v.to_str().ok());
+
+    if content_type.is_some_and(|ct| !ct.starts_with("text/html")) {
+        eprintln!("Error: content type is not text/html.");
+        return Ok(ExitCode::FAILURE);
+    }
+
+    let content = response.text()?;
+    let mut output_graph = oxrdf::Graph::new();
+    let mut processor_graph = oxrdf::Graph::new();
+    html2rdf::process(
+        &content,
+        base_iri.clone(),
+        &mut output_graph,
+        &mut processor_graph,
+    )?;
+
+    {
+        // output any warnings/errors
+        let serializer = oxttl::TurtleSerializer::new();
+        let mut locked_err = std::io::stderr().lock();
+        let mut writer = serializer.for_writer(&mut locked_err);
+        for triple in processor_graph.iter() {
+            writer.serialize_triple(triple)?;
+        }
+
+        writer.finish()?;
+        drop(processor_graph);
+    }
+
+    {
+        // use serializer with all known prefixes
+        let serializer = html2rdf::initial_context_prefixes().mappings().try_fold(
+            oxttl::TurtleSerializer::new().with_base_iri(base)?,
+            |serializer, (prefix, value)| serializer.with_prefix(prefix, value),
+        )?;
+
+        let mut locked_out = std::io::stdout().lock();
+        let mut writer = serializer.for_writer(&mut locked_out);
+        for triple in output_graph.iter() {
+            writer.serialize_triple(triple)?;
+        }
+
+        writer.finish()?;
+        drop(output_graph);
+    }
+
+    Ok(ExitCode::SUCCESS)
+}

html2rdf-cli/src/main.rs

@@ -0,0 +1,31 @@
+[package]
+name = "html2rdf"
+version = "0.1.0"
+authors.workspace = true
+edition.workspace = true
+keywords.workspace = true
+license.workspace = true
+publish.workspace = true
+readme.workspace = true
+repository.workspace = true
+
+[dependencies]
+curie = "0.1.4"
+derive_more = { version = "2.0.1", features = ["display", "error", "from", "from_str"] }
+icu = { version = "2.0.0-beta2" }
+indexmap = "2.9.0"
+itertools = "0.14.0"
+oxiri = "0.2.11"
+oxrdf = "0.2.4"
+oxsdatatypes = "0.2.2"
+rdf-canon = "0.15.1"
+rxml_validation = "0.11.0"
+scraper = "0.23.1"
+vec1 = "1.12.1"
+
+[dev-dependencies]
+insta = { version = "1.43.1", features = ["glob"] }
+oxttl = "0.1.8"
+pretty_assertions = "1.4.1"
+rstest = { version = "0.25.0", default-features = false, features = ["crate-name"] }
+sha2 = "0.10.9"