init

src/main.rs

@@ -0,0 +1,72 @@
+#[macro_use]
+extern crate structopt;
+use structopt::StructOpt;
+
+mod opts;
+
+use opts::Opts;
+use std::error::Error;
+use std::fs::File;
+use std::io;
+use std::io::{stdin, Read};
+use std::path::PathBuf;
+extern crate kuchiki;
+
+use kuchiki::traits::*;
+
+fn read_html<I: Read>(source: &mut I) -> io::Result<String> {
+    let mut html = String::new();
+    source.read_to_string(&mut html)?;
+    Ok(html)
+}
+
+fn main() -> Result<(), Box<Error>> {
+    let stdin = stdin();
+    let opt = Opts::from_args();
+
+    let html = if let Some(path) = opt.input {
+        let mut file = File::open(&path)?;
+        read_html(&mut file)?
+    } else {
+        let mut handle = stdin.lock();
+        read_html(&mut handle)?
+    };
+
+    let css_selector = opt.selector;
+
+    let document = kuchiki::parse_html().one(html);
+
+    for css_match in document.select(&css_selector).unwrap() {
+        // css_match is a NodeDataRef, but most of the interesting methods are
+        // on NodeRef. Let's get the underlying NodeRef.
+        let as_node = css_match.as_node();
+
+        // In this example, as_node represents an HTML node like
+        //
+        //   <p class='foo'>Hello world!</p>"
+        //
+        // Which is distinct from just 'Hello world!'. To get rid of that <p>
+        // tag, we're going to get each element's first child, which will be
+        // a "text" node.
+        //
+        // There are other kinds of nodes, of course. The possibilities are all
+        // listed in the `NodeData` enum in this crate.
+        //let text_node = as_node.first_child().unwrap();
+
+        // Let's get the actual text in this text node. A text node wraps around
+        // a RefCell<String>, so we need to call borrow() to get a &str out.
+        //let text = text_node.as_text().unwrap().borrow();
+
+        if let Some(child) = as_node.first_child() {
+            //TODO: Convert Nodes to String, as this only works for Nodes containing plain text
+            if let Some(text) = child.as_text() {
+                // Prints:
+                //
+                //  "Hello, world!"
+                //  "I love HTML"
+                println!("{:?}", text.borrow());
+            }
+        }
+    }
+    Ok(())
+}

src/opts.rs

@@ -0,0 +1,12 @@
+use std::path::PathBuf;
+use structopt::StructOpt;
+
+#[derive(StructOpt, Debug)]
+#[structopt(name = "htmlextract")]
+pub struct Opts {
+    #[structopt(short = "s", long = "selector")]
+    pub selector: String,
+
+    #[structopt(parse(from_os_str))]
+    pub input: Option<PathBuf>,
+}