Convert action attribute to full URL for FORM tags

5 years ago · ffa60a0ce2
parent cc09c2ba01
commit ffa60a0ce2
3 changed files with 25 additions and 5 deletions
--- a/Cargo.toml
+++ b/Cargo.toml
@ -1,6 +1,6 @@
 [package]
 name = "monolith"
-version = "2.0.1"
+version = "2.0.2"
 authors = ["Sunshine <sunshine@uberspace.net>"]

 [dependencies]
--- a/src/html.rs
+++ b/src/html.rs
@ -2,7 +2,7 @@ extern crate html5ever;

 use std::default::Default;
 use std::io;
-use http::{retrieve_asset, resolve_url};
+use http::{is_url, retrieve_asset, resolve_url};

 use self::html5ever::parse_document;
 use self::html5ever::rcdom::{Handle, NodeData, RcDom};
@ -15,6 +15,7 @@ enum NodeMatch {
    StyleSheet,
    Anchor,
    Script,
+    Form,
    Other,
 }

@ -78,6 +79,8 @@ pub fn walk_and_embed_assets(url: &str, node: &Handle, opt_no_js: bool) {
                found = NodeMatch::Anchor;
            } else if &name.local == "script" {
                found = NodeMatch::Script;
+            } else if &name.local == "form" {
+                found = NodeMatch::Form;
            }

            match found {
@ -145,6 +148,20 @@ pub fn walk_and_embed_assets(url: &str, node: &Handle, opt_no_js: bool) {
                        }
                    }
                },
+                NodeMatch::Form => {
+                    for attr in attrs_mut.iter_mut() {
+                        if &attr.name.local == "action" {
+                            // Do not touch action props which are set to a URL
+                            if is_url(&attr.value) {
+                                continue;
+                            }
+
+                            let href_full_url = resolve_url(&url, &attr.value.to_string());
+                            attr.value.clear();
+                            attr.value.push_slice(href_full_url.unwrap().as_str());
+                        }
+                    }
+                },
                NodeMatch::Other => {},
            }

@ -174,7 +191,8 @@ pub fn html_to_dom(data: &str) -> html5ever::rcdom::RcDom {
        .unwrap()
 }

-pub fn print_dom(handle: &Handle) {
+pub fn print_dom(handle: &Handle, _opt_isolate: bool) {
+    // TODO: append <meta http-equiv="Access-Control-Allow-Origin" content="'self'"/> to the <head> if opt_isolate
    serialize(&mut io::stdout(), handle, SerializeOpts::default()).unwrap();
 }

--- a/src/main.rs
+++ b/src/main.rs
@ -7,7 +7,7 @@ use monolith::html::{walk_and_embed_assets, html_to_dom, print_dom};

 fn main() {
    let command = App::new("monolith")
-        .version("2.0.1")
+        .version("2.0.2")
        .author("Sunshine <sunshine@uberspace.net>")
        .about("CLI tool to save web pages as single HTML files")
        .arg(Arg::with_name("url")
@ -16,11 +16,13 @@ fn main() {
                 .index(1)
                 .help("URL to download"))
        .args_from_usage("-j, --nojs 'Remove JavaScript'")
+        // .args_from_usage("-i, --isolate 'Isolate the document'")
        .get_matches();

    // Process the command
    let arg_target = command.value_of("url").unwrap();
    let opt_no_js = command.is_present("nojs");
+    let opt_isolate = command.is_present("isolate");

    if is_url(arg_target) {
        let data = retrieve_asset(&arg_target, false, "");
@ -28,6 +30,6 @@ fn main() {

        walk_and_embed_assets(&arg_target, &dom.document, opt_no_js);

-        print_dom(&dom.document);
+        print_dom(&dom.document, opt_isolate);
    }
 }