Revamp resolve_url() and improve code format

pull/28/head
Vincent Flyson 5 years ago
parent fdbe2499df
commit c34d77d5d8

@ -1,6 +1,6 @@
[package]
name = "monolith"
version = "2.0.8"
version = "2.0.9"
authors = ["Sunshine <sunshine@uberspace.net>"]
description = "CLI tool for saving web pages as a single HTML file"

@ -19,7 +19,8 @@ enum NodeMatch {
Other,
}
const PNG_PIXEL: &str = "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNkYAAAAAYAAjCB0C8AAAAASUVORK5CYII=";
const TRANSPARENT_PIXEL: &str = "data:image/png;base64,\
iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNkYAAAAAYAAjCB0C8AAAAASUVORK5CYII=";
const JS_DOM_EVENT_ATTRS: [&str; 21] = [
// Input
@ -76,7 +77,8 @@ pub fn walk_and_embed_assets(
NodeData::Comment { .. } => {
// Note: in case of opt_no_js being set to true, there's no need to worry about
// getting rid of comments that may contain scripts, e.g. <!--[if IE]><script>...
// since that's not part of W3C standard and gets ignored by browsers other than IE [5, 9]
// since that's not part of W3C standard and therefore gets ignored
// by browsers other than IE [5, 9]
}
NodeData::Element {
@ -130,7 +132,7 @@ pub fn walk_and_embed_assets(
if &attr.name.local == "src" {
if opt_no_images {
attr.value.clear();
attr.value.push_slice(PNG_PIXEL);
attr.value.push_slice(TRANSPARENT_PIXEL);
} else {
let src_full_url = resolve_url(&url, &attr.value.to_string());
let img_datauri = retrieve_asset(
@ -224,7 +226,13 @@ pub fn walk_and_embed_assets(
opt_user_agent,
);
let dom = html_to_dom(&iframe_data.unwrap());
walk_and_embed_assets(&src_full_url, &dom.document, opt_no_js, opt_no_images, opt_user_agent);
walk_and_embed_assets(
&src_full_url,
&dom.document,
opt_no_js,
opt_no_images,
opt_user_agent,
);
let mut buf: Vec<u8> = Vec::new();
serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
let iframe_datauri = data_to_dataurl("text/html", &buf);

@ -9,6 +9,13 @@ lazy_static! {
static ref REGEX_URL: Regex = Regex::new(r"^https?://").unwrap();
}
pub fn is_data_url(url: &str) -> Result<bool, String> {
match Url::parse(url) {
Ok(parsed_url) => Ok(parsed_url.scheme() == "data"),
Err(err) => Err(format!("{}", err)),
}
}
pub fn is_valid_url(path: &str) -> bool {
REGEX_URL.is_match(path)
}
@ -18,59 +25,19 @@ pub fn resolve_url(from: &str, to: &str) -> Result<String, ParseError> {
// (anything, http://site.com/css/main.css)
to.to_string()
} else {
let mut re = String::new();
if is_valid_url(from) {
// It's a remote resource (HTTP)
if to.starts_with('/') {
// (http://site.com/article/1, /...?)
let from_url = Url::parse(from)?;
if to.starts_with("//") {
// (http://site.com/article/1, //images/1.png)
re.push_str(from_url.scheme());
re.push_str(":");
re.push_str(to);
} else {
// (http://site.com/article/1, /css/main.css)
re.push_str(from_url.scheme());
re.push_str("://");
re.push_str(from_url.host_str().unwrap());
re.push_str(to);
}
} else {
// (http://site.com, css/main.css)
// TODO improve to ensure no // or /// ever happen
let base = Url::parse(from)?;
re = base.join(to)?.to_string();
}
} else {
// It's a local resource (fs)
// TODO improve to ensure no // or /// ever happen
// TODO for fs use basepath instead of $from
re.push_str(from);
re.push_str("/");
re.push_str(to);
}
re
Url::parse(from)?.join(to)?.to_string()
};
Ok(result)
}
pub fn url_is_data(url: &str) -> Result<bool, String> {
match Url::parse(url) {
Ok(parsed_url) => Ok(parsed_url.scheme() == "data"),
Err(err) => Err(format!("{}", err)),
}
}
pub fn retrieve_asset(
url: &str,
as_dataurl: bool,
as_mime: &str,
opt_user_agent: &str,
) -> Result<String, reqwest::Error> {
if url_is_data(&url).unwrap() {
if is_data_url(&url).unwrap() {
Ok(url.to_string())
} else {
let client = Client::builder()
@ -161,6 +128,15 @@ mod tests {
"https://www.kernel.org/theme/images/logos/tux.png"
);
let resolved_url = resolve_url(
"https://www.kernel.org",
"//another-host.org/theme/images/logos/tux.png",
)?;
assert_eq!(
resolved_url.as_str(),
"https://another-host.org/theme/images/logos/tux.png"
);
let resolved_url = resolve_url(
"https://www.kernel.org/category/signatures.html",
"/theme/images/logos/tux.png",
@ -183,12 +159,12 @@ mod tests {
}
#[test]
fn test_url_is_data() {
fn test_is_data_url() {
assert!(
url_is_data("data:text/html;base64,V2VsY29tZSBUbyBUaGUgUGFydHksIDxiPlBhbDwvYj4h")
is_data_url("data:text/html;base64,V2VsY29tZSBUbyBUaGUgUGFydHksIDxiPlBhbDwvYj4h")
.unwrap_or(false)
);
assert!(!url_is_data("https://kernel.org").unwrap_or(false));
assert!(!url_is_data("//kernel.org").unwrap_or(false));
assert!(!is_data_url("https://kernel.org").unwrap_or(false));
assert!(!is_data_url("//kernel.org").unwrap_or(false));
}
}

@ -6,7 +6,8 @@ use clap::{App, Arg};
use monolith::html::{html_to_dom, print_dom, walk_and_embed_assets};
use monolith::http::{is_valid_url, retrieve_asset};
static DEFAULT_USER_AGENT: &str = "Mozilla/5.0 (Macintosh; Intel Mac OS X 10.14; rv:66.0) Gecko/20100101 Firefox/66.0";
static DEFAULT_USER_AGENT: &str =
"Mozilla/5.0 (Macintosh; Intel Mac OS X 10.14; rv:66.0) Gecko/20100101 Firefox/66.0";
fn main() {
let command = App::new("monolith")
@ -32,10 +33,16 @@ fn main() {
let opt_user_agent = command.value_of("user-agent").unwrap_or(DEFAULT_USER_AGENT);
if is_valid_url(arg_target) {
let data = retrieve_asset(&arg_target, false, "", opt_user_agent);
let dom = html_to_dom(&data.unwrap());
let data = retrieve_asset(&arg_target, false, "", opt_user_agent).unwrap();
let dom = html_to_dom(&data);
walk_and_embed_assets(&arg_target, &dom.document, opt_no_js, opt_no_images, opt_user_agent);
walk_and_embed_assets(
&arg_target,
&dom.document,
opt_no_js,
opt_no_images,
opt_user_agent,
);
print_dom(&dom.document);
println!(); // Ensure newline at end of output

Loading…
Cancel
Save