Mirror/monolith
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity
monolith/src/tests/html/walk_and_embed_assets.rs

374 lines
13 KiB
Rust
Raw Normal View History

improve test code structure
2020-03-23 03:08:41 +01:00
// ██████╗ █████╗ ███████╗███████╗██╗███╗ ██╗ ██████╗
// ██╔══██╗██╔══██╗██╔════╝██╔════╝██║████╗ ██║██╔════╝
// ██████╔╝███████║███████╗███████╗██║██╔██╗ ██║██║ ███╗
// ██╔═══╝ ██╔══██║╚════██║╚════██║██║██║╚██╗██║██║ ██║
// ██║ ██║ ██║███████║███████║██║██║ ╚████║╚██████╔╝
// ╚═╝ ╚═╝ ╚═╝╚══════╝╚══════╝╚═╝╚═╝ ╚═══╝ ╚═════╝
Improve code structure
2019-09-29 23:15:49 +02:00
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
#[cfg(test)]
mod passing {
use html5ever::serialize::{serialize, SerializeOpts};
use reqwest::blocking::Client;
use std::collections::HashMap;
revamp codebase
2021-03-11 23:44:02 +01:00
use url::Url;
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
pass options object instead of using separate parameters
2020-06-28 07:36:41 +02:00
use crate::html;
use crate::opts::Options;
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
#[test]
fn basic() {
let cache = &mut HashMap::new();
revamp codebase
2021-03-11 23:44:02 +01:00
let html: &str = "<div><P></P></div>";
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let dom = html::html_to_dom(&html);
revamp codebase
2021-03-11 23:44:02 +01:00
let url: Url = Url::parse("http://localhost").unwrap();
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
pass options object instead of using separate parameters
2020-06-28 07:36:41 +02:00
let mut options = Options::default();
options.silent = true;
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let client = Client::new();
indent items in retrieval log to form a tree-like structure
2020-06-28 22:11:15 +02:00
html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let mut buf: Vec<u8> = Vec::new();
serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
assert_eq!(
buf.iter().map(|&c| c as char).collect::<String>(),
"<html><head></head><body><div><p></p></div></body></html>"
);
}
#[test]
fn ensure_no_recursive_iframe() {
let html = "<div><P></P><iframe src=\"\"></iframe></div>";
let dom = html::html_to_dom(&html);
revamp codebase
2021-03-11 23:44:02 +01:00
let url: Url = Url::parse("http://localhost").unwrap();
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let cache = &mut HashMap::new();
pass options object instead of using separate parameters
2020-06-28 07:36:41 +02:00
let mut options = Options::default();
options.silent = true;
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let client = Client::new();
indent items in retrieval log to form a tree-like structure
2020-06-28 22:11:15 +02:00
html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let mut buf: Vec<u8> = Vec::new();
serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
assert_eq!(
buf.iter().map(|&c| c as char).collect::<String>(),
"<html><head></head><body><div><p></p><iframe src=\"\"></iframe></div></body></html>"
);
}
#[test]
fn ensure_no_recursive_frame() {
let html = "<frameset><frame src=\"\"></frameset>";
let dom = html::html_to_dom(&html);
revamp codebase
2021-03-11 23:44:02 +01:00
let url: Url = Url::parse("http://localhost").unwrap();
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let cache = &mut HashMap::new();
pass options object instead of using separate parameters
2020-06-28 07:36:41 +02:00
let mut options = Options::default();
options.silent = true;
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let client = Client::new();
indent items in retrieval log to form a tree-like structure
2020-06-28 22:11:15 +02:00
html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let mut buf: Vec<u8> = Vec::new();
serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
assert_eq!(
buf.iter().map(|&c| c as char).collect::<String>(),
"<html><head></head><frameset><frame src=\"\"></frameset></html>"
);
}
#[test]
fn no_css() {
let html = "<link rel=\"stylesheet\" href=\"main.css\">\
add test for alternate stylesheets
2021-03-09 12:48:41 +01:00
<link rel=\"alternate stylesheet\" href=\"main.css\">\
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
<style>html{background-color: #000;}</style>\
<div style=\"display: none;\"></div>";
let dom = html::html_to_dom(&html);
revamp codebase
2021-03-11 23:44:02 +01:00
let url: Url = Url::parse("http://localhost").unwrap();
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let cache = &mut HashMap::new();
pass options object instead of using separate parameters
2020-06-28 07:36:41 +02:00
let mut options = Options::default();
options.no_css = true;
options.silent = true;
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let client = Client::new();
indent items in retrieval log to form a tree-like structure
2020-06-28 22:11:15 +02:00
html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let mut buf: Vec<u8> = Vec::new();
serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
assert_eq!(
buf.iter().map(|&c| c as char).collect::<String>(),
add support for image inputs
2020-04-03 09:30:52 +02:00
"<html>\
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
<head>\
<link rel=\"stylesheet\">\
add test for alternate stylesheets
2021-03-09 12:48:41 +01:00
<link rel=\"alternate stylesheet\">\
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
<style></style>\
</head>\
<body>\
<div></div>\
</body>\
</html>"
);
}
#[test]
fn no_images() {
let html = "<link rel=\"icon\" href=\"favicon.ico\">\
<div><img src=\"http://localhost/assets/mono_lisa.png\" /></div>";
let dom = html::html_to_dom(&html);
revamp codebase
2021-03-11 23:44:02 +01:00
let url: Url = Url::parse("http://localhost").unwrap();
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let cache = &mut HashMap::new();
pass options object instead of using separate parameters
2020-06-28 07:36:41 +02:00
let mut options = Options::default();
options.no_images = true;
options.silent = true;
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let client = Client::new();
indent items in retrieval log to form a tree-like structure
2020-06-28 22:11:15 +02:00
html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let mut buf: Vec<u8> = Vec::new();
serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
assert_eq!(
buf.iter().map(|&c| c as char).collect::<String>(),
format!(
"<html>\
automatically remove "Refresh" and "Location" META tags
2020-06-26 05:53:20 +02:00
<head>\
<link rel=\"icon\">\
</head>\
<body>\
<div>\
<img src=\"{empty_image}\">\
</div>\
</body>\
</html>",
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
empty_image = empty_image!()
)
);
}
#[test]
fn no_body_background_images() {
let html =
"<body background=\"no/such/image.png\" background=\"no/such/image2.png\"></body>";
let dom = html::html_to_dom(&html);
revamp codebase
2021-03-11 23:44:02 +01:00
let url: Url = Url::parse("http://localhost").unwrap();
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let cache = &mut HashMap::new();
pass options object instead of using separate parameters
2020-06-28 07:36:41 +02:00
let mut options = Options::default();
options.no_images = true;
options.silent = true;
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let client = Client::new();
indent items in retrieval log to form a tree-like structure
2020-06-28 22:11:15 +02:00
html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let mut buf: Vec<u8> = Vec::new();
serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
assert_eq!(
buf.iter().map(|&c| c as char).collect::<String>(),
"<html><head></head><body></body></html>"
);
}
#[test]
fn no_frames() {
let html = "<frameset><frame src=\"http://trackbook.com\"></frameset>";
let dom = html::html_to_dom(&html);
revamp codebase
2021-03-11 23:44:02 +01:00
let url: Url = Url::parse("http://localhost").unwrap();
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let cache = &mut HashMap::new();
pass options object instead of using separate parameters
2020-06-28 07:36:41 +02:00
let mut options = Options::default();
options.no_frames = true;
options.silent = true;
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let client = Client::new();
indent items in retrieval log to form a tree-like structure
2020-06-28 22:11:15 +02:00
html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let mut buf: Vec<u8> = Vec::new();
serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
assert_eq!(
buf.iter().map(|&c| c as char).collect::<String>(),
"<html><head></head><frameset><frame src=\"\"></frameset></html>"
);
}
#[test]
fn no_iframes() {
let html = "<iframe src=\"http://trackbook.com\"></iframe>";
let dom = html::html_to_dom(&html);
revamp codebase
2021-03-11 23:44:02 +01:00
let url: Url = Url::parse("http://localhost").unwrap();
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let cache = &mut HashMap::new();
pass options object instead of using separate parameters
2020-06-28 07:36:41 +02:00
let mut options = Options::default();
options.no_frames = true;
options.silent = true;
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let client = Client::new();
indent items in retrieval log to form a tree-like structure
2020-06-28 22:11:15 +02:00
html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let mut buf: Vec<u8> = Vec::new();
serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
assert_eq!(
buf.iter().map(|&c| c as char).collect::<String>(),
"<html><head></head><body><iframe src=\"\"></iframe></body></html>"
);
}
#[test]
fn no_js() {
let html = "<div onClick=\"void(0)\">\
automatically remove "Refresh" and "Location" META tags
2020-06-26 05:53:20 +02:00
<script src=\"http://localhost/assets/some.js\"></script>\
<script>alert(1)</script>\
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
</div>";
let dom = html::html_to_dom(&html);
revamp codebase
2021-03-11 23:44:02 +01:00
let url: Url = Url::parse("http://localhost").unwrap();
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let cache = &mut HashMap::new();
pass options object instead of using separate parameters
2020-06-28 07:36:41 +02:00
let mut options = Options::default();
options.no_js = true;
options.silent = true;
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let client = Client::new();
indent items in retrieval log to form a tree-like structure
2020-06-28 22:11:15 +02:00
html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
let mut buf: Vec<u8> = Vec::new();
serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
assert_eq!(
buf.iter().map(|&c| c as char).collect::<String>(),
"<html><head></head><body><div><script></script>\
<script></script></div></body></html>"
);
}
revamp codebase
2021-03-11 23:44:02 +01:00
// #[test]
// fn discards_integrity() {
// let html = "<title>No integrity</title>\
// <link integrity=\"sha384-...\" rel=\"something\"/>\
// <script integrity=\"sha384-...\" src=\"some.js\"></script>";
// let dom = html::html_to_dom(&html);
// let url: Url = Url::parse("http://localhost").unwrap();
// let cache = &mut HashMap::new();
// let mut options = Options::default();
// options.no_css = true;
// options.no_frames = true;
// options.no_js = true;
// options.no_images = true;
// options.silent = true;
// let client = Client::new();
// html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
// let mut buf: Vec<u8> = Vec::new();
// serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
// assert_eq!(
// buf.iter().map(|&c| c as char).collect::<String>(),
// "<html>\
// <head><title>No integrity</title><link rel=\"something\"><script></script></head>\
// <body></body>\
// </html>"
// );
// }
automatically remove "Refresh" and "Location" META tags
2020-06-26 05:53:20 +02:00
#[test]
fn removes_unwanted_meta_tags() {
let html = "<html>\
<head>\
<meta http-equiv=\"Refresh\" value=\"20\"/>\
<meta http-equiv=\"Location\" value=\"https://freebsd.org\"/>\
</head>\
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
<body></body>\
automatically remove "Refresh" and "Location" META tags
2020-06-26 05:53:20 +02:00
</html>";
let dom = html::html_to_dom(&html);
revamp codebase
2021-03-11 23:44:02 +01:00
let url: Url = Url::parse("http://localhost").unwrap();
automatically remove "Refresh" and "Location" META tags
2020-06-26 05:53:20 +02:00
let cache = &mut HashMap::new();
pass options object instead of using separate parameters
2020-06-28 07:36:41 +02:00
let mut options = Options::default();
options.no_css = true;
options.no_frames = true;
options.no_js = true;
options.no_images = true;
options.silent = true;
automatically remove "Refresh" and "Location" META tags
2020-06-26 05:53:20 +02:00
let client = Client::new();
pass options object instead of using separate parameters
2020-06-28 07:36:41 +02:00
indent items in retrieval log to form a tree-like structure
2020-06-28 22:11:15 +02:00
html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
automatically remove "Refresh" and "Location" META tags
2020-06-26 05:53:20 +02:00
let mut buf: Vec<u8> = Vec::new();
serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
assert_eq!(
buf.iter().map(|&c| c as char).collect::<String>(),
"<html>\
<head>\
refactor code that processes the DOM
2020-12-25 22:09:47 +01:00
<meta http-equiv=\"disabled by monolith (Refresh)\" value=\"20\">\
<meta http-equiv=\"disabled by monolith (Location)\" value=\"https://freebsd.org\">\
automatically remove "Refresh" and "Location" META tags
2020-06-26 05:53:20 +02:00
</head>\
<body></body>\
group all tests into either passing or failing groups
2020-05-23 09:49:04 +02:00
</html>"
);
}
process contents of NOSCRIPT tags
2021-02-23 06:42:39 +01:00
#[test]
fn processes_noscript_tags() {
let html = "<html>\
<body>\
<noscript>\
<img src=\"image.png\" />\
</noscript>\
</body>\
</html>";
let dom = html::html_to_dom(&html);
revamp codebase
2021-03-11 23:44:02 +01:00
let url: Url = Url::parse("http://localhost").unwrap();
process contents of NOSCRIPT tags
2021-02-23 06:42:39 +01:00
let cache = &mut HashMap::new();
let mut options = Options::default();
options.no_images = true;
options.silent = true;
let client = Client::new();
html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
let mut buf: Vec<u8> = Vec::new();
serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
assert_eq!(
buf.iter().map(|&c| c as char).collect::<String>(),
format!(
"<html>\
<head>\
</head>\
<body>\
<noscript>\
<img src=\"{}\">\
</noscript>\
</body>\
</html>",
empty_image!(),
)
);
}
wipe integrity attributes
2019-12-26 15:44:01 +01:00
}
Reference in a new issue Copy permalink