refine CLI API for white/black-listing of domains

pull/323/head
Sunshine 2 years ago
parent 7c0504c4cb
commit c1edde9b3e

@ -111,11 +111,11 @@ cat index.html | monolith -aIiFfcMv -b https://original.site/ - > result.html
- `-a`: Exclude audio sources - `-a`: Exclude audio sources
- `-b`: Use custom `base URL` - `-b`: Use custom `base URL`
- `-B`: Forbid retrieving assets from specified domain(s)
- `-c`: Exclude CSS - `-c`: Exclude CSS
- `-C`: Save document using custom `charset` - `-C`: Save document using custom `charset`
- `-d`: Allow retrieving assets only from specified `domain(s)` - `-d`: Allow retrieving assets only from specified `domain(s)`
- `-e`: Ignore network errors - `-e`: Ignore network errors
- `-E`: Avoid retrieving assets located within specified domains
- `-f`: Omit frames - `-f`: Omit frames
- `-F`: Exclude web fonts - `-F`: Exclude web fonts
- `-i`: Remove images - `-i`: Remove images

@ -5,11 +5,11 @@ use std::env;
pub struct Options { pub struct Options {
pub no_audio: bool, pub no_audio: bool,
pub base_url: Option<String>, pub base_url: Option<String>,
pub blacklist_domains: bool,
pub no_css: bool, pub no_css: bool,
pub charset: Option<String>, pub charset: Option<String>,
pub domains: Option<Vec<String>>, pub domains: Option<Vec<String>>,
pub ignore_errors: bool, pub ignore_errors: bool,
pub exclude_domains: bool,
pub no_frames: bool, pub no_frames: bool,
pub no_fonts: bool, pub no_fonts: bool,
pub no_images: bool, pub no_images: bool,
@ -50,19 +50,21 @@ impl Options {
.about(format!("{}\n{}", ASCII, env!("CARGO_PKG_DESCRIPTION")).as_str()) .about(format!("{}\n{}", ASCII, env!("CARGO_PKG_DESCRIPTION")).as_str())
.args_from_usage("-a, --no-audio 'Removes audio sources'") .args_from_usage("-a, --no-audio 'Removes audio sources'")
.args_from_usage("-b, --base-url=[http://localhost/] 'Sets custom base URL'") .args_from_usage("-b, --base-url=[http://localhost/] 'Sets custom base URL'")
.args_from_usage(
"-B, --blacklist-domains 'Treat list of specified domains as blacklist'",
)
.args_from_usage("-c, --no-css 'Removes CSS'") .args_from_usage("-c, --no-css 'Removes CSS'")
.args_from_usage("-C, --charset=[UTF-8] 'Enforces custom encoding'") .args_from_usage("-C, --charset=[UTF-8] 'Enforces custom encoding'")
.arg( .arg(
Arg::with_name("domains") Arg::with_name("domains")
.short('d') .short('d')
.long("domains") .long("domain")
.takes_value(true) .takes_value(true)
.value_name("DOMAINS") .value_name("example.com")
.action(ArgAction::Append) .action(ArgAction::Append)
.help("Whitelist of domains"), .help("Specify domains to use for white/black-listing"),
) )
.args_from_usage("-e, --ignore-errors 'Ignore network errors'") .args_from_usage("-e, --ignore-errors 'Ignore network errors'")
.args_from_usage("-E, --exclude-domains 'Treat specified domains as blacklist'")
.args_from_usage("-f, --no-frames 'Removes frames and iframes'") .args_from_usage("-f, --no-frames 'Removes frames and iframes'")
.args_from_usage("-F, --no-fonts 'Removes fonts'") .args_from_usage("-F, --no-fonts 'Removes fonts'")
.args_from_usage("-i, --no-images 'Removes images'") .args_from_usage("-i, --no-images 'Removes images'")
@ -99,6 +101,7 @@ impl Options {
if let Some(base_url) = app.value_of("base-url") { if let Some(base_url) = app.value_of("base-url") {
options.base_url = Some(base_url.to_string()); options.base_url = Some(base_url.to_string());
} }
options.blacklist_domains = app.is_present("blacklist-domains");
options.no_css = app.is_present("no-css"); options.no_css = app.is_present("no-css");
if let Some(charset) = app.value_of("charset") { if let Some(charset) = app.value_of("charset") {
options.charset = Some(charset.to_string()); options.charset = Some(charset.to_string());
@ -108,7 +111,6 @@ impl Options {
options.domains = Some(list_of_domains); options.domains = Some(list_of_domains);
} }
options.ignore_errors = app.is_present("ignore-errors"); options.ignore_errors = app.is_present("ignore-errors");
options.exclude_domains = app.is_present("exclude-domains");
options.no_frames = app.is_present("no-frames"); options.no_frames = app.is_present("no-frames");
options.no_fonts = app.is_present("no-fonts"); options.no_fonts = app.is_present("no-fonts");
options.no_images = app.is_present("no-images"); options.no_images = app.is_present("no-images");

@ -296,8 +296,8 @@ pub fn retrieve_asset(
let domain_matches = domains let domain_matches = domains
.iter() .iter()
.any(|d| domain_is_within_domain(url.host_str().unwrap(), &d.trim())); .any(|d| domain_is_within_domain(url.host_str().unwrap(), &d.trim()));
if (options.exclude_domains && domain_matches) if (options.blacklist_domains && domain_matches)
|| (!options.exclude_domains && !domain_matches) || (!options.blacklist_domains && !domain_matches)
{ {
return Err(client.get("").send().unwrap_err()); return Err(client.get("").send().unwrap_err());
} }

Loading…
Cancel
Save