feat: Add support for ranges in the --accept option / config field

fixtures/configs/smoketest.toml

@@ -46,14 +46,14 @@ timeout = 20
 retry_wait_time = 2
 
 # Comma-separated list of accepted status codes for valid links.
-accept = [200, 429]
+accept = "200, 429"
 
 # Proceed for server connections considered insecure (invalid TLS).
 insecure = false
 
 # Only test links with the given schemes (e.g. https).
 # Omit to check links with any scheme.
-scheme = [ "https" ]
+scheme = ["https"]
 
 # When links are available using HTTPS, treat HTTP links as errors.
 require_https = false
@@ -66,7 +66,7 @@ headers = []
 
 # Remap URI matching pattern to different URI.
 # This also supports (named) capturing groups.
-remap = [ 
+remap = [
     "https://example.com http://example.invalid",
     "https://example.com/(.*) http://example.org/$1",
     "https://github.com/(?P<org>.*)/(?P<repo>.*) https://gitlab.com/$org/$repo",
@@ -93,13 +93,13 @@ include_verbatim = false
 glob_ignore_case = false
 
 # Exclude URLs and mail addresses from checking (supports regex).
-exclude = [ '.*\.github.com\.*' ]
+exclude = ['.*\.github.com\.*']
 
 # Exclude these filesystem paths from getting checked.
 exclude_path = ["file/path/to/Ignore", "./other/file/path/to/Ignore"]
 
 # URLs to check (supports regex). Has preference over all excludes.
-include = [ 'gist\.github\.com.*' ]
+include = ['gist\.github\.com.*']
 
 # Exclude all private IPs from checking.
 # Equivalent to setting `exclude_private`, `exclude_link_local`, and

lychee-bin/src/client.rs

@@ -26,14 +26,15 @@ pub(crate) fn create(cfg: &Config, cookie_jar: Option<&Arc<CookieStoreMutex>>) -
         cfg.scheme.clone()
     };
 
-    let accepted = match cfg.accept {
-        Some(ref accepted) => {
-            let accepted: Result<HashSet<_>, _> = accepted
+    let accepted = match &cfg.accept {
+        Some(selector) => Some(
+            selector
+                .clone()
+                .into_set()
                 .iter()
-                .map(|code| StatusCode::from_u16(*code))
-                .collect();
-            Some(accepted?)
-        }
+                .map(|value| StatusCode::from_u16(*value))
+                .collect::<Result<HashSet<_>, _>>()?,
+        ),
         None => None,
     };
 

lychee-bin/src/commands/check.rs

@@ -44,7 +44,10 @@ where
 
     let client = params.client;
     let cache = params.cache;
-    let accept = params.cfg.accept;
+    let accept = match params.cfg.accept {
+        Some(selector) => Some(selector.into_set()),
+        None => None,
+    };
 
     let pb = if params.cfg.no_progress || params.cfg.verbose.log_level() >= log::Level::Info {
         None

lychee-bin/src/options.rs

@@ -1,17 +1,17 @@
 use crate::archive::Archive;
-use crate::parse::{parse_base, parse_statuscodes};
+use crate::parse::parse_base;
 use crate::verbosity::Verbosity;
 use anyhow::{anyhow, Context, Error, Result};
 use clap::{arg, builder::TypedValueParser, Parser};
 use const_format::{concatcp, formatcp};
 use lychee_lib::{
-    Base, BasicAuthSelector, Input, DEFAULT_MAX_REDIRECTS, DEFAULT_MAX_RETRIES,
+    AcceptSelector, Base, BasicAuthSelector, Input, DEFAULT_MAX_REDIRECTS, DEFAULT_MAX_RETRIES,
     DEFAULT_RETRY_WAIT_TIME_SECS, DEFAULT_TIMEOUT_SECS, DEFAULT_USER_AGENT,
 };
 use secrecy::{ExposeSecret, SecretString};
 use serde::Deserialize;
 use std::path::Path;
-use std::{collections::HashSet, fs, path::PathBuf, str::FromStr, time::Duration};
+use std::{fs, path::PathBuf, str::FromStr, time::Duration};
 use strum::VariantNames;
 
 pub(crate) const LYCHEE_IGNORE_FILE: &str = ".lycheeignore";
@@ -297,9 +297,9 @@ pub(crate) struct Config {
     pub(crate) header: Vec<String>,
 
     /// Comma-separated list of accepted status codes for valid links
-    #[arg(short, long, value_parser = parse_statuscodes)]
+    #[arg(short, long)]
     #[serde(default)]
-    pub(crate) accept: Option<HashSet<u16>>,
+    pub(crate) accept: Option<AcceptSelector>,
 
     /// Website timeout in seconds from connect to response finished
     #[arg(short, long, default_value = &TIMEOUT_STR)]

lychee-bin/src/parse.rs

@@ -1,7 +1,7 @@
 use anyhow::{anyhow, Context, Result};
 use headers::{HeaderMap, HeaderName};
 use lychee_lib::{remap::Remaps, Base};
-use std::{collections::HashSet, time::Duration};
+use std::time::Duration;
 
 /// Split a single HTTP header into a (key, value) tuple
 fn read_header(input: &str) -> Result<(String, String)> {
@@ -40,24 +40,8 @@ pub(crate) fn parse_base(src: &str) -> Result<Base, lychee_lib::ErrorKind> {
     Base::try_from(src)
 }
 
-/// Parse HTTP status codes into a set of `StatusCode`
-///
-/// Note that this function does not convert the status codes into
-/// `StatusCode` but rather into `u16` to avoid the need for
-/// `http` as a dependency and to support custom status codes, which are
-/// necessary for some websites, which don't adhere to the HTTP spec or IANA.
-pub(crate) fn parse_statuscodes(accept: &str) -> Result<HashSet<u16>> {
-    let mut statuscodes = HashSet::new();
-    for code in accept.split(',') {
-        let code: u16 = code.parse::<u16>()?;
-        statuscodes.insert(code);
-    }
-    Ok(statuscodes)
-}
-
 #[cfg(test)]
 mod tests {
-    use std::collections::HashSet;
 
     use headers::HeaderMap;
     use regex::Regex;
@@ -72,13 +56,6 @@ mod tests {
         assert_eq!(parse_headers(&["accept=text/html"]).unwrap(), custom);
     }
 
-    #[test]
-    fn test_parse_statuscodes() {
-        let actual = parse_statuscodes("200,204,301").unwrap();
-        let expected = IntoIterator::into_iter([200, 204, 301]).collect::<HashSet<_>>();
-        assert_eq!(actual, expected);
-    }
-
     #[test]
     fn test_parse_remap() {
         let remaps =

lychee-lib/Cargo.toml

@@ -66,6 +66,7 @@ doc-comment = "0.3.3"
 tempfile = "3.6.0"
 wiremock = "0.5.19"
 serde_json = "1.0.103"
+rstest = "0.18.1"
 
 [features]
 

lychee-lib/src/lib.rs

@@ -91,8 +91,8 @@ pub use crate::{
     collector::Collector,
     filter::{Excludes, Filter, Includes},
     types::{
-        uri::valid::Uri, Base, BasicAuthCredentials, BasicAuthSelector, CacheStatus, CookieJar,
-        ErrorKind, FileType, Input, InputContent, InputSource, Request, Response, ResponseBody,
-        Result, Status,
+        uri::valid::Uri, AcceptRange, AcceptRangeError, AcceptSelector, Base, BasicAuthCredentials,
+        BasicAuthSelector, CacheStatus, CookieJar, ErrorKind, FileType, Input, InputContent,
+        InputSource, Request, Response, ResponseBody, Result, Status,
     },
 };

lychee-lib/src/types/accept/mod.rs

@@ -0,0 +1,5 @@
+mod range;
+mod selector;
+
+pub use range::*;
+pub use selector::*;