Add support for parsing list of status codes from config file (#636)

fixtures/configs/smoketest.toml

@@ -46,7 +46,7 @@ timeout = 20
 retry_wait_time = 2
 
 # Comma-separated list of accepted status codes for valid links.
-accept = "403, 403"
+accept = [200, 429]
 
 # Proceed for server connections considered insecure (invalid TLS).
 insecure = false

lychee-bin/src/client.rs

@@ -1,9 +1,8 @@
 use crate::options::Config;
-use crate::parse::{
-    parse_basic_auth, parse_duration_secs, parse_headers, parse_remaps, parse_statuscodes,
-};
+use crate::parse::{parse_basic_auth, parse_duration_secs, parse_headers, parse_remaps};
 use anyhow::{Context, Result};
 use headers::HeaderMapExt;
+use http::StatusCode;
 use lychee_lib::{Client, ClientBuilder};
 use regex::RegexSet;
 use std::{collections::HashSet, str::FromStr};
@@ -16,7 +15,6 @@ pub(crate) fn create(cfg: &Config) -> Result<Client> {
         headers.typed_insert(auth_header);
     }
 
-    let accepted = cfg.accept.clone().and_then(|a| parse_statuscodes(&a).ok());
     let timeout = parse_duration_secs(cfg.timeout);
     let retry_wait_time = parse_duration_secs(cfg.retry_wait_time);
     let method: reqwest::Method = reqwest::Method::from_str(&cfg.method.to_uppercase())?;
@@ -32,6 +30,17 @@ pub(crate) fn create(cfg: &Config) -> Result<Client> {
         cfg.scheme.clone()
     };
 
+    let accepted = match cfg.accept {
+        Some(ref accepted) => {
+            let accepted: Result<HashSet<_>, _> = accepted
+                .iter()
+                .map(|code| StatusCode::from_u16(*code))
+                .collect();
+            Some(accepted?)
+        }
+        None => None,
+    };
+
     ClientBuilder::builder()
         .remaps(remaps)
         .includes(includes)

lychee-bin/src/options.rs

@@ -1,5 +1,4 @@
-use std::{convert::TryFrom, fs, io::ErrorKind, path::PathBuf, str::FromStr, time::Duration};
-
+use crate::parse::{parse_base, parse_statuscodes};
 use anyhow::{anyhow, Context, Error, Result};
 use const_format::{concatcp, formatcp};
 use lychee_lib::{
@@ -8,6 +7,7 @@ use lychee_lib::{
 };
 use secrecy::{ExposeSecret, SecretString};
 use serde::Deserialize;
+use std::{collections::HashSet, fs, io::ErrorKind, path::PathBuf, str::FromStr, time::Duration};
 use structopt::StructOpt;
 
 pub(crate) const LYCHEE_IGNORE_FILE: &str = ".lycheeignore";
@@ -95,10 +95,6 @@ macro_rules! fold_in {
     };
 }
 
-fn parse_base(src: &str) -> Result<Base, lychee_lib::ErrorKind> {
-    Base::try_from(src)
-}
-
 #[derive(Debug, StructOpt)]
 #[structopt(
     name = "lychee",
@@ -272,9 +268,9 @@ pub(crate) struct Config {
     pub(crate) headers: Vec<String>,
 
     /// Comma-separated list of accepted status codes for valid links
-    #[structopt(short, long)]
+    #[structopt(short, long, parse(try_from_str = parse_statuscodes))]
     #[serde(default)]
-    pub(crate) accept: Option<String>,
+    pub(crate) accept: Option<HashSet<u16>>,
 
     /// Website timeout in seconds from connect to response finished
     #[structopt(short, long, default_value = &TIMEOUT_STR)]

lychee-bin/src/parse.rs

@@ -1,7 +1,6 @@
 use anyhow::{anyhow, Context, Result};
 use headers::{authorization::Basic, Authorization, HeaderMap, HeaderName};
-use http::StatusCode;
-use lychee_lib::remap::Remaps;
+use lychee_lib::{remap::Remaps, Base};
 use std::{collections::HashSet, time::Duration};
 
 /// Split a single HTTP header into a (key, value) tuple
@@ -31,16 +30,6 @@ pub(crate) fn parse_headers<T: AsRef<str>>(headers: &[T]) -> Result<HeaderMap> {
     Ok(out)
 }
 
-/// Parse HTTP status codes into a set of `StatusCode`
-pub(crate) fn parse_statuscodes<T: AsRef<str>>(accept: T) -> Result<HashSet<StatusCode>> {
-    let mut statuscodes = HashSet::new();
-    for code in accept.as_ref().split(',') {
-        let code: StatusCode = StatusCode::from_bytes(code.as_bytes())?;
-        statuscodes.insert(code);
-    }
-    Ok(statuscodes)
-}
-
 /// Parse URI remaps
 pub(crate) fn parse_remaps(remaps: &[String]) -> Result<Remaps> {
     Remaps::try_from(remaps)
@@ -59,12 +48,25 @@ pub(crate) fn parse_basic_auth(auth: &str) -> Result<Authorization<Basic>> {
     Ok(Authorization::basic(params[0], params[1]))
 }
 
+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`
+pub(crate) fn parse_statuscodes<T: AsRef<str>>(accept: T) -> Result<HashSet<u16>> {
+    let mut statuscodes = HashSet::new();
+    for code in accept.as_ref().split(',') {
+        let code: u16 = code.parse::<u16>()?;
+        statuscodes.insert(code);
+    }
+    Ok(statuscodes)
+}
+
 #[cfg(test)]
 mod test {
     use std::collections::HashSet;
 
     use headers::{HeaderMap, HeaderMapExt};
-    use http::StatusCode;
     use regex::Regex;
     use reqwest::{header, Url};
 
@@ -80,13 +82,7 @@ mod test {
     #[test]
     fn test_parse_statuscodes() {
         let actual = parse_statuscodes("200,204,301").unwrap();
-        let expected = IntoIterator::into_iter([
-            StatusCode::OK,
-            StatusCode::NO_CONTENT,
-            StatusCode::MOVED_PERMANENTLY,
-        ])
-        .collect::<HashSet<_>>();
-
+        let expected = IntoIterator::into_iter([200, 204, 301]).collect::<HashSet<_>>();
         assert_eq!(actual, expected);
     }
 

lychee.example.toml

@@ -45,7 +45,7 @@ timeout = 20
 retry_wait_time = 2
 
 # Comma-separated list of accepted status codes for valid links.
-accept = "403, 403"
+accept = [200, 429]
 
 # Proceed for server connections considered insecure (invalid TLS).
 insecure = false