Ignore licenses for listed private sources (#391)

* Ignore licenses for listed private sources

* Clippy fix and better comments

src/lib.rs

@@ -263,6 +263,17 @@ impl Krate {
             }
         })
     }
+
+    /// Returns the normalized source URL
+    pub(crate) fn normalized_source_url(&self) -> Option<url::Url> {
+        self.source.as_ref().map(|source| {
+            let mut url = source.url().clone();
+            url.set_query(None);
+            url.set_fragment(None);
+            crate::sources::normalize_url(&mut url);
+            url
+        })
+    }
 }
 
 impl fmt::Display for Krate {

src/licenses.rs

@@ -282,13 +282,18 @@ pub fn check(
         let mut pack = Pack::with_kid(Check::Licenses, krate_lic_nfo.krate.id.clone());
 
         // If the user has set this, check if it's a private workspace
-        // crate and just print out a help message that we skipped it
+        // crate or a crate from a private registry and just print out
+        // a help message that we skipped it
         if ctx.cfg.private.ignore
-            && ctx
-                .krates
-                .workspace_members()
-                .any(|wm| wm.id == krate_lic_nfo.krate.id)
-            && krate_lic_nfo.krate.is_private(&private_registries)
+            && (krate_lic_nfo
+                .krate
+                .normalized_source_url()
+                .map_or(false, |source| ctx.cfg.ignore_sources.contains(&source))
+                || (ctx
+                    .krates
+                    .workspace_members()
+                    .any(|wm| wm.id == krate_lic_nfo.krate.id)
+                    && krate_lic_nfo.krate.is_private(&private_registries)))
         {
             pack.push(diags::SkippedPrivateWorkspaceCrate {
                 krate: krate_lic_nfo.krate,

src/licenses/cfg.rs

@@ -65,6 +65,10 @@ pub struct Private {
     /// only published to private registries
     #[serde(default)]
     pub ignore: bool,
+    /// One or more URLs to private registries, if a crate comes from one
+    /// of these registries, the crate will not have its license checked
+    #[serde(default)]
+    pub ignore_sources: Vec<Spanned<String>>,
     /// One or more private registries that you might publish crates to, if
     /// a crate is only published to private registries, and ignore is true
     /// the crate will not have its license checked
@@ -198,6 +202,24 @@ impl crate::cfg::UnvalidatedConfig for Config {
     fn validate(self, cfg_file: FileId, diags: &mut Vec<Diagnostic>) -> Self::ValidCfg {
         use rayon::prelude::*;
 
+        let mut ignore_sources = Vec::with_capacity(self.private.ignore_sources.len());
+        for aurl in &self.private.ignore_sources {
+            match url::Url::parse(aurl.as_ref()) {
+                Ok(mut url) => {
+                    crate::sources::normalize_url(&mut url);
+                    ignore_sources.push(url);
+                }
+                Err(pe) => {
+                    diags.push(
+                        Diagnostic::error()
+                            .with_message("failed to parse url")
+                            .with_labels(vec![Label::primary(cfg_file, aurl.span.clone())
+                                .with_message(pe.to_string())]),
+                    );
+                }
+            }
+        }
+
         let mut parse_license = |ls: &Spanned<String>, v: &mut Vec<Licensee>| {
             match spdx::Licensee::parse(ls.as_ref()) {
                 Ok(licensee) => {
@@ -306,6 +328,7 @@ impl crate::cfg::UnvalidatedConfig for Config {
             exceptions,
             denied,
             allowed,
+            ignore_sources,
         }
     }
 }
@@ -344,6 +367,7 @@ pub struct ValidConfig {
     pub allowed: Vec<Licensee>,
     pub clarifications: Vec<ValidClarification>,
     pub exceptions: Vec<ValidException>,
+    pub ignore_sources: Vec<url::Url>,
 }
 
 #[cfg(test)]

src/sources.rs

@@ -214,7 +214,7 @@ pub fn check(ctx: crate::CheckCtx<'_, ValidConfig>, mut sink: ErrorSink) {
     }
 }
 
-fn normalize_url(url: &mut Url) {
+pub(crate) fn normalize_url(url: &mut Url) {
     // Normalizes the URL so that different representations can be compared to each other.
     // At the moment we just remove a tailing `.git` but there are more possible optimisations.
     // See https://github.com/rust-lang/cargo/blob/1f6c6bd5e7bbdf596f7e88e6db347af5268ab113/src/cargo/util/canonical_url.rs#L31-L57