various link checker level changes

getzola · May 10, 2022 · 12dc16e · 12dc16e
1 parent 8b56180
commit 12dc16e
Show file tree

Hide file tree

Showing 6 changed files with 34 additions and 43 deletions.
diff --git a/components/markdown/src/markdown.rs b/components/markdown/src/markdown.rs
@@ -1,12 +1,13 @@
 use std::fmt::Write;
 
+use errors::bail;
 use libs::gh_emoji::Replacer as EmojiReplacer;
 use libs::once_cell::sync::Lazy;
 use libs::pulldown_cmark as cmark;
 use libs::tera;
 
 use crate::context::RenderContext;
-use errors::{anyhow, Context, Error, Result};
+use errors::{Context, Error, Result};
 use libs::pulldown_cmark::escape::escape_html;
 use utils::site::resolve_internal_link;
 use utils::slugs::slugify_anchors;
@@ -140,14 +141,14 @@ fn fix_link(
             }
             Err(_) => {
                 let msg = format!(
-                    "Dead relative link `{}` in {}",
+                    "Broken relative link `{}` in {}",
                     link,
                     context.current_page_path.unwrap_or("unknown"),
                 );
                 match context.config.link_checker.internal_level {
-                    config::LinkCheckerLevel::Error => return Err(anyhow!(msg)),
+                    config::LinkCheckerLevel::Error => bail!(msg),
                     config::LinkCheckerLevel::Warn => {
-                        console::error(&msg);
+                        console::warn(&msg);
                         link.to_string() // TODO document that broken internal relative links will show up in the HTML href as "@/path"
                     }
                 }

diff --git a/components/site/src/lib.rs b/components/site/src/lib.rs
@@ -6,7 +6,6 @@ pub mod sitemap;
 pub mod tpls;
 
 use std::collections::HashMap;
-use std::fmt::format;
 use std::fs::remove_dir_all;
 use std::path::{Path, PathBuf};
 use std::sync::{Arc, Mutex, RwLock};
@@ -306,7 +305,7 @@ impl Site {
                 .map(|(i, msg)| format!("  {}. {}", i + 1, msg))
                 .collect();
             let msg = format!(
-                "Found {} broken internal anchor links\n{}",
+                "Found {} broken internal anchor link(s)\n{}",
                 messages.len(),
                 messages.join("\n")
             );
@@ -326,7 +325,7 @@ impl Site {
                     .map(|(i, msg)| format!("  {}. {}", i + 1, msg))
                     .collect();
                 let msg = format!(
-                    "Found {} broken external links\n{}",
+                    "Found {} broken external link(s)\n{}",
                     messages.len(),
                     messages.join("\n")
                 );

diff --git a/components/site/src/link_checking.rs b/components/site/src/link_checking.rs
@@ -4,7 +4,7 @@ use std::{collections::HashMap, path::PathBuf, thread};
 use config::LinkCheckerLevel;
 use libs::rayon::prelude::*;
 
-use crate::{anyhow, Site};
+use crate::Site;
 use errors::{bail, Result};
 use libs::rayon;
 use libs::url::Url;
@@ -173,14 +173,15 @@ pub fn check_external_links(site: &Site) -> Result<(), Vec<String>> {
         }
     );
 
-    if !invalid_url_links.is_empty() {
-        for err in invalid_url_links.into_iter() {
-            let msg = err.domain.as_ref().unwrap_err();
-            match site.config.link_checker.internal_level {
-                LinkCheckerLevel::Error => console::error(&msg.to_string()),
-                LinkCheckerLevel::Warn => console::warn(&msg.to_string()),
-            }
-        }
+    for err in invalid_url_links.into_iter() {
+        let msg = err.domain.as_ref().unwrap_err().to_string();
+        messages.push(msg);
+    }
+
+    // error out if we're in error mode and any external URLs couldn't be parsed
+    match site.config.link_checker.external_level {
+        LinkCheckerLevel::Error if messages.len() > 0 => return Err(messages),
+        _ => (),
     }
 
     let mut links_by_domain: HashMap<String, Vec<&LinkDef>> = HashMap::new();
@@ -248,7 +249,7 @@ pub fn check_external_links(site: &Site) -> Result<(), Vec<String>> {
 
             for (page_path, link, check_res) in errors.iter() {
                 messages.push(format!(
-                    "Dead link in {} to {}: {}",
+                    "Broken link in {} to {}: {}",
                     page_path.to_string_lossy(),
                     link,
                     link_checker::message(check_res)
@@ -258,10 +259,5 @@ pub fn check_external_links(site: &Site) -> Result<(), Vec<String>> {
         Err(pool_err) => messages.push(pool_err.to_string()),
     }
 
-    // match site.config.link_checker.external_level {
-    //     LinkCheckerLevel::Error => Err(anyhow!("Dead links found")),
-    //     LinkCheckerLevel::Warn => Ok(()),
-    // }
-
     Err(messages)
 }
diff --git a/link-checker-level-test/config.toml b/link-checker-level-test/config.toml
@@ -16,5 +16,5 @@ highlight_code = false
 # Put all your custom variables here
 
 [link_checker]
-internal_level = "error"
-external_level = "error"
+internal_level = "warn"
+external_level = "warn"
diff --git a/link-checker-level-test/content/post.md b/link-checker-level-test/content/post.md
@@ -3,11 +3,10 @@ template = "single.html"
 +++
 
 [good internal link](./)
-[no DNS external link](https://sddfsdjf.com)
-[404 external link](https://clayto.com/1234)
-[invalid domain external link](https://cla<yto.com/)
-[invalid domain external link](https://cla<yto.com/)
-[invalid domain external link](https://cla<yto.com/)
-[invalid domain external link](https://cla<yto.com/)
-[invalid domain external link](https://cla<yto.com/)
-[invalid domain external link](https://cla<yto.com/)
+[bad anchor link 1](#foo)
+[bad anchor link 2](#bar)
+[bad internal link 1](@/foo)
+[bad internal link 2](@/bar)
+[good external link 1](https://getzola.org)
+[bad external link 2](https://cla<yto.com/)
+[external 404](https://clayto.com/foo)
diff --git a/link-checker-level-test/public/post/index.html b/link-checker-level-test/public/post/index.html
@@ -1,13 +1,9 @@
-<p><a href="https://example.com/post/#foo">bad anchor link 1</a>
+<p><a href="./">good internal link</a>
+<a href="https://example.com/post/#foo">bad anchor link 1</a>
 <a href="https://example.com/post/#bar">bad anchor link 2</a>
-<a href="https://example.com/post/#baz">bad anchor link 3</a>
-<a href="./">good internal link</a>
-<a href="https://sddfsdjf.com">no DNS external link</a>
-<a href="https://clayto.com/1234">404 external link</a>
-<a href="https://cla%3Cyto.com/">invalid domain external link</a>
-<a href="https://cla%3Cyto.com/">invalid domain external link</a>
-<a href="https://cla%3Cyto.com/">invalid domain external link</a>
-<a href="https://cla%3Cyto.com/">invalid domain external link</a>
-<a href="https://cla%3Cyto.com/">invalid domain external link</a>
-<a href="https://cla%3Cyto.com/">invalid domain external link</a></p>
+<a href="@/foo">bad internal link 1</a>
+<a href="@/bar">bad internal link 2</a>
+<a href="https://g%7Betzola.org">invalid domain external link 1</a>
+<a href="https://cla%3Cyto.com/">invalid domain external link 2</a></p>
+<!-- [external 404](https://clayto.com/foo) -->