From 8691b0166f9d046cb2490b6754834fdf87059732 Mon Sep 17 00:00:00 2001
From: Kenneth Gitere <gitere81@gmail.com>
Date: Tue, 8 Jun 2021 20:35:52 +0300
Subject: [PATCH] fix: fix panic when unwrapping a base URI

chore: add message when downloading articles to a specified output-dir
---
 src/main.rs                | 11 +++++++++++
 src/moz_readability/mod.rs | 29 +++++++++++++++++++++++------
 2 files changed, 34 insertions(+), 6 deletions(-)

diff --git a/src/main.rs b/src/main.rs
index dc4787d..e378115 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -39,6 +39,16 @@ fn main() {
 fn run(app_config: AppConfig) {
     let mut errors = Vec::new();
     let mut partial_downloads = Vec::new();
+
+    if let Some(dir_name) = &app_config.output_directory {
+        let noun = if app_config.urls.len() > 1 {
+            "articles"
+        } else {
+            "article"
+        };
+        println!("Downloading {} to {}", noun, dir_name);
+    }
+
     let bar = if app_config.can_disable_progress_bar {
         ProgressBar::hidden()
     } else {
@@ -50,6 +60,7 @@ fn run(app_config: AppConfig) {
         enabled_bar.enable_steady_tick(500);
         enabled_bar
     };
+
     let articles = download(&app_config, &bar, &mut partial_downloads, &mut errors);
     bar.finish_with_message("Downloaded articles");
 
diff --git a/src/moz_readability/mod.rs b/src/moz_readability/mod.rs
index 7549f24..0f4fc66 100644
--- a/src/moz_readability/mod.rs
+++ b/src/moz_readability/mod.rs
@@ -659,10 +659,24 @@ impl Readability {
                 .map(|node_ref| {
                     let node_attrs = node_ref.attributes.borrow();
                     let href = node_attrs.get("href").unwrap();
-                    if href.trim() == "/" {
-                        document_uri.join("/").unwrap()
-                    } else {
-                        Url::parse(href).unwrap()
+
+                    match Url::parse(href) {
+                        Ok(url) => url,
+                        Err(e) => match e {
+                            url::ParseError::RelativeUrlWithoutBase => {
+                                match document_uri.join(href) {
+                                    Ok(joined_url) => joined_url,
+                                    Err(e) => panic!(
+                                        "{:} unable to parse url {:?} on element {}",
+                                        e, href, &node_ref.name.local
+                                    ),
+                                }
+                            }
+                            e => panic!(
+                                "{:} unable to parse url {:?} on element {}",
+                                e, href, &node_ref.name.local
+                            ),
+                        },
                     }
                 })
                 .next()
@@ -1609,8 +1623,11 @@ impl Readability {
             //   // class name "comment", etc), and turn divs into P tags where they have been
             //   // used inappropriately (as in, where they contain no other block level elements.)
             let mut elements_to_score: Vec<NodeRef> = Vec::new();
-            let mut node = self.root_node.select_first("html")
-                .ok().map(|n| n.as_node().clone());
+            let mut node = self
+                .root_node
+                .select_first("html")
+                .ok()
+                .map(|n| n.as_node().clone());
 
             while let Some(node_ref) = node {
                 let node_elem = node_ref.as_element().unwrap();