Mailto support (#138)

* Add mailto suport and use try_from for parsing URLs
* Cleanup and document code

fixtures/TEST.md

@@ -22,3 +22,4 @@ https://example.com
 https://www.peerlyst.com/posts/a-list-of-static-analysis-tools-for-c-c-peerlyst
 
 test@example.com
+mailto:test2@example.com

src/extract.rs

@@ -5,8 +5,8 @@ use html5ever::tendril::{StrTendril, TendrilSink};
 use linkify::LinkFinder;
 use markup5ever_rcdom::{Handle, NodeData, RcDom};
 use pulldown_cmark::{Event as MDEvent, Parser, Tag};
-use std::collections::HashSet;
 use std::path::Path;
+use std::{collections::HashSet, convert::TryFrom};
 use url::Url;
 
 #[derive(Clone, Debug)]
@@ -152,14 +152,12 @@ pub(crate) fn extract_links(input_content: &InputContent, base_url: Option<Url>)
     // Silently ignore the parse failures for now.
     let mut uris = HashSet::new();
     for link in links {
-        match Url::parse(&link) {
-            Ok(url) => {
-                uris.insert(Uri::Website(url));
+        match Uri::try_from(link.as_str()) {
+            Ok(uri) => {
+                uris.insert(uri);
             }
             Err(_) => {
-                if link.contains('@') {
-                    uris.insert(Uri::Mail(link));
-                } else if !Path::new(&link).exists() {
+                if !Path::new(&link).exists() {
                     if let Some(base_url) = &base_url {
                         if let Ok(new_url) = base_url.join(&link) {
                             uris.insert(Uri::Website(new_url));

src/uri.rs

@@ -1,4 +1,4 @@
-use anyhow::Result;
+use anyhow::{bail, Result};
 use serde::{Deserialize, Serialize};
 use std::net::IpAddr;
 use std::{convert::TryFrom, fmt::Display};
@@ -44,7 +44,16 @@ impl TryFrom<&str> for Uri {
     type Error = anyhow::Error;
 
     fn try_from(s: &str) -> Result<Self> {
-        Ok(Uri::Website(Url::parse(s)?))
+        // Remove the `mailto` scheme if it exists
+        // to avoid parsing it as a website URL.
+        let s = s.trim_start_matches("mailto:");
+        if let Ok(uri) = Url::parse(s) {
+            return Ok(Uri::Website(uri));
+        };
+        if s.contains('@') {
+            return Ok(Uri::Mail(s.to_string()));
+        };
+        bail!("Cannot convert to Uri")
     }
 }
 
@@ -61,6 +70,23 @@ mod test {
     use super::*;
     use std::net::{IpAddr, Ipv4Addr, Ipv6Addr};
 
+    #[test]
+    fn test_uri_from_str() {
+        assert!(matches!(Uri::try_from(""), Err(_)));
+        assert_eq!(
+            Uri::try_from("http://example.com").unwrap(),
+            Uri::Website(url::Url::parse("http://example.com").unwrap())
+        );
+        assert_eq!(
+            Uri::try_from("mail@example.com").unwrap(),
+            Uri::Mail("mail@example.com".to_string())
+        );
+        assert_eq!(
+            Uri::try_from("mailto:mail@example.com").unwrap(),
+            Uri::Mail("mail@example.com".to_string())
+        );
+    }
+
     #[test]
     fn test_uri_host_ip_v4() {
         let uri =
@@ -86,4 +112,12 @@ mod test {
         let ip = uri.host_ip();
         assert!(ip.is_none());
     }
+
+    #[test]
+    fn test_mail() {
+        let uri =
+            Uri::Website(Url::parse("http://127.0.0.1").expect("Expected URI with valid IPv4"));
+        let ip = uri.host_ip().expect("Expected a valid IPv4");
+        assert_eq!(ip, IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)));
+    }
 }

tests/cli.rs

@@ -231,7 +231,7 @@ mod cli {
             .assert()
             .success();
 
-        let expected = r##"{"total":10,"successful":10,"failures":[],"timeouts":[],"redirects":[],"excludes":[],"errors":[]}"##;
+        let expected = r##"{"total":11,"successful":11,"failures":[],"timeouts":[],"redirects":[],"excludes":[],"errors":[]}"##;
         let output = fs::read_to_string(&outfile)?;
         assert_eq!(output, expected);
         fs::remove_file(outfile)?;