From 94e339c3ce019dd0e7586c0b335f4144e8b132f8 Mon Sep 17 00:00:00 2001
From: Yumi Izumi <me@yumiizumi.com>
Date: Fri, 6 Dec 2024 14:58:59 -0500
Subject: [PATCH] fix: hrefs with whitespace wrapped in <>

Refs: #34
---
 Cargo.lock     |  9 +++++++++
 Cargo.toml     |  3 +++
 src/parser.rs  |  1 +
 src/structs.rs | 10 ++++++++++
 src/to_md.rs   | 25 +++++++++++++++++++++++--
 5 files changed, 46 insertions(+), 2 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index 79a68c1..b5381e6 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -5,3 +5,12 @@ version = 3
 [[package]]
 name = "html2md-rs"
 version = "0.10.0"
+dependencies = [
+ "percent-encoding",
+]
+
+[[package]]
+name = "percent-encoding"
+version = "2.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e3148f5046208a5d56bcfc03053e3ca6334e51da8dfb19b6cdc8b306fae3283e"
diff --git a/Cargo.toml b/Cargo.toml
index ed8c3b5..3a365b9 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -11,3 +11,6 @@ repository = "https://github.com/izyumidev/html2md-rs"
 license = "MIT"
 keywords = ["html", "markdown", "html2md", "html-to-markdown", "html-to-md"]
 categories = ["parsing"]
+
+[dependencies]
+percent-encoding = "2.3.1"
diff --git a/src/parser.rs b/src/parser.rs
index 01a5686..7de0773 100644
--- a/src/parser.rs
+++ b/src/parser.rs
@@ -607,6 +607,7 @@ fn issue_31() {
         attributes: Some(Attributes {
             id: None,
             class: None,
+            href: None,
             attributes: std::collections::HashMap::from([
                 (
                     "src".to_string(),
diff --git a/src/structs.rs b/src/structs.rs
index b53619d..da09458 100644
--- a/src/structs.rs
+++ b/src/structs.rs
@@ -138,6 +138,7 @@ impl Node {
 pub struct Attributes {
     pub(crate) id: Option<String>,
     pub(crate) class: Option<String>,
+    pub(crate) href: Option<String>,
     pub(crate) attributes: HashMap<String, AttributeValues>,
 }
 
@@ -147,6 +148,7 @@ impl Attributes {
         Attributes {
             id: None,
             class: None,
+            href: None,
             attributes: HashMap::new(),
         }
     }
@@ -173,6 +175,14 @@ impl Attributes {
         self.class.as_ref()
     }
 
+    /// Return the href attribute of the element
+    pub fn get_href(&self) -> Option<String> {
+        self.get("href").and_then(|value| match value {
+            AttributeValues::String(href) => Some(href),
+            _ => None,
+        })
+    }
+
     /// Returns the attributes of the element
     pub fn contains(&self, key: &str) -> bool {
         match key {
diff --git a/src/to_md.rs b/src/to_md.rs
index 59b3be4..0ada48a 100644
--- a/src/to_md.rs
+++ b/src/to_md.rs
@@ -124,10 +124,19 @@ pub fn to_md_with_config(node: Node, config: &ToMdConfig) -> String {
                     tail.push('*');
                 }
                 A => {
-                    if let Some(link) = node.attributes.as_ref().and_then(|attrs| attrs.get("href"))
+                    if let Some(link) = node.attributes.as_ref().and_then(|attrs| attrs.get_href())
                     {
+                        let link = percent_encoding::percent_decode(link.as_bytes())
+                            .decode_utf8()
+                            .map(|s| s.to_string())
+                            .unwrap_or(link);
+
                         res.push('[');
-                        tail.push_str(&format!("]({})", link));
+                        if link.contains(' ') {
+                            tail.push_str(&format!("](<{}>)", link));
+                        } else {
+                            tail.push_str(&format!("]({})", link));
+                        }
                     } else {
                         res.push('[');
                         tail.push(']');
@@ -231,6 +240,18 @@ pub fn to_md_with_config(node: Node, config: &ToMdConfig) -> String {
     res
 }
 
+// https://github.com/izyuumi/html2md-rs/issues/34
+#[test]
+fn issue34() {
+    let input = "<p><a href=\"/my uri\">link</a></p>";
+    let expected = "[link](</my uri>)\n";
+    assert_eq!(safe_from_html_to_md(input.to_string()).unwrap(), expected);
+
+    let input = "<p><a href=\"/myuri\">link</a></p>";
+    let expected = "[link](/myuri)\n";
+    assert_eq!(safe_from_html_to_md(input.to_string()).unwrap(), expected);
+}
+
 /// Converts a string of HTML to a markdown string.
 ///
 /// Panics if the HTML is invalid.