Optimize join()

letmutex · letmutex · Jun 24, 2024 · Jun 24, 2024 · Jun 24, 2024 · Jun 24, 2024
commit 7e701fe7a6f129b1648a3230a20eef45b4e2eb66
diff --git a/benches/README.md b/benches/README.md
@@ -15,11 +15,14 @@ Memory: 15.9 GB
 # Result
 
 ```
-convert()               time:   [89.826 ms 90.520 ms 91.225 ms]
-                        change: [-0.0736% +1.0704% +2.2298%] (p = 0.06 > 0.05)
+convert()               time:   [88.528 ms 89.545 ms 90.670 ms]
+                        change: [-0.2785% +1.0525% +2.5466%] (p = 0.15 > 0.05)
                         No change in performance detected.
+Found 5 outliers among 100 measurements (5.00%)
+  4 (4.00%) high mild
+  1 (1.00%) high severe
 ```
 
-*Updated at Mon, 24 Jun 2024 04:03:55 GMT*
+*Updated at Mon, 24 Jun 2024 07:51:50 GMT*
 
 *Generated by [bench.ts](bench.ts)*
diff --git a/src/element_handler/anchor.rs b/src/element_handler/anchor.rs
@@ -5,7 +5,7 @@ use markup5ever_rcdom::Node;
 
 use crate::{
     options::{LinkReferenceStyle, LinkStyle, Options},
-    text_util::{concat_strings, StripWhitespace, TrimAsciiWhitespace},
+    text_util::{concat_strings, JoinOnStringIterator, StripWhitespace, TrimAsciiWhitespace},
 };
 
 use super::ElementHandler;
@@ -58,7 +58,6 @@ impl ElementHandler for AnchorElementHandler {
             text.lines()
                 .map(|line| line.trim_ascii_whitespace().replace("\"", "\\\""))
                 .filter(|line| !line.is_empty())
-                .collect::<Vec<String>>()
                 .join("\n")
         };
 

diff --git a/src/element_handler/blockquote.rs b/src/element_handler/blockquote.rs
@@ -1,5 +1,5 @@
 use crate::{
-    text_util::{concat_strings, TrimAsciiWhitespace},
+    text_util::{concat_strings, JoinOnStringIterator, TrimAsciiWhitespace},
     Element,
 };
 
@@ -9,7 +9,6 @@ pub(super) fn blockquote_handler(element: Element) -> Option<String> {
         .trim_end_ascii_whitespace()
         .lines()
         .map(|line| concat_strings!("> ", line))
-        .collect::<Vec<String>>()
         .join("\n");
     Some(concat_strings!("\n\n", content, "\n\n"))
 }
diff --git a/src/element_handler/code.rs b/src/element_handler/code.rs
@@ -1,7 +1,7 @@
 use crate::{
     node_util::get_parent_node_tag_name,
     options::{CodeBlockFence, CodeBlockStyle},
-    text_util::{concat_strings, TrimAsciiWhitespace},
+    text_util::{concat_strings, JoinOnStringIterator, TrimAsciiWhitespace},
     Element,
 };
 
@@ -32,7 +32,7 @@ fn handle_code_block(element: Element) -> Option<String> {
                     .to_string()
                     .split(" ")
                     .find(|cls| cls.starts_with("language-"))
-                    .map(|lang| lang.split("-").skip(1).collect::<Vec<&str>>().join("-"))
+                    .map(|lang| lang.split("-").skip(1).join("-"))
             })
             .unwrap_or(None);
         let mut result = String::from(&fence);
@@ -48,7 +48,6 @@ fn handle_code_block(element: Element) -> Option<String> {
         let code = content
             .lines()
             .map(|line| concat_strings!("    ", line))
-            .collect::<Vec<String>>()
             .join("\n");
         Some(code)
     }

diff --git a/src/element_handler/img.rs b/src/element_handler/img.rs
@@ -1,5 +1,5 @@
 use crate::{
-    text_util::{concat_strings, TrimAsciiWhitespace},
+    text_util::{concat_strings, JoinOnStringIterator, TrimAsciiWhitespace},
     Element,
 };
 
@@ -28,7 +28,6 @@ pub(super) fn img_handler(element: Element) -> Option<String> {
         text.lines()
             .map(|line| line.trim_ascii_whitespace().replace("\"", "\\\""))
             .filter(|line| !line.is_empty())
-            .collect::<Vec<String>>()
             .join("\n")
     };
 

diff --git a/src/text_util.rs b/src/text_util.rs
@@ -80,6 +80,29 @@ where
     }
 }
 
+pub(crate) trait JoinOnStringIterator {
+    fn join<S: AsRef<str>>(&mut self, separator: S) -> String;
+}
+
+impl<T, S> JoinOnStringIterator for T
+where
+    S: AsRef<str>,
+    T: Iterator<Item = S>,
+{
+    fn join<SE: AsRef<str>>(&mut self, separator: SE) -> String {
+        let Some(first) = self.next() else {
+            return String::new();
+        };
+        let separator = separator.as_ref();
+        let mut result = String::from(first.as_ref());
+        for next in self {
+            result.push_str(separator);
+            result.push_str(next.as_ref());
+        }
+        result
+    }
+}
+
 pub(crate) fn compress_whitespace(input: &str) -> String {
     let mut result = String::new();
     if input.len() == 0 {