BTreeSet is really slow; don't use it

silverpill · Apr 15, 2017 · 0c9d8af · 0c9d8af
1 parent bf576b3
commit 0c9d8af
Show file tree

Hide file tree

Showing 3 changed files with 40 additions and 50 deletions.
diff --git a/src/html.rs b/src/html.rs
@@ -1,7 +1,5 @@
 use std;
 use std::io::Write;
-use std::iter::FromIterator;
-use std::collections::BTreeMap;
 
 use nodes::{TableAlignment, NodeValue, ListType, AstNode};
 use parser::ComrakOptions;
@@ -97,27 +95,14 @@ impl<'o> HtmlFormatter<'o> {
     }
 
     fn escape(&mut self, buffer: &str) {
-        lazy_static! {
-            static ref ESCAPE_TABLE: BTreeMap<u8, &'static str> = BTreeMap::from_iter(
-                vec![(b'"', "&quot;"),
-                     (b'&', "&amp;"),
-                     // Secure mode only:
-                     // ('\'', "&#39;"),
-                     // ('/', "&#47;"),
-                     (b'<', "&lt;"),
-                     (b'>', "&gt;"),
-            ]);
-        }
-
         for c in buffer.as_bytes() {
-            match ESCAPE_TABLE.get(c) {
-                Some(s) => {
-                    self.write_all(s.as_bytes()).unwrap();
-                }
-                None => {
-                    self.write_all(&[*c]).unwrap();
-                }
-            };
+            match *c {
+                b'"' => self.write_all(b"&quot;").unwrap(),
+                b'&' => self.write_all(b"&amp;").unwrap(),
+                b'<' => self.write_all(b"&lt;").unwrap(),
+                b'>' => self.write_all(b"&gt;").unwrap(),
+                _ => self.v.push(*c),
+            }
         }
     }
 

diff --git a/src/parser/autolink.rs b/src/parser/autolink.rs
@@ -1,6 +1,4 @@
 use unicode_categories::UnicodeCategories;
-use std::iter::FromIterator;
-use std::collections::BTreeSet;
 use typed_arena::Arena;
 use nodes::{NodeValue, NodeLink, AstNode};
 use ctype::{isspace, isalpha, isalnum};
@@ -59,12 +57,17 @@ fn www_match<'a>(arena: &'a Arena<AstNode<'a>>,
                  i: usize)
                  -> Option<(&'a AstNode<'a>, usize, usize)> {
     lazy_static! {
-        static ref WWW_DELIMS: BTreeSet<u8> = BTreeSet::from_iter(
-            vec![b'*', b'_', b'~', b'(', b'[']);
+        static ref WWW_DELIMS: [bool; 256] = {
+            let mut sc = [false; 256];
+            for c in &[b'*', b'_', b'~', b'(', b'['] {
+                sc[*c as usize] = true;
+            }
+            sc
+        };
     }
 
     if i > 0 && !isspace(&contents.as_bytes()[i - 1]) &&
-       !WWW_DELIMS.contains(&contents.as_bytes()[i - 1]) {
+       !WWW_DELIMS[contents.as_bytes()[i - 1] as usize] {
         return None;
     }
 
@@ -130,9 +133,13 @@ fn is_valid_hostchar(ch: char) -> bool {
 
 fn autolink_delim(data: &str, mut link_end: usize) -> usize {
     lazy_static! {
-        static ref LINK_END_ASSORTMENT: BTreeSet<u8> = BTreeSet::from_iter(
-            vec![b'?', b'!', b'.', b',', b':', b'*',
-            b'_', b'~', b'\'', b'"']);
+        static ref LINK_END_ASSORTMENT: [bool; 256] = {
+            let mut sc = [false; 256];
+            for c in &[b'?', b'!', b'.', b',', b':', b'*', b'_', b'~', b'\'', b'"'] {
+                sc[*c as usize] = true;
+            }
+            sc
+        };
     }
 
     for i in 0..link_end {
@@ -151,7 +158,7 @@ fn autolink_delim(data: &str, mut link_end: usize) -> usize {
             None
         };
 
-        if LINK_END_ASSORTMENT.contains(&cclose) {
+        if LINK_END_ASSORTMENT[cclose as usize] {
             link_end -= 1;
         } else if cclose == b';' {
             let mut new_end = link_end - 2;
@@ -242,8 +249,13 @@ fn email_match<'a>(arena: &'a Arena<AstNode<'a>>,
                    i: usize)
                    -> Option<(&'a AstNode<'a>, usize, usize)> {
     lazy_static! {
-        static ref EMAIL_OK_SET: BTreeSet<u8> = BTreeSet::from_iter(
-            vec![b'.', b'+', b'-', b'_']);
+        static ref EMAIL_OK_SET: [bool; 256] = {
+            let mut sc = [false; 256];
+            for c in &[b'.', b'+', b'-', b'_'] {
+                sc[*c as usize] = true;
+            }
+            sc
+        };
     }
 
     let size = contents.len();
@@ -254,7 +266,7 @@ fn email_match<'a>(arena: &'a Arena<AstNode<'a>>,
     while rewind < i {
         let c = contents.as_bytes()[i - rewind - 1];
 
-        if isalnum(&c) || EMAIL_OK_SET.contains(&c) {
+        if isalnum(&c) || EMAIL_OK_SET[c as usize] {
             rewind += 1;
             continue;
         }

diff --git a/src/parser/inlines.rs b/src/parser/inlines.rs
@@ -10,7 +10,7 @@ use scanners;
 use ctype::{isspace, ispunct};
 use entity;
 use strings;
-use std::collections::{BTreeSet, HashMap};
+use std::collections::HashMap;
 
 pub struct Subject<'a, 'r, 'o> {
     pub arena: &'a Arena<AstNode<'a>>,
@@ -267,24 +267,17 @@ impl<'a, 'r, 'o> Subject<'a, 'r, 'o> {
 
     pub fn find_special_char(&self) -> usize {
         lazy_static! {
-            static ref SPECIAL_CHARS: BTreeSet<u8> =
-                [b'\n',
-                b'\r',
-                b'_',
-                b'*',
-                b'"',
-                b'`',
-                b'\\',
-                b'&',
-                b'<',
-                b'[',
-                b']',
-                b'!',
-                ].iter().cloned().collect();
+            static ref SPECIAL_CHARS: [bool; 256] = {
+                let mut sc = [false; 256];
+                for c in &[b'\n', b'\r', b'_', b'*', b'"', b'`', b'\\', b'&', b'<', b'[', b']', b'!'] {
+                    sc[*c as usize] = true;
+                }
+                sc
+            };
         }
 
         for n in self.pos..self.input.len() {
-            if SPECIAL_CHARS.contains(&self.input.as_bytes()[n]) {
+            if SPECIAL_CHARS[self.input.as_bytes()[n] as usize] {
                 return n;
             }
             if self.options.ext_strikethrough && self.input.as_bytes()[n] == b'~' {