Replace the regex highlighting by a simple algorithm

Cargo.lock

@@ -6,15 +6,6 @@ version = "0.2.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ccc9a9dd069569f212bc4330af9f17c4afb5e8ce185e83dbb14f1349dda18b10"
 
-[[package]]
-name = "aho-corasick"
-version = "0.7.13"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "043164d8ba5c4c3035fec9bbee8647c0261d788f3474306f93bb65901cae0e86"
-dependencies = [
- "memchr",
-]
-
 [[package]]
 name = "anyhow"
 version = "1.0.31"
@@ -997,7 +988,6 @@ dependencies = [
  "once_cell",
  "oxidized-mtbl",
  "rayon",
- "regex",
  "roaring",
  "serde",
  "slice-group-by",
@@ -1601,10 +1591,7 @@ version = "1.3.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9c3780fcf44b193bc4d09f36d2a3c87b251da4a046c87795a0d35f4f927ad8e6"
 dependencies = [
- "aho-corasick",
- "memchr",
  "regex-syntax",
- "thread_local 1.0.1",
 ]
 
 [[package]]
@@ -1822,7 +1809,7 @@ dependencies = [
  "chrono",
  "log 0.4.8",
  "termcolor",
- "thread_local 0.3.4",
+ "thread_local",
 ]
 
 [[package]]
@@ -1937,15 +1924,6 @@ dependencies = [
  "unreachable",
 ]
 
-[[package]]
-name = "thread_local"
-version = "1.0.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d40c6d1b69745a6ec6fb1ca717914848da4b44ae29d9b3080cbee91d72a69b14"
-dependencies = [
- "lazy_static 1.4.0",
-]
-
 [[package]]
 name = "time"
 version = "0.1.43"

Cargo.toml

@@ -28,9 +28,6 @@ smallvec = "1.4.0"
 structopt = { version = "0.3.14", default-features = false }
 tempfile = "3.1.0"
 
-# to highlight the documents
-regex = "1.3.9"
-
 # logging
 log = "0.4.8"
 stderrlog = "0.4.3"

src/bin/serve.rs

@@ -1,4 +1,5 @@
 use std::borrow::Cow;
+use std::collections::HashSet;
 use std::fs::File;
 use std::net::SocketAddr;
 use std::path::PathBuf;
@@ -7,8 +8,8 @@ use std::time::Instant;
 
 use askama_warp::Template;
 use heed::EnvOpenOptions;
-use regex::Regex;
 use serde::Deserialize;
+use slice_group_by::StrGroupBy;
 use structopt::StructOpt;
 use warp::{Filter, http::Response};
 
@@ -44,6 +45,18 @@ struct Opt {
     http_listen_addr: String,
 }
 
+fn highlight_string(string: &str, words: &HashSet<String>) -> String {
+    let mut output = String::new();
+    for token in string.linear_group_by_key(|c| c.is_alphanumeric()) {
+        let lowercase_token = token.to_lowercase();
+        let to_highlight = words.contains(&lowercase_token);
+        if to_highlight { output.push_str("<mark>") }
+        output.push_str(token);
+        if to_highlight { output.push_str("</mark>") }
+    }
+    output
+}
+
 #[derive(Template)]
 #[template(path = "index.html")]
 struct IndexTemplate {
@@ -173,15 +186,6 @@ async fn main() -> anyhow::Result<()> {
                 // We write the headers
                 body.extend_from_slice(headers);
 
-                let mut regex = format!(r"(?i)\b(");
-                let number_of_words = words.len();
-                words.into_iter().enumerate().for_each(|(i, w)| {
-                    regex.push_str(&w);
-                    if i != number_of_words - 1 { regex.push('|') }
-                });
-                regex.push_str(r")\b");
-                let re = Regex::new(&regex).unwrap();
-
                 for id in documents_ids {
                     let content = index.documents.get(&rtxn, &BEU32::new(id)).unwrap();
                     let content = content.expect(&format!("could not find document {}", id));
@@ -190,7 +194,7 @@ async fn main() -> anyhow::Result<()> {
                     let content = if disable_highlighting {
                         Cow::from(content)
                     } else {
-                        re.replace_all(content, "<mark>$1</mark>")
+                        Cow::from(highlight_string(content, &words))
                     };
 
                     body.extend_from_slice(content.as_bytes());