From 1f8c090a5f0797e6b0d593ea43ce9e1e71de51e6 Mon Sep 17 00:00:00 2001
From: Matthew Woodcraft <matthew@woodcraft.me.uk>
Date: Sun, 22 May 2022 13:37:19 +0100
Subject: [PATCH 1/2] When creating the search index, omit words longer than 80
 characters

This avoids creating deeply nested objects in searchindex.json
---
 src/renderer/html_handlebars/search.rs | 20 ++++++++++++++++++--
 1 file changed, 18 insertions(+), 2 deletions(-)

diff --git a/src/renderer/html_handlebars/search.rs b/src/renderer/html_handlebars/search.rs
index 0a59ffe9f9..b5f8c4a2ac 100644
--- a/src/renderer/html_handlebars/search.rs
+++ b/src/renderer/html_handlebars/search.rs
@@ -2,7 +2,7 @@ use std::borrow::Cow;
 use std::collections::{HashMap, HashSet};
 use std::path::Path;
 
-use elasticlunr::Index;
+use elasticlunr::{Index, IndexBuilder};
 use pulldown_cmark::*;
 
 use crate::book::{Book, BookItem};
@@ -13,9 +13,25 @@ use crate::utils;
 
 use serde::Serialize;
 
+const MAX_WORD_LENGTH_TO_INDEX: usize = 80;
+
+/// Tokenizes in the same way as elasticlunr-rs (for English), but also drops long tokens.
+fn tokenize(text: &str) -> Vec<String> {
+    text.split(|c: char| c.is_whitespace() || c == '-')
+        .filter(|s| !s.is_empty())
+        .map(|s| s.trim().to_lowercase())
+        .filter(|s| s.len() <= MAX_WORD_LENGTH_TO_INDEX)
+        .collect()
+}
+
 /// Creates all files required for search.
 pub fn create_files(search_config: &Search, destination: &Path, book: &Book) -> Result<()> {
-    let mut index = Index::new(&["title", "body", "breadcrumbs"]);
+    let mut index = IndexBuilder::new()
+        .add_field_with_tokenizer("title", Box::new(&tokenize))
+        .add_field_with_tokenizer("body", Box::new(&tokenize))
+        .add_field_with_tokenizer("breadcrumbs", Box::new(&tokenize))
+        .build();
+
     let mut doc_urls = Vec::with_capacity(book.sections.len());
 
     for item in book.iter() {

From 000a93dc777fe856bd93ccb29616250d05f99c67 Mon Sep 17 00:00:00 2001
From: Matthew Woodcraft <matthew@woodcraft.me.uk>
Date: Sun, 22 May 2022 13:57:09 +0100
Subject: [PATCH 2/2] Test that long words are omitted from the search index.

Note they do appear in the 'docs' part of searchindex.json (so they will be
visible in search teasers).
---
 tests/dummy_book/src/first/no-headers.md | 4 +++-
 tests/rendered_output.rs                 | 2 +-
 tests/searchindex_fixture.json           | 2 +-
 3 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/tests/dummy_book/src/first/no-headers.md b/tests/dummy_book/src/first/no-headers.md
index 8f9a6d17ef..5d799aa684 100644
--- a/tests/dummy_book/src/first/no-headers.md
+++ b/tests/dummy_book/src/first/no-headers.md
@@ -1,3 +1,5 @@
 Capybara capybara capybara.
 
-Capybara capybara capybara.
\ No newline at end of file
+Capybara capybara capybara.
+
+ThisLongWordIsIncludedSoWeCanCheckThatSufficientlyLongWordsAreOmittedFromTheSearchIndex.
diff --git a/tests/rendered_output.rs b/tests/rendered_output.rs
index 873a622df5..c6267830e3 100644
--- a/tests/rendered_output.rs
+++ b/tests/rendered_output.rs
@@ -772,7 +772,7 @@ mod search {
         );
         assert_eq!(
             docs[&no_headers]["body"],
-            "Capybara capybara capybara. Capybara capybara capybara."
+            "Capybara capybara capybara. Capybara capybara capybara. ThisLongWordIsIncludedSoWeCanCheckThatSufficientlyLongWordsAreOmittedFromTheSearchIndex."
         );
     }
 
diff --git a/tests/searchindex_fixture.json b/tests/searchindex_fixture.json
index 9c349b6b20..3d7062d237 100644
--- a/tests/searchindex_fixture.json
+++ b/tests/searchindex_fixture.json
@@ -229,7 +229,7 @@
           "title": "Unicode stress tests"
         },
         "18": {
-          "body": "Capybara capybara capybara. Capybara capybara capybara.",
+          "body": "Capybara capybara capybara. Capybara capybara capybara. ThisLongWordIsIncludedSoWeCanCheckThatSufficientlyLongWordsAreOmittedFromTheSearchIndex.",
           "breadcrumbs": "First Chapter » No Headers",
           "id": "18",
           "title": "First Chapter"