Use integer doc_refs to shrink the search index

This change reduced the searchindex.js of book_example from 508KB to 317KB.

src/renderer/html_handlebars/search.rs

@@ -18,12 +18,13 @@ use theme::searcher;
 /// 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 doc_urls = Vec::with_capacity(book.sections.len());
 
     for item in book.iter() {
-        render_item(&mut index, &search_config, item)?;
+        render_item(&mut index, &search_config, &mut doc_urls, item)?;
     }
 
-    let index = write_to_js(index, &search_config)?;
+    let index = write_to_js(index, &search_config, doc_urls)?;
     debug!("Writing search index ✓");
 
     if search_config.copy_js {
@@ -38,26 +39,31 @@ pub fn create_files(search_config: &Search, destination: &Path, book: &Book) ->
 }
 
 /// Uses the given arguments to construct a search document, then inserts it to the given index.
-fn add_doc<'a>(
+fn add_doc(
     index: &mut Index,
-    anchor_base: &'a str,
+    doc_urls: &mut Vec<String>, 
+    anchor_base: &str, 
     section_id: &Option<String>,
     items: &[&str],
 ) {
-    let doc_ref: Cow<'a, str> = if let &Some(ref id) = section_id {
-        format!("{}#{}", anchor_base, id).into()
+    let doc_ref = if let &Some(ref id) = section_id {
+        Cow::Owned(format!("{}#{}", anchor_base, id))
     } else {
-        anchor_base.into()
+        Cow::Borrowed(anchor_base.into())
     };
     let doc_ref = utils::collapse_whitespace(doc_ref.trim());
+    let key = doc_urls.len();
+    doc_urls.push(doc_ref.into());
+
     let items = items.iter().map(|&x| utils::collapse_whitespace(x.trim()));
-    index.add_doc(&doc_ref, items);
+    index.add_doc(&key.to_string(), items);
 }
 
 /// Renders markdown into flat unformatted text and adds it to the search index.
 fn render_item(
     index: &mut Index,
     search_config: &Search,
+    doc_urls: &mut Vec<String>,
     item: &BookItem,
 ) -> Result<()> {
     let chapter = match item {
@@ -92,6 +98,7 @@ fn render_item(
                     // Write the data to the index, and clear it for the next section
                     add_doc(
                         index,
+                        doc_urls,
                         &anchor_base,
                         &section_id,
                         &[&heading, &body, &breadcrumbs.join(" » ")],
@@ -144,6 +151,7 @@ fn render_item(
         // Make sure the last section is added to the index
         add_doc(
             index,
+            doc_urls,
             &anchor_base,
             &section_id,
             &[&heading, &body, &breadcrumbs.join(" » ")],
@@ -156,7 +164,7 @@ fn render_item(
 /// Exports the index and search options to a JS script which stores the index in `window.search`.
 /// Using a JS script is a workaround for CORS in `file://` URIs. It also removes the need for
 /// downloading/parsing JSON in JS.
-fn write_to_js(index: Index, search_config: &Search) -> Result<String> {
+fn write_to_js(index: Index, search_config: &Search, doc_urls: Vec<String>) -> Result<String> {
     use std::collections::BTreeMap;
     use self::elasticlunr::config::{SearchBool, SearchOptions, SearchOptionsField};
 
@@ -169,9 +177,11 @@ fn write_to_js(index: Index, search_config: &Search) -> Result<String> {
     #[derive(Serialize)]
     struct SearchindexJson {
         /// The options used for displaying search results
-        resultsoptions: ResultsOptions,
+        results_options: ResultsOptions,
         /// The searchoptions for elasticlunr.js
-        searchoptions: SearchOptions,
+        search_options: SearchOptions,
+        /// Used to lookup a document's URL from an integer document ref.
+        doc_urls: Vec<String>,
         /// The index for elasticlunr.js
         index: elasticlunr::Index,
     }
@@ -185,7 +195,7 @@ fn write_to_js(index: Index, search_config: &Search) -> Result<String> {
     opt.boost = Some(search_config.boost_hierarchy);
     fields.insert("breadcrumbs".into(), opt);
 
-    let searchoptions = SearchOptions {
+    let search_options = SearchOptions {
         bool: if search_config.use_boolean_and {
             SearchBool::And
         } else {
@@ -195,14 +205,15 @@ fn write_to_js(index: Index, search_config: &Search) -> Result<String> {
         fields,
     };
 
-    let resultsoptions = ResultsOptions {
+    let results_options = ResultsOptions {
         limit_results: search_config.limit_results,
         teaser_word_count: search_config.teaser_word_count,
     };
 
     let json_contents = SearchindexJson {
-        resultsoptions,
-        searchoptions,
+        results_options,
+        search_options,
+        doc_urls,
         index,
     };
 

src/theme/searcher/searcher.js

@@ -27,11 +27,12 @@ window.search = window.search || {};
         content = document.getElementById('content'),
 
         searchindex = null,
-        resultsoptions = {
+        doc_urls = [],
+        results_options = {
             teaser_word_count: 30,
             limit_results: 30,
         },
-        searchoptions = {
+        search_options = {
             bool: "AND",
             expand: true,
             fields: {
@@ -139,7 +140,7 @@ window.search = window.search || {};
         teaser_count++;
 
         // The ?URL_MARK_PARAM= parameter belongs inbetween the page and the #heading-anchor
-        var url = result.ref.split("#");
+        var url = doc_urls[result.ref].split("#");
         if (url.length == 1) { // no anchor found
             url.push("");
         }
@@ -196,7 +197,7 @@ window.search = window.search || {};
         }
 
         var window_weight = [];
-        var window_size = Math.min(weighted.length, resultsoptions.teaser_word_count);
+        var window_size = Math.min(weighted.length, results_options.teaser_word_count);
 
         var cur_sum = 0;
         for (var wordindex = 0; wordindex < window_size; wordindex++) {
@@ -247,9 +248,10 @@ window.search = window.search || {};
     }
 
     function init() {
-        resultsoptions = window.search.resultsoptions;
-        searchoptions = window.search.searchoptions;
+        results_options = window.search.results_options;
+        search_options = window.search.search_options;
         searchbar_outer = window.search.searchbar_outer;
+        doc_urls = window.search.doc_urls;
         searchindex = elasticlunr.Index.load(window.search.index);
 
         // Set up events
@@ -441,8 +443,8 @@ window.search = window.search || {};
         if (searchindex == null) { return; }
 
         // Do the actual search
-        var results = searchindex.search(searchterm, searchoptions);
-        var resultcount = Math.min(results.length, resultsoptions.limit_results);
+        var results = searchindex.search(searchterm, search_options);
+        var resultcount = Math.min(results.length, results_options.limit_results);
 
         // Display search metrics
         searchresults_header.innerText = formatSearchMetric(resultcount, searchterm);