1+ const fs = require ( "fs" ) ;
2+ const path = require ( "path" ) ;
3+ const util = require ( "util" ) ;
4+
5+ const readFileAsync = util . promisify ( fs . readFile ) ;
6+ const writeFileAsync = util . promisify ( fs . writeFile ) ;
7+
8+ const lunr = require ( "lunr" ) ;
9+ const htmlparser2 = require ( "htmlparser2" ) ;
10+
11+ function html2text ( html ) {
12+ let pageTitle = "" ;
13+ const sections = [ ] ;
14+ let sectionTitle = "" ;
15+ let sectionContent = "" ;
16+ let hashLink = "" ;
17+
18+ let insideMain = false ;
19+ let insideArticle = false ;
20+ let insideHeader = false ;
21+ let insidePre = false ;
22+ let insideH1 = false ;
23+ let insideH2 = false ;
24+ let insideH3 = false ;
25+ let insideHashLink = false ;
26+ let insideTitle = false ;
27+ let insideSVG = false ;
28+
29+ function sectionEnd ( ) {
30+ if ( sections . length === 0 ) {
31+ return ;
32+ }
33+ sections [ sections . length - 1 ] . content = sectionContent . trim ( ) ;
34+ sectionContent = "" ;
35+ }
36+
37+ function sectionStart ( ) {
38+ sections . push ( {
39+ title : sectionTitle . trim ( ) ,
40+ hash : hashLink
41+ } ) ;
42+ sectionTitle = "" ;
43+ }
44+
45+ const parser = new htmlparser2 . Parser (
46+ {
47+ onopentag ( tagname , attributes ) {
48+ if ( [ "h1" , "h2" , "h3" ] . includes ( tagname ) ) {
49+ sectionEnd ( ) ;
50+ }
51+
52+ if ( tagname === "main" ) {
53+ insideMain = true ;
54+ } else if ( tagname === "article" ) {
55+ insideArticle = true ;
56+ } else if ( tagname === "pre" ) {
57+ insidePre = true ;
58+ } else if ( tagname === "header" ) {
59+ insideHeader = true ;
60+ } else if ( tagname === "h1" ) {
61+ insideH1 = true ;
62+ } else if ( tagname === "h2" ) {
63+ insideH2 = true ;
64+ } else if ( tagname === "h3" ) {
65+ insideH3 = true ;
66+ } else if ( tagname === "a" && attributes [ "class" ] === "hash-link" ) {
67+ insideHashLink = true ;
68+ hashLink = attributes [ "href" ] ;
69+ } else if ( tagname === "title" ) {
70+ insideTitle = true ;
71+ } else if ( tagname === "svg" ) {
72+ insideSVG = true ;
73+ }
74+ } ,
75+ ontext ( text ) {
76+ if ( insideSVG ) {
77+ return ;
78+ }
79+ if ( insideMain && ! insideHashLink ) {
80+ if ( text . length ) {
81+ if ( insideH1 || insideH2 || insideH3 ) {
82+ sectionTitle += text ;
83+ } else {
84+ sectionContent += text ;
85+ }
86+ }
87+ } else if ( insideTitle ) {
88+ pageTitle += text ;
89+ }
90+ } ,
91+ onclosetag ( tagname ) {
92+ if ( tagname === "main" ) {
93+ insideMain = false ;
94+ } else if ( tagname === "article" ) {
95+ insideArticle = false ;
96+ } else if ( tagname === "pre" ) {
97+ insidePre = false ;
98+ } else if ( tagname === "header" ) {
99+ insideHeader = false ;
100+ } else if ( tagname === "h1" ) {
101+ insideH1 = false ;
102+ } else if ( tagname === "h2" ) {
103+ insideH2 = false ;
104+ } else if ( tagname === "h3" ) {
105+ insideH3 = false ;
106+ } else if ( tagname === "a" && insideHashLink ) {
107+ insideHashLink = false ;
108+ } else if ( tagname === "title" ) {
109+ insideTitle = false ;
110+ } else if ( tagname === "svg" ) {
111+ insideSVG = false ;
112+ }
113+
114+ if ( insideMain && [ "h1" , "h2" , "h3" ] . includes ( tagname ) ) {
115+ if ( insideArticle && insideHeader ) {
116+ pageTitle = sectionTitle . trim ( ) ;
117+ }
118+ sectionStart ( ) ;
119+ } else if ( ! insidePre ) {
120+ sectionTitle += " " ;
121+ sectionContent += " " ;
122+ }
123+ }
124+ } ,
125+ { decodeEntities : true , lowerCaseTags : true }
126+ ) ;
127+ parser . write ( html ) ;
128+ parser . end ( ) ;
129+
130+ sectionEnd ( ) ;
131+
132+ return { pageTitle, sections } ;
133+ }
134+
135+ module . exports = function ( context , options ) {
136+ return {
137+ name : 'docusaurus-plugin' ,
138+ getThemePath ( ) {
139+ return path . resolve ( __dirname , './theme' ) ;
140+ } ,
141+ async postBuild ( { routesPaths = [ ] , outDir, baseUrl } ) {
142+ const data = routesPaths . map ( route => {
143+ let file = route ;
144+ if ( file . startsWith ( baseUrl ) ) {
145+ file = file . replace ( baseUrl , "" ) ;
146+ }
147+ if ( ! file . endsWith ( ".html" ) ) {
148+ if ( ! file . endsWith ( "/" ) ) {
149+ file += "/" ;
150+ }
151+ file += "index.html" ;
152+ }
153+ file = path . join ( outDir , file ) ;
154+
155+ return {
156+ file,
157+ route
158+ } ;
159+ } ) ;
160+
161+ const documents = ( await Promise . all ( data . map ( async ( { file, route } ) => {
162+ const html = await readFileAsync ( file , { encoding : "utf8" } ) ;
163+
164+ const { pageTitle, sections } = html2text ( html ) ;
165+
166+ return sections . map ( section => ( {
167+ pageTitle,
168+ pageRoute : route ,
169+ sectionRoute : route + section . hash ,
170+ sectionTitle : section . title ,
171+ sectionContent : section . content
172+ } ) ) ;
173+ } ) ) ) . reduce ( ( acc , val ) => acc . concat ( val ) , [ ] ) ; // .flat()
174+
175+ const index = lunr ( function ( ) {
176+ this . ref ( "route" ) ;
177+ this . field ( "title" ) ;
178+ this . field ( "content" ) ;
179+ documents . forEach ( function ( { sectionRoute, sectionTitle, sectionContent } ) {
180+ this . add ( {
181+ route : sectionRoute ,
182+ title : sectionTitle ,
183+ content : sectionContent
184+ } ) ;
185+ } , this )
186+ } ) ;
187+
188+ await writeFileAsync (
189+ path . join ( outDir , "search-index.json" ) ,
190+ JSON . stringify ( {
191+ documents : documents . map ( ( { pageTitle, sectionTitle, sectionRoute } ) => ( { pageTitle, sectionTitle, sectionRoute } ) ) ,
192+ index
193+ } ) ,
194+ { encoding : "utf8" }
195+ ) ;
196+ } ,
197+ } ;
198+ } ;
0 commit comments