build(aio): add terms from heading to the search index

This commit is contained in:
Peter Bacon Darwin 2017-07-04 17:59:08 +01:00 committed by Pete Bacon Darwin
parent 3a0886dc12
commit e8bbf86e66
3 changed files with 106 additions and 27 deletions

View File

@ -18,7 +18,8 @@ self.onmessage = handleMessage;
function createIndex(addFn) {
return lunr(/** @this */function() {
this.ref('path');
this.field('titleWords', {boost: 50});
this.field('titleWords', {boost: 100});
this.field('headingWords', {boost: 50});
this.field('members', {boost: 40});
this.field('keywords', {boost: 20});
addFn(this);

View File

@ -50,19 +50,14 @@ module.exports = function generateKeywordsProcessor(log, readFilesProcessor) {
var ignoreWordsMap = convertToMap(wordsToIgnore);
// If the title contains a name starting with ng, e.g. "ngController", then add the module
// name
// without the ng to the title text, e.g. "controller".
function extractTitleWords(title) {
var match = /ng([A-Z]\w*)/.exec(title);
if (match) {
title = title + ' ' + match[1].toLowerCase();
}
return title;
// If the heading contains a name starting with ng, e.g. "ngController", then add the
// name without the ng to the text, e.g. "controller".
function preprocessText(text) {
return text.replace(/(^|\s)([nN]g([A-Z]\w*))/g, '$1$2 $3');
}
function extractWords(text, words, keywordMap) {
var tokens = text.toLowerCase().split(/[.\s,`'"#]+/mg);
var tokens = preprocessText(text).toLowerCase().split(/[.\s,`'"#]+/mg);
tokens.forEach(function(token) {
var match = token.match(KEYWORD_REGEX);
if (match) {
@ -82,13 +77,15 @@ module.exports = function generateKeywordsProcessor(log, readFilesProcessor) {
// Ignore internals and private exports (indicated by the ɵ prefix)
.filter(function(doc) { return !doc.internal && !doc.privateExport; });
filteredDocs.forEach(function(doc) {
filteredDocs.forEach(function(doc) {
var words = [];
var keywordMap = Object.assign({}, ignoreWordsMap);
var members = [];
var membersMap = {};
var membersMap = Object.assign({}, ignoreWordsMap);
const headingWords = [];
const headingWordMap = Object.assign({}, ignoreWordsMap);
// Search each top level property of the document for search terms
Object.keys(doc).forEach(function(key) {
@ -98,25 +95,43 @@ module.exports = function generateKeywordsProcessor(log, readFilesProcessor) {
extractWords(value, words, keywordMap);
}
// Special case properties that contain content relating to "members"
// of a doc that represents, say, a class or interface
if (key === 'methods' || key === 'properties' || key === 'events') {
value.forEach(function(member) { extractWords(member.name, members, membersMap); });
}
});
doc.searchTitle = doc.searchTitle || doc.title || doc.vFile && doc.vFile.title || doc.name;
// Extract all the keywords from the headings
if (doc.vFile && doc.vFile.headings) {
Object.keys(doc.vFile.headings).forEach(function(headingTag) {
doc.vFile.headings[headingTag].forEach(function(headingText) {
extractWords(headingText, headingWords, headingWordMap);
});
});
}
// Extract the title to use in searches
doc.searchTitle = doc.searchTitle || doc.title || doc.vFile && doc.vFile.title || doc.name || '';
// Attach all this search data to the document
doc.searchTerms = {
titleWords: extractTitleWords(doc.searchTitle),
titleWords: preprocessText(doc.searchTitle),
headingWords: headingWords.sort().join(' '),
keywords: words.sort().join(' '),
members: members.sort().join(' ')
};
});
var searchData =
filteredDocs.filter(function(page) { return page.searchTerms; }).map(function(page) {
return Object.assign(
{path: page.path, title: page.searchTitle, type: page.docType}, page.searchTerms);
// Now process all the search data and collect it up to be used in creating a new document
var searchData = filteredDocs.map(function(page) {
// Copy the properties from the searchTerms object onto the search data object
return Object.assign({
path: page.path,
title: page.searchTitle,
type: page.docType
}, page.searchTerms);
});
docs.push({

View File

@ -42,9 +42,9 @@ describe('generateKeywords processor', () => {
it('should compute `doc.searchTitle` from the doc properties if not already provided', () => {
const processor = processorFactory(mockLogger, mockReadFilesProcessor);
const docs = [
{ docType: 'class', name: 'A', searchTitle: 'searchTitle A', title: 'title A', vFile: { title: 'vFile A'} },
{ docType: 'class', name: 'B', title: 'title B', vFile: { title: 'vFile B'} },
{ docType: 'class', name: 'C', vFile: { title: 'vFile C'} },
{ docType: 'class', name: 'A', searchTitle: 'searchTitle A', title: 'title A', vFile: { headings: { h1: ['vFile A'] } } },
{ docType: 'class', name: 'B', title: 'title B', vFile: { headings: { h1: ['vFile B'] } } },
{ docType: 'class', name: 'C', vFile: { title: 'vFile C', headings: { h1: ['vFile C'] } } },
{ docType: 'class', name: 'D' },
];
processor.$process(docs);
@ -62,19 +62,82 @@ describe('generateKeywords processor', () => {
{ docType: 'class', name: 'PublicExport', searchTitle: 'class PublicExport' },
];
processor.$process(docs);
expect(docs[docs.length - 1].data).toEqual([
const keywordsDoc = docs[docs.length - 1];
expect(keywordsDoc.data).toEqual([
jasmine.objectContaining({ title: 'class PublicExport', type: 'class'})
]);
});
it('should add title words to the search terms', () => {
const processor = processorFactory(mockLogger, mockReadFilesProcessor);
const docs = [
{
docType: 'class',
name: 'PublicExport',
searchTitle: 'class PublicExport',
vFile: { headings: { h2: ['heading A', 'heading B'] } }
},
];
processor.$process(docs);
const keywordsDoc = docs[docs.length - 1];
expect(keywordsDoc.data[0].titleWords).toEqual('class PublicExport');
});
it('should add heading words to the search terms', () => {
const processor = processorFactory(mockLogger, mockReadFilesProcessor);
const docs = [
{
docType: 'class',
name: 'PublicExport',
searchTitle: 'class PublicExport',
vFile: { headings: { h2: ['Important heading', 'Secondary heading'] } }
},
];
processor.$process(docs);
const keywordsDoc = docs[docs.length - 1];
expect(keywordsDoc.data[0].headingWords).toEqual('heading important secondary');
});
it('should process terms prefixed with "ng" to include the term stripped of "ng"', () => {
const processor = processorFactory(mockLogger, mockReadFilesProcessor);
const docs = [
{
docType: 'class',
name: 'PublicExport',
searchTitle: 'ngController',
vFile: { headings: { h2: ['ngModel'] } },
content: 'Some content with ngClass in it.'
},
];
processor.$process(docs);
const keywordsDoc = docs[docs.length - 1];
expect(keywordsDoc.data[0].titleWords).toEqual('ngController Controller');
expect(keywordsDoc.data[0].headingWords).toEqual('model ngmodel');
expect(keywordsDoc.data[0].keywords).toContain('class');
expect(keywordsDoc.data[0].keywords).toContain('ngclass');
});
it('should generate renderedContent property', () => {
const processor = processorFactory(mockLogger, mockReadFilesProcessor);
const docs = [
{ docType: 'class', name: 'SomeClass', description: 'The is the documentation for the SomeClass API.' },
{
docType: 'class',
name: 'SomeClass',
description: 'The is the documentation for the SomeClass API.',
vFile: { headings: { h1: ['SomeClass'], h2: ['Some heading'] } }
},
];
processor.$process(docs);
expect(docs[docs.length - 1].renderedContent).toEqual(
'[{"title":"SomeClass","type":"class","titleWords":"SomeClass","keywords":"api class documentation for is someclass the","members":""}]'
const keywordsDoc = docs[docs.length - 1];
expect(JSON.parse(keywordsDoc.renderedContent)).toEqual(
[{
'title':'SomeClass',
'type':'class',
'titleWords':'SomeClass',
'headingWords':'heading some someclass',
'keywords':'api class documentation for is someclass the',
'members':''
}]
);
});
});