clean text before summarization

This commit is contained in:
GotPPay
2018-04-02 04:21:27 +02:00
parent e04dbe3ec9
commit 6643aafb54

View File

@@ -65,23 +65,24 @@ module.exports = {
getDataFromWPJSON (sourceUrl, 1, postCount)
.then (rawData => {
let result = '';
let htmlToTextOptions = {
wordwrap: false,
ignoreHref: true,
ignoreImage: true,
}
try {
rawData.forEach (post => {
let textData = htmlToText.fromString(post.content.rendered,htmlToTextOptions);
result += includeTitle ? post.title.rendered : '';
result += includeTitle
? `<break time="${constants.voiceResponseTimings.PAUSE_BETWEEN_TITLE_AND_CONTENT}ms"/>`
: '';
result += summarizer (post.content.rendered);
result += summarizer (textData);
result += `<break time="${constants.voiceResponseTimings.PAUSE_BETWEEN_NEWS}ms"/>`;
});
var textAnswer = htmlToText.fromString (result, {
wordwrap: false,
ignoreHref: true,
ignoreImage: true,
});
resolve (textAnswer);
resolve (result);
} catch (err) {
reject (constants.voiceResponseString.ERROR_SUMMARIZING_CONTENT);
}