4b. introduce summarizer ; use summarizer for top news reading

This commit is contained in:
GotPPay
2018-04-01 22:32:20 +02:00
parent 284cdcd7ba
commit cd74990165
8 changed files with 1179 additions and 50 deletions

View File

@@ -1,9 +1,9 @@
const constants = {}; const constants = {};
constants.skillStage = { constants.skillStage = {
IN_DEVELOPMENT : 'development', IN_DEVELOPMENT: 'development',
LIVE : 'live' LIVE: 'live',
} };
constants.amazonResultCodes = { constants.amazonResultCodes = {
OK: 200, OK: 200,
@@ -36,12 +36,17 @@ constants.voiceResponseStrings = {
QUESTION_NOT_FOUND: 'Sorry, I didnt understand', QUESTION_NOT_FOUND: 'Sorry, I didnt understand',
GENERIC_CONTINUE: 'Say something to continue', GENERIC_CONTINUE: 'Say something to continue',
DIDNT_ASK_ANYTHING: 'There was no question to answer to', DIDNT_ASK_ANYTHING: 'There was no question to answer to',
ERROR_SUMMARIZING_CONTENT: 'Sorry, there was problem with summarizing news',
ERROR_FETCHING_CONTENT: 'Failed to get content',
}; };
//Timing is given in [ms] //Timing is given in [ms]
constants.voiceResponseTimings = { constants.voiceResponseTimings = {
PAUSE_BETWEEN_QUESTIONS: 650, PAUSE_BETWEEN_QUESTIONS: 650,
PAUSE_AFTER_WELCOME_MESSAGE: 650, PAUSE_AFTER_WELCOME_MESSAGE: 650,
PAUSE_BETWEEN_TITLES: 500,
PAUSE_BETWEEN_TITLE_AND_CONTENT: 500,
PAUSE_BETWEEN_NEWS: 800,
}; };
constants.stringConstraints = { constants.stringConstraints = {
@@ -65,8 +70,8 @@ constants.stringConstraints = {
constants.answerType = { constants.answerType = {
PREDEFINED: 0, PREDEFINED: 0,
EXTERNAL_SOURCE_WP_JSON : 1, EXTERNAL_SOURCE_WP_TITLES: 1,
EXTERNAL_SOURCE_RSS : 2 EXTERNAL_SOURCE_WP_NEWS: 2,
} };
module.exports = constants; module.exports = constants;

View File

@@ -1,5 +1,7 @@
let request = require ('request'); let request = require ('request');
let Parser = require ('rss-parser'); let Parser = require ('rss-parser');
let summarizer = require ('node-summary');
const constants = require ('../config/constants');
let parser = new Parser (); let parser = new Parser ();
@@ -9,44 +11,88 @@ getDataFromRSSFeed = function (url) {
//feed.items.forEach(item => { //feed.items.forEach(item => {
// console.log(item.title + ':' + item.link) // console.log(item.title + ':' + item.link)
//}); //});
} };
getDataFromWPJSON = function (sourceUrl, page = 1, maxPosts = 10) { getDataFromWPJSON = function (sourceUrl, page = 1, maxPosts = 10) {
return new Promise ((resolve, reject) => { return new Promise ((resolve, reject) => {
var options = { var options = {
method: 'GET', method: 'GET',
url: `${sourceUrl}/wp-json/wp/v2/posts`, url: `${sourceUrl}/wp-json/wp/v2/posts`,
qs:{ qs: {
page:page, page: page,
per_page:maxPosts per_page: maxPosts,
} },
}; };
request (options, (error, response, body)=> { request (options, (error, response, body) => {
if (error) { if (error) {
reject (error); reject (error);
} else { } else {
resolve(JSON.parse (body)); resolve (JSON.parse (body));
} }
}); });
}); });
} };
summarizeText = function (title, text) {
return new Promise ((resolve, reject) => {
summarizer.summarize (title, text, (err, summary) => {
if (err) {
reject (err);
} else {
resolve (summary);
}
});
});
};
module.exports = { module.exports = {
getAnswerFromWP : function (sourceUrl){ getAnswerFromWP: function (sourceUrl) {
//This function will extract needed data from JSON, which we got from getDataFromWPJSON //This function will extract needed data from JSON, which we got from getDataFromWPJSON
//At the moment, it's taking titles and creates answer //At the moment, it's taking titles and creates answer
return new Promise((resolve,reject)=>{ return new Promise ((resolve, reject) => {
getDataFromWPJSON(sourceUrl).then(rawData=>{ getDataFromWPJSON (sourceUrl)
let result=''; .then (rawData => {
rawData.forEach(post=>{ let result = '';
result += post.title.rendered + '<break time="300ms"/> ' rawData.forEach (post => {
result += post.title.rendered + `<break time="${constants.voiceResponseTimings.PAUSE_BETWEEN_TITLES}ms"/> `;
});
resolve (result);
})
.catch (err => {
reject (constants.voiceResponseString.ERROR_FETCHING_CONTENT);
}); });
resolve(result);
}).catch(err=>{
reject('Failed to get answer');
});
}); });
} },
}
getLatestNewsFromWP: function (sourceUrl, postCount = 10, includeTitle = false) {
return new Promise ((resolve, reject) => {
getDataFromWPJSON (sourceUrl, 1, postCount)
.then (rawData => {
let result = '';
let summarizers = [];
rawData.forEach (post => {
summarizers.push (
summarizeText (post.title.rendered, post.content.rendered).then(summary=>{
result += (includeTitle) ? post.title.rendered : '';
result += (includeTitle) ? `<break time="${constants.voiceResponseTimings.PAUSE_BETWEEN_TITLE_AND_CONTENT}ms"/>` : '';
result += summary;
result += `<break time="${constants.voiceResponseTimings.PAUSE_BETWEEN_NEWS}ms"/>`;
}).catch(err=>{
result += constants.voiceResponseString.ERROR_SUMMARIZING_CONTENT;
result += `<break time="${constants.voiceResponseTimings.PAUSE_BETWEEN_NEWS}ms"/>`;
})
);
});
Promise.all(summarizers).then(()=>{
resolve(result);
}).catch(()=>{
reject(result);
});
})
.catch (err => {
reject (constants.voiceResponseString.ERROR_FETCHING_CONTENT);
});
});
},
};

View File

@@ -88,16 +88,19 @@ module.exports = {
case constants.answerType.PREDEFINED: case constants.answerType.PREDEFINED:
answerPromiseProps.resolve(intent.answer); answerPromiseProps.resolve(intent.answer);
break; break;
case constants.answerType.EXTERNAL_SOURCE_WP_JSON: case constants.answerType.EXTERNAL_SOURCE_WP_TITLES:
predefinedSourceHelper.getAnswerFromWP(intent.externalAnswerSource).then(answer=>{ predefinedSourceHelper.getAnswerFromWP(intent.externalAnswerSource).then(answer=>{
answerPromiseProps.resolve(answer); answerPromiseProps.resolve(answer);
}).catch(error=>{ }).catch(error=>{
answerPromiseProps.reject(error); answerPromiseProps.reject(error);
}); });
break; break;
case constants.answerType.EXTERNAL_SOURCE_RSS: case constants.answerType.EXTERNAL_SOURCE_WP_NEWS:
answer = 'Not implemented yet' predefinedSourceHelper.getLatestNewsFromWP(intent.externalAnswerSource).then(answer=>{
answerPromiseProps.resolve(answer); answerPromiseProps.resolve(answer);
}).catch(error=>{
answerPromiseProps.reject(error);
});
break; break;
} }

1104
backend/package-lock.json generated

File diff suppressed because it is too large Load Diff

View File

@@ -10,6 +10,7 @@
"express": "^4.13.0", "express": "^4.13.0",
"isomorphic-fetch": "^2.2.1", "isomorphic-fetch": "^2.2.1",
"mongodb": "^2.2.33", "mongodb": "^2.2.33",
"node-summary": "^1.2.0",
"nodejs-text-summarizer": "^2.0.3", "nodejs-text-summarizer": "^2.0.3",
"nodemailer": "^4.4.1", "nodemailer": "^4.4.1",
"request": "^2.83.0", "request": "^2.83.0",

View File

@@ -79,7 +79,7 @@ class AnswerSource extends Component {
value: '1', value: '1',
}, },
{ {
label: 'RSS feed - latest', label: 'WordPress latest news',
value: '2', value: '2',
}, },
]} ]}

View File

@@ -52,8 +52,8 @@ class IntentDetails extends Component {
</div> </div>
); );
break; break;
case ANSWER_TYPE.EXTERNAL_SOURCE_WP_JSON: case ANSWER_TYPE.EXTERNAL_SOURCE_WP_TITLES:
case ANSWER_TYPE.EXTERNAL_SOURCE_RSS: case ANSWER_TYPE.EXTERNAL_SOURCE_WP_NEWS:
answerBox = ( answerBox = (
<div className="QuestionBox"> <div className="QuestionBox">
<TextField <TextField

View File

@@ -24,8 +24,8 @@ export const CONTACT_SELECTED_INDEX = -3;
export const ANSWER_TYPE = { export const ANSWER_TYPE = {
PREDEFINED: 0, PREDEFINED: 0,
EXTERNAL_SOURCE_WP_JSON : 1, EXTERNAL_SOURCE_WP_TITLES : 1,
EXTERNAL_SOURCE_RSS : 2 EXTERNAL_SOURCE_WP_NEWS : 2
} }
export const RESULT_CODES = { export const RESULT_CODES = {