1 |
|
2 |
|
3 |
|
4 |
|
5 | var fetch = require('node-fetch');
|
6 | var moment = require('moment');
|
7 |
|
8 |
|
9 | const _HP = require("./classHotels.js");
|
10 |
|
11 | const keys=[];
|
12 | const nodes = ["NAME","STARS","ADRESSE"];
|
13 | var preventdouble =[]
|
14 | var sameItems =[]
|
15 | var notsameitems =[]
|
16 | var arrayofelements =[]
|
17 |
|
18 |
|
19 | function startBooking(array,city) {
|
20 | return new Promise(resolve=>{
|
21 | let i =0;
|
22 | array.forEach(t=>{
|
23 | t.source = 'Booking.com'
|
24 | t.id = i
|
25 | i++;
|
26 | })
|
27 |
|
28 | arrayofelements =array.slice();
|
29 |
|
30 | console.log('Done Booking.com fetching found: ' + arrayofelements.length)
|
31 | fetchDBData(city).then((r)=>{
|
32 | let obj={
|
33 | same:sameItems,
|
34 | notsame:notsameitems
|
35 | }
|
36 |
|
37 | resolve(obj)
|
38 | })
|
39 | })
|
40 | }
|
41 |
|
42 | function init() {
|
43 | return new Promise(resolve=>{
|
44 |
|
45 | if(keys.length>0)
|
46 | {
|
47 | foundIndex=[];
|
48 | sameItems = [];
|
49 |
|
50 | for(let skip=0;skip<arrayofelements.length;skip++){
|
51 | keys.forEach(treeitem=>{
|
52 |
|
53 | if(treeitem.index == 0)
|
54 | compareHotel(treeitem,skip,arrayofelements)
|
55 |
|
56 | })
|
57 | }
|
58 | }
|
59 | else
|
60 | {
|
61 | console.error("no Array")
|
62 | }
|
63 |
|
64 | console.log("");
|
65 | console.log("Total same found: " +sameItems.length);
|
66 | findNotSameHotels().then(res=>{
|
67 | resolve(res)
|
68 | });
|
69 | })
|
70 | }
|
71 |
|
72 | function generateTree(obj) {
|
73 | return new Promise(resolve=>{
|
74 | let i = 0, key;
|
75 | for (key in obj)
|
76 | {
|
77 | nodes.forEach(matchNode=>{
|
78 | if(key.toUpperCase().indexOf(matchNode)>=0){
|
79 |
|
80 | firstnode = {
|
81 | index:i,
|
82 | match: `${matchNode}`,
|
83 | "key":key
|
84 | }
|
85 | keys.push(firstnode)
|
86 | }
|
87 | })
|
88 | i++;
|
89 | }
|
90 | resolve(keys)
|
91 | })
|
92 |
|
93 | }
|
94 |
|
95 | function compareHotel(item,skipitem,array) {
|
96 |
|
97 | let i = skipitem;
|
98 | const word = parser(array[i][item.key]);
|
99 |
|
100 | item.points = 0;
|
101 | let currentitem = 0;
|
102 |
|
103 | array.forEach(inarray =>{
|
104 |
|
105 |
|
106 | if(skipitem==currentitem || inarray.source == arrayofelements[i].source|| inarray.id == arrayofelements[i].id ){
|
107 | currentitem++;
|
108 | return;
|
109 | }
|
110 |
|
111 | if(preventdouble.indexOf(`${inarray.source}${inarray.id}${arrayofelements[i].source}${arrayofelements[i].id}`)>=0){
|
112 | currentitem++;
|
113 | return;
|
114 | }
|
115 |
|
116 |
|
117 |
|
118 | const geo = 1;
|
119 | const email = 1;
|
120 | const street = 1;
|
121 |
|
122 |
|
123 |
|
124 | if ( parser(inarray[item.key]) == word || word.indexOf(parser(inarray[item.key]).substr(2,inarray[item.key].length-2))>=0){
|
125 | |
126 |
|
127 |
|
128 |
|
129 |
|
130 |
|
131 |
|
132 |
|
133 | let equals = {
|
134 | 0:inarray,
|
135 | 1:arrayofelements[i],
|
136 | id:`${inarray.id};${arrayofelements[i].id}`
|
137 | };
|
138 | inarray.founddb = true;
|
139 | arrayofelements[i].founddb = true
|
140 |
|
141 | preventdouble.push(`${arrayofelements[i].source}${arrayofelements[i].id}${inarray.source}${inarray.id}`)
|
142 | sameItems.push(equals)
|
143 |
|
144 | currentitem++;
|
145 | return
|
146 |
|
147 | }
|
148 |
|
149 |
|
150 |
|
151 | if(exclude(inarray[item.key]) == null || exclude(word) == null )
|
152 | { currentitem++; return; }
|
153 |
|
154 |
|
155 | const compare0 = word.split(" ");
|
156 | const compare1 = parser(inarray[item.key]).split(" ");
|
157 |
|
158 |
|
159 | try{
|
160 |
|
161 | if(compare0[0].length <= 3 || compare0[1].length <=3 || compare1[0].length<=3|| compare1[1].length<=3)
|
162 | {
|
163 | currentitem++;
|
164 | return;
|
165 | }
|
166 | }
|
167 | catch (TypeError)
|
168 | {
|
169 | currentitem++;
|
170 | return;
|
171 | }
|
172 |
|
173 |
|
174 | let machedItems =0;
|
175 | let passedmachedItems = false;
|
176 | const notmacheditems = {
|
177 | "compare0": [],
|
178 | "compare1": [],
|
179 | }
|
180 | compare0.forEach(item0=>{
|
181 | compare1.forEach(item1=>{
|
182 | if(item0.length >= item1.length)
|
183 | {
|
184 |
|
185 | if(item0.indexOf(item1)>=0){
|
186 | machedItems++;
|
187 | return;
|
188 | }
|
189 | else {
|
190 | if(notmacheditems.compare1.indexOf(item1)==-1 && notmacheditems.compare0.indexOf(item0)==-1)
|
191 | {
|
192 | notmacheditems.compare0.push(item0)
|
193 | notmacheditems.compare1.push(item1)
|
194 | }
|
195 | }
|
196 | }
|
197 | else
|
198 | {
|
199 | if(item1.indexOf(item0)>=0 && item1.length>3){
|
200 | machedItems++
|
201 | return;
|
202 | }
|
203 | else
|
204 | {
|
205 | if(notmacheditems.compare1.indexOf(item1)==-1 && notmacheditems.compare0.indexOf(item0)==-1 && item1.length>3 )
|
206 | {
|
207 | notmacheditems.compare0.push(item0)
|
208 | notmacheditems.compare1.push(item1)
|
209 | }
|
210 | }
|
211 | }
|
212 | })
|
213 | })
|
214 |
|
215 |
|
216 | if(compare0.length == machedItems )
|
217 | {
|
218 | passedmachedItems=true;
|
219 | }
|
220 |
|
221 |
|
222 | compare0.sort();
|
223 | compare1.sort();
|
224 | notmacheditems.compare0.sort();
|
225 | notmacheditems.compare1.sort();
|
226 |
|
227 | let charpoints = 0;
|
228 | let minpoints = 0
|
229 | notmacheditems.compare0.forEach(item0=>{
|
230 | notmacheditems.compare1.forEach(item1=>{
|
231 |
|
232 |
|
233 | if(item0.length<=3)return;
|
234 | if(item1.length<=3)return;
|
235 |
|
236 | if(item0.indexOf('IBIS')>=0 &&item1.indexOf('IBIS')>=0){minpoints=minpoints+2}
|
237 |
|
238 | const length = item0.length >= item1.length ?item1.length : item0.length;
|
239 |
|
240 | const charby1=[];
|
241 | const charby2=[];
|
242 | let matched2char =0;
|
243 | let matched1char =0;
|
244 |
|
245 |
|
246 | for(let i=0;i<length-1;i++)
|
247 | {
|
248 | charby2.push(`${item0.charAt(i)}${item0.charAt(i+1)} `)
|
249 | charby1.push(item0.charAt(i))
|
250 | }
|
251 |
|
252 |
|
253 | for(let i=0;i<length;i++)
|
254 | {
|
255 | if(charby2.indexOf(`${item1.charAt(i)}${item1.charAt(i+1)}`)>=0)
|
256 | {
|
257 | matched2char++;
|
258 | }
|
259 | else
|
260 | {
|
261 | minpoints= minpoints-0.3;
|
262 | }
|
263 |
|
264 | if(charby1.indexOf(item1.charAt(i))>=0)
|
265 | {
|
266 | matched1char++;
|
267 | }
|
268 | else
|
269 | {
|
270 | minpoints= minpoints -1.2;
|
271 | }
|
272 | }
|
273 |
|
274 |
|
275 | if( matched1char <= length+1 &&matched1char >= length-1 )
|
276 | {
|
277 | charpoints++
|
278 | }
|
279 | })
|
280 | })
|
281 |
|
282 |
|
283 |
|
284 | let length1 = inarray[item.key].length;
|
285 | let length2 = array[i].length;
|
286 | lengthsub = length1>=length2 ?length1- length2: length2-length1;
|
287 | if(passedmachedItems && minpoints>=-18 && charpoints>=2 && lengthsub<10){
|
288 |
|
289 |
|
290 | if(email==1 || street == 1)
|
291 | {
|
292 |
|
293 | console.log(" ")
|
294 | console.log("Found by points 0 ");
|
295 | let equals = {
|
296 | 0:inarray,
|
297 | 1:arrayofelements[i],
|
298 | id:`${inarray.id};${arrayofelements[i].id}`
|
299 | };
|
300 | inarray.founddb = true;
|
301 | arrayofelements[i].founddb = true
|
302 |
|
303 | preventdouble.push(`${arrayofelements[i].source}${arrayofelements[i].id}${inarray.source}${inarray.id}`)
|
304 | sameItems.push(equals)
|
305 |
|
306 | currentitem++;
|
307 | return
|
308 |
|
309 | }
|
310 |
|
311 | }
|
312 |
|
313 |
|
314 |
|
315 | let firstword =false;
|
316 | let secondword =false;
|
317 | try{
|
318 |
|
319 | firstword = compare1[1].substr(0,3) == compare0[1].substr(0,3);
|
320 | secondword = compare1[0].substr(0,3) == compare0[0].substr(0,3);
|
321 | }
|
322 | catch (TypeError) {
|
323 | firstword = false
|
324 | secondword = false
|
325 | }
|
326 |
|
327 |
|
328 |
|
329 | if(charpoints == compare0.length && compare0.length>1&&minpoints / compare0.length>=-5) {
|
330 |
|
331 | if(inarray.stars == arrayofelements[i].stars){
|
332 | let equals = {
|
333 | 0:inarray,
|
334 | 1:arrayofelements[i],
|
335 | id:`${inarray.id};${arrayofelements[i].id}`
|
336 | };
|
337 | inarray.founddb = true;
|
338 | arrayofelements[i].founddb = true
|
339 |
|
340 | preventdouble.push(`${arrayofelements[i].source}${arrayofelements[i].id}${inarray.source}${inarray.id}`)
|
341 | sameItems.push(equals)
|
342 |
|
343 | currentitem++;
|
344 | return
|
345 | }
|
346 | }
|
347 |
|
348 | else if(charpoints == compare0.length && minpoints / compare0.length>=-8.5 && firstword && compare1.length>1||
|
349 | charpoints == compare0.length && minpoints / compare0.length>=-8.5 && secondword && compare1.length>1) {
|
350 | inarray.points = 80;
|
351 | if(inarray.stars && arrayofelements[i].stars){
|
352 | inarray.points = inarray.stars == arrayofelements[i].stars ? 95 : 80
|
353 | }
|
354 |
|
355 | if(inarray.points >80 && (email == 1 || geo == 1))
|
356 | {
|
357 |
|
358 | let equals = {
|
359 | 0:inarray,
|
360 | 1:arrayofelements[i],
|
361 | id:`${inarray.id};${arrayofelements[i].id}`
|
362 | };
|
363 | inarray.founddb = true;
|
364 | arrayofelements[i].founddb = true
|
365 |
|
366 | preventdouble.push(`${arrayofelements[i].source}${arrayofelements[i].id}${inarray.source}${inarray.id}`)
|
367 | sameItems.push(equals)
|
368 |
|
369 | currentitem++;
|
370 | return
|
371 | }
|
372 |
|
373 |
|
374 | }
|
375 |
|
376 | else if(word.length>15 && inarray[item.key].length>15 && firstword && secondword && minpoints / compare0.length >-11 ) {
|
377 |
|
378 | if(inarray.stars && arrayofelements[i].stars ){
|
379 | inarray.points = inarray.stars == arrayofelements[i].stars ? 95 : 80;
|
380 |
|
381 | if(inarray.points>=95 ){
|
382 | let equals = {
|
383 | 0:inarray,
|
384 | 1:arrayofelements[i],
|
385 | id:`${inarray.id};${arrayofelements[i].id}`
|
386 | };
|
387 | inarray.founddb = true;
|
388 | arrayofelements[i].founddb = true
|
389 |
|
390 | preventdouble.push(`${arrayofelements[i].source}${arrayofelements[i].id}${inarray.source}${inarray.id}`)
|
391 | sameItems.push(equals)
|
392 |
|
393 | currentitem++;
|
394 | return
|
395 | }
|
396 | }
|
397 |
|
398 | }
|
399 |
|
400 | })
|
401 |
|
402 | }
|
403 |
|
404 | function compareGeo(value1, value2) {
|
405 |
|
406 | if(value1.geovalue && value2.geovalue){
|
407 | let a;
|
408 |
|
409 | if(value1.geovalue< value2.geovalue)
|
410 | a = value2.geovalue - value1.geovalue
|
411 | else
|
412 | a = value1.geovalue - value2.geovalue
|
413 |
|
414 | if (a.toString().length<5)
|
415 | {
|
416 | return 1;
|
417 | }
|
418 | }
|
419 |
|
420 | return 0
|
421 |
|
422 |
|
423 | }
|
424 |
|
425 | function compareEmail(value1, value2) {
|
426 |
|
427 | if(!value1.manualEmail || !value2.manualEmail)return 0;
|
428 |
|
429 | if(value1.manualEmail == 'NN' || value2.manualEmail == 'NN') return 0;
|
430 |
|
431 | if(value1.manualEmail && value2.manualEmail){
|
432 |
|
433 |
|
434 |
|
435 |
|
436 |
|
437 | let email1 = value1.manualEmail.replace(" ","").substr(1,value1.manualEmail.length-2).replace('.','').replace('.','')
|
438 | let email2 = value2.manualEmail.replace(" ","").substr(1,value2.manualEmail.length).replace('.','').replace('.','')
|
439 |
|
440 |
|
441 | if(email2.indexOf(email1)>=0)
|
442 | return 1
|
443 |
|
444 | }
|
445 | return 0;
|
446 | }
|
447 |
|
448 | function compareStreet(value1,value2) {
|
449 |
|
450 | if(value1.adressStreet) {
|
451 | if (value1.adressStreet == value2.adressStreet) {
|
452 | return 1;
|
453 | }
|
454 | }
|
455 | return 0
|
456 | }
|
457 |
|
458 | function parser(w) {
|
459 | let word = w.toUpperCase().replace('APARTHOTEL ','').replace('HOTEL','').replace('-STERNE').replace(/[^\w\s]/gi, '')
|
460 |
|
461 |
|
462 | if(word.indexOf(" ")== 0)
|
463 | while(word.indexOf(" ")== 0){
|
464 | word = word.substring(1,word.length)
|
465 | }
|
466 |
|
467 | if(/\s+$/.test(word))
|
468 | while(/\s+$/.test(word)){
|
469 | word = word.substring(0,word.length-1)
|
470 | }
|
471 |
|
472 |
|
473 | return word;
|
474 | }
|
475 |
|
476 | function exclude(name) {
|
477 |
|
478 |
|
479 | |
480 |
|
481 |
|
482 |
|
483 |
|
484 |
|
485 | return name;
|
486 | }
|
487 |
|
488 | function findNotSameHotels() {
|
489 |
|
490 | return new Promise(resolve=>{
|
491 | let i=0;
|
492 | notsameitems = [];
|
493 | console.log("===================================")
|
494 | console.log('findNotSameHotels:')
|
495 | arrayofelements.forEach(h=>{
|
496 |
|
497 | if(h.founddb == false && h.source != 'db'){
|
498 | console.log(h.name)
|
499 | notsameitems.push(h)
|
500 | i++;
|
501 | }
|
502 | });
|
503 |
|
504 | if(i==0)console.log("--none--")
|
505 | resolve(true);
|
506 | })
|
507 |
|
508 | }
|
509 |
|
510 | function fetchDBData(city) {
|
511 | return new Promise(resolve=>{
|
512 | _HP.db.execute(
|
513 | ` SELECT H.name , H.id, H.adressRaw, H.geodata, H.stars, H.manualEmail
|
514 | FROM CO2Hotels as H
|
515 | LEFT JOIN CO2CityConfig as C
|
516 | ON H.linkCity = C.linkCity
|
517 | WHERE C.name = ? and H.isDead = 0; `,[city],
|
518 | function(err, rows, fields)
|
519 | {
|
520 | if(rows){
|
521 | rows.forEach(item=>{
|
522 | let geo =null;
|
523 | if(item.geodata)
|
524 | {
|
525 | geo = item.geodata.split(',');
|
526 | geo =parseInt(geo[0].replace('.',''))+parseInt(geo[1].replace('.',''))
|
527 | }
|
528 |
|
529 | const hotel={
|
530 | id: item.id,
|
531 | name: item.name.replace('&','&').replace(''',"\'"),
|
532 | adressRaw: item.adressRaw,
|
533 | geodata: item.geodata,
|
534 | geovalue: geo,
|
535 |
|
536 | stars: item.stars,
|
537 | manualEmail: item.manualEmail,
|
538 | adressStreet:item.adressStreet,
|
539 | founddb: 'db',
|
540 | source: "db",
|
541 | points:0,
|
542 | }
|
543 | arrayofelements.push(hotel)
|
544 | })
|
545 | generateTree(arrayofelements[0]).then(keys=> {
|
546 |
|
547 |
|
548 | console.log(" ")
|
549 | console.log("Looking for equal Hotels. Place "+city)
|
550 | console.log("........................................")
|
551 | console.log(" ")
|
552 | init().then(res=>{
|
553 | resolve(res)
|
554 | })
|
555 | });
|
556 | }else {
|
557 | console.log(err)
|
558 | }
|
559 | }
|
560 | );
|
561 | })
|
562 | }
|
563 |
|
564 |
|
565 |
|
566 | module.exports ={
|
567 | "methods":{
|
568 | startBookingCompare:startBooking
|
569 | },
|
570 |
|
571 | }
|
572 |
|