Fixes to connection handling in daemon.
[jenirok] / src / common / eniro.cpp
1 /*
2  * This file is part of Jenirok.
3  *
4  * Jenirok is free software: you can redistribute it and/or modify
5  * it under the terms of the GNU General Public License as published by
6  * the Free Software Foundation, either version 3 of the License, or
7  * (at your option) any later version.
8  *
9  * Jenirok is distributed in the hope that it will be useful,
10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
12  * GNU General Public License for more details.
13  *
14  * You should have received a copy of the GNU General Public License
15  * along with Jenirok.  If not, see <http://www.gnu.org/licenses/>.
16  *
17  */
18
19 #include <QtCore/QDebug>
20 #include "eniro.h"
21
22 namespace
23 {
24     static const QString SITE_URLS[Eniro::SITE_COUNT] =
25     {
26             "http://wap.eniro.fi/",
27             "http://wap.eniro.se/",
28             "http://wap.eniro.dk/"
29     };
30
31     static const QString SITE_NAMES[Eniro::SITE_COUNT] =
32     {
33          "finnish",
34          "swedish",
35          "danish"
36     };
37
38     static const QString SITE_IDS[Eniro::SITE_COUNT] =
39     {
40          "fi",
41          "se",
42          "dk"
43     };
44
45     static const QString INVALID_LOGIN_STRING = "Invalid login details";
46     static const QString TIMEOUT_STRING = "Request timed out";
47     static const QString PERSON_REGEXP = "<td class=\"hTd2\">(.*)<b>(.*)</td>";
48     static const QString YELLOW_REGEXP = "<td class=\"hTd2\">(.*)<span class=\"gray\">(.*)</td>";
49     static const QString SINGLE_REGEXP = "<div class=\"header\">(.*)</div>(.*)<div class=\"callRow\">(.*)(<div class=\"block\">|</p>(.*)<br/>|</p>(.*)<br />)";
50     static const QString NUMBER_REGEXP = "<div class=\"callRow\">(.*)</div>";
51     static const QString LOGIN_CHECK = "<input class=\"inpTxt\" id=\"loginformUsername\"";
52 }
53
54 Eniro::Eniro(QObject *parent): Source(parent), site_(Eniro::FI),
55 loggedIn_(false), username_(""), password_(""),
56 timerId_(0), pendingSearches_(), pendingNumberRequests_()
57 {
58 }
59
60 Eniro::~Eniro()
61 {
62 }
63
64 void Eniro::abort()
65 {
66     Source::abort();
67
68     for(searchMap::iterator sit = pendingSearches_.begin();
69     sit != pendingSearches_.end(); sit++)
70     {
71         if(sit.value() != 0)
72         {
73             delete sit.value();
74             sit.value() = 0;
75         }
76     }
77
78     pendingSearches_.clear();
79
80     for(numberMap::iterator nit = pendingNumberRequests_.begin();
81     nit != pendingNumberRequests_.end(); nit++)
82     {
83         if(nit.value() != 0)
84         {
85             delete nit.value();
86             nit.value() = 0;
87         }
88     }
89
90     pendingNumberRequests_.clear();
91     pendingLoginRequests_.clear();
92 }
93
94 void Eniro::setSite(Eniro::Site site)
95 {
96     site_ = site;
97 }
98
99 void Eniro::timerEvent(QTimerEvent* t)
100 {
101     Q_UNUSED(t);
102
103     int currentId = http_.currentId();
104
105     if(currentId)
106     {
107         searchMap::const_iterator it = pendingSearches_.find(currentId);
108
109         if(it != pendingSearches_.end())
110         {
111             QVector <Eniro::Result> results = it.value()->results;
112             SearchDetails details = it.value()->details;
113
114             abort();
115
116             setError(TIMEOUT, TIMEOUT_STRING);
117
118             emit requestFinished(results, details, true);
119         }
120     }
121 }
122
123 void Eniro::login(QString const& username,
124                   QString const& password)
125 {
126     username_ = username;
127     password_ = password;
128     loggedIn_ = true;
129 }
130
131 void Eniro::logout()
132 {
133     username_ = "";
134     password_ = "";
135     loggedIn_ = false;
136 }
137
138 void Eniro::search(SearchDetails const& details)
139 {
140     resetTimeout();
141
142     SearchType type = details.type;
143
144     // Only logged in users can use other than person search
145     if(!loggedIn_ && site_ == FI)
146     {
147         type = PERSONS;
148     }
149
150     QUrl url = createUrl(details.query, details.location);
151     QString what;
152
153     if(loggedIn_ || site_ != FI)
154     {
155         switch(type)
156         {
157         case YELLOW_PAGES:
158             what = "mobcs";
159             break;
160
161         case PERSONS:
162             what = "mobwp";
163             break;
164
165         default:
166             what = "moball";
167             break;
168         }
169
170     }
171     else
172     {
173         what = "moball";
174     }
175
176     url.addQueryItem("what", what);
177
178     http_.setHost(url.host(), url.port(80));
179     int id = http_.get(url.encodedPath() + '?' + url.encodedQuery());
180
181     QVector <Source::Result> results;
182
183     // Store search data for later identification
184     SearchData* newData = new SearchData;
185     newData->details = details;
186     newData->results = results;
187     newData->foundNumbers = 0;
188     newData->numbersTotal = 0;
189
190     // Store request id so that it can be identified later
191     pendingSearches_[id] = newData;
192
193 }
194
195 void Eniro::handleHttpData(int id, QByteArray const& data)
196 {
197     searchMap::const_iterator searchIt;
198     numberMap::const_iterator numberIt;
199
200     // Check if request is pending search request
201     if((searchIt = pendingSearches_.find(id)) !=
202         pendingSearches_.end())
203     {
204         // Load results from html data
205         loadResults(id, data);
206     }
207
208     // Check if request is pending number requests
209     else if((numberIt = pendingNumberRequests_.find(id)) !=
210         pendingNumberRequests_.end())
211     {
212         // Load number from html data
213         loadNumber(id, data);
214     }
215
216     // Check for login request
217     else if(pendingLoginRequests_.find(id) !=
218         pendingLoginRequests_.end())
219     {
220         bool success = true;
221
222         // If html source contains LOGIN_CHECK, login failed
223         if(data.indexOf(LOGIN_CHECK) != -1)
224         {
225             success = false;
226         }
227
228         emit loginStatus(success);
229     }
230
231 }
232
233 void Eniro::handleHttpError(int id)
234 {
235     searchMap::const_iterator searchIt;
236     numberMap::const_iterator numberIt;
237
238     // Check if request is pending search request
239     if((searchIt = pendingSearches_.find(id)) !=
240         pendingSearches_.end())
241     {
242         setError(CONNECTION_FAILURE, http_.errorString());
243         emitRequestFinished(id, searchIt.value(), true);
244     }
245
246     // Check if request is pending number requests
247     else if((numberIt = pendingNumberRequests_.find(id)) !=
248         pendingNumberRequests_.end())
249     {
250         setError(CONNECTION_FAILURE, http_.errorString());
251         delete pendingNumberRequests_[id];
252         pendingNumberRequests_.remove(id);
253     }
254
255     // Check for login request
256     else if(pendingLoginRequests_.find(id) !=
257         pendingLoginRequests_.end())
258     {
259         emit loginStatus(false);
260
261     }
262 }
263
264 // Loads results from html source code
265 void Eniro::loadResults(int id, QString const& httpData)
266 {
267     searchMap::iterator it = pendingSearches_.find(id);
268
269     QRegExp rx("((" + YELLOW_REGEXP + ")|(" + PERSON_REGEXP + ")|(" + SINGLE_REGEXP + "))");
270     rx.setMinimal(true);
271
272     bool requestsPending = false;
273     int pos = 0;
274     QString data;
275
276     // Find all matches
277     while((pos = rx.indexIn(httpData, pos)) != -1)
278     {
279         pos += rx.matchedLength();
280
281         data = rx.cap(1);
282
283         data = stripTags(data);
284
285         QStringList rows = data.split('\n');
286
287         for(int i = 0; i < rows.size(); i++)
288         {
289             // Remove white spaces
290             QString trimmed = rows.at(i).trimmed().toLower();
291
292             // Remove empty strings
293             if(trimmed.isEmpty())
294             {
295                 rows.removeAt(i);
296                 i--;
297             }
298             else
299             {
300                 // Convert words to uppercase
301                 rows[i] = ucFirst(trimmed);
302             }
303         }
304
305         Result result;
306
307         switch(site_)
308         {
309         case FI:
310             result.country = "Finland";
311             break;
312         case SE:
313             result.country = "Sweden";
314             break;
315         case DK:
316             result.country = "Denmark";
317             break;
318         }
319
320         int size = rows.size();
321
322         switch(size)
323         {
324         case 1:
325             result.name = rows[0];
326             break;
327
328         case 2:
329             result.name = rows[0];
330             result.city = rows[1];
331             break;
332
333         case 3:
334             if(isPhoneNumber(rows[1]))
335             {
336                 result.name = rows[0];
337                 result.number = cleanUpNumber(rows[1]);
338                 result.city = rows[2];
339             }
340             else
341             {
342                 result.name = rows[0];
343                 result.street = rows[1];
344                 result.city = rows[2];
345             }
346             break;
347
348         case 4:
349             result.name = rows[0];
350             // Remove slashes and spaces from number
351             result.number = cleanUpNumber(rows[1]);
352             result.street = rows[2];
353             result.city = rows[3];
354             break;
355
356         default:
357             bool ok = false;
358
359             for(int a = 0; a < size && a < 8; a++)
360             {
361                 if(isPhoneNumber(rows[a]))
362                 {
363                     result.name = rows[0];
364                     result.number = cleanUpNumber(rows[a]);
365
366                     for(int i = a + 1; i < size && i < 8; i++)
367                     {
368                         if(!isPhoneNumber(rows[i]) && size > i + 1 && isStreet(rows[i]))
369                         {
370                             result.street = rows[i];
371                             result.city = rows[i+1];
372                             ok = true;
373                             break;
374                         }
375                     }
376
377                 }
378
379             }
380
381             if(ok)
382             {
383                 break;
384             }
385
386             continue;
387
388         }
389
390         it.value()->results.push_back(result);
391
392         unsigned int foundResults = ++(it.value()->numbersTotal);
393
394         // If phone number search is enabled, we have to make another
395         // request to find it out
396         if(getFindNumber() && size < 4 && (loggedIn_ || site_ != FI) &&
397                 it.value()->details.type != YELLOW_PAGES)
398         {
399             requestsPending = true;
400             getNumberForResult(id, it.value()->results.size() - 1, it.value()->details);
401         }
402         // Otherwise result is ready
403         else
404         {
405             emit resultAvailable(result, it.value()->details);
406         }
407
408         unsigned int maxResults = getMaxResults();
409
410         // Stop searching if max results is reached
411         if(maxResults && (foundResults >= maxResults))
412         {
413             break;
414         }
415     }
416
417     // If there were no results or no phone numbers needed to
418     // be fetched, the whole request is ready
419     if(it.value()->numbersTotal == 0 || !requestsPending)
420     {
421         bool error = false;
422
423         if(httpData.indexOf(LOGIN_CHECK) != -1)
424         {
425             setError(INVALID_LOGIN, INVALID_LOGIN_STRING),
426             error = true;
427         }
428
429         emitRequestFinished(it.key(), it.value(), error);
430     }
431 }
432
433 // Loads phone number from html source
434 void Eniro::loadNumber(int id, QString const& result)
435 {
436     numberMap::iterator numberIt = pendingNumberRequests_.find(id);
437
438     // Make sure that id exists in pending number requests
439     if(numberIt == pendingNumberRequests_.end() || numberIt.value() == 0)
440     {
441         return;
442     }
443
444     searchMap::iterator searchIt = pendingSearches_.find(numberIt.value()->searchId);
445
446     if(searchIt == pendingSearches_.end() || searchIt.value() == 0)
447     {
448         return;
449     }
450
451     QRegExp rx(NUMBER_REGEXP);
452     rx.setMinimal(true);
453
454     int pos = 0;
455     bool error = true;
456
457     if((pos = rx.indexIn(result, pos)) != -1)
458     {
459         QString data = rx.cap(1);
460         data = stripTags(data);
461
462         QString trimmed = data.trimmed();
463
464         if(!trimmed.isEmpty())
465         {
466             // Remove whitespaces from number
467             searchIt.value()->results[numberIt.value()->index].number = cleanUpNumber(trimmed);
468
469             emit resultAvailable(searchIt.value()->results[numberIt.value()->index], searchIt.value()->details);
470
471             unsigned int found = ++searchIt.value()->foundNumbers;
472
473             // Check if all numbers have been found
474             if(found >= searchIt.value()->numbersTotal)
475             {
476                 emitRequestFinished(searchIt.key(), searchIt.value(), false);
477             }
478
479             // If number was found, there was no error
480             error = false;
481         }
482     }
483
484     if(error)
485     {
486         setError(INVALID_LOGIN, INVALID_LOGIN_STRING);
487         emitRequestFinished(searchIt.key(), searchIt.value(), true);
488     }
489
490     // Remove number request
491     int key = numberIt.key();
492
493     delete pendingNumberRequests_[key];
494     pendingNumberRequests_[key] = 0;
495     pendingNumberRequests_.remove(key);
496
497 }
498
499 QUrl Eniro::createUrl(QString const& query, QString const& location)
500 {
501     QUrl url(SITE_URLS[site_] + "query");
502
503     if(!query.isEmpty())
504     {
505         url.addQueryItem("search_word", query);
506     }
507
508     if(!location.isEmpty())
509     {
510         url.addQueryItem("geo_area", location);
511     }
512
513     unsigned int maxResults = getMaxResults();
514
515     if(maxResults)
516     {
517         url.addQueryItem("hpp", QString::number(maxResults));
518     }
519     if(loggedIn_ && site_ == FI)
520     {
521         url.addQueryItem("login_name", username_);
522         url.addQueryItem("login_password", password_);
523     }
524
525     fixUrl(url);
526
527     return url;
528 }
529
530 // Creates a new request for phone number retrieval
531 void Eniro::getNumberForResult(int id, int index, SearchDetails const& details)
532 {
533     QUrl url = createUrl(details.query, details.location);
534     url.addQueryItem("what", "mobwpinfo");
535     url.addQueryItem("search_number", QString::number(index + 1));
536
537     http_.setHost(url.host(), url.port(80));
538     int requestId = http_.get(url.encodedPath() + '?' + url.encodedQuery());
539     NumberData* number = new NumberData;
540     number->searchId = id;
541     number->index = index;
542     pendingNumberRequests_[requestId] = number;
543
544 }
545
546 void Eniro::emitRequestFinished(int key, SearchData* data, bool error)
547 {
548     emit requestFinished(data->results, data->details, error);
549     delete pendingSearches_[key];
550     pendingSearches_[key] = 0;
551     pendingSearches_.remove(key);
552 }
553
554
555 QMap <Eniro::Site, Eniro::SiteDetails> Eniro::getSites()
556 {
557     QMap <Site, SiteDetails> sites;
558
559     for(int i = 0; i < SITE_COUNT; i++)
560     {
561         SiteDetails details;
562         details.name = SITE_NAMES[i];
563         details.id = SITE_IDS[i];
564         sites[static_cast<Site>(i)] = details;
565     }
566
567     return sites;
568 }
569
570 Eniro::Site Eniro::stringToSite(QString const& str)
571 {
572     Site site = FI;
573     QString lower = str.toLower();
574
575     for(int i = 0; i < SITE_COUNT; i++)
576     {
577         if(lower == SITE_NAMES[i] || lower == SITE_IDS[i])
578         {
579             site = static_cast <Site> (i);
580             break;
581         }
582     }
583
584     return site;
585 }
586
587 bool Eniro::isStreet(QString const& str)
588 {
589     static QRegExp number("([0-9]+)");
590     int a = number.indexIn(str);
591     int b = str.indexOf(" ");
592
593     if((a == -1 && b == -1) || (a != -1 && b != -1))
594     {
595         return true;
596     }
597
598     return false;
599 }