ilyrics/iLyrics/Magistrix.m

//
//  Magistrix.m
//  iLyrics
//
//  Created by Kim Wittenburg on 14.06.12.
//  Copyright (c) 2012 __MyCompanyName__. All rights reserved.
//
//
//TODO: Use correct Date

#import "Magistrix.h"

typedef enum {
    LyricsPage,
    ResultsPage,
    NoResultsPage,
    UnknownPage
} PageType;

@implementation Magistrix {
    NSString *query;
    int loadedResults;
    int resultCount;
}

-(NSString*) name {
    return @"Magistrix";
}

-(NSDate *)hosterVersion {
    return [NSDate dateWithString:@"2012-06-16 11:00:00 +0100"];
}

-(void) startNewSearchForQuery: (NSString*) q {
    [self resetLoadedResults];
    query = q;
}

-(BOOL) hasMoreResults {
    return loadedResults < resultCount;
}

-(NSArray*) nextResults {
    int site = (loadedResults/10) + 1;
    NSString *searchPath = [NSString stringWithFormat:@"http://www.magistrix.de/lyrics/search?q=%@&page=%i", [self stringByFormattingQuery:query], site];
    NSURL *searchURL = [NSURL URLWithString:searchPath];
    NSError *error;
    NSString *page = [NSString stringWithContentsOfURL:searchURL encoding:NSUTF8StringEncoding error:&error];
    if (error) {
        //Network error or invalid query
        return nil;
    }
    PageType pType = [self typeOfPage:page];
    if (pType == LyricsPage) {
        resultCount = 1;
        loadedResults = 1;
        return [NSArray arrayWithObject:[self searchResultFromLyricsPage:page atURL:searchURL]];
    } else if (pType == ResultsPage) {
        [self shouldSetResultCountFromPage:page];
        return [self searchResultsFromPage:page];
    } else if (pType == NoResultsPage) {
        resultCount = 0;
        return [[NSArray alloc] init];
    } else {
        NSRunAlertPanel(NSLocalizedString(@"Magistrix.messages.unknownPage.title", @""), NSLocalizedString(@"Magistrix.messages.unknownPage.detail", @""), NSLocalizedString(@"OK", @""), nil, nil);
        resultCount = 0;
        return [[NSArray alloc] init];
    }
}

-(NSString *) stringByFormattingQuery: (NSString *) q {
    NSRange stringRange = NSMakeRange(0, [q length]);
    //Can replace äöü with aou, no difference in results
    NSCharacterSet *characters = [[NSCharacterSet alphanumericCharacterSet] invertedSet];
    return [[[[[[q stringByReplacingOccurrencesOfString:@" " withString:@"+"] stringByReplacingOccurrencesOfString:@"Ö" withString:@"o" options:NSCaseInsensitiveSearch range:stringRange] stringByReplacingOccurrencesOfString:@"Ä" withString:@"a" options:NSCaseInsensitiveSearch range:stringRange] stringByReplacingOccurrencesOfString:@"Ü" withString:@"ü" options:NSCaseInsensitiveSearch range:stringRange] stringByReplacingOccurrencesOfString:@"&" withString:@"%26"] stringByTrimmingCharactersInSet:characters];
}

-(PageType) typeOfPage: (NSString *) page {
    if ([page rangeOfString:@"<title>Songtext-Suche</title>"].location != NSNotFound) {
        if ([page rangeOfString:@"<div class='empty_collection'>"].location != NSNotFound) {
            return NoResultsPage;
        } else {
            return ResultsPage;
        }
    }
    if ([page rangeOfString:@"<div id='songtext'>"].location != NSNotFound) {
        return LyricsPage;
    }
    return UnknownPage;
}

-(SearchResult*) searchResultFromLyricsPage: (NSString *) page atURL: (NSURL *) url{
    int headingStart = NSMaxRange([page rangeOfString:@"<h1>"]);
    int headingEnd = [page rangeOfString:@"</h1>"].location;
    int artistStart = NSMaxRange([page rangeOfString:@">" options:NSCaseInsensitiveSearch range:NSMakeRange(headingStart, headingEnd-headingStart)]);
    NSRange artistEndTag = [page rangeOfString:@"</a>" options:NSCaseInsensitiveSearch range:NSMakeRange(artistStart, headingEnd-artistStart)];
    int artistEnd = artistEndTag.location;
    int songNameStart = NSMaxRange(artistEndTag);
    int songNameEnd = headingEnd;
    NSString *artist = [self stringByRemovingHTMLTags:[page substringWithRange:NSMakeRange(artistStart, artistEnd-artistStart)]];
    NSString *songName = [page substringWithRange:NSMakeRange(songNameStart, songNameEnd-songNameStart)];
    //Remove the " Lyric" and the " &ndash; " from the Song name
    songName = [self stringByRemovingHTMLTags:[[songName substringToIndex:[songName length]-[@" Lyric" length]] substringFromIndex:[@" &ndash; " length]]];
    NSString *preview = [self lyricsFromPage:page];
    return [[SearchResult alloc]initWithName:songName fromArtist:artist preview:preview link:url];
}

-(NSArray*) searchResultsFromPage: (NSString *) page {
    int resultsTableStart = NSMaxRange([page rangeOfString:@"<table>"]);
    int resultsTableEnd = [page rangeOfString:@"</table>"].location;
    NSArray *resultTags = [self resultTagsFromTable:[page substringWithRange:NSMakeRange(resultsTableStart, resultsTableEnd-resultsTableStart)]];
    NSMutableArray *searchResults = [[NSMutableArray alloc] init];
    for (NSString *tag in resultTags) {
        [searchResults addObject:[self searchResultFromResultTag:tag]];
    }
    //Increase loadedResults by 10 which is a full results site
    //Even if there are less than 10 results on a page (would not have any effect)
    loadedResults += 10;
    return searchResults;
}

-(NSArray*) resultTagsFromTable: (NSString *) table {
    NSRange restRange = NSMakeRange(0, [table length]);
    NSMutableArray *tags = [[NSMutableArray alloc] init];
    NSUInteger currentIndex = [table rangeOfString:@"<td>"].location;
    while (currentIndex != NSNotFound) {
        int startIndex = currentIndex + [@"<td>" length];
        int endIndex = [table rangeOfString:@"</td>" options:NSCaseInsensitiveSearch range:restRange].location;
        NSRange tagRange = NSMakeRange(startIndex, endIndex-startIndex);
        [tags addObject:[table substringWithRange:tagRange]];
        restRange = [self restRangeFromString:table subtractingRange:tagRange];
        restRange.length -= [@"</td>" length];
        restRange.location += [@"</td>" length];
        currentIndex = [table rangeOfString:@"<td>" options:NSCaseInsensitiveSearch range:restRange].location;
    }
    return tags;
}

-(SearchResult*) searchResultFromResultTag: (NSString *) tag {
    NSRange artistStartRange = [tag rangeOfString:@">"];
    int artistEndIndex = [tag rangeOfString:@"<" options:NSCaseInsensitiveSearch range:[self restRangeFromString:tag subtractingRange:artistStartRange]].location;
    int artistStartIndex = NSMaxRange(artistStartRange);
    NSString *artist = [self stringByRemovingHTMLTags:[tag substringWithRange:NSMakeRange(artistStartIndex, artistEndIndex-artistStartIndex)]];
    
    NSRange restRange = [self restRangeFromString:tag subtractingRange:NSMakeRange(artistEndIndex, [@"</a>" length])];
    NSRange songNameTagStartRange = [tag rangeOfString:@"<a href=\"" options:NSCaseInsensitiveSearch range:restRange];
    int linkStart = NSMaxRange(songNameTagStartRange);
    NSRange linkEndRage = [tag rangeOfString:@"\"" options:NSCaseInsensitiveSearch range:[self restRangeFromString:tag subtractingRange:songNameTagStartRange]];
    NSURL *link = [self urlFromHref:[tag substringWithRange:NSMakeRange(linkStart, linkEndRage.location-linkStart)]];
    int songNameStart = NSMaxRange([tag rangeOfString:@">" options:NSCaseInsensitiveSearch range:[self restRangeFromString:tag subtractingRange:songNameTagStartRange]]);
    int songNameEnd = [tag rangeOfString:@"</a>" options:NSCaseInsensitiveSearch range:[self restRangeFromString:tag subtractingRange:songNameTagStartRange]].location;
    NSString *songName = [self stringByRemovingHTMLTags:[tag substringWithRange:NSMakeRange(songNameStart, songNameEnd-songNameStart)]];
    int previewStart = songNameEnd + [@"</a>" length] + [@"\n<p>" length];
    NSRange previewRestRange = NSMakeRange(previewStart, [tag length]-previewStart);
    int previewEnd = [tag rangeOfString:@"</p>" options:NSCaseInsensitiveSearch range:previewRestRange].location;
    NSString *preview = [self stringByRemovingHTMLTags:[tag substringWithRange:NSMakeRange(previewStart, previewEnd-previewStart)]];
    return [[SearchResult alloc] initWithName:songName fromArtist:artist preview:preview link:link];
}

-(NSURL*) urlFromHref: (NSString *) link {
    if ([link hasPrefix:@"/"]) {
        return [NSURL URLWithString:[NSString stringWithFormat:@"http://www.Magistrix.de%@", link]];
    } else {
        return [NSURL URLWithString:link];
    }
}

-(NSString*) stringByRemovingHTMLTags: (NSString *)string {
    return [[[[[[[[[[[string stringByReplacingOccurrencesOfString:@"<strong>" withString:@""] stringByReplacingOccurrencesOfString:@"</strong>" withString:@""] stringByReplacingOccurrencesOfString:@"<b>" withString:@""] stringByReplacingOccurrencesOfString:@"</b>" withString:@""] stringByReplacingOccurrencesOfString:@"<i>" withString:@""] stringByReplacingOccurrencesOfString:@"</i>" withString:@""] stringByReplacingOccurrencesOfString:@"<p>" withString:@""] stringByReplacingOccurrencesOfString:@"</p>" withString:@""] stringByReplacingOccurrencesOfString:@"<br />" withString:@""] stringByReplacingOccurrencesOfString:@"&quot;" withString:@"\""] stringByReplacingOccurrencesOfString:@"&amp;" withString:@"&"];
}

-(NSRange) restRangeFromString: (NSString *) page subtractingRange: (NSRange) aRange {
    int loc = NSMaxRange(aRange);
    return NSMakeRange(loc, [page length]-loc);
}

-(void) shouldSetResultCountFromPage: (NSString *) page {
    if (resultCount == 0) {
        //Nothing loaded before
        NSRange resultsLabelStartRange = [page rangeOfString:@"<p>Ergebnisse 1 bis "];
        int resultsLabelEnd = [page rangeOfString:@"</p>" options:NSCaseInsensitiveSearch range:[self restRangeFromString:page subtractingRange:resultsLabelStartRange]].location;
        NSString *resultsLabel = [page substringWithRange:NSMakeRange(NSMaxRange(resultsLabelStartRange), resultsLabelEnd-resultsLabelStartRange.location)];
        if ([resultsLabel rangeOfString:@"ungefähr"].location != NSNotFound){
            //Needs not more than 100 which means 10 sites (which is the maximum)
            resultCount = 100;
        } else {
            int resultsStart = NSMaxRange([resultsLabel rangeOfString:@"von "]);
            int resultsEnd = [resultsLabel rangeOfString:@"<"].location;
            NSString *results = [resultsLabel substringWithRange:NSMakeRange(resultsStart, resultsEnd-resultsStart)];
            resultCount = [results intValue];
        }
        if (resultCount > 100) {
            resultCount = 100;
        }
    }
}

-(void) resetLoadedResults {
    loadedResults = 0;
}

-(Lyrics*) lyricsBySearchResult: (SearchResult *) result {
    NSError *error = nil;
    NSString *page = [NSString stringWithContentsOfURL:[result link] encoding:NSUTF8StringEncoding error:&error];
    if (error) {
        return nil;
    }
    NSString *lyrics = [self lyricsFromPage:page];
    return [[Lyrics alloc] initWithName:[result name] byArtist:[result artist] withLyrics:lyrics];
}


-(NSString *) lyricsFromPage: (NSString *)page {
    int lyricsStart = NSMaxRange([page rangeOfString:@"<div id='songtext'>"]) + [@"\n" length];
    NSRange restRange = NSMakeRange(lyricsStart, [page length]-lyricsStart);
    int lyricsEnd = [page rangeOfString:@"<div class='clear'></div>" options:NSCaseInsensitiveSearch range:restRange].location;
    NSString *lyrics = [self stringByRemovingHTMLTags:[page substringWithRange:NSMakeRange(lyricsStart, lyricsEnd-lyricsStart)]];
    return lyrics;
}
@end