Chris@0: /* -*- c-basic-offset: 4 indent-tabs-mode: nil -*-  vi:set ts=8 sts=4 sw=4: */
Chris@0: 
Chris@0: #include "ImportWikipediaWorksList.h"
Chris@0: 
Chris@0: #include <dataquay/Debug.h>
Chris@0: 
Chris@0: #include <QFile>
Chris@0: #include <QFileInfo>
Chris@0: #include <QTextStream>
Chris@0: #include <QRegExp>
Chris@0: #include <QVariant>
Chris@0: 
Chris@0: #include <exception>
Chris@0: 
Chris@0: using namespace Dataquay;
Chris@0: 
Chris@0: namespace ClassicalData {
Chris@0: 
Chris@0: void
Chris@0: WikipediaWorksListImporter::setSource(QUrl source)
Chris@0: {
Chris@0:     DEBUG << "WikipediaWorksListImporter::setSource: " << source << endl;
Chris@0:     import(source);
Chris@0: }
Chris@0: 
Chris@0: static QString
Chris@0: sanitise(QString field, QString &linkText)
Chris@0: {
Chris@0:     int mp;
Chris@0: 
Chris@0:     field.replace(QString::fromUtf8("\342\200\222"), "-");
Chris@0:     field.replace(QString::fromUtf8("\342\200\223"), "-");
Chris@0:     field.replace(QString::fromUtf8("\342\200\224"), "-");
Chris@0:     field.replace(QString::fromUtf8("\342\200\225"), "-");
Chris@0: 
Chris@0:     field.replace(QString::fromUtf8("\342\231\255"), "-flat");
Chris@0:     field.replace(QString::fromUtf8("\342\231\257"), "-sharp");
Chris@0: 
Chris@0:     QRegExp link2("([^A-Za-z]*)\\[\\[([^\\]\\|]+)\\|([^\\]]+)\\]\\]");
Chris@0:     if ((mp = link2.indexIn(field)) >= 0) {
Chris@0: 	if (linkText == "" && mp < 4) linkText = link2.cap(2);
Chris@0: 	field.replace(mp, link2.matchedLength(), link2.cap(1) + link2.cap(3));
Chris@0: 	return sanitise(field, linkText);
Chris@0:     }
Chris@0: 
Chris@0:     QRegExp link1("^([^A-Za-z]*)\\[\\[([^\\]]+)\\]\\]");
Chris@0:     if ((mp = link1.indexIn(field)) >= 0) {
Chris@0: 	if (linkText == "") linkText = link1.cap(2);
Chris@0: 	field.replace(mp, link1.matchedLength(), link1.cap(1) + link1.cap(2));
Chris@0: 	return sanitise(field, linkText);
Chris@0:     }
Chris@0: 
Chris@0:     field = field.trimmed();
Chris@0: 
Chris@0:     field.replace("[", "");
Chris@0:     field.replace("]", "");
Chris@0:     field.replace(QRegExp("\\{+[^\\}]*\\}+ *"), " ");
Chris@0:     field.replace("'''", "\"");
Chris@0:     field.replace("''", "\"");
Chris@0:     field.replace("&quot;", "\"");
Chris@0:     field.replace("\"\"", "\"");
Chris@0:     field.replace(QRegExp("^[\'\"] (\")?"), "\"");
Chris@0:     field.replace(QRegExp("&lt;[^&]*&gt;"), "");
Chris@0:     field.replace(QRegExp("^\\**"), "");
Chris@0:     
Chris@0:     if (field.endsWith("c.")) {
Chris@0:         // historical artifact from removal of Bruckner year indication (c. 1856)
Chris@0:         field = field.left(field.length()-2);
Chris@0:     }
Chris@0: 
Chris@0:     while (field.endsWith(".") || field.endsWith(",")) {
Chris@0:         field = field.left(field.length()-1);
Chris@0:     }
Chris@0: 
Chris@0:     if (field.startsWith(";") || field.startsWith(":") || field.startsWith(",")
Chris@0:         || field.startsWith("-")) {
Chris@0:         field = field.right(field.length()-1);
Chris@0:     }
Chris@0: 
Chris@0:     if (field.startsWith("(") && field.endsWith(")")) {
Chris@0:         DEBUG << "before: " << field;
Chris@0:         field = field.mid(1, field.length()-2);
Chris@0:         DEBUG << "after: " << field;
Chris@0:     }
Chris@0: 
Chris@0:     field.replace(QRegExp("^\\**"), "");
Chris@0:     if (field == ")" || field == "(") {
Chris@0:         field = "";
Chris@0:     }
Chris@0: 
Chris@0:     field.replace(" - ,", ",");
Chris@0:     field.replace("  ", " ");
Chris@0: 
Chris@0:     return field.trimmed();
Chris@0: }
Chris@0: 
Chris@0: static QString
Chris@0: extractYear(QString datefield)
Chris@0: {
Chris@0:     QRegExp re("[0-9]{4}");
Chris@0:     if (re.indexIn(datefield) >= 0) {
Chris@0: 	return re.cap(0);
Chris@0:     }
Chris@0:     return "";
Chris@0: }
Chris@0: 
Chris@0: static QString
Chris@0: extractKey(QString titlefield)
Chris@0: {
Chris@0:     QRegExp re("in ([A-H]([ -][a-z]+)? (major|minor))");
Chris@0:     if (re.indexIn(titlefield) >= 0) {
Chris@0: 	return re.cap(1);
Chris@0:     }
Chris@0:     return "";
Chris@0: }
Chris@0: 
Chris@0: static Work *
Chris@0: makeWork(QString composerName, QString opfield, QString numfield,
Chris@0:          int partNumber, QString titlefield, QString datefield,
Chris@0:          QString placefield, QString remarksfield, Work *main)
Chris@0: {
Chris@0:     if (titlefield.contains("List of ") || titlefield.contains("http:")) return 0;
Chris@0: 
Chris@0:     QString linkText;
Chris@0: 
Chris@0:     Work *w = new Work;
Chris@0: 
Chris@0:     QRegExp embeddedOpMatcher("([Oo]pus|[Oo]p.|WAB) (posth[a-z\\.]* *)?([0-9][^ ;:,]*)(,? *([Nn]umber|[Nn]o.|[Nn]r.) ([0-9][^ ;:,]*))?,?");
Chris@0:     if (embeddedOpMatcher.indexIn(titlefield) >= 0) {
Chris@0:         QString opf = embeddedOpMatcher.cap(0);
Chris@0:         if (opfield == "") opfield = opf;
Chris@0:         titlefield.replace(opf, "");
Chris@0:     } else if (embeddedOpMatcher.indexIn(remarksfield) >= 0) {
Chris@0:         opfield = embeddedOpMatcher.cap(0);
Chris@0:     }
Chris@0:     if (main && numfield == "") {
Chris@0:         QRegExp embeddedNumMatcher("(Number|No.|Nr.) ([0-9][^ ;:,]*)");
Chris@0:         if (embeddedNumMatcher.indexIn(titlefield) >= 0) {
Chris@0:             numfield = embeddedNumMatcher.cap(2);
Chris@0:         } else if (embeddedNumMatcher.indexIn(remarksfield) >= 0) {
Chris@0:             numfield = embeddedNumMatcher.cap(2);
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     QString op = sanitise(opfield, linkText);
Chris@0:     if (op != "") {
Chris@0:         if (op.toLower().contains("op")) {
Chris@0:             op.replace("Opus ", "");
Chris@0:             op.replace("Op. ", "");
Chris@0:             op.replace("Op.", "");
Chris@0:             op.replace("Op ", "");
Chris@0:             op.replace("opus ", "");
Chris@0:             op.replace("op. ", "");
Chris@0:             op.replace("op.", "");
Chris@0:             op.replace("op ", "");
Chris@0:             w->setOpus(op);
Chris@0:         } else if (QRegExp("^[0-9]*$").indexIn(op) >= 0) {
Chris@0:             w->setOpus(op);
Chris@0:         } else {
Chris@0:             w->setCatalogue(op);
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     QString num = sanitise(numfield, linkText);
Chris@0:     if (num != "") {
Chris@0:         num.replace("No. ", "");
Chris@0:         num.replace("No ", "");
Chris@0:         w->setNumber(num);
Chris@0:     } else if (partNumber > 0) {
Chris@0:         w->setNumber(QString("%1").arg(partNumber));
Chris@0:     }
Chris@0: 
Chris@0:     QString key = extractKey(titlefield);
Chris@0:     if (key != "") {
Chris@0:         w->setKey(key);
Chris@0:     }
Chris@0: 
Chris@0:     DEBUG << "title before sanitise: " << titlefield << endl;
Chris@0: 
Chris@0:     remarksfield = remarksfield.trimmed();
Chris@0: 
Chris@0:     QString title = sanitise(titlefield, linkText);
Chris@0:     title.replace(QRegExp(", which.*$"), "");
Chris@0:     if (linkText != "") {
Chris@0:         if (remarksfield == "" && title.startsWith(linkText)) {
Chris@0:             remarksfield = title.right(title.length() - linkText.length());
Chris@0:             title = linkText;
Chris@0:         }
Chris@0: 	linkText.replace(" ", "_");
Chris@0: 	QUrl url;
Chris@0: 	url.setScheme("http");
Chris@0: 	url.setHost("en.wikipedia.org");
Chris@0: 	url.setPath("/wiki/" + QUrl::toPercentEncoding(linkText));
Chris@0: 	Document *d = new Document;
Chris@18: 	d->setUri(Uri(url));
Chris@0: 	d->setSiteName("Wikipedia");
Chris@0: 	d->setTopic(w);
Chris@0: 	w->addPage(d);
Chris@0:     }
Chris@0: 
Chris@0:     DEBUG << "title after sanitise: " << title << ", link text " << linkText << ", remarks " << remarksfield << endl;
Chris@0: 
Chris@0:     QRegExp explicationRE("^(\"[^-]+\") - (.+)$");
Chris@0:     int pos;
Chris@0:     if ((pos = explicationRE.indexIn(title)) >= 0) {
Chris@0:         QString part = explicationRE.cap(2);
Chris@0:         if (part[0].isUpper()) w->addAlias(explicationRE.cap(2));
Chris@0:         else if (remarksfield == "") remarksfield = explicationRE.cap(2);
Chris@0:         title = explicationRE.cap(1);
Chris@0:     }
Chris@0: 
Chris@0:     QRegExp remarksRE1("^(\"[^-]+\") (for .*)$");
Chris@0:     if ((pos = remarksRE1.indexIn(title)) >= 0) {
Chris@0:         if (remarksfield != "") {
Chris@0:             remarksfield = QString("%1 - %2")
Chris@0:                 .arg(remarksRE1.cap(2)).arg(remarksfield);
Chris@0:         } else {
Chris@0:             remarksfield = remarksRE1.cap(2);
Chris@0:         }
Chris@0:         title = remarksRE1.cap(1);
Chris@0:     }
Chris@0:     
Chris@0:     QRegExp remarksRE2("^(\"[^\"]+\"), (.*)$");
Chris@0:     if ((pos = remarksRE2.indexIn(title)) >= 0) {
Chris@0:         if (remarksfield != "") {
Chris@0:             remarksfield = QString("%1 - %2")
Chris@0:                 .arg(remarksRE2.cap(2)).arg(remarksfield);
Chris@0:         } else {
Chris@0:             remarksfield = remarksRE2.cap(2);
Chris@0:         }
Chris@0:         title = remarksRE2.cap(1);
Chris@0:     }
Chris@0: 
Chris@0:     QRegExp explicationRE2("^([^\\(]*\") \\(([^\\)]*)\\)(.*)$");
Chris@0:     if ((pos = explicationRE2.indexIn(title)) >= 0) {
Chris@0:         w->addAlias(explicationRE2.cap(2));
Chris@0:         if (remarksfield == "") remarksfield = explicationRE2.cap(3);
Chris@0:         title = explicationRE2.cap(1);
Chris@0:     }
Chris@0: 
Chris@0:     if (title.startsWith("Song \"")) {
Chris@0:         title = title.right(title.length() - 5);
Chris@0:         w->addForm(Form::getFormByName("song"));
Chris@0:     }
Chris@0:     if (!main && title.startsWith("Song cycle \"")) {
Chris@0:         title = title.right(title.length() - 11);
Chris@0:         w->addForm(Form::getFormByName("song cycle"));
Chris@0:     }
Chris@0:     if (main && main->forms().contains(Form::getFormByName("song cycle"))) {
Chris@0:         w->addForm(Form::getFormByName("song"));
Chris@0:     }
Chris@0: 
Chris@0:     if (title == "" && !main) {
Chris@0:         delete w;
Chris@0:         return 0;
Chris@0:     }
Chris@0: 
Chris@0:     w->setName(title);
Chris@0:     
Chris@0:     QString remarks = sanitise(remarksfield, linkText);
Chris@0:     if (remarks != "") {
Chris@0:         w->setRemarks(remarks);
Chris@0:     }
Chris@0: 
Chris@0:     QString year = extractYear(datefield);
Chris@0:     QString place = sanitise(placefield, linkText);
Chris@0: 
Chris@0:     DEBUG << "title = " << title << endl;
Chris@0: 
Chris@0:     if (main) {
Chris@0:         main->addPart(w);
Chris@0:         w->setPartOf(main);
Chris@0:         w->setComposition(main->composition());
Chris@0:         main->composition()->addWork(w);
Chris@0:     }
Chris@0: 
Chris@0:     if (!main || !main->composition() ||
Chris@0:         (year != "" && (main->composition()->year() != year.toInt()))) {
Chris@0:         Composition *c = new Composition;
Chris@0:         c->setComposerName(composerName);
Chris@0:         c->addWork(w);
Chris@0:         c->setYear(year.toInt());
Chris@0:         c->setPlace(place);
Chris@0:         w->setComposition(c);
Chris@0:     }
Chris@0: 
Chris@0:     return w;
Chris@0: }
Chris@0: 
Chris@0: 
Chris@0: void
Chris@0: WikipediaWorksListImporter::import(QUrl source)
Chris@0: {
Chris@0:     //!!! for now
Chris@0:     QString filename = source.toLocalFile();
Chris@0: 
Chris@0:     QFile file(filename);
Chris@0:     if (!file.open(QFile::ReadOnly | QFile::Text)) {
Chris@0: 	throw std::exception();
Chris@0:     }
Chris@0: 
Chris@0:     QTextStream stream(&file);
Chris@0:     stream.setCodec("UTF-8");
Chris@0:     
Chris@0:     QString composerName;
Chris@0:     if (filename.contains("K%C3%B6chel")) {
Chris@0: 	composerName = "Wolfgang Amadeus Mozart";
Chris@0:     } else if (filename.contains("/Schubert_")) {
Chris@0:         composerName = "Franz Schubert";
Chris@0:     } else {
Chris@0: 	QRegExp byby("by_(.*)_by");
Chris@0: 	if (byby.indexIn(filename) >= 0) {
Chris@0: 	    composerName = byby.cap(1).replace('_', ' ');
Chris@0: 	} else {
Chris@0:             QRegExp bybr("by_(.*)_\\(");
Chris@0:             if (bybr.indexIn(filename) >= 0) {
Chris@0:                 composerName = bybr.cap(1).replace('_', ' ');
Chris@0:             } else {
Chris@0:                 QRegExp by("by_(.*)");
Chris@0:                 if (by.indexIn(filename) >= 0) {
Chris@0:                     composerName = by.cap(1).replace('_', ' ');
Chris@0:                 } else {
Chris@0:                     QRegExp of("of_([A-Z].*)");
Chris@0:                     if (of.indexIn(filename) >= 0) {
Chris@0:                         composerName = of.cap(1).replace('_', ' ');
Chris@0:                     }
Chris@0:                 }
Chris@0:             }
Chris@0: 	}
Chris@0:     }
Chris@0:     composerName = QUrl::fromPercentEncoding(composerName.toLocal8Bit());
Chris@0: 
Chris@0:     DEBUG << "composerName = " << composerName << endl;
Chris@0: 
Chris@0: 
Chris@0:     // We try to keep these matchers specific enough that we can be
Chris@0:     // sure the title field will come out containing _at least_ the
Chris@0:     // title.  i.e. the title field should never end up with just the
Chris@0:     // opus number or date or whatever, even if the line is formatted
Chris@0:     // in a way we hadn't anticipated.  Thus it helps if the title is
Chris@0:     // bookended by '' or [[]], etc
Chris@0: 
Chris@0:     // e.g. Beethoven
Chris@0:     // *Opus 84: ''[[Egmont (Beethoven)|Egmont]]'', overture and incidental music (1810)
Chris@0:     // opus field - n/a - title - date - n/a - remarks
Chris@0:     QRegExp workMatcher1("^\\* *(([Oo]pus|[Oo]p\\.|WoO|Anh|H|D|G) *[0-9][^ ,:{]*)[:,] *(.*) *\\([^\\)]*([0-9]{4}(-[0-9]+)*)[^0-9\\)]*\\) *(.*)$");
Chris@0: 
Chris@0:     // e.g. Tchaikovsky
Chris@0:     // *'''Op. 19''' 6 Pieces, for piano (1873)
Chris@0:     // or Ravel
Chris@0:     // * '''1''', Piano Sonata movement (1888), lost
Chris@0: 
Chris@0: /*
Chris@0:     // opus field - n/a - title - date - n/a - remarks
Chris@0:     QRegExp workMatcher1a("^\\* *'''(([Oo]pus|[Oo]p\\.|WoO|Anh|H|D|G)? *[0-9][^ ,:'{]*)'''[:, ] *(.*) *\\([^\\)]*([0-9]{4}(-[0-9]+)*)[^0-9\\)]*\\) *(.*)$");
Chris@0: */
Chris@0:     // opus field - n/a - title
Chris@0:     QRegExp workMatcher1a("^\\* *'''(([Oo]pus|[Oo]p\\.|WoO|Anh|[A-Z]{1,2})?\\.? *[0-9][^ ,:'{]*),?'''[:, ] *(.*)$");
Chris@0: 
Chris@0:     // e.g. Copland
Chris@0:     // * ''Four Motets'' for mixed voices (1921)
Chris@0:     // title - date field
Chris@0:     // (no opus)
Chris@0:     QRegExp workMatcher2("^\\* *(''.*''\\)?) *(.*)$");
Chris@0:     workMatcher2.setMinimal(true); // avoid matching multiple ''...'' substrings
Chris@0: 
Chris@0:     // e.g. Copland
Chris@0:     // * Arrangement of ''Lincoln Portrait'' for concert band (1942)
Chris@0:     // or Mendelssohn
Chris@0:     // * [[Christe du Lamm Gottes]] (1827), SATB, strings
Chris@0:     // title - date field - remarks
Chris@0:     // (no opus)
Chris@0:     QRegExp workMatcher3("^\\* *([^\\*].*) *\\(([^\\)]*[0-9]{4}[^\\)]*)\\) *(.*)$");
Chris@0:     
Chris@0:     // e.g. Scriabin
Chris@0:     // *[[Sonata No. 2 (Scriabin)|Sonata No. 2 in G sharp minor]], Op. 19  (also known as ''Sonata-Fantasy'')"
Chris@0:     // title - opus field - n/a - remarks
Chris@0:     QRegExp workMatcher4("^\\* *(\\[\\[.*\\]\\]),* (([Oo]pus|[Oo]p\\.|WoO|Anh|H|D|G) *[0-9][^ ,:'{]*) *(.*)$");
Chris@0: 
Chris@0:     // e.g. Scriabin
Chris@0:     // *Opus 35: [[Opus 35 (Scriabin)|Three Preludes]]
Chris@0:     // opus field - n/a - title - remarks
Chris@0:     QRegExp workMatcher5("^\\* *(([Oo]pus|[Oo]p\\.|WoO|Anh|H|D|G) *[0-9][^ ,:'{]*)[:,]* *([\\[']+.*[\\]']+) *(.*)$");
Chris@0: 
Chris@0:     // e.g. Boccherini
Chris@0:     // *G 1: Cello Sonata in F major
Chris@0:     // or weird Schubert layout
Chris@0:     // * D 505{{nbsp|4}}Adagio in D-flat for Piano
Chris@0:     // or Glazunov
Chris@0:     // :Op. 67: ''[[The Seasons (ballet)|The Seasons]]'', ballet in one act (1900)
Chris@0:     // or even
Chris@0:     // ::Op. 77: ''[[Symphony No. 7 (Glazunov)|Symphony No. 7]]'' &quot;Pastorale&quot; in F major (1902-1903)
Chris@0:     // This one is a real mess, for really messy pages.  Needs to go near
Chris@0:     // the end of the matchers in case it catches something it shouldn't
Chris@0:     // n/a - opus field - n/a - n/a - n/a - title
Chris@0:     QRegExp workMatcher6("^([\\*:]|::) *(([Oo]pus|[Oo]p\\.|WoO|Anh|H|D|G) *[0-9][^ ,:'{]*)(([:,]| *\\{+[^\\}]+\\}+) *(.*))?$");
Chris@0: 
Chris@0:     // e.g. Bruch
Chris@0:     // * Adagio appassionato for violin and orchestra in C sharp minor, Op. 57
Chris@0:     // title - opus field - date field
Chris@0:     QRegExp workMatcher7("^\\* *(.*),? (([Oo]pus|[Oo]p\\.|WoO|Anh|H|D|G) *[0-9][^ ,:'{]*|[Oo]p. posth[a-z.]*) *(\\([^\\)]*([0-9]{4}(-[0-9]+)*)[^0-9\\)]*\\))? *$");
Chris@0: 
Chris@0:     // e.g. Bruckner
Chris@0:     // * Symphony No. 0 in D minor 1869 WAB 100
Chris@0:     // title - date field - opus field
Chris@0:     QRegExp workMatcher8("^\\* *(.*) ([0-9]{4}[0-9/-]*) *(WAB [0-9][^ ]*)$");
Chris@0: 
Chris@0:     // e.g. Bach
Chris@0:     // * BWV 506 ? Was bist du doch, o Seele, so betruebet
Chris@0:     // opus field - title
Chris@0:     QRegExp workMatcher9("^\\* *(BWV [^ ]+)(.*)$");
Chris@0: 
Chris@0:     // Catch-all for things that look at all promising (anything that
Chris@0:     // starts with ' or [ after bullet: take the whole as title)
Chris@0:     QRegExp workMatcher10("^[\\*:] *((['\\[]|&quot;).*)$");
Chris@0: 
Chris@0: 
Chris@0: 
Chris@0:     // e.g. Beethoven
Chris@0:     // **No. 1: [[Piano Trio No. 1 (Beethoven)|Piano Trio No. 1]] in E-flat major
Chris@0:     // number field - n/a - title, remarks etc
Chris@0:     QRegExp partMatcher1("^[\\*:]{2} *((No\\.? *)?[0-9][^ ,:'{]*)[:, ] *(.*)$");
Chris@0: 
Chris@0:     // e.g. Copland
Chris@0:     // ** ''Help us, O Lord''
Chris@0:     // title - remarks
Chris@0:     QRegExp partMatcher2("^\\*\\* *(''.*'') *(.*)$");
Chris@0:     partMatcher2.setMinimal(true); // avoid matching multiple ''...'' substrings
Chris@0: 
Chris@0:     // e.g. Scriabin
Chris@0:     // **[[Mazurka Op. 40 No. 1 (Scriabin)|Mazurka in D flat major]]
Chris@0:     // title - remarks
Chris@0:     QRegExp partMatcher3("^\\*\\* *(\\[\\[.*\\]\\])(.*)$");
Chris@0: 
Chris@0:     // e.g. Berlioz
Chris@0:     // ** 1: ''Méditation religieuse''
Chris@0:     // number - title - remarks
Chris@0:     QRegExp partMatcher4("^\\*\\* *([0-9][0-9a-z]*)[\\.: ] *([\\[]*''.*''[\\]]*) *(.*)$");
Chris@0: 
Chris@0:     // e.g. Tchaikovsky
Chris@0:     // **4. Nocturne [???????] (C? minor)
Chris@0:     // number - title - remarks
Chris@0:     QRegExp partMatcher5("^\\*\\* *([0-9][0-9a-z]*)[\\.: ] *(.*\\[[^\\]]+\\])(.*)$");
Chris@0: 
Chris@0:     // e.g. Schubert
Chris@0:     // **2. &quot;Wohin?&quot;
Chris@0:     // n/a - number - title
Chris@0:     QRegExp partMatcher6("^\\*\\* *(([0-9][0-9a-z]*)[\\.:])? *((&quot;|'').*)$");
Chris@0:    
Chris@0:     // e.g. Mendelssohn
Chris@0:     // ** Notturno
Chris@0:     // title only
Chris@0:     QRegExp partMatcher7("^\\*\\* *(.*)$");
Chris@0: 
Chris@0: 
Chris@0:     // Date and remarks within titlefield or remarksfield
Chris@0:     QRegExp matcherDate("\\([^\\)]*([0-9]{4})[^0-9\\)]*\\),?(.*)");
Chris@0: 
Chris@0:     
Chris@0:     Work *main = 0;
Chris@0:     int partNumber = 0;
Chris@0: 
Chris@0:     QString line;
Chris@0:     QString opfield, numfield, titlefield, remarksfield, datefield;
Chris@0: 
Chris@0:     while (!stream.atEnd()) {
Chris@0: 
Chris@0:         if (line == "") {
Chris@0:             line = stream.readLine();
Chris@0:             DEBUG << "line: " << line << endl;
Chris@0:         }
Chris@0: 
Chris@0:         opfield = "";
Chris@0:         numfield = "";
Chris@0:         titlefield = "";
Chris@0:         datefield = "";
Chris@0:         remarksfield = "";
Chris@0:         partNumber = 0;
Chris@0: 
Chris@0:         if (workMatcher1.indexIn(line) >= 0) {
Chris@0: 
Chris@0:             DEBUG << "matcher 1" << endl;
Chris@0:             opfield = workMatcher1.cap(1);
Chris@0:             titlefield = workMatcher1.cap(3);
Chris@0:             datefield = workMatcher1.cap(4);
Chris@0:             remarksfield = workMatcher1.cap(6);
Chris@0: 
Chris@0:         } else if (workMatcher1a.indexIn(line) >= 0) {
Chris@0: 
Chris@0:             DEBUG << "matcher 1a" << endl;
Chris@0:             opfield = workMatcher1a.cap(1);
Chris@0:             titlefield = workMatcher1a.cap(3);
Chris@0: /*
Chris@0:             datefield = workMatcher1a.cap(4);
Chris@0:             remarksfield = workMatcher1a.cap(6);
Chris@0: */
Chris@0: 
Chris@0:         } else if (workMatcher2.indexIn(line) >= 0) {
Chris@0: 
Chris@0:             DEBUG << "matcher 2" << endl;
Chris@0:             titlefield = workMatcher2.cap(1);
Chris@0:             remarksfield = workMatcher2.cap(2);
Chris@0:             
Chris@0:         } else if (workMatcher3.indexIn(line) >= 0) {
Chris@0: 
Chris@0:             DEBUG << "matcher 3" << endl;
Chris@0:             titlefield = workMatcher3.cap(1);
Chris@0:             datefield = workMatcher3.cap(2);
Chris@0:             remarksfield = workMatcher3.cap(3);
Chris@0: 
Chris@0:         } else if (workMatcher4.indexIn(line) >= 0) {
Chris@0: 
Chris@0:             DEBUG << "matcher 4" << endl;
Chris@0:             titlefield = workMatcher4.cap(1);
Chris@0:             opfield = workMatcher4.cap(2);
Chris@0:             remarksfield = workMatcher4.cap(4);
Chris@0: 
Chris@0:         } else if (workMatcher5.indexIn(line) >= 0) {
Chris@0: 
Chris@0:             DEBUG << "matcher 5" << endl;
Chris@0:             opfield = workMatcher5.cap(1);
Chris@0:             titlefield = workMatcher5.cap(3);
Chris@0:             remarksfield = workMatcher5.cap(4);
Chris@0: 
Chris@0:         } else if (workMatcher6.indexIn(line) >= 0) {
Chris@0: 
Chris@0:             DEBUG << "matcher 6" << endl;
Chris@0:             opfield = workMatcher6.cap(2);
Chris@0:             titlefield = workMatcher6.cap(6);
Chris@0: 
Chris@0:         } else if (workMatcher7.indexIn(line) >= 0) {
Chris@0: 
Chris@0:             DEBUG << "matcher 7" << endl;
Chris@0:             titlefield = workMatcher7.cap(1);
Chris@0:             opfield = workMatcher7.cap(2);
Chris@0:             datefield = workMatcher7.cap(3);
Chris@0: 
Chris@0:         } else if (workMatcher8.indexIn(line) >= 0) {
Chris@0: 
Chris@0:             DEBUG << "matcher 8" << endl;
Chris@0:             titlefield = workMatcher8.cap(1);
Chris@0:             datefield = workMatcher8.cap(2);
Chris@0:             opfield = workMatcher8.cap(3);
Chris@0: 
Chris@0:         } else if (workMatcher9.indexIn(line) >= 0) {
Chris@0: 
Chris@0:             DEBUG << "matcher 9" << endl;
Chris@0:             opfield = workMatcher9.cap(1);
Chris@0:             titlefield = workMatcher9.cap(2);
Chris@0: 
Chris@0:         } else if (workMatcher10.indexIn(line) >= 0) {
Chris@0: 
Chris@0:             DEBUG << "matcher 10" << endl;
Chris@0:             titlefield = workMatcher10.cap(1);
Chris@0: 
Chris@0:         } else {
Chris@0:             if (line.startsWith("*") || line.startsWith(":")) {
Chris@0:                 DEBUG << "Failed to match promising works list line: " << line << endl;
Chris@0:             }
Chris@0:             line = "";
Chris@0:             continue;
Chris@0:         }
Chris@0: 
Chris@0:         if (titlefield != "" && datefield == "") {
Chris@0:             int dpos;
Chris@0:             if ((dpos = matcherDate.indexIn(titlefield)) != -1) {
Chris@0:                 datefield = matcherDate.cap(1);
Chris@0:                 remarksfield = matcherDate.cap(2);
Chris@0:                 titlefield = titlefield.left(dpos);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         if (remarksfield != "" && datefield == "") {
Chris@0:             int dpos;
Chris@0:             if ((dpos = matcherDate.indexIn(remarksfield)) != -1) {
Chris@0:                 datefield = matcherDate.cap(1);
Chris@0:                 remarksfield = remarksfield.left(dpos);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         main = makeWork(composerName, opfield, "", 0,
Chris@0:                         titlefield, datefield, "", remarksfield, 0);
Chris@0: 
Chris@0:         if (main) m_objects.push_back(main);
Chris@0: 
Chris@0:         line = "";
Chris@0: 
Chris@0:         while (!stream.atEnd()) {
Chris@0:             
Chris@0:             ++partNumber;
Chris@0:             line = stream.readLine();
Chris@0:             DEBUG << "line: " << line << endl;
Chris@0:             
Chris@0:             if (partMatcher1.indexIn(line) >= 0) {
Chris@0:                 
Chris@0:                 DEBUG << "part matcher 1" << endl;
Chris@0:                 numfield = partMatcher1.cap(1);
Chris@0:                 titlefield = partMatcher1.cap(3);
Chris@0:                 remarksfield = "";
Chris@0: 
Chris@0:             } else if (partMatcher2.indexIn(line) >= 0) {
Chris@0:                 
Chris@0:                 DEBUG << "part matcher 2" << endl;
Chris@0:                 titlefield = partMatcher2.cap(1);
Chris@0:                 remarksfield = partMatcher2.cap(2);
Chris@0: 
Chris@0:             } else if (partMatcher3.indexIn(line) >= 0) {
Chris@0:                 
Chris@0:                 DEBUG << "part matcher 3" << endl;
Chris@0:                 titlefield = partMatcher3.cap(1);
Chris@0:                 remarksfield = partMatcher3.cap(2);
Chris@0: 
Chris@0:             } else if (partMatcher4.indexIn(line) >= 0) {
Chris@0:                 
Chris@0:                 DEBUG << "part matcher 4" << endl;
Chris@0:                 numfield = partMatcher4.cap(1);
Chris@0:                 titlefield = partMatcher4.cap(2);
Chris@0:                 remarksfield = partMatcher4.cap(3);
Chris@0: 
Chris@0:             } else if (partMatcher5.indexIn(line) >= 0) {
Chris@0:                 
Chris@0:                 DEBUG << "part matcher 5" << endl;
Chris@0:                 numfield = partMatcher5.cap(1);
Chris@0:                 titlefield = partMatcher5.cap(2);
Chris@0:                 remarksfield = partMatcher5.cap(3);
Chris@0: 
Chris@0:             } else if (partMatcher6.indexIn(line) >= 0) {
Chris@0:                 
Chris@0:                 DEBUG << "part matcher 6" << endl;
Chris@0:                 numfield = partMatcher6.cap(2);
Chris@0:                 titlefield = partMatcher6.cap(3);
Chris@0: 
Chris@0:             } else if (partMatcher7.indexIn(line) >= 0) {
Chris@0:                 
Chris@0:                 DEBUG << "part matcher 7" << endl;
Chris@0:                 titlefield = partMatcher7.cap(1);
Chris@0: 
Chris@0:             } else {
Chris@0:                 if (line.startsWith("**") || line.startsWith("::")) {
Chris@0:                     DEBUG << "Failed to match promising part line: " << line << endl;
Chris@0:                 }
Chris@0:                 break;
Chris@0:             }
Chris@0: 
Chris@0:             if (titlefield != "" && datefield == "") {
Chris@0:                 int dpos;
Chris@0:                 if ((dpos = matcherDate.indexIn(titlefield)) != -1) {
Chris@0:                     datefield = matcherDate.cap(1);
Chris@0:                     remarksfield = matcherDate.cap(2);
Chris@0:                     titlefield = titlefield.left(dpos);
Chris@0:                 }
Chris@0:             }
Chris@0: 
Chris@0:             Work *part = makeWork(composerName, opfield, numfield, partNumber,
Chris@0:                                   titlefield, datefield, "", remarksfield,
Chris@0:                                   main);
Chris@0: 
Chris@0:             if (part) m_objects.push_back(part);
Chris@0:         }
Chris@0:     }        
Chris@0: 
Chris@0:     DEBUG << "Found " << m_objects.size() << " things" << endl;
Chris@0: }
Chris@0: 
Chris@0: 
Chris@0: }
Chris@0: 
Chris@0: