To check out this repository please hg clone the following URL, or open the URL using EasyMercurial or your preferred Mercurial client.
root / import / ImportWikipediaWorksList.cpp
History | View | Annotate | Download (21.3 KB)
| 1 |
/* -*- c-basic-offset: 4 indent-tabs-mode: nil -*- vi:set ts=8 sts=4 sw=4: */
|
|---|---|
| 2 |
|
| 3 |
#include "ImportWikipediaWorksList.h" |
| 4 |
|
| 5 |
#include <dataquay/Debug.h> |
| 6 |
|
| 7 |
#include <QFile> |
| 8 |
#include <QFileInfo> |
| 9 |
#include <QTextStream> |
| 10 |
#include <QRegExp> |
| 11 |
#include <QVariant> |
| 12 |
|
| 13 |
#include <exception> |
| 14 |
|
| 15 |
using namespace Dataquay; |
| 16 |
|
| 17 |
namespace ClassicalData {
|
| 18 |
|
| 19 |
void
|
| 20 |
WikipediaWorksListImporter::setSource(QUrl source) |
| 21 |
{
|
| 22 |
DEBUG << "WikipediaWorksListImporter::setSource: " << source << endl;
|
| 23 |
import(source); |
| 24 |
} |
| 25 |
|
| 26 |
static QString
|
| 27 |
sanitise(QString field, QString &linkText) |
| 28 |
{
|
| 29 |
int mp;
|
| 30 |
|
| 31 |
field.replace(QString::fromUtf8("\342\200\222"), "-"); |
| 32 |
field.replace(QString::fromUtf8("\342\200\223"), "-"); |
| 33 |
field.replace(QString::fromUtf8("\342\200\224"), "-"); |
| 34 |
field.replace(QString::fromUtf8("\342\200\225"), "-"); |
| 35 |
|
| 36 |
field.replace(QString::fromUtf8("\342\231\255"), "-flat"); |
| 37 |
field.replace(QString::fromUtf8("\342\231\257"), "-sharp"); |
| 38 |
|
| 39 |
QRegExp link2("([^A-Za-z]*)\\[\\[([^\\]\\|]+)\\|([^\\]]+)\\]\\]");
|
| 40 |
if ((mp = link2.indexIn(field)) >= 0) { |
| 41 |
if (linkText == "" && mp < 4) linkText = link2.cap(2); |
| 42 |
field.replace(mp, link2.matchedLength(), link2.cap(1) + link2.cap(3)); |
| 43 |
return sanitise(field, linkText);
|
| 44 |
} |
| 45 |
|
| 46 |
QRegExp link1("^([^A-Za-z]*)\\[\\[([^\\]]+)\\]\\]");
|
| 47 |
if ((mp = link1.indexIn(field)) >= 0) { |
| 48 |
if (linkText == "") linkText = link1.cap(2); |
| 49 |
field.replace(mp, link1.matchedLength(), link1.cap(1) + link1.cap(2)); |
| 50 |
return sanitise(field, linkText);
|
| 51 |
} |
| 52 |
|
| 53 |
field = field.trimmed(); |
| 54 |
|
| 55 |
field.replace("[", ""); |
| 56 |
field.replace("]", ""); |
| 57 |
field.replace(QRegExp("\\{+[^\\}]*\\}+ *"), " "); |
| 58 |
field.replace("'''", "\""); |
| 59 |
field.replace("''", "\""); |
| 60 |
field.replace(""", "\""); |
| 61 |
field.replace("\"\"", "\""); |
| 62 |
field.replace(QRegExp("^[\'\"] (\")?"), "\""); |
| 63 |
field.replace(QRegExp("<[^&]*>"), ""); |
| 64 |
field.replace(QRegExp("^\\**"), ""); |
| 65 |
|
| 66 |
if (field.endsWith("c.")) { |
| 67 |
// historical artifact from removal of Bruckner year indication (c. 1856)
|
| 68 |
field = field.left(field.length()-2);
|
| 69 |
} |
| 70 |
|
| 71 |
while (field.endsWith(".") || field.endsWith(",")) { |
| 72 |
field = field.left(field.length()-1);
|
| 73 |
} |
| 74 |
|
| 75 |
if (field.startsWith(";") || field.startsWith(":") || field.startsWith(",") |
| 76 |
|| field.startsWith("-")) {
|
| 77 |
field = field.right(field.length()-1);
|
| 78 |
} |
| 79 |
|
| 80 |
if (field.startsWith("(") && field.endsWith(")")) { |
| 81 |
DEBUG << "before: " << field;
|
| 82 |
field = field.mid(1, field.length()-2); |
| 83 |
DEBUG << "after: " << field;
|
| 84 |
} |
| 85 |
|
| 86 |
field.replace(QRegExp("^\\**"), ""); |
| 87 |
if (field == ")" || field == "(") { |
| 88 |
field = "";
|
| 89 |
} |
| 90 |
|
| 91 |
field.replace(" - ,", ","); |
| 92 |
field.replace(" ", " "); |
| 93 |
|
| 94 |
return field.trimmed();
|
| 95 |
} |
| 96 |
|
| 97 |
static QString
|
| 98 |
extractYear(QString datefield) |
| 99 |
{
|
| 100 |
QRegExp re("[0-9]{4}");
|
| 101 |
if (re.indexIn(datefield) >= 0) { |
| 102 |
return re.cap(0); |
| 103 |
} |
| 104 |
return ""; |
| 105 |
} |
| 106 |
|
| 107 |
static QString
|
| 108 |
extractKey(QString titlefield) |
| 109 |
{
|
| 110 |
QRegExp re("in ([A-H]([ -][a-z]+)? (major|minor))");
|
| 111 |
if (re.indexIn(titlefield) >= 0) { |
| 112 |
return re.cap(1); |
| 113 |
} |
| 114 |
return ""; |
| 115 |
} |
| 116 |
|
| 117 |
static Work *
|
| 118 |
makeWork(QString composerName, QString opfield, QString numfield, |
| 119 |
int partNumber, QString titlefield, QString datefield,
|
| 120 |
QString placefield, QString remarksfield, Work *main) |
| 121 |
{
|
| 122 |
if (titlefield.contains("List of ") || titlefield.contains("http:")) return 0; |
| 123 |
|
| 124 |
QString linkText; |
| 125 |
|
| 126 |
Work *w = new Work;
|
| 127 |
|
| 128 |
QRegExp embeddedOpMatcher("([Oo]pus|[Oo]p.|WAB) (posth[a-z\\.]* *)?([0-9][^ ;:,]*)(,? *([Nn]umber|[Nn]o.|[Nn]r.) ([0-9][^ ;:,]*))?,?");
|
| 129 |
if (embeddedOpMatcher.indexIn(titlefield) >= 0) { |
| 130 |
QString opf = embeddedOpMatcher.cap(0);
|
| 131 |
if (opfield == "") opfield = opf; |
| 132 |
titlefield.replace(opf, "");
|
| 133 |
} else if (embeddedOpMatcher.indexIn(remarksfield) >= 0) { |
| 134 |
opfield = embeddedOpMatcher.cap(0);
|
| 135 |
} |
| 136 |
if (main && numfield == "") { |
| 137 |
QRegExp embeddedNumMatcher("(Number|No.|Nr.) ([0-9][^ ;:,]*)");
|
| 138 |
if (embeddedNumMatcher.indexIn(titlefield) >= 0) { |
| 139 |
numfield = embeddedNumMatcher.cap(2);
|
| 140 |
} else if (embeddedNumMatcher.indexIn(remarksfield) >= 0) { |
| 141 |
numfield = embeddedNumMatcher.cap(2);
|
| 142 |
} |
| 143 |
} |
| 144 |
|
| 145 |
QString op = sanitise(opfield, linkText); |
| 146 |
if (op != "") { |
| 147 |
if (op.toLower().contains("op")) { |
| 148 |
op.replace("Opus ", ""); |
| 149 |
op.replace("Op. ", ""); |
| 150 |
op.replace("Op.", ""); |
| 151 |
op.replace("Op ", ""); |
| 152 |
op.replace("opus ", ""); |
| 153 |
op.replace("op. ", ""); |
| 154 |
op.replace("op.", ""); |
| 155 |
op.replace("op ", ""); |
| 156 |
w->setOpus(op); |
| 157 |
} else if (QRegExp("^[0-9]*$").indexIn(op) >= 0) { |
| 158 |
w->setOpus(op); |
| 159 |
} else {
|
| 160 |
w->setCatalogue(op); |
| 161 |
} |
| 162 |
} |
| 163 |
|
| 164 |
QString num = sanitise(numfield, linkText); |
| 165 |
if (num != "") { |
| 166 |
num.replace("No. ", ""); |
| 167 |
num.replace("No ", ""); |
| 168 |
w->setNumber(num); |
| 169 |
} else if (partNumber > 0) { |
| 170 |
w->setNumber(QString("%1").arg(partNumber));
|
| 171 |
} |
| 172 |
|
| 173 |
QString key = extractKey(titlefield); |
| 174 |
if (key != "") { |
| 175 |
w->setKey(key); |
| 176 |
} |
| 177 |
|
| 178 |
DEBUG << "title before sanitise: " << titlefield << endl;
|
| 179 |
|
| 180 |
remarksfield = remarksfield.trimmed(); |
| 181 |
|
| 182 |
QString title = sanitise(titlefield, linkText); |
| 183 |
title.replace(QRegExp(", which.*$"), ""); |
| 184 |
if (linkText != "") { |
| 185 |
if (remarksfield == "" && title.startsWith(linkText)) { |
| 186 |
remarksfield = title.right(title.length() - linkText.length()); |
| 187 |
title = linkText; |
| 188 |
} |
| 189 |
linkText.replace(" ", "_"); |
| 190 |
QUrl url; |
| 191 |
url.setScheme("http");
|
| 192 |
url.setHost("en.wikipedia.org");
|
| 193 |
url.setPath("/wiki/" + QUrl::toPercentEncoding(linkText));
|
| 194 |
Document *d = new Document;
|
| 195 |
d->setUri(Uri(url)); |
| 196 |
d->setSiteName("Wikipedia");
|
| 197 |
d->setTopic(w); |
| 198 |
w->addPage(d); |
| 199 |
} |
| 200 |
|
| 201 |
DEBUG << "title after sanitise: " << title << ", link text " << linkText << ", remarks " << remarksfield << endl; |
| 202 |
|
| 203 |
QRegExp explicationRE("^(\"[^-]+\") - (.+)$");
|
| 204 |
int pos;
|
| 205 |
if ((pos = explicationRE.indexIn(title)) >= 0) { |
| 206 |
QString part = explicationRE.cap(2);
|
| 207 |
if (part[0].isUpper()) w->addAlias(explicationRE.cap(2)); |
| 208 |
else if (remarksfield == "") remarksfield = explicationRE.cap(2); |
| 209 |
title = explicationRE.cap(1);
|
| 210 |
} |
| 211 |
|
| 212 |
QRegExp remarksRE1("^(\"[^-]+\") (for .*)$");
|
| 213 |
if ((pos = remarksRE1.indexIn(title)) >= 0) { |
| 214 |
if (remarksfield != "") { |
| 215 |
remarksfield = QString("%1 - %2")
|
| 216 |
.arg(remarksRE1.cap(2)).arg(remarksfield);
|
| 217 |
} else {
|
| 218 |
remarksfield = remarksRE1.cap(2);
|
| 219 |
} |
| 220 |
title = remarksRE1.cap(1);
|
| 221 |
} |
| 222 |
|
| 223 |
QRegExp remarksRE2("^(\"[^\"]+\"), (.*)$");
|
| 224 |
if ((pos = remarksRE2.indexIn(title)) >= 0) { |
| 225 |
if (remarksfield != "") { |
| 226 |
remarksfield = QString("%1 - %2")
|
| 227 |
.arg(remarksRE2.cap(2)).arg(remarksfield);
|
| 228 |
} else {
|
| 229 |
remarksfield = remarksRE2.cap(2);
|
| 230 |
} |
| 231 |
title = remarksRE2.cap(1);
|
| 232 |
} |
| 233 |
|
| 234 |
QRegExp explicationRE2("^([^\\(]*\") \\(([^\\)]*)\\)(.*)$");
|
| 235 |
if ((pos = explicationRE2.indexIn(title)) >= 0) { |
| 236 |
w->addAlias(explicationRE2.cap(2));
|
| 237 |
if (remarksfield == "") remarksfield = explicationRE2.cap(3); |
| 238 |
title = explicationRE2.cap(1);
|
| 239 |
} |
| 240 |
|
| 241 |
if (title.startsWith("Song \"")) { |
| 242 |
title = title.right(title.length() - 5);
|
| 243 |
w->addForm(Form::getFormByName("song"));
|
| 244 |
} |
| 245 |
if (!main && title.startsWith("Song cycle \"")) { |
| 246 |
title = title.right(title.length() - 11);
|
| 247 |
w->addForm(Form::getFormByName("song cycle"));
|
| 248 |
} |
| 249 |
if (main && main->forms().contains(Form::getFormByName("song cycle"))) { |
| 250 |
w->addForm(Form::getFormByName("song"));
|
| 251 |
} |
| 252 |
|
| 253 |
if (title == "" && !main) { |
| 254 |
delete w;
|
| 255 |
return 0; |
| 256 |
} |
| 257 |
|
| 258 |
w->setName(title); |
| 259 |
|
| 260 |
QString remarks = sanitise(remarksfield, linkText); |
| 261 |
if (remarks != "") { |
| 262 |
w->setRemarks(remarks); |
| 263 |
} |
| 264 |
|
| 265 |
QString year = extractYear(datefield); |
| 266 |
QString place = sanitise(placefield, linkText); |
| 267 |
|
| 268 |
DEBUG << "title = " << title << endl;
|
| 269 |
|
| 270 |
if (main) {
|
| 271 |
main->addPart(w); |
| 272 |
w->setPartOf(main); |
| 273 |
w->setComposition(main->composition()); |
| 274 |
main->composition()->addWork(w); |
| 275 |
} |
| 276 |
|
| 277 |
if (!main || !main->composition() ||
|
| 278 |
(year != "" && (main->composition()->year() != year.toInt()))) {
|
| 279 |
Composition *c = new Composition;
|
| 280 |
c->setComposerName(composerName); |
| 281 |
c->addWork(w); |
| 282 |
c->setYear(year.toInt()); |
| 283 |
c->setPlace(place); |
| 284 |
w->setComposition(c); |
| 285 |
} |
| 286 |
|
| 287 |
return w;
|
| 288 |
} |
| 289 |
|
| 290 |
|
| 291 |
void
|
| 292 |
WikipediaWorksListImporter::import(QUrl source) |
| 293 |
{
|
| 294 |
//!!! for now
|
| 295 |
QString filename = source.toLocalFile(); |
| 296 |
|
| 297 |
QFile file(filename); |
| 298 |
if (!file.open(QFile::ReadOnly | QFile::Text)) {
|
| 299 |
throw std::exception();
|
| 300 |
} |
| 301 |
|
| 302 |
QTextStream stream(&file); |
| 303 |
stream.setCodec("UTF-8");
|
| 304 |
|
| 305 |
QString composerName; |
| 306 |
if (filename.contains("K%C3%B6chel")) { |
| 307 |
composerName = "Wolfgang Amadeus Mozart";
|
| 308 |
} else if (filename.contains("/Schubert_")) { |
| 309 |
composerName = "Franz Schubert";
|
| 310 |
} else {
|
| 311 |
QRegExp byby("by_(.*)_by");
|
| 312 |
if (byby.indexIn(filename) >= 0) { |
| 313 |
composerName = byby.cap(1).replace('_', ' '); |
| 314 |
} else {
|
| 315 |
QRegExp bybr("by_(.*)_\\(");
|
| 316 |
if (bybr.indexIn(filename) >= 0) { |
| 317 |
composerName = bybr.cap(1).replace('_', ' '); |
| 318 |
} else {
|
| 319 |
QRegExp by("by_(.*)");
|
| 320 |
if (by.indexIn(filename) >= 0) { |
| 321 |
composerName = by.cap(1).replace('_', ' '); |
| 322 |
} else {
|
| 323 |
QRegExp of("of_([A-Z].*)");
|
| 324 |
if (of.indexIn(filename) >= 0) { |
| 325 |
composerName = of.cap(1).replace('_', ' '); |
| 326 |
} |
| 327 |
} |
| 328 |
} |
| 329 |
} |
| 330 |
} |
| 331 |
composerName = QUrl::fromPercentEncoding(composerName.toLocal8Bit()); |
| 332 |
|
| 333 |
DEBUG << "composerName = " << composerName << endl;
|
| 334 |
|
| 335 |
|
| 336 |
// We try to keep these matchers specific enough that we can be
|
| 337 |
// sure the title field will come out containing _at least_ the
|
| 338 |
// title. i.e. the title field should never end up with just the
|
| 339 |
// opus number or date or whatever, even if the line is formatted
|
| 340 |
// in a way we hadn't anticipated. Thus it helps if the title is
|
| 341 |
// bookended by '' or [[]], etc
|
| 342 |
|
| 343 |
// e.g. Beethoven
|
| 344 |
// *Opus 84: ''[[Egmont (Beethoven)|Egmont]]'', overture and incidental music (1810)
|
| 345 |
// opus field - n/a - title - date - n/a - remarks
|
| 346 |
QRegExp workMatcher1("^\\* *(([Oo]pus|[Oo]p\\.|WoO|Anh|H|D|G) *[0-9][^ ,:{]*)[:,] *(.*) *\\([^\\)]*([0-9]{4}(-[0-9]+)*)[^0-9\\)]*\\) *(.*)$");
|
| 347 |
|
| 348 |
// e.g. Tchaikovsky
|
| 349 |
// *'''Op. 19''' 6 Pieces, for piano (1873)
|
| 350 |
// or Ravel
|
| 351 |
// * '''1''', Piano Sonata movement (1888), lost
|
| 352 |
|
| 353 |
/*
|
| 354 |
// opus field - n/a - title - date - n/a - remarks
|
| 355 |
QRegExp workMatcher1a("^\\* *'''(([Oo]pus|[Oo]p\\.|WoO|Anh|H|D|G)? *[0-9][^ ,:'{]*)'''[:, ] *(.*) *\\([^\\)]*([0-9]{4}(-[0-9]+)*)[^0-9\\)]*\\) *(.*)$");
|
| 356 |
*/
|
| 357 |
// opus field - n/a - title
|
| 358 |
QRegExp workMatcher1a("^\\* *'''(([Oo]pus|[Oo]p\\.|WoO|Anh|[A-Z]{1,2})?\\.? *[0-9][^ ,:'{]*),?'''[:, ] *(.*)$");
|
| 359 |
|
| 360 |
// e.g. Copland
|
| 361 |
// * ''Four Motets'' for mixed voices (1921)
|
| 362 |
// title - date field
|
| 363 |
// (no opus)
|
| 364 |
QRegExp workMatcher2("^\\* *(''.*''\\)?) *(.*)$");
|
| 365 |
workMatcher2.setMinimal(true); // avoid matching multiple ''...'' substrings |
| 366 |
|
| 367 |
// e.g. Copland
|
| 368 |
// * Arrangement of ''Lincoln Portrait'' for concert band (1942)
|
| 369 |
// or Mendelssohn
|
| 370 |
// * [[Christe du Lamm Gottes]] (1827), SATB, strings
|
| 371 |
// title - date field - remarks
|
| 372 |
// (no opus)
|
| 373 |
QRegExp workMatcher3("^\\* *([^\\*].*) *\\(([^\\)]*[0-9]{4}[^\\)]*)\\) *(.*)$");
|
| 374 |
|
| 375 |
// e.g. Scriabin
|
| 376 |
// *[[Sonata No. 2 (Scriabin)|Sonata No. 2 in G sharp minor]], Op. 19 (also known as ''Sonata-Fantasy'')"
|
| 377 |
// title - opus field - n/a - remarks
|
| 378 |
QRegExp workMatcher4("^\\* *(\\[\\[.*\\]\\]),* (([Oo]pus|[Oo]p\\.|WoO|Anh|H|D|G) *[0-9][^ ,:'{]*) *(.*)$");
|
| 379 |
|
| 380 |
// e.g. Scriabin
|
| 381 |
// *Opus 35: [[Opus 35 (Scriabin)|Three Preludes]]
|
| 382 |
// opus field - n/a - title - remarks
|
| 383 |
QRegExp workMatcher5("^\\* *(([Oo]pus|[Oo]p\\.|WoO|Anh|H|D|G) *[0-9][^ ,:'{]*)[:,]* *([\\[']+.*[\\]']+) *(.*)$");
|
| 384 |
|
| 385 |
// e.g. Boccherini
|
| 386 |
// *G 1: Cello Sonata in F major
|
| 387 |
// or weird Schubert layout
|
| 388 |
// * D 505{{nbsp|4}}Adagio in D-flat for Piano
|
| 389 |
// or Glazunov
|
| 390 |
// :Op. 67: ''[[The Seasons (ballet)|The Seasons]]'', ballet in one act (1900)
|
| 391 |
// or even
|
| 392 |
// ::Op. 77: ''[[Symphony No. 7 (Glazunov)|Symphony No. 7]]'' "Pastorale" in F major (1902-1903)
|
| 393 |
// This one is a real mess, for really messy pages. Needs to go near
|
| 394 |
// the end of the matchers in case it catches something it shouldn't
|
| 395 |
// n/a - opus field - n/a - n/a - n/a - title
|
| 396 |
QRegExp workMatcher6("^([\\*:]|::) *(([Oo]pus|[Oo]p\\.|WoO|Anh|H|D|G) *[0-9][^ ,:'{]*)(([:,]| *\\{+[^\\}]+\\}+) *(.*))?$");
|
| 397 |
|
| 398 |
// e.g. Bruch
|
| 399 |
// * Adagio appassionato for violin and orchestra in C sharp minor, Op. 57
|
| 400 |
// title - opus field - date field
|
| 401 |
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\\)]*\\))? *$");
|
| 402 |
|
| 403 |
// e.g. Bruckner
|
| 404 |
// * Symphony No. 0 in D minor 1869 WAB 100
|
| 405 |
// title - date field - opus field
|
| 406 |
QRegExp workMatcher8("^\\* *(.*) ([0-9]{4}[0-9/-]*) *(WAB [0-9][^ ]*)$");
|
| 407 |
|
| 408 |
// e.g. Bach
|
| 409 |
// * BWV 506 ? Was bist du doch, o Seele, so betruebet
|
| 410 |
// opus field - title
|
| 411 |
QRegExp workMatcher9("^\\* *(BWV [^ ]+)(.*)$");
|
| 412 |
|
| 413 |
// Catch-all for things that look at all promising (anything that
|
| 414 |
// starts with ' or [ after bullet: take the whole as title)
|
| 415 |
QRegExp workMatcher10("^[\\*:] *((['\\[]|").*)$");
|
| 416 |
|
| 417 |
|
| 418 |
|
| 419 |
// e.g. Beethoven
|
| 420 |
// **No. 1: [[Piano Trio No. 1 (Beethoven)|Piano Trio No. 1]] in E-flat major
|
| 421 |
// number field - n/a - title, remarks etc
|
| 422 |
QRegExp partMatcher1("^[\\*:]{2} *((No\\.? *)?[0-9][^ ,:'{]*)[:, ] *(.*)$");
|
| 423 |
|
| 424 |
// e.g. Copland
|
| 425 |
// ** ''Help us, O Lord''
|
| 426 |
// title - remarks
|
| 427 |
QRegExp partMatcher2("^\\*\\* *(''.*'') *(.*)$");
|
| 428 |
partMatcher2.setMinimal(true); // avoid matching multiple ''...'' substrings |
| 429 |
|
| 430 |
// e.g. Scriabin
|
| 431 |
// **[[Mazurka Op. 40 No. 1 (Scriabin)|Mazurka in D flat major]]
|
| 432 |
// title - remarks
|
| 433 |
QRegExp partMatcher3("^\\*\\* *(\\[\\[.*\\]\\])(.*)$");
|
| 434 |
|
| 435 |
// e.g. Berlioz
|
| 436 |
// ** 1: ''Méditation religieuse''
|
| 437 |
// number - title - remarks
|
| 438 |
QRegExp partMatcher4("^\\*\\* *([0-9][0-9a-z]*)[\\.: ] *([\\[]*''.*''[\\]]*) *(.*)$");
|
| 439 |
|
| 440 |
// e.g. Tchaikovsky
|
| 441 |
// **4. Nocturne [???????] (C? minor)
|
| 442 |
// number - title - remarks
|
| 443 |
QRegExp partMatcher5("^\\*\\* *([0-9][0-9a-z]*)[\\.: ] *(.*\\[[^\\]]+\\])(.*)$");
|
| 444 |
|
| 445 |
// e.g. Schubert
|
| 446 |
// **2. "Wohin?"
|
| 447 |
// n/a - number - title
|
| 448 |
QRegExp partMatcher6("^\\*\\* *(([0-9][0-9a-z]*)[\\.:])? *(("|'').*)$");
|
| 449 |
|
| 450 |
// e.g. Mendelssohn
|
| 451 |
// ** Notturno
|
| 452 |
// title only
|
| 453 |
QRegExp partMatcher7("^\\*\\* *(.*)$");
|
| 454 |
|
| 455 |
|
| 456 |
// Date and remarks within titlefield or remarksfield
|
| 457 |
QRegExp matcherDate("\\([^\\)]*([0-9]{4})[^0-9\\)]*\\),?(.*)");
|
| 458 |
|
| 459 |
|
| 460 |
Work *main = 0;
|
| 461 |
int partNumber = 0; |
| 462 |
|
| 463 |
QString line; |
| 464 |
QString opfield, numfield, titlefield, remarksfield, datefield; |
| 465 |
|
| 466 |
while (!stream.atEnd()) {
|
| 467 |
|
| 468 |
if (line == "") { |
| 469 |
line = stream.readLine(); |
| 470 |
DEBUG << "line: " << line << endl;
|
| 471 |
} |
| 472 |
|
| 473 |
opfield = "";
|
| 474 |
numfield = "";
|
| 475 |
titlefield = "";
|
| 476 |
datefield = "";
|
| 477 |
remarksfield = "";
|
| 478 |
partNumber = 0;
|
| 479 |
|
| 480 |
if (workMatcher1.indexIn(line) >= 0) { |
| 481 |
|
| 482 |
DEBUG << "matcher 1" << endl;
|
| 483 |
opfield = workMatcher1.cap(1);
|
| 484 |
titlefield = workMatcher1.cap(3);
|
| 485 |
datefield = workMatcher1.cap(4);
|
| 486 |
remarksfield = workMatcher1.cap(6);
|
| 487 |
|
| 488 |
} else if (workMatcher1a.indexIn(line) >= 0) { |
| 489 |
|
| 490 |
DEBUG << "matcher 1a" << endl;
|
| 491 |
opfield = workMatcher1a.cap(1);
|
| 492 |
titlefield = workMatcher1a.cap(3);
|
| 493 |
/*
|
| 494 |
datefield = workMatcher1a.cap(4);
|
| 495 |
remarksfield = workMatcher1a.cap(6);
|
| 496 |
*/
|
| 497 |
|
| 498 |
} else if (workMatcher2.indexIn(line) >= 0) { |
| 499 |
|
| 500 |
DEBUG << "matcher 2" << endl;
|
| 501 |
titlefield = workMatcher2.cap(1);
|
| 502 |
remarksfield = workMatcher2.cap(2);
|
| 503 |
|
| 504 |
} else if (workMatcher3.indexIn(line) >= 0) { |
| 505 |
|
| 506 |
DEBUG << "matcher 3" << endl;
|
| 507 |
titlefield = workMatcher3.cap(1);
|
| 508 |
datefield = workMatcher3.cap(2);
|
| 509 |
remarksfield = workMatcher3.cap(3);
|
| 510 |
|
| 511 |
} else if (workMatcher4.indexIn(line) >= 0) { |
| 512 |
|
| 513 |
DEBUG << "matcher 4" << endl;
|
| 514 |
titlefield = workMatcher4.cap(1);
|
| 515 |
opfield = workMatcher4.cap(2);
|
| 516 |
remarksfield = workMatcher4.cap(4);
|
| 517 |
|
| 518 |
} else if (workMatcher5.indexIn(line) >= 0) { |
| 519 |
|
| 520 |
DEBUG << "matcher 5" << endl;
|
| 521 |
opfield = workMatcher5.cap(1);
|
| 522 |
titlefield = workMatcher5.cap(3);
|
| 523 |
remarksfield = workMatcher5.cap(4);
|
| 524 |
|
| 525 |
} else if (workMatcher6.indexIn(line) >= 0) { |
| 526 |
|
| 527 |
DEBUG << "matcher 6" << endl;
|
| 528 |
opfield = workMatcher6.cap(2);
|
| 529 |
titlefield = workMatcher6.cap(6);
|
| 530 |
|
| 531 |
} else if (workMatcher7.indexIn(line) >= 0) { |
| 532 |
|
| 533 |
DEBUG << "matcher 7" << endl;
|
| 534 |
titlefield = workMatcher7.cap(1);
|
| 535 |
opfield = workMatcher7.cap(2);
|
| 536 |
datefield = workMatcher7.cap(3);
|
| 537 |
|
| 538 |
} else if (workMatcher8.indexIn(line) >= 0) { |
| 539 |
|
| 540 |
DEBUG << "matcher 8" << endl;
|
| 541 |
titlefield = workMatcher8.cap(1);
|
| 542 |
datefield = workMatcher8.cap(2);
|
| 543 |
opfield = workMatcher8.cap(3);
|
| 544 |
|
| 545 |
} else if (workMatcher9.indexIn(line) >= 0) { |
| 546 |
|
| 547 |
DEBUG << "matcher 9" << endl;
|
| 548 |
opfield = workMatcher9.cap(1);
|
| 549 |
titlefield = workMatcher9.cap(2);
|
| 550 |
|
| 551 |
} else if (workMatcher10.indexIn(line) >= 0) { |
| 552 |
|
| 553 |
DEBUG << "matcher 10" << endl;
|
| 554 |
titlefield = workMatcher10.cap(1);
|
| 555 |
|
| 556 |
} else {
|
| 557 |
if (line.startsWith("*") || line.startsWith(":")) { |
| 558 |
DEBUG << "Failed to match promising works list line: " << line << endl;
|
| 559 |
} |
| 560 |
line = "";
|
| 561 |
continue;
|
| 562 |
} |
| 563 |
|
| 564 |
if (titlefield != "" && datefield == "") { |
| 565 |
int dpos;
|
| 566 |
if ((dpos = matcherDate.indexIn(titlefield)) != -1) { |
| 567 |
datefield = matcherDate.cap(1);
|
| 568 |
remarksfield = matcherDate.cap(2);
|
| 569 |
titlefield = titlefield.left(dpos); |
| 570 |
} |
| 571 |
} |
| 572 |
|
| 573 |
if (remarksfield != "" && datefield == "") { |
| 574 |
int dpos;
|
| 575 |
if ((dpos = matcherDate.indexIn(remarksfield)) != -1) { |
| 576 |
datefield = matcherDate.cap(1);
|
| 577 |
remarksfield = remarksfield.left(dpos); |
| 578 |
} |
| 579 |
} |
| 580 |
|
| 581 |
main = makeWork(composerName, opfield, "", 0, |
| 582 |
titlefield, datefield, "", remarksfield, 0); |
| 583 |
|
| 584 |
if (main) m_objects.push_back(main);
|
| 585 |
|
| 586 |
line = "";
|
| 587 |
|
| 588 |
while (!stream.atEnd()) {
|
| 589 |
|
| 590 |
++partNumber; |
| 591 |
line = stream.readLine(); |
| 592 |
DEBUG << "line: " << line << endl;
|
| 593 |
|
| 594 |
if (partMatcher1.indexIn(line) >= 0) { |
| 595 |
|
| 596 |
DEBUG << "part matcher 1" << endl;
|
| 597 |
numfield = partMatcher1.cap(1);
|
| 598 |
titlefield = partMatcher1.cap(3);
|
| 599 |
remarksfield = "";
|
| 600 |
|
| 601 |
} else if (partMatcher2.indexIn(line) >= 0) { |
| 602 |
|
| 603 |
DEBUG << "part matcher 2" << endl;
|
| 604 |
titlefield = partMatcher2.cap(1);
|
| 605 |
remarksfield = partMatcher2.cap(2);
|
| 606 |
|
| 607 |
} else if (partMatcher3.indexIn(line) >= 0) { |
| 608 |
|
| 609 |
DEBUG << "part matcher 3" << endl;
|
| 610 |
titlefield = partMatcher3.cap(1);
|
| 611 |
remarksfield = partMatcher3.cap(2);
|
| 612 |
|
| 613 |
} else if (partMatcher4.indexIn(line) >= 0) { |
| 614 |
|
| 615 |
DEBUG << "part matcher 4" << endl;
|
| 616 |
numfield = partMatcher4.cap(1);
|
| 617 |
titlefield = partMatcher4.cap(2);
|
| 618 |
remarksfield = partMatcher4.cap(3);
|
| 619 |
|
| 620 |
} else if (partMatcher5.indexIn(line) >= 0) { |
| 621 |
|
| 622 |
DEBUG << "part matcher 5" << endl;
|
| 623 |
numfield = partMatcher5.cap(1);
|
| 624 |
titlefield = partMatcher5.cap(2);
|
| 625 |
remarksfield = partMatcher5.cap(3);
|
| 626 |
|
| 627 |
} else if (partMatcher6.indexIn(line) >= 0) { |
| 628 |
|
| 629 |
DEBUG << "part matcher 6" << endl;
|
| 630 |
numfield = partMatcher6.cap(2);
|
| 631 |
titlefield = partMatcher6.cap(3);
|
| 632 |
|
| 633 |
} else if (partMatcher7.indexIn(line) >= 0) { |
| 634 |
|
| 635 |
DEBUG << "part matcher 7" << endl;
|
| 636 |
titlefield = partMatcher7.cap(1);
|
| 637 |
|
| 638 |
} else {
|
| 639 |
if (line.startsWith("**") || line.startsWith("::")) { |
| 640 |
DEBUG << "Failed to match promising part line: " << line << endl;
|
| 641 |
} |
| 642 |
break;
|
| 643 |
} |
| 644 |
|
| 645 |
if (titlefield != "" && datefield == "") { |
| 646 |
int dpos;
|
| 647 |
if ((dpos = matcherDate.indexIn(titlefield)) != -1) { |
| 648 |
datefield = matcherDate.cap(1);
|
| 649 |
remarksfield = matcherDate.cap(2);
|
| 650 |
titlefield = titlefield.left(dpos); |
| 651 |
} |
| 652 |
} |
| 653 |
|
| 654 |
Work *part = makeWork(composerName, opfield, numfield, partNumber, |
| 655 |
titlefield, datefield, "", remarksfield,
|
| 656 |
main); |
| 657 |
|
| 658 |
if (part) m_objects.push_back(part);
|
| 659 |
} |
| 660 |
} |
| 661 |
|
| 662 |
DEBUG << "Found " << m_objects.size() << " things" << endl; |
| 663 |
} |
| 664 |
|
| 665 |
|
| 666 |
} |
| 667 |
|
| 668 |
|