| 1 | /* |
| 2 | SPDX-FileCopyrightText: 2018 Stefan BrĂ¼ns <stefan.bruens@rwth-aachen.de> |
| 3 | |
| 4 | SPDX-License-Identifier: LGPL-2.1-or-later |
| 5 | */ |
| 6 | |
| 7 | |
| 8 | #include "postscriptdscextractor.h" |
| 9 | #include "kfilemetadata_debug.h" |
| 10 | |
| 11 | #include <QFile> |
| 12 | #include <QTimeZone> |
| 13 | |
| 14 | namespace KFileMetaData |
| 15 | { |
| 16 | |
| 17 | DscExtractor::(QObject* parent) |
| 18 | : ExtractorPlugin(parent) |
| 19 | { |
| 20 | |
| 21 | } |
| 22 | |
| 23 | QStringList DscExtractor::() const |
| 24 | { |
| 25 | QStringList list; |
| 26 | list << QStringLiteral("application/postscript" ) |
| 27 | << QStringLiteral("image/x-eps" ); |
| 28 | |
| 29 | return list; |
| 30 | } |
| 31 | |
| 32 | void DscExtractor::(ExtractionResult* result) |
| 33 | { |
| 34 | QFile file(result->inputUrl()); |
| 35 | if (!file.open(flags: QIODevice::ReadOnly)) { |
| 36 | qCWarning(KFILEMETADATA_LOG) << "Document is not a valid file" ; |
| 37 | return; |
| 38 | } |
| 39 | |
| 40 | // A little bit heuristic - assume EPS files are images, PS complete documents |
| 41 | if (result->inputMimetype() == QLatin1String("application/postscript" )) { |
| 42 | result->addType(type: Type::Document); |
| 43 | } else { |
| 44 | result->addType(type: Type::Image); |
| 45 | } |
| 46 | |
| 47 | if (!(result->inputFlags() & ExtractionResult::ExtractMetaData)) { |
| 48 | return; |
| 49 | } |
| 50 | // Try to find some DSC (PostScript Language Document Structuring Conventions) conforming data |
| 51 | QTextStream stream(&file); |
| 52 | QString line; |
| 53 | |
| 54 | while (stream.readLineInto(line: &line)) { |
| 55 | if (!line.startsWith(s: QLatin1String("%%" ))) { |
| 56 | continue; |
| 57 | } |
| 58 | |
| 59 | if (const auto tag = QLatin1String("%%Pages:" ); line.startsWith(s: tag)) { |
| 60 | bool ok = false; |
| 61 | int pages = QStringView(line).mid(pos: tag.size()).toInt(ok: &ok, base: 10); |
| 62 | if (ok) { |
| 63 | result->add(property: Property::PageCount, value: pages); |
| 64 | } |
| 65 | |
| 66 | } else if (const auto tag = QLatin1String("%%Title:" ); line.startsWith(s: tag)) { |
| 67 | QStringView title = QStringView(line).mid(pos: tag.size()).trimmed(); |
| 68 | if (title.startsWith(c: QLatin1Char('(')) && title.endsWith(c: QLatin1Char(')'))) { |
| 69 | title = title.mid(pos: 1, n: title.size() - 2); |
| 70 | } |
| 71 | if (!title.isEmpty()) { |
| 72 | result->add(property: Property::Title, value: title.toString()); |
| 73 | } |
| 74 | |
| 75 | } else if (const auto tag = QLatin1String("%%CreationDate:" ); line.startsWith(s: tag)) { |
| 76 | // "Neither the date nor time need be in any standard format." |
| 77 | QStringView date = QStringView(line).mid(pos: tag.size()).trimmed(); |
| 78 | if (date.startsWith(c: QLatin1Char('(')) && date.endsWith(c: QLatin1Char(')'))) { |
| 79 | date = date.mid(pos: 1, n: date.size() - 2); |
| 80 | date = date.trimmed(); |
| 81 | } |
| 82 | if (date.startsWith(s: QLatin1String("D:" )) && date.size() >= 23) { |
| 83 | // Standard PDF date format, ASN.1 like - (D:YYYYMMDDHHmmSSOHH'mm') |
| 84 | auto dt = QDateTime::fromString(string: date.mid(pos: 2, n: 14).toString(), format: QLatin1String("yyyyMMddhhmmss" )); |
| 85 | auto offset = QTime::fromString(string: date.mid(pos: 17, n: 5).toString(), format: QLatin1String("hh'\\''mm" )); |
| 86 | if (date.at(n: 16) == QLatin1Char('+')) { |
| 87 | dt.setTimeZone(toZone: QTimeZone::fromSecondsAheadOfUtc(offset: QTime(0, 0).secsTo(t: offset))); |
| 88 | } else { |
| 89 | dt.setTimeZone(toZone: QTimeZone::fromSecondsAheadOfUtc(offset: -1 * QTime(0, 0).secsTo(t: offset))); |
| 90 | } |
| 91 | result->add(property: Property::CreationDate, value: dt); |
| 92 | } else { |
| 93 | auto dt = QDateTime::fromString(string: date.toString()); |
| 94 | if (dt.isValid()) { |
| 95 | result->add(property: Property::CreationDate, value: dt); |
| 96 | } |
| 97 | } |
| 98 | |
| 99 | } else if (line.startsWith(s: QLatin1String("%%EndComments" ))) { |
| 100 | break; |
| 101 | } |
| 102 | } |
| 103 | } |
| 104 | |
| 105 | } // namespace KFileMetaData |
| 106 | |
| 107 | #include "moc_postscriptdscextractor.cpp" |
| 108 | |