| 1 | /* |
| 2 | Implementation of the data protocol (rfc 2397) |
| 3 | |
| 4 | SPDX-FileCopyrightText: 2002, 2003 Leo Savernik <l.savernik@aon.at> |
| 5 | |
| 6 | SPDX-License-Identifier: LGPL-2.0-only |
| 7 | */ |
| 8 | |
| 9 | #include "dataprotocol_p.h" |
| 10 | |
| 11 | #include "global.h" |
| 12 | #include "metadata.h" |
| 13 | |
| 14 | #include <QByteArray> |
| 15 | #include <QStringDecoder> |
| 16 | |
| 17 | using namespace KIO; |
| 18 | |
| 19 | /* structure containing header information */ |
| 20 | struct { |
| 21 | QString ; // MIME type of content (lowercase) |
| 22 | MetaData ; // attribute/value pairs (attribute lowercase, |
| 23 | // value unchanged) |
| 24 | bool ; // true if data is base64 encoded |
| 25 | QByteArray ; // reference to decoded url |
| 26 | int ; // zero-indexed position within url |
| 27 | // where the real data begins. May point beyond |
| 28 | // the end to indicate that there is no data |
| 29 | }; |
| 30 | |
| 31 | /* returns the position of the first occurrence of any of the given |
| 32 | * characters c1 or comma (',') or semicolon (';') or buf.length() |
| 33 | * if none is contained. |
| 34 | * |
| 35 | * \a buf buffer where to look for c |
| 36 | * \a begin zero-indexed starting position |
| 37 | * \a c1 character to find or '\0' to ignore |
| 38 | */ |
| 39 | static int find(const QByteArray &buf, int begin, const char c1) |
| 40 | { |
| 41 | static const char comma = ','; |
| 42 | static const char semicolon = ';'; |
| 43 | int pos = begin; |
| 44 | int size = buf.length(); |
| 45 | while (pos < size) { |
| 46 | const char ch = buf[pos]; |
| 47 | if (ch == comma || ch == semicolon || (c1 != '\0' && ch == c1)) { |
| 48 | break; |
| 49 | } |
| 50 | pos++; |
| 51 | } /*wend*/ |
| 52 | return pos; |
| 53 | } |
| 54 | |
| 55 | /* extracts the string between the current position pos and the first |
| 56 | * occurrence of either c1 or comma (',') or semicolon (';') exclusively |
| 57 | * and updates pos to point at the found delimiter or at the end of the |
| 58 | * buffer if neither character occurred. |
| 59 | * \a buf buffer where to look for |
| 60 | * \a pos zero-indexed position within buffer |
| 61 | * \a c1 character to find or '\0' to ignore |
| 62 | */ |
| 63 | static inline QString (const QByteArray &buf, int &pos, const char c1 = '\0') |
| 64 | { |
| 65 | int oldpos = pos; |
| 66 | pos = find(buf, begin: oldpos, c1); |
| 67 | return QString::fromLatin1(ba: buf.mid(index: oldpos, len: pos - oldpos)); |
| 68 | } |
| 69 | |
| 70 | /* ignores all whitespaces |
| 71 | * \a buf buffer to operate on |
| 72 | * \a pos position to shift to first non-whitespace character |
| 73 | * Upon return pos will either point to the first non-whitespace |
| 74 | * character or to the end of the buffer. |
| 75 | */ |
| 76 | static inline void ignoreWS(const QByteArray &buf, int &pos) |
| 77 | { |
| 78 | int size = buf.length(); |
| 79 | while (pos < size && (buf[pos] == ' ' || buf[pos] == '\t')) { |
| 80 | ++pos; |
| 81 | } |
| 82 | } |
| 83 | |
| 84 | /* parses a quoted string as per rfc 822. |
| 85 | * |
| 86 | * If trailing quote is missing, the whole rest of the buffer is returned. |
| 87 | * \a buf buffer to operate on |
| 88 | * \a pos position pointing to the leading quote |
| 89 | * Returns the extracted string. pos will be updated to point to the |
| 90 | * character following the trailing quote. |
| 91 | */ |
| 92 | static QString parseQuotedString(const QByteArray &buf, int &pos) |
| 93 | { |
| 94 | int size = buf.length(); |
| 95 | QString res; |
| 96 | res.reserve(asize: size); // can't be larger than buf |
| 97 | pos++; // jump over leading quote |
| 98 | bool escaped = false; // if true means next character is literal |
| 99 | bool parsing = true; // true as long as end quote not found |
| 100 | while (parsing && pos < size) { |
| 101 | const QChar ch = QLatin1Char(buf[pos++]); |
| 102 | if (escaped) { |
| 103 | res += ch; |
| 104 | escaped = false; |
| 105 | } else { |
| 106 | switch (ch.unicode()) { |
| 107 | case '"': |
| 108 | parsing = false; |
| 109 | break; |
| 110 | case '\\': |
| 111 | escaped = true; |
| 112 | break; |
| 113 | default: |
| 114 | res += ch; |
| 115 | break; |
| 116 | } /*end switch*/ |
| 117 | } /*end if*/ |
| 118 | } /*wend*/ |
| 119 | res.squeeze(); |
| 120 | return res; |
| 121 | } |
| 122 | |
| 123 | /* parses the header of a data url |
| 124 | * \a url the data url |
| 125 | * \a mimeOnly if the only interesting information is the MIME type |
| 126 | * Returns DataHeader structure with the header information |
| 127 | */ |
| 128 | static DataHeader (const QUrl &url, const bool mimeOnly) |
| 129 | { |
| 130 | DataHeader ; |
| 131 | |
| 132 | // initialize header info members |
| 133 | header_info.mime_type = QStringLiteral("text/plain" ); |
| 134 | header_info.attributes.insert(QStringLiteral("charset" ), QStringLiteral("us-ascii" )); |
| 135 | header_info.is_base64 = false; |
| 136 | |
| 137 | // decode url and save it |
| 138 | const QByteArray &raw_url = header_info.url = QByteArray::fromPercentEncoding(pctEncoded: url.path(options: QUrl::FullyEncoded).toLatin1()); |
| 139 | const int raw_url_len = raw_url.length(); |
| 140 | |
| 141 | header_info.data_offset = 0; |
| 142 | |
| 143 | // read MIME type |
| 144 | if (raw_url_len == 0) { |
| 145 | return header_info; |
| 146 | } |
| 147 | const QString mime_type = extract(buf: raw_url, pos&: header_info.data_offset).trimmed(); |
| 148 | if (!mime_type.isEmpty()) { |
| 149 | header_info.mime_type = mime_type; |
| 150 | } |
| 151 | if (mimeOnly) { |
| 152 | return header_info; |
| 153 | } |
| 154 | |
| 155 | if (header_info.data_offset >= raw_url_len) { |
| 156 | return header_info; |
| 157 | } |
| 158 | // jump over delimiter token and return if data reached |
| 159 | if (raw_url[header_info.data_offset++] == ',') { |
| 160 | return header_info; |
| 161 | } |
| 162 | |
| 163 | // read all attributes and store them |
| 164 | bool data_begin_reached = false; |
| 165 | while (!data_begin_reached && header_info.data_offset < raw_url_len) { |
| 166 | // read attribute |
| 167 | const QString attribute = extract(buf: raw_url, pos&: header_info.data_offset, c1: '=').trimmed(); |
| 168 | if (header_info.data_offset >= raw_url_len || raw_url[header_info.data_offset] != '=') { |
| 169 | // no assignment, must be base64 option |
| 170 | if (attribute == QLatin1String("base64" )) { |
| 171 | header_info.is_base64 = true; |
| 172 | } |
| 173 | } else { |
| 174 | header_info.data_offset++; // jump over '=' token |
| 175 | |
| 176 | // read value |
| 177 | ignoreWS(buf: raw_url, pos&: header_info.data_offset); |
| 178 | if (header_info.data_offset >= raw_url_len) { |
| 179 | return header_info; |
| 180 | } |
| 181 | |
| 182 | QString value; |
| 183 | if (raw_url[header_info.data_offset] == '"') { |
| 184 | value = parseQuotedString(buf: raw_url, pos&: header_info.data_offset); |
| 185 | ignoreWS(buf: raw_url, pos&: header_info.data_offset); |
| 186 | } else { |
| 187 | value = extract(buf: raw_url, pos&: header_info.data_offset).trimmed(); |
| 188 | } |
| 189 | |
| 190 | // add attribute to map |
| 191 | header_info.attributes[attribute.toLower()] = value; |
| 192 | |
| 193 | } /*end if*/ |
| 194 | if (header_info.data_offset < raw_url_len && raw_url[header_info.data_offset] == ',') { |
| 195 | data_begin_reached = true; |
| 196 | } |
| 197 | header_info.data_offset++; // jump over separator token |
| 198 | } /*wend*/ |
| 199 | |
| 200 | return header_info; |
| 201 | } |
| 202 | |
| 203 | DataProtocol::DataProtocol() |
| 204 | { |
| 205 | } |
| 206 | |
| 207 | DataProtocol::~DataProtocol() = default; |
| 208 | |
| 209 | void DataProtocol::get(const QUrl &url) |
| 210 | { |
| 211 | ref(); |
| 212 | // qDebug() << this; |
| 213 | |
| 214 | const DataHeader hdr = parseDataHeader(url, mimeOnly: false); |
| 215 | |
| 216 | const int size = hdr.url.length(); |
| 217 | const int data_ofs = qMin(a: hdr.data_offset, b: size); |
| 218 | // FIXME: string is copied, would be nice if we could have a reference only |
| 219 | const QByteArray url_data = hdr.url.mid(index: data_ofs); |
| 220 | QByteArray outData; |
| 221 | |
| 222 | if (hdr.is_base64) { |
| 223 | // base64 stuff is expected to contain the correct charset, so we just |
| 224 | // decode it and pass it to the receiver |
| 225 | outData = QByteArray::fromBase64(base64: url_data); |
| 226 | } else { |
| 227 | QStringDecoder codec(hdr.attributes[QStringLiteral("charset" )].toLatin1().constData()); |
| 228 | if (codec.isValid()) { |
| 229 | outData = QString(codec.decode(ba: url_data)).toUtf8(); |
| 230 | } else { |
| 231 | outData = url_data; |
| 232 | } /*end if*/ |
| 233 | } /*end if*/ |
| 234 | |
| 235 | // qDebug() << "emit mimeType@"<<this; |
| 236 | Q_EMIT mimeType(hdr.mime_type); |
| 237 | // qDebug() << "emit totalSize@"<<this; |
| 238 | Q_EMIT totalSize(outData.size()); |
| 239 | |
| 240 | // qDebug() << "emit setMetaData@"<<this; |
| 241 | setAllMetaData(hdr.attributes); |
| 242 | |
| 243 | // qDebug() << "emit sendMetaData@"<<this; |
| 244 | sendMetaData(); |
| 245 | // qDebug() << "(1) queue size " << dispatchQueue.size(); |
| 246 | // empiric studies have shown that this shouldn't be queued & dispatched |
| 247 | Q_EMIT data(outData); |
| 248 | // qDebug() << "(2) queue size " << dispatchQueue.size(); |
| 249 | dispatch_data(ba: QByteArray{}); |
| 250 | // qDebug() << "(3) queue size " << dispatchQueue.size(); |
| 251 | dispatch_finished(); |
| 252 | // qDebug() << "(4) queue size " << dispatchQueue.size(); |
| 253 | deref(); |
| 254 | } |
| 255 | |
| 256 | /* --------------------------------------------------------------------- */ |
| 257 | |
| 258 | void DataProtocol::mimetype(const QUrl &url) |
| 259 | { |
| 260 | ref(); |
| 261 | Q_EMIT mimeType(parseDataHeader(url, mimeOnly: true).mime_type); |
| 262 | Q_EMIT finished(); |
| 263 | deref(); |
| 264 | } |
| 265 | |
| 266 | /* --------------------------------------------------------------------- */ |
| 267 | |
| 268 | #if !defined(TESTKIO) |
| 269 | #include "moc_dataprotocol_p.cpp" |
| 270 | #endif |
| 271 | |