|
a/src/internfile/internfile.cpp |
|
b/src/internfile/internfile.cpp |
|
... |
|
... |
| 149 |
} else {
|
149 |
} else {
|
| 150 |
return ipath;
|
150 |
return ipath;
|
| 151 |
}
|
151 |
}
|
| 152 |
}
|
152 |
}
|
| 153 |
|
153 |
|
| 154 |
// Uncompress input file into a temporary one, by executing the appropriate
|
|
|
| 155 |
// script.
|
|
|
| 156 |
static bool uncompressfile(RclConfig *conf, const string& ifn,
|
|
|
| 157 |
const vector<string>& cmdv, TempDir& tdir,
|
|
|
| 158 |
string& tfile)
|
|
|
| 159 |
{
|
|
|
| 160 |
// Make sure tmp dir is empty. we guarantee this to filters
|
|
|
| 161 |
if (!tdir.ok() || !tdir.wipe()) {
|
|
|
| 162 |
LOGERR(("uncompressfile: can't clear temp dir %s\n", tdir.dirname()));
|
|
|
| 163 |
return false;
|
|
|
| 164 |
}
|
|
|
| 165 |
string cmd = cmdv.front();
|
|
|
| 166 |
|
|
|
| 167 |
// Substitute file name and temp dir in command elements
|
|
|
| 168 |
vector<string>::const_iterator it = cmdv.begin();
|
|
|
| 169 |
++it;
|
|
|
| 170 |
vector<string> args;
|
|
|
| 171 |
map<char, string> subs;
|
|
|
| 172 |
subs['f'] = ifn;
|
|
|
| 173 |
subs['t'] = tdir.dirname();
|
|
|
| 174 |
for (; it != cmdv.end(); it++) {
|
|
|
| 175 |
string ns;
|
|
|
| 176 |
pcSubst(*it, ns, subs);
|
|
|
| 177 |
args.push_back(ns);
|
|
|
| 178 |
}
|
|
|
| 179 |
|
|
|
| 180 |
// Execute command and retrieve output file name, check that it exists
|
|
|
| 181 |
ExecCmd ex;
|
|
|
| 182 |
int status = ex.doexec(cmd, args, 0, &tfile);
|
|
|
| 183 |
if (status || tfile.empty()) {
|
|
|
| 184 |
LOGERR(("uncompressfile: doexec: failed for [%s] status 0x%x\n",
|
|
|
| 185 |
ifn.c_str(), status));
|
|
|
| 186 |
if (!tdir.wipe()) {
|
|
|
| 187 |
LOGERR(("uncompressfile: wipedir failed\n"));
|
|
|
| 188 |
}
|
|
|
| 189 |
return false;
|
|
|
| 190 |
}
|
|
|
| 191 |
if (tfile[tfile.length() - 1] == '\n')
|
|
|
| 192 |
tfile.erase(tfile.length() - 1, 1);
|
|
|
| 193 |
return true;
|
|
|
| 194 |
}
|
|
|
| 195 |
|
|
|
| 196 |
// Delete temporary uncompressed file
|
|
|
| 197 |
void FileInterner::tmpcleanup()
|
|
|
| 198 |
{
|
|
|
| 199 |
if (m_tfile.empty())
|
|
|
| 200 |
return;
|
|
|
| 201 |
if (unlink(m_tfile.c_str()) < 0) {
|
|
|
| 202 |
LOGERR(("FileInterner::tmpcleanup: unlink(%s) errno %d\n",
|
|
|
| 203 |
m_tfile.c_str(), errno));
|
|
|
| 204 |
return;
|
|
|
| 205 |
}
|
|
|
| 206 |
}
|
|
|
| 207 |
|
|
|
| 208 |
// Constructor: identify the input file, possibly create an
|
154 |
// Constructor: identify the input file, possibly create an
|
| 209 |
// uncompressed temporary copy, and create the top filter for the
|
155 |
// uncompressed temporary copy, and create the top filter for the
|
| 210 |
// uncompressed file type.
|
156 |
// uncompressed file type.
|
| 211 |
//
|
157 |
//
|
| 212 |
// Empty handler on return says that we're in error, this will be
|
158 |
// Empty handler on return says that we're in error, this will be
|
| 213 |
// processed by the first call to internfile().
|
159 |
// processed by the first call to internfile().
|
| 214 |
// Split into "constructor calls init()" to allow use from other constructor
|
160 |
// Split into "constructor calls init()" to allow use from other constructor
|
| 215 |
FileInterner::FileInterner(const string &fn, const struct stat *stp,
|
161 |
FileInterner::FileInterner(const string &fn, const struct stat *stp,
|
| 216 |
RclConfig *cnf,
|
|
|
| 217 |
TempDir& td, int flags, const string *imime)
|
162 |
RclConfig *cnf, int flags, const string *imime)
|
| 218 |
: m_tdir(td), m_ok(false), m_missingdatap(0)
|
163 |
: m_ok(false), m_missingdatap(0), m_uncomp((flags & FIF_forPreview) != 0)
|
| 219 |
{
|
164 |
{
|
| 220 |
LOGDEB0(("FileInterner::FileInterner(fn=%s)\n", fn.c_str()));
|
165 |
LOGDEB0(("FileInterner::FileInterner(fn=%s)\n", fn.c_str()));
|
| 221 |
if (fn.empty()) {
|
166 |
if (fn.empty()) {
|
| 222 |
LOGERR(("FileInterner::FileInterner: empty file name!\n"));
|
167 |
LOGERR(("FileInterner::FileInterner: empty file name!\n"));
|
| 223 |
return;
|
168 |
return;
|
|
... |
|
... |
| 283 |
if (m_cfg->getUncompressor(l_mime, ucmd)) {
|
228 |
if (m_cfg->getUncompressor(l_mime, ucmd)) {
|
| 284 |
// Check for compressed size limit
|
229 |
// Check for compressed size limit
|
| 285 |
int maxkbs = -1;
|
230 |
int maxkbs = -1;
|
| 286 |
if (!m_cfg->getConfParam("compressedfilemaxkbs", &maxkbs) ||
|
231 |
if (!m_cfg->getConfParam("compressedfilemaxkbs", &maxkbs) ||
|
| 287 |
maxkbs < 0 || !stp || int(stp->st_size / 1024) < maxkbs) {
|
232 |
maxkbs < 0 || !stp || int(stp->st_size / 1024) < maxkbs) {
|
| 288 |
if (!uncompressfile(m_cfg, m_fn, ucmd, m_tdir, m_tfile)) {
|
233 |
if (!m_uncomp.uncompressfile(m_fn, ucmd, m_tfile)) {
|
| 289 |
return;
|
234 |
return;
|
| 290 |
}
|
235 |
}
|
| 291 |
LOGDEB1(("FileInterner:: after ucomp: m_tdir %s, tfile %s\n",
|
236 |
LOGDEB1(("FileInterner:: after ucomp: tfile %s\n",
|
| 292 |
m_tdir.dirname(), m_tfile.c_str()));
|
237 |
m_tfile.c_str()));
|
| 293 |
m_fn = m_tfile;
|
238 |
m_fn = m_tfile;
|
| 294 |
// Stat the uncompressed file, mainly to get the size
|
239 |
// Stat the uncompressed file, mainly to get the size
|
| 295 |
struct stat ucstat;
|
240 |
struct stat ucstat;
|
| 296 |
if (stat(m_fn.c_str(), &ucstat) != 0) {
|
241 |
if (stat(m_fn.c_str(), &ucstat) != 0) {
|
| 297 |
LOGERR(("FileInterner: can't stat the uncompressed file"
|
242 |
LOGERR(("FileInterner: can't stat the uncompressed file"
|
|
... |
|
... |
| 350 |
m_ok = true;
|
295 |
m_ok = true;
|
| 351 |
}
|
296 |
}
|
| 352 |
|
297 |
|
| 353 |
// Setup from memory data (ie: out of the web cache). imime needs to be set.
|
298 |
// Setup from memory data (ie: out of the web cache). imime needs to be set.
|
| 354 |
FileInterner::FileInterner(const string &data, RclConfig *cnf,
|
299 |
FileInterner::FileInterner(const string &data, RclConfig *cnf,
|
| 355 |
TempDir& td, int flags, const string& imime)
|
300 |
int flags, const string& imime)
|
| 356 |
: m_tdir(td), m_ok(false), m_missingdatap(0)
|
301 |
: m_ok(false), m_missingdatap(0), m_uncomp((flags & FIF_forPreview) != 0)
|
| 357 |
{
|
302 |
{
|
| 358 |
LOGDEB0(("FileInterner::FileInterner(data)\n"));
|
303 |
LOGDEB0(("FileInterner::FileInterner(data)\n"));
|
| 359 |
initcommon(cnf, flags);
|
304 |
initcommon(cnf, flags);
|
| 360 |
init(data, cnf, flags, imime);
|
305 |
init(data, cnf, flags, imime);
|
| 361 |
}
|
306 |
}
|
|
... |
|
... |
| 414 |
for (unsigned int i = 0; i < MAXHANDLERS; i++)
|
359 |
for (unsigned int i = 0; i < MAXHANDLERS; i++)
|
| 415 |
m_tmpflgs[i] = false;
|
360 |
m_tmpflgs[i] = false;
|
| 416 |
m_targetMType = cstr_textplain;
|
361 |
m_targetMType = cstr_textplain;
|
| 417 |
}
|
362 |
}
|
| 418 |
|
363 |
|
| 419 |
FileInterner::FileInterner(const Rcl::Doc& idoc, RclConfig *cnf,
|
364 |
FileInterner::FileInterner(const Rcl::Doc& idoc, RclConfig *cnf, int flags)
|
| 420 |
TempDir& td, int flags)
|
365 |
: m_ok(false), m_missingdatap(0), m_uncomp(((flags & FIF_forPreview) != 0))
|
| 421 |
: m_tdir(td), m_ok(false), m_missingdatap(0)
|
|
|
| 422 |
{
|
366 |
{
|
| 423 |
LOGDEB0(("FileInterner::FileInterner(idoc)\n"));
|
367 |
LOGDEB0(("FileInterner::FileInterner(idoc)\n"));
|
| 424 |
initcommon(cnf, flags);
|
368 |
initcommon(cnf, flags);
|
| 425 |
|
369 |
|
| 426 |
DocFetcher *fetcher = docFetcherMake(idoc);
|
370 |
DocFetcher *fetcher = docFetcherMake(idoc);
|
|
... |
|
... |
| 460 |
return ret;
|
404 |
return ret;
|
| 461 |
}
|
405 |
}
|
| 462 |
|
406 |
|
| 463 |
FileInterner::~FileInterner()
|
407 |
FileInterner::~FileInterner()
|
| 464 |
{
|
408 |
{
|
| 465 |
tmpcleanup();
|
|
|
| 466 |
for (vector<Dijon::Filter*>::iterator it = m_handlers.begin();
|
409 |
for (vector<Dijon::Filter*>::iterator it = m_handlers.begin();
|
| 467 |
it != m_handlers.end(); it++) {
|
410 |
it != m_handlers.end(); it++) {
|
| 468 |
returnMimeHandler(*it);
|
411 |
returnMimeHandler(*it);
|
| 469 |
}
|
412 |
}
|
| 470 |
// m_tempfiles will take care of itself
|
413 |
// m_tempfiles will take care of itself
|
|
... |
|
... |
| 1010 |
RclConfig *cnf, const Rcl::Doc& idoc)
|
953 |
RclConfig *cnf, const Rcl::Doc& idoc)
|
| 1011 |
{
|
954 |
{
|
| 1012 |
LOGDEB(("FileInterner::idocToFile\n"));
|
955 |
LOGDEB(("FileInterner::idocToFile\n"));
|
| 1013 |
idoc.dump();
|
956 |
idoc.dump();
|
| 1014 |
|
957 |
|
| 1015 |
TempDir tmpdir;
|
|
|
| 1016 |
|
|
|
| 1017 |
// We set FIF_forPreview for consistency with the previous version
|
958 |
// We set FIF_forPreview for consistency with the previous version
|
| 1018 |
// which determined this by looking at mtype!=null. Probably
|
959 |
// which determined this by looking at mtype!=null. Probably
|
| 1019 |
// doesn't change anything in this case.
|
960 |
// doesn't change anything in this case.
|
| 1020 |
FileInterner interner(idoc, cnf, tmpdir, FIF_forPreview);
|
961 |
FileInterner interner(idoc, cnf, FIF_forPreview);
|
| 1021 |
interner.setTargetMType(idoc.mimetype);
|
962 |
interner.setTargetMType(idoc.mimetype);
|
| 1022 |
return interner.interntofile(otemp, tofile, idoc.ipath, idoc.mimetype);
|
963 |
return interner.interntofile(otemp, tofile, idoc.ipath, idoc.mimetype);
|
| 1023 |
}
|
964 |
}
|
| 1024 |
|
965 |
|
| 1025 |
bool FileInterner::interntofile(TempFile& otemp, const string& tofile,
|
966 |
bool FileInterner::interntofile(TempFile& otemp, const string& tofile,
|
|
... |
|
... |
| 1100 |
return true;
|
1041 |
return true;
|
| 1101 |
}
|
1042 |
}
|
| 1102 |
return false;
|
1043 |
return false;
|
| 1103 |
}
|
1044 |
}
|
| 1104 |
|
1045 |
|
|
|
1046 |
// Static.
|
| 1105 |
bool FileInterner::maybeUncompressToTemp(TempFile& temp, const string& fn,
|
1047 |
bool FileInterner::maybeUncompressToTemp(TempFile& temp, const string& fn,
|
| 1106 |
RclConfig *cnf, const Rcl::Doc& doc)
|
1048 |
RclConfig *cnf, const Rcl::Doc& doc)
|
| 1107 |
{
|
1049 |
{
|
| 1108 |
LOGDEB(("FileInterner::maybeUncompressToTemp: [%s]\n", fn.c_str()));
|
1050 |
LOGDEB(("FileInterner::maybeUncompressToTemp: [%s]\n", fn.c_str()));
|
| 1109 |
struct stat st;
|
1051 |
struct stat st;
|
|
... |
|
... |
| 1129 |
maxkbs >= 0 && int(st.st_size / 1024) > maxkbs) {
|
1071 |
maxkbs >= 0 && int(st.st_size / 1024) > maxkbs) {
|
| 1130 |
LOGINFO(("FileInterner:: %s over size limit %d kbs\n",
|
1072 |
LOGINFO(("FileInterner:: %s over size limit %d kbs\n",
|
| 1131 |
fn.c_str(), maxkbs));
|
1073 |
fn.c_str(), maxkbs));
|
| 1132 |
return false;
|
1074 |
return false;
|
| 1133 |
}
|
1075 |
}
|
| 1134 |
TempDir tmpdir;
|
|
|
| 1135 |
temp =
|
1076 |
temp =
|
| 1136 |
TempFile(new TempFileInternal(cnf->getSuffixFromMimeType(doc.mimetype)));
|
1077 |
TempFile(new TempFileInternal(cnf->getSuffixFromMimeType(doc.mimetype)));
|
| 1137 |
if (!tmpdir.ok() || !temp->ok()) {
|
1078 |
if (!temp->ok()) {
|
| 1138 |
LOGERR(("FileInterner: cant create temporary file/dir"));
|
1079 |
LOGERR(("FileInterner: cant create temporary file"));
|
| 1139 |
return false;
|
1080 |
return false;
|
| 1140 |
}
|
1081 |
}
|
| 1141 |
|
1082 |
|
|
|
1083 |
Uncomp uncomp;
|
| 1142 |
string uncomped;
|
1084 |
string uncomped;
|
| 1143 |
if (!uncompressfile(cnf, fn, ucmd, tmpdir, uncomped)) {
|
1085 |
if (!uncomp.uncompressfile(fn, ucmd, uncomped)) {
|
| 1144 |
return false;
|
1086 |
return false;
|
| 1145 |
}
|
1087 |
}
|
| 1146 |
|
1088 |
|
| 1147 |
// uncompressfile choses the output file name, there is good
|
1089 |
// uncompressfile choses the output file name, there is good
|
| 1148 |
// reason for this, but it's not nice here. Have to move, the
|
1090 |
// reason for this, but it's not nice here. Have to move, the
|
|
... |
|
... |
| 1231 |
struct stat st;
|
1173 |
struct stat st;
|
| 1232 |
if (stat(fn.c_str(), &st)) {
|
1174 |
if (stat(fn.c_str(), &st)) {
|
| 1233 |
perror("stat");
|
1175 |
perror("stat");
|
| 1234 |
exit(1);
|
1176 |
exit(1);
|
| 1235 |
}
|
1177 |
}
|
| 1236 |
TempDir tmp;
|
|
|
| 1237 |
FileInterner interner(fn, &st, config, tmp, 0);
|
1178 |
FileInterner interner(fn, &st, config, 0);
|
| 1238 |
Rcl::Doc doc;
|
1179 |
Rcl::Doc doc;
|
| 1239 |
FileInterner::Status status = interner.internfile(doc, ipath);
|
1180 |
FileInterner::Status status = interner.internfile(doc, ipath);
|
| 1240 |
switch (status) {
|
1181 |
switch (status) {
|
| 1241 |
case FileInterner::FIDone:
|
1182 |
case FileInterner::FIDone:
|
| 1242 |
case FileInterner::FIAgain:
|
1183 |
case FileInterner::FIAgain:
|