|
a/src/internfile/mh_execm.cpp |
|
b/src/internfile/mh_execm.cpp |
|
... |
|
... |
| 23 |
#include "mh_execm.h"
|
23 |
#include "mh_execm.h"
|
| 24 |
#include "mh_html.h"
|
24 |
#include "mh_html.h"
|
| 25 |
#include "debuglog.h"
|
25 |
#include "debuglog.h"
|
| 26 |
#include "cancelcheck.h"
|
26 |
#include "cancelcheck.h"
|
| 27 |
#include "smallut.h"
|
27 |
#include "smallut.h"
|
| 28 |
#include "transcode.h"
|
|
|
| 29 |
#include "md5.h"
|
28 |
#include "md5.h"
|
| 30 |
#include "rclconfig.h"
|
29 |
#include "rclconfig.h"
|
| 31 |
#include "mimetype.h"
|
30 |
#include "mimetype.h"
|
| 32 |
#include "idfile.h"
|
31 |
#include "idfile.h"
|
| 33 |
|
32 |
|
|
... |
|
... |
| 281 |
m_fn.c_str(), reason.c_str()));
|
280 |
m_fn.c_str(), reason.c_str()));
|
| 282 |
}
|
281 |
}
|
| 283 |
}
|
282 |
}
|
| 284 |
|
283 |
|
| 285 |
// Charset. For many document types it doesn't matter. For text
|
284 |
// Charset. For many document types it doesn't matter. For text
|
| 286 |
// and html it does. We supply a default from the
|
285 |
// and html it does. We supply a default from the configuration.
|
| 287 |
// configuration.
|
|
|
| 288 |
bool trustcharset = true;
|
|
|
| 289 |
if (charset.empty()) {
|
286 |
if (charset.empty()) {
|
| 290 |
charset = cfgFilterOutputCharset.empty() ? "utf-8" :
|
287 |
charset = cfgFilterOutputCharset.empty() ? "utf-8" :
|
| 291 |
cfgFilterOutputCharset;
|
288 |
cfgFilterOutputCharset;
|
| 292 |
if (!stringlowercmp("default", charset)) {
|
289 |
if (!stringlowercmp("default", charset)) {
|
| 293 |
trustcharset = false;
|
|
|
| 294 |
charset = m_dfltInputCharset;
|
290 |
charset = m_dfltInputCharset;
|
| 295 |
}
|
291 |
}
|
| 296 |
}
|
292 |
}
|
|
|
293 |
m_metaData[cstr_origcharset] = charset;
|
|
|
294 |
m_metaData[cstr_charset] = charset;
|
| 297 |
|
295 |
|
| 298 |
string& output = m_metaData[cstr_content];
|
|
|
| 299 |
if (!m_metaData[cstr_mimetype].compare(cstr_textplain) &&
|
296 |
if (!m_metaData[cstr_mimetype].compare(cstr_textplain)) {
|
| 300 |
(!trustcharset || stringlowercmp("utf-8", charset))) {
|
297 |
(void)txtdcode("mh_execm");
|
| 301 |
string transcoded;
|
|
|
| 302 |
int ecnt;
|
|
|
| 303 |
if (!transcode(output, transcoded, charset, "UTF-8", &ecnt)) {
|
|
|
| 304 |
LOGERR(("mh_execm: transcode failed from [%s] to UTF-8\n",
|
|
|
| 305 |
charset.c_str()));
|
|
|
| 306 |
// Erase text in this case: it's garbage
|
|
|
| 307 |
output.clear();
|
|
|
| 308 |
} else {
|
|
|
| 309 |
if (ecnt) {
|
|
|
| 310 |
LOGDEB(("mh_exec: %d transcoding errors from [%s] to UTF-8\n",
|
|
|
| 311 |
ecnt, charset.c_str()));
|
|
|
| 312 |
}
|
|
|
| 313 |
output = transcoded;
|
|
|
| 314 |
charset = "utf-8";
|
|
|
| 315 |
}
|
|
|
| 316 |
}
|
298 |
}
|
| 317 |
|
299 |
|
| 318 |
m_metaData[cstr_charset] = charset;
|
|
|
| 319 |
|
300 |
|
| 320 |
if (eofnext_received)
|
301 |
if (eofnext_received)
|
| 321 |
m_havedoc = false;
|
302 |
m_havedoc = false;
|
| 322 |
|
303 |
|
| 323 |
LOGDEB0(("MHExecMultiple: returning %d bytes of content,"
|
304 |
LOGDEB0(("MHExecMultiple: returning %d bytes of content,"
|