tris/recoll
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

internfile: do not process set_document_xx errors. Wait until the next_document() call so that the file names or ipaths are indexed

Browse Source
This commit is contained in:
Jean-Francois Dockes 2019-03-12 14:55:31 +01:00
parent 0086f57162
commit bfa786dfeb
1 changed files with 14 additions and 33 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

47
src/internfile/internfile.cpp
View File

@ -245,12 +245,9 @@ void FileInterner::init(const string &f, const struct stat *stp, RclConfig *cnf,
df->set_property(Dijon::Filter::DJF_UDI, udi); df->set_property(Dijon::Filter::DJF_UDI, udi);
df->set_docsize(docsize); df->set_docsize(docsize);
if (!df->set_document_file(l_mime, m_fn)) { // Don't process init errors here: doing it later allows indexing
delete df; // the file name of even a totally unparsable file
LOGERR("FileInterner:: error converting " << m_fn << "\n"); (void)df->set_document_file(l_mime, m_fn);
return;
}
m_handlers.push_back(df); m_handlers.push_back(df);
LOGDEB("FileInterner:: init ok " << l_mime << " [" << m_fn << "]\n"); LOGDEB("FileInterner:: init ok " << l_mime << " [" << m_fn << "]\n");
m_ok = true; m_ok = true;
@ -286,26 +283,21 @@ void FileInterner::init(const string &data, RclConfig *cnf,
df->set_property(Dijon::Filter::OPERATING_MODE, df->set_property(Dijon::Filter::OPERATING_MODE,
m_forPreview ? "view" : "index"); m_forPreview ? "view" : "index");
bool result = false;
df->set_docsize(data.length()); df->set_docsize(data.length());
if (df->is_data_input_ok(Dijon::Filter::DOCUMENT_STRING)) { if (df->is_data_input_ok(Dijon::Filter::DOCUMENT_STRING)) {
result = df->set_document_string(m_mimetype, data); (void)df->set_document_string(m_mimetype, data);
} else if (df->is_data_input_ok(Dijon::Filter::DOCUMENT_DATA)) { } else if (df->is_data_input_ok(Dijon::Filter::DOCUMENT_DATA)) {
result = df->set_document_data(m_mimetype, data.c_str(), data.length()); (void)df->set_document_data(m_mimetype, data.c_str(), data.length());
} else if (df->is_data_input_ok(Dijon::Filter::DOCUMENT_FILE_NAME)) { } else if (df->is_data_input_ok(Dijon::Filter::DOCUMENT_FILE_NAME)) {
TempFile temp = dataToTempFile(data, m_mimetype); TempFile temp = dataToTempFile(data, m_mimetype);
if (temp.ok() && if (temp.ok()) {
(result = df->set_document_file(m_mimetype, temp.filename()))) { (void)df->set_document_file(m_mimetype, temp.filename());
m_tmpflgs[m_handlers.size()] = true; m_tmpflgs[m_handlers.size()] = true;
m_tempfiles.push_back(temp); m_tempfiles.push_back(temp);
} }
} }
if (!result) { // Don't process init errors here: doing it later allows indexing
LOGINFO("FileInterner:: set_doc failed inside for mtype " << // the file name of even a totally unparsable file
m_mimetype << "\n");
delete df;
return;
}
m_handlers.push_back(df); m_handlers.push_back(df);
m_ok = true; m_ok = true;
} }
@ -763,15 +755,11 @@ int FileInterner::addHandler()
if (!setres) { if (!setres) {
LOGINFO("FileInterner::addHandler: set_doc failed inside [" << m_fn << LOGINFO("FileInterner::addHandler: set_doc failed inside [" << m_fn <<
"] for mtype " << mimetype << "\n"); "] for mtype " << mimetype << "\n");
delete newflt;
if (m_forPreview)
return ADD_ERROR;
return ADD_CONTINUE;
} }
// add handler and go on, maybe this one will give us text... // Add handler and go on, maybe this one will give us text...
m_handlers.push_back(newflt); m_handlers.push_back(newflt);
LOGDEB1("FileInterner::addHandler: added\n"); LOGDEB1("FileInterner::addHandler: added\n");
return ADD_OK; return setres ? ADD_OK : ADD_BREAK;
} }
// Information and debug after a next_document error // Information and debug after a next_document error
@ -799,19 +787,12 @@ FileInterner::Status FileInterner::internfile(Rcl::Doc& doc,const string& ipath)
} }
// Input Ipath vector when retrieving a given subdoc for previewing // Input Ipath vector when retrieving a given subdoc for previewing
// Note that the vector is big enough for the maximum stack. All values
// over the last significant one are ""
// We set the ipath for the first handler here, others are set
// when they're pushed on the stack
vector<string> vipath; vector<string> vipath;
if (!ipath.empty() && !m_direct) { if (!ipath.empty() && !m_direct) {
vector<string> lipath; stringToTokens(ipath, vipath, cstr_isep, true);
stringToTokens(ipath, lipath, cstr_isep, true); for (auto& entry: vipath) {
for (vector<string>::iterator it = lipath.begin(); entry = colon_restore(entry);
it != lipath.end(); it++) {
*it = colon_restore(*it);
} }
vipath.insert(vipath.begin(), lipath.begin(), lipath.end());
if (!m_handlers.back()->skip_to_document(vipath[m_handlers.size()-1])){ if (!m_handlers.back()->skip_to_document(vipath[m_handlers.size()-1])){
LOGERR("FileInterner::internfile: can't skip\n"); LOGERR("FileInterner::internfile: can't skip\n");
return FIError; return FIError;