recoll / Code / Diff of /src/internfile/internfile.cpp

Diff of /src/internfile/internfile.cpp [6cdf9a] .. [638d46]

Switch to unified view


...
    }
    LOGDEB2(("getKeyValue: no value for [%s]\n", key.c_str()));
    return false;
}









bool FileInterner::dijontorcl(Rcl::Doc& doc)
{
    Dijon::Filter *df = m_handlers.back();
    if (df == 0) {
    //??
...
    }
    const map<string, string>& docdata = df->get_meta_data();

    for (map<string,string>::const_iterator it = docdata.begin(); 
     it != docdata.end(); it++) {
    if (it->first == cstr_dj_keycontent) {
        doc.text = it->second;
    } else if (it->first == cstr_dj_keymd) {
        doc.dmtime = it->second;
  } else if (it->first == cstr_dj_keyorigcharset) {
        doc.origcharset = it->second;
    } else if (it->first == cstr_dj_keymt || it->first == cstr_dj_keycharset) {
        // don't need/want these.
    } else {
        doc.meta[it->first] = it->second;
    }
    }
    if (doc.meta[Rcl::Doc::keyabs].empty() && !doc.meta[cstr_dj_keyds].empty()) {
    doc.meta[Rcl::Doc::keyabs] = doc.meta[cstr_dj_keyds];
    doc.meta.erase(cstr_dj_keyds);
    }
    return true;
}

// Collect the ipath from the current path in the document tree.
...

    string ipathel;
    for (vector<Dijon::Filter*>::const_iterator hit = m_handlers.begin();
     hit != m_handlers.end(); hit++) {
    const map<string, string>& docdata = (*hit)->get_meta_data();
    if (getKeyValue(docdata, cstr_dj_keyipath, ipathel)) {
        if (!ipathel.empty()) {
        // We have a non-empty ipath
        hasipath = true;
        getKeyValue(docdata, cstr_dj_keymt, doc.mimetype);
        getKeyValue(docdata, cstr_dj_keyfn, doc.utf8fn);
        }
        doc.ipath += colon_hide(ipathel) + cstr_isep;
    } else {
        doc.ipath += cstr_isep;
    }
    getKeyValue(docdata, cstr_dj_keyauthor, doc.meta[Rcl::Doc::keyau]);
    getKeyValue(docdata, cstr_dj_keymd, doc.dmtime);
    }

    // Trim empty tail elements in ipath.
    if (hasipath) {
    LOGDEB2(("IPATH [%s]\n", doc.ipath.c_str()));
...
// and possibly add a filter/handler to the stack
int FileInterner::addHandler()
{
    const map<string, string>& docdata = m_handlers.back()->get_meta_data();
    string charset, mimetype;
    getKeyValue(docdata, cstr_dj_keycharset, charset);
    getKeyValue(docdata, cstr_dj_keymt, mimetype);

    LOGDEB(("FileInterner::addHandler: next_doc is %s\n", mimetype.c_str()));

    // If we find a document of the target type (text/plain in
    // general), we're done decoding. If we hit text/plain, we're done
...
    // copying the text, which may be big.
    string ns;
    const string *txt = &ns;
    {
    map<string,string>::const_iterator it;
    it = docdata.find(cstr_dj_keycontent);
    if (it != docdata.end())
        txt = &it->second;
    }
    bool setres = false;
    if (newflt->is_data_input_ok(Dijon::Filter::DOCUMENT_STRING)) {

	a/src/internfile/internfile.cpp		b/src/internfile/internfile.cpp
	...		...
636	}	636	}
637	LOGDEB2(("getKeyValue: no value for [%s]\n", key.c_str()));	637	LOGDEB2(("getKeyValue: no value for [%s]\n", key.c_str()));
638	return false;	638	return false;
639	}	639	}
640		640
641	// These defs are for the Dijon meta array. Rcl::Doc predefined field
642	// names are used where appropriate. In some cases, Rcl::Doc names are
643	// used inside the Dijon metadata (ex: origcharset)
644	static const string cstr_keyds("description");
645	static const string cstr_keyfn("filename");
646	static const string cstr_keymd("modificationdate");
647	static const string cstr_keytt("title");
648
649	bool FileInterner::dijontorcl(Rcl::Doc& doc)	641	bool FileInterner::dijontorcl(Rcl::Doc& doc)
650	{	642	{
651	Dijon::Filter *df = m_handlers.back();	643	Dijon::Filter *df = m_handlers.back();
652	if (df == 0) {	644	if (df == 0) {
653	//??	645	//??
	...		...
656	}	648	}
657	const map<string, string>& docdata = df->get_meta_data();	649	const map<string, string>& docdata = df->get_meta_data();
658		650
659	for (map<string,string>::const_iterator it = docdata.begin();	651	for (map<string,string>::const_iterator it = docdata.begin();
660	it != docdata.end(); it++) {	652	it != docdata.end(); it++) {
661	if (it->first == cstr_content) {	653	if (it->first == cstr_dj_keycontent) {
662	doc.text = it->second;	654	doc.text = it->second;
663	} else if (it->first == cstr_keymd) {	655	} else if (it->first == cstr_dj_keymd) {
664	doc.dmtime = it->second;	656	doc.dmtime = it->second;
665	} else if (it->first == Rcl::Doc::keyoc) {	657	} else if (it->first == cstr_dj_keyorigcharset) {
666	doc.origcharset = it->second;	658	doc.origcharset = it->second;
667	} else if (it->first == cstr_mimetype \|\| it->first == cstr_charset) {	659	} else if (it->first == cstr_dj_keymt \|\| it->first == cstr_dj_keycharset) {
668	// don't need/want these.	660	// don't need/want these.
669	} else {	661	} else {
670	doc.meta[it->first] = it->second;	662	doc.meta[it->first] = it->second;
671	}	663	}
672	}	664	}
673	if (doc.meta[Rcl::Doc::keyabs].empty() && !doc.meta[cstr_keyds].empty()) {	665	if (doc.meta[Rcl::Doc::keyabs].empty() && !doc.meta[cstr_dj_keyds].empty()) {
674	doc.meta[Rcl::Doc::keyabs] = doc.meta[cstr_keyds];	666	doc.meta[Rcl::Doc::keyabs] = doc.meta[cstr_dj_keyds];
675	doc.meta.erase(cstr_keyds);	667	doc.meta.erase(cstr_dj_keyds);
676	}	668	}
677	return true;	669	return true;
678	}	670	}
679		671
680	// Collect the ipath from the current path in the document tree.	672	// Collect the ipath from the current path in the document tree.
	...		...
702		694
703	string ipathel;	695	string ipathel;
704	for (vector<Dijon::Filter*>::const_iterator hit = m_handlers.begin();	696	for (vector<Dijon::Filter*>::const_iterator hit = m_handlers.begin();
705	hit != m_handlers.end(); hit++) {	697	hit != m_handlers.end(); hit++) {
706	const map<string, string>& docdata = (*hit)->get_meta_data();	698	const map<string, string>& docdata = (*hit)->get_meta_data();
707	if (getKeyValue(docdata, cstr_ipath, ipathel)) {	699	if (getKeyValue(docdata, cstr_dj_keyipath, ipathel)) {
708	if (!ipathel.empty()) {	700	if (!ipathel.empty()) {
709	// We have a non-empty ipath	701	// We have a non-empty ipath
710	hasipath = true;	702	hasipath = true;
711	getKeyValue(docdata, cstr_mimetype, doc.mimetype);	703	getKeyValue(docdata, cstr_dj_keymt, doc.mimetype);
712	getKeyValue(docdata, cstr_keyfn, doc.utf8fn);	704	getKeyValue(docdata, cstr_dj_keyfn, doc.utf8fn);
713	}	705	}
714	doc.ipath += colon_hide(ipathel) + cstr_isep;	706	doc.ipath += colon_hide(ipathel) + cstr_isep;
715	} else {	707	} else {
716	doc.ipath += cstr_isep;	708	doc.ipath += cstr_isep;
717	}	709	}
718	getKeyValue(docdata, cstr_author, doc.meta[Rcl::Doc::keyau]);	710	getKeyValue(docdata, cstr_dj_keyauthor, doc.meta[Rcl::Doc::keyau]);
719	getKeyValue(docdata, cstr_keymd, doc.dmtime);	711	getKeyValue(docdata, cstr_dj_keymd, doc.dmtime);
720	}	712	}
721		713
722	// Trim empty tail elements in ipath.	714	// Trim empty tail elements in ipath.
723	if (hasipath) {	715	if (hasipath) {
724	LOGDEB2(("IPATH [%s]\n", doc.ipath.c_str()));	716	LOGDEB2(("IPATH [%s]\n", doc.ipath.c_str()));
	...		...
752	// and possibly add a filter/handler to the stack	744	// and possibly add a filter/handler to the stack
753	int FileInterner::addHandler()	745	int FileInterner::addHandler()
754	{	746	{
755	const map<string, string>& docdata = m_handlers.back()->get_meta_data();	747	const map<string, string>& docdata = m_handlers.back()->get_meta_data();
756	string charset, mimetype;	748	string charset, mimetype;
757	getKeyValue(docdata, cstr_charset, charset);	749	getKeyValue(docdata, cstr_dj_keycharset, charset);
758	getKeyValue(docdata, cstr_mimetype, mimetype);	750	getKeyValue(docdata, cstr_dj_keymt, mimetype);
759		751
760	LOGDEB(("FileInterner::addHandler: next_doc is %s\n", mimetype.c_str()));	752	LOGDEB(("FileInterner::addHandler: next_doc is %s\n", mimetype.c_str()));
761		753
762	// If we find a document of the target type (text/plain in	754	// If we find a document of the target type (text/plain in
763	// general), we're done decoding. If we hit text/plain, we're done	755	// general), we're done decoding. If we hit text/plain, we're done
	...		...
794	// copying the text, which may be big.	786	// copying the text, which may be big.
795	string ns;	787	string ns;
796	const string *txt = &ns;	788	const string *txt = &ns;
797	{	789	{
798	map<string,string>::const_iterator it;	790	map<string,string>::const_iterator it;
799	it = docdata.find(cstr_content);	791	it = docdata.find(cstr_dj_keycontent);
800	if (it != docdata.end())	792	if (it != docdata.end())
801	txt = &it->second;	793	txt = &it->second;
802	}	794	}
803	bool setres = false;	795	bool setres = false;
804	if (newflt->is_data_input_ok(Dijon::Filter::DOCUMENT_STRING)) {	796	if (newflt->is_data_input_ok(Dijon::Filter::DOCUMENT_STRING)) {