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

Diff of /src/internfile/internfile.cpp [ce2d3c] .. [9b44f9]

Switch to unified view


...
#include "execmd.h"
#include "pathut.h"
#include "wipedir.h"
#include "rclconfig.h"
#include "mh_html.h"
#ifdef RCL_USE_XATTR
#include "pxattr.h"
#endif // RCL_USE_XATTR

// The internal path element separator. This can't be the same as the rcldb 
// file to ipath separator : "|"
static const string isep(":");
static const string stxtplain("text/plain");

set<string> FileInterner::o_missingExternal;
map<string, set<string> >  FileInterner::o_typesForMissing;

#ifdef RCL_USE_XATTR
void FileInterner::reapXAttrs(const string& path)
{
    vector<string> xnames;
    if (!pxattr::list(path, &xnames)) {
  LOGERR(("FileInterner::reapXattrs: pxattr::list: errno %d\n", errno));
  return;
    }
    const map<string, string>& xtof = m_cfg->getXattrToField();
    for (vector<string>::const_iterator it = xnames.begin();
   it != xnames.end(); it++) {
  map<string, string>::const_iterator mit;
  if ((mit = xtof.find(*it)) != xtof.end()) {
      string value;
      if (!pxattr::get(path, *it, &value, pxattr::PXATTR_NOFOLLOW)) {
      LOGERR(("FileInterner::reapXattrs: pxattr::get failed"
          "for %s, errno %d\n", (*it).c_str(), errno));
      continue;
      }
      // Encode should we ?
      m_XAttrsFields[mit->second] = value;
  }
    }
}
#endif // RCL_USE_XATTR

// This is used when the user wants to retrieve a search result doc's parent
// (ie message having a given attachment)
bool FileInterner::getEnclosing(const string &url, const string &ipath,
                string &eurl, string &eipath)
...
    Dijon::Filter *df = getMimeHandler(l_mime, m_cfg, !m_forPreview);

    if (!df) {
    // No handler for this type, for now :( if indexallfilenames
    // is set in the config, this normally wont happen (we get mh_unknown)
    LOGERR(("FileInterner:: ignored: [%s] mime [%s]\n", 
      f.c_str(), l_mime.c_str()));
    return;
    }
    df->set_property(Dijon::Filter::OPERATING_MODE, 
                m_forPreview ? "view" : "index");

    string charset = m_cfg->getDefCharset();
    df->set_property(Dijon::Filter::DEFAULT_CHARSET, charset);

#ifdef RCL_USE_XATTR
    // Get fields computed from extended attributes. We use the
    // original file, not the m_fn which may be the uncompressed temp
    // file
    reapXAttrs(f);
#endif //RCL_USE_XATTR

    if (!df->set_document_file(m_fn)) {
    LOGERR(("FileInterner:: error parsing %s\n", m_fn.c_str()));
    return;
    }
    m_handlers.reserve(MAXHANDLERS);
...
    }
    if (doc.meta[Rcl::Doc::keyabs].empty() && !doc.meta[keyds].empty()) {
    doc.meta[Rcl::Doc::keyabs] = doc.meta[keyds];
    doc.meta.erase(keyds);
    }












    return true;
}

// Collect the ipath from the current path in the document tree.
// While we're at it, we also set the mimetype and filename, which are special 
...
// We also set the author and modification time from the last doc
// which has them.
void FileInterner::collectIpathAndMT(Rcl::Doc& doc, string& ipath) const
{
    bool hasipath = false;

#ifdef RCL_USE_XATTR
    // Set fields from extended file attributes.
    // These can be overriden by values from inside the file
    for (map<string,string>::const_iterator it = m_XAttrsFields.begin(); 
   it != m_XAttrsFields.end(); it++) {
  doc.meta[it->first] = it->second;
    }
#endif //RCL_USE_XATTR

    // If there is no ipath stack, the mimetype is the one from the file
    doc.mimetype = m_mimetype;

    string ipathel;
...
        LOGERR(("FileInterner::internfile: can't skip\n"));
        return FIError;
    }
    }
 breakloop:

    if (m_handlers.empty()) {
    LOGDEB(("FileInterner::internfile: conversion ended with no doc\n"));
    return FIError;
    }

    // If indexing compute ipath and significant mimetype.  ipath is
    // returned through the parameter not doc.ipath We also retrieve
    // some metadata fields from the ancesters (like date or
    // author). This is useful for email attachments. The values will
    // be replaced by those internal to the document (by dijontorcl())
    // if any, so the order of calls is important.
    if (!m_forPreview)
    collectIpathAndMT(doc, ipath);
    else
    doc.mimetype = m_reachedMType;


	a/src/internfile/internfile.cpp		b/src/internfile/internfile.cpp
	...		...
42	#include "execmd.h"	42	#include "execmd.h"
43	#include "pathut.h"	43	#include "pathut.h"
44	#include "wipedir.h"	44	#include "wipedir.h"
45	#include "rclconfig.h"	45	#include "rclconfig.h"
46	#include "mh_html.h"	46	#include "mh_html.h"
		47	#ifdef RCL_USE_XATTR
		48	#include "pxattr.h"
		49	#endif // RCL_USE_XATTR
47		50
48	// The internal path element separator. This can't be the same as the rcldb	51	// The internal path element separator. This can't be the same as the rcldb
49	// file to ipath separator : "\|"	52	// file to ipath separator : "\|"
50	static const string isep(":");	53	static const string isep(":");
51	static const string stxtplain("text/plain");	54	static const string stxtplain("text/plain");
52		55
53	set<string> FileInterner::o_missingExternal;	56	set<string> FileInterner::o_missingExternal;
54	map<string, set<string> > FileInterner::o_typesForMissing;	57	map<string, set<string> > FileInterner::o_typesForMissing;
		58
		59	#ifdef RCL_USE_XATTR
		60	void FileInterner::reapXAttrs(const string& path)
		61	{
		62	vector<string> xnames;
		63	if (!pxattr::list(path, &xnames)) {
		64	LOGERR(("FileInterner::reapXattrs: pxattr::list: errno %d\n", errno));
		65	return;
		66	}
		67	const map<string, string>& xtof = m_cfg->getXattrToField();
		68	for (vector<string>::const_iterator it = xnames.begin();
		69	it != xnames.end(); it++) {
		70	map<string, string>::const_iterator mit;
		71	if ((mit = xtof.find(*it)) != xtof.end()) {
		72	string value;
		73	if (!pxattr::get(path, *it, &value, pxattr::PXATTR_NOFOLLOW)) {
		74	LOGERR(("FileInterner::reapXattrs: pxattr::get failed"
		75	"for %s, errno %d\n", (*it).c_str(), errno));
		76	continue;
		77	}
		78	// Encode should we ?
		79	m_XAttrsFields[mit->second] = value;
		80	}
		81	}
		82	}
		83	#endif // RCL_USE_XATTR
55		84
56	// This is used when the user wants to retrieve a search result doc's parent	85	// This is used when the user wants to retrieve a search result doc's parent
57	// (ie message having a given attachment)	86	// (ie message having a given attachment)
58	bool FileInterner::getEnclosing(const string &url, const string &ipath,	87	bool FileInterner::getEnclosing(const string &url, const string &ipath,
59	string &eurl, string &eipath)	88	string &eurl, string &eipath)
	...		...
191	Dijon::Filter *df = getMimeHandler(l_mime, m_cfg, !m_forPreview);	220	Dijon::Filter *df = getMimeHandler(l_mime, m_cfg, !m_forPreview);
192		221
193	if (!df) {	222	if (!df) {
194	// No handler for this type, for now :( if indexallfilenames	223	// No handler for this type, for now :( if indexallfilenames
195	// is set in the config, this normally wont happen (we get mh_unknown)	224	// is set in the config, this normally wont happen (we get mh_unknown)
196	LOGERR(("FileInterner:: ignored: [%s] mime [%s]\n", f.c_str(), l_mime.c_str()));	225	LOGERR(("FileInterner:: ignored: [%s] mime [%s]\n",
		226	f.c_str(), l_mime.c_str()));
197	return;	227	return;
198	}	228	}
199	df->set_property(Dijon::Filter::OPERATING_MODE,	229	df->set_property(Dijon::Filter::OPERATING_MODE,
200	m_forPreview ? "view" : "index");	230	m_forPreview ? "view" : "index");
201		231
202	string charset = m_cfg->getDefCharset();	232	string charset = m_cfg->getDefCharset();
203	df->set_property(Dijon::Filter::DEFAULT_CHARSET, charset);	233	df->set_property(Dijon::Filter::DEFAULT_CHARSET, charset);
		234
		235	#ifdef RCL_USE_XATTR
		236	// Get fields computed from extended attributes. We use the
		237	// original file, not the m_fn which may be the uncompressed temp
		238	// file
		239	reapXAttrs(f);
		240	#endif //RCL_USE_XATTR
		241
204	if (!df->set_document_file(m_fn)) {	242	if (!df->set_document_file(m_fn)) {
205	LOGERR(("FileInterner:: error parsing %s\n", m_fn.c_str()));	243	LOGERR(("FileInterner:: error parsing %s\n", m_fn.c_str()));
206	return;	244	return;
207	}	245	}
208	m_handlers.reserve(MAXHANDLERS);	246	m_handlers.reserve(MAXHANDLERS);
	...		...
361	}	399	}
362	if (doc.meta[Rcl::Doc::keyabs].empty() && !doc.meta[keyds].empty()) {	400	if (doc.meta[Rcl::Doc::keyabs].empty() && !doc.meta[keyds].empty()) {
363	doc.meta[Rcl::Doc::keyabs] = doc.meta[keyds];	401	doc.meta[Rcl::Doc::keyabs] = doc.meta[keyds];
364	doc.meta.erase(keyds);	402	doc.meta.erase(keyds);
365	}	403	}
366	#ifdef RCL_USE_XATTR
367	// Finally set any data possibly coming out of the extended file attributes
368	// these override any values from inside the file.
369	RecollFilter rf = dynamic_cast<RecollFilter>(df);
370	if (rf != 0) {
371	const map<string, string>& ffa = rf->getFieldsFromAttrs();
372	for (map<string,string>::const_iterator it = ffa.begin();
373	it != ffa.end(); it++) {
374	doc.meta[it->first] = it->second;
375	}
376	}
377	#endif //RCL_USE_XATTR
378	return true;	404	return true;
379	}	405	}
380		406
381	// Collect the ipath from the current path in the document tree.	407	// Collect the ipath from the current path in the document tree.
382	// While we're at it, we also set the mimetype and filename, which are special	408	// While we're at it, we also set the mimetype and filename, which are special
	...		...
385	// We also set the author and modification time from the last doc	411	// We also set the author and modification time from the last doc
386	// which has them.	412	// which has them.
387	void FileInterner::collectIpathAndMT(Rcl::Doc& doc, string& ipath) const	413	void FileInterner::collectIpathAndMT(Rcl::Doc& doc, string& ipath) const
388	{	414	{
389	bool hasipath = false;	415	bool hasipath = false;
		416
		417	#ifdef RCL_USE_XATTR
		418	// Set fields from extended file attributes.
		419	// These can be overriden by values from inside the file
		420	for (map<string,string>::const_iterator it = m_XAttrsFields.begin();
		421	it != m_XAttrsFields.end(); it++) {
		422	doc.meta[it->first] = it->second;
		423	}
		424	#endif //RCL_USE_XATTR
390		425
391	// If there is no ipath stack, the mimetype is the one from the file	426	// If there is no ipath stack, the mimetype is the one from the file
392	doc.mimetype = m_mimetype;	427	doc.mimetype = m_mimetype;
393		428
394	string ipathel;	429	string ipathel;
	...		...
608	LOGERR(("FileInterner::internfile: can't skip\n"));	643	LOGERR(("FileInterner::internfile: can't skip\n"));
609	return FIError;	644	return FIError;
610	}	645	}
611	}	646	}
612	breakloop:	647	breakloop:
613
614	if (m_handlers.empty()) {	648	if (m_handlers.empty()) {
615	LOGDEB(("FileInterner::internfile: conversion ended with no doc\n"));	649	LOGDEB(("FileInterner::internfile: conversion ended with no doc\n"));
616	return FIError;	650	return FIError;
617	}	651	}
618		652
619	// If indexing compute ipath and significant mimetype.	653	// If indexing compute ipath and significant mimetype. ipath is
620	// ipath is returned through the parameter not doc.ipath We also	654	// returned through the parameter not doc.ipath We also retrieve
621	// retrieve some metadata fields from the ancesters (like date or	655	// some metadata fields from the ancesters (like date or
622	// author). This is useful for email attachments. The values will	656	// author). This is useful for email attachments. The values will
623	// be replaced by those found by dijontorcl if any, so the order	657	// be replaced by those internal to the document (by dijontorcl())
624	// of calls is important.	658	// if any, so the order of calls is important.
625	if (!m_forPreview)	659	if (!m_forPreview)
626	collectIpathAndMT(doc, ipath);	660	collectIpathAndMT(doc, ipath);
627	else	661	else
628	doc.mimetype = m_reachedMType;	662	doc.mimetype = m_reachedMType;
629		663