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

Diff of /src/internfile/internfile.cpp [63a29c] .. [d392d3]

Switch to unified view


#ifndef lint
static char rcsid[] = "@(#$Id: internfile.cpp,v 1.4 2005-03-25 09:40:27 dockes Exp $ (C) 2004 J.F.Dockes";
#endif
#include <unistd.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <errno.h>
...
    if (!tfile.empty() && tfile[tfile.length() - 1] == '\n')
    tfile.erase(tfile.length() - 1, 1);
    return true;
}

void FileInterner::tmpcleanup()
{
    if (tdir.empty() || tfile.empty())
    return;
    if (unlink(tfile.c_str()) < 0) {
    LOGERR(("FileInterner::tmpcleanup: unlink(%s) errno %d\n", 
      tfile.c_str(), errno));
    return;
    }
}

// Handler==0 on return says we're in error
FileInterner::FileInterner(const std::string &f, RclConfig *cnf, 
             const string& td)
    : fn(f), config(cnf), tdir(td), handler(0) 
{





    mime = mimetype(fn, config->getMimeMap());
    if (mime.empty()) {
    // No mime type: not listed in our map.
    LOGDEB(("FileInterner::FileInterner: (no mime) [%s]\n", fn.c_str()));
    return;
    }

    // First check for a compressed file. If so, create a temporary
    // uncompressed file, and rerun the mime type identification, then do the
    // rest with the temp file.
    list<string>ucmd;
    if (getUncompressor(mime, config->getMimeConf(), ucmd)) {
    if (!uncompressfile(config, fn, ucmd, tdir, tfile)) {
        return;
  }
    LOGDEB(("internfile: after ucomp: tdir %s, tfile %s\n", 
        tdir.c_str(), tfile.c_str()));
    fn = tfile;
    mime = mimetype(fn, config->getMimeMap());
    if (mime.empty()) {
        // No mime type ?? pass on.
        LOGDEB(("internfile: (no mime) [%s]\n", fn.c_str()));
      return;
    }

    }


    // Look for appropriate handler
    handler = getMimeHandler(mime, config->getMimeConf());
    if (!handler) {
    // No handler for this type, for now :(
    LOGDEB(("FileInterner::FileInterner: %s: no handler\n", mime.c_str()));
  return;
    }

    LOGDEB(("FileInterner::FileInterner: %s [%s]\n",mime.c_str(), fn.c_str()));
}

FileInterner::Status FileInterner::internfile(Rcl::Doc& doc, string& ipath)
{
    if (!handler)
  return FIError;

    // Turn file into a document. The document has fields for title, body 
    // etc.,  all text converted to utf8
    MimeHandler::Status mhs = handler->worker(config, fn,  mime, doc, ipath);
    FileInterner::Status ret = FIError;
    switch (mhs) {
    case MimeHandler::MHError: break;
    case MimeHandler::MHDone: ret = FIDone;break;
    case MimeHandler::MHAgain: ret = FIAgain;break;
    }

    doc.mimetype = mime;






    return ret;
}

FileInterner::~FileInterner()
{
    delete handler; 
    handler = 0;
    tmpcleanup();
}

	a/src/internfile/internfile.cpp		b/src/internfile/internfile.cpp
1	#ifndef lint	1	#ifndef lint
2	static char rcsid[] = "@(#$Id: internfile.cpp,v 1.3 2005-03-17 15:35:49 dockes Exp $ (C) 2004 J.F.Dockes";	2	static char rcsid[] = "@(#$Id: internfile.cpp,v 1.4 2005-03-25 09:40:27 dockes Exp $ (C) 2004 J.F.Dockes";
3	#endif	3	#endif
4	#include <unistd.h>	4	#include <unistd.h>
5	#include <sys/types.h>	5	#include <sys/types.h>
6	#include <sys/stat.h>	6	#include <sys/stat.h>
7	#include <errno.h>	7	#include <errno.h>
	...		...
68	if (!tfile.empty() && tfile[tfile.length() - 1] == '\n')	68	if (!tfile.empty() && tfile[tfile.length() - 1] == '\n')
69	tfile.erase(tfile.length() - 1, 1);	69	tfile.erase(tfile.length() - 1, 1);
70	return true;	70	return true;
71	}	71	}
72		72
73	static void tmpcleanup(const string& tdir, const string& tfile)	73	void FileInterner::tmpcleanup()
74	{	74	{
75	if (tdir.empty() \|\| tfile.empty())	75	if (tdir.empty() \|\| tfile.empty())
76	return;	76	return;
77	if (unlink(tfile.c_str()) < 0) {	77	if (unlink(tfile.c_str()) < 0) {
78	LOGERR(("tmpcleanup: unlink(%s) errno %d\n", tfile.c_str(),	78	LOGERR(("FileInterner::tmpcleanup: unlink(%s) errno %d\n",
79	errno));	79	tfile.c_str(), errno));
80	return;	80	return;
81	}	81	}
82	}	82	}
83		83
84	bool internfile(const std::string &ifn, RclConfig *config, Rcl::Doc& doc,	84	// Handler==0 on return says we're in error
		85	FileInterner::FileInterner(const std::string &f, RclConfig *cnf,
85	const string& tdir)	86	const string& td)
		87	: fn(f), config(cnf), tdir(td), handler(0)
86	{	88	{
87	string fn = ifn;
88	string tfile;
89	MimeHandler *handler = 0;
90	bool ret = false;
91
92	string mime = mimetype(fn, config->getMimeMap());	89	mime = mimetype(fn, config->getMimeMap());
93	if (mime.empty()) {	90	if (mime.empty()) {
94	// No mime type: not listed in our map.	91	// No mime type: not listed in our map.
95	LOGDEB(("internfile: (no mime) [%s]\n", fn.c_str()));	92	LOGDEB(("FileInterner::FileInterner: (no mime) [%s]\n", fn.c_str()));
96	return false;	93	return;
97	}	94	}
98		95
99	// First check for a compressed file. If so, create a temporary	96	// First check for a compressed file. If so, create a temporary
100	// uncompressed file, and rerun the mime type identification, then do the	97	// uncompressed file, and rerun the mime type identification, then do the
101	// rest with the temp file.	98	// rest with the temp file.
102	list<string>ucmd;	99	list<string>ucmd;
103	if (getUncompressor(mime, config->getMimeConf(), ucmd)) {	100	if (getUncompressor(mime, config->getMimeConf(), ucmd)) {
104	if (!uncompressfile(config, fn, ucmd, tdir, tfile))	101	if (!uncompressfile(config, fn, ucmd, tdir, tfile)) {
105	return false;	102	return;
		103	}
106	LOGDEB(("internfile: after ucomp: tdir %s, tfile %s\n",	104	LOGDEB(("internfile: after ucomp: tdir %s, tfile %s\n",
107	tdir.c_str(), tfile.c_str()));	105	tdir.c_str(), tfile.c_str()));
108	fn = tfile;	106	fn = tfile;
109	mime = mimetype(fn, config->getMimeMap());	107	mime = mimetype(fn, config->getMimeMap());
110	if (mime.empty()) {	108	if (mime.empty()) {
111	// No mime type ?? pass on.	109	// No mime type ?? pass on.
112	LOGDEB(("internfile: (no mime) [%s]\n", fn.c_str()));	110	LOGDEB(("internfile: (no mime) [%s]\n", fn.c_str()));
113	goto out;	111	return;
114	}	112	}
115
116	}	113	}
117
118		114
119	// Look for appropriate handler	115	// Look for appropriate handler
120	handler = getMimeHandler(mime, config->getMimeConf());	116	handler = getMimeHandler(mime, config->getMimeConf());
121	if (!handler) {	117	if (!handler) {
122	// No handler for this type, for now :(	118	// No handler for this type, for now :(
123	LOGDEB(("internfile: %s : no handler\n", mime.c_str()));	119	LOGDEB(("FileInterner::FileInterner: %s: no handler\n", mime.c_str()));
124	goto out;	120	return;
125	}	121	}
126		122
127	LOGDEB(("internfile: %s [%s]\n", mime.c_str(), fn.c_str()));	123	LOGDEB(("FileInterner::FileInterner: %s [%s]\n",mime.c_str(), fn.c_str()));
		124	}
		125
		126	FileInterner::Status FileInterner::internfile(Rcl::Doc& doc, string& ipath)
		127	{
		128	if (!handler)
		129	return FIError;
128		130
129	// Turn file into a document. The document has fields for title, body	131	// Turn file into a document. The document has fields for title, body
130	// etc., all text converted to utf8	132	// etc., all text converted to utf8
131	if (!handler->worker(config, fn, mime, doc)) {	133	MimeHandler::Status mhs = handler->worker(config, fn, mime, doc, ipath);
132	goto out;	134	FileInterner::Status ret = FIError;
		135	switch (mhs) {
		136	case MimeHandler::MHError: break;
		137	case MimeHandler::MHDone: ret = FIDone;break;
		138	case MimeHandler::MHAgain: ret = FIAgain;break;
133	}	139	}
		140
134	doc.mimetype = mime;	141	doc.mimetype = mime;
135
136	// Clean up. We delete the temp file and its father directory
137	ret = true;
138	out:
139	delete handler;
140	tmpcleanup(tdir, tfile);
141	return ret;	142	return ret;
142	}	143	}
		144
		145	FileInterner::~FileInterner()
		146	{
		147	delete handler;
		148	handler = 0;
		149	tmpcleanup();
		150	}