recoll / Code / Diff of /src/internfile/mh

Diff of /src/internfile/mh_mail.cpp [2a3075] .. [e11182]

Switch to unified view


#ifndef lint
static char rcsid[] = "@(#$Id: mh_mail.cpp,v 1.14 2006-04-07 08:51:15 dockes Exp $ (C) 2005 J.F.Dockes";
#endif
/*
 *   This program is free software; you can redistribute it and/or modify
 *   it under the terms of the GNU General Public License as published by
 *   the Free Software Foundation; either version 2 of the License, or
...
static void 
walkmime(RclConfig *cnf, string &out, Binc::MimePart& doc, int depth);

MimeHandlerMail::~MimeHandlerMail()
{
    if (m_vfp) {
    fclose((FILE *)m_vfp);
    m_vfp = 0;
    }
}

// We are called for two different file types: mbox-type folders
// holding multiple messages, and maildir-type files with one message
// ipath is non empty only when we are called for retrieving a single message
// for preview. It is always empty during indexing, and we fill it up with 
// the message number for the returned doc
MimeHandler::Status 
MimeHandlerMail::mkDoc(RclConfig *cnf, const string &fn, 
            const string &mtype, Rcl::Doc &docout, string& ipath)
{
    LOGDEB2(("MimeHandlerMail::mkDoc: %s [%s]\n", mtype.c_str(), fn.c_str()));
    m_conf = cnf;

    if (!stringlowercmp("message/rfc822", mtype)) {
    ipath = "";
    int fd;
    if ((fd = open(fn.c_str(), 0)) < 0) {
...
    }
    LOGDEB2(("MimeHandlerMail::processmbox: fn %s, mtarg %d\n", fn.c_str(),
        mtarg));

    FILE *fp;
    // Open the file on first call, then save/reuse the file pointer
    if (!m_vfp) {


    fp = fopen(fn.c_str(), "r");
    if (fp == 0) {
        LOGERR(("MimeHandlerMail::processmbox: error opening %s\n", 
            fn.c_str()));
        return MimeHandler::MHError;
    }
    m_vfp = fp;
    } else {
  fp = (FILE *)m_vfp;
    }

    // If we are called to retrieve a specific message, seek to bof
    // (then scan up to the message). This is for the case where the
    // same object is reused to fetch several messages (else the fp is
    // just opened no need for a seek).  We could also check if the
    // current message number is lower than the requested one and
    // avoid rereading the whole thing in this case. But I'm not sure
    // we're ever used in this way (multiple retrieves on same
    // object).  So:
    if (mtarg > 0) {
    fseek(fp, 0, SEEK_SET);
    m_msgnum = 0;
    }

    off_t start, end;
    bool iseof = false;
    do  {
...
        if (ferror(fp) || feof(fp))
            iseof = true;
        break;
        }
    }
    m_msgnum++;
    fseek(fp, end, SEEK_SET);
    } while (mtarg > 0 && m_msgnum < mtarg);


    size_t size = end - start;
    fseek(fp, start, SEEK_SET);
    char *cp = (char *)malloc(size);
...
    doc.parseFull(s);
    MimeHandler::Status ret = processone(fn, doc, docout);
    if (ret == MimeHandler::MHError)
    return ret;
    char buf[20];
    sprintf(buf, "%d", m_msgnum);
    ipath = buf;
    return iseof ? MimeHandler::MHDone : 
    (mtarg > 0) ? MimeHandler::MHDone : MimeHandler::MHAgain;
}

...
    docout.text += string("Subject: ") + transcoded + string("\n");
    }

    LOGDEB2(("MimeHandlerMail::processone: ismultipart %d mime subtype '%s'\n",
        doc.isMultipart(), doc.getSubType().c_str()));
    walkmime(m_conf, docout.text, doc, 0);

    // LOGDEB(("MimeHandlerMail::processone: text: '%s'\n", docout.text.c_str()));
    return MimeHandler::MHDone;
}


	a/src/internfile/mh_mail.cpp		b/src/internfile/mh_mail.cpp
1	#ifndef lint	1	#ifndef lint
2	static char rcsid[] = "@(#$Id: mh_mail.cpp,v 1.13 2006-01-23 13:32:28 dockes Exp $ (C) 2005 J.F.Dockes";	2	static char rcsid[] = "@(#$Id: mh_mail.cpp,v 1.14 2006-04-07 08:51:15 dockes Exp $ (C) 2005 J.F.Dockes";
3	#endif	3	#endif
4	/*	4	/*
5	* This program is free software; you can redistribute it and/or modify	5	* This program is free software; you can redistribute it and/or modify
6	* it under the terms of the GNU General Public License as published by	6	* it under the terms of the GNU General Public License as published by
7	* the Free Software Foundation; either version 2 of the License, or	7	* the Free Software Foundation; either version 2 of the License, or
	...		...
50	static void	50	static void
51	walkmime(RclConfig *cnf, string &out, Binc::MimePart& doc, int depth);	51	walkmime(RclConfig *cnf, string &out, Binc::MimePart& doc, int depth);
52		52
53	MimeHandlerMail::~MimeHandlerMail()	53	MimeHandlerMail::~MimeHandlerMail()
54	{	54	{
55	if (vfp) {	55	if (m_vfp) {
56	fclose((FILE *)vfp);	56	fclose((FILE *)m_vfp);
57	vfp = 0;	57	m_vfp = 0;
58	}	58	}
59	}	59	}
60		60
61	// We are called for two different file types: mbox-type folders	61	// We are called for two different file types: mbox-type folders
62	// holding multiple messages, and maildir-type files with one message	62	// holding multiple messages, and maildir-type files with one message
		63	// ipath is non empty only when we are called for retrieving a single message
		64	// for preview. It is always empty during indexing, and we fill it up with
		65	// the message number for the returned doc
63	MimeHandler::Status	66	MimeHandler::Status
64	MimeHandlerMail::mkDoc(RclConfig *cnf, const string &fn,	67	MimeHandlerMail::mkDoc(RclConfig *cnf, const string &fn,
65	const string &mtype, Rcl::Doc &docout, string& ipath)	68	const string &mtype, Rcl::Doc &docout, string& ipath)
66	{	69	{
67	LOGDEB2(("MimeHandlerMail::mkDoc: %s [%s]\n", mtype.c_str(), fn.c_str()));	70	LOGDEB2(("MimeHandlerMail::mkDoc: %s [%s]\n", mtype.c_str(), fn.c_str()));
68	conf = cnf;	71	m_conf = cnf;
69		72
70	if (!stringlowercmp("message/rfc822", mtype)) {	73	if (!stringlowercmp("message/rfc822", mtype)) {
71	ipath = "";	74	ipath = "";
72	int fd;	75	int fd;
73	if ((fd = open(fn.c_str(), 0)) < 0) {	76	if ((fd = open(fn.c_str(), 0)) < 0) {
	...		...
95	}	98	}
96	LOGDEB2(("MimeHandlerMail::processmbox: fn %s, mtarg %d\n", fn.c_str(),	99	LOGDEB2(("MimeHandlerMail::processmbox: fn %s, mtarg %d\n", fn.c_str(),
97	mtarg));	100	mtarg));
98		101
99	FILE *fp;	102	FILE *fp;
		103	// Open the file on first call, then save/reuse the file pointer
100	if (vfp) {	104	if (!m_vfp) {
101	fp = (FILE *)vfp;
102	} else {
103	fp = fopen(fn.c_str(), "r");	105	fp = fopen(fn.c_str(), "r");
104	if (fp == 0) {	106	if (fp == 0) {
105	LOGERR(("MimeHandlerMail::processmbox: error opening %s\n",	107	LOGERR(("MimeHandlerMail::processmbox: error opening %s\n",
106	fn.c_str()));	108	fn.c_str()));
107	return MimeHandler::MHError;	109	return MimeHandler::MHError;
108	}	110	}
109	vfp = fp;	111	m_vfp = fp;
		112	} else {
		113	fp = (FILE *)m_vfp;
110	}	114	}
		115
		116	// If we are called to retrieve a specific message, seek to bof
		117	// (then scan up to the message). This is for the case where the
		118	// same object is reused to fetch several messages (else the fp is
		119	// just opened no need for a seek). We could also check if the
		120	// current message number is lower than the requested one and
		121	// avoid rereading the whole thing in this case. But I'm not sure
		122	// we're ever used in this way (multiple retrieves on same
		123	// object). So:
111	if (mtarg > 0) {	124	if (mtarg > 0) {
112	fseek(fp, 0, SEEK_SET);	125	fseek(fp, 0, SEEK_SET);
113	msgnum = 0;	126	m_msgnum = 0;
114	}	127	}
115		128
116	off_t start, end;	129	off_t start, end;
117	bool iseof = false;	130	bool iseof = false;
118	do {	131	do {
	...		...
140	if (ferror(fp) \|\| feof(fp))	153	if (ferror(fp) \|\| feof(fp))
141	iseof = true;	154	iseof = true;
142	break;	155	break;
143	}	156	}
144	}	157	}
145	msgnum++;	158	m_msgnum++;
146	fseek(fp, end, SEEK_SET);	159	fseek(fp, end, SEEK_SET);
147	} while (mtarg > 0 && msgnum < mtarg);	160	} while (mtarg > 0 && m_msgnum < mtarg);
148		161
149		162
150	size_t size = end - start;	163	size_t size = end - start;
151	fseek(fp, start, SEEK_SET);	164	fseek(fp, start, SEEK_SET);
152	char cp = (char )malloc(size);	165	char cp = (char )malloc(size);
	...		...
167	doc.parseFull(s);	180	doc.parseFull(s);
168	MimeHandler::Status ret = processone(fn, doc, docout);	181	MimeHandler::Status ret = processone(fn, doc, docout);
169	if (ret == MimeHandler::MHError)	182	if (ret == MimeHandler::MHError)
170	return ret;	183	return ret;
171	char buf[20];	184	char buf[20];
172	sprintf(buf, "%d", msgnum);	185	sprintf(buf, "%d", m_msgnum);
173	ipath = buf;	186	ipath = buf;
174	return iseof ? MimeHandler::MHDone :	187	return iseof ? MimeHandler::MHDone :
175	(mtarg > 0) ? MimeHandler::MHDone : MimeHandler::MHAgain;	188	(mtarg > 0) ? MimeHandler::MHDone : MimeHandler::MHAgain;
176	}	189	}
177		190
	...		...
228	docout.text += string("Subject: ") + transcoded + string("\n");	241	docout.text += string("Subject: ") + transcoded + string("\n");
229	}	242	}
230		243
231	LOGDEB2(("MimeHandlerMail::processone: ismultipart %d mime subtype '%s'\n",	244	LOGDEB2(("MimeHandlerMail::processone: ismultipart %d mime subtype '%s'\n",
232	doc.isMultipart(), doc.getSubType().c_str()));	245	doc.isMultipart(), doc.getSubType().c_str()));
233	walkmime(conf, docout.text, doc, 0);	246	walkmime(m_conf, docout.text, doc, 0);
234		247
235	// LOGDEB(("MimeHandlerMail::processone: text: '%s'\n", docout.text.c_str()));	248	// LOGDEB(("MimeHandlerMail::processone: text: '%s'\n", docout.text.c_str()));
236	return MimeHandler::MHDone;	249	return MimeHandler::MHDone;
237	}	250	}
238		251