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

Diff of /src/internfile/mh_mbox.cpp [7221b9] .. [f10e14]

Switch to unified view


...
    }
};

const size_t MboxCache::o_b1size = 1024;
static class MboxCache mcache;
static const string keyquirks("mhmboxquirks");

MimeHandlerMbox::~MimeHandlerMbox()
{
    clear();
}
...
    fseek((FILE *)m_vfp, 0, SEEK_END);
    m_fsize = ftell((FILE*)m_vfp);
    fseek((FILE*)m_vfp, 0, SEEK_SET);
    m_havedoc = true;
    m_offsets.clear();
    m_quirks = 0;

    // Check for location-based quirks:
    RclConfig *config = RclConfig::getMainConfig();
    string quirks;
    if (config && config->getConfParam(keyquirks, quirks)) {
  if (quirks == "tbird") {
      LOGDEB(("MimeHandlerMbox: setting quirks TBIRD\n"));
      m_quirks |= MBOXQUIRK_TBIRD;
  }
    }

    return true;
}

#define LL 1024
typedef char line_type[LL+10];
...
// in "Date: " header format, like:   From - Mon, 8 May 2006 10:57:32
// This was added as an alternative format. By the way it also fools "mail" and
// emacs-vm, Recoll is not alone
// Update: 2009-11-27: word after From may be quoted string: From "john bull"
static const  char *frompat =  




"^From[ ]+([^ ]+|\"[^\"]+\")[ ]+"    // 'From (toto@tutu|"john bull") '
"[[:alpha:]]{3}[ ]+[[:alpha:]]{3}[ ]+[0-3 ][0-9][ ]+" // Fri Oct 26
"[0-2][0-9]:[0-5][0-9](:[0-5][0-9])?[ ]+"             // Time, seconds optional
"([^ ]+[ ]+)?"                                        // Optional tz
"[12][0-9][0-9][0-9]"            // Year, unanchored, more data may follow
...
"^From[ ]+[^ ]+[ ]+"                                   // From toto@tutu
"[[:alpha:]]{3},[ ]+[0-3]?[0-9][ ]+[[:alpha:]]{3}[ ]+" // Mon, 8 May
"[12][0-9][0-9][0-9][ ]+"                              // Year
"[0-2][0-9]:[0-5][0-9](:[0-5][0-9])?"                  // Time, secs optional
    ;


static regex_t fromregex;
static bool regcompiled;

bool MimeHandlerMbox::next_document()
{
...
            hademptyline, ll, line));
        if (ll <= 0) {
        hademptyline = true;
        continue;
        }
      // Non empty line. If the previous one was empty, check regex
      if (hademptyline) {
      // Tbird sometimes omits the empty line, so avoid resetting
      // state (initially true) and hope for the best
      if (!(m_quirks & MBOXQUIRK_TBIRD))
          hademptyline = false;
      if (!regexec(&fromregex, line, 0, 0, 0)) {
          LOGDEB0(("MimeHandlerMbox: msgnum %d, "
           "From_ at line %d: [%s]\n", m_msgnum, m_lineno, line));
            start = ftello(fp);
            m_offsets.push_back(off_From);
            m_msgnum++;
            break;
      }
        }

    }

    // Look for next 'From ' line or eof, end of message.
    for (;;) {
        end = ftello(fp);
...
        m_lineno++;
        int ll;
        stripendnl(line, ll);
        LOGDEB2(("End: hadempty %d ll %d Line: [%s]\n", 
            hademptyline, ll, line));
      if (hademptyline) {
      if (ll > 0) {
          if (!(m_quirks & MBOXQUIRK_TBIRD))
          hademptyline = false;
         if (!regexec(&fromregex, line, 0, 0, 0)) {
         // Rewind to start of "From " line
         fseek(fp, end, SEEK_SET);
         m_lineno--;
         break;
          }
      }
      } else if (ll <= 0) {
      hademptyline = true;
        }
        if (mtarg <= 0 || m_msgnum == mtarg) {
        line[ll] = '\n';
        line[ll+1] = 0;
        msgtxt += line;





        }
    }

    } while (mtarg > 0 && m_msgnum < mtarg);


	a/src/internfile/mh_mbox.cpp		b/src/internfile/mh_mbox.cpp
	...		...
223	}	223	}
224	};	224	};
225		225
226	const size_t MboxCache::o_b1size = 1024;	226	const size_t MboxCache::o_b1size = 1024;
227	static class MboxCache mcache;	227	static class MboxCache mcache;
		228	static const string keyquirks("mhmboxquirks");
228		229
229	MimeHandlerMbox::~MimeHandlerMbox()	230	MimeHandlerMbox::~MimeHandlerMbox()
230	{	231	{
231	clear();	232	clear();
232	}	233	}
	...		...
263	fseek((FILE *)m_vfp, 0, SEEK_END);	264	fseek((FILE *)m_vfp, 0, SEEK_END);
264	m_fsize = ftell((FILE*)m_vfp);	265	m_fsize = ftell((FILE*)m_vfp);
265	fseek((FILE*)m_vfp, 0, SEEK_SET);	266	fseek((FILE*)m_vfp, 0, SEEK_SET);
266	m_havedoc = true;	267	m_havedoc = true;
267	m_offsets.clear();	268	m_offsets.clear();
		269	m_quirks = 0;
		270
		271	// Check for location-based quirks:
		272	RclConfig *config = RclConfig::getMainConfig();
		273	string quirks;
		274	if (config && config->getConfParam(keyquirks, quirks)) {
		275	if (quirks == "tbird") {
		276	LOGDEB(("MimeHandlerMbox: setting quirks TBIRD\n"));
		277	m_quirks \|= MBOXQUIRK_TBIRD;
		278	}
		279	}
		280
268	return true;	281	return true;
269	}	282	}
270		283
271	#define LL 1024	284	#define LL 1024
272	typedef char line_type[LL+10];	285	typedef char line_type[LL+10];
	...		...
323	// in "Date: " header format, like: From - Mon, 8 May 2006 10:57:32	336	// in "Date: " header format, like: From - Mon, 8 May 2006 10:57:32
324	// This was added as an alternative format. By the way it also fools "mail" and	337	// This was added as an alternative format. By the way it also fools "mail" and
325	// emacs-vm, Recoll is not alone	338	// emacs-vm, Recoll is not alone
326	// Update: 2009-11-27: word after From may be quoted string: From "john bull"	339	// Update: 2009-11-27: word after From may be quoted string: From "john bull"
327	static const char *frompat =	340	static const char *frompat =
328	#if 0 //1.9.0
329	"^From .* [1-2][0-9][0-9][0-9]$";
330	#endif
331	#if 1
332	"^From[ ]+([^ ]+\|\"[^\"]+\")[ ]+" // 'From (toto@tutu\|"john bull") '	341	"^From[ ]+([^ ]+\|\"[^\"]+\")[ ]+" // 'From (toto@tutu\|"john bull") '
333	"[[:alpha:]]{3}[ ]+[[:alpha:]]{3}[ ]+[0-3 ][0-9][ ]+" // Fri Oct 26	342	"[[:alpha:]]{3}[ ]+[[:alpha:]]{3}[ ]+[0-3 ][0-9][ ]+" // Fri Oct 26
334	"[0-2][0-9]:[0-5][0-9](:[0-5][0-9])?[ ]+" // Time, seconds optional	343	"[0-2][0-9]:[0-5][0-9](:[0-5][0-9])?[ ]+" // Time, seconds optional
335	"([^ ]+[ ]+)?" // Optional tz	344	"([^ ]+[ ]+)?" // Optional tz
336	"[12][0-9][0-9][0-9]" // Year, unanchored, more data may follow	345	"[12][0-9][0-9][0-9]" // Year, unanchored, more data may follow
	...		...
338	"^From[ ]+[^ ]+[ ]+" // From toto@tutu	347	"^From[ ]+[^ ]+[ ]+" // From toto@tutu
339	"[[:alpha:]]{3},[ ]+[0-3]?[0-9][ ]+[[:alpha:]]{3}[ ]+" // Mon, 8 May	348	"[[:alpha:]]{3},[ ]+[0-3]?[0-9][ ]+[[:alpha:]]{3}[ ]+" // Mon, 8 May
340	"[12][0-9][0-9][0-9][ ]+" // Year	349	"[12][0-9][0-9][0-9][ ]+" // Year
341	"[0-2][0-9]:[0-5][0-9](:[0-5][0-9])?" // Time, secs optional	350	"[0-2][0-9]:[0-5][0-9](:[0-5][0-9])?" // Time, secs optional
342	;	351	;
343	#endif	352
344	// "([ ]+[-+][0-9]{4})?$"
345	static regex_t fromregex;	353	static regex_t fromregex;
346	static bool regcompiled;	354	static bool regcompiled;
347		355
348	bool MimeHandlerMbox::next_document()	356	bool MimeHandlerMbox::next_document()
349	{	357	{
	...		...
423	hademptyline, ll, line));	431	hademptyline, ll, line));
424	if (ll <= 0) {	432	if (ll <= 0) {
425	hademptyline = true;	433	hademptyline = true;
426	continue;	434	continue;
427	}	435	}
		436	// Non empty line. If the previous one was empty, check regex
		437	if (hademptyline) {
		438	// Tbird sometimes omits the empty line, so avoid resetting
		439	// state (initially true) and hope for the best
		440	if (!(m_quirks & MBOXQUIRK_TBIRD))
		441	hademptyline = false;
428	if (hademptyline && !regexec(&fromregex, line, 0, 0, 0)) {	442	if (!regexec(&fromregex, line, 0, 0, 0)) {
429	LOGDEB0(("MimeHandlerMbox: msgnum %d, From_ at line %d: [%s]\n",	443	LOGDEB0(("MimeHandlerMbox: msgnum %d, "
430	m_msgnum, m_lineno, line));	444	"From_ at line %d: [%s]\n", m_msgnum, m_lineno, line));
431	start = ftello(fp);	445	start = ftello(fp);
432	m_offsets.push_back(off_From);	446	m_offsets.push_back(off_From);
433	m_msgnum++;	447	m_msgnum++;
434	break;	448	break;
		449	}
435	}	450	}
436	hademptyline = false;
437	}	451	}
438		452
439	// Look for next 'From ' line or eof, end of message.	453	// Look for next 'From ' line or eof, end of message.
440	for (;;) {	454	for (;;) {
441	end = ftello(fp);	455	end = ftello(fp);
	...		...
447	m_lineno++;	461	m_lineno++;
448	int ll;	462	int ll;
449	stripendnl(line, ll);	463	stripendnl(line, ll);
450	LOGDEB2(("End: hadempty %d ll %d Line: [%s]\n",	464	LOGDEB2(("End: hadempty %d ll %d Line: [%s]\n",
451	hademptyline, ll, line));	465	hademptyline, ll, line));
		466	if (hademptyline) {
		467	if (ll > 0) {
		468	if (!(m_quirks & MBOXQUIRK_TBIRD))
		469	hademptyline = false;
452	if (hademptyline && !regexec(&fromregex, line, 0, 0, 0)) {	470	if (!regexec(&fromregex, line, 0, 0, 0)) {
453	// Rewind to start of "From " line	471	// Rewind to start of "From " line
454	fseek(fp, end, SEEK_SET);	472	fseek(fp, end, SEEK_SET);
455	m_lineno--;	473	m_lineno--;
456	break;	474	break;
		475	}
		476	}
		477	} else if (ll <= 0) {
		478	hademptyline = true;
457	}	479	}
458	if (mtarg <= 0 \|\| m_msgnum == mtarg) {	480	if (mtarg <= 0 \|\| m_msgnum == mtarg) {
459	line[ll] = '\n';	481	line[ll] = '\n';
460	line[ll+1] = 0;	482	line[ll+1] = 0;
461	msgtxt += line;	483	msgtxt += line;
462	}
463	if (ll <= 0) {
464	hademptyline = true;
465	} else {
466	hademptyline = false;
467	}	484	}
468	}	485	}
469		486
470	} while (mtarg > 0 && m_msgnum < mtarg);	487	} while (mtarg > 0 && m_msgnum < mtarg);
471		488