-/*
- * Routines to encode message headers using RFC 2047-encoding.
+/* encode_rfc2047.c -- encode message headers using RFC 2047 encoding.
*
* This code is Copyright (c) 2002, by the authors of nmh. See the
* COPYRIGHT file in the root directory of the nmh distribution for
#include <h/mhparse.h>
#include <h/addrsbr.h>
#include <h/utils.h>
+#include "base64.h"
+#include "unquote.h"
/*
* List of headers that contain addresses and as a result require special
charset = write_charset_8bit();
if (strcasecmp(charset, "US-ASCII") == 0) {
- advise(NULL, "Cannot use US-ASCII with 8 bit characters in header");
+ inform("Cannot use US-ASCII with 8 bit characters in header");
return 1;
}
* On the encoding we choose, and the specifics of encoding:
*
* - If a specified encoding is passed in, we use that.
- * - If more than 50% of the characters are high-bit, we use base64
- * and encode the whole field as one atom (possibly split).
- * - Otherwise, we use quoted-printable.
+ * - Otherwise, pick which encoding is shorter.
+ *
+ * We don't quite handle continuation right here, but it should be
+ * pretty close.
*/
if (encoding == CE_UNKNOWN)
- encoding = (eightbitcount * 10 / (asciicount + eightbitcount) > 5) ?
- CE_BASE64 : CE_QUOTED;
+ encoding = pref_encoding(asciicount, qpspecialcount, eightbitcount);
unfold_header(value, asciicount + eightbitcount);
eightbitcount + qpspecialcount, 0);
default:
- advise(NULL, "Internal error: unknown RFC-2047 encoding type");
+ inform("Internal error: unknown RFC-2047 encoding type");
return 1;
}
}
{
int prefixlen = name ? strlen(name) + 2: 0, outlen = 0, column, newline = 1;
int charsetlen = strlen(charset), utf8;
- char *output = NULL, *p, *q;
+ char *output = NULL, *p, *q = NULL;
/*
* Right now we just encode the whole thing. Maybe later on we'll
* allow for the encoded output.
*/
if (column + (utf8len(p) * 3) > ENCODELINELIMIT - 2) {
- newline = 1;
+ newline = 1;
}
}
}
- strcat(q, "?=");
+ if (q == NULL) {
+ /* This should never happen, but just in case. Found by
+ clang static analyzer. */
+ inform("null output encoding for %s, continuing...", *value);
+ return 1;
+ }
+ *q++ = '?';
+ *q++ = '=';
if (prefixlen)
- strcat(q, "\n");
+ *q++ = '\n';
+
+ *q = '\0';
free(*value);
{
int prefixlen = name ? strlen(name) + 2 : 0, charsetlen = strlen(charset);
int outlen = 0, numencode, curlen;
- char *output = NULL, *p = *value, *q = NULL, *linestart;
+ char *output = NULL, *p = *value, *q = NULL, *linestart = NULL;
/*
* Skip over any leading white space.
numencode = strbase64(ENCODELINELIMIT - (q - linestart) - 2);
if (numencode <= 0) {
- advise(NULL, "Internal error: tried to encode %d characters "
+ inform("Internal error: tried to encode %d characters "
"in base64", numencode);
return 1;
}
numencode--;
if (numencode == 0) {
- advise(NULL, "Internal error: could not find start of "
+ inform("Internal error: could not find start of "
"UTF-8 character when base64 encoding header");
return 1;
}
if (writeBase64raw((unsigned char *) p, numencode,
(unsigned char *) q) != OK) {
- advise(NULL, "Internal error: base64 encoding of header failed");
+ inform("Internal error: base64 encoding of header failed");
return 1;
}
if (writeBase64raw((unsigned char *) p, strlen(p),
(unsigned char *) q) != OK) {
- advise(NULL, "Internal error: base64 encoding of header failed");
+ inform("Internal error: base64 encoding of header failed");
return 1;
}
* This has the side effect of stripping off the final newline
* for the header; we put it back in the encoding routine.
*/
- while (is_fws(*q++))
- ;
+ while (is_fws(*q))
+ q++;
if (*q == '\0')
break;
{
int prefixlen = strlen(name) + 2, column = prefixlen, groupflag;
int asciichars, specialchars, eightbitchars, reformat = 0, errflag = 0;
- int retval;
size_t len;
char *mp, *cp = NULL, *output = NULL;
char *tmpbuf = NULL;
size_t tmpbufsize = 0;
struct mailname *mn;
+ char errbuf[BUFSIZ];
/*
* Because these are addresses, we need to handle them individually.
output = add(" ", output);
for (groupflag = 0; (mp = getname(*value)); ) {
- if ((mn = getm(mp, NULL, 0, AD_HOST, NULL)) == NULL) {
+ if ((mn = getm(mp, NULL, 0, errbuf, sizeof(errbuf))) == NULL) {
+ inform("%s: %s", errbuf, mp);
errflag++;
continue;
}
switch (encoding) {
case CE_BASE64:
- retval = field_encode_base64(NULL, &mn->m_pers, charset);
+ if (field_encode_base64(NULL, &mn->m_pers, charset)) {
+ errflag++;
+ goto out;
+ }
break;
case CE_QUOTED:
- retval = field_encode_quoted(NULL, &mn->m_pers, charset,
- asciichars,
- eightbitchars + specialchars, 1);
+ if (field_encode_quoted(NULL, &mn->m_pers, charset, asciichars,
+ eightbitchars + specialchars, 1)) {
+ errflag++;
+ goto out;
+ }
break;
default:
- advise(NULL, "Internal error: unknown RFC-2047 encoding type");
+ inform("Internal error: unknown RFC-2047 encoding type");
errflag++;
goto out;
}
if (! mn->m_note)
goto do_reformat;
- len = strlen(mn->m_note);
-
- if ((len = strlen(mn->m_pers)) + 1 > tmpbufsize) {
+ if ((len = strlen(mn->m_note)) + 1 > tmpbufsize) {
tmpbuf = mh_xrealloc(tmpbuf, tmpbufsize = len + 1);
}
if (mn->m_note[0] != '(' || mn->m_note[len - 1] != ')') {
- advise(NULL, "Internal error: Invalid note field \"%s\"",
+ inform("Internal error: Invalid note field \"%s\"",
mn->m_note);
errflag++;
goto out;
switch (encoding) {
case CE_BASE64:
- retval = field_encode_base64(NULL, &tmpbuf, charset);
+ if (field_encode_base64(NULL, &tmpbuf, charset)) {
+ errflag++;
+ goto out;
+ }
break;
case CE_QUOTED:
- retval = field_encode_quoted(NULL, &tmpbuf, charset,
- asciichars,
- eightbitchars + specialchars, 1);
+ if (field_encode_quoted(NULL, &tmpbuf, charset, asciichars,
+ eightbitchars + specialchars, 1)) {
+ errflag++;
+ goto out;
+ }
break;
default:
- advise(NULL, "Internal error: unknown RFC-2047 encoding type");
+ inform("Internal error: unknown RFC-2047 encoding type");
errflag++;
goto out;
}
* we can use m_text directly.
*/
+ /*
+ * If we were in a group but are no longer, make sure we add a
+ * semicolon (which needs to be FIRST, as it needs to be at the end
+ * of the last address).
+ */
+
+ if (groupflag && ! mn->m_ingrp) {
+ output = add(";", output);
+ column++;
+ }
+
+ groupflag = mn->m_ingrp;
+
+ if (mn->m_gname) {
+ cp = mh_xstrdup(mn->m_gname);
+ }
+
if (reformat) {
- if (mn->m_gname) {
- cp = add(mn->m_gname, NULL);
- }
cp = add(adrformat(mn), cp);
} else {
- cp = add(mn->m_text, NULL);
+ cp = add(mn->m_text, cp);
}
len = strlen(cp);
column += len;
free(cp);
cp = NULL;
+ }
- /*
- * If we were in a group but are no longer, make sure we add a
- * trailing semicolon.
- */
-
- if (groupflag && ! mn->m_ingrp) {
- output = add(";", output);
- }
+ /*
+ * Just in case we're at the end of a list
+ */
- groupflag = mn->m_ingrp;
+ if (groupflag) {
+ output = add(";", output);
}
- *value = output;
- output = NULL;
+ output = add("\n", output);
-out:
+ free(*value);
+ *value = output;
+ output = NULL;
- if (tmpbuf)
- free(tmpbuf);
- if (output)
- free(output);
+out:
+ free(tmpbuf);
+ free(output);
return errflag > 0;
}
for (; *string != '\0'; string++) {
if ((isascii((unsigned char) *string))) {
(*asciilen)++;
- if (!qphrasevalid((unsigned char) *string))
+ /*
+ * So, a space is not a valid phrase character, but we're counting
+ * an exception here, because in q-p a space can be directly
+ * encoded as an underscore.
+ */
+ if (!qphrasevalid((unsigned char) *string) && *string != ' ')
(*specialchars)++;
} else {
(*eightbitchars)++;
projects / nmh / blobdiff