X-Git-Url: https://diplodocus.org/git/nmh/blobdiff_plain/2734acd65d1667d99902db2ce35741dd97e69f1c..4b2ba897bbf4c545b33e42bd1e06bee26e665dc4:/uip/mhparse.c?ds=inline diff --git a/uip/mhparse.c b/uip/mhparse.c index 79fbb716..49c1aba6 100644 --- a/uip/mhparse.c +++ b/uip/mhparse.c @@ -9,19 +9,19 @@ #include #include -#include #include #include #include #include #include #include +#ifdef HAVE_ICONV +# include +#endif /* HAVE_ICONV */ extern int debugsw; -extern pid_t xpid; /* mhshowsbr.c */ - /* cache policies */ extern int rcachesw; /* mhcachesbr.c */ extern int wcachesw; /* mhcachesbr.c */ @@ -33,10 +33,21 @@ int checksw = 0; /* check Content-MD5 field */ * 1) Instruct parser not to detect invalid Content-Transfer-Encoding * in a multipart. * 2) Suppress the warning about bogus multipart content, and report it. + * 3) Suppress the warning about extraneous trailing ';' in header parameter + * lists, and report it. */ int skip_mp_cte_check; int suppress_bogus_mp_content_warning; int bogus_mp_content; +int suppress_extraneous_trailing_semicolon_warning; +int extraneous_trailing_semicolon; +int suppress_multiple_mime_version_warning = 1; + +/* list of preferred type/subtype pairs, for -prefer */ +char *preferred_types[NPREFS], + *preferred_subtypes[NPREFS]; +int npreferred; + /* * Structures for TEXT messages @@ -58,6 +69,7 @@ struct k2v SubMultiPart[] = { { "alternative", MULTI_ALTERNATE }, { "digest", MULTI_DIGEST }, { "parallel", MULTI_PARALLEL }, + { "related", MULTI_RELATED }, { NULL, MULTI_UNKNOWN } /* this one must be last! */ }; @@ -98,7 +110,7 @@ static struct k2v EncodingType[] = { int find_cache (CT, int, int *, char *, char *, int); /* mhmisc.c */ -int part_ok (CT, int); +int part_ok (CT); int type_ok (CT, int); void content_error (char *, CT, char *, ...); @@ -114,7 +126,8 @@ static int get_comment (const char *, const char *, char **, char **); static int InitGeneric (CT); static int InitText (CT); static int InitMultiPart (CT); -void reverse_parts (CT); +static void reverse_parts (CT); +static void prefer_parts(CT ct); static int InitMessage (CT); static int InitApplication (CT); static int init_encoding (CT, OpenCEFunc); @@ -135,10 +148,11 @@ static int readDigest (CT, char *); static int get_leftover_mp_content (CT, int); static int InitURL (CT); static int openURL (CT, char **); -static size_t param_len(PM, int, size_t, int *); -static size_t encode_param(PM, char *, size_t, size_t, size_t, int); +static int parse_header_attrs (const char *, const char *, char **, PM *, + PM *, char **); +static size_t param_len(PM, int, size_t, int *, int *, size_t *); static size_t normal_param(PM, char *, size_t, size_t, size_t); -static int get_dispo (char *, CT); +static int get_dispo (char *, CT, int); struct str2init str2cts[] = { { "application", CT_APPLICATION, InitApplication }, @@ -178,19 +192,6 @@ struct str2init str2methods[] = { }; -int -pidcheck (int status) -{ - if ((status & 0xff00) == 0xff00 || (status & 0x007f) != SIGQUIT) - return status; - - fflush (stdout); - fflush (stderr); - done (1); - return 1; -} - - /* * Main entry point for parsing a MIME message or file. * It returns the Content structure for the top level @@ -204,6 +205,7 @@ parse_mime (char *file) char buffer[BUFSIZ]; FILE *fp; CT ct; + size_t n; /* * Check if file is actually standard input @@ -217,8 +219,13 @@ parse_mime (char *file) } file = add (tfile, NULL); - while (fgets (buffer, sizeof(buffer), stdin)) - fputs (buffer, fp); + while ((n = fread(buffer, 1, sizeof(buffer), stdin)) > 0) { + if (fwrite(buffer, 1, n, fp) != n) { + (void) m_unlink (file); + advise (file, "error copying to temporary file"); + return NULL; + } + } fflush (fp); if (ferror (stdin)) { @@ -287,9 +294,7 @@ get_content (FILE *in, char *file, int toplevel) m_getfld_state_t gstate = 0; /* allocate the content structure */ - if (!(ct = (CT) calloc (1, sizeof(*ct)))) - adios (NULL, "out of memory"); - + NEW0(ct); ct->c_fp = in; ct->c_file = add (file, NULL); ct->c_begin = ftell (ct->c_fp) + 1; @@ -358,16 +363,12 @@ get_content (FILE *in, char *file, int toplevel) if (!strcasecmp (hp->name, VRSN_FIELD)) { int ucmp; char c, *cp, *dp; + char *vrsn; - if (ct->c_vrsn) { - advise (NULL, "message %s has multiple %s: fields", - ct->c_file, VRSN_FIELD); - goto next_header; - } - ct->c_vrsn = add (hp->value, NULL); + vrsn = add (hp->value, NULL); /* Now, cleanup this field */ - cp = ct->c_vrsn; + cp = vrsn; while (isspace ((unsigned char) *cp)) cp++; @@ -394,6 +395,14 @@ get_content (FILE *in, char *file, int toplevel) admonish (NULL, "message %s has unknown value for %s: field (%s)", ct->c_file, VRSN_FIELD, cp); } + if (!ct->c_vrsn) { + ct->c_vrsn = vrsn; + } else { + if (! suppress_multiple_mime_version_warning) + advise (NULL, "message %s has multiple %s: fields", + ct->c_file, VRSN_FIELD); + free(vrsn); + } } else if (!strcasecmp (hp->name, TYPE_FIELD)) { /* Get Content-Type field */ @@ -514,7 +523,7 @@ get_content (FILE *in, char *file, int toplevel) } else if (!strcasecmp (hp->name, DISPO_FIELD)) { /* Get Content-Disposition field */ - if (get_dispo(hp->value, ct) == NOTOK) + if (get_dispo(hp->value, ct, 0) == NOTOK) goto out; } @@ -572,7 +581,7 @@ add_header (CT ct, char *name, char *value) HF hp; /* allocate header field structure */ - hp = mh_xmalloc (sizeof(*hp)); + NEW(hp); /* link data into header structure */ hp->name = name; @@ -592,88 +601,6 @@ add_header (CT ct, char *name, char *value) } -/* Make sure that buf contains at least one appearance of name, - followed by =. If not, insert both name and value, just after - first semicolon, if any. Note that name should not contain a - trailing =. And quotes will be added around the value. Typical - usage: make sure that a Content-Disposition header contains - filename="foo". If it doesn't and value does, use value from - that. */ -static char * -incl_name_value (char *buf, char *name, char *value) { - char *newbuf = buf; - - /* Assume that name is non-null. */ - if (buf && value) { - char *name_plus_equal = concat (name, "=", NULL); - - if (! strstr (buf, name_plus_equal)) { - char *insertion; - char *cp, *prefix, *suffix; - - /* Trim trailing space, esp. newline. */ - for (cp = &buf[strlen (buf) - 1]; - cp >= buf && isspace ((unsigned char) *cp); - --cp) { - *cp = '\0'; - } - - insertion = concat ("; ", name, "=", "\"", value, "\"", NULL); - - /* Insert at first semicolon, if any. If none, append to - end. */ - prefix = add (buf, NULL); - if ((cp = strchr (prefix, ';'))) { - suffix = concat (cp, NULL); - *cp = '\0'; - newbuf = concat (prefix, insertion, suffix, "\n", NULL); - free (suffix); - } else { - /* Append to end. */ - newbuf = concat (buf, insertion, "\n", NULL); - } - - free (prefix); - free (insertion); - free (buf); - } - - free (name_plus_equal); - } - - return newbuf; -} - -/* Extract just name_suffix="foo", if any, from value. If there isn't - one, return the entire value. Note that, for example, a name_suffix - of name will match filename="foo", and return foo. */ -static char * -extract_name_value (char *name_suffix, char *value) { - char *extracted_name_value = value; - char *name_suffix_plus_quote = concat (name_suffix, "=\"", NULL); - char *name_suffix_equals = strstr (value, name_suffix_plus_quote); - char *cp; - - free (name_suffix_plus_quote); - if (name_suffix_equals) { - char *name_suffix_begin; - - /* Find first \". */ - for (cp = name_suffix_equals; *cp != '"'; ++cp) /* empty */; - name_suffix_begin = ++cp; - /* Find second \". */ - for (; *cp != '"'; ++cp) /* empty */; - - extracted_name_value = mh_xmalloc (cp - name_suffix_begin + 1); - memcpy (extracted_name_value, - name_suffix_begin, - cp - name_suffix_begin); - extracted_name_value[cp - name_suffix_begin] = '\0'; - } - - return extracted_name_value; -} - /* * Parse Content-Type line and (if `magic' is non-zero) mhbuild composition * directives. Fills in the information of the CTinfo structure. @@ -718,7 +645,7 @@ get_ctinfo (char *cp, CT ct, int magic) *dp = c, cp = dp; if (!*ci->ci_type) { - advise (NULL, "invalid %s: field in message %s (empty type)", + advise (NULL, "invalid %s: field in message %s (empty type)", TYPE_FIELD, ct->c_file); return NOTOK; } @@ -837,7 +764,7 @@ magic_skip: * Get any {Content-Disposition} given in buffer. */ if (magic && *cp == '{') { - ct->c_dispo = ++cp; + ++cp; for (dp = cp + strlen (cp) - 1; dp >= cp; dp--) if (*dp == '}') break; @@ -849,10 +776,10 @@ magic_skip: c = *dp; *dp = '\0'; - if (*ct->c_dispo) - ct->c_dispo = concat (ct->c_dispo, "\n", NULL); - else - ct->c_dispo = NULL; + + if (get_dispo(cp, ct, 1) != OK) + return NOTOK; + *dp++ = c; cp = dp; @@ -912,14 +839,11 @@ magic_skip: have a *filename=, extract it from the magic contents. The r1bindex call skips any leading directory components. */ - if (ct->c_dispo) - ct->c_dispo = - incl_name_value (ct->c_dispo, - "filename", - r1bindex (extract_name_value ("name", - ci-> - ci_magic), - '/')); + if (ct->c_dispo_type && + !get_param(ct->c_dispo_first, "filename", '_', 1)) { + add_param(&ct->c_dispo_first, &ct->c_dispo_last, "filename", + r1bindex(ci->ci_magic, '/'), 0); + } } else advise (NULL, @@ -936,15 +860,20 @@ magic_skip: * get_ctinfo(). */ static int -get_dispo (char *cp, CT ct) +get_dispo (char *cp, CT ct, int buildflag) { - char *dp; + char *dp, *dispoheader; char c; int status; - /* Save the whole copy of the Content-Disposition header */ + /* + * Save the whole copy of the Content-Disposition header, unless we're + * processing a mhbuild directive. A NULL c_dispo will be a flag to + * mhbuild that the disposition header needs to be generated at that + * time. + */ - cp = ct->c_dispo = add(cp, NULL); + dispoheader = cp = add(cp, NULL); while (isspace ((unsigned char) *cp)) /* trim leading spaces */ cp++; @@ -962,8 +891,11 @@ get_dispo (char *cp, CT ct) if (debugsw) fprintf (stderr, "%s: %s\n", DISPO_FIELD, cp); - if (*cp == '(' && get_comment (ct->c_file, DISPO_FIELD, &cp, NULL) == NOTOK) + if (*cp == '(' && get_comment (ct->c_file, DISPO_FIELD, &cp, NULL) == + NOTOK) { + free(dispoheader); return NOTOK; + } for (dp = cp; istoken (*dp); dp++) continue; @@ -977,15 +909,21 @@ get_dispo (char *cp, CT ct) if ((status = parse_header_attrs (ct->c_file, DISPO_FIELD, &cp, &ct->c_dispo_first, &ct->c_dispo_last, NULL)) != OK) { - return status == NOTOK ? NOTOK : OK; - } - - if (*cp) { + if (status == NOTOK) { + free(dispoheader); + return NOTOK; + } + } else if (*cp) { advise (NULL, "extraneous information in message %s's %s: field\n%*s(%s)", ct->c_file, DISPO_FIELD, strlen(invo_name) + 2, "", cp); } + if (buildflag) + free(dispoheader); + else + ct->c_dispo = dispoheader; + return OK; } @@ -1078,7 +1016,6 @@ InitText (CT ct) char *chset = NULL; char *cp; PM pm; - struct k2v *kv; struct text *t; CI ci = &ct->c_ctinfo; @@ -1087,14 +1024,10 @@ InitText (CT ct) ci->ci_subtype = add ("plain", ci->ci_subtype); /* match subtype */ - for (kv = SubText; kv->kv_key; kv++) - if (!strcasecmp (ci->ci_subtype, kv->kv_key)) - break; - ct->c_subtype = kv->kv_value; + ct->c_subtype = ct_str_subtype (CT_TEXT, ci->ci_subtype); /* allocate text character set structure */ - if ((t = (struct text *) calloc (1, sizeof(*t))) == NULL) - adios (NULL, "out of memory"); + NEW0(t); ct->c_ctparams = (void *) t; /* scan for charset parameter */ @@ -1138,9 +1071,11 @@ InitMultiPart (CT ct) long last, pos; char *cp, *dp; PM pm; - char *bp, buffer[BUFSIZ]; + char *bp; + char *bufp = NULL; + size_t buflen; + ssize_t gotlen; struct multipart *m; - struct k2v *kv; struct part *part, **next; CI ci = &ct->c_ctinfo; CT p; @@ -1162,8 +1097,9 @@ InitMultiPart (CT ct) admonish (NULL, "\"%s/%s\" type in message %s must be encoded in\n" - "7bit, 8bit, or binary, per RFC 2045 (6.4). One workaround " - "is to\nmanually edit the file and change the \"%s\"\n" + "7bit, 8bit, or binary, per RFC 2045 (6.4). " + "mhfixmsg -fixcte can fix it, or\n" + "manually edit the file and change the \"%s\"\n" "Content-Transfer-Encoding to one of those. For now", ci->ci_type, ci->ci_subtype, ct->c_file, bp); free (cte); @@ -1172,10 +1108,7 @@ InitMultiPart (CT ct) } /* match subtype */ - for (kv = SubMultiPart; kv->kv_key; kv++) - if (!strcasecmp (ci->ci_subtype, kv->kv_key)) - break; - ct->c_subtype = kv->kv_value; + ct->c_subtype = ct_str_subtype (CT_MULTIPART, ci->ci_subtype); /* * Check for "boundary" parameter, which is @@ -1198,8 +1131,7 @@ InitMultiPart (CT ct) } /* allocate primary structure for multipart info */ - if ((m = (struct multipart *) calloc (1, sizeof(*m))) == NULL) - adios (NULL, "out of memory"); + NEW0(m); ct->c_ctparams = (void *) m; /* check if boundary parameter contains only whitespace characters */ @@ -1232,24 +1164,24 @@ InitMultiPart (CT ct) part = NULL; inout = 1; - while (fgets (buffer, sizeof(buffer) - 1, fp)) { + while ((gotlen = getline(&bufp, &buflen, fp)) != -1) { if (pos > last) break; - pos += strlen (buffer); - if (buffer[0] != '-' || buffer[1] != '-') + pos += gotlen; + if (bufp[0] != '-' || bufp[1] != '-') continue; if (inout) { - if (strcmp (buffer + 2, m->mp_start)) + if (strcmp (bufp + 2, m->mp_start)) continue; next_part: - if ((part = (struct part *) calloc (1, sizeof(*part))) == NULL) - adios (NULL, "out of memory"); + NEW0(part); *next = part; next = &part->mp_next; if (!(p = get_content (fp, ct->c_file, ct->c_subtype == MULTI_DIGEST ? -1 : 0))) { + free(bufp); ct->c_fp = NULL; return NOTOK; } @@ -1259,18 +1191,18 @@ next_part: fseek (fp, pos, SEEK_SET); inout = 0; } else { - if (strcmp (buffer + 2, m->mp_start) == 0) { + if (strcmp (bufp + 2, m->mp_start) == 0) { inout = 1; end_part: p = part->mp_part; - p->c_end = ftell(fp) - (strlen(buffer) + 1); + p->c_end = ftell(fp) - (gotlen + 1); if (p->c_end < p->c_begin) p->c_begin = p->c_end; if (inout) goto next_part; goto last_part; } else { - if (strcmp (buffer + 2, m->mp_stop) == 0) + if (strcmp (bufp + 2, m->mp_stop) == 0) goto end_part; } } @@ -1297,8 +1229,10 @@ end_part: last_part: /* reverse the order of the parts for multipart/alternative */ - if (ct->c_subtype == MULTI_ALTERNATE) + if (ct->c_subtype == MULTI_ALTERNATE) { reverse_parts (ct); + prefer_parts (ct); + } /* * label all subparts with part number, and @@ -1325,6 +1259,7 @@ last_part: /* initialize the content of the subparts */ if (p->c_ctinitfnx && (*p->c_ctinitfnx) (p) == NOTOK) { + free(bufp); fclose (ct->c_fp); ct->c_fp = NULL; return NOTOK; @@ -1335,6 +1270,7 @@ last_part: get_leftover_mp_content (ct, 1); get_leftover_mp_content (ct, 0); + free(bufp); fclose (ct->c_fp); ct->c_fp = NULL; return OK; @@ -1342,10 +1278,16 @@ last_part: /* - * reverse the order of the parts of a multipart/alternative + * reverse the order of the parts of a multipart/alternative, + * presumably to put the "most favored" alternative first, for + * ease of choosing/displaying it later on. from a mail message on + * nmh-workers, from kenh: + * "Stock" MH 6.8.5 did not have a reverse_parts() function, but I + * see code in mhn that did the same thing... Acccording to the RCS + * logs, that code was around from the initial checkin of mhn.c by + * John Romine in 1992, which is as far back as we have." */ - -void +static void reverse_parts (CT ct) { struct multipart *m = (struct multipart *) ct->c_ctparams; @@ -1361,6 +1303,84 @@ reverse_parts (CT ct) } } +static void +move_preferred_part (CT ct, char *type, char *subtype) +{ + struct multipart *m = (struct multipart *) ct->c_ctparams; + struct part *part, *prev, *head, *nhead, *ntail; + struct part h, n; + CI ci; + + /* move the matching part(s) to the head of the list: walk the + * list of parts, move matching parts to a new list (maintaining + * their order), and finally, concatenate the old list onto the + * new. + */ + + head = &h; + nhead = &n; + + head->mp_next = m->mp_parts; + nhead->mp_next = NULL; + ntail = nhead; + + prev = head; + part = head->mp_next; + while (part != NULL) { + ci = &part->mp_part->c_ctinfo; + if (!strcasecmp(ci->ci_type, type) && + (!subtype || !strcasecmp(ci->ci_subtype, subtype))) { + prev->mp_next = part->mp_next; + part->mp_next = NULL; + ntail->mp_next = part; + ntail = part; + part = prev->mp_next; + } else { + prev = part; + part = prev->mp_next; + } + } + ntail->mp_next = head->mp_next; + m->mp_parts = nhead->mp_next; + +} + +/* + * move parts that match the user's preferences (-prefer) to the head + * of the line. process preferences in reverse so first one given + * ends up first in line + */ +static void +prefer_parts(CT ct) +{ + int i; + for (i = npreferred-1; i >= 0; i--) + move_preferred_part(ct, preferred_types[i], preferred_subtypes[i]); +} + + + +/* parse_mime() arranges alternates in reverse (priority) order. This + function can be used to reverse them back. This will put, for + example, a text/plain part before a text/html part in a + multipart/alternative part, for example, where it belongs. */ +void +reverse_alternative_parts (CT ct) { + if (ct->c_type == CT_MULTIPART) { + struct multipart *m = (struct multipart *) ct->c_ctparams; + struct part *part; + + if (ct->c_subtype == MULTI_ALTERNATE) { + reverse_parts (ct); + } + + /* And call recursively on each part of a multipart. */ + for (part = m->mp_parts; part; part = part->mp_next) { + reverse_alternative_parts (part->mp_part); + } + } +} + /* * MESSAGE @@ -1369,7 +1389,6 @@ reverse_parts (CT ct) static int InitMessage (CT ct) { - struct k2v *kv; CI ci = &ct->c_ctinfo; if ((ct->c_encoding != CE_7BIT) && (ct->c_encoding != CE_8BIT)) { @@ -1384,10 +1403,7 @@ InitMessage (CT ct) ci->ci_subtype = add ("rfc822", ci->ci_subtype); /* match subtype */ - for (kv = SubMessage; kv->kv_key; kv++) - if (!strcasecmp (ci->ci_subtype, kv->kv_key)) - break; - ct->c_subtype = kv->kv_value; + ct->c_subtype = ct_str_subtype (CT_MESSAGE, ci->ci_subtype); switch (ct->c_subtype) { case MESSAGE_RFC822: @@ -1398,8 +1414,7 @@ InitMessage (CT ct) PM pm; struct partial *p; - if ((p = (struct partial *) calloc (1, sizeof(*p))) == NULL) - adios (NULL, "out of memory"); + NEW0(p); ct->c_ctparams = (void *) p; /* scan for parameters "id", "number", and "total" */ @@ -1447,8 +1462,7 @@ invalid_param: CT p; FILE *fp; - if ((e = (struct exbody *) calloc (1, sizeof(*e))) == NULL) - adios (NULL, "out of memory"); + NEW0(e); ct->c_ctparams = (void *) e; if (!ct->c_fp @@ -1472,14 +1486,14 @@ invalid_param: && p->c_ceopenfnx == openMail) { int cc, size; char *bp; - + if ((size = ct->c_end - p->c_begin) <= 0) { if (!e->eb_subject) content_error (NULL, ct, "empty body for access-type=mail-server"); goto no_body; } - + e->eb_body = bp = mh_xmalloc ((unsigned) size); fseek (p->c_fp, p->c_begin, SEEK_SET); while (size > 0) @@ -1637,14 +1651,10 @@ params_external (CT ct, int composing) static int InitApplication (CT ct) { - struct k2v *kv; CI ci = &ct->c_ctinfo; /* match subtype */ - for (kv = SubApplication; kv->kv_key; kv++) - if (!strcasecmp (ci->ci_subtype, kv->kv_key)) - break; - ct->c_subtype = kv->kv_value; + ct->c_subtype = ct_str_subtype (CT_APPLICATION, ci->ci_subtype); return OK; } @@ -1717,26 +1727,6 @@ size_encoding (CT ct) * BASE64 */ -static unsigned char b642nib[0x80] = { - 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, - 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, - 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, - 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, - 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, - 0xff, 0xff, 0xff, 0x3e, 0xff, 0xff, 0xff, 0x3f, - 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3a, 0x3b, - 0x3c, 0x3d, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, - 0xff, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, - 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, - 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, - 0x17, 0x18, 0x19, 0xff, 0xff, 0xff, 0xff, 0xff, - 0xff, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20, - 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, - 0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f, 0x30, - 0x31, 0x32, 0x33, 0xff, 0xff, 0xff, 0xff, 0xff -}; - - static int InitBase64 (CT ct) { @@ -1747,15 +1737,15 @@ InitBase64 (CT ct) static int openBase64 (CT ct, char **file) { - int bitno, cc, digested; - int fd, len, skip, own_ct_fp = 0, text = ct->c_type == CT_TEXT; - uint32_t bits; - unsigned char value, b; - char *cp, *ep, buffer[BUFSIZ]; + ssize_t cc, len; + int fd, own_ct_fp = 0; + char *cp, *buffer = NULL; /* sbeck -- handle suffixes */ CI ci; CE ce = &ct->c_cefile; - MD5_CTX mdContext; + unsigned char *decoded; + size_t decoded_len; + unsigned char digest[16]; if (ce->ce_fp) { fseek (ce->ce_fp, 0L, SEEK_SET); @@ -1779,15 +1769,7 @@ openBase64 (CT ct, char **file) /* sbeck@cise.ufl.edu -- handle suffixes */ ci = &ct->c_ctinfo; - snprintf (buffer, sizeof(buffer), "%s-suffix-%s/%s", - invo_name, ci->ci_type, ci->ci_subtype); - cp = context_find (buffer); - if (cp == NULL || *cp == '\0') { - snprintf (buffer, sizeof(buffer), "%s-suffix-%s", invo_name, - ci->ci_type); - cp = context_find (buffer); - } - if (cp != NULL && *cp != '\0') { + if ((cp = context_find_by_type ("suffix", ci->ci_type, ci->ci_subtype))) { if (ce->ce_unlink) { /* Create temporary file with filename extension. */ if ((ce->ce_file = m_mktemps(invo_name, cp, NULL, NULL)) == NULL) { @@ -1814,6 +1796,8 @@ openBase64 (CT ct, char **file) if ((len = ct->c_end - ct->c_begin) < 0) adios (NULL, "internal error(1)"); + buffer = mh_xmalloc (len + 1); + if (! ct->c_fp) { if ((ct->c_fp = fopen (ct->c_file, "r")) == NULL) { content_error (ct->c_file, ct, "unable to open for reading"); @@ -1821,17 +1805,11 @@ openBase64 (CT ct, char **file) } own_ct_fp = 1; } - - if ((digested = ct->c_digested)) - MD5Init (&mdContext); - - bitno = 18; - bits = 0L; - skip = 0; lseek (fd = fileno (ct->c_fp), (off_t) ct->c_begin, SEEK_SET); + cp = buffer; while (len > 0) { - switch (cc = read (fd, buffer, sizeof(buffer) - 1)) { + switch (cc = read (fd, cp, len)) { case NOTOK: content_error (ct->c_file, ct, "error reading from"); goto clean_up; @@ -1844,75 +1822,41 @@ openBase64 (CT ct, char **file) if (cc > len) cc = len; len -= cc; - - for (ep = (cp = buffer) + cc; cp < ep; cp++) { - switch (*cp) { - default: - if (isspace ((unsigned char) *cp)) - break; - if (skip || (((unsigned char) *cp) & 0x80) - || (value = b642nib[((unsigned char) *cp) & 0x7f]) > 0x3f) { - if (debugsw) { - fprintf (stderr, "*cp=0x%x pos=%ld skip=%d\n", - (unsigned char) *cp, - (long) (lseek (fd, (off_t) 0, SEEK_CUR) - (ep - cp)), - skip); - } - content_error (NULL, ct, - "invalid BASE64 encoding -- continuing"); - continue; - } - - bits |= value << bitno; -test_end: - if ((bitno -= 6) < 0) { - b = (bits >> 16) & 0xff; - if (!text || b != '\r') - putc ((char) b, ce->ce_fp); - if (digested) - MD5Update (&mdContext, &b, 1); - if (skip < 2) { - b = (bits >> 8) & 0xff; - if (! text || b != '\r') - putc ((char) b, ce->ce_fp); - if (digested) - MD5Update (&mdContext, &b, 1); - if (skip < 1) { - b = bits & 0xff; - if (! text || b != '\r') - putc ((char) b, ce->ce_fp); - if (digested) - MD5Update (&mdContext, &b, 1); - } - } - - if (ferror (ce->ce_fp)) { - content_error (ce->ce_file, ct, - "error writing to"); - goto clean_up; - } - bitno = 18, bits = 0L, skip = 0; - } - break; - - case '=': - if (++skip > 3) - goto self_delimiting; - goto test_end; - } - } - } + cp += cc; + } } - if (bitno != 18) { - if (debugsw) - fprintf (stderr, "premature ending (bitno %d)\n", bitno); + /* decodeBase64() requires null-terminated input. */ + *cp = '\0'; - content_error (NULL, ct, "invalid BASE64 encoding"); - goto clean_up; + if (decodeBase64 (buffer, &decoded, &decoded_len, ct->c_type == CT_TEXT, + ct->c_digested ? digest : NULL) == OK) { + size_t i; + unsigned char *decoded_p = decoded; + for (i = 0; i < decoded_len; ++i) { + putc (*decoded_p++, ce->ce_fp); + } + free ((char *) decoded); + if (ferror (ce->ce_fp)) { + content_error (ce->ce_file, ct, "error writing to"); + goto clean_up; + } + + if (ct->c_digested) { + if (memcmp(digest, ct->c_digest, + sizeof(digest) / sizeof(digest[0]))) { + content_error (NULL, ct, + "content integrity suspect (digest mismatch) -- continuing"); + } else { + if (debugsw) { + fprintf (stderr, "content integrity confirmed\n"); + } + } + } + } else { + goto clean_up; } -self_delimiting: fseek (ct->c_fp, 0L, SEEK_SET); if (fflush (ce->ce_fp)) { @@ -1920,19 +1864,6 @@ self_delimiting: goto clean_up; } - if (digested) { - unsigned char digest[16]; - - MD5Final (digest, &mdContext); - if (memcmp((char *) digest, (char *) ct->c_digest, - sizeof(digest) / sizeof(digest[0]))) - content_error (NULL, ct, - "content integrity suspect (digest mismatch) -- continuing"); - else - if (debugsw) - fprintf (stderr, "content integrity confirmed\n"); - } - fseek (ce->ce_fp, 0L, SEEK_SET); ready_to_go: @@ -1941,6 +1872,7 @@ ready_to_go: fclose (ct->c_fp); ct->c_fp = NULL; } + free (buffer); return fileno (ce->ce_fp); clean_up: @@ -1949,6 +1881,7 @@ clean_up: ct->c_fp = NULL; } free_encoding (ct, 0); + free (buffer); return NOTOK; } @@ -1966,18 +1899,18 @@ static char hex2nib[0x80] = { 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, - 0x00, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F, 0x00, + 0x00, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, - 0x00, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x00, + 0x00, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 }; -static int +static int InitQuoted (CT ct) { return init_encoding (ct, openQuoted); @@ -1989,7 +1922,9 @@ openQuoted (CT ct, char **file) { int cc, digested, len, quoted, own_ct_fp = 0; char *cp, *ep; - char buffer[BUFSIZ]; + char *bufp = NULL; + size_t buflen; + ssize_t gotlen; unsigned char mask; CE ce = &ct->c_cefile; /* sbeck -- handle suffixes */ @@ -2018,15 +1953,7 @@ openQuoted (CT ct, char **file) /* sbeck@cise.ufl.edu -- handle suffixes */ ci = &ct->c_ctinfo; - snprintf (buffer, sizeof(buffer), "%s-suffix-%s/%s", - invo_name, ci->ci_type, ci->ci_subtype); - cp = context_find (buffer); - if (cp == NULL || *cp == '\0') { - snprintf (buffer, sizeof(buffer), "%s-suffix-%s", invo_name, - ci->ci_type); - cp = context_find (buffer); - } - if (cp != NULL && *cp != '\0') { + if ((cp = context_find_by_type ("suffix", ci->ci_type, ci->ci_subtype))) { if (ce->ce_unlink) { /* Create temporary file with filename extension. */ if ((ce->ce_file = m_mktemps(invo_name, cp, NULL, NULL)) == NULL) { @@ -2071,16 +1998,16 @@ openQuoted (CT ct, char **file) fseek (ct->c_fp, ct->c_begin, SEEK_SET); while (len > 0) { - if (fgets (buffer, sizeof(buffer) - 1, ct->c_fp) == NULL) { + if ((gotlen = getline(&bufp, &buflen, ct->c_fp)) == -1) { content_error (NULL, ct, "premature eof"); goto clean_up; } - if ((cc = strlen (buffer)) > len) + if ((cc = gotlen) > len) cc = len; len -= cc; - for (ep = (cp = buffer) + cc - 1; cp <= ep; ep--) + for (ep = (cp = bufp) + cc - 1; cp <= ep; ep--) if (!isspace ((unsigned char) *ep)) break; *++ep = '\n', ep++; @@ -2185,6 +2112,7 @@ ready_to_go: fclose (ct->c_fp); ct->c_fp = NULL; } + free (bufp); return fileno (ce->ce_fp); clean_up: @@ -2193,6 +2121,7 @@ clean_up: fclose (ct->c_fp); ct->c_fp = NULL; } + free (bufp); return NOTOK; } @@ -2244,15 +2173,7 @@ open7Bit (CT ct, char **file) /* sbeck@cise.ufl.edu -- handle suffixes */ ci = &ct->c_ctinfo; - snprintf (buffer, sizeof(buffer), "%s-suffix-%s/%s", - invo_name, ci->ci_type, ci->ci_subtype); - cp = context_find (buffer); - if (cp == NULL || *cp == '\0') { - snprintf (buffer, sizeof(buffer), "%s-suffix-%s", invo_name, - ci->ci_type); - cp = context_find (buffer); - } - if (cp != NULL && *cp != '\0') { + if ((cp = context_find_by_type ("suffix", ci->ci_type, ci->ci_subtype))) { if (ce->ce_unlink) { /* Create temporary file with filename extension. */ if ((ce->ce_file = m_mktemps(invo_name, cp, NULL, NULL)) == NULL) { @@ -2284,7 +2205,7 @@ open7Bit (CT ct, char **file) fprintf (ce->ce_fp, "%s: %s/%s", TYPE_FIELD, ci->ci_type, ci->ci_subtype); len += strlen (TYPE_FIELD) + 2 + strlen (ci->ci_type) + 1 + strlen (ci->ci_subtype); - buffer = output_params(len, ci->ci_first_pm, &len); + buffer = output_params(len, ci->ci_first_pm, &len, 0); if (buffer) { fputs (buffer, ce->ce_fp); @@ -2340,7 +2261,9 @@ open7Bit (CT ct, char **file) cc = len; len -= cc; - fwrite (buffer, sizeof(*buffer), cc, ce->ce_fp); + if ((int) fwrite (buffer, sizeof(*buffer), cc, ce->ce_fp) < cc) { + advise ("open7Bit", "fwrite"); + } if (ferror (ce->ce_fp)) { content_error (ce->ce_file, ct, "error writing to"); goto clean_up; @@ -2407,6 +2330,7 @@ openExternal (CT ct, CT cb, CE ce, char **file, int *fd) } } + *fd = fileno (ce->ce_fp); return OK; ready_already: @@ -2474,7 +2398,9 @@ openFile (CT ct, char **file) while ((cc = fread (buffer, sizeof(*buffer), sizeof(buffer), gp)) > 0) - fwrite (buffer, sizeof(*buffer), cc, fp); + if ((int) fwrite (buffer, sizeof(*buffer), cc, fp) < cc) { + advise ("openFile", "fwrite"); + } fflush (fp); if (ferror (gp)) { @@ -2544,13 +2470,6 @@ openFTP (CT ct, char **file) return NOTOK; } - if (xpid) { - if (xpid < 0) - xpid = -xpid; - pidcheck (pidwait (xpid, NOTOK)); - xpid = 0; - } - /* Get the buffer ready to go */ bp = buffer; buflen = sizeof(buffer); @@ -2587,7 +2506,7 @@ openFTP (CT ct, char **file) /* * Now, check the answer */ - if (!getanswer (buffer)) + if (!read_yes_or_no_if_tty (buffer)) return NOTOK; if (e->eb_flags) { @@ -2690,7 +2609,9 @@ openFTP (CT ct, char **file) while ((cc= fread (buffer, sizeof(*buffer), sizeof(buffer), gp)) > 0) - fwrite (buffer, sizeof(*buffer), cc, fp); + if ((int) fwrite (buffer, sizeof(*buffer), cc, fp) < cc) { + advise ("openFTP", "fwrite"); + } fflush (fp); if (ferror (gp)) { @@ -2750,13 +2671,6 @@ openMail (CT ct, char **file) return NOTOK; } - if (xpid) { - if (xpid < 0) - xpid = -xpid; - pidcheck (pidwait (xpid, NOTOK)); - xpid = 0; - } - /* Get buffer ready to go */ bp = buffer; buflen = sizeof(buffer); @@ -2779,7 +2693,7 @@ openMail (CT ct, char **file) e->eb_subject ? e->eb_subject : e->eb_body); /* Now, check answer */ - if (!getanswer (buffer)) + if (!read_yes_or_no_if_tty (buffer)) return NOTOK; vecp = 0; @@ -2887,13 +2801,6 @@ openURL (CT ct, char **file) return NOTOK; } - if (xpid) { - if (xpid < 0) - xpid = -xpid; - pidcheck (pidwait (xpid, NOTOK)); - xpid = 0; - } - ce->ce_unlink = (*file == NULL); caching = 0; cachefile[0] = '\0'; @@ -2964,7 +2871,9 @@ openURL (CT ct, char **file) while ((cc = fread(buffer, sizeof(*buffer), sizeof(buffer), gp)) > 0) - fwrite(buffer, sizeof(*buffer), cc, fp); + if ((int) fwrite(buffer, sizeof(*buffer), cc, fp) < cc) { + advise ("openURL", "fwrite"); + } fflush(fp); @@ -2982,79 +2891,45 @@ openURL (CT ct, char **file) return fd; } + +/* + * Stores MD5 digest (in cp, from Content-MD5 header) in ct->c_digest. It + * has to be base64 decoded. + */ static int readDigest (CT ct, char *cp) { - int bitno, skip; - uint32_t bits; - char *bp = cp; - unsigned char *dp, value, *ep; - - bitno = 18; - bits = 0L; - skip = 0; - - for (ep = (dp = ct->c_digest) - + sizeof(ct->c_digest) / sizeof(ct->c_digest[0]); *cp; cp++) - switch (*cp) { - default: - if (skip - || (*cp & 0x80) - || (value = b642nib[*cp & 0x7f]) > 0x3f) { - if (debugsw) - fprintf (stderr, "invalid BASE64 encoding\n"); - return NOTOK; - } + unsigned char *digest; - bits |= value << bitno; -test_end: - if ((bitno -= 6) < 0) { - if (dp + (3 - skip) > ep) - goto invalid_digest; - *dp++ = (bits >> 16) & 0xff; - if (skip < 2) { - *dp++ = (bits >> 8) & 0xff; - if (skip < 1) - *dp++ = bits & 0xff; - } - bitno = 18; - bits = 0L; - skip = 0; - } - break; + size_t len; + if (decodeBase64 (cp, &digest, &len, 0, NULL) == OK) { + const size_t maxlen = sizeof ct->c_digest / sizeof ct->c_digest[0]; - case '=': - if (++skip > 3) - goto self_delimiting; - goto test_end; - } - if (bitno != 18) { - if (debugsw) - fprintf (stderr, "premature ending (bitno %d)\n", bitno); + if (strlen ((char *) digest) <= maxlen) { + memcpy (ct->c_digest, digest, maxlen); - return NOTOK; - } -self_delimiting: - if (dp != ep) { -invalid_digest: - if (debugsw) { - while (*cp) - cp++; - fprintf (stderr, "invalid MD5 digest (got %d octets)\n", - (int)(cp - bp)); - } + if (debugsw) { + size_t i; - return NOTOK; - } + fprintf (stderr, "MD5 digest="); + for (i = 0; i < maxlen; ++i) { + fprintf (stderr, "%02x", ct->c_digest[i] & 0xff); + } + fprintf (stderr, "\n"); + } - if (debugsw) { - fprintf (stderr, "MD5 digest="); - for (dp = ct->c_digest; dp < ep; dp++) - fprintf (stderr, "%02x", *dp & 0xff); - fprintf (stderr, "\n"); - } + return OK; + } else { + if (debugsw) { + fprintf (stderr, "invalid MD5 digest (got %d octets)\n", + (int) strlen ((char *) digest)); + } - return OK; + return NOTOK; + } + } else { + return NOTOK; + } } @@ -3062,12 +2937,15 @@ invalid_digest: after the last subpart that hasn't been stored anywhere else, so do that. */ int -get_leftover_mp_content (CT ct, int before /* or after */) { +get_leftover_mp_content (CT ct, int before /* or after */) +{ struct multipart *m = (struct multipart *) ct->c_ctparams; char *boundary; int found_boundary = 0; - char buffer[BUFSIZ]; int max = BUFSIZ; + char *bufp = NULL; + size_t buflen; + ssize_t gotlen; int read = 0; char *content = NULL; @@ -3100,18 +2978,18 @@ get_leftover_mp_content (CT ct, int before /* or after */) { } /* Back up by 1 to pick up the newline. */ - while (fgets (buffer, sizeof(buffer) - 1, ct->c_fp)) { - read += strlen (buffer); + while ((gotlen = getline(&bufp, &buflen, ct->c_fp)) != -1) { + read += gotlen; /* Don't look beyond beginning of first subpart (before) or next part (after). */ - if (read > max) buffer[read-max] = '\0'; + if (read > max) bufp[read-max] = '\0'; if (before) { - if (! strcmp (buffer, boundary)) { + if (! strcmp (bufp, boundary)) { found_boundary = 1; } } else { - if (! found_boundary && ! strcmp (buffer, boundary)) { + if (! found_boundary && ! strcmp (bufp, boundary)) { found_boundary = 1; continue; } @@ -3120,12 +2998,12 @@ get_leftover_mp_content (CT ct, int before /* or after */) { if ((before && ! found_boundary) || (! before && found_boundary)) { if (content) { char *old_content = content; - content = concat (content, buffer, NULL); + content = concat (content, bufp, NULL); free (old_content); } else { content = before - ? concat ("\n", buffer, NULL) - : concat (buffer, NULL); + ? concat ("\n", bufp, NULL) + : concat (bufp, NULL); } } @@ -3159,6 +3037,7 @@ get_leftover_mp_content (CT ct, int before /* or after */) { } free (boundary); + free (bufp); return OK; } @@ -3222,6 +3101,8 @@ ct_subtype_str (int type, int subtype) { return "digest"; case MULTI_PARALLEL: return "parallel"; + case MULTI_RELATED: + return "related"; default: return "unknown_multipart_subtype"; } @@ -3242,6 +3123,62 @@ ct_subtype_str (int type, int subtype) { } +int +ct_str_type (const char *type) { + struct str2init *s2i; + + for (s2i = str2cts; s2i->si_key; ++s2i) { + if (! strcasecmp (type, s2i->si_key)) { + break; + } + } + if (! s2i->si_key && ! uprf (type, "X-")) { + ++s2i; + } + + return s2i->si_val; +} + + +int +ct_str_subtype (int type, const char *subtype) { + struct k2v *kv; + + switch (type) { + case CT_APPLICATION: + for (kv = SubApplication; kv->kv_key; ++kv) { + if (! strcasecmp (subtype, kv->kv_key)) { + break; + } + } + return kv->kv_value; + case CT_MESSAGE: + for (kv = SubMessage; kv->kv_key; ++kv) { + if (! strcasecmp (subtype, kv->kv_key)) { + break; + } + } + return kv->kv_value; + case CT_MULTIPART: + for (kv = SubMultiPart; kv->kv_key; ++kv) { + if (! strcasecmp (subtype, kv->kv_key)) { + break; + } + } + return kv->kv_value; + case CT_TEXT: + for (kv = SubText; kv->kv_key; ++kv) { + if (! strcasecmp (subtype, kv->kv_key)) { + break; + } + } + return kv->kv_value; + default: + return 0; + } +} + + /* Find the content type and InitFunc for the CT. */ const struct str2init * get_ct_init (int type) { @@ -3292,16 +3229,49 @@ get_ce_method (const char *method) { return NULL; } -int +/* + * Parse a series of MIME attributes (or parameters) given a header as + * input. + * + * Arguments include: + * + * filename - Name of input file (for error messages) + * fieldname - Name of field being processed + * headerp - Pointer to pointer of the beginning of the MIME attributes. + * Updated to point to end of attributes when finished. + * param_head - Pointer to head of parameter list + * param_tail - Pointer to tail of parameter list + * commentp - Pointer to header comment pointer (may be NULL) + * + * Returns OK if parsing was successful, NOTOK if parsing failed, and + * DONE to indicate a benign error (minor parsing error, but the program + * should continue). + */ + +static int parse_header_attrs (const char *filename, const char *fieldname, char **header_attrp, PM *param_head, PM *param_tail, char **commentp) { char *cp = *header_attrp; PM pm; + struct sectlist { + char *value; + int index; + int len; + struct sectlist *next; + } *sp, *sp2; + struct parmlist { + char *name; + char *charset; + char *lang; + struct sectlist *sechead; + struct parmlist *next; + } *pp, *pp2, *phead = NULL; while (*cp == ';') { - char *dp, *vp, *up, c; + char *dp, *vp, *up, *nameptr, *valptr, *charset = NULL, *lang = NULL; + int encoded = 0, partial = 0, len = 0, index = 0; cp++; while (isspace ((unsigned char) *cp)) @@ -3313,10 +3283,13 @@ parse_header_attrs (const char *filename, const char *fieldname, } if (*cp == 0) { - advise (NULL, - "extraneous trailing ';' in message %s's %s: " - "parameter list", - filename, fieldname); + if (! suppress_extraneous_trailing_semicolon_warning) { + advise (NULL, + "extraneous trailing ';' in message %s's %s: " + "parameter list", + filename, fieldname); + } + extraneous_trailing_semicolon = 1; return DONE; } @@ -3335,60 +3308,299 @@ parse_header_attrs (const char *filename, const char *fieldname, return NOTOK; } - pm = mh_xmalloc(sizeof(*pm)); - memset(pm, 0, sizeof(*pm)); + /* + * To handle RFC 2231, we have to deal with the following extensions: + * + * name*=encoded-value + * name*=part-N-of-a-parameter-value + * name**=encoded-part-N-of-a-parameter-value + * + * So the rule is: + * If there's a * right before the equal sign, it's encoded. + * If there's a * and one or more digits, then it's section N. + * + * Remember we can have one or the other, or both. cp points to + * beginning of name, up points past the last character in the + * parameter name. + */ + + for (vp = cp; vp < up; vp++) { + if (*vp == '*' && vp < up - 1) { + partial = 1; + continue; + } else if (*vp == '*' && vp == up - 1) { + encoded = 1; + } else if (partial) { + if (isdigit((unsigned char) *vp)) + index = *vp - '0' + index * 10; + else { + advise (NULL, "invalid parameter index in message %s's " + "%s: field\n%*s(parameter %s)", filename, + fieldname, strlen(invo_name) + 2, "", cp); + return NOTOK; + } + } else { + len++; + } + } + + /* + * Break out the parameter name and value sections and allocate + * memory for each. + */ + + nameptr = mh_xmalloc(len + 1); + strncpy(nameptr, cp, len); + nameptr[len] = '\0'; - /* This is all mega-bozo and needs cleanup */ - vp = (pm->pm_name = add (cp, NULL)) + (up - cp); - *vp = '\0'; for (dp++; isspace ((unsigned char) *dp);) dp++; - /* Now store the attribute value. */ + if (encoded) { + /* + * Single quotes delimit the character set and language tag. + * They are required on the first section (or a complete + * parameter). + */ + if (index == 0) { + vp = dp; + while (*vp != '\'' && !isspace((unsigned char) *vp) && + *vp != '\0') + vp++; + if (*vp == '\'') { + if (vp != dp) { + len = vp - dp; + charset = mh_xmalloc(len + 1); + strncpy(charset, dp, len); + charset[len] = '\0'; + } else { + charset = NULL; + } + vp++; + } else { + advise(NULL, "missing charset in message %s's %s: " + "field\n%*s(parameter %s)", filename, fieldname, + strlen(invo_name) + 2, "", nameptr); + free(nameptr); + return NOTOK; + } + dp = vp; + + while (*vp != '\'' && !isspace((unsigned char) *vp) && + *vp != '\0') + vp++; + + if (*vp == '\'') { + if (vp != dp) { + len = vp - dp; + lang = mh_xmalloc(len + 1); + strncpy(lang, dp, len); + lang[len] = '\0'; + } else { + lang = NULL; + } + vp++; + } else { + advise(NULL, "missing language tag in message %s's %s: " + "field\n%*s(parameter %s)", filename, fieldname, + strlen(invo_name) + 2, "", nameptr); + free(nameptr); + if (charset) + free(charset); + return NOTOK; + } + + dp = vp; + } + + /* + * At this point vp should be pointing at the beginning + * of the encoded value/section. Continue until we reach + * the end or get whitespace. But first, calculate the + * length so we can allocate the correct buffer size. + */ + + for (vp = dp, len = 0; istoken(*vp); vp++) { + if (*vp == '%') { + if (*(vp + 1) == '\0' || + !isxdigit((unsigned char) *(vp + 1)) || + *(vp + 2) == '\0' || + !isxdigit((unsigned char) *(vp + 2))) { + advise(NULL, "invalid encoded sequence in message " + "%s's %s: field\n%*s(parameter %s)", + filename, fieldname, strlen(invo_name) + 2, + "", nameptr); + free(nameptr); + if (charset) + free(charset); + if (lang) + free(lang); + return NOTOK; + } + vp += 2; + } + len++; + } + + up = valptr = mh_xmalloc(len + 1); + + for (vp = dp; istoken(*vp); vp++) { + if (*vp == '%') { + *up++ = decode_qp(*(vp + 1), *(vp + 2)); + vp += 2; + } else { + *up++ = *vp; + } + } + + *up = '\0'; + cp = vp; + } else { + /* + * A "normal" string. If it's got a leading quote, then we + * strip the quotes out. Otherwise go until we reach the end + * or get whitespace. Note we scan it twice; once to get the + * length, then the second time copies it into the destination + * buffer. + */ - vp = pm->pm_name + (dp - cp); + len = 0; - if (*dp == '"') { - for (cp = ++dp, dp = vp;;) { - switch (c = *cp++) { + if (*dp == '"') { + for (cp = dp + 1;;) { + switch (*cp++) { case '\0': bad_quote: advise (NULL, "invalid quoted-string in message %s's %s: " "field\n%*s(parameter %s)", filename, fieldname, strlen(invo_name) + 2, "", - pm->pm_name); + nameptr); + free(nameptr); + if (charset) + free(charset); + if (lang) + free(lang); return NOTOK; + case '"': + break; case '\\': - *dp++ = c; - if ((c = *cp++) == '\0') + if (*++cp == '\0') goto bad_quote; - /* else fall... */ - + /* FALL THROUGH */ default: - *dp++ = c; + len++; continue; + } + break; + } - case '"': - *dp = '\0'; + } else { + for (cp = dp; istoken (*cp); cp++) { + len++; + } + } + + valptr = mh_xmalloc(len + 1); + + if (*dp == '"') { + int i; + for (cp = dp + 1, vp = valptr, i = 0; i < len; i++) { + if (*cp == '\\') { + cp++; + } + *vp++ = *cp++; + } + cp++; + } else { + strncpy(valptr, cp = dp, len); + cp += len; + } + + valptr[len] = '\0'; + } + + /* + * If 'partial' is set, we don't allocate a parameter now. We + * put it on the parameter linked list to be reassembled later. + * + * "phead" points to a list of all parameters we need to reassemble. + * Each parameter has a list of sections. We insert the sections in + * order. + */ + + if (partial) { + for (pp = phead; pp != NULL; pp = pp->next) { + if (strcasecmp(nameptr, pp->name) == 0) + break; + } + + if (pp == NULL) { + NEW(pp); + memset(pp, 0, sizeof(*pp)); + pp->name = nameptr; + pp->next = phead; + phead = pp; + } + + /* + * Insert this into the section linked list + */ + + NEW(sp); + memset(sp, 0, sizeof(*sp)); + sp->value = valptr; + sp->index = index; + sp->len = len; + + if (pp->sechead == NULL || pp->sechead->index > index) { + sp->next = pp->sechead; + pp->sechead = sp; + } else { + for (sp2 = pp->sechead; sp2 != NULL; sp2 = sp2->next) { + if (sp2->index == sp->index) { + advise (NULL, "duplicate index (%d) in message " + "%s's %s: field\n%*s(parameter %s)", sp->index, + filename, fieldname, strlen(invo_name) + 2, "", + nameptr); + free (nameptr); + return NOTOK; + } + if (sp2->index < sp->index && + (sp2->next == NULL || sp2->next->index > sp->index)) { + sp->next = sp2->next; + sp2->next = sp; break; + } } - break; + + if (sp2 == NULL) { + advise(NULL, "Internal error: cannot insert partial " + "param in message %s's %s: field\n%*s(parameter %s)", + filename, fieldname, strlen(invo_name) + 2, "", + nameptr); + free (nameptr); + return NOTOK; + } + } + + /* + * Save our charset and lang tags. + */ + + if (index == 0 && encoded) { + if (pp->charset) + free(pp->charset); + pp->charset = charset; + if (pp->lang) + free(pp->lang); + pp->lang = lang; } } else { - for (cp = dp, dp = vp; istoken (*cp); cp++, dp++) - continue; - *dp = '\0'; - } - pm->pm_value = getcpy(vp); - if (!*vp) { - advise (NULL, - "invalid parameter in message %s's %s: " - "field\n%*s(parameter %s)", - filename, fieldname, strlen(invo_name) + 2, "", - pm->pm_name); - return NOTOK; + pm = add_param(param_head, param_tail, nameptr, valptr, 1); + pm->pm_charset = charset; + pm->pm_lang = lang; } while (isspace ((unsigned char) *cp)) @@ -3398,20 +3610,65 @@ bad_quote: get_comment (filename, fieldname, &cp, commentp) == NOTOK) { return NOTOK; } + } - if (*param_head == NULL) { - *param_head = pm; - *param_tail = pm; - } else { - (*param_tail)->pm_next = pm; - *param_tail = pm; + /* + * Now that we're done, reassemble all of the partial parameters. + */ + + for (pp = phead; pp != NULL; ) { + char *p, *q; + size_t tlen = 0; + int pindex = 0; + for (sp = pp->sechead; sp != NULL; sp = sp->next) { + if (sp->index != pindex++) { + advise(NULL, "missing section %d for parameter in " + "message %s's %s: field\n%*s(parameter %s)", pindex - 1, + filename, fieldname, strlen(invo_name) + 2, "", + pp->name); + return NOTOK; + } + tlen += sp->len; + } + + p = q = mh_xmalloc(tlen + 1); + for (sp = pp->sechead; sp != NULL; ) { + memcpy(q, sp->value, sp->len); + q += sp->len; + free(sp->value); + sp2 = sp->next; + free(sp); + sp = sp2; } + + p[tlen] = '\0'; + + pm = add_param(param_head, param_tail, pp->name, p, 1); + pm->pm_charset = pp->charset; + pm->pm_lang = pp->lang; + pp2 = pp->next; + free(pp); + pp = pp2; } *header_attrp = cp; return OK; } +/* + * Return the charset for a particular content type. + */ + +char * +content_charset (CT ct) { + char *ret_charset = NULL; + + ret_charset = get_param(ct->c_ctinfo.ci_first_pm, "charset", '?', 0); + + return ret_charset ? ret_charset : getcpy ("US-ASCII"); +} + + /* * Create a string based on a list of output parameters. Assume that this * parameter string will be appended to an existing header, so start out @@ -3419,12 +3676,12 @@ bad_quote: */ char * -output_params(size_t initialwidth, PM params, int *offsetout) +output_params(size_t initialwidth, PM params, int *offsetout, int external) { char *paramout = NULL; char line[CPERLIN * 2], *q; - int curlen, index, eightbit, encode, i; - size_t valoff; + int curlen, index, cont, encode, i; + size_t valoff, numchars; while (params != NULL) { encode = 0; @@ -3432,6 +3689,9 @@ output_params(size_t initialwidth, PM params, int *offsetout) valoff = 0; q = line; + if (external && strcasecmp(params->pm_name, "body") == 0) + continue; + if (strlen(params->pm_name) > CPERLIN) { advise(NULL, "Parameter name \"%s\" is too long", params->pm_name); if (paramout) @@ -3439,7 +3699,7 @@ output_params(size_t initialwidth, PM params, int *offsetout) return NULL; } - curlen = param_len(params, index, valoff, &eightbit); + curlen = param_len(params, index, valoff, &encode, &cont, &numchars); /* * Loop until we get a parameter that fits within a line. We @@ -3447,22 +3707,11 @@ output_params(size_t initialwidth, PM params, int *offsetout) * on that. */ - while (curlen + 8 > CPERLIN - 1) { - int curvallen = strlen(params->pm_value + valoff) - - (initialwidth + curlen - (CPERLIN - 1)); - + while (cont) { *q++ = ';'; *q++ = '\n'; *q++ = '\t'; - /* - * curvallen holds how many characters we take from this - * current value. Make sure it's at least 1. - */ - - if (curvallen < 1) - curvallen = 1; - /* * At this point we're definitely continuing the line, so * be sure to include the parameter name and section index. @@ -3471,28 +3720,16 @@ output_params(size_t initialwidth, PM params, int *offsetout) q += snprintf(q, sizeof(line) - (q - line), "%s*%d", params->pm_name, index); - /* - * If eightbit was set and we're on index 0, we need to include - * the character set and encode the first section. Otherwise - * only encode if the section we're on contains an 8bit character. - */ - - if (eightbit && index == 0) - encode = 1; - else if (eightbit && contains8bit(params->pm_value + valoff, - params->pm_value + valoff + curvallen)) - encode = 1; - /* * Both of these functions do a NUL termination */ if (encode) i = encode_param(params, q, sizeof(line) - (q - line), - curvallen, valoff, index); + numchars, valoff, index); else i = normal_param(params, q, sizeof(line) - (q - line), - curvallen, valoff); + numchars, valoff); if (i == 0) { if (paramout) @@ -3500,9 +3737,10 @@ output_params(size_t initialwidth, PM params, int *offsetout) return NULL; } - valoff += curvallen; + valoff += numchars; index++; - curlen = param_len(params, index, valoff, &eightbit); + curlen = param_len(params, index, valoff, &encode, &cont, + &numchars); q = line; /* @@ -3546,10 +3784,11 @@ output_params(size_t initialwidth, PM params, int *offsetout) q += snprintf(q, sizeof(line) - (q - line), "%s*%d", params->pm_name, index); } else { - q = strncpy(q, params->pm_name, sizeof(line) - (q - line)); + strncpy(q, params->pm_name, sizeof(line) - (q - line)); + q += strlen(q); } - if (eightbit) + if (encode) i = encode_param(params, q, sizeof(line) - (q - line), strlen(params->pm_value + valoff), valoff, index); else @@ -3575,65 +3814,125 @@ output_params(size_t initialwidth, PM params, int *offsetout) } /* - * Calculate the size of a parameter. Include any necessary encoding. - * Start the length computation from where "offset" is marked. + * Calculate the size of a parameter. + * + * Arguments include + * + * pm - The parameter being output + * index - If continuing the parameter, the index of the section + * we're on. + * valueoff - The current offset into the parameter value that we're + * working on (previous sections have consumed valueoff bytes). + * encode - Set if we should perform encoding on this parameter section + * (given that we're consuming bytesfit bytes). + * cont - Set if the remaining data in value will not fit on a single + * line and will need to be continued. + * bytesfit - The number of bytes that we can consume from the parameter + * value and still fit on a completely new line. The + * calculation assumes the new line starts with a tab, + * includes the parameter name and any encoding, and fits + * within CPERLIN bytes. Will always be at least 1. */ static size_t -param_len(PM pm, int index, size_t valueoff, int *eightbit) +param_len(PM pm, int index, size_t valueoff, int *encode, int *cont, + size_t *bytesfit) { - char *start = pm->pm_value + valueoff, *p; - size_t len = 0; + char *start = pm->pm_value + valueoff, *p, indexchar[32]; + size_t len = 0, fit = 0; + int fitlimit = 0, eightbit, maxfit; + + *encode = 0; /* - * Add up the length. First, start with the parameter name, and include - * the equal sign. + * Add up the length. First, start with the parameter name. */ - len += strlen(pm->pm_name) + 1; + len = strlen(pm->pm_name); /* - * Scan the parameter value. If we find an 8-bit character, then - * we need to compute the locale name for the length. + * Scan the parameter value and see if we need to do encoding for this + * section. */ - *eightbit = contains8bit(start, NULL); + eightbit = contains8bit(start, NULL); /* - * If we've got 8-bit character, put the locale on the front (if we're - * doing part 0. Also compute the length of the string based on the - * encoding we need to do. + * Determine if we need to encode this section. Encoding is necessary if: + * + * - There are any 8-bit characters at all and we're on the first + * section. + * - There are 8-bit characters within N bytes of our section start. + * N is calculated based on the number of bytes it would take to + * reach CPERLIN. Specifically: + * 8 (starting tab) + + * strlen(param name) + + * 4 ('* for section marker, '=', opening/closing '"') + * strlen (index) + * is the number of bytes used by everything that isn't part of the + * value. So that gets subtracted from CPERLIN. */ - if (*eightbit) { + snprintf(indexchar, sizeof(indexchar), "%d", index); + maxfit = CPERLIN - (12 + len + strlen(indexchar)); + if ((eightbit && index == 0) || contains8bit(start, start + maxfit)) { + *encode = 1; + } + + len++; /* Add in equal sign */ + + if (*encode) { + /* + * We're using maxfit as a marker for how many characters we can + * fit into the line. Bump it by two because we're not using quotes + * when encoding. + */ + + maxfit += 2; + /* * If we don't have a charset or language tag in this parameter, * add them now. */ - if (! pm->pm_charset) + if (! pm->pm_charset) { pm->pm_charset = getcpy(write_charset_8bit()); + if (strcasecmp(pm->pm_charset, "US-ASCII") == 0) + adios(NULL, "8-bit characters in parameter \"%s\", but " + "local character set is US-ASCII", pm->pm_name); + } if (! pm->pm_lang) pm->pm_lang = getcpy(NULL); /* Default to a blank lang tag */ - len++; /* For the encoding we need to do */ + len++; /* For the encoding marker */ + maxfit--; if (index == 0) { - len += strlen(pm->pm_charset) + strlen(pm->pm_lang) + 2; + int enclen = strlen(pm->pm_charset) + strlen(pm->pm_lang) + 2; + len += enclen; + maxfit-= enclen; } else { /* - * We know we definitely need to include an index. - * This will get the length wrong if we have more than 99 - * sections. I can live with that. + * We know we definitely need to include an index. maxfit already + * includes the section marker. */ - len += 2; /* * */ - if (index > 9) - len++; + len += strlen(indexchar); } for (p = start; *p != '\0'; p++) { - if (isparamencode(*p)) + if (isparamencode(*p)) { len += 3; - else + maxfit -= 3; + } else { len++; + maxfit--; + } + /* + * Just so there's no confusion: maxfit is counting OUTPUT + * characters (post-encoding). fit is counting INPUT characters. + */ + if (! fitlimit && maxfit >= 0) + fit++; + else if (! fitlimit) + fitlimit++; } } else { /* @@ -3646,15 +3945,27 @@ param_len(PM pm, int index, size_t valueoff, int *eightbit) case '"': case '\\': len++; + maxfit--; /* FALL THROUGH */ default: len++; + maxfit--; } + if (! fitlimit && maxfit >= 0) + fit++; + else if (! fitlimit) + fitlimit++; } len += 2; } + if (fit < 1) + fit = 1; + + *cont = fitlimit; + *bytesfit = fit; + return len; } @@ -3662,7 +3973,7 @@ param_len(PM pm, int index, size_t valueoff, int *eightbit) * Output an encoded parameter string. */ -static size_t +size_t encode_param(PM pm, char *output, size_t len, size_t valuelen, size_t valueoff, int index) { @@ -3769,14 +4080,14 @@ normal_param(PM pm, char *output, size_t len, size_t valuelen, */ PM -add_param(PM *first, PM *last, const char *name, const char *value) +add_param(PM *first, PM *last, char *name, char *value, int nocopy) { - PM pm = mh_xmalloc(sizeof(*pm)); + PM pm; + NEW(pm); memset(pm, 0, sizeof(*pm)); - - pm->pm_name = getcpy(name); - pm->pm_value = getcpy(value); + pm->pm_name = nocopy ? name : getcpy(name); + pm->pm_value = nocopy ? value : getcpy(value); if (*first) { (*last)->pm_next = pm; @@ -3788,3 +4099,165 @@ add_param(PM *first, PM *last, const char *name, const char *value) return pm; } + +/* + * Either replace a current parameter with a new value, or add the parameter + * to the parameter linked list. + */ + +PM +replace_param(PM *first, PM *last, char *name, char *value, int nocopy) +{ + PM pm; + + for (pm = *first; pm != NULL; pm = pm->pm_next) { + if (strcasecmp(name, pm->pm_name) == 0) { + /* + * If nocopy is set, it's assumed that we own both name + * and value. We don't need name, so we discard it now. + */ + if (nocopy) + free(name); + free(pm->pm_value); + pm->pm_value = nocopy ? value : getcpy(value); + return pm; + } + } + + return add_param(first, last, name, value, nocopy); +} + +/* + * Retrieve a parameter value from a parameter linked list. If the parameter + * value needs converted to the local character set, do that now. + */ + +char * +get_param(PM first, const char *name, char replace, int fetchonly) +{ + while (first != NULL) { + if (strcasecmp(name, first->pm_name) == 0) { + if (fetchonly) + return first->pm_value; + else + return getcpy(get_param_value(first, replace)); + } + first = first->pm_next; + } + + return NULL; +} + +/* + * Return a parameter value, converting to the local character set if + * necessary + */ + +char *get_param_value(PM pm, char replace) +{ + static char buffer[4096]; /* I hope no parameters are larger */ + size_t bufsize = sizeof(buffer); +#ifdef HAVE_ICONV + size_t inbytes; + int utf8; + iconv_t cd; + ICONV_CONST char *p; +#else /* HAVE_ICONV */ + char *p; +#endif /* HAVE_ICONV */ + + char *q; + + /* + * If we don't have a character set indicated, it's assumed to be + * US-ASCII. If it matches our character set, we don't need to convert + * anything. + */ + + if (!pm->pm_charset || check_charset(pm->pm_charset, + strlen(pm->pm_charset))) { + return pm->pm_value; + } + + /* + * In this case, we need to convert. If we have iconv support, use + * that. Otherwise, go through and simply replace every non-ASCII + * character with the substitution character. + */ + +#ifdef HAVE_ICONV + q = buffer; + bufsize = sizeof(buffer); + utf8 = strcasecmp(pm->pm_charset, "UTF-8") == 0; + + cd = iconv_open(get_charset(), pm->pm_charset); + if (cd == (iconv_t) -1) { + goto noiconv; + } + + inbytes = strlen(pm->pm_value); + p = pm->pm_value; + + while (inbytes) { + if (iconv(cd, &p, &inbytes, &q, &bufsize) == (size_t)-1) { + if (errno != EILSEQ) { + iconv_close(cd); + goto noiconv; + } + /* + * Reset shift state, substitute our character, + * try to restart conversion. + */ + + iconv(cd, NULL, NULL, &q, &bufsize); + + if (bufsize == 0) { + iconv_close(cd); + goto noiconv; + } + *q++ = replace; + bufsize--; + if (bufsize == 0) { + iconv_close(cd); + goto noiconv; + } + if (utf8) { + for (++p, --inbytes; + inbytes > 0 && (((unsigned char) *p) & 0xc0) == 0x80; + ++p, --inbytes) + continue; + } else { + p++; + inbytes--; + } + } + } + + iconv_close(cd); + + if (bufsize == 0) + q--; + *q = '\0'; + + return buffer; + +noiconv: +#endif /* HAVE_ICONV */ + + /* + * Take everything non-ASCII and substituite the replacement character + */ + + q = buffer; + bufsize = sizeof(buffer); + for (p = pm->pm_value; *p != '\0' && bufsize > 1; p++, q++, bufsize--) { + if (isascii((unsigned char) *p) && !iscntrl((unsigned char) *p)) + *q = *p; + else + *q = replace; + } + + *q = '\0'; + + return buffer; +}