X-Git-Url: https://diplodocus.org/git/nmh/blobdiff_plain/29997d403aaba5b114ca7dc783930f96d3615ebc..ef1ba39e8dae81091b6c3e73e72825ef6edea3c6:/uip/mhbuildsbr.c diff --git a/uip/mhbuildsbr.c b/uip/mhbuildsbr.c index 42526cf3..fc90dc01 100644 --- a/uip/mhbuildsbr.c +++ b/uip/mhbuildsbr.c @@ -17,7 +17,6 @@ #include #include -#include #include #include #include @@ -48,6 +47,13 @@ struct attach_list { struct attach_list *next; }; +typedef struct convert_list { + char *type; + char *filename; + char *argstring; + struct convert_list *next; +} convert_list; + /* * Maximum size of URL token in message/external-body */ @@ -62,6 +68,8 @@ void content_error (char *, CT, char *, ...); int find_cache (CT, int, int *, char *, char *, int); /* mhfree.c */ +extern CT *cts; +void freects_done (int) NORETURN; void free_ctinfo (CT); void free_encoding (CT, int); @@ -70,6 +78,12 @@ void free_encoding (CT, int); */ static int init_decoded_content (CT, const char *); static void setup_attach_content(CT, char *); +static void set_disposition (CT); +static void set_charset (CT, int); +static void expand_pseudoheaders (CT, struct multipart *, const char *, + const convert_list *); +static void expand_pseudoheader (CT, CT *, struct multipart *, const char *, + const char *, const char *); static char *fgetstr (char *, int, FILE *); static int user_content (FILE *, char *, CT *, const char *infilename); static void set_id (CT, int); @@ -77,6 +91,7 @@ static int compose_content (CT, int); static int scan_content (CT, size_t); static int build_headers (CT, int); static char *calculate_digest (CT, int); +static int extract_headers (CT, char *, FILE **); static unsigned char directives_stack[32]; @@ -133,6 +148,7 @@ build_mime (char *infile, int autobuild, int dist, int directives, HF hp; m_getfld_state_t gstate = 0; struct attach_list *attach_head = NULL, *attach_tail = NULL, *at_entry; + convert_list *convert_head = NULL, *convert_tail = NULL, *convert; directive_init(directives); @@ -145,8 +161,7 @@ build_mime (char *infile, int autobuild, int dist, int directives, /* * Allocate space for primary (outside) content */ - if ((ct = (CT) calloc (1, sizeof(*ct))) == NULL) - adios (NULL, "out of memory"); + NEW0(ct); /* * Allocate structure for handling decoded content @@ -170,13 +185,13 @@ build_mime (char *infile, int autobuild, int dist, int directives, /* abort if draft has Mime-Version or C-T-E header field */ if (strcasecmp (name, VRSN_FIELD) == 0 || - strcasecmp (name, ENCODING_FIELD) == 0) { + strcasecmp (name, ENCODING_FIELD) == 0) { if (autobuild) { fclose(in); + free (ct); return NULL; - } else { - adios (NULL, "draft shouldn't contain %s: field", name); } + adios (NULL, "draft shouldn't contain %s: field", name); } /* ignore any Content-Type fields in the header */ @@ -204,10 +219,11 @@ build_mime (char *infile, int autobuild, int dist, int directives, * header; in that case, add it to our attach list */ - if (strcasecmp(ATTACH_FIELD, np) == 0) { - struct attach_list *entry; + if (strcasecmp(ATTACH_FIELD, np) == 0 || + strcasecmp(ATTACH_FIELD_ALT, np) == 0) { + struct attach_list *entry; char *s = vp, *e = vp + strlen(vp) - 1; - free(np); + free(np); /* * Make sure we can find the start of this filename. @@ -226,8 +242,8 @@ build_mime (char *infile, int autobuild, int dist, int directives, goto finish_field; } - entry = mh_xmalloc(sizeof(*entry)); - entry->filename = getcpy(s); + NEW(entry); + entry->filename = mh_xstrdup(s); entry->next = NULL; free(vp); @@ -237,6 +253,86 @@ build_mime (char *infile, int autobuild, int dist, int directives, } else { attach_head = attach_tail = entry; } + } else if (strncasecmp(MHBUILD_FILE_PSEUDOHEADER, np, + strlen (MHBUILD_FILE_PSEUDOHEADER)) == 0) { + /* E.g., + * Nmh-mhbuild-file-text/calendar: /home/user/Mail/inbox/9 + */ + char *type = np + strlen (MHBUILD_FILE_PSEUDOHEADER); + char *filename = vp; + + /* vp should begin with a space because m_getfld() + includes the space after the colon in buf. */ + while (isspace((unsigned char) *filename)) { ++filename; } + /* Trim trailing newline and any other whitespace. */ + rtrim (filename); + + for (convert = convert_head; convert; convert = convert->next) { + if (strcasecmp (convert->type, type) == 0) { break; } + } + if (convert) { + if (convert->filename && + strcasecmp (convert->filename, filename)) { + adios (NULL, "Multiple %s headers with different files" + " not allowed", type); + } else { + convert->filename = mh_xstrdup(filename); + } + } else { + NEW0(convert); + convert->filename = mh_xstrdup(filename); + convert->type = mh_xstrdup(type); + + if (convert_tail) { + convert_tail->next = convert; + } else { + convert_head = convert; + } + convert_tail = convert; + } + + free (vp); + free (np); + } else if (strncasecmp(MHBUILD_ARGS_PSEUDOHEADER, np, + strlen (MHBUILD_ARGS_PSEUDOHEADER)) == 0) { + /* E.g., + * Nmh-mhbuild-args-text/calendar: -reply accept + */ + char *type = np + strlen (MHBUILD_ARGS_PSEUDOHEADER); + char *argstring = vp; + + /* vp should begin with a space because m_getfld() + includes the space after the colon in buf. */ + while (isspace((unsigned char) *argstring)) { ++argstring; } + /* Trim trailing newline and any other whitespace. */ + rtrim (argstring); + + for (convert = convert_head; convert; convert = convert->next) { + if (strcasecmp (convert->type, type) == 0) { break; } + } + if (convert) { + if (convert->argstring && + strcasecmp (convert->argstring, argstring)) { + adios (NULL, "Multiple %s headers with different " + "argstrings not allowed", type); + } else { + convert->argstring = mh_xstrdup(argstring); + } + } else { + NEW0(convert); + convert->type = mh_xstrdup(type); + convert->argstring = mh_xstrdup(argstring); + + if (convert_tail) { + convert_tail->next = convert; + } else { + convert_head = convert; + } + convert_tail = convert; + } + + free (vp); + free (np); } else { add_header (ct, np, vp); } @@ -262,15 +358,17 @@ finish_field: } m_getfld_state_destroy (&gstate); - /* - * Iterate through the list of headers and call the function to MIME-ify - * them if required. - */ + if (header_encoding != CE_8BIT) { + /* + * Iterate through the list of headers and call the function to MIME-ify + * them if required. + */ - for (hp = ct->c_first_hf; hp != NULL; hp = hp->next) { - if (encode_rfc2047(hp->name, &hp->value, header_encoding, NULL)) { - adios(NULL, "Unable to encode header \"%s\"", hp->name); - } + for (hp = ct->c_first_hf; hp != NULL; hp = hp->next) { + if (encode_rfc2047(hp->name, &hp->value, header_encoding, NULL)) { + adios(NULL, "Unable to encode header \"%s\"", hp->name); + } + } } /* @@ -294,8 +392,7 @@ finish_field: ct->c_type = CT_MULTIPART; ct->c_subtype = MULTI_MIXED; - if ((m = (struct multipart *) calloc (1, sizeof(*m))) == NULL) - adios (NULL, "out of memory"); + NEW0(m); ct->c_ctparams = (void *) m; pp = &m->mp_parts; @@ -314,8 +411,7 @@ finish_field: if (!p) continue; - if ((part = (struct part *) calloc (1, sizeof(*part))) == NULL) - adios (NULL, "out of memory"); + NEW0(part); *pp = part; pp = &part->mp_next; part->mp_part = p; @@ -327,17 +423,15 @@ finish_field: */ for (at_entry = attach_head; at_entry; ) { - struct attach_list *at_prev = at_entry; - struct part *part; + struct attach_list *at_prev = at_entry; + struct part *part; CT p; if (access(at_entry->filename, R_OK) != 0) { adios("reading", "Unable to open %s for", at_entry->filename); } - if ((p = (CT) calloc (1, sizeof(*p))) == NULL) - adios(NULL, "out of memory"); - + NEW0(p); init_decoded_content(p, infile); /* @@ -348,8 +442,7 @@ finish_field: setup_attach_content(p, at_entry->filename); - if ((part = (struct part *) calloc (1, sizeof(*part))) == NULL) - adios (NULL, "out of memory"); + NEW0(part); *pp = part; pp = &part->mp_next; part->mp_part = p; @@ -359,19 +452,50 @@ finish_field: free(at_prev); } + /* + * Handle the mhbuild pseudoheaders, which deal with specific + * content types. + */ + if (convert_head) { + CT *ctp; + convert_list *next; + + done = freects_done; + + /* In case there are multiple calls that land here, prevent leak. */ + for (ctp = cts; ctp && *ctp; ++ctp) { free_content (*ctp); } + free (cts); + + /* Extract the type part (as a CT) from filename. */ + cts = mh_xcalloc(2, sizeof *cts); + if (! (cts[0] = parse_mime (convert_head->filename))) { + adios (NULL, "failed to parse %s", convert_head->filename); + } + + expand_pseudoheaders (cts[0], m, infile, convert_head); + + /* Free the convert list. */ + for (convert = convert_head; convert; convert = next) { + next = convert->next; + free (convert->type); + free (convert->filename); + free (convert->argstring); + free (convert); + } + convert_head = NULL; + } + /* * To allow for empty message bodies, if we've found NO content at all * yet cook up an empty text/plain part. */ if (!m->mp_parts) { - CT p; - struct part *part; + CT p; + struct part *part; struct text *t; - if ((p = (CT) calloc (1, sizeof(*p))) == NULL) - adios(NULL, "out of memory"); - + NEW0(p); init_decoded_content(p, infile); if (get_ctinfo ("text/plain", p, 0) == NOTOK) @@ -385,20 +509,16 @@ finish_field: * So this seems like the best option available since we're going * to call scan_content() on this. */ - p->c_cefile.ce_file = getcpy("/dev/null"); + p->c_cefile.ce_file = mh_xstrdup("/dev/null"); p->c_begin = ftell(in); p->c_end = ftell(in); - if ((t = (struct text *) calloc (1, sizeof (*t))) == NULL) - adios (NULL, "out of memory"); - + NEW0(t); t->tx_charset = CHARSET_SPECIFIED; p->c_ctparams = t; - if ((part = (struct part *) calloc (1, sizeof(*part))) == NULL) - adios (NULL, "out of memory"); + NEW0(part); *pp = part; - pp = &part->mp_next; part->mp_part = p; } @@ -439,6 +559,13 @@ finish_field: if ((cp = strchr(prefix, 'a')) == NULL) adios (NULL, "internal error(4)"); + /* + * If using EAI, force 8-bit charset. + */ + if (header_encoding == CE_8BIT) { + set_charset (ct, 1); + } + /* * Scan the contents. Choose a transfer encoding, and * check if prefix for multipart boundary clashes with @@ -450,7 +577,7 @@ finish_field: } else { if (*++cp == 0) adios (NULL, "giving up trying to find a unique delimiter string"); - else + else (*cp)++; } } @@ -544,8 +671,7 @@ user_content (FILE *in, char *buf, CT *ctp, const char *infilename) } /* allocate basic Content structure */ - if ((ct = (CT) calloc (1, sizeof(*ct))) == NULL) - adios (NULL, "out of memory"); + NEW0(ct); *ctp = ct; /* allocate basic structure for handling decoded content */ @@ -584,9 +710,8 @@ user_content (FILE *in, char *buf, CT *ctp, const char *infilename) strncpy (content, buf + 2, sizeof(content)); inlineD = 1; goto rock_and_roll; - } else { - inlineD = 0; } + inlineD = 0; /* the directive is implicit */ strncpy (content, "text/plain", sizeof(content)); @@ -776,17 +901,15 @@ use_forw: * reference, we need to create another Content structure * for the message/external-body to wrap it in. */ - if ((ct = (CT) calloc (1, sizeof(*ct))) == NULL) - adios (NULL, "out of memory"); + NEW0(ct); + init_decoded_content(ct, infilename); *ctp = ct; - ci = &ct->c_ctinfo; if (get_ctinfo (buffer, ct, 0) == NOTOK) done (1); ct->c_type = CT_MESSAGE; ct->c_subtype = MESSAGE_EXTERNAL; - if ((e = (struct exbody *) calloc (1, sizeof(*e))) == NULL) - adios (NULL, "out of memory"); + NEW0(e); ct->c_ctparams = (void *) e; e->eb_parent = ct; @@ -826,14 +949,10 @@ use_forw: * No [file] argument, so check profile for * method to compose content. */ - snprintf (buffer, sizeof(buffer), "%s-compose-%s/%s", - invo_name, ci->ci_type, ci->ci_subtype); - if ((cp = context_find (buffer)) == NULL || *cp == '\0') { - snprintf (buffer, sizeof(buffer), "%s-compose-%s", invo_name, ci->ci_type); - if ((cp = context_find (buffer)) == NULL || *cp == '\0') { - content_error (NULL, ct, "don't know how to compose content"); - done (1); - } + cp = context_find_by_type ("compose", ci->ci_type, ci->ci_subtype); + if (cp == NULL) { + content_error (NULL, ct, "don't know how to compose content"); + done (1); } ci->ci_magic = add (cp, NULL); return OK; @@ -899,8 +1018,7 @@ use_forw: ct->c_type = CT_MULTIPART; ct->c_subtype = MULTI_DIGEST; - if ((m = (struct multipart *) calloc (1, sizeof(*m))) == NULL) - adios (NULL, "out of memory"); + NEW0(m); ct->c_ctparams = (void *) m; pp = &m->mp_parts; @@ -910,8 +1028,7 @@ use_forw: CT p; CE pe; - if ((p = (CT) calloc (1, sizeof(*p))) == NULL) - adios (NULL, "out of memory"); + NEW0(p); init_decoded_content (p, infilename); pe = &p->c_cefile; if (get_ctinfo ("message/rfc822", p, 0) == NOTOK) @@ -924,8 +1041,7 @@ use_forw: if (listsw && stat (pe->ce_file, &st) != NOTOK) p->c_end = (long) st.st_size; - if ((part = (struct part *) calloc (1, sizeof(*part))) == NULL) - adios (NULL, "out of memory"); + NEW0(part); *pp = part; pp = &part->mp_next; part->mp_part = p; @@ -985,8 +1101,7 @@ use_forw: ct->c_type = CT_MULTIPART; ct->c_subtype = vrsn; - if ((m = (struct multipart *) calloc (1, sizeof(*m))) == NULL) - adios (NULL, "out of memory"); + NEW0(m); ct->c_ctparams = (void *) m; pp = &m->mp_parts; @@ -1002,8 +1117,7 @@ use_forw: if (!p) continue; - if ((part = (struct part *) calloc (1, sizeof(*part))) == NULL) - adios (NULL, "out of memory"); + NEW0(part); *pp = part; pp = &part->mp_next; part->mp_part = p; @@ -1032,10 +1146,10 @@ set_id (CT ct, int top) time (&clock); snprintf (contentid, sizeof(contentid), "%s\n", message_id (clock, 1)); partno = 0; - msgfmt = getcpy(contentid); + msgfmt = mh_xstrdup(contentid); } snprintf (contentid, sizeof(contentid), msgfmt, top ? 0 : ++partno); - ct->c_id = getcpy (contentid); + ct->c_id = mh_xstrdup(contentid); } @@ -1161,7 +1275,7 @@ compose_content (CT ct, int verbose) for (pm = ci->ci_first_pm; pm; pm = pm->pm_next) { snprintf (bp, buflen, "%s%s=\"%s\"", s, - pm->pm_name, get_param_value(pm, '?')); + pm->pm_name, get_param_value(pm, '?')); len = strlen (bp); bp += len; buflen -= len; @@ -1278,13 +1392,17 @@ raw: static int scan_content (CT ct, size_t maxunencoded) { - int len; + int prefix_len; int check8bit = 0, contains8bit = 0; /* check if contains 8bit data */ - int checklinelen = 0, linelen = 0; /* check for long lines */ + int checknul = 0, containsnul = 0; /* check if contains NULs */ + int checklinelen = 0, linelen = 0; /* check for long lines */ int checkllinelen = 0; /* check for extra-long lines */ int checkboundary = 0, boundaryclash = 0; /* check if clashes with multipart boundary */ int checklinespace = 0, linespace = 0; /* check if any line ends with space */ - char *cp = NULL, buffer[BUFSIZ]; + char *cp = NULL; + char *bufp = NULL; + size_t buflen; + ssize_t gotlen; struct text *t = NULL; FILE *in = NULL; CE ce = &ct->c_cefile; @@ -1320,27 +1438,31 @@ scan_content (CT ct, size_t maxunencoded) * Decide what to check while scanning this content. Note that * for text content we always check for 8bit characters if the * charset is unspecified, because that controls whether or not the - * character set is us-ascii or retrieved from the locale. + * character set is us-ascii or retrieved from the locale. And + * we check even if the charset is specified, to allow setting + * the proper Content-Transfer-Encoding. */ if (ct->c_type == CT_TEXT) { t = (struct text *) ct->c_ctparams; - if (t->tx_charset == CHARSET_UNSPECIFIED) - check8bit = 1; + if (t->tx_charset == CHARSET_UNSPECIFIED) { + checknul = 1; + } + check8bit = 1; } switch (ct->c_reqencoding) { case CE_8BIT: - checkllinelen = 1; + checkllinelen = 1; checkboundary = 1; break; case CE_QUOTED: - checkboundary = 1; + checkboundary = 1; break; case CE_BASE64: break; case CE_UNKNOWN: - /* Use the default rules based on content-type */ + /* Use the default rules based on content-type */ switch (ct->c_type) { case CT_TEXT: checkboundary = 1; @@ -1354,21 +1476,24 @@ scan_content (CT ct, size_t maxunencoded) case CT_APPLICATION: check8bit = 1; + checknul = 1; checklinelen = 1; checklinespace = 1; checkboundary = 1; break; case CT_MESSAGE: - check8bit = 0; checklinelen = 0; checklinespace = 0; /* don't check anything for message/external */ - if (ct->c_subtype == MESSAGE_EXTERNAL) + if (ct->c_subtype == MESSAGE_EXTERNAL) { checkboundary = 0; - else + check8bit = 0; + } else { checkboundary = 1; + check8bit = 1; + } break; case CT_AUDIO: @@ -1391,28 +1516,31 @@ scan_content (CT ct, size_t maxunencoded) * Scan the unencoded content */ if (check8bit || checklinelen || checklinespace || checkboundary || - checkllinelen) { + checkllinelen || checknul) { if ((in = fopen (ce->ce_file, "r")) == NULL) adios (ce->ce_file, "unable to open for reading"); - len = strlen (prefix); + prefix_len = strlen (prefix); - while (fgets (buffer, sizeof(buffer) - 1, in)) { + while ((gotlen = getline(&bufp, &buflen, in)) != -1) { /* - * Check for 8bit data. + * Check for 8bit and NUL data. */ - if (check8bit) { - for (cp = buffer; *cp; cp++) { - if (!isascii ((unsigned char) *cp)) { - contains8bit = 1; - check8bit = 0; /* no need to keep checking */ - } + for (cp = bufp; (check8bit || checknul) && + cp < bufp + gotlen; cp++) { + if (!isascii ((unsigned char) *cp)) { + contains8bit = 1; + check8bit = 0; /* no need to keep checking */ + } + if (!*cp) { + containsnul = 1; + checknul = 0; /* no need to keep checking */ } } /* * Check line length. */ - if (checklinelen && (strlen (buffer) > maxunencoded + 1)) { + if (checklinelen && ((size_t)gotlen > maxunencoded + 1)) { linelen = 1; checklinelen = 0; /* no need to keep checking */ } @@ -1424,7 +1552,7 @@ scan_content (CT ct, size_t maxunencoded) */ if (checkllinelen && !linelen && - (strlen(buffer) > MAXLONGLINE + 1)) { + (gotlen > MAXLONGLINE + 1)) { adios(NULL, "Line in content exceeds maximum line limit (%d)", MAXLONGLINE); } @@ -1432,7 +1560,8 @@ scan_content (CT ct, size_t maxunencoded) /* * Check if line ends with a space. */ - if (checklinespace && (cp = buffer + strlen (buffer) - 2) > buffer && isspace ((unsigned char) *cp)) { + if (checklinespace && (cp = bufp + gotlen - 2) > bufp && + isspace ((unsigned char) *cp)) { linespace = 1; checklinespace = 0; /* no need to keep checking */ } @@ -1441,64 +1570,62 @@ scan_content (CT ct, size_t maxunencoded) * Check if content contains a line that clashes * with our standard boundary for multipart messages. */ - if (checkboundary && buffer[0] == '-' && buffer[1] == '-') { - for (cp = buffer + strlen (buffer) - 1; cp >= buffer; cp--) + if (checkboundary && bufp[0] == '-' && bufp[1] == '-') { + for (cp = bufp + gotlen - 1; cp >= bufp; cp--) if (!isspace ((unsigned char) *cp)) break; *++cp = '\0'; - if (!strncmp(buffer + 2, prefix, len) && isdigit((unsigned char) buffer[2 + len])) { + if (!strncmp(bufp + 2, prefix, prefix_len) && + isdigit((unsigned char) bufp[2 + prefix_len])) { boundaryclash = 1; checkboundary = 0; /* no need to keep checking */ } } } fclose (in); + free(bufp); } /* * If the content is text and didn't specify a character set, * we need to figure out which one was used. */ - - if (ct->c_type == CT_TEXT) { - t = (struct text *) ct->c_ctparams; - if (t->tx_charset == CHARSET_UNSPECIFIED) { - CI ci = &ct->c_ctinfo; - - add_param(&ci->ci_first_pm, &ci->ci_last_pm, "charset", - contains8bit ? write_charset_8bit() : "us-ascii", 0); - t->tx_charset = CHARSET_SPECIFIED; - } - } + set_charset (ct, contains8bit); /* * Decide which transfer encoding to use. */ if (ct->c_reqencoding != CE_UNKNOWN) - ct->c_encoding = ct->c_reqencoding; - else + ct->c_encoding = ct->c_reqencoding; + else { + int wants_q_p = (containsnul || linelen || linespace || checksw); + switch (ct->c_type) { case CT_TEXT: - if (contains8bit && !linelen && !linespace && !checksw) - ct->c_encoding = CE_8BIT; - else if (contains8bit || linelen || linespace || checksw) - ct->c_encoding = CE_QUOTED; - else - ct->c_encoding = CE_7BIT; + if (wants_q_p) + ct->c_encoding = CE_QUOTED; + else if (contains8bit) + ct->c_encoding = CE_8BIT; + else + ct->c_encoding = CE_7BIT; + break; case CT_APPLICATION: /* For application type, use base64, except when postscript */ - if (contains8bit || linelen || linespace || checksw) - ct->c_encoding = (ct->c_subtype == APPLICATION_POSTSCRIPT) - ? CE_QUOTED : CE_BASE64; - else + if (wants_q_p || contains8bit) { + if (ct->c_subtype == APPLICATION_POSTSCRIPT) + ct->c_encoding = CE_QUOTED; /* historical */ + else + ct->c_encoding = CE_BASE64; + } else { ct->c_encoding = CE_7BIT; + } break; case CT_MESSAGE: - ct->c_encoding = CE_7BIT; + ct->c_encoding = contains8bit ? CE_8BIT : CE_7BIT; break; case CT_AUDIO: @@ -1508,6 +1635,7 @@ scan_content (CT ct, size_t maxunencoded) ct->c_encoding = CE_BASE64; break; } + } return (boundaryclash ? NOTOK : OK); } @@ -1570,7 +1698,7 @@ build_headers (CT ct, int header_encoding) if (!s) adios(NULL, "Internal error: failed outputting Content-Type " - "parameters"); + "parameters"); vp = add (s, vp); free(s); @@ -1609,8 +1737,11 @@ build_headers (CT ct, int header_encoding) if (ct->c_descr) { np = add (DESCR_FIELD, NULL); vp = concat (" ", ct->c_descr, NULL); - if (encode_rfc2047(DESCR_FIELD, &vp, header_encoding, NULL)) - adios(NULL, "Unable to encode %s header", DESCR_FIELD); + if (header_encoding != CE_8BIT) { + if (encode_rfc2047(DESCR_FIELD, &vp, header_encoding, NULL)) { + adios(NULL, "Unable to encode %s header", DESCR_FIELD); + } + } add_header (ct, np, vp); } @@ -1628,9 +1759,8 @@ build_headers (CT ct, int header_encoding) np = output_params(len, ct->c_dispo_first, NULL, 0); vp = add(np, vp); vp = add("\n", vp); - if (np) - free(np); - add_header (ct, getcpy(DISPO_FIELD), vp); + mh_xfree(np); + add_header (ct, mh_xstrdup(DISPO_FIELD), vp); } skip_headers: @@ -1653,16 +1783,18 @@ skip_headers: /* * output the Content-Transfer-Encoding + * If using EAI and message body is 7-bit, force 8-bit C-T-E. */ + if (header_encoding == CE_8BIT && ct->c_encoding == CE_7BIT) { + ct->c_encoding = CE_8BIT; + } + switch (ct->c_encoding) { case CE_7BIT: /* Nothing to output */ break; case CE_8BIT: - if (ct->c_type == CT_MESSAGE) - adios (NULL, "internal error, invalid encoding"); - np = add (ENCODING_FIELD, NULL); vp = concat (" ", "8bit", "\n", NULL); add_header (ct, np, vp); @@ -1744,7 +1876,7 @@ static char * calculate_digest (CT ct, int asciiP) { int cc; - char buffer[BUFSIZ], *vp, *op; + char *vp, *op; unsigned char *dp; unsigned char digest[16]; unsigned char outbuf[25]; @@ -1762,20 +1894,24 @@ calculate_digest (CT ct, int asciiP) /* calculate md5 message digest */ if (asciiP) { - while (fgets (buffer, sizeof(buffer) - 1, in)) { + char *bufp = NULL; + size_t buflen; + ssize_t gotlen; + while ((gotlen = getline(&bufp, &buflen, in)) != -1) { char c, *cp; - cp = buffer + strlen (buffer) - 1; + cp = bufp + gotlen - 1; if ((c = *cp) == '\n') - *cp = '\0'; + gotlen--; - MD5Update (&mdContext, (unsigned char *) buffer, - (unsigned int) strlen (buffer)); + MD5Update (&mdContext, (unsigned char *) bufp, + (unsigned int) gotlen); if (c == '\n') MD5Update (&mdContext, (unsigned char *) "\r\n", 2); } } else { + char buffer[BUFSIZ]; while ((cc = fread (buffer, sizeof(*buffer), sizeof(buffer), in)) > 0) MD5Update (&mdContext, (unsigned char *) buffer, (unsigned int) cc); } @@ -1799,7 +1935,7 @@ calculate_digest (CT ct, int asciiP) /* encode the digest using base64 */ for (dp = digest, op = (char *) outbuf, - cc = sizeof(digest) / sizeof(digest[0]); + cc = sizeof(digest) / sizeof(digest[0]); cc > 0; cc -= 3, op += 4) { unsigned long bits; char *bp; @@ -1841,7 +1977,7 @@ setup_attach_content(CT ct, char *filename) PM pm; if (! (type = mime_type(filename))) { - adios(NULL, "Unable to determine MIME type of \"%s\"", filename); + adios(NULL, "Unable to determine MIME type of \"%s\"", filename); } /* @@ -1852,12 +1988,12 @@ setup_attach_content(CT ct, char *filename) */ if (get_ctinfo(type, ct, 0) == NOTOK) - done(1); + done(1); free(type); for (s2i = str2cts; s2i->si_key; s2i++) - if (strcasecmp(ct->c_ctinfo.ci_type, s2i->si_key) == 0) + if (strcasecmp(ct->c_ctinfo.ci_type, s2i->si_key) == 0) break; if (!s2i->si_key && !uprf(ct->c_ctinfo.ci_type, "X-")) s2i++; @@ -1868,23 +2004,23 @@ setup_attach_content(CT ct, char *filename) switch (ct->c_type = s2i->si_val) { case CT_MULTIPART: - adios (NULL, "multipart types must be specified by mhbuild directives"); + adios (NULL, "multipart types must be specified by mhbuild directives"); /* NOTREACHED */ case CT_MESSAGE: - if (strcasecmp(ct->c_ctinfo.ci_subtype, "partial") == 0) + if (strcasecmp(ct->c_ctinfo.ci_subtype, "partial") == 0) adios(NULL, "Sorry, %s/%s isn't supported", ct->c_ctinfo.ci_type, - ct->c_ctinfo.ci_subtype); + ct->c_ctinfo.ci_subtype); if (strcasecmp(ct->c_ctinfo.ci_subtype, "external-body") == 0) adios(NULL, "external-body messages must be specified " - "by mhbuild directives"); + "by mhbuild directives"); /* Fall through */ default: - /* + /* * This sets the subtype, if it's significant */ - if ((ct->c_ctinitfnx = s2i->si_init)) + if ((ct->c_ctinitfnx = s2i->si_init)) (*ct->c_ctinitfnx)(ct); break; } @@ -1896,33 +2032,468 @@ setup_attach_content(CT ct, char *filename) for (pm = ct->c_ctinfo.ci_first_pm; pm; pm = pm->pm_next) { if (strcasecmp(pm->pm_name, "name") == 0) { - if (pm->pm_value) - free(pm->pm_value); - pm->pm_value = getcpy(simplename); + mh_xfree(pm->pm_value); + pm->pm_value = mh_xstrdup(simplename); break; } } if (pm == NULL) - add_param(&ct->c_ctinfo.ci_first_pm, &ct->c_ctinfo.ci_last_pm, + add_param(&ct->c_ctinfo.ci_first_pm, &ct->c_ctinfo.ci_last_pm, "name", simplename, 0); - ct->c_descr = getcpy(simplename); + ct->c_descr = mh_xstrdup(simplename); ct->c_descr = add("\n", ct->c_descr); - ct->c_cefile.ce_file = getcpy(filename); + ct->c_cefile.ce_file = mh_xstrdup(filename); - /* - * If it's a text/calendar, we need to make sure it's an inline, - * otherwise it won't work with some calendar programs. Otherwise - * assume attachment - */ + set_disposition (ct); + + add_param(&ct->c_dispo_first, &ct->c_dispo_last, "filename", simplename, 0); +} - if (strcasecmp(ct->c_ctinfo.ci_type, "text") == 0 && - strcasecmp(ct->c_ctinfo.ci_subtype, "calendar") == 0) { - ct->c_dispo_type = getcpy("inline"); +/* + * If disposition type hasn't already been set in ct: + * Look for mhbuild-disposition-/ entry + * that specifies Content-Disposition type. Only + * 'attachment' and 'inline' are allowed. Default to + * 'attachment'. + */ +void +set_disposition (CT ct) { + if (ct->c_dispo_type == NULL) { + char *cp = context_find_by_type ("disposition", ct->c_ctinfo.ci_type, + ct->c_ctinfo.ci_subtype); + + if (cp && strcasecmp (cp, "attachment") && + strcasecmp (cp, "inline")) { + admonish (NULL, "configuration problem: %s-disposition-%s%s%s " + "specifies '%s' but only 'attachment' and 'inline' are " + "allowed", invo_name, + ct->c_ctinfo.ci_type, + ct->c_ctinfo.ci_subtype ? "/" : "", + ct->c_ctinfo.ci_subtype ? ct->c_ctinfo.ci_subtype : "", + cp); + } + + if (!cp) + cp = "attachment"; + ct->c_dispo_type = mh_xstrdup(cp); + } +} + +/* + * Set text content charset if it was unspecified. contains8bit + * selctions: + * 0: content does not contain 8-bit characters + * 1: content contains 8-bit characters + * -1: ignore content and use user's locale to determine charset + */ +void +set_charset (CT ct, int contains8bit) { + if (ct->c_type == CT_TEXT) { + struct text *t; + + if (ct->c_ctparams == NULL) { + NEW0(t); + ct->c_ctparams = t; + t->tx_charset = CHARSET_UNSPECIFIED; + } else { + t = (struct text *) ct->c_ctparams; + } + + if (t->tx_charset == CHARSET_UNSPECIFIED) { + CI ci = &ct->c_ctinfo; + char *eightbitcharset = write_charset_8bit(); + char *charset = contains8bit ? eightbitcharset : "us-ascii"; + + if (contains8bit == 1 && + strcasecmp (eightbitcharset, "US-ASCII") == 0) { + adios (NULL, "Text content contains 8 bit characters, but " + "character set is US-ASCII"); + } + + add_param (&ci->ci_first_pm, &ci->ci_last_pm, "charset", charset, + 0); + + t->tx_charset = CHARSET_SPECIFIED; + } + } +} + + +/* + * Look at all of the replied-to message parts and expand any that + * are matched by a pseudoheader. Except don't descend into + * message parts. + */ +void +expand_pseudoheaders (CT ct, struct multipart *m, const char *infile, + const convert_list *convert_head) { + /* text_plain_ct is used to concatenate all of the text/plain + replies into one part, instead of having each one in a separate + part. */ + CT text_plain_ct = NULL; + + switch (ct->c_type) { + case CT_MULTIPART: { + struct multipart *mp = (struct multipart *) ct->c_ctparams; + struct part *part; + + if (ct->c_subtype == MULTI_ALTERNATE) { + int matched = 0; + + /* The parts are in descending priority order (defined by + RFC 2046 Sec. 5.1.4) because they were reversed by + parse_mime (). So, stop looking for matches with + immediate subparts after the first match of an + alternative. */ + for (part = mp->mp_parts; ! matched && part; part = part->mp_next) { + char *type_subtype = + concat (part->mp_part->c_ctinfo.ci_type, "/", + part->mp_part->c_ctinfo.ci_subtype, NULL); + + if (part->mp_part->c_type == CT_MULTIPART) { + expand_pseudoheaders (part->mp_part, m, infile, + convert_head); + } else { + const convert_list *c; + + for (c = convert_head; c; c = c->next) { + if (! strcasecmp (type_subtype, c->type)) { + expand_pseudoheader (part->mp_part, &text_plain_ct, + m, infile, + c->type, c->argstring); + matched = 1; + break; + } + } + } + free (type_subtype); + } + } else { + for (part = mp->mp_parts; part; part = part->mp_next) { + expand_pseudoheaders (part->mp_part, m, infile, convert_head); + } + } + break; + } + + default: { + char *type_subtype = + concat (ct->c_ctinfo.ci_type, "/", ct->c_ctinfo.ci_subtype, + NULL); + const convert_list *c; + + for (c = convert_head; c; c = c->next) { + if (! strcasecmp (type_subtype, c->type)) { + expand_pseudoheader (ct, &text_plain_ct, m, infile, c->type, + c->argstring); + break; + } + } + free (type_subtype); + break; + } + } +} + + +/* + * Expand a single pseudoheader. It's for the specified type. + */ +void +expand_pseudoheader (CT ct, CT *text_plain_ct, struct multipart *m, + const char *infile, const char *type, + const char *argstring) { + char *reply_file; + FILE *reply_fp = NULL; + char *convert, *type_p, *subtype_p; + char *convert_command; + char *charset = NULL; + char *cp; + struct str2init *s2i; + CT reply_ct; + struct part *part; + int eightbit = 0; + int status; + + type_p = getcpy (type); + if ((subtype_p = strchr (type_p, '/'))) { + *subtype_p++ = '\0'; + convert = context_find_by_type ("convert", type_p, subtype_p); } else { - ct->c_dispo_type = getcpy("attachment"); + free (type_p); + type_p = concat ("mhbuild-convert-", type, NULL); + convert = context_find (type_p); } + free (type_p); - add_param(&ct->c_dispo_first, &ct->c_dispo_last, "filename", simplename, 0); + if (! (convert)) { + /* No mhbuild-convert- entry in mhn.defaults or profile + for type. */ + return; + } + /* reply_file is used to pass the output of the convert. */ + reply_file = getcpy (m_mktemp2 (NULL, invo_name, NULL, NULL)); + convert_command = + concat (convert, " ", argstring ? argstring : "", " >", reply_file, + NULL); + + /* Convert here . . . */ + ct->c_storeproc = mh_xstrdup(convert_command); + ct->c_umask = ~m_gmprot (); + + if ((status = show_content_aux (ct, 0, convert_command, NULL, NULL)) != + OK) { + admonish (NULL, "store of %s content failed", type); + } + free (convert_command); + + /* Fill out the the new ct, reply_ct. */ + NEW0(reply_ct); + init_decoded_content (reply_ct, infile); + + if (extract_headers (reply_ct, reply_file, &reply_fp) == NOTOK) { + free (reply_file); + admonish (NULL, + "failed to extract headers from convert output in %s", + reply_file); + return; + } + + /* For text content only, see if it is 8-bit text. */ + if (reply_ct->c_type == CT_TEXT) { + int fd; + + if ((fd = open (reply_file, O_RDONLY)) == NOTOK || + scan_input (fd, &eightbit) == NOTOK) { + free (reply_file); + admonish (NULL, "failed to read %s", reply_file); + return; + } + (void) close (fd); + } + + /* This sets reply_ct->c_ctparams, and reply_ct->c_termproc if the + charset can't be handled natively. */ + for (s2i = str2cts; s2i->si_key; s2i++) { + if (strcasecmp(reply_ct->c_ctinfo.ci_type, s2i->si_key) == 0) { + break; + } + } + + if ((reply_ct->c_ctinitfnx = s2i->si_init)) { + (*reply_ct->c_ctinitfnx)(reply_ct); + } + + if ((cp = + get_param (reply_ct->c_ctinfo.ci_first_pm, "charset", '?', 1))) { + /* The reply Content-Type had the charset. */ + charset = cp; + } else { + set_charset (reply_ct, -1); + charset = get_param (reply_ct->c_ctinfo.ci_first_pm, "charset", '?', 1); + if (reply_ct->c_reqencoding == CE_UNKNOWN && + reply_ct->c_type == CT_TEXT) { + /* Assume that 8bit is sufficient (for text). In other words, + don't allow it to be encoded as quoted printable if lines + are too long. This also sidesteps the check for whether + it needs to be encoded as binary; instead, it relies on + the applicable mhbuild-convert-text directive to ensure + that the resultant text is not binary. */ + reply_ct->c_reqencoding = eightbit ? CE_8BIT : CE_7BIT; + } + } + + /* Concatenate text/plain parts. */ + if (reply_ct->c_type == CT_TEXT && + reply_ct->c_subtype == TEXT_PLAIN) { + if (! *text_plain_ct && m->mp_parts && m->mp_parts->mp_part && + m->mp_parts->mp_part->c_type == CT_TEXT && + m->mp_parts->mp_part->c_subtype == TEXT_PLAIN) { + *text_plain_ct = m->mp_parts->mp_part; + /* Make sure that the charset is set in the text/plain + part. */ + set_charset (*text_plain_ct, -1); + if ((*text_plain_ct)->c_reqencoding == CE_UNKNOWN) { + /* Assume that 8bit is sufficient (for text). In other words, + don't allow it to be encoded as quoted printable if lines + are too long. This also sidesteps the check for whether + it needs to be encoded as binary; instead, it relies on + the applicable mhbuild-convert-text directive to ensure + that the resultant text is not binary. */ + (*text_plain_ct)->c_reqencoding = + eightbit ? CE_8BIT : CE_7BIT; + } + } + + if (*text_plain_ct) { + /* Only concatenate if the charsets are identical. */ + char *text_plain_ct_charset = + get_param ((*text_plain_ct)->c_ctinfo.ci_first_pm, "charset", + '?', 1); + + if (strcasecmp (text_plain_ct_charset, charset) == 0) { + /* Append this text/plain reply to the first one. + If there's a problem anywhere along the way, + instead attach it is a separate part. */ + int text_plain_reply = + open ((*text_plain_ct)->c_cefile.ce_file, + O_WRONLY | O_APPEND); + int addl_reply = open (reply_file, O_RDONLY); + + if (text_plain_reply != NOTOK && addl_reply != NOTOK) { + /* Insert blank line before each addl part. */ + /* It would be nice not to do this for the first one. */ + if (write (text_plain_reply, "\n", 1) == 1) { + /* Copy the text from the new reply and + then free its Content struct. */ + cpydata (addl_reply, text_plain_reply, + (*text_plain_ct)->c_cefile.ce_file, + reply_file); + if (close (text_plain_reply) == OK && + close (addl_reply) == OK) { + /* If appended text needed 8-bit but first text didn't, + propagate the 8-bit indication. */ + if ((*text_plain_ct)->c_reqencoding == CE_7BIT && + reply_ct->c_reqencoding == CE_8BIT) { + (*text_plain_ct)->c_reqencoding = CE_8BIT; + } + + if (reply_fp) { fclose (reply_fp); } + free (reply_file); + free_content (reply_ct); + return; + } + } + } + } + } else { + *text_plain_ct = reply_ct; + } + } + + reply_ct->c_cefile.ce_file = reply_file; + reply_ct->c_cefile.ce_fp = reply_fp; + reply_ct->c_cefile.ce_unlink = 1; + + /* Attach the new part to the parent multipart/mixed, "m". */ + NEW0(part); + part->mp_part = reply_ct; + if (m->mp_parts) { + struct part *p; + + for (p = m->mp_parts; p && p->mp_next; p = p->mp_next) { continue; } + p->mp_next = part; + } else { + m->mp_parts = part; + } +} + + +/* Extract any Content-Type header from beginning of convert output. */ +int +extract_headers (CT ct, char *reply_file, FILE **reply_fp) { + char *buffer = NULL, *cp, *end_of_header; + int found_header = 0; + struct stat statbuf; + + /* Read the convert reply from the file to memory. */ + if (stat (reply_file, &statbuf) == NOTOK) { + admonish (reply_file, "failed to stat"); + goto failed_to_extract_ct; + } + + buffer = mh_xmalloc (statbuf.st_size + 1); + + if ((*reply_fp = fopen (reply_file, "r+")) == NULL || + fread (buffer, 1, (size_t) statbuf.st_size, *reply_fp) < + (size_t) statbuf.st_size) { + admonish (reply_file, "failed to read"); + goto failed_to_extract_ct; + } + buffer[statbuf.st_size] = '\0'; + + /* Look for a header in the convert reply. */ + if (strncasecmp (buffer, TYPE_FIELD, strlen (TYPE_FIELD)) == 0 && + buffer[strlen (TYPE_FIELD)] == ':') { + if ((end_of_header = strstr (buffer, "\r\n\r\n"))) { + end_of_header += 2; + found_header = 1; + } else if ((end_of_header = strstr (buffer, "\n\n"))) { + ++end_of_header; + found_header = 1; + } + } + + if (found_header) { + CT tmp_ct; + char *tmp_file; + FILE *tmp_f; + size_t n; + + /* Truncate buffer to just the C-T. */ + *end_of_header = '\0'; + n = strlen (buffer); + + if (get_ctinfo (buffer + 14, ct, 0) != OK) { + admonish (NULL, "unable to get content info for reply"); + goto failed_to_extract_ct; + } + + /* Hack. Use parse_mime() to detect the type/subtype of the + reply, which we'll use below. */ + tmp_file = getcpy (m_mktemp2 (NULL, invo_name, NULL, NULL)); + if ((tmp_f = fopen (tmp_file, "w")) && + fwrite (buffer, 1, n, tmp_f) == n) { + fclose (tmp_f); + } else { + goto failed_to_extract_ct; + } + tmp_ct = parse_mime (tmp_file); + + if (tmp_ct) { + /* The type and subtype were detected from the reply + using parse_mime() above. */ + ct->c_type = tmp_ct->c_type; + ct->c_subtype = tmp_ct->c_subtype; + free_content (tmp_ct); + } + + free (tmp_file); + + /* Rewrite the content without the header. */ + cp = end_of_header + 1; + rewind (*reply_fp); + + if (fwrite (cp, 1, statbuf.st_size - (cp - buffer), *reply_fp) < + (size_t) (statbuf.st_size - (cp - buffer))) { + admonish (reply_file, "failed to write"); + goto failed_to_extract_ct; + } + + if (ftruncate (fileno (*reply_fp), statbuf.st_size - (cp - buffer)) != + 0) { + advise (reply_file, "ftruncate"); + goto failed_to_extract_ct; + } + } else { + /* No header section, assume the reply is text/plain. */ + ct->c_type = CT_TEXT; + ct->c_subtype = TEXT_PLAIN; + if (get_ctinfo ("text/plain", ct, 0) == NOTOK) { + /* This never should fail, but just in case. */ + adios (NULL, "unable to get content info for reply"); + } + } + + /* free_encoding() will close reply_fp, which is passed through + ct->c_cefile.ce_fp. */ + free (buffer); + return OK; + +failed_to_extract_ct: + if (*reply_fp) { fclose (*reply_fp); } + free (buffer); + return NOTOK; }