X-Git-Url: https://diplodocus.org/git/nmh/blobdiff_plain/337b4e616e8f53ba06285b1645e1df9918ed5c16..e35fb433:/uip/mhbuildsbr.c?ds=inline diff --git a/uip/mhbuildsbr.c b/uip/mhbuildsbr.c index 68c555d3..4af6e590 100644 --- a/uip/mhbuildsbr.c +++ b/uip/mhbuildsbr.c @@ -1,6 +1,4 @@ - -/* - * mhbuildsbr.c -- routines to expand/translate MIME composition files +/* mhbuildsbr.c -- routines to expand/translate MIME composition files * * This code is Copyright (c) 2002, by the authors of nmh. See the * COPYRIGHT file in the root directory of the nmh distribution for @@ -17,13 +15,20 @@ #include #include -#include #include #include #include +#include #include #include #include +#include "h/mhcachesbr.h" +#include "mhmisc.h" +#include "sbr/m_mktemp.h" +#include "sbr/message_id.h" +#include "sbr/mime_type.h" +#include "mhfree.h" +#include "mhshowsbr.h" #ifdef HAVE_SYS_TIME_H # include @@ -37,10 +42,6 @@ extern int listsw; extern int rfc934sw; extern int contentidsw; -/* cache policies */ -extern int rcachesw; /* mhcachesbr.c */ -extern int wcachesw; /* mhcachesbr.c */ - static char prefix[] = "----- =_aaaaaaaaaa"; struct attach_list { @@ -48,28 +49,25 @@ struct attach_list { struct attach_list *next; }; -/* - * Maximum size of URL token in message/external-body - */ - -#define MAXURLTOKEN 40 - - -/* mhmisc.c */ -void content_error (char *, CT, char *, ...); - -/* mhcachesbr.c */ -int find_cache (CT, int, int *, char *, char *, int); +typedef struct convert_list { + char *type; + char *filename; + char *argstring; + struct convert_list *next; +} convert_list; -/* mhfree.c */ -void free_ctinfo (CT); -void free_encoding (CT, int); /* * static prototypes */ 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 +75,7 @@ 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]; @@ -124,15 +123,16 @@ build_mime (char *infile, int autobuild, int dist, int directives, int header_encoding, size_t maxunencoded, int verbose) { int compnum, state; - char buf[BUFSIZ], name[NAMESZ]; + char buf[NMH_BUFSIZ], name[NAMESZ]; char *cp, *np, *vp; struct multipart *m; struct part **pp; CT ct; FILE *in; HF hp; - m_getfld_state_t gstate = 0; + m_getfld_state_t gstate; 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 +145,7 @@ 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 @@ -160,10 +159,11 @@ build_mime (char *infile, int autobuild, int dist, int directives, * draft into the linked list of header fields for * the new MIME message. */ - m_getfld_track_filepos (&gstate, in); + gstate = m_getfld_state_init(in); + m_getfld_track_filepos2(&gstate); for (compnum = 1;;) { int bufsz = sizeof buf; - switch (state = m_getfld (&gstate, name, buf, &bufsz, in)) { + switch (state = m_getfld2(&gstate, name, buf, &bufsz)) { case FLD: case FLDPLUS: compnum++; @@ -175,28 +175,27 @@ build_mime (char *infile, int autobuild, int dist, int directives, 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 */ if (!strcasecmp (name, TYPE_FIELD)) { while (state == FLDPLUS) { bufsz = sizeof buf; - state = m_getfld (&gstate, name, buf, &bufsz, in); + state = m_getfld2(&gstate, name, buf, &bufsz); } goto finish_field; } /* get copies of the buffers */ - np = add (name, NULL); - vp = add (buf, NULL); + np = mh_xstrdup(name); + vp = mh_xstrdup(buf); /* if necessary, get rest of field */ while (state == FLDPLUS) { bufsz = sizeof buf; - state = m_getfld (&gstate, name, buf, &bufsz, in); + state = m_getfld2(&gstate, name, buf, &bufsz); vp = add (buf, vp); /* add to previous value */ } @@ -205,7 +204,8 @@ 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) { + 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); @@ -227,8 +227,8 @@ 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); @@ -238,6 +238,86 @@ build_mime (char *infile, int autobuild, int dist, int directives, } else { attach_head = attach_tail = entry; } + } else if (strncasecmp(MHBUILD_FILE_PSEUDOHEADER, np, + LEN(MHBUILD_FILE_PSEUDOHEADER)) == 0) { + /* E.g., + * Nmh-mhbuild-file-text/calendar: /home/user/Mail/inbox/9 + */ + char *type = np + LEN(MHBUILD_FILE_PSEUDOHEADER); + char *filename = vp; + + /* vp should begin with a space because m_getfld2() + 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, + LEN(MHBUILD_ARGS_PSEUDOHEADER)) == 0) { + /* E.g., + * Nmh-mhbuild-args-text/calendar: -reply accept + */ + char *type = np + LEN(MHBUILD_ARGS_PSEUDOHEADER); + char *argstring = vp; + + /* vp should begin with a space because m_getfld2() + 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); } @@ -248,7 +328,7 @@ finish_field: case BODY: fseek (in, (long) (-strlen (buf)), SEEK_CUR); - /* fall through */ + break; case FILEEOF: break; @@ -263,15 +343,17 @@ 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); + } + } } /* @@ -280,13 +362,13 @@ finish_field: */ if (! dist) { - np = add (VRSN_FIELD, NULL); + np = mh_xstrdup(VRSN_FIELD); vp = concat (" ", VRSN_VALUE, "\n", NULL); add_header (ct, np, vp); } /* - * We initally assume we will find multiple contents in the + * We initially assume we will find multiple contents in the * draft. So create a multipart/mixed content to hold everything. * We can remove this later, if it is not needed. */ @@ -295,8 +377,7 @@ 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; @@ -309,14 +390,13 @@ finish_field: CT p; if (user_content (in, buf, &p, infile) == DONE) { - admonish (NULL, "ignoring spurious #end"); + inform("ignoring spurious #end, continuing..."); continue; } 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; @@ -336,9 +416,7 @@ finish_field: 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); /* @@ -349,8 +427,7 @@ 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; @@ -360,6 +437,39 @@ 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. @@ -370,9 +480,7 @@ finish_field: 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) @@ -386,18 +494,15 @@ 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; part->mp_part = p; } @@ -439,6 +544,13 @@ 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 @@ -450,8 +562,7 @@ finish_field: } else { if (*++cp == 0) adios (NULL, "giving up trying to find a unique delimiter string"); - else - (*cp)++; + (*cp)++; } } @@ -475,7 +586,7 @@ init_decoded_content (CT ct, const char *filename) ct->c_ceclosefnx = close_encoding; ct->c_cesizefnx = NULL; /* since unencoded */ ct->c_encoding = CE_7BIT; /* Seems like a reasonable default */ - ct->c_file = add(filename, NULL); + ct->c_file = mh_xstrdup(FENDNULL(filename)); return OK; } @@ -485,23 +596,26 @@ static char * fgetstr (char *s, int n, FILE *stream) { char *cp, *ep; - int o_n = n; + ep = s + n; while(1) { - for (ep = (cp = s) + o_n; cp < ep; ) { - int i; + for (cp = s; cp < ep;) { + int len; if (!fgets (cp, n, stream)) - return (cp != s ? s : NULL); + return cp == s ? NULL : s; /* "\\\nEOF" ignored. */ - if (cp == s && *cp != '#') - return s; + if (! do_direct() || (cp == s && *cp != '#')) + return s; /* Plaintext line. */ - cp += (i = strlen (cp)) - 1; - if (i <= 1 || *cp-- != '\n' || *cp != '\\') + len = strlen(cp); + if (len <= 1) + break; /* Can't contain "\\\n". */ + cp += len - 1; /* Just before NUL. */ + if (*cp-- != '\n' || *cp != '\\') break; - *cp = '\0'; - n -= (i - 2); + *cp = '\0'; /* Erase the trailing "\\\n". */ + n -= (len - 2); } if (strcmp(s, "#on\n") == 0) { @@ -511,11 +625,9 @@ fgetstr (char *s, int n, FILE *stream) } else if (strcmp(s, "#pop\n") == 0) { directive_pop(); } else { - break; + return s; } } - - return s; } @@ -529,7 +641,7 @@ user_content (FILE *in, char *buf, CT *ctp, const char *infilename) { int extrnal, vrsn; char *cp, **ap; - char buffer[BUFSIZ]; + char buffer[NMH_BUFSIZ]; struct multipart *m; struct part **pp; struct stat st; @@ -544,8 +656,7 @@ 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 */ @@ -577,16 +688,15 @@ user_content (FILE *in, char *buf, CT *ctp, const char *infilename) } /* use a temp file to collect the plain text lines */ - ce->ce_file = add (cp, NULL); + ce->ce_file = mh_xstrdup(cp); ce->ce_unlink = 1; if (do_direct() && (buf[0] == '#' && buf[1] == '<')) { 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)); @@ -596,7 +706,7 @@ user_content (FILE *in, char *buf, CT *ctp, const char *infilename) int i; if (headers >= 0 && do_direct() && uprf (buffer, DESCR_FIELD) - && buffer[i = strlen (DESCR_FIELD)] == ':') { + && buffer[i = LEN(DESCR_FIELD)] == ':') { headers = 1; again_descr: @@ -619,7 +729,7 @@ again_descr: } if (headers >= 0 && do_direct() && uprf (buffer, DISPO_FIELD) - && buffer[i = strlen (DISPO_FIELD)] == ':') { + && buffer[i = LEN(DISPO_FIELD)] == ':') { headers = 1; again_dispo: @@ -683,7 +793,7 @@ rock_and_roll: ct->c_encoding = CE_7BIT; goto call_init; } - /* else fall... */ + /* FALLTHRU */ case CT_MULTIPART: adios (NULL, "it doesn't make sense to define an in-line %s content", ct->c_type == CT_MESSAGE ? "message" : "multipart"); @@ -776,8 +886,7 @@ 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; if (get_ctinfo (buffer, ct, 0) == NOTOK) @@ -785,8 +894,7 @@ use_forw: 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; @@ -807,7 +915,7 @@ use_forw: continue; if (!*cp) adios (NULL, "empty pipe command for #%s directive", ci->ci_type); - cp = add (cp, NULL); + cp = mh_xstrdup(cp); free (ci->ci_magic); ci->ci_magic = cp; } else { @@ -831,7 +939,7 @@ use_forw: content_error (NULL, ct, "don't know how to compose content"); done (1); } - ci->ci_magic = add (cp, NULL); + ci->ci_magic = mh_xstrdup(cp); return OK; } @@ -862,14 +970,13 @@ use_forw: if (*cp == '+' || *cp == '@') { if (folder) adios (NULL, "only one folder per #forw directive"); - else - folder = pluspath (cp); + folder = pluspath (cp); } } /* else, use the current folder */ if (!folder) - folder = add (getfolder (1), NULL); + folder = mh_xstrdup(getfolder(1)); if (!(mp = folder_read (folder, 0))) adios (NULL, "unable to read folder %s", folder); @@ -895,8 +1002,7 @@ 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; @@ -906,8 +1012,7 @@ 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) @@ -916,12 +1021,11 @@ use_forw: p->c_subtype = MESSAGE_RFC822; snprintf (buffer, sizeof(buffer), "%s/%d", mp->foldpath, msgnum); - pe->ce_file = add (buffer, NULL); + pe->ce_file = mh_xstrdup(buffer); 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; @@ -936,7 +1040,7 @@ use_forw: msgnum = mp->lowsel; snprintf (buffer, sizeof(buffer), "%s/%d", mp->foldpath, msgnum); - ce->ce_file = add (buffer, NULL); + ce->ce_file = mh_xstrdup(buffer); if (listsw && stat (ce->ce_file, &st) != NOTOK) ct->c_end = (long) st.st_size; } @@ -981,8 +1085,7 @@ 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; @@ -998,13 +1101,12 @@ 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; } - admonish (NULL, "premature end-of-file, missing #end"); + inform("premature end-of-file, missing #end, continuing..."); return OK; } @@ -1028,10 +1130,10 @@ 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); } @@ -1067,7 +1169,7 @@ compose_content (CT ct, int verbose) CT p = part->mp_part; sprintf (pp, "%d", partnum); - p->c_partno = add (partnam, NULL); + p->c_partno = mh_xstrdup(partnam); if (compose_content (p, verbose) == NOTOK) return NOTOK; } @@ -1101,7 +1203,7 @@ compose_content (CT ct, int verbose) if (listsw) { ct->c_end = (partnum = strlen (prefix) + 2) + 2; if (ct->c_rfc934) - ct->c_end += 1; + ct->c_end++; for (part = m->mp_parts; part; part = part->mp_next) ct->c_end += part->mp_part->c_end + partnum; @@ -1119,7 +1221,7 @@ compose_content (CT ct, int verbose) default: if (!ce->ce_file) { pid_t child_id; - int i, xstdout, len, buflen; + int xstdout, len, buflen; char *bp, *cp; char *vec[4], buffer[BUFSIZ]; FILE *out; @@ -1133,7 +1235,7 @@ compose_content (CT ct, int verbose) adios("mhbuildsbr", "unable to create temporary file in %s", get_temp_dir()); } - ce->ce_file = add (tfile, NULL); + ce->ce_file = mh_xstrdup(tfile); ce->ce_unlink = 1; xstdout = 0; @@ -1169,7 +1271,7 @@ compose_content (CT ct, int verbose) case 'F': /* %f, and stdout is not-redirected */ xstdout = 1; - /* and fall... */ + /* FALLTHRU */ case 'f': /* @@ -1219,8 +1321,7 @@ raw: if ((out = fopen (ce->ce_file, "w")) == NULL) adios (ce->ce_file, "unable to open for writing"); - for (i = 0; (child_id = fork()) == NOTOK && i > 5; i++) - sleep (5); + child_id = fork(); switch (child_id) { case NOTOK: adios ("fork", "unable to fork"); @@ -1320,15 +1421,17 @@ 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; checknul = 1; } + check8bit = 1; } switch (ct->c_reqencoding) { @@ -1470,24 +1573,7 @@ scan_content (CT ct, size_t maxunencoded) * 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; - char *eightbitcharset = write_charset_8bit(); - - if (contains8bit && 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", - contains8bit ? eightbitcharset : "us-ascii", 0); - - t->tx_charset = CHARSET_SPECIFIED; - } - } + set_charset (ct, contains8bit); /* * Decide which transfer encoding to use. @@ -1575,11 +1661,11 @@ build_headers (CT ct, int header_encoding) /* * output the content type and subtype */ - np = add (TYPE_FIELD, NULL); + np = mh_xstrdup(TYPE_FIELD); vp = concat (" ", ci->ci_type, "/", ci->ci_subtype, NULL); /* keep track of length of line */ - len = strlen (TYPE_FIELD) + strlen (ci->ci_type) + len = LEN(TYPE_FIELD) + strlen (ci->ci_type) + strlen (ci->ci_subtype) + 3; extbody = ct->c_type == CT_MESSAGE && ct->c_subtype == MESSAGE_EXTERNAL; @@ -1621,10 +1707,12 @@ build_headers (CT ct, int header_encoding) add_header (ct, np, vp); /* - * output the Content-ID, unless disabled by -nocontentid + * output the Content-ID, unless disabled by -nocontentid. Note that + * RFC 2045 always requires a Content-ID header for message/external-body + * entities. */ - if (contentidsw && ct->c_id) { - np = add (ID_FIELD, NULL); + if ((contentidsw || ct->c_ctexbody) && ct->c_id) { + np = mh_xstrdup(ID_FIELD); vp = concat (" ", ct->c_id, NULL); add_header (ct, np, vp); } @@ -1632,10 +1720,13 @@ build_headers (CT ct, int header_encoding) * output the Content-Description */ if (ct->c_descr) { - np = add (DESCR_FIELD, NULL); + np = mh_xstrdup(DESCR_FIELD); 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); } @@ -1644,18 +1735,17 @@ build_headers (CT ct, int header_encoding) * set, then we need to build it. */ if (ct->c_dispo) { - np = add (DISPO_FIELD, NULL); + np = mh_xstrdup(DISPO_FIELD); vp = concat (" ", ct->c_dispo, NULL); add_header (ct, np, vp); } else if (ct->c_dispo_type) { vp = concat (" ", ct->c_dispo_type, NULL); - len = strlen(DISPO_FIELD) + strlen(vp) + 1; + len = LEN(DISPO_FIELD) + strlen(vp) + 1; 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: @@ -1671,21 +1761,26 @@ skip_headers: * output the Content-MD5 */ if (checksw) { - np = add (MD5_FIELD, NULL); - vp = calculate_digest (ct, (ct->c_encoding == CE_QUOTED) ? 1 : 0); + np = mh_xstrdup(MD5_FIELD); + vp = calculate_digest (ct, ct->c_encoding == CE_QUOTED); add_header (ct, np, vp); } /* * 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: - np = add (ENCODING_FIELD, NULL); + np = mh_xstrdup(ENCODING_FIELD); vp = concat (" ", "8bit", "\n", NULL); add_header (ct, np, vp); break; @@ -1694,7 +1789,7 @@ skip_headers: if (ct->c_type == CT_MESSAGE || ct->c_type == CT_MULTIPART) adios (NULL, "internal error, invalid encoding"); - np = add (ENCODING_FIELD, NULL); + np = mh_xstrdup(ENCODING_FIELD); vp = concat (" ", "quoted-printable", "\n", NULL); add_header (ct, np, vp); break; @@ -1703,7 +1798,7 @@ skip_headers: if (ct->c_type == CT_MESSAGE || ct->c_type == CT_MULTIPART) adios (NULL, "internal error, invalid encoding"); - np = add (ENCODING_FIELD, NULL); + np = mh_xstrdup(ENCODING_FIELD); vp = concat (" ", "base64", "\n", NULL); add_header (ct, np, vp); break; @@ -1712,7 +1807,7 @@ skip_headers: if (ct->c_type == CT_MESSAGE) adios (NULL, "internal error, invalid encoding"); - np = add (ENCODING_FIELD, NULL); + np = mh_xstrdup(ENCODING_FIELD); vp = concat (" ", "binary", "\n", NULL); add_header (ct, np, vp); break; @@ -1817,7 +1912,7 @@ calculate_digest (CT ct, int asciiP) unsigned char *ep; fprintf (stderr, "MD5 digest="); - for (ep = (dp = digest) + sizeof(digest) / sizeof(digest[0]); + for (ep = (dp = digest) + sizeof digest; dp < ep; dp++) fprintf (stderr, "%02x", *dp & 0xff); fprintf (stderr, "\n"); @@ -1825,7 +1920,7 @@ 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; cc > 0; cc -= 3, op += 4) { unsigned long bits; char *bp; @@ -1865,7 +1960,6 @@ setup_attach_content(CT ct, char *filename) char *type, *simplename = r1bindex(filename, '/'); struct str2init *s2i; PM pm; - char *cp; if (! (type = mime_type(filename))) { adios(NULL, "Unable to determine MIME type of \"%s\"", filename); @@ -1873,7 +1967,7 @@ setup_attach_content(CT ct, char *filename) /* * Parse the Content-Type. get_ctinfo() parses MIME parameters, but - * since we're just feeding it a MIME type we have to add those ourself. + * since we're just feeding it a MIME type we have to add those ourselves. * Map that to a valid content-type label and call any initialization * function. */ @@ -1905,7 +1999,7 @@ setup_attach_content(CT ct, char *filename) if (strcasecmp(ct->c_ctinfo.ci_subtype, "external-body") == 0) adios(NULL, "external-body messages must be specified " "by mhbuild directives"); - /* Fall through */ + /* FALLTHRU */ default: /* @@ -1923,9 +2017,8 @@ 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; } } @@ -1934,36 +2027,419 @@ setup_attach_content(CT ct, char *filename) 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); - /* - * Look for mhbuild-disposition-/ entry - * that specifies Content-Disposition type. Only - * 'attachment' and 'inline' are allowed. Default to - * 'attachment'. - */ + set_disposition (ct); + + add_param(&ct->c_dispo_first, &ct->c_dispo_last, "filename", simplename, 0); +} + +/* + * 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")) { + inform("configuration problem: %s-disposition-%s%s%s specifies " + "'%s' but only 'attachment' and 'inline' are allowed, " + "continuing...", invo_name, + ct->c_ctinfo.ci_type, + ct->c_ctinfo.ci_subtype ? "/" : "", + FENDNULL(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 + * selections: + * 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"; - cp = context_find_by_type ("disposition", ct->c_ctinfo.ci_type, - ct->c_ctinfo.ci_subtype); - if (cp != NULL) { - if (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 (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 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 { + free (type_p); + type_p = concat ("mhbuild-convert-", type, NULL); + convert = context_find (type_p); + } + free (type_p); + + 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, " ", FENDNULL(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) { + inform("store of %s content failed, continuing...", 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) { + inform("failed to extract headers from convert output in %s, " + "continuing...", reply_file); + free(reply_file); + return; + } + + /* 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 (cp) { - ct->c_dispo_type = getcpy(cp); + 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 { - ct->c_dispo_type = getcpy("attachment"); + set_charset (reply_ct, -1); + charset = get_param (reply_ct->c_ctinfo.ci_first_pm, "charset", '?', 1); } - add_param(&ct->c_dispo_first, &ct->c_dispo_last, "filename", simplename, 0); + /* 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) { + /* 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, LEN(TYPE_FIELD)) == 0 && + buffer[LEN(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, written; + + /* Truncate buffer to just the C-T. */ + *end_of_header = '\0'; + n = strlen (buffer); + + if (get_ctinfo (buffer + 14, ct, 0) != OK) { + inform("unable to get content info for reply, continuing..."); + 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)); + tmp_f = fopen(tmp_file, "w"); + if (!tmp_f) + goto failed_to_extract_ct; + written = fwrite(buffer, 1, n, tmp_f); + fclose(tmp_f); + if (written != n) + 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; }