X-Git-Url: https://diplodocus.org/git/nmh/blobdiff_plain/6779979cf3a57b458d28e4d1aaac026a61d73572..f1920d78123667716f2321d37ce37628603b2700:/uip/mhfixmsg.c diff --git a/uip/mhfixmsg.c b/uip/mhfixmsg.c index c93a71fd..65abd513 100644 --- a/uip/mhfixmsg.c +++ b/uip/mhfixmsg.c @@ -16,6 +16,9 @@ #define MHFIXMSG_SWITCHES \ X("decodetext 8bit|7bit", 0, DECODETEXTSW) \ X("nodecodetext", 0, NDECODETEXTSW) \ + X("decodetypes", 0, DECODETYPESW) \ + X("crlflinebreaks", 0, CRLFLINEBREAKSSW) \ + X("nocrlflinebreaks", 0, NCRLFLINEBREAKSSW) \ X("textcharset", 0, TEXTCHARSETSW) \ X("notextcharset", 0, NTEXTCHARSETSW) \ X("reformat", 0, REFORMATSW) \ @@ -26,10 +29,13 @@ X("nofixboundary", 0, NFIXBOUNDARYSW) \ X("fixcte", 0, FIXCTESW) \ X("nofixcte", 0, NFIXCTESW) \ + X("fixtype mimetype", 0, FIXTYPESW) \ X("file file", 0, FILESW) \ X("outfile file", 0, OUTFILESW) \ X("rmmproc program", 0, RPROCSW) \ X("normmproc", 0, NRPRCSW) \ + X("changecur", 0, CHGSW) \ + X("nochangecur", 0, NCHGSW) \ X("verbose", 0, VERBSW) \ X("noverbose", 0, NVERBSW) \ X("version", 0, VERSIONSW) \ @@ -53,7 +59,9 @@ int debugsw; /* Needed by mhparse.c. */ extern int skip_mp_cte_check; /* flag to InitMultiPart */ extern int suppress_bogus_mp_content_warning; /* flag to InitMultiPart */ extern int bogus_mp_content; /* flag from InitMultiPart */ -void reverse_parts (CT); +/* flags to/from parse_header_attrs */ +extern int suppress_extraneous_trailing_semicolon_warning; +extern int extraneous_trailing_semicolon; /* mhoutsbr.c */ int output_message (CT, char *); @@ -71,21 +79,30 @@ void freects_done (int) NORETURN; typedef struct fix_transformations { int fixboundary; int fixcte; + svector_t fixtypes; int reformat; int replacetextplain; int decodetext; + char *decodetypes; + /* Whether to use CRLF linebreaks, per RFC 2046 Sec. 4.1.1, par.1. */ + int lf_line_endings; char *textcharset; } fix_transformations; int mhfixmsgsbr (CT *, const fix_transformations *, char *); -static void reverse_alternative_parts (CT); static int fix_boundary (CT *, int *); static int get_multipart_boundary (CT, char **); static int replace_boundary (CT, char *, char *); +static int fix_types (CT, svector_t, int *); +static char *replace_substring (char **, const char *, const char *); +static char *remove_parameter (char *, const char *); static int fix_multipart_cte (CT, int *); static int set_ce (CT, int); static int ensure_text_plain (CT *, CT, int *, int); +static int find_textplain_sibling (CT, int, int *); +static int insert_new_text_plain_part (CT, int, CT); static CT build_text_plain_part (CT); +static int insert_into_new_mp_alt (CT *, int *); static CT divide_part (CT); static void copy_ctinfo (CI, CI); static int decode_part (CT); @@ -95,11 +112,14 @@ static CT build_multipart_alt (CT, CT, int, int); static int boundary_in_content (FILE **, char *, const char *); static void transfer_noncontent_headers (CT, CT); static int set_ct_type (CT, int type, int subtype, int encoding); -static int decode_text_parts (CT, int, int *); +static int decode_text_parts (CT, int, const char *, int *); +static int should_decode(const char *, const char *, const char *); static int content_encoding (CT, const char **); static int strip_crs (CT, int *); static int convert_charsets (CT, char *, int *); +static int fix_always (CT, int *); static int write_content (CT, char *, char *, int, int); +static void set_text_ctparams(CT, char *, int); static int remove_file (char *); static void report (char *, char *, char *, char *, ...); static void pipeser (int); @@ -116,11 +136,15 @@ main (int argc, char **argv) { CT *ctp; FILE *fp; int using_stdin = 0; + int chgflag = 1; int status = OK; fix_transformations fx; fx.reformat = fx.fixcte = fx.fixboundary = 1; + fx.fixtypes = NULL; fx.replacetextplain = 0; fx.decodetext = CE_8BIT; + fx.decodetypes = "text,application/ics"; /* Default, per man page. */ + fx.lf_line_endings = 0; fx.textcharset = NULL; if (nmh_init(argv[0], 1)) { return 1; } @@ -165,6 +189,17 @@ main (int argc, char **argv) { case NDECODETEXTSW: fx.decodetext = 0; continue; + case DECODETYPESW: + if (! (cp = *argp++) || *cp == '-') + adios (NULL, "missing argument to %s", argp[-2]); + fx.decodetypes = cp; + continue; + case CRLFLINEBREAKSSW: + fx.lf_line_endings = 0; + continue; + case NCRLFLINEBREAKSSW: + fx.lf_line_endings = 1; + continue; case TEXTCHARSETSW: if (! (cp = *argp++) || (*cp == '-' && cp[1])) adios (NULL, "missing argument to %s", argp[-2]); @@ -185,6 +220,18 @@ main (int argc, char **argv) { case NFIXCTESW: fx.fixcte = 0; continue; + case FIXTYPESW: + if (! (cp = *argp++) || (*cp == '-' && cp[1])) + adios (NULL, "missing argument to %s", argp[-2]); + if (! strncasecmp (cp, "multipart/", 10) || + ! strncasecmp (cp, "message/", 8)) { + adios (NULL, "-fixtype %s not allowed", cp); + } else if (! strchr (cp, '/')) { + adios (NULL, "-fixtype requires type/subtype"); + } + if (fx.fixtypes == NULL) { fx.fixtypes = svector_create (10); } + svector_push_back (fx.fixtypes, cp); + continue; case REFORMATSW: fx.reformat = 1; continue; @@ -214,6 +261,12 @@ main (int argc, char **argv) { case NRPRCSW: rmmproc = NULL; continue; + case CHGSW: + chgflag = 1; + continue; + case NCHGSW: + chgflag = 0; + continue; case VERBSW: verbosw = 1; continue; @@ -249,6 +302,7 @@ main (int argc, char **argv) { } suppress_bogus_mp_content_warning = skip_mp_cte_check = 1; + suppress_extraneous_trailing_semicolon_warning = 1; if (! context_find ("path")) free (path ("./", TFOLDER)); @@ -287,11 +341,15 @@ main (int argc, char **argv) { } } - if (! (cts = (CT *) calloc ((size_t) 2, sizeof *cts))) + if (! (cts = (CT *) mh_xcalloc ((size_t) 2, sizeof *cts))) { adios (NULL, "out of memory"); + } ctp = cts; - if ((ct = parse_mime (file))) *ctp++ = ct; + if ((ct = parse_mime (file))) { + set_text_ctparams(ct, fx.decodetypes, fx.lf_line_endings); + *ctp++ = ct; + } } else { /* * message(s) are coming from a folder @@ -321,8 +379,10 @@ main (int argc, char **argv) { done (1); seq_setprev (mp); /* set the previous-sequence */ - if (! (cts = (CT *) calloc ((size_t) (mp->numsel + 1), sizeof *cts))) + if (! (cts = + (CT *) mh_xcalloc ((size_t) (mp->numsel + 1), sizeof *cts))) { adios (NULL, "out of memory"); + } ctp = cts; for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) { @@ -330,11 +390,16 @@ main (int argc, char **argv) { char *msgnam; msgnam = m_name (msgnum); - if ((ct = parse_mime (msgnam))) *ctp++ = ct; + if ((ct = parse_mime (msgnam))) { + set_text_ctparams(ct, fx.decodetypes, fx.lf_line_endings); + *ctp++ = ct; + } } } - seq_setcur (mp, mp->hghsel); /* update current message */ + if (chgflag) { + seq_setcur (mp, mp->hghsel); /* update current message */ + } seq_save (mp); /* synchronize sequences */ context_replace (pfolder, folder);/* update current folder */ context_save (); /* save the context file */ @@ -357,6 +422,7 @@ main (int argc, char **argv) { status = 1; } + if (fx.fixtypes != NULL) { svector_free (fx.fixtypes); } free (outfile); free (file); @@ -391,9 +457,13 @@ mhfixmsgsbr (CT *ctp, const fix_transformations *fx, char *outfile) { } reverse_alternative_parts (*ctp); + status = fix_always (*ctp, &message_mods); if (status == OK && fx->fixboundary) { status = fix_boundary (ctp, &message_mods); } + if (status == OK && fx->fixtypes != NULL) { + status = fix_types (*ctp, fx->fixtypes, &message_mods); + } if (status == OK && fx->fixcte) { status = fix_multipart_cte (*ctp, &message_mods); } @@ -402,13 +472,13 @@ mhfixmsgsbr (CT *ctp, const fix_transformations *fx, char *outfile) { ensure_text_plain (ctp, NULL, &message_mods, fx->replacetextplain); } if (status == OK && fx->decodetext) { - status = decode_text_parts (*ctp, fx->decodetext, &message_mods); + status = decode_text_parts (*ctp, fx->decodetext, fx->decodetypes, &message_mods); } if (status == OK && fx->textcharset != NULL) { status = convert_charsets (*ctp, fx->textcharset, &message_mods); } - if (! (*ctp)->c_umask) { + if (status == OK && ! (*ctp)->c_umask) { /* Set the umask for the contents file. This currently isn't used but just in case it is in the future. */ struct stat st; @@ -446,7 +516,7 @@ mhfixmsgsbr (CT *ctp, const fix_transformations *fx, char *outfile) { } if (modify_inplace) { - if (status != OK) (void) m_unlink (outfile); + if (status != OK) { (void) m_unlink (outfile); } free (outfile); outfile = NULL; } @@ -457,33 +527,12 @@ mhfixmsgsbr (CT *ctp, const fix_transformations *fx, char *outfile) { } -/* parse_mime() arranges alternates in reverse (priority) order, so - reverse them back. This will put a text/plain part at the front of - a multipart/alternative part, for example, where it belongs. */ -static 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); - } - } -} - - static int fix_boundary (CT *ct, int *message_mods) { struct multipart *mp; int status = OK; - if (bogus_mp_content) { + if (ct && (*ct)->c_type == CT_MULTIPART && bogus_mp_content) { mp = (struct multipart *) (*ct)->c_ctparams; /* @@ -505,9 +554,11 @@ fix_boundary (CT *ct, int *message_mods) { if ((fixed = m_mktemp2 (NULL, invo_name, NULL, &(*ct)->c_fp))) { if (replace_boundary (*ct, fixed, part_boundary) == OK) { char *filename = add ((*ct)->c_file, NULL); + CT fixed_ct; free_content (*ct); - if ((*ct = parse_mime (fixed))) { + if ((fixed_ct = parse_mime (fixed))) { + *ct = fixed_ct; (*ct)->c_unlink = 1; ++*message_mods; @@ -515,6 +566,10 @@ fix_boundary (CT *ct, int *message_mods) { report (NULL, NULL, filename, "fix multipart boundary"); } + } else { + *ct = NULL; + advise (NULL, "unable to parse fixed part"); + status = NOTOK; } free (filename); } else { @@ -528,7 +583,16 @@ fix_boundary (CT *ct, int *message_mods) { } free (part_boundary); + } else { + /* Couldn't fix the boundary. Report failure so that mhfixmsg + doesn't modify the message. */ + status = NOTOK; } + } else { + /* No multipart struct, even though the content type is + CT_MULTIPART. Report failure so that mhfixmsg doesn't modify + the message. */ + status = NOTOK; } } @@ -691,7 +755,9 @@ replace_boundary (CT ct, char *file, char *boundary) { case BODY: fputs ("\n", fpout); /* buf will have a terminating NULL, skip it. */ - fwrite (buf, 1, bufsz-1, fpout); + if ((int) fwrite (buf, 1, bufsz-1, fpout) < bufsz-1) { + advise (file, "fwrite"); + } continue; case FILEEOF: @@ -720,6 +786,189 @@ replace_boundary (CT ct, char *file, char *boundary) { } +static int +fix_types (CT ct, svector_t fixtypes, int *message_mods) { + int status = OK; + + switch (ct->c_type) { + case CT_MULTIPART: { + struct multipart *m = (struct multipart *) ct->c_ctparams; + struct part *part; + + for (part = m->mp_parts; status == OK && part; part = part->mp_next) { + status = fix_types (part->mp_part, fixtypes, message_mods); + } + break; + } + + case CT_MESSAGE: + if (ct->c_subtype == MESSAGE_EXTERNAL) { + struct exbody *e = (struct exbody *) ct->c_ctparams; + + status = fix_types (e->eb_content, fixtypes, message_mods); + } + break; + + default: { + char **typep, *type; + + if (ct->c_ctinfo.ci_type && ct->c_ctinfo.ci_subtype) { + for (typep = svector_strs (fixtypes); + typep && (type = *typep); + ++typep) { + char *type_subtype = + concat (ct->c_ctinfo.ci_type, "/", ct->c_ctinfo.ci_subtype, + NULL); + + if (! strcasecmp (type, type_subtype) && + decode_part (ct) == OK && + ct->c_cefile.ce_file != NULL) { + char *ct_type_subtype = mime_type (ct->c_cefile.ce_file); + char *cp; + + if ((cp = strchr (ct_type_subtype, ';'))) { + /* Truncate to remove any parameter list from + mime_type () result. */ + *cp = '\0'; + } + + if (strcasecmp (type, ct_type_subtype)) { + char *ct_type, *ct_subtype; + HF hf; + + /* The Content-Type header does not match the + content, so update these struct Content + fields to match: + * c_type, c_subtype + * c_ctinfo.ci_type, c_ctinfo.ci_subtype + * c_ctline + */ + /* Extract type and subtype from type/subtype. */ + ct_type = getcpy (ct_type_subtype); + if ((cp = strchr (ct_type, '/'))) { + *cp = '\0'; + ct_subtype = getcpy (++cp); + } else { + advise (NULL, "missing / in MIME type of %s %s", + ct->c_file, ct->c_partno); + free (ct_type); + return NOTOK; + } + + ct->c_type = ct_str_type (ct_type); + ct->c_subtype = ct_str_subtype (ct->c_type, ct_subtype); + + free (ct->c_ctinfo.ci_type); + ct->c_ctinfo.ci_type = ct_type; + free (ct->c_ctinfo.ci_subtype); + ct->c_ctinfo.ci_subtype = ct_subtype; + if (! replace_substring (&ct->c_ctline, type, + ct_type_subtype)) { + advise (NULL, "did not find %s in %s", + type, ct->c_ctline); + } + + /* Update Content-Type header field. */ + for (hf = ct->c_first_hf; hf; hf = hf->next) { + if (! strcasecmp (TYPE_FIELD, hf->name)) { + if (replace_substring (&hf->value, type, + ct_type_subtype)) { + ++*message_mods; + if (verbosw) { + report (NULL, ct->c_partno, ct->c_file, + "change Content-Type in header " + "from %s to %s", + type, ct_type_subtype); + } + break; + } else { + advise (NULL, "did not find %s in %s", + type, hf->value); + } + } + } + } + free (ct_type_subtype); + } + free (type_subtype); + } + } + }} + + return status; +} + +char * +replace_substring (char **str, const char *old, const char *new) { + char *cp; + + if ((cp = strstr (*str, old))) { + char *remainder = cp + strlen (old); + char *prefix, *new_str; + + if (cp - *str) { + prefix = getcpy (*str); + *(prefix + (cp - *str)) = '\0'; + new_str = concat (prefix, new, remainder, NULL); + free (prefix); + } else { + new_str = concat (new, remainder, NULL); + } + + free (*str); + + return *str = new_str; + } else { + return NULL; + } +} + +/* + * Remove a name=value parameter, given just its name, from a header value. + */ +char * +remove_parameter (char *str, const char *name) { + /* It looks to me, based on the BNF in RFC 2045, than there can't + be whitespace betwwen the parameter name and the "=", or + between the "=" and the parameter value. */ + char *param_name = concat (name, "=", NULL); + char *cp; + + if ((cp = strstr (str, param_name))) { + char *start, *end; + size_t count = 1; + + /* Remove any leading spaces, before the parameter name. */ + for (start = cp; + start > str && isspace ((unsigned char) *(start-1)); + --start) { + continue; + } + /* Remove a leading semicolon. */ + if (start > str && *(start-1) == ';') { --start; } + + end = cp + strlen (name) + 1; + if (*end == '"') { + /* Skip past the quoted value, and then the final quote. */ + for (++end ; *end && *end != '"'; ++end) { continue; } + ++end; + } else { + /* Skip past the value. */ + for (++end ; *end && ! isspace ((unsigned char) *end); ++end) {} + } + + /* Count how many characters need to be moved. Include + trailing null, which is accounted for by the + initialization of count to 1. */ + for (cp = end; *cp; ++cp) { ++count; } + (void) memmove (start, end, count); + } + + free (param_name); + + return str; +} + static int fix_multipart_cte (CT ct, int *message_mods) { int status = OK; @@ -843,102 +1092,126 @@ ensure_text_plain (CT *ct, CT parent, int *message_mods, int replacetextplain) { switch ((*ct)->c_type) { case CT_TEXT: { - int has_text_plain = 0; - /* Nothing to do for text/plain. */ - if ((*ct)->c_subtype == TEXT_PLAIN) return OK; + if ((*ct)->c_subtype == TEXT_PLAIN) { return OK; } if (parent && parent->c_type == CT_MULTIPART && parent->c_subtype == MULTI_ALTERNATE) { - struct multipart *mp = (struct multipart *) parent->c_ctparams; - struct part *part, *prev; int new_subpart_number = 1; - - /* See if there is a sibling text/plain. */ - for (prev = part = mp->mp_parts; part; part = part->mp_next) { - ++new_subpart_number; - if (part->mp_part->c_type == CT_TEXT && - part->mp_part->c_subtype == TEXT_PLAIN) { - if (replacetextplain) { - struct part *old_part; - if (part == mp->mp_parts) { - old_part = mp->mp_parts; - mp->mp_parts = part->mp_next; - } else { - old_part = prev->mp_next; - prev->mp_next = part->mp_next; - } - if (verbosw) { - report (NULL, parent->c_partno, parent->c_file, - "remove text/plain part %s", - old_part->mp_part->c_partno); - } - free_content (old_part->mp_part); - free (old_part); - } else { - has_text_plain = 1; - } - break; - } - prev = part; - } + int has_text_plain = + find_textplain_sibling (parent, replacetextplain, + &new_subpart_number); if (! has_text_plain) { /* Parent is a multipart/alternative. Insert a new text/plain subpart. */ - struct part *new_part = mh_xmalloc (sizeof *new_part); - - if ((new_part->mp_part = build_text_plain_part (*ct))) { - char buffer[16]; - snprintf (buffer, sizeof buffer, "%d", new_subpart_number); - - new_part->mp_next = mp->mp_parts; - mp->mp_parts = new_part; - new_part->mp_part->c_partno = - concat (parent->c_partno ? parent->c_partno : "1", ".", - buffer, NULL); - + const int inserted = + insert_new_text_plain_part (*ct, new_subpart_number, + parent); + if (inserted) { ++*message_mods; if (verbosw) { report (NULL, parent->c_partno, parent->c_file, "insert text/plain part"); } } else { - free_content (new_part->mp_part); - free (new_part); status = NOTOK; } } - } else { - /* Slip new text/plain part into a new multipart/alternative. */ - CT tp_part = build_text_plain_part (*ct); + } else if (parent && parent->c_type == CT_MULTIPART && + parent->c_subtype == MULTI_RELATED) { + char *type_subtype = + concat ((*ct)->c_ctinfo.ci_type, "/", + (*ct)->c_ctinfo.ci_subtype, NULL); + const char *parent_type = + get_param (parent->c_ctinfo.ci_first_pm, "type", '?', 1); + int new_subpart_number = 1; + int has_text_plain = 0; + + /* Have to do string comparison on the subtype because we + don't enumerate all of them in c_subtype values. + parent_type will be NULL if the multipart/related part + doesn't have a type parameter. The type parameter must + be specified according to RFC 2387 Sec. 3.1 but not all + messages comply. */ + if (parent_type && strcasecmp (type_subtype, parent_type) == 0) { + /* The type of this part matches the root type of the + parent multipart/related. Look to see if there's + text/plain sibling. */ + has_text_plain = + find_textplain_sibling (parent, replacetextplain, + &new_subpart_number); + } - if (tp_part) { - CT mp_alt = build_multipart_alt (*ct, tp_part, CT_MULTIPART, - MULTI_ALTERNATE); - if (mp_alt) { - struct multipart *mp = - (struct multipart *) mp_alt->c_ctparams; + free (type_subtype); - if (mp && mp->mp_parts) { - mp->mp_parts->mp_part = tp_part; - /* Make the new multipart/alternative the parent. */ - *ct = mp_alt; + if (! has_text_plain) { + struct multipart *mp = (struct multipart *) parent->c_ctparams; + struct part *part; + int siblings = 0; - ++*message_mods; - if (verbosw) { - report (NULL, (*ct)->c_partno, (*ct)->c_file, - "insert text/plain part"); + for (part = mp->mp_parts; part; part = part->mp_next) { + if (*ct != part->mp_part) { + ++siblings; + } + } + + if (siblings) { + /* Parent is a multipart/related. Insert a new + text/plain subpart in a new multipart/alternative. */ + if (insert_into_new_mp_alt (ct, message_mods)) { + /* Not an error if text/plain couldn't be added. */ + } + } else { + /* There are no siblings, so insert a new text/plain + subpart, and change the parent type from + multipart/related to multipart/alternative. */ + const int inserted = + insert_new_text_plain_part (*ct, new_subpart_number, + parent); + + if (inserted) { + HF hf; + + parent->c_subtype = MULTI_ALTERNATE; + parent->c_ctinfo.ci_subtype = getcpy ("alternative"); + if (! replace_substring (&parent->c_ctline, "/related", + "/alternative")) { + advise (NULL, + "did not find multipart/related in %s", + parent->c_ctline); + } + + /* Update Content-Type header field. */ + for (hf = parent->c_first_hf; hf; hf = hf->next) { + if (! strcasecmp (TYPE_FIELD, hf->name)) { + if (replace_substring (&hf->value, "/related", + "/alternative")) { + ++*message_mods; + if (verbosw) { + report (NULL, parent->c_partno, + parent->c_file, + "insert text/plain part"); + } + + /* Remove, e.g., type="text/html" from + multipart/alternative. */ + remove_parameter (hf->value, "type"); + break; + } else { + advise (NULL, "did not find multipart/" + "related in header %s", + hf->value); + } + } } } else { - free_content (tp_part); - free_content (mp_alt); - status = NOTOK; + /* Not an error if text/plain couldn't be inserted. */ } - } else { - status = NOTOK; } - } else { + } + } else { + if (insert_into_new_mp_alt (ct, message_mods)) { status = NOTOK; } } @@ -960,9 +1233,8 @@ ensure_text_plain (CT *ct, CT parent, int *message_mods, int replacetextplain) { case CT_MESSAGE: if ((*ct)->c_subtype == MESSAGE_EXTERNAL) { - struct exbody *e; + struct exbody *e = (struct exbody *) (*ct)->c_ctparams; - e = (struct exbody *) (*ct)->c_ctparams; status = ensure_text_plain (&e->eb_content, *ct, message_mods, replacetextplain); } @@ -973,6 +1245,71 @@ ensure_text_plain (CT *ct, CT parent, int *message_mods, int replacetextplain) { } +/* See if there is a sibling text/plain. */ +static int +find_textplain_sibling (CT parent, int replacetextplain, + int *new_subpart_number) { + struct multipart *mp = (struct multipart *) parent->c_ctparams; + struct part *part, *prev; + int has_text_plain = 0; + + for (prev = part = mp->mp_parts; part; part = part->mp_next) { + ++*new_subpart_number; + if (part->mp_part->c_type == CT_TEXT && + part->mp_part->c_subtype == TEXT_PLAIN) { + if (replacetextplain) { + struct part *old_part; + if (part == mp->mp_parts) { + old_part = mp->mp_parts; + mp->mp_parts = part->mp_next; + } else { + old_part = prev->mp_next; + prev->mp_next = part->mp_next; + } + if (verbosw) { + report (NULL, parent->c_partno, parent->c_file, + "remove text/plain part %s", + old_part->mp_part->c_partno); + } + free_content (old_part->mp_part); + free (old_part); + } else { + has_text_plain = 1; + } + break; + } + prev = part; + } + + return has_text_plain; +} + + +static int +insert_new_text_plain_part (CT ct, int new_subpart_number, CT parent) { + struct multipart *mp = (struct multipart *) parent->c_ctparams; + struct part *new_part = mh_xmalloc (sizeof *new_part); + + if ((new_part->mp_part = build_text_plain_part (ct))) { + char buffer[16]; + snprintf (buffer, sizeof buffer, "%d", new_subpart_number); + + new_part->mp_next = mp->mp_parts; + mp->mp_parts = new_part; + new_part->mp_part->c_partno = + concat (parent->c_partno ? parent->c_partno : "1", ".", + buffer, NULL); + + return 1; + } else { + free_content (new_part->mp_part); + free (new_part); + + return 0; + } +} + + static CT build_text_plain_part (CT encoded_part) { CT tp_part = divide_part (encoded_part); @@ -988,29 +1325,67 @@ build_text_plain_part (CT encoded_part) { if ((tempfile = m_mktemp2 (NULL, invo_name, NULL, NULL)) == NULL) { advise (NULL, "unable to create temporary file in %s", get_temp_dir()); - } - tmp_plain_file = add (tempfile, NULL); - if (reformat_part (tp_part, tmp_plain_file, - tp_part->c_ctinfo.ci_type, - tp_part->c_ctinfo.ci_subtype, - tp_part->c_type) == OK) { - return tp_part; + } else { + tmp_plain_file = add (tempfile, NULL); + if (reformat_part (tp_part, tmp_plain_file, + tp_part->c_ctinfo.ci_type, + tp_part->c_ctinfo.ci_subtype, + tp_part->c_type) == OK) { + return tp_part; + } } } free_content (tp_part); - (void) m_unlink (tmp_plain_file); + if (tmp_plain_file) { (void) m_unlink (tmp_plain_file); } free (tmp_plain_file); return NULL; } +/* Slip new text/plain part into a new multipart/alternative. */ +static int +insert_into_new_mp_alt (CT *ct, int *message_mods) { + CT tp_part = build_text_plain_part (*ct); + int status = OK; + + if (tp_part) { + CT mp_alt = build_multipart_alt (*ct, tp_part, CT_MULTIPART, + MULTI_ALTERNATE); + if (mp_alt) { + struct multipart *mp = (struct multipart *) mp_alt->c_ctparams; + + if (mp && mp->mp_parts) { + mp->mp_parts->mp_part = tp_part; + /* Make the new multipart/alternative the parent. */ + *ct = mp_alt; + + ++*message_mods; + if (verbosw) { + report (NULL, (*ct)->c_partno, (*ct)->c_file, + "insert text/plain part"); + } + } else { + free_content (tp_part); + free_content (mp_alt); + status = NOTOK; + } + } else { + status = NOTOK; + } + } else { + /* Not an error if text/plain couldn't be built. */ + } + + return status; +} + static CT divide_part (CT ct) { CT new_part; - if ((new_part = (CT) calloc (1, sizeof *new_part)) == NULL) + if ((new_part = (CT) mh_xcalloc (1, sizeof *new_part)) == NULL) adios (NULL, "out of memory"); /* Just copy over what is needed for decoding. c_vrsn and @@ -1091,37 +1466,22 @@ reformat_part (CT ct, char *file, char *type, char *subtype, int c_type) { Could show_multi() in mhshowsbr.c avoid this? */ /* Check for invo_name-format-type/subtype. */ - cp = concat (invo_name, "-format-", type, "/", subtype, NULL); - if ((cf = context_find (cp)) && *cf != '\0') { + if ((cf = context_find_by_type ("format", type, subtype)) == NULL) { + if (verbosw) { + advise (NULL, "Don't know how to convert %s, there is no " + "%s-format-%s/%s profile entry", + ct->c_file, invo_name, type, subtype); + } + return NOTOK; + } else { if (strchr (cf, '>')) { - free (cp); advise (NULL, "'>' prohibited in \"%s\",\nplease fix your " "%s-format-%s/%s profile entry", cf, invo_name, type, - subtype); - return NOTOK; - } - } else { - free (cp); - - /* Check for invo_name-format-type. */ - cp = concat (invo_name, "-format-", type, NULL); - if (! (cf = context_find (cp)) || *cf == '\0') { - free (cp); - if (verbosw) { - advise (NULL, "Don't know how to convert %s, there is no " - "%s-format-%s/%s profile entry", - ct->c_file, invo_name, type, subtype); - } - return NOTOK; - } + subtype ? subtype : ""); - if (strchr (cf, '>')) { - free (cp); - advise (NULL, "'>' prohibited in \"%s\"", cf); return NOTOK; } } - free (cp); cp = concat (cf, " >", file, NULL); status = show_content_aux (ct, 0, cp, NULL, NULL); @@ -1159,8 +1519,10 @@ reformat_part (CT ct, char *file, char *type, char *subtype, int c_type) { /* Identifies 7bit or 8bit content based on charset. */ static int charset_encoding (CT ct) { - int encoding = - strcasecmp (content_charset (ct), "US-ASCII") ? CE_8BIT : CE_7BIT; + char *ct_charset = content_charset (ct); + int encoding = strcasecmp (ct_charset, "US-ASCII") ? CE_8BIT : CE_7BIT; + + free (ct_charset); return encoding; } @@ -1177,11 +1539,11 @@ build_multipart_alt (CT first_alt, CT new_part, int type, int subtype) { struct multipart *m; const struct str2init *ctinit; - if ((ct = (CT) calloc (1, sizeof *ct)) == NULL) + if ((ct = (CT) mh_xcalloc (1, sizeof *ct)) == NULL) adios (NULL, "out of memory"); /* Set up the multipart/alternative part. These fields of *ct were - initialized to 0 by calloc(): + initialized to 0 by mh_xcalloc(): c_fp, c_unlink, c_begin, c_end, c_vrsn, c_ctline, c_celine, c_id, c_descr, c_dispo, c_partno, @@ -1278,20 +1640,20 @@ build_multipart_alt (CT first_alt, CT new_part, int type, int subtype) { } add_param(&ct->c_ctinfo.ci_first_pm, &ct->c_ctinfo.ci_last_pm, - "boundary", boundary, 0); + "boundary", boundary, 0); p = (struct part *) mh_xmalloc (sizeof *p); p->mp_next = (struct part *) mh_xmalloc (sizeof *p->mp_next); p->mp_next->mp_next = NULL; p->mp_next->mp_part = first_alt; - if ((m = (struct multipart *) calloc (1, sizeof (struct multipart))) == + if ((m = (struct multipart *) mh_xcalloc (1, sizeof (struct multipart))) == NULL) adios (NULL, "out of memory"); m->mp_start = concat (boundary, "\n", NULL); m->mp_stop = concat (boundary, "--\n", NULL); m->mp_parts = p; - ct->c_ctparams = (void *) m; + ct->c_ctparams = m; free (boundary); @@ -1421,11 +1783,39 @@ set_ct_type (CT ct, int type, int subtype, int encoding) { static int -decode_text_parts (CT ct, int encoding, int *message_mods) { +decode_text_parts (CT ct, int encoding, const char *decodetypes, int *message_mods) { int status = OK; + int lf_line_endings = 0; switch (ct->c_type) { - case CT_TEXT: + case CT_MULTIPART: { + struct multipart *m = (struct multipart *) ct->c_ctparams; + struct part *part; + + /* Should check to see if the body for this part is encoded? + For now, it gets passed along as-is by InitMultiPart(). */ + for (part = m->mp_parts; status == OK && part; part = part->mp_next) { + status = decode_text_parts (part->mp_part, encoding, decodetypes, message_mods); + } + break; + } + + case CT_MESSAGE: + if (ct->c_subtype == MESSAGE_EXTERNAL) { + struct exbody *e = (struct exbody *) ct->c_ctparams; + + status = decode_text_parts (e->eb_content, encoding, decodetypes, message_mods); + } + break; + + default: + if (! should_decode(decodetypes, ct->c_ctinfo.ci_type, ct->c_ctinfo.ci_subtype)) { + break; + } + + lf_line_endings = + ct->c_ctparams && ((struct text *) ct->c_ctparams)->lf_line_endings; + switch (ct->c_encoding) { case CE_BASE64: case CE_QUOTED: { @@ -1477,7 +1867,9 @@ decode_text_parts (CT ct, int encoding, int *message_mods) { report (NULL, ct->c_partno, ct->c_file, "decode%s", ct->c_ctline ? ct->c_ctline : ""); } - strip_crs (ct, message_mods); + if (lf_line_endings) { + strip_crs (ct, message_mods); + } } else { status = NOTOK; } @@ -1489,40 +1881,49 @@ decode_text_parts (CT ct, int encoding, int *message_mods) { } case CE_8BIT: case CE_7BIT: - strip_crs (ct, message_mods); + if (lf_line_endings) { + strip_crs (ct, message_mods); + } break; default: break; } break; - - case CT_MULTIPART: { - struct multipart *m = (struct multipart *) ct->c_ctparams; - struct part *part; - - /* Should check to see if the body for this part is encoded? - For now, it gets passed along as-is by InitMultiPart(). */ - for (part = m->mp_parts; status == OK && part; part = part->mp_next) { - status = decode_text_parts (part->mp_part, encoding, message_mods); - } - break; } - case CT_MESSAGE: - if (ct->c_subtype == MESSAGE_EXTERNAL) { - struct exbody *e; + return status; +} - e = (struct exbody *) ct->c_ctparams; - status = decode_text_parts (e->eb_content, encoding, message_mods); - } - break; - default: - break; +/* Determine if the part with type[/subtype] should be decoded, according to + decodetypes (which came from the -decodetypes switch). */ +static int +should_decode(const char *decodetypes, const char *type, const char *subtype) { + /* Quick search for matching type[/subtype] in decodetypes: bracket + decodetypes with commas, then search for ,type, and ,type/subtype, in + it. */ + + int found_match = 0; + char *delimited_decodetypes = concat(",", decodetypes, ",", NULL); + char *delimited_type = concat(",", type, ",", NULL); + + if (nmh_strcasestr(delimited_decodetypes, delimited_type)) { + found_match = 1; + } else if (subtype != NULL) { + char *delimited_type_subtype = + concat(",", type, "/", subtype, ",", NULL); + + if (nmh_strcasestr(delimited_decodetypes, delimited_type_subtype)) { + found_match = 1; + } + free(delimited_type_subtype); } - return status; + free(delimited_type); + free(delimited_decodetypes); + + return found_match; } @@ -1645,7 +2046,7 @@ strip_crs (CT ct, int *message_mods) { size_t i; int last_char_was_cr = 0; - if (end > 0) bytes_to_read -= bytes_read; + if (end > 0) { bytes_to_read -= bytes_read; } for (i = 0, cp = buffer; i < bytes_read; ++i, ++cp) { if (*cp == '\n' && last_char_was_cr) { @@ -1680,11 +2081,19 @@ strip_crs (CT ct, int *message_mods) { if (*cp == '\r') { last_char_was_cr = 1; } else if (last_char_was_cr) { - if (*cp != '\n') write (fd, "\r", 1); - write (fd, cp, 1); + if (*cp != '\n') { + if (write (fd, "\r", 1) < 0) { + advise (tempfile, "CR write"); + } + } + if (write (fd, cp, 1) < 0) { + advise (tempfile, "write"); + } last_char_was_cr = 0; } else { - write (fd, cp, 1); + if (write (fd, cp, 1) < 0) { + advise (tempfile, "write"); + } last_char_was_cr = 0; } } @@ -1722,6 +2131,8 @@ strip_crs (CT ct, int *message_mods) { } } + free (charset); + return status; } @@ -1736,14 +2147,18 @@ convert_charsets (CT ct, char *dest_charset, int *message_mods) { status = convert_charset (ct, dest_charset, message_mods); if (status == OK) { if (verbosw) { + char *ct_charset = content_charset (ct); + report (NULL, ct->c_partno, ct->c_file, - "convert %s to %s", - content_charset(ct), dest_charset); + "convert %s to %s", ct_charset, dest_charset); + free (ct_charset); } } else { + char *ct_charset = content_charset (ct); + report ("iconv", ct->c_partno, ct->c_file, - "failed to convert %s to %s", - content_charset(ct), dest_charset); + "failed to convert %s to %s", ct_charset, dest_charset); + free (ct_charset); } } break; @@ -1763,9 +2178,8 @@ convert_charsets (CT ct, char *dest_charset, int *message_mods) { case CT_MESSAGE: if (ct->c_subtype == MESSAGE_EXTERNAL) { - struct exbody *e; + struct exbody *e = (struct exbody *) ct->c_ctparams; - e = (struct exbody *) ct->c_ctparams; status = convert_charsets (e->eb_content, dest_charset, message_mods); } @@ -1779,6 +2193,87 @@ convert_charsets (CT ct, char *dest_charset, int *message_mods) { } +/* + * Fix various problems that aren't handled elsewhere. These + * are fixed unconditionally: there are no switches to disable + * them. (Currently, "problems" is just one: an extraneous + * semicolon at the end of a header parameter list.) + */ +static int +fix_always (CT ct, int *message_mods) { + int status = OK; + + switch (ct->c_type) { + case CT_MULTIPART: { + struct multipart *m = (struct multipart *) ct->c_ctparams; + struct part *part; + + for (part = m->mp_parts; status == OK && part; part = part->mp_next) { + status = fix_always (part->mp_part, message_mods); + } + break; + } + + case CT_MESSAGE: + if (ct->c_subtype == MESSAGE_EXTERNAL) { + struct exbody *e = (struct exbody *) ct->c_ctparams; + + status = fix_always (e->eb_content, message_mods); + } + break; + + default: { + HF hf; + + for (hf = ct->c_first_hf; hf; hf = hf->next) { + size_t len = strlen (hf->value); + + if (strcasecmp (hf->name, TYPE_FIELD) != 0 && + strcasecmp (hf->name, DISPO_FIELD) != 0) { + /* Only do this for Content-Type and + Content-Disposition fields because those are the + only headers that parse_mime() warns about. */ + continue; + } + + /* whitespace following a trailing ';' will be nuked as well */ + if (hf->value[len - 1] == '\n') { + while (isspace((unsigned char)(hf->value[len - 2]))) { + if (len-- == 0) { break; } + } + } + + if (hf->value[len - 2] == ';') { + /* Remove trailing ';' from parameter value. */ + hf->value[len - 2] = '\n'; + hf->value[len - 1] = '\0'; + + /* Also, if Content-Type parameter, remove trailing ';' + from ct->c_ctline. This probably isn't necessary + but can't hurt. */ + if (strcasecmp(hf->name, TYPE_FIELD) == 0 && ct->c_ctline) { + size_t l = strlen(ct->c_ctline) - 1; + while (isspace((unsigned char)(ct->c_ctline[l])) || + ct->c_ctline[l] == ';') { + ct->c_ctline[l--] = '\0'; + if (l == 0) { break; } + } + } + + ++*message_mods; + if (verbosw) { + report (NULL, ct->c_partno, ct->c_file, + "remove trailing ; from %s parameter value", + hf->name); + } + } + } + }} + + return status; +} + + static int write_content (CT ct, char *input_filename, char *outfile, int modify_inplace, int message_mods) { @@ -1812,8 +2307,8 @@ write_content (CT ct, char *input_filename, char *outfile, int modify_inplace, } } } - if (new != -1) close (new); - if (old != -1) close (old); + if (new != -1) { close (new); } + if (old != -1) { close (old); } (void) m_unlink (outfile); if (i < 0) { @@ -1853,6 +2348,44 @@ write_content (CT ct, char *input_filename, char *outfile, int modify_inplace, } +/* + * parse_mime() does not set lf_line_endings in struct text, so use this function to do it. + * It touches the parts the decodetypes identifies. + */ +static void +set_text_ctparams(CT ct, char *decodetypes, int lf_line_endings) { + switch (ct->c_type) { + case CT_MULTIPART: { + struct multipart *m = (struct multipart *) ct->c_ctparams; + struct part *part; + + for (part = m->mp_parts; part; part = part->mp_next) { + set_text_ctparams(part->mp_part, decodetypes, lf_line_endings); + } + break; + } + + case CT_MESSAGE: + if (ct->c_subtype == MESSAGE_EXTERNAL) { + struct exbody *e = (struct exbody *) ct->c_ctparams; + + set_text_ctparams(e->eb_content, decodetypes, lf_line_endings); + } + break; + + default: + if (should_decode(decodetypes, ct->c_ctinfo.ci_type, ct->c_ctinfo.ci_subtype)) { + if (ct->c_ctparams == NULL) { + if ((ct->c_ctparams = (struct text *) mh_xcalloc (1, sizeof (struct text))) == NULL) { + adios (NULL, "out of memory"); + } + } + ((struct text *) ct->c_ctparams)->lf_line_endings = lf_line_endings; + } + } +} + + /* * If "rmmproc" is defined, call that to remove the file. Otherwise, * use the standard MH backup file.