X-Git-Url: https://diplodocus.org/git/nmh/blobdiff_plain/f4e84752c9c8de7dcb266054135147434b0dc3b8..3789a22b9fd7711d55e9b54997e4677d2b96dc78:/uip/mhfixmsg.c diff --git a/uip/mhfixmsg.c b/uip/mhfixmsg.c index d6903974..af6cd889 100644 --- a/uip/mhfixmsg.c +++ b/uip/mhfixmsg.c @@ -11,26 +11,28 @@ #include #include #include -#include -#include #include -#ifdef HAVE_ICONV -# include -#endif #define MHFIXMSG_SWITCHES \ X("decodetext 8bit|7bit", 0, DECODETEXTSW) \ X("nodecodetext", 0, NDECODETEXTSW) \ - X("textcodeset", 0, TEXTCODESETSW) \ - X("notextcodeset", 0, NTEXTCODESETSW) \ + X("textcharset", 0, TEXTCHARSETSW) \ + X("notextcharset", 0, NTEXTCHARSETSW) \ X("reformat", 0, REFORMATSW) \ X("noreformat", 0, NREFORMATSW) \ + X("replacetextplain", 0, REPLACETEXTPLAINSW) \ + X("noreplacetextplain", 0, NREPLACETEXTPLAINSW) \ X("fixboundary", 0, FIXBOUNDARYSW) \ 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) \ @@ -51,19 +53,16 @@ int debugsw; /* Needed by mhparse.c. */ #define quitser pipeser /* mhparse.c */ -extern char *tmp; /* directory to place tmp files */ 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 */ -CT parse_mime (char *); -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 *); -/* mhshowsbr.c */ -int show_content_aux (CT, int, int, char *, char *); - /* mhmisc.c */ void flush_errors (void); @@ -77,20 +76,24 @@ void freects_done (int) NORETURN; typedef struct fix_transformations { int fixboundary; int fixcte; + svector_t fixtypes; int reformat; + int replacetextplain; int decodetext; - char *textcodeset; + 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 *, const char *); -static char *update_attr (char *, const char *, const char *e); +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 int fix_multipart_cte (CT, int *); static int set_ce (CT, int); -static int ensure_text_plain (CT *, 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 CT divide_part (CT); static void copy_ctinfo (CI, CI); @@ -102,13 +105,13 @@ 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 content_encoding (CT); -static int convert_codesets (CT, char *, int *); -static int convert_codeset (CT, char *, int *); +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 int remove_file (char *); -static void report (char *, char *, char *, ...); -static char *upcase (char *); +static void report (char *, char *, char *, char *, ...); static void pipeser (int); @@ -123,21 +126,18 @@ 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.textcodeset = NULL; - - done = freects_done; + fx.textcharset = NULL; -#ifdef LOCALE - setlocale(LC_ALL, ""); -#endif - invo_name = r1bindex (argv[0], '/'); + if (nmh_init(argv[0], 1)) { return 1; } - /* read user profile/context */ - context_read(); + done = freects_done; arguments = getarguments (invo_name, argc, argv, 1); argp = arguments; @@ -177,13 +177,13 @@ main (int argc, char **argv) { case NDECODETEXTSW: fx.decodetext = 0; continue; - case TEXTCODESETSW: + case TEXTCHARSETSW: if (! (cp = *argp++) || (*cp == '-' && cp[1])) adios (NULL, "missing argument to %s", argp[-2]); - fx.textcodeset = cp; + fx.textcharset = cp; continue; - case NTEXTCODESETSW: - fx.textcodeset = 0; + case NTEXTCHARSETSW: + fx.textcharset = 0; continue; case FIXBOUNDARYSW: fx.fixboundary = 1; @@ -197,25 +197,53 @@ 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; case NREFORMATSW: fx.reformat = 0; continue; - + case REPLACETEXTPLAINSW: + fx.replacetextplain = 1; + continue; + case NREPLACETEXTPLAINSW: + fx.replacetextplain = 0; + continue; case FILESW: if (! (cp = *argp++) || (*cp == '-' && cp[1])) adios (NULL, "missing argument to %s", argp[-2]); file = *cp == '-' ? add (cp, NULL) : path (cp, TFILE); continue; - case OUTFILESW: if (! (cp = *argp++) || (*cp == '-' && cp[1])) adios (NULL, "missing argument to %s", argp[-2]); outfile = *cp == '-' ? add (cp, NULL) : path (cp, TFILE); continue; - + case RPROCSW: + if (!(rmmproc = *argp++) || *rmmproc == '-') + adios (NULL, "missing argument to %s", argp[-2]); + continue; + case NRPRCSW: + rmmproc = NULL; + continue; + case CHGSW: + chgflag = 1; + continue; + case NCHGSW: + chgflag = 0; + continue; case VERBSW: verbosw = 1; continue; @@ -229,8 +257,14 @@ main (int argc, char **argv) { adios (NULL, "only one folder at a time!"); else folder = pluspath (cp); - } else - app_msgarg(&msgs, cp); + } else { + if (*cp == '/') { + /* Interpret a full path as a filename, not a message. */ + file = add (cp, NULL); + } else { + app_msgarg (&msgs, cp); + } + } } SIGNAL (SIGQUIT, quitser); @@ -244,17 +278,8 @@ main (int argc, char **argv) { fclose (fp); } - /* - * Check for storage directory. If specified, - * then store temporary files there. Else we - * store them in standard nmh directory. - */ - if ((cp = context_find (nmhstorage)) && *cp) - tmp = concat (cp, "/", invo_name, NULL); - else - tmp = add (m_maildir (invo_name), NULL); - suppress_bogus_mp_content_warning = skip_mp_cte_check = 1; + suppress_extraneous_trailing_semicolon_warning = 1; if (! context_find ("path")) free (path ("./", TFOLDER)); @@ -278,27 +303,27 @@ main (int argc, char **argv) { using_stdin = 1; - if ((cp = m_mktemp2 (tmp, invo_name, &fd, NULL)) == NULL) { - adios (NULL, "unable to create temporary file"); + if ((cp = m_mktemp2 (NULL, invo_name, &fd, NULL)) == NULL) { + adios (NULL, "unable to create temporary file in %s", + get_temp_dir()); } else { free (file); file = add (cp, NULL); - chmod (file, 0600); cpydata (STDIN_FILENO, fd, "-", file); } if (close (fd)) { - unlink (file); + (void) m_unlink (file); adios (NULL, "failed to write temporary file"); } } - 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))) { *ctp++ = ct; } } else { /* * message(s) are coming from a folder @@ -315,7 +340,7 @@ main (int argc, char **argv) { adios (maildir, "unable to change directory to"); /* read folder and create message structure */ - if (! (mp = folder_read (folder))) + if (! (mp = folder_read (folder, 1))) adios (NULL, "unable to read folder %s", folder); /* check for empty folder */ @@ -328,8 +353,10 @@ 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++) { @@ -337,24 +364,16 @@ main (int argc, char **argv) { char *msgnam; msgnam = m_name (msgnum); - if ((ct = parse_mime (msgnam))) - *ctp++ = ct; + if ((ct = parse_mime (msgnam))) { *ctp++ = ct; } } } - /* - * This is a hack. If we are using an external rmmproc, - * then save the current folder to the context file, - * so the external rmmproc will remove files from the correct - * directory. This should be moved to folder_delmsgs(). - */ - if (rmmproc) { - context_replace (pfolder, folder);/* update current folder */ - seq_setcur (mp, mp->hghsel); /* update current message */ - seq_save (mp); /* synchronize sequences */ - context_save (); /* save the context file */ - fflush (stdout); + 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 */ } if (*cts) { @@ -362,7 +381,7 @@ main (int argc, char **argv) { status += mhfixmsgsbr (ctp, &fx, outfile); if (using_stdin) { - unlink (file); + (void) m_unlink (file); if (! outfile) { /* Just calling m_backup() unlinks the backup file. */ @@ -374,8 +393,8 @@ main (int argc, char **argv) { status = 1; } + if (fx.fixtypes != NULL) { svector_free (fx.fixtypes); } free (outfile); - free (tmp); free (file); /* done is freects_done, which will clean up all of cts. */ @@ -397,30 +416,40 @@ mhfixmsgsbr (CT *ctp, const fix_transformations *fx, char *outfile) { modify_inplace = 1; if ((*ctp)->c_file) { - outfile = add (m_mktemp2 (tmp, invo_name, NULL, NULL), NULL); + char *tempfile; + if ((tempfile = m_mktemp2 (NULL, invo_name, NULL, NULL)) == NULL) { + adios (NULL, "unable to create temporary file in %s", + get_temp_dir()); + } + outfile = add (tempfile, NULL); } else { adios (NULL, "missing both input and output filenames\n"); } } 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); } if (status == OK && fx->reformat) { - status = ensure_text_plain (ctp, NULL, &message_mods); + status = + ensure_text_plain (ctp, NULL, &message_mods, fx->replacetextplain); } if (status == OK && fx->decodetext) { status = decode_text_parts (*ctp, fx->decodetext, &message_mods); } - if (status == OK && fx->textcodeset != NULL) { - status = convert_codesets (*ctp, fx->textcodeset, &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; @@ -458,6 +487,7 @@ mhfixmsgsbr (CT *ctp, const fix_transformations *fx, char *outfile) { } if (modify_inplace) { + if (status != OK) { (void) m_unlink (outfile); } free (outfile); outfile = NULL; } @@ -468,33 +498,12 @@ 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; /* @@ -513,17 +522,25 @@ fix_boundary (CT *ct, int *message_mods) { if (get_multipart_boundary (*ct, &part_boundary) == OK) { char *fixed; - if ((fixed = m_mktemp2 (tmp, invo_name, NULL, &(*ct)->c_fp))) { + 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); - *ct = parse_mime (fixed); - (*ct)->c_unlink = 1; - - ++*message_mods; - if (verbosw) { - report (NULL, filename, "fix multipart boundary"); + if ((fixed_ct = parse_mime (fixed))) { + *ct = fixed_ct; + (*ct)->c_unlink = 1; + + ++*message_mods; + if (verbosw) { + report (NULL, NULL, filename, + "fix multipart boundary"); + } + } else { + *ct = NULL; + advise (NULL, "unable to parse fixed part"); + status = NOTOK; } free (filename); } else { @@ -531,7 +548,8 @@ fix_boundary (CT *ct, int *message_mods) { status = NOTOK; } } else { - advise (NULL, "unable to create temporary file"); + advise (NULL, "unable to create temporary file in %s", + get_temp_dir()); status = NOTOK; } @@ -567,10 +585,11 @@ get_multipart_boundary (CT ct, char **part_boundary) { while (begin >= (off_t) ct->c_begin) { fseeko (ct->c_fp, begin, SEEK_SET); while ((bytes_read = fread (buffer, 1, sizeof buffer, ct->c_fp)) > 0) { - char *end = buffer + bytes_read - 1; - char *cp; + char *cp = rfind_str (buffer, bytes_read, "--"); + + if (cp) { + char *end; - if ((cp = rfind_str (buffer, bytes_read, "--"))) { /* Trim off trailing "--" and anything beyond. */ *cp-- = '\0'; if ((end = rfind_str (buffer, cp - buffer, "\n"))) { @@ -599,15 +618,18 @@ get_multipart_boundary (CT ct, char **part_boundary) { if (cp && cp - buffer >= 2 && *--cp == '-' && *--cp == '-' && (cp > buffer && *--cp == '\n')) { + status = OK; break; } - /* Else the start and end boundaries didn't match, or - the start boundary doesn't begin with "\n--" (or - "--" if at the beginning of buffer). Keep trying. */ } else { + /* The start and end boundaries didn't match, or the + start boundary doesn't begin with "\n--" (or "--" + if at the beginning of buffer). Keep trying. */ status = NOTOK; } } + } else { + status = NOTOK; } if (status == OK) { @@ -623,7 +645,7 @@ get_multipart_boundary (CT ct, char **part_boundary) { /* Open and copy ct->c_file to file, replacing the multipart boundary. */ static int -replace_boundary (CT ct, char *file, const char *boundary) { +replace_boundary (CT ct, char *file, char *boundary) { FILE *fpin, *fpout; int compnum, state; char buf[BUFSIZ], name[NAMESZ]; @@ -669,10 +691,22 @@ replace_boundary (CT ct, char *file, const char *boundary) { if (strcasecmp (TYPE_FIELD, np)) { fprintf (fpout, "%s:%s", np, vp); } else { - char *new_boundary = update_attr (vp, "boundary=", boundary); - - fprintf (fpout, "%s:%s\n", np, new_boundary); - free (new_boundary); + char *new_ctline, *new_params; + + replace_param(&ct->c_ctinfo.ci_first_pm, + &ct->c_ctinfo.ci_last_pm, "boundary", + boundary, 0); + + new_ctline = concat(" ", ct->c_ctinfo.ci_type, "/", + ct->c_ctinfo.ci_subtype, NULL); + new_params = output_params(strlen(TYPE_FIELD) + + strlen(new_ctline) + 1, + ct->c_ctinfo.ci_first_pm, NULL, 0); + fprintf (fpout, "%s:%s%s\n", np, new_ctline, + new_params ? new_params : ""); + free(new_ctline); + if (new_params) + free(new_params); } free (vp); @@ -683,7 +717,9 @@ replace_boundary (CT ct, char *file, const 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: @@ -712,41 +748,141 @@ replace_boundary (CT ct, char *file, const char *boundary) { } -/* Change the value of a name=value pair in a header field body. - If the name isn't there, append them. In any case, a new - string will be allocated and must be free'd by the caller. - Trims any trailing newlines. */ -static char * -update_attr (char *body, const char *name, const char *value) { - char *bp = nmh_strcasestr (body, name); - char *new_body; +static int +fix_types (CT ct, svector_t fixtypes, int *message_mods) { + int status = OK; - if (bp) { - char *other_attrs = strchr (bp, ';'); + switch (ct->c_type) { + case CT_MULTIPART: { + struct multipart *m = (struct multipart *) ct->c_ctparams; + struct part *part; - *(bp + strlen (name)) = '\0'; - new_body = concat (body, "\"", value, "\"", NULL); + for (part = m->mp_parts; status == OK && part; part = part->mp_next) { + status = fix_types (part->mp_part, fixtypes, message_mods); + } + break; + } - if (other_attrs) { - char *cp; + case CT_MESSAGE: + if (ct->c_subtype == MESSAGE_EXTERNAL) { + struct exbody *e = (struct exbody *) ct->c_ctparams; - /* Trim any trailing newlines. */ - for (cp = &other_attrs[strlen (other_attrs) - 1]; - cp > other_attrs && *cp == '\n'; - *cp-- = '\0') continue; - new_body = add (other_attrs, new_body); + status = fix_types (e->eb_content, fixtypes, message_mods); } - } else { - char *cp; + break; - /* Append name/value pair, after first removing a final newline - and (extraneous) semicolon. */ - if (*(cp = &body[strlen (body) - 1]) == '\n') *cp = '\0'; - if (*(cp = &body[strlen (body) - 1]) == ';') *cp = '\0'; - new_body = concat (body, "; ", name, "\"", value, "\"", NULL); - } + 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); + } - return new_body; + free (*str); + + return *str = new_str; + } else { + return NULL; + } } @@ -785,7 +921,7 @@ fix_multipart_cte (CT ct, int *message_mods) { ++*message_mods; if (verbosw) { char *encoding = cpytrim (hf->value); - report (ct->c_partno, ct->c_file, + report (NULL, ct->c_partno, ct->c_file, "replace Content-Transfer-Encoding of %s " "with 8 bit", encoding); free (encoding); @@ -868,82 +1004,110 @@ set_ce (CT ct, int encoding) { /* Make sure each text part has a corresponding text/plain part. */ static int -ensure_text_plain (CT *ct, CT parent, int *message_mods) { +ensure_text_plain (CT *ct, CT parent, int *message_mods, int replacetextplain) { int status = OK; 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; int new_subpart_number = 1; + int has_text_plain = + find_textplain_sibling (parent, replacetextplain, + &new_subpart_number); - /* See if there is a sibling text/plain. */ - for (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) { - has_text_plain = 1; - break; + if (! has_text_plain) { + /* Parent is a multipart/alternative. Insert a new + text/plain subpart. */ + 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 { + status = NOTOK; } } + } 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); + } + + free (type_subtype); 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 (parent->c_partno, parent->c_file, + 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); - CT mp_alt = build_multipart_alt (*ct, tp_part, CT_MULTIPART, - MULTI_ALTERNATE); - 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; + 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; - ++*message_mods; - if (verbosw) { - report ((*ct)->c_partno, (*ct)->c_file, - "insert text/plain part"); + 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 { - free_content (tp_part); - - /* Undo enough of what build_multipart_alt() did so - that free_content() can be called on mp_alt. */ - mp->mp_parts->mp_part = NULL; - mp->mp_parts->mp_next->mp_part = NULL; - free_content (mp_alt); status = NOTOK; } } @@ -956,7 +1120,8 @@ ensure_text_plain (CT *ct, CT parent, int *message_mods) { for (part = mp->mp_parts; status == OK && part; part = part->mp_next) { if ((*ct)->c_type == CT_MULTIPART) { - status = ensure_text_plain (&part->mp_part, *ct, message_mods); + status = ensure_text_plain (&part->mp_part, *ct, message_mods, + replacetextplain); } } break; @@ -964,10 +1129,10 @@ ensure_text_plain (CT *ct, CT parent, 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 = ensure_text_plain (&e->eb_content, *ct, message_mods); + status = ensure_text_plain (&e->eb_content, *ct, message_mods, + replacetextplain); } break; } @@ -976,6 +1141,71 @@ ensure_text_plain (CT *ct, CT parent, int *message_mods) { } +/* 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); @@ -986,17 +1216,24 @@ build_text_plain_part (CT encoded_part) { contains the decoded contents. And the decoding function, such as openQuoted, will have set ...->ce_unlink to 1 so that it will be unlinked by free_content (). */ - tmp_plain_file = add (m_mktemp2 (tmp, invo_name, NULL, NULL), 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; + char *tempfile; + + if ((tempfile = m_mktemp2 (NULL, invo_name, NULL, NULL)) == NULL) { + advise (NULL, "unable to create temporary file in %s", + get_temp_dir()); + } 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); - unlink (tmp_plain_file); + if (tmp_plain_file) { (void) m_unlink (tmp_plain_file); } free (tmp_plain_file); return NULL; @@ -1007,7 +1244,7 @@ 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 @@ -1034,19 +1271,19 @@ divide_part (CT ct) { static void copy_ctinfo (CI dest, CI src) { - char **s_ap, **d_ap, **s_vp, **d_vp; + PM s_pm, d_pm; dest->ci_type = src->ci_type ? add (src->ci_type, NULL) : NULL; dest->ci_subtype = src->ci_subtype ? add (src->ci_subtype, NULL) : NULL; - for (s_ap = src->ci_attrs, d_ap = dest->ci_attrs, - s_vp = src->ci_values, d_vp = dest->ci_values; - *s_ap; - ++s_ap, ++d_ap, ++s_vp, ++d_vp) { - *d_ap = add (*s_ap, NULL); - *d_vp = *s_vp; + for (s_pm = src->ci_first_pm; s_pm; s_pm = s_pm->pm_next) { + d_pm = add_param(&dest->ci_first_pm, &dest->ci_last_pm, s_pm->pm_name, + s_pm->pm_value, 0); + if (s_pm->pm_charset) + d_pm->pm_charset = getcpy(s_pm->pm_charset); + if (s_pm->pm_lang) + d_pm->pm_lang = getcpy(s_pm->pm_lang); } - *d_ap = NULL; dest->ci_comment = src->ci_comment ? add (src->ci_comment, NULL) : NULL; dest->ci_magic = src->ci_magic ? add (src->ci_magic, NULL) : NULL; @@ -1057,13 +1294,17 @@ static int decode_part (CT ct) { char *tmp_decoded; int status; + char *tempfile; - tmp_decoded = add (m_mktemp2 (tmp, invo_name, NULL, NULL), NULL); + if ((tempfile = m_mktemp2 (NULL, invo_name, NULL, NULL)) == NULL) { + adios (NULL, "unable to create temporary file in %s", get_temp_dir()); + } + tmp_decoded = add (tempfile, NULL); /* The following call will load ct->c_cefile.ce_file with the tmp filename of the decoded content. tmp_decoded will contain the encoded output, get rid of that. */ status = output_message (ct, tmp_decoded); - unlink (tmp_decoded); + (void) m_unlink (tmp_decoded); free (tmp_decoded); return status; @@ -1084,46 +1325,31 @@ 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, 1, 0, cp, NULL); + status = show_content_aux (ct, 0, cp, NULL, NULL); free (cp); /* Unlink decoded content tmp file and free its filename to avoid leaks. The file stream should already have been closed. */ if (ct->c_cefile.ce_unlink) { - unlink (ct->c_cefile.ce_file); + (void) m_unlink (ct->c_cefile.ce_file); free (ct->c_cefile.ce_file); ct->c_cefile.ce_file = NULL; ct->c_cefile.ce_unlink = 0; @@ -1149,23 +1375,13 @@ reformat_part (CT ct, char *file, char *type, char *subtype, int c_type) { } -/* Identifies 7bit or 8bit content based on charset, if specified. */ +/* Identifies 7bit or 8bit content based on charset. */ static int charset_encoding (CT ct) { - int encoding = CE_8BIT; - CI ctinfo = &ct->c_ctinfo; - char **ap, **vp; - - for (ap = ctinfo->ci_attrs, vp = ctinfo->ci_values; *ap; ++ap, ++vp) { - if (! strcasecmp (*ap, "charset")) { - /* norm_charmap() is case sensitive. */ - char *ch = upcase (*vp); + char *ct_charset = content_charset (ct); + int encoding = strcasecmp (ct_charset, "US-ASCII") ? CE_8BIT : CE_7BIT; - if (! strcmp (norm_charmap (ch), "US-ASCII")) encoding = CE_7BIT; - free (ch); - break; - } - } + free (ct_charset); return encoding; } @@ -1180,14 +1396,13 @@ build_multipart_alt (CT first_alt, CT new_part, int type, int subtype) { CT ct; struct part *p; struct multipart *m; - char *cp; 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, @@ -1195,7 +1410,7 @@ build_multipart_alt (CT first_alt, CT new_part, int type, int subtype) { c_cefile, c_encoding, c_digested, c_digest[16], c_ctexbody, c_ctinitfnx, c_ceopenfnx, c_ceclosefnx, c_cesizefnx, - c_umask, c_pid, c_rfc934, + c_umask, c_rfc934, c_showproc, c_termproc, c_storeproc, c_storage, c_folder */ @@ -1222,6 +1437,7 @@ build_multipart_alt (CT first_alt, CT new_part, int type, int subtype) { boundary_in_content (&new_part->c_cefile.ce_fp, new_part->c_cefile.ce_file, boundary)) == -1) { + free (ct); return NULL; } } @@ -1232,6 +1448,7 @@ build_multipart_alt (CT first_alt, CT new_part, int type, int subtype) { if ((found_boundary = boundary_in_content (&new_part->c_fp, new_part->c_file, boundary)) == -1) { + free (ct); return NULL; } } @@ -1252,6 +1469,7 @@ build_multipart_alt (CT first_alt, CT new_part, int type, int subtype) { if (found_boundary) { advise (NULL, "giving up trying to find a unique boundary"); + free (ct); return NULL; } } @@ -1280,27 +1498,21 @@ build_multipart_alt (CT first_alt, CT new_part, int type, int subtype) { ct->c_ctinfo.ci_subtype = add (subtypename, NULL); } - name = concat (" ", typename, "/", subtypename, boundary_indicator, - boundary, NULL); - if ((cp = strstr (name, boundary_indicator))) { - ct->c_ctinfo.ci_attrs[0] = name; - ct->c_ctinfo.ci_attrs[1] = NULL; - /* ci_values don't get free'd, so point into ci_attrs. */ - ct->c_ctinfo.ci_values[0] = cp + strlen (boundary_indicator); - } + add_param(&ct->c_ctinfo.ci_first_pm, &ct->c_ctinfo.ci_last_pm, + "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); @@ -1441,39 +1653,52 @@ decode_text_parts (CT ct, int encoding, int *message_mods) { int ct_encoding; if (decode_part (ct) == OK && ct->c_cefile.ce_file) { - if ((ct_encoding = content_encoding (ct)) == CE_BINARY && - encoding != CE_BINARY) { + const char *reason = NULL; + + if ((ct_encoding = content_encoding (ct, &reason)) == CE_BINARY + && encoding != CE_BINARY) { + /* The decoding isn't acceptable so discard it. + Leave status as OK to allow other transformations. */ if (verbosw) { - report (ct->c_partno, ct->c_file, - "will not decode%s because it is binary", + report (NULL, ct->c_partno, ct->c_file, + "will not decode%s because it is binary (%s)", ct->c_partno ? "" : ct->c_ctline ? ct->c_ctline - : ""); + : "", + reason); } - unlink (ct->c_cefile.ce_file); + (void) m_unlink (ct->c_cefile.ce_file); free (ct->c_cefile.ce_file); ct->c_cefile.ce_file = NULL; - } else if (ct_encoding == CE_8BIT && encoding == CE_7BIT) { + } else if (ct->c_encoding == CE_QUOTED && + ct_encoding == CE_8BIT && encoding == CE_7BIT) { + /* The decoding isn't acceptable so discard it. + Leave status as OK to allow other transformations. */ if (verbosw) { - report (ct->c_partno, ct->c_file, + report (NULL, ct->c_partno, ct->c_file, "will not decode%s because it is 8bit", ct->c_partno ? "" : ct->c_ctline ? ct->c_ctline : ""); } - unlink (ct->c_cefile.ce_file); + (void) m_unlink (ct->c_cefile.ce_file); free (ct->c_cefile.ce_file); ct->c_cefile.ce_file = NULL; } else { - int enc = ct_encoding == CE_BINARY - ? CE_BINARY - : charset_encoding (ct); + int enc; + if (ct_encoding == CE_BINARY) + enc = CE_BINARY; + else if (ct_encoding == CE_8BIT && encoding == CE_7BIT) + enc = CE_QUOTED; + else + enc = charset_encoding (ct); if (set_ce (ct, enc) == OK) { ++*message_mods; if (verbosw) { - report (ct->c_partno, ct->c_file, "decode%s", + report (NULL, ct->c_partno, ct->c_file, "decode%s", ct->c_ctline ? ct->c_ctline : ""); } + strip_crs (ct, message_mods); } else { status = NOTOK; } @@ -1483,9 +1708,14 @@ decode_text_parts (CT ct, int encoding, int *message_mods) { } break; } + case CE_8BIT: + case CE_7BIT: + strip_crs (ct, message_mods); + break; default: break; } + break; case CT_MULTIPART: { @@ -1502,9 +1732,8 @@ decode_text_parts (CT ct, int encoding, 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 = decode_text_parts (e->eb_content, encoding, message_mods); } break; @@ -1519,13 +1748,15 @@ decode_text_parts (CT ct, int encoding, int *message_mods) { /* See if the decoded content is 7bit, 8bit, or binary. It's binary if it has any NUL characters, a CR not followed by a LF, or lines - greater than 998 characters in length. */ + greater than 998 characters in length. If binary, reason is set + to a string explaining why. */ static int -content_encoding (CT ct) { +content_encoding (CT ct, const char **reason) { CE ce = &ct->c_cefile; int encoding = CE_7BIT; if (ce->ce_file) { + size_t line_len = 0; char buffer[BUFSIZ]; size_t inbytes; @@ -1539,13 +1770,22 @@ content_encoding (CT ct) { (inbytes = fread (buffer, 1, sizeof buffer, ce->ce_fp)) > 0) { char *cp; size_t i; - size_t line_len = 0; int last_char_was_cr = 0; for (i = 0, cp = buffer; i < inbytes; ++i, ++cp) { if (*cp == '\0' || ++line_len > 998 || (*cp != '\n' && last_char_was_cr)) { encoding = CE_BINARY; + if (*cp == '\0') { + *reason = "null character"; + } else if (line_len > 998) { + *reason = "line length > 998"; + } else if (*cp != '\n' && last_char_was_cr) { + *reason = "CR not followed by LF"; + } else { + /* Should not reach this. */ + *reason = ""; + } break; } else if (*cp == '\n') { line_len = 0; @@ -1566,13 +1806,179 @@ content_encoding (CT ct) { static int -convert_codesets (CT ct, char *dest_codeset, int *message_mods) { +strip_crs (CT ct, int *message_mods) { + char *charset = content_charset (ct); + int status = OK; + + /* Only strip carriage returns if content is ASCII or another + charset that has the same readily recognizable CR followed by a + LF. We can include UTF-8 here because if the high-order bit of + a UTF-8 byte is 0, then it must be a single-byte ASCII + character. */ + if (! strcasecmp (charset, "US-ASCII") || + ! strcasecmp (charset, "UTF-8") || + ! strncasecmp (charset, "ISO-8859-", 9) || + ! strncasecmp (charset, "WINDOWS-12", 10)) { + char **file = NULL; + FILE **fp = NULL; + size_t begin; + size_t end; + int has_crs = 0; + int opened_input_file = 0; + + if (ct->c_cefile.ce_file) { + file = &ct->c_cefile.ce_file; + fp = &ct->c_cefile.ce_fp; + begin = end = 0; + } else if (ct->c_file) { + file = &ct->c_file; + fp = &ct->c_fp; + begin = (size_t) ct->c_begin; + end = (size_t) ct->c_end; + } /* else don't know where the content is */ + + if (file && *file && fp) { + if (! *fp) { + if ((*fp = fopen (*file, "r")) == NULL) { + advise (*file, "unable to open for reading"); + status = NOTOK; + } else { + opened_input_file = 1; + } + } + } + + if (fp && *fp) { + char buffer[BUFSIZ]; + size_t bytes_read; + size_t bytes_to_read = + end > 0 && end > begin ? end - begin : sizeof buffer; + + fseeko (*fp, begin, SEEK_SET); + while ((bytes_read = fread (buffer, 1, + min (bytes_to_read, sizeof buffer), + *fp)) > 0) { + /* Look for CR followed by a LF. This is supposed to + be text so there should be LF's. If not, don't + modify the content. */ + char *cp; + size_t i; + int last_char_was_cr = 0; + + 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) { + has_crs = 1; + break; + } + + last_char_was_cr = *cp == '\r' ? 1 : 0; + } + } + + if (has_crs) { + int fd; + char *stripped_content_file; + char *tempfile = m_mktemp2 (NULL, invo_name, &fd, NULL); + + if (tempfile == NULL) { + adios (NULL, "unable to create temporary file in %s", + get_temp_dir()); + } + stripped_content_file = add (tempfile, NULL); + + /* Strip each CR before a LF from the content. */ + fseeko (*fp, begin, SEEK_SET); + while ((bytes_read = fread (buffer, 1, sizeof buffer, *fp)) > + 0) { + char *cp; + size_t i; + int last_char_was_cr = 0; + + for (i = 0, cp = buffer; i < bytes_read; ++i, ++cp) { + if (*cp == '\r') { + last_char_was_cr = 1; + } else if (last_char_was_cr) { + 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 { + if (write (fd, cp, 1) < 0) { + advise (tempfile, "write"); + } + last_char_was_cr = 0; + } + } + } + + if (close (fd)) { + admonish (NULL, "unable to write temporary file %s", + stripped_content_file); + (void) m_unlink (stripped_content_file); + status = NOTOK; + } else { + /* Replace the decoded file with the converted one. */ + if (ct->c_cefile.ce_file) { + if (ct->c_cefile.ce_unlink) { + (void) m_unlink (ct->c_cefile.ce_file); + } + free (ct->c_cefile.ce_file); + } + ct->c_cefile.ce_file = stripped_content_file; + ct->c_cefile.ce_unlink = 1; + + ++*message_mods; + if (verbosw) { + report (NULL, ct->c_partno, + begin == 0 && end == 0 ? "" : *file, + "stripped CRs"); + } + } + } + + if (opened_input_file) { + fclose (*fp); + *fp = NULL; + } + } + } + + free (charset); + + return status; +} + + +static int +convert_charsets (CT ct, char *dest_charset, int *message_mods) { int status = OK; switch (ct->c_type) { case CT_TEXT: if (ct->c_subtype == TEXT_PLAIN) { - status = convert_codeset (ct, dest_codeset, 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", 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", ct_charset, dest_charset); + free (ct_charset); + } } break; @@ -1584,18 +1990,17 @@ convert_codesets (CT ct, char *dest_codeset, int *message_mods) { For now, it gets passed along as-is by InitMultiPart(). */ for (part = m->mp_parts; status == OK && part; part = part->mp_next) { status = - convert_codesets (part->mp_part, dest_codeset, message_mods); + convert_charsets (part->mp_part, dest_charset, message_mods); } break; } 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_codesets (e->eb_content, dest_codeset, message_mods); + convert_charsets (e->eb_content, dest_charset, message_mods); } break; @@ -1607,154 +2012,82 @@ convert_codesets (CT ct, char *dest_codeset, 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 -convert_codeset (CT ct, char *dest_codeset, int *message_mods) { - const char *const charset = "charset"; - char **src_codeset = NULL; - char *default_codeset = NULL; - CI ctinfo = &ct->c_ctinfo; - char **ap, **vp; +fix_always (CT ct, int *message_mods) { int status = OK; - for (ap = ctinfo->ci_attrs, vp = ctinfo->ci_values; *ap; ++ap, ++vp) { - if (! strcasecmp (*ap, charset)) { - src_codeset = vp; - break; - } - } - /* RFC 2045, Sec. 5.2: default to us-ascii. */ - if (src_codeset == NULL) src_codeset = &default_codeset; - if (*src_codeset == NULL) *src_codeset = "US-ASCII"; - - if (strcmp (norm_charmap (*src_codeset), norm_charmap (dest_codeset))) { -#ifdef HAVE_ICONV - iconv_t conv_desc = NULL; - char *dest; - int fd = -1; - char **file = NULL; - FILE **fp = NULL; - long begin; - long end; - int opened_input_file = 0; - char src_buffer[BUFSIZ]; - HF hf; + switch (ct->c_type) { + case CT_MULTIPART: { + struct multipart *m = (struct multipart *) ct->c_ctparams; + struct part *part; - if ((conv_desc = iconv_open (dest_codeset, *src_codeset)) == - (iconv_t) -1) { - advise (NULL, "Can't convert %s to %s", *src_codeset, dest_codeset); - return -1; + for (part = m->mp_parts; status == OK && part; part = part->mp_next) { + status = fix_always (part->mp_part, message_mods); } + break; + } - dest = add (m_mktemp2 (tmp, invo_name, &fd, NULL), NULL); - - if (ct->c_cefile.ce_file) { - file = &ct->c_cefile.ce_file; - fp = &ct->c_cefile.ce_fp; - begin = 0; - end = -1; - } else if (ct->c_file) { - file = &ct->c_file; - fp = &ct->c_fp; - begin = ct->c_begin; - end = ct->c_end; - } /* else no input file: shouldn't happen */ + case CT_MESSAGE: + if (ct->c_subtype == MESSAGE_EXTERNAL) { + struct exbody *e = (struct exbody *) ct->c_ctparams; - if (file && *file && fp) { - if (! *fp) { - if ((*fp = fopen (*file, "r")) == NULL) { - advise (*file, "unable to open for reading"); - status = NOTOK; - } else { - opened_input_file = 1; - } - } + status = fix_always (e->eb_content, message_mods); } + break; - if (fp && *fp) { - size_t inbytes; - size_t max = end > 0 ? (size_t) (end-begin) : sizeof src_buffer; - - fseeko (*fp, begin, SEEK_SET); - while (status == OK && max > 0 && - (inbytes = fread (src_buffer, 1, max, *fp)) > 0) { - char dest_buffer[BUFSIZ]; - char *ib = src_buffer, *ob = dest_buffer; - size_t outbytes = sizeof dest_buffer; - size_t outbytes_before = outbytes; - - if (end > 0) max -= inbytes; + default: { + HF hf; - if (iconv (conv_desc, &ib, &inbytes, &ob, &outbytes) == - (size_t) -1) { - status = NOTOK; - break; - } else { - write (fd, dest_buffer, outbytes_before - outbytes); - } - } + for (hf = ct->c_first_hf; hf; hf = hf->next) { + size_t len = strlen (hf->value); - if (opened_input_file) { - fclose (*fp); - *fp = NULL; + 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; } - } - iconv_close (conv_desc); - close (fd); - - if (status == OK) { - /* Replace the decoded file with the converted one. */ - if (ct->c_cefile.ce_file) { - if (ct->c_cefile.ce_unlink) { - unlink (ct->c_cefile.ce_file); + /* 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; } } - free (ct->c_cefile.ce_file); - } - ct->c_cefile.ce_file = dest; - ct->c_cefile.ce_unlink = 1; - - ++*message_mods; - if (verbosw) { - report (ct->c_partno, ct->c_file, "convert %s to %s", - *src_codeset, dest_codeset); } - /* Update ci_attrs. */ - *src_codeset = dest_codeset; - - /* Update ct->c_ctline. */ - if (ct->c_ctline) { - char *ctline = - update_attr (ct->c_ctline, "charset=", dest_codeset); - - free (ct->c_ctline); - ct->c_ctline = ctline; - } /* else no CT line, which is odd */ + 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; } + } + } - /* Update Content-Type header field. */ - for (hf = ct->c_first_hf; hf; hf = hf->next) { - if (! strcasecmp (TYPE_FIELD, hf->name)) { - char *ctline_less_newline = - update_attr (hf->value, "charset=", dest_codeset); - char *ctline = concat (ctline_less_newline, "\n", NULL); - free (ctline_less_newline); - - free (hf->value); - hf->value = ctline; - break; + ++*message_mods; + if (verbosw) { + report (NULL, ct->c_partno, ct->c_file, + "remove trailing ; from %s parameter value", + hf->name); } } - } else { - unlink (dest); } -#else /* ! HAVE_ICONV */ - NMH_UNUSED (message_mods); - - advise (NULL, "Can't convert %s to %s without iconv", *src_codeset, - dest_codeset); - status = NOTOK; -#endif /* ! HAVE_ICONV */ - } + }} return status; } @@ -1774,27 +2107,54 @@ write_content (CT ct, char *input_filename, char *outfile, int modify_inplace, if (remove_file (infile) == OK) { if (rename (outfile, infile)) { - /* The -file argument processing used path() to - expand filename to absolute path. */ - int file = ct->c_file && ct->c_file[0] == '/'; - - admonish (NULL, "unable to rename %s %s to %s", - file ? "file" : "message", outfile, infile); - unlink (outfile); - status = NOTOK; + /* Rename didn't work, possibly because of an + attempt to rename across filesystems. Try + brute force copy. */ + int old = open (outfile, O_RDONLY); + int new = + open (infile, O_WRONLY | O_CREAT, m_gmprot ()); + int i = -1; + + if (old != -1 && new != -1) { + char buffer[BUFSIZ]; + + while ((i = read (old, buffer, sizeof buffer)) > + 0) { + if (write (new, buffer, i) != i) { + i = -1; + break; + } + } + } + if (new != -1) { close (new); } + if (old != -1) { close (old); } + (void) m_unlink (outfile); + + if (i < 0) { + /* The -file argument processing used path() to + expand filename to absolute path. */ + int file = ct->c_file && ct->c_file[0] == '/'; + + admonish (NULL, "unable to rename %s %s to %s", + file ? "file" : "message", outfile, + infile); + status = NOTOK; + } } } else { admonish (NULL, "unable to remove input file %s, " "not modifying it", infile); - unlink (outfile); + (void) m_unlink (outfile); status = NOTOK; } free (infile); + } else { + status = NOTOK; } } else { /* No modifications and didn't need the tmp outfile. */ - unlink (outfile); + (void) m_unlink (outfile); } } else { /* Output is going to some file. Produce it whether or not @@ -1829,7 +2189,7 @@ remove_file (char *file) { static void -report (char *partno, char *filename, char *message, ...) { +report (char *what, char *partno, char *filename, char *message, ...) { va_list args; char *fmt; @@ -1838,7 +2198,7 @@ report (char *partno, char *filename, char *message, ...) { fmt = concat (filename, partno ? " part " : ", ", partno ? partno : "", partno ? ", " : "", message, NULL); - advertise (NULL, NULL, fmt, args); + advertise (what, NULL, fmt, args); free (fmt); va_end (args); @@ -1846,17 +2206,6 @@ report (char *partno, char *filename, char *message, ...) { } -static char * -upcase (char *str) { - char *up = cpytrim (str); - char *cp; - - for (cp = up; *cp; ++cp) *cp = toupper ((unsigned char) *cp); - - return up; -} - - static void pipeser (int i) {