/*
- * mhfixmsg.c -- rewrite a message with various tranformations
+ * mhfixmsg.c -- rewrite a message with various transformations
*
* This code is Copyright (c) 2002 and 2013, by the authors of nmh.
* See the COPYRIGHT file in the root directory of the nmh
#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) \
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 int fix_boundary (CT *, int *);
+static int copy_input_to_output (const char *, const char *);
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);
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 int remove_file (char *);
+static int write_content (CT, const char *, char *, int, int);
+static void set_text_ctparams(CT, char *, int);
+static int remove_file (const char *);
static void report (char *, char *, char *, char *, ...);
static void pipeser (int);
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; }
done (0);
case DECODETEXTSW:
- if (! (cp = *argp++) || *cp == '-')
+ if (! (cp = *argp++) || *cp == '-') {
adios (NULL, "missing argument to %s", argp[-2]);
+ }
if (! strcasecmp (cp, "8bit")) {
fx.decodetext = CE_8BIT;
} else if (! strcasecmp (cp, "7bit")) {
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]))
+ if (! (cp = *argp++) || (*cp == '-' && cp[1])) {
adios (NULL, "missing argument to %s", argp[-2]);
+ }
fx.textcharset = cp;
continue;
case NTEXTCHARSETSW:
fx.fixcte = 0;
continue;
case FIXTYPESW:
- if (! (cp = *argp++) || (*cp == '-' && cp[1]))
+ 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);
fx.replacetextplain = 0;
continue;
case FILESW:
- if (! (cp = *argp++) || (*cp == '-' && cp[1]))
+ 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]))
+ 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 == '-')
+ if (!(rmmproc = *argp++) || *rmmproc == '-') {
adios (NULL, "missing argument to %s", argp[-2]);
+ }
continue;
case NRPRCSW:
rmmproc = NULL;
}
}
if (*cp == '+' || *cp == '@') {
- if (folder)
+ if (folder) {
adios (NULL, "only one folder at a time!");
- else
+ } else {
folder = pluspath (cp);
+ }
} else {
if (*cp == '/') {
/* Interpret a full path as a filename, not a message. */
suppress_bogus_mp_content_warning = skip_mp_cte_check = 1;
suppress_extraneous_trailing_semicolon_warning = 1;
- if (! context_find ("path"))
+ if (! context_find ("path")) {
free (path ("./", TFOLDER));
+ }
- if (file && msgs.size)
+ if (file && msgs.size) {
adios (NULL, "cannot specify msg and file at same time!");
+ }
/*
* check if message is coming from file
}
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 {
+ advise (NULL, "unable to parse message from file %s", file);
+ status = NOTOK;
+
+ /* If there's an outfile, pass the input message unchanged, so the message won't
+ get dropped from a pipeline. */
+ if (outfile) {
+ /* Something went wrong. Output might be expected, such as if this were run
+ as a filter. Just copy the input to the output. */
+ if (copy_input_to_output (file, outfile) != OK) {
+ advise (NULL, "unable to copy message to %s, it might be lost\n", outfile);
+ }
+ }
+ }
} else {
/*
* message(s) are coming from a folder
*/
CT ct;
- if (! msgs.size)
+ if (! msgs.size) {
app_msgarg(&msgs, "cur");
- if (! folder)
+ }
+ if (! folder) {
folder = getfolder (1);
+ }
maildir = m_maildir (folder);
- if (chdir (maildir) == NOTOK)
+ if (chdir (maildir) == NOTOK) {
adios (maildir, "unable to change directory to");
+ }
/* read folder and create message structure */
- if (! (mp = folder_read (folder, 1)))
+ if (! (mp = folder_read (folder, 1))) {
adios (NULL, "unable to read folder %s", folder);
+ }
/* check for empty folder */
- if (mp->nummsg == 0)
+ if (mp->nummsg == 0) {
adios (NULL, "no messages in %s", folder);
+ }
/* parse all the message ranges/sequences and set SELECTED */
for (msgnum = 0; msgnum < msgs.size; msgnum++)
- if (! m_convert (mp, msgs.msgs[msgnum]))
+ if (! m_convert (mp, msgs.msgs[msgnum])) {
done (1);
+ }
seq_setprev (mp); /* set the previous-sequence */
if (! (cts =
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;
+ } else {
+ advise (NULL, "unable to parse message %s", msgnam);
+ status = NOTOK;
+
+ /* If there's an outfile, pass the input message unchanged, so the message won't
+ get dropped from a pipeline. */
+ if (outfile) {
+ /* Something went wrong. Output might be expected, such as if this were run
+ as a filter. Just copy the input to the output. */
+ const char *input_filename = path (msgnam, TFILE);
+
+ if (copy_input_to_output (input_filename, outfile) != OK) {
+ advise (NULL, "unable to copy message to %s, it might be lost\n", outfile);
+ }
+ }
+ }
}
}
if (fx.fixtypes != NULL) { svector_free (fx.fixtypes); }
free (outfile);
free (file);
+ free (folder);
+ free (arguments);
/* done is freects_done, which will clean up all of cts. */
done (status);
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);
/* Something went wrong. Output might be expected, such
as if this were run as a filter. Just copy the input
to the output. */
- int in = open (input_filename, O_RDONLY);
- int out = strcmp (outfile, "-")
- ? open (outfile, O_WRONLY | O_CREAT, m_gmprot ())
- : STDOUT_FILENO;
-
- if (in != -1 && out != -1) {
- cpydata (in, out, input_filename, outfile);
- } else {
- status = NOTOK;
+ if (copy_input_to_output (input_filename, outfile) != OK) {
+ advise (NULL, "unable to copy message to %s, it might be lost\n", outfile);
}
-
- close (out);
- close (in);
}
if (modify_inplace) {
}
+/* Copy input message to output. Assumes not modifying in place, so this
+ might be running as part of a pipeline. */
+static int
+copy_input_to_output (const char *input_filename, const char *output_filename) {
+ int in = open (input_filename, O_RDONLY);
+ int out = strcmp (output_filename, "-")
+ ? open (output_filename, O_WRONLY | O_CREAT, m_gmprot ())
+ : STDOUT_FILENO;
+ int status = OK;
+
+ if (in != -1 && out != -1) {
+ cpydata (in, out, input_filename, output_filename);
+ } else {
+ status = NOTOK;
+ }
+
+ close (out);
+ close (in);
+
+ return status;
+}
+
+
static int
fix_boundary (CT *ct, int *message_mods) {
struct multipart *mp;
}
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;
}
}
fprintf (fpout, "%s:%s%s\n", np, new_ctline,
new_params ? new_params : "");
free(new_ctline);
- if (new_params)
+ if (new_params) {
free(new_params);
+ }
}
free (vp);
}
}
+/*
+ * 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) {
free (type_subtype);
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");
+ struct multipart *mp = (struct multipart *) parent->c_ctparams;
+ struct part *part;
+ int siblings = 0;
+
+ for (part = mp->mp_parts; part; part = part->mp_next) {
+ if (*ct != part->mp_part) {
+ ++siblings;
}
- } else {
- status = NOTOK;
}
- }
- } else {
- /* Slip new text/plain part into a new multipart/alternative. */
- CT tp_part = build_text_plain_part (*ct);
- 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 (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;
- if (mp && mp->mp_parts) {
- mp->mp_parts->mp_part = tp_part;
- /* Make the new multipart/alternative the parent. */
- *ct = mp_alt;
+ 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);
+ }
- ++*message_mods;
- if (verbosw) {
- report (NULL, (*ct)->c_partno, (*ct)->c_file,
- "insert text/plain part");
+ /* 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;
}
}
}
+/* 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;
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: {
ct->c_cefile.ce_file = NULL;
} else {
int enc;
- if (ct_encoding == CE_BINARY)
+ if (ct_encoding == CE_BINARY) {
enc = CE_BINARY;
- else if (ct_encoding == CE_8BIT && encoding == CE_7BIT)
+ } else if (ct_encoding == CE_8BIT && encoding == CE_7BIT) {
enc = CE_QUOTED;
- else
+ } else {
enc = charset_encoding (ct);
+ }
if (set_ce (ct, enc) == OK) {
++*message_mods;
if (verbosw) {
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;
}
}
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 = (struct exbody *) ct->c_ctparams;
+ return status;
+}
- 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;
}
static int
-write_content (CT ct, char *input_filename, char *outfile, int modify_inplace,
+write_content (CT ct, const char *input_filename, char *outfile, int modify_inplace,
int message_mods) {
int status = OK;
}
+/*
+ * 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.
*/
static int
-remove_file (char *file) {
+remove_file (const char *file) {
if (rmmproc) {
char *rmm_command = concat (rmmproc, " ", file, NULL);
int status = system (rmm_command);
projects / nmh / blobdiff