* Exim - an Internet mail transport agent *
*************************************************/
-/* Copyright (c) Tom Kistner <tom@duncanthrax.net> 2004 */
-/* License: GPL */
+/*
+ * Copyright (c) The Exim Maintainers 2015 - 2023
+ * Copyright (c) Tom Kistner <tom@duncanthrax.net> 2004 - 2015
+ * License: GPL
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ */
#include "exim.h"
#ifdef WITH_CONTENT_SCAN /* entire file */
FILE *mime_stream = NULL;
uschar *mime_current_boundary = NULL;
+static mime_header mime_header_list[] = {
+ /* name namelen value */
+ { US"content-type:", 13, &mime_content_type },
+ { US"content-disposition:", 20, &mime_content_disposition },
+ { US"content-transfer-encoding:", 26, &mime_content_transfer_encoding },
+ { US"content-id:", 11, &mime_content_id },
+ { US"content-description:", 20, &mime_content_description }
+};
+
+static int mime_header_list_size = nelem(mime_header_list);
+
+static mime_parameter mime_parameter_list[] = {
+ /* name namelen value */
+ { US"name=", 5, &mime_filename },
+ { US"filename=", 9, &mime_filename },
+ { US"charset=", 8, &mime_charset },
+ { US"boundary=", 9, &mime_boundary }
+};
+
+
/*************************************************
* set MIME anomaly level + text *
*************************************************/
/* Small wrapper to set the two expandables which
give info on detected "problems" in MIME
- encodings. Those are defined in mime.h. */
+ encodings. Indexes are defined in mime.h. */
-static void
-mime_set_anomaly(int level, const char *text)
+void
+mime_set_anomaly(int idx)
{
- mime_anomaly_level = level;
- mime_anomaly_text = CUS text;
+struct anom {
+ int level;
+ const uschar * text;
+} anom[] = { {1, CUS"Broken Quoted-Printable encoding detected"},
+ {2, CUS"Broken BASE64 encoding detected"} };
+
+mime_anomaly_level = anom[idx].level;
+mime_anomaly_text = anom[idx].text;
}
static ssize_t
mime_decode_asis(FILE* in, FILE* out, uschar* boundary)
{
- ssize_t len, size = 0;
- uschar buffer[MIME_MAX_LINE_LENGTH];
+ssize_t len, size = 0;
+uschar buffer[MIME_MAX_LINE_LENGTH];
- while(fgets(CS buffer, MIME_MAX_LINE_LENGTH, mime_stream) != NULL)
- {
- if (boundary != NULL
- && Ustrncmp(buffer, "--", 2) == 0
- && Ustrncmp((buffer+2), boundary, Ustrlen(boundary)) == 0
- )
- break;
+while(fgets(CS buffer, MIME_MAX_LINE_LENGTH, mime_stream) != NULL)
+ {
+ if (boundary != NULL
+ && Ustrncmp(buffer, "--", 2) == 0
+ && Ustrncmp((buffer+2), boundary, Ustrlen(boundary)) == 0
+ )
+ break;
- len = Ustrlen(buffer);
- if (fwrite(buffer, 1, (size_t)len, out) < len)
- return -1;
- size += len;
- } /* while */
- return size;
+ len = Ustrlen(buffer);
+ if (fwrite(buffer, 1, (size_t)len, out) < len)
+ return -1;
+ size += len;
+ } /* while */
+return size;
}
-/* decode base64 MIME part */
-static ssize_t
-mime_decode_base64(FILE* in, FILE* out, uschar* boundary)
-{
- uschar ibuf[MIME_MAX_LINE_LENGTH], obuf[MIME_MAX_LINE_LENGTH];
- uschar *ipos, *opos;
- ssize_t len, size = 0;
- int bytestate = 0;
-
- opos = obuf;
-
- while (Ufgets(ibuf, MIME_MAX_LINE_LENGTH, in) != NULL)
- {
- if (boundary != NULL
- && Ustrncmp(ibuf, "--", 2) == 0
- && Ustrncmp((ibuf+2), boundary, Ustrlen(boundary)) == 0
- )
- break;
-
- for (ipos = ibuf ; *ipos != '\r' && *ipos != '\n' && *ipos != 0; ++ipos)
- {
- if (*ipos == '=') /* skip padding */
- {
- ++bytestate;
- continue;
- }
- if (mime_b64[*ipos] == 128) /* skip bad characters */
- {
- mime_set_anomaly(MIME_ANOMALY_BROKEN_BASE64);
- continue;
- }
-
- /* simple state-machine */
- switch((bytestate++) & 3)
- {
- case 0:
- *opos = mime_b64[*ipos] << 2;
- break;
- case 1:
- *opos |= mime_b64[*ipos] >> 4;
- ++opos;
- *opos = mime_b64[*ipos] << 4;
- break;
- case 2:
- *opos |= mime_b64[*ipos] >> 2;
- ++opos;
- *opos = mime_b64[*ipos] << 6;
- break;
- case 3:
- *opos |= mime_b64[*ipos];
- ++opos;
- break;
- } /* switch */
- } /* for */
-
- /* something to write? */
- len = opos - obuf;
- if (len > 0)
- {
- if (fwrite(obuf, 1, len, out) != len) return -1; /* error */
- size += len;
- /* copy incomplete last byte to start of obuf, where we continue */
- if ((bytestate & 3) != 0)
- *obuf = *opos;
- opos = obuf;
- }
- } /* while */
-
- /* write out last byte if it was incomplete */
- if (bytestate & 3)
- {
- if (fwrite(obuf, 1, 1, out) != 1) return -1;
- ++size;
- }
-
- return size;
-}
-
/* decode quoted-printable MIME part */
static ssize_t
{
/* Error from decoder. ipos is unchanged. */
mime_set_anomaly(MIME_ANOMALY_BROKEN_QP);
- *opos = '=';
- ++opos;
+ *opos++ = '=';
++ipos;
}
else if (decode_qp_result == -1)
break;
else if (decode_qp_result >= 0)
- {
- *opos = decode_qp_result;
- ++opos;
- }
+ *opos++ = decode_qp_result;
}
else
- {
- *opos = *ipos;
- ++opos;
- ++ipos;
- }
+ *opos++ = *ipos++;
}
/* something to write? */
len = opos - obuf;
}
+/*
+ * Return open filehandle for combo of path and file.
+ * Side-effect: set mime_decoded_filename, to copy in allocated mem
+ */
static FILE *
mime_get_decode_file(uschar *pname, uschar *fname)
{
-FILE *f = NULL;
-uschar *filename;
-
-filename = (uschar *)malloc(2048);
-
if (pname && fname)
- {
- (void)string_format(filename, 2048, "%s/%s", pname, fname);
- f = modefopen(filename,"wb+",SPOOL_MODE);
- }
+ mime_decoded_filename = string_sprintf("%s/%s", pname, fname);
else if (!pname)
- f = modefopen(fname,"wb+",SPOOL_MODE);
+ mime_decoded_filename = string_copy(fname);
else if (!fname)
{
int file_nr = 0;
do
{
struct stat mystat;
- (void)string_format(filename, 2048,
- "%s/%s-%05u", pname, message_id, file_nr++);
+ mime_decoded_filename = string_sprintf("%s/%s-%05u", pname, message_id, file_nr++);
/* security break */
if (file_nr >= 1024)
break;
- result = stat(CS filename, &mystat);
+ result = stat(CS mime_decoded_filename, &mystat);
} while(result != -1);
-
- f = modefopen(filename, "wb+", SPOOL_MODE);
}
-/* set expansion variable */
-mime_decoded_filename = filename;
-
-return f;
+return modefopen(mime_decoded_filename, "wb+", SPOOL_MODE);
}
{
int sep = 0;
const uschar *list = *listptr;
-uschar *option;
-uschar option_buffer[1024];
-uschar decode_path[1024];
+uschar * option;
+uschar * decode_path;
FILE *decode_file = NULL;
long f_pos = 0;
ssize_t size_counter = 0;
ssize_t (*decode_function)(FILE*, FILE*, uschar*);
-if (mime_stream == NULL)
+if (!mime_stream || (f_pos = ftell(mime_stream)) < 0)
return FAIL;
-f_pos = ftell(mime_stream);
-
/* build default decode path (will exist since MBOX must be spooled up) */
-(void)string_format(decode_path,1024,"%s/scan/%s",spool_directory,message_id);
+decode_path = string_sprintf("%s/scan/%s", spool_directory, message_id);
/* try to find 1st option */
-if ((option = string_nextinlist(&list, &sep,
- option_buffer,
- sizeof(option_buffer))) != NULL)
+if ((option = string_nextinlist(&list, &sep, NULL, 0)))
{
/* parse 1st option */
- if ( (Ustrcmp(option,"false") == 0) || (Ustrcmp(option,"0") == 0) )
+ if ((Ustrcmp(option,"false") == 0) || (Ustrcmp(option,"0") == 0))
/* explicitly no decoding */
return FAIL;
size_counter = decode_function(mime_stream, decode_file, mime_current_boundary);
clearerr(mime_stream);
-fseek(mime_stream, f_pos, SEEK_SET);
+if (fseek(mime_stream, f_pos, SEEK_SET))
+ return DEFER;
if (fclose(decode_file) != 0 || size_counter < 0)
return DEFER;
if ( ((c == '\t') || (c == ' ')) && (header_value_mode == 1) )
continue;
- /* we have hit a non-whitespace char, start copying value data */
- header_value_mode = 2;
+ /* we have hit a non-whitespace char, start copying value data */
+ header_value_mode = 2;
- if (c == '"') /* flip "quoted" mode */
- header_value_mode = header_value_mode==2 ? 3 : 2;
+ if (c == '"') /* flip "quoted" mode */
+ header_value_mode = header_value_mode==2 ? 3 : 2;
- /* leave value mode on unquoted ';' */
- if (header_value_mode == 2 && c == ';') {
- header_value_mode = 0;
- };
- /* -------------------------------- */
+ /* leave value mode on unquoted ';' */
+ if (header_value_mode == 2 && c == ';')
+ header_value_mode = 0;
+ /* -------------------------------- */
}
else
{
header[num_copied] = '\0';
/* return 0 for EOF or empty line */
-if ((c == EOF) || (num_copied == 1))
- return 0;
-else
- return 1;
+return c == EOF || num_copied == 1 ? 0 : 1;
}
+/* reset all per-part mime variables */
static void
mime_vars_reset(void)
{
mime_param_val(uschar ** sp)
{
uschar * s = *sp;
-uschar * val = NULL;
-int size = 0, ptr = 0;
+gstring * val = NULL;
-/* debug_printf(" considering paramval '%s'\n", s); */
+/* debug_printf_indent(" considering paramval '%s'\n", s); */
while (*s && *s != ';') /* ; terminates */
if (*s == '"')
{
s++; /* skip opening " */
while (*s && *s != '"') /* " protects ; */
- val = string_cat(val, &size, &ptr, s++, 1);
+ val = string_catn(val, s++, 1);
if (*s) s++; /* skip closing " */
}
else
- val = string_cat(val, &size, &ptr, s++, 1);
-if (val) val[ptr] = '\0';
+ val = string_catn(val, s++, 1);
*sp = s;
-return val;
+return string_from_gstring(val);
}
static uschar *
static uschar *
rfc2231_to_2047(const uschar * fname, const uschar * charset, int * len)
{
-int size = 0, ptr = 0;
-uschar * val = string_cat(NULL, &size, &ptr, US"=?", 2);
+gstring * val = string_catn(NULL, US"=?", 2);
uschar c;
-val = string_cat(val, &size, &ptr, charset, Ustrlen(charset));
-val = string_cat(val, &size, &ptr, US"?Q?", 3);
+if (charset)
+ val = string_cat(val, charset);
+val = string_catn(val, US"?Q?", 3);
while ((c = *fname))
if (c == '%' && isxdigit(fname[1]) && isxdigit(fname[2]))
{
- val = string_cat(val, &size, &ptr, US"=", 1);
- val = string_cat(val, &size, &ptr, ++fname, 2);
+ val = string_catn(val, US"=", 1);
+ val = string_catn(val, ++fname, 2);
fname += 2;
}
else
- val = string_cat(val, &size, &ptr, fname++, 1);
+ val = string_catn(val, fname++, 1);
-val = string_cat(val, &size, &ptr, US"?=", 2);
-val[*len = ptr] = '\0';
-return val;
+val = string_catn(val, US"?=", 2);
+*len = gstring_length(val);
+return string_from_gstring(val);
}
uschar * header = NULL;
struct mime_boundary_context nested_context;
-/* reserve a line buffer to work in */
-header = store_get(MIME_MAX_HEADER_SIZE+1);
+/* reserve a line buffer to work in. Assume tainted data. */
+header = store_get(MIME_MAX_HEADER_SIZE+1, GET_TAINTED);
/* Not actually used at the moment, but will be vital to fixing
* some RFC 2046 nonconformance later... */
if (!fgets(CS header, MIME_MAX_HEADER_SIZE, f))
{
/* Hit EOF or read error. Ugh. */
- DEBUG(D_acl) debug_printf("Hit EOF ...\n");
+ DEBUG(D_acl) debug_printf_indent("MIME: Hit EOF ...\n");
return rc;
}
if (Ustrncmp((header+2+Ustrlen(context->boundary)), "--", 2) == 0)
{
/* END boundary found */
- DEBUG(D_acl) debug_printf("End boundary found %s\n",
+ DEBUG(D_acl) debug_printf_indent("MIME: End boundary found %s\n",
context->boundary);
return rc;
}
- DEBUG(D_acl) debug_printf("Next part with boundary %s\n",
+ DEBUG(D_acl) debug_printf_indent("MIME: Next part with boundary %s\n",
context->boundary);
break;
}
/* parse headers, set up expansion variables */
while (mime_get_header(f, header))
- {
- struct mime_header * mh;
/* look for interesting headers */
- for (mh = mime_header_list;
+ for (struct mime_header * mh = mime_header_list;
mh < mime_header_list + mime_header_list_size;
mh++) if (strncmpic(mh->name, header, mh->namelen) == 0)
{
- uschar * header_value = NULL;
- int header_value_len = 0;
uschar * p = header + mh->namelen;
uschar * q;
for (q = p; *q != ';' && *q; q++) ;
*mh->value = string_copynlc(p, q-p);
- DEBUG(D_acl) debug_printf("found %s MIME header, value is '%s'\n",
+ DEBUG(D_acl) debug_printf_indent("MIME: found %s header, value is '%s'\n",
mh->name, *mh->value);
if (*(p = q)) p++; /* jump past the ; */
while (*p)
{
- mime_parameter * mp;
-
- DEBUG(D_acl) debug_printf(" considering paramlist '%s'\n", p);
+ DEBUG(D_acl) debug_printf_indent("MIME: considering paramlist '%s'\n", p);
if ( !mime_filename
- && strncmpic("content-disposition:", header, 20) == 0
- && strncmpic("filename*", p, 9) == 0
+ && strncmpic(CUS"content-disposition:", header, 20) == 0
+ && strncmpic(CUS"filename*", p, 9) == 0
)
{ /* RFC 2231 filename */
uschar * q;
uschar * s = q;
/* look for a ' in the "filename" */
- while(*s != '\'' && *s) s++; /* s is ' or NUL */
+ while(*s != '\'' && *s) s++; /* s is 1st ' or NUL */
if ((size = s-q) > 0)
- {
mime_filename_charset = string_copyn(q, size);
- p = s;
- while(*p == '\'' && *p) p++; /* p is after ' */
- }
+ if (*(p = s)) p++;
+ while(*p == '\'') p++; /* p is after 2nd ' */
}
else
p = q;
+ DEBUG(D_acl) debug_printf_indent("MIME: charset %s fname '%s'\n",
+ mime_filename_charset ? mime_filename_charset : US"<NULL>", p);
+
temp_string = rfc2231_to_2047(p, mime_filename_charset, &slen);
- temp_string = rfc2047_decode(temp_string, FALSE, NULL, 32,
+ DEBUG(D_acl) debug_printf_indent("MIME: 2047-name %s\n", temp_string);
+
+ temp_string = rfc2047_decode(temp_string, FALSE, NULL, ' ',
NULL, &err_msg);
- size = Ustrlen(temp_string);
+ DEBUG(D_acl) debug_printf_indent("MIME: plain-name %s\n", temp_string);
- if (size == slen)
+ if (!temp_string || (size = Ustrlen(temp_string)) == slen)
decoding_failed = TRUE;
else
/* build up a decoded filename over successive
else
/* look for interesting parameters */
- for (mp = mime_parameter_list;
+ for (mime_parameter * mp = mime_parameter_list;
mp < mime_parameter_list + nelem(mime_parameter_list);
mp++
) if (strncmpic(mp->name, p, mp->namelen) == 0)
? rfc2047_decode(q, check_rfc2047_length, NULL, 32, NULL,
&dummy_errstr)
: NULL;
- DEBUG(D_acl) debug_printf(
- " found %s MIME parameter in %s header, value '%s'\n",
+ DEBUG(D_acl) debug_printf_indent(
+ "MIME: found %s parameter in %s header, value '%s'\n",
mp->name, mh->name, *mp->value);
break; /* done matching param names */
if (*p) p++;
} /* param scan on line */
- if (strncmpic("content-disposition:", header, 20) == 0)
+ if (strncmpic(CUS"content-disposition:", header, 20) == 0)
{
if (decoding_failed) mime_filename = mime_fname_rfc2231;
- DEBUG(D_acl) debug_printf(
- " found %s MIME parameter in %s header, value is '%s'\n",
+ DEBUG(D_acl) debug_printf_indent(
+ "MIME: found %s parameter in %s header, value is '%s'\n",
"filename", mh->name, mime_filename);
}
}
}
- }
/* set additional flag variables (easier access) */
if ( mime_content_type
if (rc != OK) break;
/* If we have a multipart entity and a boundary, go recursive */
- if ( (mime_content_type != NULL) &&
- (nested_context.boundary != NULL) &&
- (Ustrncmp(mime_content_type,"multipart",9) == 0) )
+ if ( mime_content_type && nested_context.boundary
+ && Ustrncmp(mime_content_type,"multipart",9) == 0)
{
- DEBUG(D_acl) debug_printf("Entering multipart recursion, boundary '%s'\n",
- nested_context.boundary);
+ DEBUG(D_acl)
+ debug_printf_indent("MIME: Entering multipart recursion, boundary '%s'\n",
+ nested_context.boundary);
nested_context.context =
context && context->context == MBC_ATTACHMENT
rc = mime_acl_check(acl, f, &nested_context, user_msgptr, log_msgptr);
if (rc != OK) break;
}
- else if ( (mime_content_type != NULL) &&
- (Ustrncmp(mime_content_type,"message/rfc822",14) == 0) )
+ else if ( mime_content_type
+ && Ustrncmp(mime_content_type,"message/rfc822",14) == 0)
{
- const uschar *rfc822name = NULL;
- uschar filename[2048];
+ const uschar * rfc822name = NULL;
+ uschar * filename;
int file_nr = 0;
int result = 0;
/* must find first free sequential filename */
- do
+ for (gstring * g = string_get(64); result != -1; gstring_reset(g))
{
struct stat mystat;
- (void)string_format(filename, 2048,
+ g = string_fmt_append(g,
"%s/scan/%s/__rfc822_%05u", spool_directory, message_id, file_nr++);
/* security break */
if (file_nr >= 128)
goto NO_RFC822;
- result = stat(CS filename,&mystat);
- } while (result != -1);
+ result = stat(CS (filename = string_from_gstring(g)), &mystat);
+ }
rfc822name = filename;
if (!mime_decoded_filename) /* decoding failed */
{
log_write(0, LOG_MAIN,
- "mime_regex acl condition warning - could not decode RFC822 MIME part to file.");
+ "MIME acl condition warning - could not decode RFC822 MIME part to file.");
rc = DEFER;
goto out;
}