/*
* $Id$
- *
- * This is the MIME parser for Citadel. Sometimes it actually works.
+ */
+/**
+ * \defgroup MIME This is the MIME parser for Citadel.
*
* Copyright (c) 1998-2005 by Art Cancro
* This code is distributed under the terms of the GNU General Public License.
- *
+ * \ingroup WebcitHttpServer
*/
-
+/*@{*/
#include "webcit.h"
#include "webserver.h"
#include "mime_parser.h"
-
void extract_key(char *target, char *source, char *key)
{
- int a, b;
-
- strcpy(target, source);
- for (a = 0; a < strlen(target); ++a) {
- if ((!strncasecmp(&target[a], key, strlen(key)))
- && (target[a + strlen(key)] == '=')) {
- strcpy(target, &target[a + strlen(key) + 1]);
- if (target[0] == 34)
- strcpy(target, &target[1]);
- for (b = 0; b < strlen(target); ++b)
- if (target[b] == 34)
- target[b] = 0;
- return;
+ char *ptr;
+ char looking_for[256];
+ int double_quotes = 0;
+
+ snprintf(looking_for, sizeof looking_for, "%s=", key);
+
+ ptr = bmstrcasestr(source, looking_for);
+ if (ptr == NULL) {
+ strcpy(target, "");
+ return;
+ }
+ strcpy(target, (ptr + strlen(looking_for)));
+
+ for (ptr=target; (*ptr != 0); ++ptr) {
+
+ /* A semicolon means we've hit the end of the key, unless we're inside double quotes */
+ if ( (double_quotes != 1) && (*ptr == ';')) {
+ *ptr = 0;
+ }
+
+ /* if we find double quotes, we've got a great set of string boundaries */
+ if (*ptr == '\"') {
+ ++double_quotes;
+ if (double_quotes == 1) {
+ strcpy(ptr, ptr+1);
+ }
+ else {
+ *ptr = 0;
+ }
}
}
- strcpy(target, "");
}
-/*
- * Convert "quoted-printable" to binary. Returns number of bytes decoded.
- */
-int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
- char buf[SIZ];
- int buf_length = 0;
- int soft_line_break = 0;
- unsigned int ch;
- int decoded_length = 0;
- int i;
-
- decoded[0] = 0;
- decoded_length = 0;
- buf[0] = 0;
- buf_length = 0;
-
- for (i = 0; i < sourcelen; ++i) {
-
- buf[buf_length++] = encoded[i];
-
- if ( (encoded[i] == '\n')
- || (encoded[i] == 0)
- || (i == (sourcelen-1)) ) {
- buf[buf_length++] = 0;
-
- /*** begin -- process one line ***/
-
- if (buf[strlen(buf)-1] == '\n') {
- buf[strlen(buf)-1] = 0;
- }
- if (buf[strlen(buf)-1] == '\r') {
- buf[strlen(buf)-1] = 0;
- }
- while (isspace(buf[strlen(buf)-1])) {
- buf[strlen(buf)-1] = 0;
- }
- soft_line_break = 0;
-
- while (strlen(buf) > 0) {
- if (!strcmp(buf, "=")) {
- soft_line_break = 1;
- strcpy(buf, "");
- } else if ((strlen(buf)>=3) && (buf[0]=='=')) {
- sscanf(&buf[1], "%02x", &ch);
- decoded[decoded_length++] = ch;
- strcpy(buf, &buf[3]);
- } else {
- decoded[decoded_length++] = buf[0];
- strcpy(buf, &buf[1]);
- }
- }
- if (soft_line_break == 0) {
- decoded[decoded_length++] = '\r';
- decoded[decoded_length++] = '\n';
- }
- buf_length = 0;
- /*** end -- process one line ***/
- }
- }
-
- decoded[decoded_length++] = 0;
- return(decoded_length);
-}
-
/*
* Given a message or message-part body and a length, handle any necessary
* decoding and pass the request up the stack.
return;
}
+ /* Fail silently if we hit an unknown encoding. */
if ((strcasecmp(encoding, "base64"))
&& (strcasecmp(encoding, "quoted-printable"))) {
return;
}
+
/*
- * Allocate a buffer for the decoded data. The output buffer is the
- * same size as the input buffer; this assumes that the decoded data
- * will never be larger than the encoded data. This is a safe
- * assumption with base64, uuencode, and quoted-printable.
+ * Allocate a buffer for the decoded data. The output buffer is slightly
+ * larger than the input buffer; this assumes that the decoded data
+ * will never be significantly larger than the encoded data. This is a
+ * safe assumption with base64, uuencode, and quoted-printable.
*/
- decoded = malloc(length+2048);
+ decoded = malloc(length + 32768);
if (decoded == NULL) {
return;
}
bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
}
else if (!strcasecmp(encoding, "quoted-printable")) {
- bytes_decoded = CtdlDecodeQuotedPrintable(decoded,
- part_start, length);
+ bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
}
if (bytes_decoded > 0) if (CallBack != NULL) {
int part_seq = 0;
int i;
size_t length;
- char nested_partnum[SIZ];
+ char nested_partnum[256];
+ int crlf_in_use = 0;
+ char *evaluate_crlf_ptr = NULL;
+ int buflen = 0;
+ int headerlen = 0;
ptr = content_start;
content_length = 0;
/* Learn interesting things from the headers */
strcpy(header, "");
+ headerlen = 0;
do {
- ptr = memreadline(ptr, buf, SIZ);
+ ptr = memreadlinelen(ptr, buf, SIZ, &buflen);
if (ptr >= content_end) {
goto end_parser;
}
- for (i = 0; i < strlen(buf); ++i) {
+ for (i = 0; i < buflen; ++i) {
if (isspace(buf[i])) {
buf[i] = ' ';
}
}
if (!isspace(buf[0])) {
- if (!strncasecmp(header, "Content-type: ", 14)) {
- strcpy(content_type, &header[14]);
+ if (!strncasecmp(header, "Content-type:", 13)) {
+ strcpy(content_type, &header[13]);
+ striplt(content_type);
extract_key(content_type_name, content_type, "name");
extract_key(charset, content_type, "charset");
+ extract_key(boundary, header, "boundary");
/* Deal with weird headers */
if (strchr(content_type, ' '))
*(strchr(content_type, ' ')) = '\0';
if (strchr(content_type, ';'))
*(strchr(content_type, ';')) = '\0';
}
- if (!strncasecmp(header, "Content-Disposition: ", 21)) {
- strcpy(disposition, &header[21]);
+ if (!strncasecmp(header, "Content-Disposition:", 20)) {
+ strcpy(disposition, &header[20]);
+ striplt(disposition);
extract_key(content_disposition_name, disposition, "name");
extract_key(filename, disposition, "filename");
}
- if (!strncasecmp(header, "Content-length: ", 16)) {
- content_length = (size_t) atol(&header[16]);
+ if (!strncasecmp(header, "Content-length: ", 15)) {
+ char clbuf[10];
+ safestrncpy(clbuf, &header[15], sizeof clbuf);
+ striplt(clbuf);
+ content_length = (size_t) atol(clbuf);
+ }
+ if (!strncasecmp(header, "Content-transfer-encoding: ", 26)) {
+ strcpy(encoding, &header[26]);
+ striplt(encoding);
}
- if (!strncasecmp(header,
- "Content-transfer-encoding: ", 27))
- strcpy(encoding, &header[27]);
- if (strlen(boundary) == 0)
- extract_key(boundary, header, "boundary");
strcpy(header, "");
+ headerlen = 0;
+ }
+ if ((headerlen + buflen + 2) < SIZ) {
+ memcpy(&header[headerlen], buf, buflen);
+ headerlen += buflen;
+ header[headerlen] = '\0';
}
- if ((strlen(header) + strlen(buf) + 2) < SIZ)
- strcat(header, buf);
- } while ((strlen(buf) > 0) && (*ptr != 0));
+ } while ((!IsEmptyStr(buf)) && (*ptr != 0));
if (strchr(disposition, ';'))
*(strchr(disposition, ';')) = '\0';
*(strchr(content_type, ';')) = '\0';
striplt(content_type);
- if (strlen(boundary) > 0) {
+ if (!IsEmptyStr(boundary)) {
is_multipart = 1;
} else {
is_multipart = 0;
part_start = NULL;
do {
-
- /* next_boundary = bmstrstr(ptr, startary, memcmp); */
next_boundary = NULL;
for (srch=ptr; srch<content_end; ++srch) {
if (!memcmp(srch, startary, startary_len)) {
if ( (part_start != NULL) && (next_boundary != NULL) ) {
part_end = next_boundary;
- --part_end;
+ --part_end; /* omit the trailing LF */
+ if (crlf_in_use) {
+ --part_end; /* omit the trailing CR */
+ }
- if (strlen(partnum) > 0) {
+ if (!IsEmptyStr(partnum)) {
snprintf(nested_partnum,
sizeof nested_partnum,
"%s.%d", partnum,
else {
/* Set up for the next part. */
part_start = strstr(next_boundary, "\n");
+
+ /* Determine whether newlines are LF or CRLF */
+ evaluate_crlf_ptr = part_start;
+ --evaluate_crlf_ptr;
+ if (!memcmp(evaluate_crlf_ptr, "\r\n", 2)) {
+ crlf_in_use = 1;
+ }
+ else {
+ crlf_in_use = 0;
+ }
+
+ /* Advance past the LF ... now we're in the next part */
++part_start;
ptr = part_start;
}
++length;
}
part_end = content_end;
- /* fix an off-by-one error */
- --part_end;
- --length;
+
+ /******
+ * I thought there was an off-by-one error here, but there isn't.
+ * This probably means that there's an off-by-one error somewhere
+ * else ... or maybe only in certain messages?
+ --part_end;
+ --length;
+ ******/
/* Truncate if the header told us to */
if ( (content_length > 0) && (length > content_length) ) {
else {
name = content_type_name;
}
-
+
+ /* lprintf(CTDL_DEBUG, "mime_decode part=%s, len=%d, type=%s, charset=%s, encoding=%s\n",
+ partnum, length, content_type, charset, encoding); */
+
+ /* Ok, we've got a non-multipart part here, so do something with it.
+ */
mime_decode(partnum,
- part_start, length,
- content_type, charset, encoding, disposition,
- name, filename,
- CallBack, NULL, NULL,
- userdata, dont_decode);
+ part_start, length,
+ content_type, charset, encoding, disposition,
+ name, filename,
+ CallBack, NULL, NULL,
+ userdata, dont_decode
+ );
+
+ /*
+ * Now if it's an encapsulated message/rfc822 then we have to recurse into it
+ */
+ if (!strcasecmp(content_type, "message/rfc822")) {
+
+ if (PreMultiPartCallBack != NULL) {
+ PreMultiPartCallBack("", "", partnum, "",
+ NULL, content_type, charset,
+ 0, encoding, userdata);
+ }
+ if (CallBack != NULL) {
+ if (strlen(partnum) > 0) {
+ snprintf(nested_partnum,
+ sizeof nested_partnum,
+ "%s.%d", partnum,
+ ++part_seq);
+ }
+ else {
+ snprintf(nested_partnum,
+ sizeof nested_partnum,
+ "%d", ++part_seq);
+ }
+ the_mime_parser(nested_partnum,
+ part_start, part_end,
+ CallBack,
+ PreMultiPartCallBack,
+ PostMultiPartCallBack,
+ userdata,
+ dont_decode
+ );
+ }
+ if (PostMultiPartCallBack != NULL) {
+ PostMultiPartCallBack("", "", partnum, "", NULL,
+ content_type, charset, 0, encoding, userdata);
+ }
+
+
+ }
+
}
end_parser: /* free the buffers! end the oppression!! */