/*
* $Id$
*
- * This is the MIME parser for Citadel. Sometimes it actually works.
+ * This is the MIME parser for Citadel.
*
- * Copyright (c) 1998-2001 by Art Cancro
- * This code is distributed under the terms of the GNU General Public License.
+ * Copyright (c) 1998-2006 by Art Cancro
+ * This code is distributed under the GNU General Public License v2.
*
*/
-#ifdef DLL_EXPORT
-#define IN_LIBCIT
-#endif
-
#include <stdlib.h>
#include <unistd.h>
#include <stdio.h>
#include <string.h>
#include <sys/stat.h>
#include <errno.h>
+
#include "citadel.h"
#include "server.h"
-#include "dynloader.h"
#include "sysdep_decls.h"
-#include "mime_parser.h"
#include "tools.h"
+#include "mime_parser.h"
+
void extract_key(char *target, char *source, char *key)
{
/*
* Convert "quoted-printable" to binary. Returns number of bytes decoded.
+ * according to RFC2045 section 6.7
*/
-int decode_quoted_printable(char *decoded, char *encoded, int sourcelen) {
- char buf[SIZ];
- int buf_length = 0;
- int soft_line_break = 0;
- int ch;
+int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
+ unsigned int ch;
int decoded_length = 0;
- int i;
-
- decoded[0] = 0;
- decoded_length = 0;
- buf[0] = 0;
- buf_length = 0;
+ int pos = 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 ***/
+ while (pos < sourcelen)
+ {
+ if (!strncmp(&encoded[pos], "=\r\n", 3))
+ {
+ pos += 3;
+ }
+ else if (!strncmp(&encoded[pos], "=\n", 2))
+ {
+ pos += 2;
+ }
+ else if (encoded[pos] == '=')
+ {
+ ch = 0;
+ sscanf(&encoded[pos+1], "%02x", &ch);
+ pos += 3;
+ decoded[decoded_length++] = ch;
+ }
+ else
+ {
+ decoded[decoded_length++] = encoded[pos];
+ pos += 1;
}
}
-
- decoded[decoded_length++] = 0;
+ 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.
*/
void mime_decode(char *partnum,
char *part_start, size_t length,
- char *content_type, char *encoding,
+ char *content_type, char *charset, char *encoding,
char *disposition,
char *name, char *filename,
void (*CallBack)
char *cbdisp,
void *cbcontent,
char *cbtype,
+ char *cbcharset,
size_t cblength,
char *cbencoding,
void *cbuserdata),
char *cbdisp,
void *cbcontent,
char *cbtype,
+ char *cbcharset,
size_t cblength,
char *cbencoding,
void *cbuserdata),
char *cbdisp,
void *cbcontent,
char *cbtype,
+ char *cbcharset,
size_t cblength,
char *cbencoding,
void *cbuserdata),
char *decoded;
size_t bytes_decoded = 0;
- lprintf(9, "mime_decode() called\n");
-
/* Some encodings aren't really encodings */
if (!strcasecmp(encoding, "7bit"))
strcpy(encoding, "");
if (CallBack != NULL) {
CallBack(name, filename, fixed_partnum(partnum),
disposition, part_start,
- content_type, length, encoding, userdata);
+ content_type, charset, length, encoding, userdata);
}
return;
}
+ /* Fail silently if we hit an unknown encoding. */
if ((strcasecmp(encoding, "base64"))
&& (strcasecmp(encoding, "quoted-printable"))) {
- lprintf(9, "ERROR: unknown MIME encoding '%s'\n", encoding);
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.
*/
- lprintf(9, "About to allocate %d bytes for decoded part\n",
- length+2048);
- decoded = mallok(length+2048);
+ decoded = malloc(length + 32768);
if (decoded == NULL) {
- lprintf(9, "ERROR: cannot allocate memory.\n");
return;
}
- lprintf(9, "Got it!\n");
- lprintf(9, "Decoding %s\n", encoding);
if (!strcasecmp(encoding, "base64")) {
- bytes_decoded = decode_base64(decoded, part_start, length);
+ bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
}
else if (!strcasecmp(encoding, "quoted-printable")) {
- bytes_decoded = decode_quoted_printable(decoded,
- part_start, length);
+ bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
}
- lprintf(9, "Bytes decoded: %d\n", bytes_decoded);
if (bytes_decoded > 0) if (CallBack != NULL) {
CallBack(name, filename, fixed_partnum(partnum),
disposition, decoded,
- content_type, bytes_decoded, "binary", userdata);
+ content_type, charset, bytes_decoded, "binary", userdata);
}
- phree(decoded);
+ free(decoded);
}
/*
char *cbdisp,
void *cbcontent,
char *cbtype,
+ char *cbcharset,
size_t cblength,
char *cbencoding,
void *cbuserdata),
char *cbdisp,
void *cbcontent,
char *cbtype,
+ char *cbcharset,
size_t cblength,
char *cbencoding,
void *cbuserdata),
char *cbdisp,
void *cbcontent,
char *cbtype,
+ char *cbcharset,
size_t cblength,
char *cbencoding,
void *cbuserdata),
{
char *ptr;
+ char *srch = NULL;
char *part_start, *part_end = NULL;
char buf[SIZ];
char *header;
char *boundary;
char *startary;
+ size_t startary_len = 0;
char *endary;
+ char *next_boundary;
char *content_type;
+ char *charset;
size_t content_length;
char *encoding;
char *disposition;
- char *name;
+ char *name = NULL;
+ char *content_type_name;
+ char *content_disposition_name;
char *filename;
int is_multipart;
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;
- lprintf(9, "the_mime_parser() called\n");
ptr = content_start;
content_length = 0;
- boundary = mallok(SIZ);
+ boundary = malloc(SIZ);
memset(boundary, 0, SIZ);
- startary = mallok(SIZ);
+ startary = malloc(SIZ);
memset(startary, 0, SIZ);
- endary = mallok(SIZ);
+ endary = malloc(SIZ);
memset(endary, 0, SIZ);
- header = mallok(SIZ);
+ header = malloc(SIZ);
memset(header, 0, SIZ);
- content_type = mallok(SIZ);
+ content_type = malloc(SIZ);
memset(content_type, 0, SIZ);
- encoding = mallok(SIZ);
+ charset = malloc(SIZ);
+ memset(charset, 0, SIZ);
+
+ encoding = malloc(SIZ);
memset(encoding, 0, SIZ);
- name = mallok(SIZ);
- memset(name, 0, SIZ);
+ content_type_name = malloc(SIZ);
+ memset(content_type_name, 0, SIZ);
- filename = mallok(SIZ);
+ content_disposition_name = malloc(SIZ);
+ memset(content_disposition_name, 0, SIZ);
+
+ filename = malloc(SIZ);
memset(filename, 0, SIZ);
- disposition = mallok(SIZ);
+ disposition = malloc(SIZ);
memset(disposition, 0, SIZ);
/* If the caller didn't supply an endpointer, generate one by measure */
goto end_parser;
}
- for (i = 0; i < strlen(buf); ++i)
- if (isspace(buf[i]))
+ for (i = 0; i < strlen(buf); ++i) {
+ if (isspace(buf[i])) {
buf[i] = ' ';
+ }
+ }
+
if (!isspace(buf[0])) {
- if (!strncasecmp(header, "Content-type: ", 14)) {
- strcpy(content_type, &header[14]);
- extract_key(name, content_type, "name");
- lprintf(9, "Extracted content-type <%s>\n",
- content_type);
+ 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");
+ /* 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, "");
}
- if ((strlen(header) + strlen(buf) + 2) < SIZ)
+ if ((strlen(header) + strlen(buf) + 2) < SIZ) {
strcat(header, buf);
+ }
} while ((strlen(buf) > 0) && (*ptr != 0));
- for (i = 0; i < strlen(disposition); ++i)
- if (disposition[i] == ';')
- disposition[i] = 0;
- while (isspace(disposition[0]))
- strcpy(disposition, &disposition[1]);
- for (i = 0; i < strlen(content_type); ++i)
- if (content_type[i] == ';')
- content_type[i] = 0;
- while (isspace(content_type[0]))
- strcpy(content_type, &content_type[1]);
+ if (strchr(disposition, ';'))
+ *(strchr(disposition, ';')) = '\0';
+ striplt(disposition);
+ if (strchr(content_type, ';'))
+ *(strchr(content_type, ';')) = '\0';
+ striplt(content_type);
if (strlen(boundary) > 0) {
is_multipart = 1;
is_multipart = 0;
}
- lprintf(9, "is_multipart=%d, boundary=<%s>\n",
- is_multipart, boundary);
-
/* If this is a multipart message, then recursively process it */
part_start = NULL;
if (is_multipart) {
/* Tell the client about this message's multipartedness */
if (PreMultiPartCallBack != NULL) {
PreMultiPartCallBack("", "", partnum, "",
- NULL, content_type,
+ NULL, content_type, charset,
0, encoding, userdata);
}
/* Figure out where the boundaries are */
snprintf(startary, SIZ, "--%s", boundary);
snprintf(endary, SIZ, "--%s--", boundary);
+ startary_len = strlen(startary);
+
+ part_start = NULL;
do {
- if ( (!strncasecmp(ptr, startary, strlen(startary)))
- || (!strncasecmp(ptr, endary, strlen(endary))) ) {
- lprintf(9, "hit boundary!\n");
- if (part_start != NULL) {
- if (strlen(partnum) > 0) {
- snprintf(nested_partnum,
- sizeof nested_partnum,
- "%s.%d", partnum,
- ++part_seq);
+ next_boundary = NULL;
+ for (srch=ptr; srch<content_end; ++srch) {
+ if (!memcmp(srch, startary, startary_len)) {
+ next_boundary = srch;
+ srch = content_end;
+ }
+ }
+
+ if ( (part_start != NULL) && (next_boundary != NULL) ) {
+ part_end = next_boundary;
+ --part_end; /* omit the trailing LF */
+ if (crlf_in_use) {
+ --part_end; /* omit the trailing CR */
+ }
+
+ 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 (next_boundary != NULL) {
+ /* If we pass out of scope, don't attempt to
+ * read past the end boundary. */
+ if (!strcmp(next_boundary, endary)) {
+ ptr = content_end;
+ }
+ 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 {
- snprintf(nested_partnum,
- sizeof nested_partnum,
- "%d", ++part_seq);
+ crlf_in_use = 0;
}
- the_mime_parser(nested_partnum,
- part_start, part_end,
- CallBack,
- PreMultiPartCallBack,
- PostMultiPartCallBack,
- userdata,
- dont_decode);
+
+ /* Advance past the LF ... now we're in the next part */
+ ++part_start;
+ ptr = part_start;
}
- ptr = memreadline(ptr, buf, SIZ);
- part_start = ptr;
}
else {
- part_end = ptr;
- ++ptr;
+ /* Invalid end of multipart. Bail out! */
+ ptr = content_end;
}
- } while ( (strcasecmp(ptr, endary)) && (ptr <= content_end) );
+ } while ( (ptr < content_end) && (next_boundary != NULL) );
+
if (PostMultiPartCallBack != NULL) {
PostMultiPartCallBack("", "", partnum, "", NULL,
- content_type, 0, encoding, userdata);
+ content_type, charset, 0, encoding, userdata);
}
goto end_parser;
}
/* If it's not a multipart message, then do something with it */
if (!is_multipart) {
- lprintf(9, "doing non-multipart thing\n");
part_start = ptr;
length = 0;
while (ptr < content_end) {
++length;
}
part_end = content_end;
+
+ /******
+ * 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) ) {
length = content_length;
- lprintf(9, "truncated to %ld\n", (long)content_length);
}
-
+
+ /* Sometimes the "name" field is tacked on to Content-type,
+ * and sometimes it's tacked on to Content-disposition. Use
+ * whichever one we have.
+ */
+ if (strlen(content_disposition_name) > strlen(content_type_name)) {
+ name = content_disposition_name;
+ }
+ 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, 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!! */
- phree(boundary);
- phree(startary);
- phree(endary);
- phree(header);
- phree(content_type);
- phree(encoding);
- phree(name);
- phree(filename);
- phree(disposition);
+ free(boundary);
+ free(startary);
+ free(endary);
+ free(header);
+ free(content_type);
+ free(charset);
+ free(encoding);
+ free(content_type_name);
+ free(content_disposition_name);
+ free(filename);
+ free(disposition);
}
char *cbdisp,
void *cbcontent,
char *cbtype,
+ char *cbcharset,
size_t cblength,
char *cbencoding,
void *cbuserdata),
char *cbdisp,
void *cbcontent,
char *cbtype,
+ char *cbcharset,
size_t cblength,
char *cbencoding,
void *cbuserdata),
char *cbdisp,
void *cbcontent,
char *cbtype,
+ char *cbcharset,
size_t cblength,
char *cbencoding,
void *cbuserdata),
)
{
- lprintf(9, "mime_parser() called\n");
the_mime_parser("", content_start, content_end,
CallBack,
PreMultiPartCallBack,