/*
- * mime_parser.c
+ * $Id$
*
- * This is a really bad attempt at writing a parser to handle MIME-encoded
- * messages.
+ * This is the MIME parser for Citadel. Sometimes it actually works.
*
- * Copyright (c) 1998-1999 by Art Cancro
+ * Copyright (c) 1998-2001 by Art Cancro
* This code is distributed under the terms of the GNU General Public License.
*
*/
+#ifdef DLL_EXPORT
+#define IN_LIBCIT
+#endif
+
#include <stdlib.h>
#include <unistd.h>
-#include <fcntl.h> /* stu */
#include <stdio.h>
#include <signal.h>
#include <sys/types.h>
#include <ctype.h>
#include <string.h>
#include <sys/stat.h>
-#include <pthread.h>
+#include <errno.h>
#include "citadel.h"
-#include "mime_parser.h"
-#include "sysdep_decls.h"
#include "server.h"
+#include "dynloader.h"
+#include "sysdep_decls.h"
+#include "mime_parser.h"
+#include "tools.h"
-
-void extract_key(char *target, char *source, char *key) {
+void extract_key(char *target, char *source, char *key)
+{
int a, b;
strcpy(target, source);
- for (a=0; a<strlen(target); ++a) {
+ 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;
+ && (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;
- }
}
- strcpy(target, "");
}
+ strcpy(target, "");
+}
-
-/*
- * Utility function to "readline" from memory
- * (returns new pointer)
+/*
+ * For non-multipart messages, we need to generate a quickie partnum of "1"
+ * to return to callback functions. Some callbacks demand it.
*/
-char *memreadline(char *start, char *buf, int maxlen) {
- char ch;
- char *ptr;
-
- ptr = start;
- memset(buf, 0, maxlen);
+char *fixed_partnum(char *supplied_partnum) {
+ if (supplied_partnum == NULL) return "1";
+ if (strlen(supplied_partnum)==0) return "1";
+ return supplied_partnum;
+}
- while(1) {
- ch = *ptr++;
- if ((ch==10)||(ch==0)) {
- if (strlen(buf)>0)
- if (buf[strlen(buf)-1]==13)
- buf[strlen(buf)-1] = 0;
- return ptr;
- }
- if (strlen(buf) < (maxlen-1)) {
- buf[strlen(buf)+1] = 0;
- buf[strlen(buf)] = ch;
- }
- }
- }
/*
* 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 *name, char *filename,
- void (*CallBack)
- (char *cbname,
- char *cbfilename,
- char *cbpartnum,
- void *cbcontent,
- char *cbtype,
- size_t cblength)
- ) {
+ char *part_start, size_t length,
+ char *content_type, char *encoding,
+ char *disposition,
+ char *name, char *filename,
+ void (*CallBack)
+ (char *cbname,
+ char *cbfilename,
+ char *cbpartnum,
+ char *cbdisp,
+ void *cbcontent,
+ char *cbtype,
+ size_t cblength,
+ char *cbencoding,
+ void *cbuserdata),
+ void (*PreMultiPartCallBack)
+ (char *cbname,
+ char *cbfilename,
+ char *cbpartnum,
+ char *cbdisp,
+ void *cbcontent,
+ char *cbtype,
+ size_t cblength,
+ char *cbencoding,
+ void *cbuserdata),
+ void (*PostMultiPartCallBack)
+ (char *cbname,
+ char *cbfilename,
+ char *cbpartnum,
+ char *cbdisp,
+ void *cbcontent,
+ char *cbtype,
+ size_t cblength,
+ char *cbencoding,
+ void *cbuserdata),
+ void *userdata,
+ int dont_decode
+)
+{
char *decoded;
struct stat statbuf;
size_t blocksize;
int write_error = 0;
+ lprintf(9, "mime_decode() called\n");
- int stu;
- int stu2 = 0;
-
+ /* Some encodings aren't really encodings */
+ if (!strcasecmp(encoding, "7bit"))
+ strcpy(encoding, "");
+ if (!strcasecmp(encoding, "8bit"))
+ strcpy(encoding, "");
+ if (!strcasecmp(encoding, "binary"))
+ strcpy(encoding, "");
/* If this part is not encoded, send as-is */
- if (strlen(encoding)==0) {
- CallBack(name, filename, partnum, part_start,
- content_type, length);
+ if ( (strlen(encoding) == 0) || (dont_decode)) {
+ if (CallBack != NULL) {
+ CallBack(name, filename, fixed_partnum(partnum),
+ disposition, part_start,
+ content_type, length, encoding, userdata);
+ }
return;
- }
-
- if ( (strcasecmp(encoding, "base64"))
- && (strcasecmp(encoding, "7bit")) ) {
- lprintf(5, "ERROR: unknown MIME encoding '%s'\n", 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.
+ * assumption with base64, uuencode, and quoted-printable. Just to
+ * be safe, we still pad the buffer a bit.
*/
- decoded = mallok(length);
+ decoded = malloc(length + 1024);
if (decoded == NULL) {
- lprintf(5, "ERROR: cannot allocate memory.\n");
+ lprintf(9, "ERROR: cannot allocate memory.\n");
+ return;
+ }
+ if (pipe(sendpipe) != 0)
+ return;
+ if (pipe(recvpipe) != 0)
return;
- }
- if (pipe(sendpipe) != 0) return;
- if (pipe(recvpipe) != 0) return;
childpid = fork();
- lprintf(9, "fork() returned %d\n", childpid);
if (childpid < 0) {
- phree(decoded);
+ free(decoded);
return;
- }
-
+ }
if (childpid == 0) {
- /* close(2); FIX uncomment this when solid */
+ close(2);
/* send stdio to the pipes */
- if (dup2(sendpipe[0], 0)<0) lprintf(5, "ERROR dup2()\n");
- if (dup2(recvpipe[1], 1)<0) lprintf(5, "ERROR dup2()\n");
- close(sendpipe[1]); /* Close the ends we're not using */
+ if (dup2(sendpipe[0], 0) < 0)
+ lprintf(9, "ERROR dup2()\n");
+ if (dup2(recvpipe[1], 1) < 0)
+ lprintf(9, "ERROR dup2()\n");
+ close(sendpipe[1]); /* Close the ends we're not using */
close(recvpipe[0]);
if (!strcasecmp(encoding, "base64"))
- execlp("/usr/local/base64/base64", "base64", "-d", NULL);
- else if (!strcasecmp(encoding, "7bit"))
- execlp("/bin/dd", "dd", NULL);
- lprintf(5, "ERROR: cannot exec decoder for %s\n", encoding);
+ execlp("./base64", "base64", "-d", NULL);
+ else if (!strcasecmp(encoding, "quoted-printable"))
+ execlp("./qpdecode", "qpdecode", NULL);
+ lprintf(9, "ERROR: cannot exec decoder for %s\n", encoding);
exit(1);
- }
-
- close(sendpipe[0]); /* Close the ends we're not using */
+ }
+ close(sendpipe[0]); /* Close the ends we're not using */
close(recvpipe[1]);
- lprintf(9, "ready to send %d bytes\n", length);
-
- while ( (bytes_sent < length) && (write_error == 0) ) {
+ while ((bytes_sent < length) && (write_error == 0)) {
/* Empty the input pipe FIRST */
- while (fstat(recvpipe[0], &statbuf), (statbuf.st_size > 0) ) {
+ while (fstat(recvpipe[0], &statbuf), (statbuf.st_size > 0)) {
blocksize = read(recvpipe[0], &decoded[bytes_recv],
- statbuf.st_size);
- if (blocksize < 0)
- lprintf(5, "ERROR: cannot read from pipe\n");
+ statbuf.st_size);
+ if (blocksize < 0)
+ lprintf(9, "ERROR: cannot read from pipe\n");
else
bytes_recv = bytes_recv + blocksize;
- }
+ }
/* Then put some data into the output pipe */
blocksize = length - bytes_sent;
- if (blocksize > 2048) blocksize = 2048;
- if (write(sendpipe[1], &part_start[bytes_sent], blocksize) <0) {
- lprintf(5, "ERROR: cannot write to pipe: %s\n",
+ if (blocksize > 2048)
+ blocksize = 2048;
+ if (write(sendpipe[1], &part_start[bytes_sent], blocksize) < 0) {
+ lprintf(9, "ERROR: cannot write to pipe: %s\n",
strerror(errno));
write_error = 1;
- }
- bytes_sent = bytes_sent + blocksize;
}
+ bytes_sent = bytes_sent + blocksize;
+ }
close(sendpipe[1]);
/* Empty the input pipe */
- while ( (blocksize = read(recvpipe[0], &decoded[bytes_recv], 1)),
- (blocksize > 0) ) {
+ while ((blocksize = read(recvpipe[0], &decoded[bytes_recv], 1)),
+ (blocksize > 0)) {
bytes_recv = bytes_recv + blocksize;
- }
-
- lprintf(9, "Decoded length = %d\n", bytes_recv);
-
- if (strlen(filename) ==0) sprintf(filename, "noname%d", stu2++);
- chdir("stu");
- stu = creat(filename, 0666);
- write(stu, decoded, bytes_recv);
- close(stu);
- chdir("..");
+ }
- if (bytes_recv > 0)
- CallBack(name, filename, partnum, decoded,
- content_type, bytes_recv);
- phree(decoded);
+ if (bytes_recv > 0) if (CallBack != NULL) {
+ CallBack(name, filename, fixed_partnum(partnum),
+ disposition, decoded,
+ content_type, bytes_recv, "binary", userdata);
}
+ free(decoded);
+}
+
/*
* Break out the components of a multipart message
* (This function expects to be fed HEADERS + CONTENT)
* considered to have ended when the parser encounters a 0x00 byte.
*/
void the_mime_parser(char *partnum,
- char *content_start, char *content_end,
- void (*CallBack)
- (char *cbname,
- char *cbfilename,
- char *cbpartnum,
- void *cbcontent,
- char *cbtype,
- size_t cblength)
- ) {
+ char *content_start, char *content_end,
+ void (*CallBack)
+ (char *cbname,
+ char *cbfilename,
+ char *cbpartnum,
+ char *cbdisp,
+ void *cbcontent,
+ char *cbtype,
+ size_t cblength,
+ char *cbencoding,
+ void *cbuserdata),
+ void (*PreMultiPartCallBack)
+ (char *cbname,
+ char *cbfilename,
+ char *cbpartnum,
+ char *cbdisp,
+ void *cbcontent,
+ char *cbtype,
+ size_t cblength,
+ char *cbencoding,
+ void *cbuserdata),
+ void (*PostMultiPartCallBack)
+ (char *cbname,
+ char *cbfilename,
+ char *cbpartnum,
+ char *cbdisp,
+ void *cbcontent,
+ char *cbtype,
+ size_t cblength,
+ char *cbencoding,
+ void *cbuserdata),
+ void *userdata,
+ int dont_decode
+)
+{
char *ptr;
- char *part_start, *part_end;
- char buf[256];
- char header[256];
- char boundary[256];
- char startary[256];
- char endary[256];
- char content_type[256];
- char encoding[256];
- char name[256];
- char filename[256];
+ char *part_start, *part_end = NULL;
+ char buf[SIZ];
+ char header[SIZ];
+ char boundary[SIZ];
+ char startary[SIZ];
+ char endary[SIZ];
+ char content_type[SIZ];
+ size_t content_length;
+ char encoding[SIZ];
+ char disposition[SIZ];
+ char name[SIZ];
+ char filename[SIZ];
int is_multipart;
int part_seq = 0;
int i;
size_t length;
- char nested_partnum[256];
+ char nested_partnum[SIZ];
+ lprintf(9, "the_mime_parser() called\n");
ptr = content_start;
memset(boundary, 0, sizeof boundary);
memset(content_type, 0, sizeof content_type);
memset(encoding, 0, sizeof encoding);
memset(name, 0, sizeof name);
memset(filename, 0, sizeof filename);
+ memset(disposition, 0, sizeof disposition);
+ content_length = 0;
+
+ /* If the caller didn't supply an endpointer, generate one by measure */
+ if (content_end == NULL) {
+ content_end = &content_start[strlen(content_start)];
+ }
/* Learn interesting things from the headers */
strcpy(header, "");
do {
ptr = memreadline(ptr, buf, sizeof buf);
- if (*ptr == 0) return; /* premature end of message */
- if (content_end != NULL)
- if (ptr >= content_end) return;
+ if (ptr >= content_end)
+ return;
- for (i=0; i<strlen(buf); ++i)
- if (isspace(buf[i])) 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");
- }
+ }
if (!strncasecmp(header, "Content-Disposition: ", 21)) {
- extract_key(filename, header, "filename");
- }
+ strcpy(disposition, &header[21]);
+ extract_key(filename, disposition, "filename");
+ }
+ if (!strncasecmp(header, "Content-length: ", 16)) {
+ content_length = (size_t) atol(&header[16]);
+ }
if (!strncasecmp(header,
- "Content-transfer-encoding: ", 27))
- strcpy(encoding, &header[27]);
- if (strlen(boundary)==0)
+ "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)<sizeof(header))
+ }
+ if ((strlen(header) + strlen(buf) + 2) < sizeof(header))
strcat(header, buf);
- } while ((strlen(buf) > 0) && (*ptr != 0));
-
- for (i=0; i<strlen(content_type); ++i)
- if (content_type[i]==';') content_type[i] = 0;
+ } 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 (strlen(boundary) > 0) {
is_multipart = 1;
- }
- else {
+ } else {
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,
+ 0, encoding, userdata);
+ }
+
+ /* Figure out where the boundaries are */
sprintf(startary, "--%s", boundary);
sprintf(endary, "--%s--", boundary);
do {
- part_end = ptr;
- ptr = memreadline(ptr, buf, sizeof buf);
- if (*ptr == 0) return; /* premature end of message */
- if (content_end != NULL)
- if (ptr >= content_end) return;
- if ((!strcasecmp(buf, startary))
- ||(!strcasecmp(buf, endary))) {
+ if ( (!strncasecmp(ptr, startary, strlen(startary)))
+ || (!strncasecmp(ptr, endary, strlen(endary))) ) {
+ lprintf(9, "hit boundary!\n");
if (part_start != NULL) {
- sprintf(nested_partnum, "%s.%d",
- partnum, ++part_seq);
- the_mime_parser(nested_partnum,
- part_start, part_end,
- CallBack);
+ if (strlen(partnum) > 0) {
+ sprintf(nested_partnum, "%s.%d",
+ partnum, ++part_seq);
}
- part_start = ptr;
+ else {
+ sprintf(nested_partnum, "%d",
+ ++part_seq);
+ }
+ the_mime_parser(nested_partnum,
+ part_start, part_end,
+ CallBack,
+ PreMultiPartCallBack,
+ PostMultiPartCallBack,
+ userdata,
+ dont_decode);
}
- } while (strcasecmp(buf, endary));
+ ptr = memreadline(ptr, buf, sizeof(buf));
+ part_start = ptr;
+ }
+ else {
+ part_end = ptr;
+ ++ptr;
+ }
+ } while ( (strcasecmp(ptr, endary)) && (ptr <= content_end) );
+ if (PostMultiPartCallBack != NULL) {
+ PostMultiPartCallBack("", "", partnum, "", NULL,
+ content_type, 0, encoding, userdata);
}
+ return;
+ }
/* 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 != 0)&&((content_end==NULL)||(ptr<content_end))) {
+ while (ptr < content_end) {
+ ++ptr;
++length;
- part_end = ptr++;
- }
- mime_decode(partnum,
- part_start, length,
- content_type, encoding,
- name, filename, CallBack);
}
-
+ part_end = content_end;
+
+ /* Truncate if the header told us to */
+ if ( (content_length > 0) && (length > content_length) ) {
+ length = content_length;
+ lprintf(9, "truncated to %d\n", content_length);
+ }
+
+ mime_decode(partnum,
+ part_start, length,
+ content_type, encoding, disposition,
+ name, filename,
+ CallBack, NULL, NULL,
+ userdata, dont_decode);
}
+}
+
+
/*
* Entry point for the MIME parser.
* Note: NULL can be supplied as content_end; in this case, the message is
* considered to have ended when the parser encounters a 0x00 byte.
*/
-void mime_parser(char *content_start, char *content_end,
- void (*CallBack)
- (char *cbname,
- char *cbfilename,
- char *cbpartnum,
- void *cbcontent,
- char *cbtype,
- size_t cblength)
- ) {
-
- the_mime_parser("1", content_start, content_end, CallBack);
- }
+void mime_parser(char *content_start,
+ char *content_end,
+
+ void (*CallBack)
+ (char *cbname,
+ char *cbfilename,
+ char *cbpartnum,
+ char *cbdisp,
+ void *cbcontent,
+ char *cbtype,
+ size_t cblength,
+ char *cbencoding,
+ void *cbuserdata),
+
+ void (*PreMultiPartCallBack)
+ (char *cbname,
+ char *cbfilename,
+ char *cbpartnum,
+ char *cbdisp,
+ void *cbcontent,
+ char *cbtype,
+ size_t cblength,
+ char *cbencoding,
+ void *cbuserdata),
+
+ void (*PostMultiPartCallBack)
+ (char *cbname,
+ char *cbfilename,
+ char *cbpartnum,
+ char *cbdisp,
+ void *cbcontent,
+ char *cbtype,
+ size_t cblength,
+ char *cbencoding,
+ void *cbuserdata),
+
+ void *userdata,
+ int dont_decode
+)
+{
+
+ lprintf(9, "mime_parser() called\n");
+ the_mime_parser("", content_start, content_end,
+ CallBack,
+ PreMultiPartCallBack,
+ PostMultiPartCallBack,
+ userdata, dont_decode);
+}