* mime_parser.c
*
* This is a really bad attempt at writing a parser to handle MIME-encoded
- * data, including multipart messages. In the case of WebCit, the input data
- * might be a form containing uploaded files. In the Citadel server, the data
- * is more likely to be an actual MIME-encoded message.
+ * messages.
*
* Copyright (c) 1998-1999 by Art Cancro
* This code is distributed under the terms of the GNU General Public License.
*
*/
+#include "sysdep.h"
#include <stdlib.h>
#include <unistd.h>
#include <stdio.h>
#include <sys/types.h>
#include <ctype.h>
#include <string.h>
+#include <sys/stat.h>
+#include <errno.h>
+#ifdef HAVE_PTHREAD_H
+#include <pthread.h>
+#endif
+#include "citadel.h"
#include "mime_parser.h"
+#include "sysdep_decls.h"
+#include "server.h"
+/*
+ * Utility function to "readline" from memory
+ * (returns new pointer)
+ */
+char *memreadline(char *start, char *buf, int maxlen) {
+ char ch;
+ char *ptr;
+
+ ptr = start;
+ memset(buf, 0, maxlen);
+
+ 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;
+ }
+ }
+ }
+
/*
- * The very back end for the component handler
- * (This function expects to be fed CONTENT ONLY, no headers)
+ * Given a message or message-part body and a length, handle any necessary
+ * decoding and pass the request up the stack.
*/
-void do_something_with_it(char *content,
- int length,
- char *content_type,
- char *content_disposition,
+void mime_decode(char *partnum,
+ char *part_start, size_t length,
+ char *content_type, char *encoding,
+ char *disposition,
+ char *name, char *filename,
void (*CallBack)
(char *cbname,
char *cbfilename,
- char *cbencoding,
+ char *cbpartnum,
+ char *cbdisp,
void *cbcontent,
char *cbtype,
size_t cblength)
) {
- char name[256];
- char filename[256];
- extract_key(name, content_disposition, " name");
- extract_key(filename, content_disposition, "filename");
+ char *decoded;
+ struct stat statbuf;
+ int sendpipe[2];
+ int recvpipe[2];
+ int childpid;
+ size_t bytes_sent = 0;
+ size_t bytes_recv = 0;
+ size_t blocksize;
+ int write_error = 0;
+
+ lprintf(9, "mime_decode() called\n");
+
+ /* 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, disposition, part_start,
+ content_type, length);
+ return;
+ }
+
+ if ( (strcasecmp(encoding, "base64"))
+ && (strcasecmp(encoding, "quoted-printable")) ) {
+ lprintf(5, "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. Just to
+ * be safe, we still pad the buffer a bit.
+ */
+ decoded = mallok(length + 1024);
+ if (decoded == NULL) {
+ lprintf(5, "ERROR: cannot allocate memory.\n");
+ return;
+ }
+ if (pipe(sendpipe) != 0) return;
+ if (pipe(recvpipe) != 0) return;
- /* Nested multipart gets recursively fed back into the parser */
- if (!strncasecmp(content_type, "multipart", 9)) {
- mime_parser(content, length, content_type, CallBack);
+ childpid = fork();
+ if (childpid < 0) {
+ phree(decoded);
+ return;
}
- /**** OTHERWISE, HERE'S WHERE WE HANDLE THE STUFF!! *****/
+ if (childpid == 0) {
+ 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 */
+ close(recvpipe[0]);
+ if (!strcasecmp(encoding, "base64"))
+ execlp("./base64", "base64", "-d", NULL);
+ else if (!strcasecmp(encoding, "quoted-printable"))
+ execlp("./qpdecode", "qpdecode", NULL);
+ lprintf(5, "ERROR: cannot exec decoder for %s\n", encoding);
+ exit(1);
+ }
- CallBack(name, filename, "", content, content_type, length);
+ close(sendpipe[0]); /* Close the ends we're not using */
+ close(recvpipe[1]);
+
+ while ( (bytes_sent < length) && (write_error == 0) ) {
+ /* Empty the input pipe FIRST */
+ 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");
+ 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",
+ strerror(errno));
+ write_error = 1;
+ }
+ bytes_sent = bytes_sent + blocksize;
+ }
+ close(sendpipe[1]);
+ /* Empty the input pipe */
+ while ( (blocksize = read(recvpipe[0], &decoded[bytes_recv], 1)),
+ (blocksize > 0) ) {
+ bytes_recv = bytes_recv + blocksize;
+ }
- /**** END OF STUFF-HANDLER ****/
+ if (bytes_recv > 0)
+ CallBack(name, filename, partnum, disposition, decoded,
+ content_type, bytes_recv);
+ phree(decoded);
}
-
/*
- * Take a part, figure out its length, and do something with it
- * (This function expects to be fed HEADERS+CONTENT)
+ * Break out the components of a multipart message
+ * (This function expects to be fed HEADERS + CONTENT)
+ * 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 handle_part(char *content,
- int part_length,
- char *supplied_content_type,
+void the_mime_parser(char *partnum,
+ char *content_start, char *content_end,
void (*CallBack)
(char *cbname,
char *cbfilename,
- char *cbencoding,
+ char *cbpartnum,
+ char *cbdisp,
void *cbcontent,
char *cbtype,
size_t cblength)
) {
+
+ 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 content_disposition[256];
- char *start;
- char buf[512];
- int crlf = 0; /* set to 1 for crlf-style newlines */
- int actual_length;
-
- strcpy(content_type, supplied_content_type);
-
- /* Strip off any leading blank lines. */
- start = content;
- while ((!strncmp(start, "\r", 1)) || (!strncmp(start, "\n", 1))) {
- ++start;
- --part_length;
+ char encoding[256];
+ char disposition[256];
+ char name[256];
+ char filename[256];
+ int is_multipart;
+ int part_seq = 0;
+ int i;
+ size_t length;
+ char nested_partnum[256];
+
+ 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);
+
+ /* 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;
+
+ 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)) {
+ strcpy(disposition, &header[21]);
+ extract_key(filename, disposition, "filename");
+ }
+ 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)<sizeof(header))
+ strcat(header, buf);
+ } while ((strlen(buf) > 0) && (*ptr != 0));
+
+ for (i=0; i<strlen(disposition); ++i)
+ if (disposition[i]==';') disposition[i] = 0;
+ for (i=0; i<strlen(content_type); ++i)
+ if (content_type[i]==';') content_type[i] = 0;
+
+ if (strlen(boundary) > 0) {
+ is_multipart = 1;
+ }
+ else {
+ is_multipart = 0;
}
- /* At this point all we have left is the headers and the content. */
- do {
- strcpy(buf, "");
+ /* If this is a multipart message, then recursively process it */
+ part_start = NULL;
+ if (is_multipart) {
+ sprintf(startary, "--%s", boundary);
+ sprintf(endary, "--%s--", boundary);
do {
- buf[strlen(buf)+1] = 0;
- if (strlen(buf)<((sizeof buf)-1)) {
- strncpy(&buf[strlen(buf)], start, 1);
+ 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 (part_start != NULL) {
+ sprintf(nested_partnum, "%s.%d",
+ partnum, ++part_seq);
+ the_mime_parser(nested_partnum,
+ part_start, part_end,
+ CallBack);
+ }
+ part_start = ptr;
}
- ++start;
- --part_length;
- } while((buf[strlen(buf)-1] != 10) && (part_length>0));
- if (part_length <= 0) return;
- buf[strlen(buf)-1] = 0;
- if (buf[strlen(buf)-1]==13) {
- buf[strlen(buf)-1] = 0;
- crlf = 1;
- }
- if (!strncasecmp(buf, "Content-type: ", 14)) {
- strcpy(content_type, &buf[14]);
- }
- if (!strncasecmp(buf, "Content-disposition: ", 21)) {
- strcpy(content_disposition, &buf[21]);
+ } while (strcasecmp(buf, endary));
+ }
+
+ /* If it's not a multipart message, then do something with it */
+ if (!is_multipart) {
+ part_start = ptr;
+ length = 0;
+ while ((*ptr != 0)&&((content_end==NULL)||(ptr<content_end))) {
+ ++length;
+ part_end = ptr++;
}
- } while (strlen(buf)>0);
+ mime_decode(partnum,
+ part_start, length,
+ content_type, encoding, disposition,
+ name, filename, CallBack);
+ }
- if (crlf) actual_length = part_length - 2;
- else actual_length = part_length - 1;
-
- /* Now that we've got this component isolated, what to do with it? */
- do_something_with_it(start, actual_length,
- content_type, content_disposition, CallBack);
-
}
-
/*
- * Break out the components of a multipart message
- * (This function expects to be fed CONTENT ONLY, no headers)
+ * Entry point for the MIME parser.
+ * (This function expects to be fed HEADERS + CONTENT)
+ * 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,
- int ContentLength,
- char *ContentType,
+void mime_parser(char *content_start, char *content_end,
void (*CallBack)
(char *cbname,
char *cbfilename,
- char *cbencoding,
+ char *cbpartnum,
+ char *cbdisp,
void *cbcontent,
char *cbtype,
size_t cblength)
) {
- char boundary[256];
- char endary[256];
- int have_boundary = 0;
- int a;
- char *ptr;
- char *beginning;
- int bytes_processed = 0;
- int part_length;
-
- /* If it's not multipart, don't process it as multipart */
- if (strncasecmp(ContentType, "multipart", 9)) {
- do_something_with_it(content, ContentLength,
- ContentType, "", CallBack);
- return;
- }
-
- /* Figure out what the boundary is */
- strcpy(boundary, ContentType);
- for (a=0; a<strlen(boundary); ++a) {
- if (!strncasecmp(&boundary[a], "boundary=", 9)) {
- boundary[0]='-';
- boundary[1]='-';
- strcpy(&boundary[2], &boundary[a+9]);
- have_boundary = 1;
- a = 0;
- }
- if ((boundary[a]==13) || (boundary[a]==10)) {
- boundary[a] = 0;
- }
- }
- if (boundary[2]==34) {
- strcpy(&boundary[2], &boundary[3]);
- for (a=2; a<strlen(boundary); ++a)
- if (boundary[a]==34) boundary[a]=0;
- }
-
- /* We can't process multipart messages without a boundary. */
- if (have_boundary == 0) return;
- strcpy(endary, boundary);
- strcat(endary, "--");
- fprintf(stderr, "BOUNDARY: %s\n", boundary);
-
- ptr = content;
-
- /* Seek to the beginning of the next boundary */
- while (bytes_processed < ContentLength) {
- /* && (strncasecmp(ptr, boundary, strlen(boundary))) ) { */
- if (strncasecmp(ptr, boundary, strlen(boundary))) {
- ++ptr;
- ++bytes_processed;
- }
-
- /* See if we're at the end */
- if (!strncasecmp(ptr, endary, strlen(endary))) {
- return;
- }
-
- /* Seek to the end of the boundary string */
- if (!strncasecmp(ptr, boundary, strlen(boundary))) {
- fprintf(stderr, "FOUNDA BOUNDA\n");
- while ( (bytes_processed < ContentLength)
- && (strncasecmp(ptr, "\n", 1)) ) {
- ++ptr;
- ++bytes_processed;
- }
- beginning = ptr;
- part_length = 0;
- while ( (bytes_processed < ContentLength)
- && (strncasecmp(ptr, boundary, strlen(boundary))) ) {
- ++ptr;
- ++bytes_processed;
- ++part_length;
- }
- handle_part(beginning, part_length, "", CallBack);
- /* Back off so we can see the next boundary */
- --ptr;
- --bytes_processed;
- }
- }
+ lprintf(9, "mime_parser() called\n");
+ the_mime_parser("1", content_start, content_end, CallBack);
}
projects / citadel.git / blobdiff