* 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.
-/*
- * The very back end for the component handler
- * (This function expects to be fed CONTENT ONLY, no headers)
- */
-void do_something_with_it(char *content,
- int length,
- char *content_type,
- char *content_disposition,
- void (*CallBack)
- (char *cbname,
- char *cbfilename,
- char *cbencoding,
- 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");
-
- /* Nested multipart gets recursively fed back into the parser */
- if (!strncasecmp(content_type, "multipart", 9)) {
- mime_parser(content, length, content_type, CallBack);
- }
-
- /**** OTHERWISE, HERE'S WHERE WE HANDLE THE STUFF!! *****/
+ /**** OTHERWISE, HERE'S WHERE WE HANDLE THE STUFF!! *****
CallBack(name, filename, "", content, content_type, length);
- /**** END OF STUFF-HANDLER ****/
-
- }
+ **** END OF STUFF-HANDLER ****/
-/*
- * Take a part, figure out its length, and do something with it
- * (This function expects to be fed HEADERS+CONTENT)
+/*
+ * Utility function to "readline" from memory
+ * (returns new pointer)
*/
-void handle_part(char *content,
- int part_length,
- char *supplied_content_type,
- void (*CallBack)
- (char *cbname,
- char *cbfilename,
- char *cbencoding,
- void *cbcontent,
- char *cbtype,
- size_t cblength)
- ) {
- 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 *memreadline(char *start, char *buf, int maxlen) {
+ char ch;
+ char *ptr;
- /* At this point all we have left is the headers and the content. */
- do {
- strcpy(buf, "");
- do {
- buf[strlen(buf)+1] = 0;
- if (strlen(buf)<((sizeof buf)-1)) {
- strncpy(&buf[strlen(buf)], start, 1);
- }
- ++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]);
+ ptr = start;
+ bzero(buf, 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 (!strncasecmp(buf, "Content-disposition: ", 21)) {
- strcpy(content_disposition, &buf[21]);
+ if (strlen(buf) < (maxlen-1)) {
+ buf[strlen(buf)+1] = 0;
+ buf[strlen(buf)] = ch;
}
- } while (strlen(buf)>0);
-
- 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)
+ * (This function expects to be fed HEADERS + CONTENT)
*/
-
-
void mime_parser(char *content,
- int ContentLength,
- char *ContentType,
void (*CallBack)
(char *cbname,
char *cbfilename,
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);
+ char *ptr;
+ char *part_start, *part_end;
+ char buf[256];
+ char header[256];
+ char boundary[256];
+ char content_type[256];
+ char encoding[256];
+ int content_length;
+ int i;
ptr = content;
+ bzero(boundary, sizeof boundary);
+ bzero(content_type, sizeof content_type);
+ bzero(encoding, sizeof encoding);
+ content_length = 0;
- /* 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;
+ /* Learn interesting things from the headers */
+ strcpy(header, "");
+ do {
+ ptr = memreadline(ptr, buf, sizeof buf);
+ 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]);
+ if (!strncasecmp(header, "Content-length: ", 16))
+ content_length = atoi(&header[16]);
+ 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));
- /* See if we're at the end */
- if (!strncasecmp(ptr, endary, strlen(endary))) {
- return;
- }
+ cprintf("Content type is <%s>\n", content_type);
+ cprintf("Encoding is <%s>\n", encoding);
+ cprintf("Content length is %d\n", content_length);
+ cprintf("Boundary is <%s>\n", boundary);
- /* 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;
- }
+ if (*ptr == 0) return; /* premature end of message */
+
+ /* If this is a multipart message, then recursively process it */
+ if (strlen(boundary)>0) {
}
+
+
}
-/*
- * Feed MIME-encoded stuff to the mime_parser
- */
-void output_mime_parts(char *msg) {
- char content_type[256];
- int content_length = (-1);
- char buf[256];
- CIT_UBYTE rch;
- char *mptr, *meas;
- int i;
-
- strcpy(content_type, "");
- mptr = msg;
-
- while(1) {
- buf[0] = 0;
- do {
- do {
- buf[strlen(buf)+1] = 0;
- rch = *mptr++;
- if (strlen(buf)<((sizeof buf)-1))
- buf[strlen(buf)] = rch;
- } while ( (rch > 0) && (rch != 10) );
- if (buf[strlen(buf)-1]==10) {
- buf[strlen(buf)-1] = 0;
- }
- else {
- return;
- }
- if (buf[strlen(buf)-1]==13) buf[strlen(buf)-1]=0;
- } while (buf[strlen(buf)-1]==';');
- for (i=0; i<strlen(buf); ++i) if (isspace(buf[i])) buf[i]=' ';
- if (!strncasecmp(buf, "Content-type: ", 14))
- strcpy(content_type, &buf[14]);
- if (!strncasecmp(buf, "Content-length: ", 16))
- content_length = atoi(&buf[16]);
- if (strlen(buf)==0) {
- if (content_length < 0) {
- content_length = 0;
- meas = mptr;
- while (*mptr++ != 0) ++content_length;
- }
- cprintf("mime=type=%s\n", content_type);
- cprintf("mime=len=%d\n", content_length);
- mime_parser(mptr, content_length, content_type,
- *part_handler);
- return;
- }
- }
- }
-
-
-
/*
* Get a message off disk. (return value is the message's timestamp)
/* do some sort of MIME output */
if ( (mode == MT_MIME) && (format_type == 4) ) {
- output_mime_parts(mptr);
+ mime_parser(mptr, *part_handler);
cprintf("000\n");
cdb_free(dmsgtext);
return(xtime);