]> code.citadel.org Git - citadel.git/blobdiff - citadel/mime_parser.c
- port to Cygwin (DLL support, etc.)
[citadel.git] / citadel / mime_parser.c
index 9904f4607a6fae866bfce4e8782b9df94397f993..1d35ce7a7fd731f6c84bd757f2b792a993e265f3 100644 (file)
@@ -1,15 +1,17 @@
 /*
  * $Id$
  *
- * 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.
  *
  */
 
-#include "sysdep.h"
+#ifdef DLL_EXPORT
+#define IN_LIBCIT
+#endif
+
 #include <stdlib.h>
 #include <unistd.h>
 #include <stdio.h>
 #include <sys/stat.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)
@@ -47,34 +50,17 @@ void extract_key(char *target, char *source, char *key)
 }
 
 
-
-/* 
- * 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);
-
-       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;
-               }
-       }
+char *fixed_partnum(char *supplied_partnum) {
+       if (supplied_partnum == NULL) return "1";
+       if (strlen(supplied_partnum)==0) return "1";
+       return supplied_partnum;
 }
 
+
 /*
  * Given a message or message-part body and a length, handle any necessary
  * decoding and pass the request up the stack.
@@ -92,8 +78,30 @@ void mime_decode(char *partnum,
                   void *cbcontent,
                   char *cbtype,
                   size_t cblength,
+                  char *cbencoding,
                   void *cbuserdata),
-                 void *userdata
+                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
 )
 {
 
@@ -118,14 +126,17 @@ void mime_decode(char *partnum,
                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, userdata);
+       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, "quoted-printable"))) {
-               lprintf(5, "ERROR: unknown MIME encoding '%s'\n", encoding);
+               lprintf(9, "ERROR: unknown MIME encoding '%s'\n", encoding);
                return;
        }
        /*
@@ -135,9 +146,9 @@ void mime_decode(char *partnum,
         * assumption with base64, uuencode, and quoted-printable.  Just to
         * be safe, we still pad the buffer a bit.
         */
-       decoded = mallok(length + 1024);
+       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)
@@ -147,23 +158,23 @@ void mime_decode(char *partnum,
 
        childpid = fork();
        if (childpid < 0) {
-               phree(decoded);
+               free(decoded);
                return;
        }
        if (childpid == 0) {
                close(2);
                /* send stdio to the pipes */
                if (dup2(sendpipe[0], 0) < 0)
-                       lprintf(5, "ERROR dup2()\n");
+                       lprintf(9, "ERROR dup2()\n");
                if (dup2(recvpipe[1], 1) < 0)
-                       lprintf(5, "ERROR dup2()\n");
+                       lprintf(9, "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);
+               lprintf(9, "ERROR: cannot exec decoder for %s\n", encoding);
                exit(1);
        }
        close(sendpipe[0]);     /* Close the ends we're not using  */
@@ -175,7 +186,7 @@ void mime_decode(char *partnum,
                        blocksize = read(recvpipe[0], &decoded[bytes_recv],
                                         statbuf.st_size);
                        if (blocksize < 0)
-                               lprintf(5, "ERROR: cannot read from pipe\n");
+                               lprintf(9, "ERROR: cannot read from pipe\n");
                        else
                                bytes_recv = bytes_recv + blocksize;
                }
@@ -184,7 +195,7 @@ void mime_decode(char *partnum,
                if (blocksize > 2048)
                        blocksize = 2048;
                if (write(sendpipe[1], &part_start[bytes_sent], blocksize) < 0) {
-                       lprintf(5, "ERROR: cannot write to pipe: %s\n",
+                       lprintf(9, "ERROR: cannot write to pipe: %s\n",
                                strerror(errno));
                        write_error = 1;
                }
@@ -197,11 +208,13 @@ void mime_decode(char *partnum,
                bytes_recv = bytes_recv + blocksize;
        }
 
-       if (bytes_recv > 0)
-               CallBack(name, filename, partnum, disposition, decoded,
-                        content_type, bytes_recv, userdata);
+       if (bytes_recv > 0) if (CallBack != NULL) {
+               CallBack(name, filename, fixed_partnum(partnum),
+                       disposition, decoded,
+                       content_type, bytes_recv, "binary", userdata);
+       }
 
-       phree(decoded);
+       free(decoded);
 }
 
 /*
@@ -220,28 +233,51 @@ void the_mime_parser(char *partnum,
                       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 *userdata
+                    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 disposition[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;
@@ -251,16 +287,19 @@ void the_mime_parser(char *partnum,
        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]))
@@ -274,6 +313,9 @@ void the_mime_parser(char *partnum,
                                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]);
@@ -302,26 +344,27 @@ void the_mime_parser(char *partnum,
                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 */
-               CallBack("", "", partnum, "", NULL, content_type, 0, userdata);
+               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) {
                                        if (strlen(partnum) > 0) {
                                                sprintf(nested_partnum, "%s.%d",
@@ -333,29 +376,51 @@ void the_mime_parser(char *partnum,
                                        }
                                        the_mime_parser(nested_partnum,
                                                    part_start, part_end,
-                                                       CallBack, userdata);
+                                                       CallBack,
+                                                       PreMultiPartCallBack,
+                                                       PostMultiPartCallBack,
+                                                       userdata,
+                                                       dont_decode);
                                }
+                               ptr = memreadline(ptr, buf, sizeof(buf));
                                part_start = ptr;
                        }
-               } while (strcasecmp(buf, endary));
+                       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++;
                }
+               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, userdata);
+                           name, filename,
+                           CallBack, NULL, NULL,
+                           userdata, dont_decode);
        }
-
-
 }
 
 
@@ -366,7 +431,9 @@ void the_mime_parser(char *partnum,
  * 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 mime_parser(char *content_start,
+               char *content_end,
+
                 void (*CallBack)
                  (char *cbname,
                   char *cbfilename,
@@ -375,11 +442,40 @@ void mime_parser(char *content_start, char *content_end,
                   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 *userdata
+
+                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, userdata);
+       the_mime_parser("", content_start, content_end,
+                       CallBack,
+                       PreMultiPartCallBack,
+                       PostMultiPartCallBack,
+                       userdata, dont_decode);
 }