* fix freeing.. well, most of it
[citadel.git] / webcit / html2html.c
index f96c23908da49607ef397fd2f183e10ab10cc219..cab626d3ee48e618ddb05f097553a43f26e36be6 100644 (file)
@@ -8,7 +8,6 @@
  */
 /*@{*/
 #include "webcit.h"
-#include "vcard.h"
 #include "webserver.h"
 
 
@@ -79,21 +78,20 @@ void extract_charset_from_meta(char *charset, char *meta_http_equiv, char *meta_
 /**
  * \brief Sanitize and enhance an HTML message for display.
  *        Also convert weird character sets to UTF-8 if necessary.
+ *        Also fixup img src="cid:..." type inline images to fetch the image
  *
  * \param supplied_charset the input charset as declared in the MIME headers
  */
-void output_html(char *supplied_charset, int treat_as_wiki) {
+void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, StrBuf *Source, StrBuf *Target) {
        char buf[SIZ];
        char *msg;
        char *ptr;
        char *msgstart;
        char *msgend;
-       char *converted_msg;
-       size_t converted_alloc = 0;
+       StrBuf *converted_msg;
        int buffer_length = 1;
        int line_length = 0;
        int content_length = 0;
-       int output_length = 0;
        char new_window[SIZ];
        int brak = 0;
        int alevel = 0;
@@ -110,21 +108,23 @@ void output_html(char *supplied_charset, int treat_as_wiki) {
        size_t obuflen;               /**< Length of output buffer              */
        char *osav;                   /**< Saved pointer to output buffer       */
 #endif
+       if (Target == NULL)
+               Target = WC->WBuf;
 
        safestrncpy(charset, supplied_charset, sizeof charset);
        msg = strdup("");
        sprintf(new_window, "<a target=\"%s\" href=", TARGET);
 
-       while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
+       if (Source == NULL) while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
                line_length = strlen(buf);
                buffer_length = content_length + line_length + 2;
                ptr = realloc(msg, buffer_length);
                if (ptr == NULL) {
-                       wprintf("<b>");
-                       wprintf(_("realloc() error! couldn't get %d bytes: %s"),
+                       StrBufAppendPrintf(Target, "<b>");
+                       StrBufAppendPrintf(Target, _("realloc() error! couldn't get %d bytes: %s"),
                                buffer_length + 1,
                                strerror(errno));
-                       wprintf("</b><br /><br />\n");
+                       StrBufAppendPrintf(Target, "</b><br /><br />\n");
                        while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
                                /** flush */
                        }
@@ -137,6 +137,12 @@ void output_html(char *supplied_charset, int treat_as_wiki) {
                strcpy(&msg[content_length], "\n");
                content_length += 1;
        }
+       else {
+               content_length = StrLength(Source);
+               free(msg);
+               msg = (char*) ChrPtr(Source);/* TODO: remove cast */
+               buffer_length = content_length;
+       }
 
        /** Do a first pass to isolate the message body */
        ptr = msg + 1;
@@ -229,6 +235,10 @@ void output_html(char *supplied_charset, int treat_as_wiki) {
                strcpy(msg, msgstart);
        }
 
+       /** Now go through the message, parsing tags as necessary. */
+       converted_msg = NewStrBufPlain(NULL, content_length + 8192);
+
+
        /** Convert foreign character sets to UTF-8 if necessary. */
 #ifdef HAVE_ICONV
        if ( (strcasecmp(charset, "us-ascii"))
@@ -236,25 +246,36 @@ void output_html(char *supplied_charset, int treat_as_wiki) {
           && (strcasecmp(charset, ""))
        ) {
                lprintf(9, "Converting %s to UTF-8\n", charset);
-               ic = ctdl_iconv_open("UTF-8", charset);
+               ctdl_iconv_open("UTF-8", charset, &ic);
                if (ic == (iconv_t)(-1) ) {
                        lprintf(5, "%s:%d iconv_open() failed: %s\n",
                                __FILE__, __LINE__, strerror(errno));
                }
        }
-       if (ic != (iconv_t)(-1) ) {
-               ibuf = msg;
-               ibuflen = content_length;
-               obuflen = content_length + (content_length / 2) ;
-               obuf = (char *) malloc(obuflen);
-               osav = obuf;
-               iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
-               content_length = content_length + (content_length / 2) - obuflen;
-               osav[content_length] = 0;
-               free(msg);
-               msg = osav;
-               iconv_close(ic);
+       if  (Source == NULL) {
+               if (ic != (iconv_t)(-1) ) {
+                       ibuf = msg;
+                       ibuflen = content_length;
+                       obuflen = content_length + (content_length / 2) ;
+                       obuf = (char *) malloc(obuflen);
+                       osav = obuf;
+                       iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
+                       content_length = content_length + (content_length / 2) - obuflen;
+                       osav[content_length] = 0;
+                       free(msg);
+                       msg = osav;
+                       iconv_close(ic);
+               }
+       }
+       else {
+               if (ic != (iconv_t)(-1) ) {
+                       StrBuf *Buf = NewStrBufPlain(NULL, StrLength(Source) + 8096);;
+                       StrBufConvert(Source, Buf, &ic);
+                       FreeStrBuf(&Buf);
+                       iconv_close(ic);
+               }
        }
+               
 #endif
 
        /**
@@ -265,15 +286,11 @@ void output_html(char *supplied_charset, int treat_as_wiki) {
         *      terminated now.
         */
 
-       /** Now go through the message, parsing tags as necessary. */
-       converted_alloc = content_length + 8192;
-       converted_msg = malloc(converted_alloc);
        if (converted_msg == NULL) {
-               wprintf("Error %d: %s<br />%s:%s", errno, strerror(errno), __FILE__, __LINE__);
+               StrBufAppendPrintf(Target, "Error %d: %s<br />%s:%d", errno, strerror(errno), __FILE__, __LINE__);
                goto BAIL;
        }
 
-       strcpy(converted_msg, "");
        ptr = msg;
        msgend = strchr(msg, 0);
        while (ptr < msgend) {
@@ -281,7 +298,7 @@ void output_html(char *supplied_charset, int treat_as_wiki) {
                /** Try to sanitize the html of any rogue scripts */
                if (!strncasecmp(ptr, "<script", 7)) {
                        if (scriptlevel == 0) {
-                               script_start_pos = output_length;
+                               script_start_pos = StrLength(converted_msg);
                        }
                        ++scriptlevel;
                }
@@ -297,16 +314,8 @@ void output_html(char *supplied_charset, int treat_as_wiki) {
                 */
                if (!strncasecmp(ptr, "<a href=\"mailto:", 16)) {
                        content_length += 64;
-                       if (content_length >= converted_alloc) {
-                               converted_alloc += 8192;
-                               converted_msg = realloc(converted_msg, converted_alloc);
-                               if (converted_msg == NULL) {
-                                       abort();
-                               }
-                       }
-                       sprintf(&converted_msg[output_length],
+                       StrBufAppendPrintf(converted_msg,
                                "<a href=\"display_enter?force_room=_MAIL_&recp=");
-                       output_length += 46;
                        ptr = &ptr[16];
                        ++alevel;
                        ++brak;
@@ -319,37 +328,47 @@ void output_html(char *supplied_charset, int treat_as_wiki) {
                             &&  ((strchr(ptr, '/') < strchr(ptr, '>'))) 
                             ) {
                                /* open external links to new window */
-                               content_length += 64;
-                               if (content_length >= converted_alloc) {
-                                       converted_alloc += 8192;
-                                       converted_msg = realloc(converted_msg, converted_alloc);
-                                       if (converted_msg == NULL) {
-                                               abort();
-                                       }
-                               }
-                               sprintf(&converted_msg[output_length], new_window);
-                               output_length += strlen(new_window);
+                               StrBufAppendPrintf(converted_msg, new_window);
                                ptr = &ptr[8];
                        }
                        else if ( (treat_as_wiki) && (strncasecmp(ptr, "<a href=\"wiki?", 14)) ) {
                                content_length += 64;
-                               if (content_length >= converted_alloc) {
-                                       converted_alloc += 8192;
-                                       converted_msg = realloc(converted_msg, converted_alloc);
-                                       if (converted_msg == NULL) {
-                                               abort();
-                                       }
-                               }
-                               sprintf(&converted_msg[output_length], "<a href=\"wiki?page=");
-                               output_length += 19;
+                               StrBufAppendPrintf(converted_msg, "<a href=\"wiki?page=");
                                ptr = &ptr[9];
                        }
                        else {
-                               sprintf(&converted_msg[output_length], "<a href=\"");
-                               output_length += 9;
+                               StrBufAppendPrintf(converted_msg, "<a href=\"");
                                ptr = &ptr[9];
                        }
                }
+               /** Fixup <img src="cid:... ...> to fetch the mime part */
+               else if (!strncasecmp(ptr, "<img ", 5)) {
+                       char* tag_end=strchr(ptr,'>');
+                       char* src=strstr(ptr, " src=\"cid:");
+                       char *cid_start, *cid_end;
+                       ++brak;
+
+                       if (src && 
+                               (cid_start=strchr(src,':')) && 
+                               (cid_end=strchr(cid_start,'"')) &&
+                               (cid_end < tag_end)) {
+
+                               /* copy tag and attributes up to src="cid: */
+                               StrBufAppendBufPlain(converted_msg, ptr, src - ptr, 0);
+                               cid_start++;
+
+                               /* add in /webcit/mimepart/<msgno>/CID/ 
+                                  trailing / stops dumb URL filters getting excited */
+                               StrBufAppendPrintf(converted_msg,
+                                       "src=\"/webcit/mimepart/%d/",msgnum);
+                               StrBufAppendBufPlain(converted_msg, cid_start, cid_end - cid_start, 0);
+                               StrBufAppendBufPlain(converted_msg, "/\"", -1, 0);
+                               
+                               ptr = cid_end+1;
+                       }
+                       StrBufAppendBufPlain(converted_msg, ptr, tag_end - ptr, 0);
+                       ptr = tag_end;
+               }
 
                /**
                 * Turn anything that looks like a URL into a real link, as long
@@ -360,6 +379,7 @@ void output_html(char *supplied_charset, int treat_as_wiki) {
                                /** Find the end of the link */
                                int strlenptr;
                                linklen = 0;
+                               
                                strlenptr = strlen(ptr);
                                for (i=0; i<=strlenptr; ++i) {
                                        if ((ptr[i]==0)
@@ -372,13 +392,30 @@ void output_html(char *supplied_charset, int treat_as_wiki) {
                                           ||(ptr[i]=='>')
                                           ||(ptr[i]=='[')
                                           ||(ptr[i]==']')
+                                          ||(ptr[i]=='"')
+                                          ||(ptr[i]=='\'')
                                        ) linklen = i;
+                                       /* did s.b. send us an entity? */
+                                       if (ptr[i] == '&') {
+                                               if ((ptr[i+2] ==';') ||
+                                                   (ptr[i+3] ==';') ||
+                                                   (ptr[i+5] ==';') ||
+                                                   (ptr[i+6] ==';') ||
+                                                   (ptr[i+7] ==';'))
+                                                       linklen = i;
+                                       }
                                        if (linklen > 0) break;
                                }
                                if (linklen > 0) {
                                        char *ltreviewptr;
                                        char *nbspreviewptr;
-                                       //* spot for some subject strings tinymce tends to give us.
+                                       char linkedchar;
+                                       int len = linklen;
+                                       
+                                       len = linklen;
+                                       linkedchar = ptr[len];
+                                       ptr[len] = '\0';
+                                       /* spot for some subject strings tinymce tends to give us. */
                                        ltreviewptr = strchr(ptr, '<');
                                        if (ltreviewptr != NULL) {
                                                *ltreviewptr = '\0';
@@ -393,35 +430,20 @@ void output_html(char *supplied_charset, int treat_as_wiki) {
                                        if (ltreviewptr != 0)
                                                *ltreviewptr = '<';
 
+                                       ptr[len] = linkedchar;
+
                                        content_length += (32 + linklen);
-                                       if (content_length >= converted_alloc) {
-                                               converted_alloc += 8192;
-                                               converted_msg = realloc(converted_msg, converted_alloc);
-                                               if (converted_msg == NULL) {
-                                                       abort();
-                                               }
-                                       }
-                                       sprintf(&converted_msg[output_length], new_window);
-                                       output_length += strlen(new_window);
-                                       converted_msg[output_length] = '\"';
-                                       converted_msg[++output_length] = 0;
-                                       for (i=0; i<linklen; ++i) {
-                                               converted_msg[output_length] = ptr[i];
-                                               converted_msg[++output_length] = 0;
-                                       }
-                                       sprintf(&converted_msg[output_length], "\">");
-                                       output_length += 2;
-                                       for (i=0; i<linklen; ++i) {
-                                               converted_msg[output_length] = *ptr++;
-                                               converted_msg[++output_length] = 0;
-                                       }
-                                       sprintf(&converted_msg[output_length], "</A>");
-                                       output_length += 4;
+                                        StrBufAppendPrintf(converted_msg, "%s\"", new_window);
+                                       StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
+                                       StrBufAppendPrintf(converted_msg, "\">");
+                                       StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
+                                       ptr += linklen;
+                                       StrBufAppendPrintf(converted_msg, "</A>");
                                }
                }
                else {
-                       converted_msg[output_length] = *ptr++;
-                       converted_msg[++output_length] = 0;
+                       StrBufAppendBufPlain(converted_msg, ptr, 1, 0);
+                       ptr++;
                }
 
                /**
@@ -435,8 +457,7 @@ void output_html(char *supplied_charset, int treat_as_wiki) {
                if (*(ptr-1) == '>') {
                        --brak;
                        if ((scriptlevel == 0) && (script_start_pos >= 0)) {
-                               output_length = script_start_pos;
-                               converted_msg[output_length] = 0;
+                               StrBufCutRight(converted_msg, StrLength(converted_msg) - script_start_pos);
                                script_start_pos = (-1);
                        }
                }
@@ -448,14 +469,14 @@ void output_html(char *supplied_charset, int treat_as_wiki) {
        /**     output_length = content_length;                         */
 
        /** Output our big pile of markup */
-       client_write(converted_msg, output_length);
+       StrBufAppendBuf(Target, converted_msg, 0);
 
 BAIL:  /** A little trailing vertical whitespace... */
-       wprintf("<br /><br />\n");
+       StrBufAppendPrintf(Target, "<br /><br />\n");
 
        /** Now give back the memory */
-       if (converted_msg != NULL) free(converted_msg);
-       if (msg != NULL) free(msg);
+       FreeStrBuf(&converted_msg);
+       if ((msg != NULL) && (Source == NULL)) free(msg);
 }
 
 /*@}*/