]> code.citadel.org Git - citadel.git/blobdiff - webcit/html2html.c
* Changed some of the '&' between url parameters to '?' (ongoing)
[citadel.git] / webcit / html2html.c
index e331cda67cc21f04347124d8399c9846b7601cf0..1daf7b4103218872572b09c5e9186b3d27f10485 100644 (file)
@@ -70,6 +70,9 @@ void extract_charset_from_meta(char *charset, char *meta_http_equiv, char *meta_
                        strcpy(charset, "UTF-8");
                }
 
+               /* Remove wandering punctuation */
+               if ((ptr=strchr(charset, '\"'))) *ptr = 0;
+               striplt(charset);
        }
 }
 
@@ -100,6 +103,7 @@ void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, St
        int i;
        int linklen;
        char charset[128];
+       StrBuf *BodyArea = NULL;
 #ifdef HAVE_ICONV
        iconv_t ic = (iconv_t)(-1) ;
        char *ibuf;                   /**< Buffer of characters to be converted */
@@ -122,8 +126,8 @@ void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, St
                if (ptr == NULL) {
                        StrBufAppendPrintf(Target, "<b>");
                        StrBufAppendPrintf(Target, _("realloc() error! couldn't get %d bytes: %s"),
-                               buffer_length + 1,
-                               strerror(errno));
+                                       buffer_length + 1,
+                                       strerror(errno));
                        StrBufAppendPrintf(Target, "</b><br /><br />\n");
                        while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
                                /** flush */
@@ -192,7 +196,7 @@ void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, St
                                                        stripquotes(meta_http_equiv);
                                                        stripquotes(meta_content);
                                                        extract_charset_from_meta(charset,
-                                                               meta_http_equiv, meta_content);
+                                                                       meta_http_equiv, meta_content);
                                                }
                                                free(meta_content);
                                        }
@@ -207,11 +211,55 @@ void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, St
                 * the tag to be removed.
                 */     
                if ( (!strncasecmp(ptr, "HTML", 4))
-                  ||(!strncasecmp(ptr, "HEAD", 4))
-                  ||(!strncasecmp(ptr, "/HEAD", 5))
-                  ||(!strncasecmp(ptr, "BODY", 4)) ) {
+                               ||(!strncasecmp(ptr, "HEAD", 4))
+                               ||(!strncasecmp(ptr, "/HEAD", 5))
+                               ||(!strncasecmp(ptr, "BODY", 4)) ) {
+                       char *pBody = NULL;
+
+                       if (!strncasecmp(ptr, "BODY", 4)) {
+                               pBody = ptr;
+                       }
                        ptr = strchr(ptr, '>');
                        if ((ptr == NULL) || (ptr >= msgend)) break;
+                       if ((pBody != NULL) && (ptr - pBody > 4)) {
+                               char* src;
+                               char *cid_start, *cid_end;
+
+                               *ptr = '\0';
+                               pBody += 4; 
+                               while ((isspace(*pBody)) && (pBody < ptr))
+                                       pBody ++;
+                               BodyArea = NewStrBufPlain(NULL,  ptr - pBody);
+
+                               if (pBody < ptr) {
+                                       src = strstr(pBody, "cid:");
+                                       if (src) {
+                                               cid_start = src + 4;
+                                               cid_end = cid_start;
+                                               while ((*cid_end != '"') && 
+                                                               !isspace(*cid_end) &&
+                                                               (cid_end < ptr))
+                                                       cid_end ++;
+
+                                               /* copy tag and attributes up to src="cid: */
+                                               StrBufAppendBufPlain(BodyArea, pBody, src - pBody, 0);
+
+                                               /* add in /webcit/mimepart/<msgno>/CID/ 
+                                                  trailing / stops dumb URL filters getting excited */
+                                               StrBufAppendPrintf(BodyArea,
+                                                               "/webcit/mimepart/%d/",msgnum);
+                                               StrBufAppendBufPlain(BodyArea, cid_start, cid_end - cid_start, 0);
+
+                                               if (ptr - cid_end > 0)
+                                                       StrBufAppendBufPlain(BodyArea, 
+                                                                       cid_end + 1, 
+                                                                       ptr - cid_end, 0);
+                                       }
+                                       else 
+                                               StrBufAppendBufPlain(BodyArea, pBody, ptr - pBody, 0);
+                               }
+                               *ptr = '>';
+                       }
                        ++ptr;
                        if ((ptr == NULL) || (ptr >= msgend)) break;
                        msgstart = ptr;
@@ -222,11 +270,11 @@ void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, St
                 * the tag to be removed.
                 */
                if ( (!strncasecmp(ptr, "/HTML", 5))
-                  ||(!strncasecmp(ptr, "/BODY", 5)) ) {
+                               ||(!strncasecmp(ptr, "/BODY", 5)) ) {
                        --ptr;
                        msgend = ptr;
                        strcpy(ptr, "");
-                       
+
                }
 
                ++ptr;
@@ -242,14 +290,14 @@ void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, St
        /** Convert foreign character sets to UTF-8 if necessary. */
 #ifdef HAVE_ICONV
        if ( (strcasecmp(charset, "us-ascii"))
-          && (strcasecmp(charset, "UTF-8"))
-          && (strcasecmp(charset, ""))
-       ) {
+                       && (strcasecmp(charset, "UTF-8"))
+                       && (strcasecmp(charset, ""))
+          ) {
                lprintf(9, "Converting %s to UTF-8\n", 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));
+                                       __FILE__, __LINE__, strerror(errno));
                }
        }
        if  (Source == NULL) {
@@ -276,7 +324,7 @@ void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, St
                        msg = (char*)ChrPtr(Source); /* TODO: get rid of this. */
                }
        }
-               
+
 #endif
 
        /**
@@ -292,6 +340,11 @@ void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, St
                goto BAIL;
        }
 
+       if (BodyArea != NULL) {
+               StrBufAppendBufPlain(converted_msg, HKEY("<table "), 0);  
+               StrBufAppendBuf(converted_msg, BodyArea, 0);
+               StrBufAppendBufPlain(converted_msg, HKEY(" width=\"100%\"><tr><td>"), 0);
+       }
        ptr = msg;
        msgend = strchr(msg, 0);
        while (ptr < msgend) {
@@ -316,7 +369,7 @@ void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, St
                if (!strncasecmp(ptr, "<a href=\"mailto:", 16)) {
                        content_length += 64;
                        StrBufAppendPrintf(converted_msg,
-                               "<a href=\"display_enter?force_room=_MAIL_&recp=");
+                                       "<a href=\"display_enter?force_room=_MAIL_?recp=");
                        ptr = &ptr[16];
                        ++alevel;
                        ++brak;
@@ -326,8 +379,8 @@ void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, St
                        ++alevel;
                        ++brak;
                        if ( ((strchr(ptr, ':') < strchr(ptr, '/')))
-                            &&  ((strchr(ptr, '/') < strchr(ptr, '>'))) 
-                            ) {
+                                       &&  ((strchr(ptr, '/') < strchr(ptr, '>'))) 
+                          ) {
                                /* open external links to new window */
                                StrBufAppendPrintf(converted_msg, new_window);
                                ptr = &ptr[8];
@@ -344,27 +397,38 @@ void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, St
                }
                /** 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;
+                       char* tag_end=strchr(ptr,'>');
+                       char* src;
+                       /* FIXME - handle this situation (maybe someone opened an <img cid... 
+                        * and then ended the message)
+                        */
+                       if (!tag_end) {
+                               lprintf(9, "tag_end is null and ptr is:\n");
+                               lprintf(9, "%s\n", ptr);
+                               lprintf(9, "Theoretical bytes remaining: %d\n", msgend - ptr);
+                       }
 
-                       if (src && 
-                               (cid_start=strchr(src,':')) && 
-                               (cid_end=strchr(cid_start,'"')) &&
-                               (cid_end < tag_end)) {
+                       src=strstr(ptr, " src=\"cid:");
+                       ++brak;
 
+                       if (src
+                               && tag_end
+                               && (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 */
+                                  trailing / stops dumb URL filters getting excited */
                                StrBufAppendPrintf(converted_msg,
-                                       "src=\"/webcit/mimepart/%d/",msgnum);
+                                               " 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);
@@ -411,7 +475,7 @@ void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, St
                                        char *ltreviewptr;
                                        char *nbspreviewptr;
                                        char linkedchar;
-                                       int len = linklen;
+                                       int len;
                                        
                                        len = linklen;
                                        linkedchar = ptr[len];
@@ -447,22 +511,30 @@ void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, St
                        ptr++;
                }
 
-               /**
-                * We need to know when we're inside a tag,
-                * so we don't turn things that look like URL's into
-                * links, when they're already links - or image sources.
-                */
-               if (*(ptr-1) == '<') {
-                       ++brak;
-               }
-               if (*(ptr-1) == '>') {
-                       --brak;
-                       if ((scriptlevel == 0) && (script_start_pos >= 0)) {
-                               StrBufCutRight(converted_msg, StrLength(converted_msg) - script_start_pos);
-                               script_start_pos = (-1);
+
+               if ((ptr >= msg) && (ptr <= msgend)) {
+                       /*
+                        * We need to know when we're inside a tag,
+                        * so we don't turn things that look like URL's into
+                        * links, when they're already links - or image sources.
+                        */
+                       if ((ptr > msg) && (*(ptr-1) == '<')) {
+                               ++brak;
+                       }
+                       if ((ptr > msg) && (*(ptr-1) == '>')) {
+                               --brak;
+                               if ((scriptlevel == 0) && (script_start_pos >= 0)) {
+                                       StrBufCutRight(converted_msg, StrLength(converted_msg) - script_start_pos);
+                                       script_start_pos = (-1);
+                               }
                        }
+                       if (!strncasecmp(ptr, "</A>", 3)) --alevel;
                }
-               if (!strncasecmp(ptr, "</A>", 3)) --alevel;
+       }
+
+       if (BodyArea != NULL) {
+               StrBufAppendBufPlain(converted_msg, HKEY("</td></tr></table>"), 0);  
+               FreeStrBuf(&BodyArea);
        }
 
        /**     uncomment these two lines to override conversion        */
@@ -542,7 +614,7 @@ void UrlizeText(StrBuf* Target, StrBuf *Source, StrBuf *WrkBuf)
                           LB, QU, ChrPtr(WrkBuf), QU, QU, TARGET, 
                           QU, RB, ChrPtr(WrkBuf), LB, RB);
 
-       TrailerLen = len - (end - start);
+       TrailerLen = StrLength(Source) - (end - ChrPtr(Source));
        if (TrailerLen > 0)
                StrBufAppendBufPlain(Target, end, TrailerLen, 0);
 }