// ============================================================================
// C servlet sample for the G-WAN Web Application Server (http://trustleap.ch/)
// ----------------------------------------------------------------------------
// tidy.c: use the 'libtidy' C library to help clean-up/validate HTML pages
//         just like the mod_tidy Apache module.
//
//         Installation of the libtidy library (Debian):
//            sudo apt-get install -y libtidy-dev libtidy-0.99-0
// ----------------------------------------------------------------------------
// Usage: http://127.0.0.1:8080/?tidy&file=index.html
//
// Output:
//    Tidying '/gwan/0.0.0.0_8080/#0.0.0.0/www/index.html'
//    Diagnostics:
//    line  42 column  3 - Warning: missing </a> before <div>
//    line  46 column  3 - Warning: discarding unexpected </a>
//    line  82 column 66 - Warning: unescaped & or unknown entity "&sort_by"
//    line   7 column  1 - Warning: <meta> proprietary attribute "charset"
//    line   7 column  1 - Warning: <meta> lacks "content" attribute
//    line   8 column  1 - Warning: <link> inserting "type" attribute
//    line  30 column  6 - Warning: <img> lacks "alt" attribute
//    line  47 column  3 - Warning: <img> lacks "alt" attribute
//    line 163 column  1 - Warning: <img> lacks "alt" attribute
//    line  65 column  3 - Warning: trimming empty <b>
//    Info: Document content looks like HTML Proprietary
//    10 warnings, 0 errors were found!
// ============================================================================
#include "gwan.h" // G-WAN definitions

#pragma link    "tidy"
#pragma include "/usr/include/tidy/"

#include <tidy.h>
#include <buffio.h>

int main(int argc, char **argv )
{
   xbuf_t *reply = get_reply(argv);

   // -------------------------------------------------------------------------
   // get the URI parameter
   // -------------------------------------------------------------------------
   char *file = ""; get_arg("file=", &file, argc, argv);
   if(!file || !*file)
   {
      xbuf_cat(reply, "<p>usage: http://127.0.0.1:8080/?tidy&file=index.html</p>");
      return HTTP_200_OK;
   }
   
   // -------------------------------------------------------------------------
   // build the local file name   
   // -------------------------------------------------------------------------
   char *wwwpath = (char*)get_env(argv, WWW_ROOT);
   char filename[512] = {0};
   snprintf(filename, sizeof(filename), "%s/%s", wwwpath, file);
   u32 i = 1;
   for(; i < argc; i++) // just in case we have subdirectories
   {
       strcat(filename, "/");
       strcat(filename, argv[i]);
   }
   
   // -------------------------------------------------------------------------
   // try to load the file
   // -------------------------------------------------------------------------
   //printf("loading '%s' file...\n", filename);
   xbuf_t f; xbuf_init(&f);
   xbuf_frfile(&f, filename);
   xbuf_xcat  (reply, "<p>%s file <tt>%s</tt></p>", 
               f.len ? "Tidying" : "Can't find",
               filename);
   //printf("loaded %d bytes\n", f.len);
   if(!f.len) return HTTP_200_OK;

   // -------------------------------------------------------------------------
   // now we have loaded the file, process it with libtidy
   // -------------------------------------------------------------------------
   //puts("running libtidy...");
   TidyBuffer output = {0};
   TidyBuffer errbuf = {0};
   TidyDoc    tdoc   = tidyCreate(); // initialize "document"
   
   int ret = tidyOptSetBool(tdoc, TidyXhtmlOut, no);     // convert to XHTML
   if(ret != 0) ret = tidySetErrorBuffer(tdoc, &errbuf); // capture diagnostics
   if(ret >= 0) ret = tidyParseString(tdoc, f.ptr);      // parse the file
   if(ret >= 0) ret = tidyCleanAndRepair(tdoc);          // tidy it up!
   if(ret >= 0) ret = tidyRunDiagnostics(tdoc);
   if(ret >  1) ret = tidyOptSetBool(tdoc, TidyForceOutput, yes) ? ret : -1;
   if(ret >= 0) ret = tidySaveBuffer(tdoc, &output);
   if(ret >= 0)
   {
      if(ret > 0)
         xbuf_xcat(reply, "<hr><p>Diagnostics:</p><pre>\n%H</pre><hr>", errbuf.bp);
      
      // save the 'enhanced' HTML to a new file for easier comparison
      snprintf(filename, sizeof(filename), "%s/___%s", wwwpath, file);
      f.len = 0;
      xbuf_cat(&f, (char*)output.bp);
      xbuf_tofile(&f, filename);
   }
   else
      xbuf_xcat(reply, "<p>A severe error (%d) occurred.</p><br>", ret);

   // -------------------------------------------------------------------------
   // reformat the tidy reply
   // -------------------------------------------------------------------------
   //puts("reformating...");
   char *p = reply->ptr, 
        *e = reply->ptr + reply->len;
      while(p && *p && p < e)
         p = xbuf_replfrto(reply, p, e, "&lt;br&gt;line", "\n\nline");
   
   p = strstr(reply->ptr, "&lt;br&gt;Info:");
   if(p)
   {
      xbuf_replfrto(reply, p, e, "&lt;br&gt;Info:", "&lt;br&gt;\n\nInfo:");
   
      while(p && *p && p < e)
         p = xbuf_replfrto(reply, p, e, "&lt;br&gt;", "\n");
   }
   p = reply->ptr;
   while(p && *p && p < e)
      p = xbuf_replfrto(reply, p, e, " - ", "\n- ");
      
   // -------------------------------------------------------------------------
   // footer: add link to new file
   // -------------------------------------------------------------------------
   if(ret >= 0)
   {
      xbuf_xcat(reply, "<p>Click <a href=\"%s\"  target=\"_blank\"> here</a>"
      " to see the 'enhanced' <tt>%s</tt> file.</p>", file, file);
   }
   
   // -------------------------------------------------------------------------
   // free Tidy context
   // -------------------------------------------------------------------------
   tidyBufFree(&output);
   tidyBufFree(&errbuf);
   tidyRelease(tdoc);
   
   return HTTP_200_OK; // return an HTTP code
}
// ============================================================================
// End of Source Code
// ============================================================================