src/xrc/xmlexpat.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        xmlexpat.cpp
   3 // Purpose:     wxXmlDocument - XML reader via Expat
   4 // Author:      Vaclav Slavik
   5 // Created:     2001/04/30
   6 // RCS-ID:      $Id$
   7 // Copyright:   (c) 2001 Vaclav Slavik
   8 // Licence:     wxWindows licence
   9 /////////////////////////////////////////////////////////////////////////////
  10
  11 #ifdef __GNUG__
  12 // nothing - already in xml.cpp
  13 #endif
  14
  15 // For compilers that support precompilation, includes "wx.h".
  16 #include "wx/wxprec.h"
  17
  18 #ifdef __BORLANDC__
  19     #pragma hdrstop
  20 #endif
  21
  22 #include "wx/wfstream.h"
  23 #include "wx/intl.h"
  24 #include "wx/log.h"
  25 #include "wx/strconv.h"
  26 #include "wx/xrc/xmlio.h"
  27
  28 #include "xmlparse.h"
  29
  30 /*
  31
  32     FIXME:
  33      - process all elements, including CDATA
  34
  35  */
  36
  37
  38 // converts Expat-produced string in UTF-8 into wxString.
  39 inline static wxString CharToString(wxMBConv *conv,
  40                                     const char *s, size_t len = wxSTRING_MAXLEN)
  41 {
  42 #if wxUSE_UNICODE
  43     (void)conv;
  44     return wxString(s, wxConvUTF8, len);
  45 #else
  46     if ( conv )
  47     {
  48         size_t nLen = (len != wxSTRING_MAXLEN) ? len :
  49                           nLen = wxConvUTF8.MB2WC((wchar_t*) NULL, s, 0);
  50
  51         wchar_t *buf = new wchar_t[nLen+1];
  52         wxConvUTF8.MB2WC(buf, s, nLen);
  53         buf[nLen] = 0;
  54         return wxString(buf, *conv, len);
  55         delete[] buf;
  56     }
  57     else
  58         return wxString(s, len);
  59 #endif
  60 }
  61
  62 bool wxXmlIOHandlerExpat::CanLoad(wxInputStream& stream)
  63 {
  64     char cheader[7];
  65     cheader[6] = 0;
  66     stream.Read(cheader, 6);
  67     stream.SeekI(-6, wxFromCurrent);
  68     return (strcmp(cheader, "<?xml ") == 0);
  69 }
  70
  71
  72 struct wxXmlParsingContext
  73 {
  74     wxMBConv  *conv;
  75
  76     wxXmlNode *root;
  77     wxXmlNode *node;
  78     wxXmlNode *lastAsText;
  79     wxString   encoding;
  80     wxString   version;
  81 };
  82
  83 static void StartElementHnd(void *userData, const char *name, const char **atts)
  84 {
  85     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
  86     wxXmlNode *node = new wxXmlNode(wxXML_ELEMENT_NODE, CharToString(ctx->conv, name));
  87     const char **a = atts;
  88     while (*a)
  89     {
  90         node->AddProperty(CharToString(ctx->conv, a[0]), CharToString(ctx->conv, a[1]));
  91         a += 2;
  92     }
  93     if (ctx->root == NULL)
  94         ctx->root = node;
  95     else
  96         ctx->node->AddChild(node);
  97     ctx->node = node;
  98     ctx->lastAsText = NULL;
  99 }
 100
 101 static void EndElementHnd(void *userData, const char* WXUNUSED(name))
 102 {
 103     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 104
 105     ctx->node = ctx->node->GetParent();
 106     ctx->lastAsText = NULL;
 107 }
 108
 109 static void TextHnd(void *userData, const char *s, int len)
 110 {
 111     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 112     char *buf = new char[len + 1];
 113
 114     buf[len] = '\0';
 115     memcpy(buf, s, (size_t)len);
 116
 117     if (ctx->lastAsText)
 118     {
 119         ctx->lastAsText->SetContent(ctx->lastAsText->GetContent() +
 120                                     CharToString(ctx->conv, buf));
 121     }
 122     else
 123     {
 124         bool whiteOnly = TRUE;
 125         for (char *c = buf; *c != '\0'; c++)
 126             if (*c != ' ' && *c != '\t' && *c != '\n' && *c != '\r')
 127             {
 128                 whiteOnly = FALSE;
 129                 break;
 130             }
 131         if (!whiteOnly)
 132         {
 133             ctx->lastAsText = new wxXmlNode(wxXML_TEXT_NODE, wxT("text"),
 134                                             CharToString(ctx->conv, buf));
 135             ctx->node->AddChild(ctx->lastAsText);
 136         }
 137     }
 138
 139     delete[] buf;
 140 }
 141
 142 static void CommentHnd(void *userData, const char *data)
 143 {
 144     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 145
 146     if (ctx->node)
 147     {
 148         // VS: ctx->node == NULL happens if there is a comment before
 149         //     the root element (e.g. wxDesigner's output). We ignore such
 150         //     comments, no big deal...
 151         ctx->node->AddChild(new wxXmlNode(wxXML_COMMENT_NODE,
 152                             wxT("comment"), CharToString(ctx->conv, data)));
 153     }
 154     ctx->lastAsText = NULL;
 155 }
 156
 157 static void DefaultHnd(void *userData, const char *s, int len)
 158 {
 159     // XML header:
 160     if (len > 6 && memcmp(s, "<?xml ", 6) == 0)
 161     {
 162         wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 163
 164         wxString buf = CharToString(ctx->conv, s, (size_t)len);
 165         int pos;
 166         pos = buf.Find(wxT("encoding="));
 167         if (pos != wxNOT_FOUND)
 168             ctx->encoding = buf.Mid(pos + 10).BeforeFirst(buf[(size_t)pos+9]);
 169         pos = buf.Find(wxT("version="));
 170         if (pos != wxNOT_FOUND)
 171             ctx->version = buf.Mid(pos + 9).BeforeFirst(buf[(size_t)pos+8]);
 172     }
 173 }
 174
 175 static int UnknownEncodingHnd(void * WXUNUSED(encodingHandlerData),
 176                                const XML_Char *name, XML_Encoding *info)
 177 {
 178     // We must build conversion table for expat. The easiest way to do so
 179     // is to let wxCSConv convert as string containing all characters to
 180     // wide character representation:
 181     wxCSConv conv(name);
 182     char mbBuf[255];
 183     wchar_t wcBuf[255];
 184     size_t i;
 185
 186     for (i = 0; i < 255; i++)
 187         mbBuf[i] = i+1;
 188     mbBuf[255] = 0;
 189     conv.MB2WC(wcBuf, mbBuf, 255);
 190     wcBuf[255] = 0;
 191
 192     info->map[0] = 0;
 193     for (i = 0; i < 255; i++)
 194         info->map[i+1] = (int)wcBuf[i];
 195
 196     info->data = NULL;
 197     info->convert = NULL;
 198     info->release = NULL;
 199
 200     return 1;
 201 }
 202
 203 bool wxXmlIOHandlerExpat::Load(wxInputStream& stream, wxXmlDocument& doc,
 204                                const wxString& encoding)
 205 {
 206     const size_t BUFSIZE = 1024;
 207     char buf[BUFSIZE];
 208     wxXmlParsingContext ctx;
 209     bool done;
 210     XML_Parser parser = XML_ParserCreate(NULL);
 211
 212     ctx.root = ctx.node = NULL;
 213     ctx.encoding = wxT("UTF-8"); // default in absence of encoding=""
 214     ctx.conv = NULL;
 215 #if !wxUSE_UNICODE
 216     if ( encoding != wxT("UTF-8") && encoding != wxT("utf-8") )
 217         ctx.conv = new wxCSConv(encoding);
 218 #endif
 219
 220     XML_SetUserData(parser, (void*)&ctx);
 221     XML_SetElementHandler(parser, StartElementHnd, EndElementHnd);
 222     XML_SetCharacterDataHandler(parser, TextHnd);
 223     XML_SetCommentHandler(parser, CommentHnd);
 224     XML_SetDefaultHandler(parser, DefaultHnd);
 225     XML_SetUnknownEncodingHandler(parser, UnknownEncodingHnd, NULL);
 226
 227     do
 228     {
 229         size_t len = stream.Read(buf, BUFSIZE).LastRead();
 230         done = (len < BUFSIZE);
 231         if (!XML_Parse(parser, buf, len, done))
 232         {
 233             wxLogError(_("XML parsing error: '%s' at line %d"),
 234                        XML_ErrorString(XML_GetErrorCode(parser)),
 235                        XML_GetCurrentLineNumber(parser));
 236           return FALSE;
 237         }
 238     } while (!done);
 239
 240     doc.SetVersion(ctx.version);
 241     doc.SetFileEncoding(ctx.encoding);
 242     doc.SetRoot(ctx.root);
 243
 244     XML_ParserFree(parser);
 245 #if !wxUSE_UNICODE
 246     if ( ctx.conv )
 247         delete ctx.conv;
 248 #endif
 249
 250     return TRUE;
 251 }