Changeset 1233

Timestamp:

2000-06-23T15:21:39+12:00 (24 years ago)

Author:

sjboddie

Message:

Created converter classes for simple 8 bit encodings that use a
simple textual map file. Instances of these classes are used to handle
the Windows 1256 (Arabic) encoding.

Location:

trunk/gsdl

Files:

• lib/gsdlunicode.cpp (modified) (3 diffs)
• lib/gsdlunicode.h (modified) (1 diff)
• src/recpt/librarymain.cpp (modified) (2 diffs)

trunk/gsdl/lib/gsdlunicode.cpp

@@ -28 +28 @@
 /*
    $Log$
+   Revision 1.13  2000/06/23 03:21:38  sjboddie
+   Created converter classes for simple 8 bit encodings that use a
+   simple textual map file. Instances of these classes are used to handle
+   the Windows 1256 (Arabic) encoding.
+
    Revision 1.12  2000/04/06 19:58:02  cs025
    Correcting a correction - reinstated all lib files due to silly
@@ -70 +75 @@
 #include <stdio.h>
 
+#if defined(GSDL_USE_OBJECTSPACE)
+#  include <ospace\std\iostream>
+#  include <ospace\std\fstream>
+#elif defined(GSDL_USE_IOS_H)
+#  include <iostream.h>
+#  include <fstream.h>
+#else
+#  include <iostream>
+#  include <fstream>
+#endif
 
 
@@ -551 +566 @@
   else status = unfinished;
 }
+
+
+bool simplemapconvert::loadmapfile (bool in) {
+  if (loaded) return true;
+  if (mapfile.empty()) return false;
+
+  char *cfilename = mapfile.getcstr();
+#ifdef GSDL_USE_IOS_H
+  ifstream mapfilein (cfilename, ios::in | ios::nocreate);
+#else
+  ifstream mapfilein (cfilename, ios::in);
+#endif
+  delete cfilename;
+  if (!mapfilein) return false;
+
+  char cline[2048];
+  text_t line;
+
+  while (!mapfilein.eof()) {
+    mapfilein.getline (cline, 2048);
+    line.clear();
+    line.appendcstr (cline);
+    if (line.empty()) continue;
+    // remove comments
+    text_t::iterator end = line.end();
+    text_t::iterator here = findchar (line.begin(), end, '#');
+    if (here != end) {
+      line.erase (here, end);
+      if (line.empty()) continue;
+    }
+    
+    text_tarray parts;
+    splitchar (line.begin(), line.end(), '\t', parts);
+    
+    // do some simple sanity checks
+    if (parts.size() < 2) continue;
+    text_t::iterator begin1 = parts[0].begin();
+    text_t::iterator begin2 = parts[1].begin();
+    if (*begin1 != '0' || *(begin1+1) != 'x') continue;
+    if (*begin2 != '0' || *(begin2+1) != 'x') continue;
+    char *from = parts[0].getcstr();
+    char *to = parts[1].getcstr();
+    unsigned short f = 0, t = 0;
+    sscanf (from, "%i", &f);
+    sscanf (to, "%i", &t);
+    delete from;
+    delete to;
+    
+    if (in) mapping[f] = t;
+    else mapping[t] = f;
+  }
+
+  loaded = true;
+  return true;
+}
+
+unsigned short simplemapconvert::convert (unsigned short c, bool in) {
+
+  if (!loaded) 
+    if (!loadmapfile(in)) return absentc;
+  
+  return mapping[c];
+}
+
+
+void simplemapinconvertclass::convert (text_t &output, status_t &status) {
+  output.clear();
+  
+  if (start == NULL || len == 0) {
+    status = finished;
+    return;
+  }
+
+  // don't want any funny sign conversions happening
+  unsigned char *here = (unsigned char *)start;
+  while (len > 0) {
+
+    if (*here < 0x80)
+      output.push_back (*here); // append this character
+    else 
+      output.push_back (converter.convert(*here, true));
+
+    ++here;
+    --len;
+  }
+
+  start = (char *)here; // save current position
+  status = finished;
+}
+
+
+void simplemapoutconvertclass::convert (char *output, size_t maxlen, 
+                    size_t &len, status_t &status) {
+
+  if (input == NULL || output == NULL) {
+    status = finished;
+    return;
+  }
+
+  // don't want any funny sign conversions happening
+  unsigned char *uoutput = (unsigned char *)output;
+  text_t::iterator textend = input->end();
+  len = 0;
+  while ((len < maxlen) && (texthere != textend)) {
+
+    if (*texthere < 0x80) *uoutput = (unsigned char)(*texthere);
+    else *uoutput = converter.convert (*texthere, false);
+
+    ++uoutput;
+    ++len;
+    ++texthere;
+  }
+  
+  if (texthere == textend) status = finished;
+  else status = unfinished;
+}

trunk/gsdl/lib/gsdlunicode.h

@@ -215 +215 @@
 };
 
+
+// Simple input and output converter classes for use with 8 bit encodings
+// using simple textual map files. Map files should contain (at least) two
+// tab-separated fields. The first field is the mapped value and the second
+// field is the unicode value.
+
+struct ltus_t 
+{
+  bool operator()(const unsigned short &t1, const unsigned short &t2) const
+  { return t1 < t2; }
+};
+
+
+class simplemapconvert {
+public:
+  simplemapconvert () {absentc=0; loaded=false;}
+  unsigned short convert (unsigned short c, bool in);
+  void setmapfile (const text_t &themapfile) {mapfile = themapfile;}
+
+protected:
+  bool loadmapfile (bool in);
+
+  map <unsigned short, unsigned short, ltus_t> mapping;
+  bool loaded;
+  text_t mapfile;
+  unsigned short absentc;
+};
+
+
+class simplemapinconvertclass : public inconvertclass {
+public:
+  void convert (text_t &output, status_t &status);
+
+  void setmapfile (const text_t &themapfile) {converter.setmapfile(themapfile);}
+  
+protected:
+  simplemapconvert converter;
+};
+
+class simplemapoutconvertclass : public rzwsoutconvertclass {
+public:
+  void convert (char *output, size_t maxlen, 
+        size_t &len, status_t &status);
+
+  void setmapfile (const text_t &themapfile) {converter.setmapfile(themapfile);}
+  
+protected:
+  simplemapconvert converter;
+};
+
+
+
+
 #endif

trunk/gsdl/src/recpt/librarymain.cpp

@@ -28 +28 @@
 /*
    $Log$
+   Revision 1.27  2000/06/23 03:21:39  sjboddie
+   Created converter classes for simple 8 bit encodings that use a
+   simple textual map file. Instances of these classes are used to handle
+   the Windows 1256 (Arabic) encoding.
+
    Revision 1.26  2000/05/29 03:30:03  sjboddie
    fixed a bug preventing GB encoded text from being displayed correctly
@@ -230 +235 @@
   recpt.add_converter ("g", &gbinconvert, &gboutconvert);
 
+  text_t armapfile = filename_cat (gsdlhome, "unicode", "MAPPINGS");
+  armapfile = filename_cat (armapfile, "WINDOWS", "1256.TXT");
+  simplemapinconvertclass arinconvert;
+  arinconvert.setmapfile (armapfile);
+  simplemapoutconvertclass aroutconvert;
+  aroutconvert.setmapfile (armapfile);
+  recpt.add_converter ("a", &arinconvert, &aroutconvert);  
+
+
   // the list of actions. Note: these actions will become invalid
   // at the end of this function.