stardict-tools/stardict-tools-3.0.1/src/cedict.cpp

   1 #include "stdio.h"
   2 #include "stdlib.h"
   3 #include <string.h>
   4 #include <sys/stat.h>
   5
   6
   7 #include <gtk/gtk.h>
   8 #include <glib.h>
   9
  10 struct _worditem
  11 {
  12         gchar *word;
  13         gchar *pinyin;
  14         gchar *definition;
  15 };
  16
  17 gint stardict_strcmp(const gchar *s1, const gchar *s2)
  18 {
  19         gint a;
  20         a = g_ascii_strcasecmp(s1, s2);
  21         if (a == 0)
  22                 return strcmp(s1, s2);
  23         else
  24                 return a;
  25 }
  26
  27 gint comparefunc(gconstpointer a,gconstpointer b)
  28 {
  29         gint x;
  30         x = stardict_strcmp(((struct _worditem *)a)->word,((struct _worditem *)b)->word);
  31         if (x == 0)
  32                 return ((struct _worditem *)a)->definition - ((struct _worditem *)b)->definition;
  33         else
  34                 return x;
  35 }
  36
  37 void to_pinyin(gchar *str)
  38 {
  39         gchar *p1 = str;
  40         gchar *p2 = str;
  41         while (*p1) {
  42                 if (*p1 == 'u' && *(p1+1) == ':') {
  43                         *p2 = 'v';
  44                         p2++;
  45                         p1+=2;
  46                         continue;
  47                 } else {
  48                         *p2 = *p1;
  49                         p2++;
  50                         p1++;
  51                         continue;
  52                 }
  53         }
  54         *p2 = '\0';
  55 }
  56
  57 void to_definition(gchar *str)
  58 {
  59         while (*str) {
  60                 if (*str=='/')
  61                         *str='\n';
  62                 str++;
  63         }
  64 }
  65
  66 void convert(char *filename)
  67 {
  68         struct stat stats;
  69         if (stat (filename, &stats) == -1)
  70         {
  71                 printf("file not exist!\n");
  72                 return;
  73         }
  74         gchar *basefilename = g_path_get_basename(filename);
  75         FILE *tabfile;
  76         tabfile = fopen(filename,"r");
  77
  78         gchar *buffer = (gchar *)g_malloc (stats.st_size + 1);
  79         fread (buffer, 1, stats.st_size, tabfile);
  80         fclose (tabfile);
  81         buffer[stats.st_size] = '\0';
  82
  83         GArray *array = g_array_sized_new(FALSE,FALSE, sizeof(struct _worditem),20000);
  84
  85         gchar *p, *p1, *p2, *p3;
  86         p = buffer;
  87         if ((guchar)*p==0xEF && (guchar)*(p+1)==0xBB && (guchar)*(p+2)==0xBF) // UTF-8 order characters.
  88                 p+=3;
  89         struct _worditem worditem;
  90         glong linenum=1;
  91         while (1) {
  92                 if (*p == '\0') {
  93                         g_print("over\n");
  94                         break;
  95                 }
  96                 p1 = strchr(p,'\n');
  97                 if (!p1) {
  98                         g_print("error, no end line\n");
  99                         return;
 100                 }
 101                 *p1 = '\0';
 102                 p1++;
 103                 p2 = strchr(p,'[');
 104                 if (!p2) {
 105                         g_print("error, no [, %ld\n", linenum);
 106                         return;
 107                 }
 108                 *p2 = '\0';
 109                 p2++;
 110                 p3 = strchr(p2, ']');
 111                 if (!p3) {
 112                         g_print("error, no ], %ld\n", linenum);
 113                         return;
 114                 }
 115                 *p3 = '\0';
 116                 p3++;
 117                 worditem.word = p;
 118                 to_pinyin(p2);
 119                 worditem.pinyin = p2;
 120                 to_definition(p3);
 121                 worditem.definition = p3;
 122                 g_strstrip(worditem.word);
 123                 g_strstrip(worditem.pinyin);
 124                 g_strstrip(worditem.definition);
 125                 if (!worditem.word[0]) {
 126                         g_print("%s-%ld, bad word!!!\n", basefilename, linenum);
 127                         p= p1;
 128                         linenum++;
 129                         continue;
 130                 }
 131                 if (!worditem.pinyin[0]) {
 132                         g_print("%s-%ld, bad pinyin!!!\n", basefilename, linenum);
 133                 }
 134                 if (!worditem.definition[0]) {
 135                         g_print("%s-%ld, bad definition!!!\n", basefilename, linenum);
 136                 }
 137                 if (!worditem.pinyin[0] && !worditem.definition[0]) {
 138                         g_print("%s-%ld, bad pinyin and definition!!!\n", basefilename, linenum);
 139                         p= p1;
 140                         linenum++;
 141                         continue;
 142                 }
 143                 g_array_append_val(array, worditem);
 144                 p= p1;
 145                 linenum++;
 146         }
 147         g_array_sort(array,comparefunc);
 148
 149         gchar idxfilename[256];
 150         gchar dicfilename[256];
 151         sprintf(idxfilename, "%s.idx", basefilename);
 152         sprintf(dicfilename, "%s.dict", basefilename);
 153         FILE *idxfile = fopen(idxfilename,"w");
 154         FILE *dicfile = fopen(dicfilename,"w");
 155
 156         guint32 offset_old;
 157         guint32 tmpglong;
 158         struct _worditem *pworditem;
 159         gint pinyin_len;
 160         gint definition_len;
 161         gulong i;
 162         for (i=0; i< array->len; i++) {
 163                 offset_old = ftell(dicfile);
 164                 pworditem = &g_array_index(array, struct _worditem, i);
 165                 pinyin_len = strlen(pworditem->pinyin);
 166                 fwrite(pworditem->pinyin, 1 , pinyin_len+1,dicfile);
 167                 definition_len = strlen(pworditem->definition);
 168                 fwrite(pworditem->definition, 1 ,definition_len,dicfile);
 169                 fwrite(pworditem->word,sizeof(gchar),strlen(pworditem->word)+1,idxfile);
 170                 tmpglong = g_htonl(offset_old);
 171                 fwrite(&(tmpglong),sizeof(guint32),1,idxfile);
 172                 tmpglong = g_htonl(pinyin_len+1+ definition_len);
 173                 fwrite(&(tmpglong),sizeof(guint32),1,idxfile);
 174         }
 175         fclose(idxfile);
 176         fclose(dicfile);
 177         g_print("%s wordcount: %d\n", basefilename, array->len);
 178
 179         g_free(buffer);
 180         g_array_free(array,TRUE);
 181
 182         gchar command[256];
 183         sprintf(command, "dictzip %s.dict", basefilename);
 184         system(command);
 185
 186         g_free(basefilename);
 187 }
 188
 189 int main(int argc,char * argv [])
 190 {
 191         if (argc<2) {
 192                 printf("please type this:\n./cedict cedict.gb.utf8\n");
 193                 return FALSE;
 194         }
 195
 196         gtk_set_locale ();
 197         g_type_init ();
 198         for (int i=1; i< argc; i++)
 199                 convert (argv[i]);
 200         return FALSE;
 201 }
 202