plugins/dif/dif.c

   1 /*
   2  * dif.c: read/write sheets using a DIF encoding.
   3  *
   4  * Authors:
   5  *   Kevin Handy <kth@srv.net>
   6  *   Zbigniew Chyla <cyba@gnome.pl>
   7  *
   8  *      Based on ff-csv code.
   9  */
  10 #include <gnumeric-config.h>
  11 #include <glib/gi18n-lib.h>
  12 #include <gnumeric.h>
  13 #include <libgnumeric.h>
  14
  15 #include <cell.h>
  16 #include <sheet.h>
  17 #include <value.h>
  18 #include <numbers.h>
  19 #include <gutils.h>
  20 #include <goffice/goffice.h>
  21 #include <workbook-view.h>
  22 #include <workbook.h>
  23 #include <gnm-plugin.h>
  24
  25 #include <gsf/gsf-input-textline.h>
  26 #include <gsf/gsf-output.h>
  27 #include <gsf/gsf-utils.h>
  28 #include <string.h>
  29 #include <stdlib.h>
  30
  31 GNM_PLUGIN_MODULE_HEADER;
  32
  33 #define N_INPUT_LINES_BETWEEN_UPDATES   50
  34
  35 void dif_file_open (GOFileOpener const *fo, GOIOContext *io_context,
  36                     WorkbookView *wbv, GsfInput *input);
  37 void dif_file_save (GOFileSaver const *fs, GOIOContext *io_context,
  38                     WorkbookView const *wbv, GsfOutput *out);
  39
  40 typedef struct {
  41         GOIOContext *io_context;
  42
  43         GsfInputTextline *input;
  44         gint   line_no;
  45         gsize  line_len;
  46         gchar *line;
  47
  48         Sheet *sheet;
  49
  50         GIConv converter;
  51 } DifInputContext;
  52
  53
  54 static DifInputContext *
  55 dif_input_context_new (GOIOContext *io_context, Workbook *wb, GsfInput *input)
  56 {
  57         DifInputContext *ctxt = NULL;
  58
  59         ctxt = g_new (DifInputContext, 1);
  60         ctxt->io_context     = io_context;
  61
  62         ctxt->input          = (GsfInputTextline *) gsf_input_textline_new (input);
  63
  64         ctxt->line_no        = 1;
  65         ctxt->line           = NULL;
  66         ctxt->sheet          = workbook_sheet_add (wb, -1, GNM_DEFAULT_COLS, GNM_DEFAULT_ROWS);
  67         ctxt->converter      = g_iconv_open ("UTF-8", "ISO-8859-1");
  68
  69         go_io_progress_message (io_context, _("Reading file..."));
  70
  71         return ctxt;
  72 }
  73
  74 static void
  75 dif_input_context_destroy (DifInputContext *ctxt)
  76 {
  77         go_io_progress_unset (ctxt->io_context);
  78         g_object_unref (ctxt->input); ctxt->input = NULL;
  79         gsf_iconv_close (ctxt->converter);
  80         g_free (ctxt->line);
  81         g_free (ctxt);
  82 }
  83
  84 static gboolean
  85 dif_get_line (DifInputContext *ctxt)
  86 {
  87         char *raw;
  88
  89         if (NULL == (raw = gsf_input_textline_ascii_gets (ctxt->input)))
  90                 return FALSE;
  91
  92         g_free (ctxt->line);
  93         ctxt->line = g_convert_with_iconv (raw, -1, ctxt->converter,
  94                                            NULL, &ctxt->line_len, NULL);
  95
  96         ctxt->line_no++;
  97         return ctxt->line != NULL;
  98 }
  99
 100 /*
 101  * Raturns FALSE on EOF.
 102  */
 103 static gboolean
 104 dif_parse_header (DifInputContext *ctxt)
 105 {
 106         while (1) {
 107                 gchar *topic = NULL, *num_line = NULL, *str_line = NULL;
 108                 size_t str_line_len;
 109                 int res = -1;
 110
 111                 if (!dif_get_line (ctxt)) {
 112                         res = FALSE;
 113                         goto out;
 114                 }
 115                 topic = g_strdup (ctxt->line);
 116
 117                 if (!dif_get_line (ctxt)) {
 118                         res = FALSE;
 119                         goto out;
 120                 }
 121                 num_line = g_strdup (ctxt->line);
 122
 123                 if (!dif_get_line (ctxt)) {
 124                         res = FALSE;
 125                         goto out;
 126                 }
 127                 str_line = g_strdup (ctxt->line);
 128                 str_line_len = strlen (str_line);
 129
 130                 if (strcmp (topic, "TABLE") == 0) {
 131                         gchar *name;
 132
 133                         if (str_line_len > 2 && str_line[0] == '"' && str_line[str_line_len - 1] == '"') {
 134                                 str_line[str_line_len - 1] = '\0';
 135                                 name = str_line + 1;
 136                         }  else {
 137                                 name = str_line;
 138                         }
 139                         if (name[0] != '\0') {
 140                                 /* FIXME - rename the sheet */
 141                         }
 142                 } else if (strcmp (topic, "DATA") == 0) {
 143                         res = TRUE;
 144                 }
 145
 146                 /*
 147                  * Other "standard" header entry items are:
 148                  * SIZE, LABEL, UNITS, TUPLES, VECTORS, COMMENT, MINORSTART,
 149                  * TRUELENGTH, PERIODICITY, DISPLAYUNITS
 150                  */
 151
 152         out:
 153                 g_free (topic);
 154                 g_free (num_line);
 155                 g_free (str_line);
 156                 if (res >= 0)
 157                         return res;
 158         }
 159 }
 160
 161 /*
 162  * Raturns FALSE on EOF.
 163  */
 164 static gboolean
 165 dif_parse_data (DifInputContext *ctxt)
 166 {
 167         gboolean too_many_rows = FALSE, too_many_columns = FALSE;
 168         gint row = -1, col = 0;
 169         gint val_type;
 170         GnmCell *cell;
 171         gchar *msg;
 172
 173         while (1) {
 174                 if (!dif_get_line (ctxt))
 175                         return FALSE;
 176
 177                 val_type = atoi (ctxt->line);
 178                 if (val_type == 0) {
 179                         gchar const *comma = strchr (ctxt->line, ',');
 180                         if (comma == NULL)
 181                                 go_io_warning (ctxt->io_context,
 182                                                 _("Syntax error at line %d. Ignoring."),
 183                                                 ctxt->line_no);
 184                         else if (col > gnm_sheet_get_max_cols (ctxt->sheet)) {
 185                                 too_many_columns = TRUE;
 186                                 break;
 187                         } else {
 188                                 gnm_float num = gnm_strto (comma+1, NULL);
 189                                 GnmValue *v = NULL;
 190
 191                                 if (!dif_get_line (ctxt))
 192                                         return FALSE;
 193
 194                                 if (0 == strcmp (ctxt->line, "V")) {            /* V       value */
 195                                         v = value_new_float (num);
 196                                 } else if (0 == strcmp (ctxt->line, "NA")) {    /* NA      not available res must be O */
 197                                         v = value_new_error_NA (NULL);
 198                                 } else if (0 == strcmp (ctxt->line, "TRUE")) {  /* TRUE    bool T        res must be 1 */
 199                                         v = value_new_bool (TRUE);
 200                                 } else if (0 == strcmp (ctxt->line, "FALSE")) { /* FALSE   bool F        res must be O */
 201                                         v = value_new_bool (TRUE);
 202                                 } else if (0 == strcmp (ctxt->line, "ERROR")) { /* ERROR   err           res must be O */
 203                                         go_io_warning (ctxt->io_context,
 204                                                         _("Unknown value type '%s' at line %d. Ignoring."),
 205                                                         ctxt->line, ctxt->line_no);
 206                                 }
 207
 208                                 if (NULL != v) {
 209                                         cell = sheet_cell_fetch (ctxt->sheet, col, row);
 210                                         gnm_cell_set_value (cell, v);
 211                                 }
 212                                 col++;
 213                         }
 214                 } else if (val_type == 1) {
 215                         if (!dif_get_line (ctxt))
 216                                 return FALSE;
 217                         if (col > gnm_sheet_get_max_cols (ctxt->sheet)) {
 218                                 too_many_columns = TRUE;
 219                                 continue;
 220                         }
 221                         cell = sheet_cell_fetch (ctxt->sheet, col, row);
 222                         if (ctxt->line_len >= 2 &&
 223                             ctxt->line[0] == '"' && ctxt->line[ctxt->line_len - 1] == '"') {
 224                                 ctxt->line[ctxt->line_len - 1] = '\0';
 225                                 gnm_cell_set_text (cell, ctxt->line + 1);
 226                         } else
 227                                 gnm_cell_set_text (cell, ctxt->line);
 228                         col++;
 229                 } else if (val_type == -1) {
 230                         if (!dif_get_line (ctxt))
 231                                 return FALSE;
 232                         if (strcmp (ctxt->line, "BOT") == 0) {
 233                                 col = 0;
 234                                 row++;
 235                                 if (row > gnm_sheet_get_max_rows (ctxt->sheet)) {
 236                                         too_many_rows = TRUE;
 237                                         break;
 238                                 }
 239                         } else if (strcmp (ctxt->line, "EOD") == 0) {
 240                                 break;
 241                         } else {
 242                                 msg = g_strdup_printf (
 243                                       _("Unknown data value \"%s\" at line %d. Ignoring."),
 244                                       ctxt->line, ctxt->line_no);
 245                                 g_warning ("%s", msg);
 246                                 g_free (msg);
 247                         }
 248                 } else {
 249                         msg = g_strdup_printf (
 250                               _("Unknown value type %d at line %d. Ignoring."),
 251                               val_type, ctxt->line_no);
 252                         g_warning ("%s", msg);
 253                         g_free (msg);
 254                         (void) dif_get_line (ctxt);
 255                 }
 256         }
 257
 258         if (too_many_rows) {
 259                 g_warning (_("DIF file has more than the maximum number of rows %d. "
 260                              "Ignoring remaining rows."), gnm_sheet_get_max_rows (ctxt->sheet));
 261         }
 262         if (too_many_columns) {
 263                 g_warning (_("DIF file has more than the maximum number of columns %d. "
 264                              "Ignoring remaining columns."), gnm_sheet_get_max_cols (ctxt->sheet));
 265         }
 266
 267         return TRUE;
 268 }
 269
 270 static void
 271 dif_parse_sheet (DifInputContext *ctxt)
 272 {
 273         GnmLocale *locale = gnm_push_C_locale ();
 274
 275         if (!dif_parse_header (ctxt)) {
 276                 go_io_error_info_set (ctxt->io_context, go_error_info_new_printf (
 277                 _("Unexpected end of file at line %d while reading header."),
 278                 ctxt->line_no));
 279         } else if (!dif_parse_data(ctxt)) {
 280                 go_io_error_info_set (ctxt->io_context, go_error_info_new_printf (
 281                 _("Unexpected end of file at line %d while reading data."),
 282                 ctxt->line_no));
 283         }
 284
 285         gnm_pop_C_locale (locale);
 286 }
 287
 288 void
 289 dif_file_open (GOFileOpener const *fo, GOIOContext *io_context,
 290                WorkbookView *wbv, GsfInput *input)
 291 {
 292         Workbook *wb = wb_view_get_workbook (wbv);
 293         DifInputContext *ctxt = dif_input_context_new (io_context, wb, input);
 294
 295         workbook_set_saveinfo (wb, GO_FILE_FL_MANUAL_REMEMBER,
 296                 go_file_saver_for_id ("Gnumeric_dif:dif"));
 297         if (ctxt != NULL) {
 298                 dif_parse_sheet (ctxt);
 299                 if (go_io_error_occurred (io_context))
 300                         go_io_error_push (io_context,
 301                                 go_error_info_new_str (_("Error while reading DIF file.")));
 302                 dif_input_context_destroy (ctxt);
 303         } else if (!go_io_error_occurred (io_context))
 304                 go_io_error_unknown (io_context);
 305 }
 306
 307 /*
 308  * Write _current_ sheet of the workbook to a DIF format file
 309  */
 310 void
 311 dif_file_save (GOFileSaver const *fs, GOIOContext *io_context,
 312                WorkbookView const *wbv, GsfOutput *out)
 313 {
 314         GnmLocale *locale;
 315         Sheet *sheet;
 316         GnmRange r;
 317         gint row, col;
 318         gboolean ok = TRUE;
 319
 320         sheet = wb_view_cur_sheet (wbv);
 321         if (sheet == NULL) {
 322                 go_io_error_string (io_context, _("Cannot get default sheet."));
 323                 return;
 324         }
 325
 326         r = sheet_get_extent (sheet, FALSE, TRUE);
 327
 328         /* Write out the standard headers */
 329         gsf_output_puts   (out, "TABLE\n"   "0,1\n" "\"GNUMERIC\"\n");
 330         gsf_output_printf (out, "VECTORS\n" "0,%d\n" "\"\"\n", r.end.col+1);
 331         gsf_output_printf (out, "TUPLES\n"  "0,%d\n" "\"\"\n", r.end.row+1);
 332         gsf_output_puts   (out, "DATA\n"    "0,0\n"  "\"\"\n");
 333
 334         locale = gnm_push_C_locale ();
 335
 336         /* Process all cells */
 337         for (row = r.start.row; ok && row <= r.end.row; row++) {
 338                 gsf_output_puts (out, "-1,0\n" "BOT\n");
 339                 for (col = r.start.col; col <= r.end.col; col++) {
 340                         GnmCell *cell = sheet_cell_get (sheet, col, row);
 341                         if (gnm_cell_is_empty (cell)) {
 342                                 gsf_output_puts(out, "1,0\n" "\"\"\n");
 343                         } else if (VALUE_IS_BOOLEAN (cell->value)) {
 344                                 if (value_get_as_checked_bool (cell->value))
 345                                         gsf_output_puts(out, "0,1\n" "TRUE\n");
 346                                 else
 347                                         gsf_output_puts(out, "0,0\n" "FALSE\n");
 348                         } else if (VALUE_IS_ERROR (cell->value)) {
 349                                 if (value_error_classify (cell->value) == GNM_ERROR_NA)
 350                                         gsf_output_puts(out, "0,0\n" "NA\n");
 351                                 else
 352                                         gsf_output_puts(out, "0,0\n" "ERROR\n");
 353                         } else if (VALUE_IS_FLOAT (cell->value))
 354                                 gsf_output_printf (out, "0,%" GNM_FORMAT_g "\n" "V\n",
 355                                         value_get_as_float (cell->value));
 356                         else {
 357                                 gchar *str = gnm_cell_get_rendered_text (cell);
 358                                 ok = gsf_output_printf (out,
 359                                                          "1,0\n" "\"%s\"\n",
 360                                                          str);
 361                                 g_free (str);
 362                         }
 363                 }
 364         }
 365
 366         gsf_output_puts (out, "-1,0\n" "EOD\n");
 367
 368         gnm_pop_C_locale (locale);
 369
 370         if (!ok)
 371                 go_io_error_string (io_context, _("Error while saving DIF file."));
 372 }