1 /* PSPPIRE - a graphical user interface for PSPP.
2 Copyright (C) 2008, 2009, 2010, 2011, 2012, 2013, 2015 Free Software Foundation
4 This program is free software: you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation, either version 3 of the License, or
7 (at your option) any later version.
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program. If not, see <http://www.gnu.org/licenses/>. */
19 #include "page-separators.h"
21 #include "ui/gui/text-data-import-dialog.h"
30 #include "data/data-in.h"
31 #include "data/data-out.h"
32 #include "data/format-guesser.h"
33 #include "data/value-labels.h"
34 #include "language/data-io/data-parser.h"
35 #include "language/lexer/lexer.h"
36 #include "libpspp/assertion.h"
37 #include "libpspp/i18n.h"
38 #include "libpspp/line-reader.h"
39 #include "libpspp/message.h"
40 #include "ui/gui/dialog-common.h"
41 #include "ui/gui/executor.h"
42 #include "ui/gui/helper.h"
43 #include "ui/gui/builder-wrapper.h"
44 #include "ui/gui/psppire-data-window.h"
45 #include "ui/gui/psppire-dialog.h"
46 #include "ui/gui/psppire-encoding-selector.h"
47 #include "ui/gui/psppire-empty-list-store.h"
48 #include "ui/gui/psppire-var-sheet.h"
49 #include "ui/gui/psppire-scanf.h"
50 #include "ui/syntax-gen.h"
52 #include "gl/intprops.h"
53 #include "gl/xalloc.h"
56 #define _(msgid) gettext (msgid)
57 #define N_(msgid) msgid
59 /* Page where the user chooses field separators. */
60 struct separators_page
62 /* How to break lines into columns. */
63 struct string separators; /* Field separators. */
64 struct string quotes; /* Quote characters. */
68 GtkWidget *custom_entry;
70 GtkWidget *quote_combo;
71 GtkEntry *quote_entry;
72 PsppSheetView *fields_tree_view;
75 /* The "separators" page of the assistant. */
77 static void revise_fields_preview (struct import_assistant *ia);
78 static void choose_likely_separators (struct import_assistant *ia);
79 static void find_commonest_chars (unsigned long int histogram[UCHAR_MAX + 1],
80 const char *targets, const char *def,
81 struct string *result);
82 static void clear_fields (struct import_assistant *ia);
83 static void revise_fields_preview (struct import_assistant *);
84 static void set_separators (struct import_assistant *);
85 static void get_separators (struct import_assistant *);
86 static void on_separators_custom_entry_notify (GObject *UNUSED,
88 struct import_assistant *);
89 static void on_separators_custom_cb_toggle (GtkToggleButton *custom_cb,
90 struct import_assistant *);
91 static void on_quote_combo_change (GtkComboBox *combo,
92 struct import_assistant *);
93 static void on_quote_cb_toggle (GtkToggleButton *quote_cb,
94 struct import_assistant *);
95 static void on_separator_toggle (GtkToggleButton *, struct import_assistant *);
97 /* A common field separator and its identifying name. */
100 const char *name; /* Name (for use with get_widget_assert). */
101 int c; /* Separator character. */
104 /* All the separators in the dialog box. */
105 static const struct separator separators[] =
117 #define SEPARATOR_CNT (sizeof separators / sizeof *separators)
120 set_quote_list (GtkComboBox *cb)
122 GtkListStore *list = gtk_list_store_new (1, G_TYPE_STRING);
125 const gchar *seperator[3] = {"'\"", "\'", "\""};
127 for (i = 0; i < 3; i++)
129 const gchar *s = seperator[i];
131 /* Add a new row to the model */
132 gtk_list_store_append (list, &iter);
133 gtk_list_store_set (list, &iter,
139 gtk_combo_box_set_model (GTK_COMBO_BOX (cb), GTK_TREE_MODEL (list));
140 g_object_unref (list);
142 gtk_combo_box_set_entry_text_column (cb, 0);
145 /* Initializes IA's separators substructure. */
147 struct separators_page *
148 separators_page_create (struct import_assistant *ia)
150 GtkBuilder *builder = ia->asst.builder;
154 struct separators_page *p = xzalloc (sizeof *p);
156 p->page = add_page_to_assistant (ia, get_widget_assert (builder, "Separators"),
157 GTK_ASSISTANT_PAGE_CONTENT);
159 p->custom_cb = get_widget_assert (builder, "custom-cb");
160 p->custom_entry = get_widget_assert (builder, "custom-entry");
161 p->quote_combo = get_widget_assert (builder, "quote-combo");
162 p->quote_entry = GTK_ENTRY (gtk_bin_get_child (GTK_BIN (p->quote_combo)));
163 p->quote_cb = get_widget_assert (builder, "quote-cb");
165 set_quote_list (GTK_COMBO_BOX (p->quote_combo));
166 p->fields_tree_view = PSPP_SHEET_VIEW (get_widget_assert (builder, "fields"));
167 g_signal_connect (p->quote_combo, "changed",
168 G_CALLBACK (on_quote_combo_change), ia);
169 g_signal_connect (p->quote_cb, "toggled",
170 G_CALLBACK (on_quote_cb_toggle), ia);
171 g_signal_connect (p->custom_entry, "notify::text",
172 G_CALLBACK (on_separators_custom_entry_notify), ia);
173 g_signal_connect (p->custom_cb, "toggled",
174 G_CALLBACK (on_separators_custom_cb_toggle), ia);
175 for (i = 0; i < SEPARATOR_CNT; i++)
176 g_signal_connect (get_widget_assert (builder, separators[i].name),
177 "toggled", G_CALLBACK (on_separator_toggle), ia);
182 /* Frees IA's separators substructure. */
184 destroy_separators_page (struct import_assistant *ia)
186 struct separators_page *s = ia->separators;
188 ds_destroy (&s->separators);
189 ds_destroy (&s->quotes);
193 /* Called just before the separators page becomes visible in the
196 prepare_separators_page (struct import_assistant *ia)
198 revise_fields_preview (ia);
201 /* Called when the Reset button is clicked on the separators
202 page, resets the separators to the defaults. */
204 reset_separators_page (struct import_assistant *ia)
206 choose_likely_separators (ia);
210 /* Frees and clears the column data in IA's separators
213 clear_fields (struct import_assistant *ia)
215 if (ia->column_cnt > 0)
220 for (row = 0; row < ia->file.line_cnt; row++)
222 const struct string *line = &ia->file.lines[row];
223 const char *line_start = ds_data (line);
224 const char *line_end = ds_end (line);
226 for (col = ia->columns; col < &ia->columns[ia->column_cnt]; col++)
228 char *s = ss_data (col->contents[row]);
229 if (!(s >= line_start && s <= line_end))
230 ss_dealloc (&col->contents[row]);
234 for (col = ia->columns; col < &ia->columns[ia->column_cnt]; col++)
237 free (col->contents);
246 /* Breaks the file data in IA into columns based on the
247 separators set in IA's separators substructure. */
249 split_fields (struct import_assistant *ia)
251 struct separators_page *s = ia->separators;
252 size_t columns_allocated;
258 /* Is space in the set of separators? */
259 space_sep = ss_find_byte (ds_ss (&s->separators), ' ') != SIZE_MAX;
261 /* Split all the lines, not just those from
262 ia->first_line.skip_lines on, so that we split the line that
263 contains variables names if ia->first_line.variable_names is
265 columns_allocated = 0;
266 for (row = 0; row < ia->file.line_cnt; row++)
268 struct string *line = &ia->file.lines[row];
269 struct substring text = ds_ss (line);
272 for (column_idx = 0; ; column_idx++)
274 struct substring field;
275 struct column *column;
278 ss_ltrim (&text, ss_cstr (" "));
279 if (ss_is_empty (text))
285 else if (!ds_is_empty (&s->quotes)
286 && ds_find_byte (&s->quotes, text.string[0]) != SIZE_MAX)
288 int quote = ss_get_byte (&text);
293 while ((c = ss_get_byte (&text)) != EOF)
296 else if (ss_match_byte (&text, quote))
297 ds_put_byte (&s, quote);
303 ss_get_bytes (&text, ss_cspan (text, ds_ss (&s->separators)),
306 if (column_idx >= ia->column_cnt)
308 struct column *column;
310 if (ia->column_cnt >= columns_allocated)
311 ia->columns = x2nrealloc (ia->columns, &columns_allocated,
312 sizeof *ia->columns);
313 column = &ia->columns[ia->column_cnt++];
316 column->contents = xcalloc (ia->file.line_cnt,
317 sizeof *column->contents);
319 column = &ia->columns[column_idx];
320 column->contents[row] = field;
321 if (ss_length (field) > column->width)
322 column->width = ss_length (field);
325 ss_ltrim (&text, ss_cstr (" "));
326 if (ss_is_empty (text))
328 if (ss_find_byte (ds_ss (&s->separators), ss_first (text))
330 ss_advance (&text, 1);
335 /* Chooses a name for each column on the separators page */
337 choose_column_names (struct import_assistant *ia)
339 struct dictionary *dict;
340 unsigned long int generated_name_count = 0;
344 dict = dict_create (get_default_encoding ());
345 name_row = ia->variable_names && ia->skip_lines ? ia->skip_lines : 0;
346 for (col = ia->columns; col < &ia->columns[ia->column_cnt]; col++)
350 hint = name_row ? ss_xstrdup (col->contents[name_row - 1]) : NULL;
351 name = dict_make_unique_var_name (dict, hint, &generated_name_count);
355 dict_create_var_assert (dict, name, 0);
360 /* Picks the most likely separator and quote characters based on
363 choose_likely_separators (struct import_assistant *ia)
365 unsigned long int histogram[UCHAR_MAX + 1] = { 0 };
368 /* Construct a histogram of all the characters used in the
370 for (row = 0; row < ia->file.line_cnt; row++)
372 struct substring line = ds_ss (&ia->file.lines[row]);
373 size_t length = ss_length (line);
375 for (i = 0; i < length; i++)
376 histogram[(unsigned char) line.string[i]]++;
379 find_commonest_chars (histogram, "\"'", "", &ia->separators->quotes);
380 find_commonest_chars (histogram, ",;:/|!\t-", ",", &ia->separators->separators);
383 /* Chooses the most common character among those in TARGETS,
384 based on the frequency data in HISTOGRAM, and stores it in
385 RESULT. If there is a tie for the most common character among
386 those in TARGETS, the earliest character is chosen. If none
387 of the TARGETS appear at all, then DEF is used as a
390 find_commonest_chars (unsigned long int histogram[UCHAR_MAX + 1],
391 const char *targets, const char *def,
392 struct string *result)
394 unsigned char max = 0;
395 unsigned long int max_count = 0;
397 for (; *targets != '\0'; targets++)
399 unsigned char c = *targets;
400 unsigned long int count = histogram[c];
401 if (count > max_count)
410 ds_put_byte (result, max);
413 ds_assign_cstr (result, def);
416 /* Revises the contents of the fields tree view based on the
417 currently chosen set of separators. */
419 revise_fields_preview (struct import_assistant *ia)
423 push_watch_cursor (ia);
425 w = GTK_WIDGET (ia->separators->fields_tree_view);
426 gtk_widget_destroy (w);
429 choose_column_names (ia);
430 ia->separators->fields_tree_view = create_data_tree_view (
432 GTK_CONTAINER (get_widget_assert (ia->asst.builder, "fields-scroller")),
435 pop_watch_cursor (ia);
438 /* Sets the widgets to match IA's separators substructure. */
440 set_separators (struct import_assistant *ia)
442 struct separators_page *s = ia->separators;
444 struct string custom;
449 ds_init_empty (&custom);
451 for (i = 0; i < ds_length (&s->separators); i++)
453 unsigned char c = ds_at (&s->separators, i);
456 for (j = 0; j < SEPARATOR_CNT; j++)
458 const struct separator *s = &separators[j];
466 ds_put_byte (&custom, c);
470 for (i = 0; i < SEPARATOR_CNT; i++)
472 const struct separator *s = &separators[i];
473 GtkWidget *button = get_widget_assert (ia->asst.builder, s->name);
474 gtk_toggle_button_set_active (GTK_TOGGLE_BUTTON (button),
475 (seps & (1u << i)) != 0);
477 any_custom = !ds_is_empty (&custom);
478 gtk_entry_set_text (GTK_ENTRY (s->custom_entry), ds_cstr (&custom));
479 gtk_toggle_button_set_active (GTK_TOGGLE_BUTTON (s->custom_cb),
481 gtk_widget_set_sensitive (s->custom_entry, any_custom);
482 ds_destroy (&custom);
484 any_quotes = !ds_is_empty (&s->quotes);
486 gtk_entry_set_text (s->quote_entry,
487 any_quotes ? ds_cstr (&s->quotes) : "\"");
488 gtk_toggle_button_set_active (GTK_TOGGLE_BUTTON (s->quote_cb),
490 gtk_widget_set_sensitive (s->quote_combo, any_quotes);
493 /* Sets IA's separators substructure to match the widgets. */
495 get_separators (struct import_assistant *ia)
497 struct separators_page *s = ia->separators;
500 ds_clear (&s->separators);
501 for (i = 0; i < SEPARATOR_CNT; i++)
503 const struct separator *sep = &separators[i];
504 GtkWidget *button = get_widget_assert (ia->asst.builder, sep->name);
505 if (gtk_toggle_button_get_active (GTK_TOGGLE_BUTTON (button)))
506 ds_put_byte (&s->separators, sep->c);
509 if (gtk_toggle_button_get_active (GTK_TOGGLE_BUTTON (s->custom_cb)))
510 ds_put_cstr (&s->separators,
511 gtk_entry_get_text (GTK_ENTRY (s->custom_entry)));
513 if (gtk_toggle_button_get_active (GTK_TOGGLE_BUTTON (s->quote_cb)))
515 const gchar *text = gtk_entry_get_text (GTK_ENTRY (gtk_bin_get_child (GTK_BIN (s->quote_combo))));
516 ds_assign_cstr (&s->quotes, text);
519 ds_clear (&s->quotes);
522 /* Called when the user changes the entry field for custom
525 on_separators_custom_entry_notify (GObject *gobject UNUSED,
526 GParamSpec *arg1 UNUSED,
527 struct import_assistant *ia)
529 revise_fields_preview (ia);
532 /* Called when the user toggles the checkbox that enables custom
535 on_separators_custom_cb_toggle (GtkToggleButton *custom_cb,
536 struct import_assistant *ia)
538 bool is_active = gtk_toggle_button_get_active (custom_cb);
539 gtk_widget_set_sensitive (ia->separators->custom_entry, is_active);
540 revise_fields_preview (ia);
543 /* Called when the user changes the selection in the combo box
544 that selects a quote character. */
546 on_quote_combo_change (GtkComboBox *combo, struct import_assistant *ia)
548 revise_fields_preview (ia);
551 /* Called when the user toggles the checkbox that enables
554 on_quote_cb_toggle (GtkToggleButton *quote_cb, struct import_assistant *ia)
556 bool is_active = gtk_toggle_button_get_active (quote_cb);
557 gtk_widget_set_sensitive (ia->separators->quote_combo, is_active);
558 revise_fields_preview (ia);
561 /* Called when the user toggles one of the separators
564 on_separator_toggle (GtkToggleButton *toggle UNUSED,
565 struct import_assistant *ia)
567 revise_fields_preview (ia);
573 separators_append_syntax (const struct import_assistant *ia, struct string *s)
576 ds_put_cstr (s, " /DELIMITERS=\"");
577 if (ds_find_byte (&ia->separators->separators, '\t') != SIZE_MAX)
578 ds_put_cstr (s, "\\t");
579 if (ds_find_byte (&ia->separators->separators, '\\') != SIZE_MAX)
580 ds_put_cstr (s, "\\\\");
581 for (i = 0; i < ds_length (&ia->separators->separators); i++)
583 char c = ds_at (&ia->separators->separators, i);
585 ds_put_cstr (s, "\"\"");
586 else if (c != '\t' && c != '\\')
589 ds_put_cstr (s, "\"\n");
590 if (!ds_is_empty (&ia->separators->quotes))
591 syntax_gen_pspp (s, " /QUALIFIER=%sq\n", ds_cstr (&ia->separators->quotes));