#include "psppire-delimited-text.h"
#include "psppire-text-file.h"
+#include "language/data-io/data-parser.h"
#include "libpspp/str.h"
+#include "libpspp/string-array.h"
#include "libpspp/i18n.h"
#include <gtk/gtk.h>
PROP_0,
PROP_CHILD,
PROP_DELIMITERS,
- PROP_QUOTES,
+ PROP_QUOTE,
PROP_FIRST_LINE
};
+static struct data_parser *
+make_data_parser (PsppireDelimitedText *tf)
+{
+ struct data_parser *parser = data_parser_create ();
+ data_parser_set_type (parser, DP_DELIMITED);
+ data_parser_set_span (parser, false);
+ data_parser_set_quotes (parser, ss_empty ());
+ data_parser_set_quote_escape (parser, true);
+ data_parser_set_empty_line_has_field (parser, true);
+
+ bool space = false;
+ struct string hard_delimiters = DS_EMPTY_INITIALIZER;
+ GSList *del;
+ for (del = tf->delimiters; del; del = g_slist_next (del))
+ {
+ gunichar c = GPOINTER_TO_INT (del->data);
+ if (c == ' ')
+ space = true;
+ else
+ ds_put_unichar (&hard_delimiters, c);
+ }
+ data_parser_set_soft_delimiters (parser, ss_cstr (space ? " " : ""));
+ data_parser_set_hard_delimiters (parser, ds_ss (&hard_delimiters));
+ ds_destroy (&hard_delimiters);
+
+ if (tf->quote)
+ {
+ struct string quote = DS_EMPTY_INITIALIZER;
+ ds_put_unichar ("e, tf->quote);
+ data_parser_set_quotes (parser, ds_ss ("e));
+ ds_destroy ("e);
+ }
+ return parser;
+}
+
static void
count_delims (PsppireDelimitedText *tf)
{
if (tf->child == NULL)
return;
- tf->max_delimiters = 0;
+ struct data_parser *parser = make_data_parser (tf);
+
+ tf->max_fields = 0;
GtkTreeIter iter;
gboolean valid;
for (valid = gtk_tree_model_get_iter_first (tf->child, &iter);
valid;
valid = gtk_tree_model_iter_next (tf->child, &iter))
{
- gunichar quote = -1;
- // FIXME: Box these lines to avoid constant allocation/deallocation
gchar *line = NULL;
gtk_tree_model_get (tf->child, &iter, 1, &line, -1);
- {
- char *p;
- gint count = 0;
- for (p = line; ; p = g_utf8_find_next_char (p, NULL))
- {
- const gunichar c = g_utf8_get_char (p);
- if (c == 0)
- break;
-
- if (c == quote)
- quote = -1;
- else if (c == tf->quotes[0] || c == tf->quotes[1])
- quote = c;
-
- if (quote == -1)
- {
- GSList *del;
- for (del = tf->delimiters; del; del = g_slist_next (del))
- {
- if (c == GPOINTER_TO_INT (del->data))
- count++;
- }
- }
- }
- tf->max_delimiters = MAX (tf->max_delimiters, count);
- }
+ size_t n_fields = data_parser_split (parser, ss_cstr (line), NULL);
+ if (n_fields > tf->max_fields)
+ tf->max_fields = n_fields;
g_free (line);
}
+
+ data_parser_destroy (parser);
}
static void
cache_invalidate (PsppireDelimitedText *tf)
{
- memset (tf->cache_starts, 0, sizeof tf->cache_starts);
- if (tf->const_cache.string)
- {
- ss_dealloc (&tf->const_cache);
- tf->const_cache.string = NULL;
- tf->cache_row = -1;
- }
+ tf->cache_row = -1;
+ data_parser_destroy (tf->parser);
+ tf->parser = make_data_parser (tf);
}
static void
g_slist_free (tf->delimiters);
tf->delimiters = g_slist_copy (g_value_get_pointer (value));
break;
- case PROP_QUOTES:
- {
- tf->quotes[0] = tf->quotes[1] = -1;
-
- const gchar *s = g_value_get_string (value);
- for (size_t i = 0; i < 2 && s && s[0]; i++)
- {
- tf->quotes[i] = g_utf8_get_char (s);
- s = g_utf8_find_next_char (s, NULL);
- }
- }
+ case PROP_QUOTE:
+ tf->quote = g_value_get_uint (value);
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
case PROP_DELIMITERS:
g_value_set_pointer (value, text_file->delimiters);
break;
- case PROP_QUOTES:
- {
- GString *s = g_string_new (NULL);
- for (size_t i = 0; i < 2; i++)
- {
- gunichar quote = text_file->quotes[i];
- if (quote && quote != -1)
- g_string_append_unichar (s, quote);
- }
- }
+ case PROP_QUOTE:
+ g_value_set_uint (value, text_file->quote);
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
{
PsppireDelimitedText *tf = PSPPIRE_DELIMITED_TEXT (tree_model);
- /* + 1 for the trailing field and +1 for the leading line number column */
- return tf->max_delimiters + 1 + 1;
+ /* +1 for the leading line number column */
+ return tf->max_fields + 1;
}
return TRUE;
}
-
-static void
-nullify_char (struct substring cs)
-{
- int char_len = ss_first_mblen (cs);
- while (char_len > 0)
- {
- cs.string[char_len - 1] = '\0';
- char_len--;
- }
-}
-
-
/* Split row N into it's delimited fields (if it is not already cached)
and set this row as the current cache. */
static void
split_row_into_fields (PsppireDelimitedText *file, gint n)
{
if (n == file->cache_row) /* Cache hit */
- {
- return;
- }
-
- memset (file->cache_starts, 0, sizeof file->cache_starts);
- /* Cache miss */
- if (file->const_cache.string)
- {
- ss_dealloc (&file->const_cache);
- }
- ss_alloc_substring_pool (&file->const_cache,
- PSPPIRE_TEXT_FILE (file->child)->lines[n], NULL);
- struct substring cs = file->const_cache;
- int field = 0;
- file->cache_starts[0] = cs.string;
- gunichar quote = -1;
- for (;
- UINT32_MAX != ss_first_mb (cs);
- ss_get_mb (&cs))
- {
- ucs4_t character = ss_first_mb (cs);
- gboolean char_is_quote = FALSE;
- if (quote == -1)
- {
- if (character == file->quotes[0] || character == file->quotes[1])
- {
- quote = character;
- char_is_quote = TRUE;
- file->cache_starts[field] += ss_first_mblen (cs);
- }
- }
- else if (character == quote)
- {
- char_is_quote = TRUE;
- nullify_char (cs);
- quote = -1;
- }
-
- if (quote == -1 && char_is_quote == FALSE)
- {
- GSList *del;
- for (del = file->delimiters; del; del = g_slist_next (del))
- {
- if (character == GPOINTER_TO_INT (del->data))
- {
- field++;
- int char_len = ss_first_mblen (cs);
- file->cache_starts[field] = cs.string + char_len;
- nullify_char (cs);
- break;
- }
- }
- }
- }
+ return;
+ if (!file->parser)
+ file->parser = make_data_parser (file);
+ string_array_clear (&file->cache);
+ data_parser_split (file->parser, PSPPIRE_TEXT_FILE (file->child)->lines[n],
+ &file->cache);
file->cache_row = n;
}
split_row_into_fields (file, file->first_line - 1);
- return file->cache_starts [column];
+ return column < file->cache.n ? file->cache.strings[column] : "";
}
static void
split_row_into_fields (file, n);
- g_value_set_string (value, file->cache_starts [column - 1]);
+ size_t idx = column - 1;
+ const char *s = idx < file->cache.n ? file->cache.strings[idx] : "";
+ g_value_set_string (value, s);
}
P_("A GSList of gunichars which delimit the fields."),
G_PARAM_READWRITE);
- GParamSpec *quotes_spec =
- g_param_spec_string ("quotes",
- "Field Quotes",
- P_("A string of characters that quote the fields."),
- P_(""),
+ GParamSpec *quote_spec =
+ g_param_spec_unichar ("quote",
+ "Quote Character",
+ P_("A character that quotes the field, or 0 to disable quoting."),
+ 0,
G_PARAM_READWRITE);
GParamSpec *child_spec =
delimiters_spec);
g_object_class_install_property (object_class,
- PROP_QUOTES,
- quotes_spec);
+ PROP_QUOTE,
+ quote_spec);
g_object_class_install_property (object_class,
PROP_FIRST_LINE,
text_file->first_line = 0;
text_file->delimiters = g_slist_prepend (NULL, GINT_TO_POINTER (':'));
- text_file->const_cache.string = NULL;
- text_file->const_cache.length = 0;
text_file->cache_row = -1;
- memset (text_file->cache_starts, 0, sizeof text_file->cache_starts);
+ string_array_init (&text_file->cache);
+ text_file->parser = NULL;
- text_file->max_delimiters = 0;
+ text_file->max_fields = 0;
- text_file->quotes[0] = text_file->quotes[1] = -1;
+ text_file->quote = 0;
text_file->dispose_has_run = FALSE;
text_file->stamp = g_random_int ();
PsppireDelimitedText *tf = PSPPIRE_DELIMITED_TEXT (object);
g_slist_free (tf->delimiters);
-
- ss_dealloc (&tf->const_cache);
+ string_array_destroy (&tf->cache);
+ data_parser_destroy (tf->parser);
/* must chain up */
(* parent_class->finalize) (object);