Added e-table/e-table-sorting-utils.lo.

2001-02-23  Christopher James Lahey  <clahey@ximian.com>

	* gal/Makefile.am (libgal_la_LIBADD): Added
	e-table/e-table-sorting-utils.lo.

From gal/e-table/ChangeLog:

2001-02-23  Christopher James Lahey  <clahey@ximian.com>

	* Makefile.am (libetable_la_SOURCES): Added
	e-table-sorting-utils.c.
	(libetableinclude_HEADERS): Added e-table-sorting-utils.h.

	* e-table-group-container.c: Implemented add_array method.  Use
	add_array to implement add_all instead of using add.  Did some
	general clean up.

	* e-table-group-leaf.c, e-table-group-leaf.h: Implemented
	add_array method.  Changed everything to use an ETableSubset
	variable instead of an ETableSortedVariable.  Implemented it as
	ETableSortedVariable if there's grouping involved and ETableSorted
	if there isn't.

	* e-table-group.c, e-table-group.h: Added add_array method.  Added
	unused get_position code.

	* e-table-sorted-variable.c, e-table-sorted-variable.h: Moved a
	lot of the actual implementation of sorting to
	e-table-sorting-utils.c.  Got rid of the signal handlers here
	since we weren't using them and ETableSubset has the proxy virtual
	methods.

	* e-table-sorted.c, e-table-sorted.h: Completely replaced these
	files with a variation on ETableSortedVariable that doesn't have
	the add methods and such but instead just mimics the source
	model's set of rows completely.

	* e-table-sorter.c, e-table-sorter.h: Added functions to get at
	the arrays here.

	* e-table-sorting-utils.c, e-table-sorting-utils.h: Utility
	functions for doing sorting.

	* e-table-subset-variable.c, e-table-subset-variable.h: Added
	add_array method.

	* e-table-subset.c, e-table-subset.h: Made it so that the signal
	hookups to the source model's signals are virtual methods.

	* e-table.c, e-table.h: Don't call add_all and add and remove type
	functions if not grouped.  Don't rebuild for sort changes unless
	it was either grouped before or will be grouped now.

svn path=/trunk/; revision=8374

1 files changed, 372 insertions, 0 deletions

diff --git a/widgets/table/e-table-sorting-utils.c b/widgets/table/e-table-sorting-utils.c
new file mode 100644
index 0000000000..c5f58b8795
--- /dev/null
+++ b/widgets/table/e-table-sorting-utils.c
@@ -0,0 +1,372 @@
+/* -*- Mode: C; tab-width: 8; indent-tabs-mode: t; c-basic-offset: 8 -*- */
+
+#include <string.h>
+#include <e-table-sorting-utils.h>
+
+#define d(x)
+
+static ETableSortInfo *sort_info_closure;
+
+static void **vals_closure;
+static int cols_closure;
+static int *ascending_closure;
+static GCompareFunc *compare_closure;
+
+/* FIXME: Make it not cache the second and later columns (as if anyone cares.) */
+
+static int
+qsort_callback(const void *data1, const void *data2)
+{
+	gint row1 = *(int *)data1;
+	gint row2 = *(int *)data2;
+	int j;
+	int sort_count = e_table_sort_info_sorting_get_count(sort_info_closure);
+	int comp_val = 0;
+	int ascending = 1;
+	for (j = 0; j < sort_count; j++) {
+		comp_val = (*(compare_closure[j]))(vals_closure[cols_closure * row1 + j], vals_closure[cols_closure * row2 + j]);
+		ascending = ascending_closure[j];
+		if (comp_val != 0)
+			break;
+	}
+	if (comp_val == 0) {
+		if (row1 < row2)
+			comp_val = -1;
+		if (row1 > row2)
+			comp_val = 1;
+	}
+	if (!ascending)
+		comp_val = -comp_val;
+	return comp_val;
+}
+
+struct _subinfo {
+	int start;
+	GArray *rowsort;	/* an array of row info's */
+};
+
+struct _rowinfo {
+	int row;
+	struct _subinfo *subinfo;
+	struct _group_info *groupinfo;
+};
+
+static int
+qsort_callback_complex(const void *data1, const void *data2)
+{
+	gint row1 = ((struct _rowinfo *)data1)->row;
+	gint row2 = ((struct _rowinfo *)data2)->row;
+	int j;
+	int sort_count = e_table_sort_info_sorting_get_count(sort_info_closure);
+	int comp_val = 0;
+	int ascending = 1;
+	for (j = 0; j < sort_count; j++) {
+		comp_val = (*(compare_closure[j]))(vals_closure[cols_closure * row1 + j], vals_closure[cols_closure * row2 + j]);
+		ascending = ascending_closure[j];
+		if (comp_val != 0)
+			break;
+	}
+	if (comp_val == 0) {
+		if (row1 < row2)
+			comp_val = -1;
+		if (row1 > row2)
+			comp_val = 1;
+	}
+	if (!ascending)
+		comp_val = -comp_val;
+	return comp_val;
+}
+
+/* if sortgroup is like:
+0   1  	     1 	1
+1   1	     2 	  2
+2    2	     3 	  2
+3    2	     4 	   3
+4     3	     5 	  3
+5    2 	     6 	1
+6   1	     0	1
+
+   Want to sort the 1's first
+   Then sort each group of 2's, offsetting into the output by the new root 1 location
+   ... Recursively ...
+*/
+
+struct _group_info {
+	char *group;
+	int row;
+};
+
+#ifdef DEBUG
+#undef DEBUG
+#endif
+/*#define DEBUG*/
+
+#ifdef DEBUG
+static int total=0;
+static int total_sorted=0;
+#endif
+
+/* builds the info needed to sort everything */
+static struct _subinfo *
+etsu_sort_build_subset(int rows, struct _group_info *groupinfo, int start, int *end)
+{
+	int i, lastinsert;
+	GArray *rowsort = g_array_new(0, 0, sizeof(struct _rowinfo));
+	struct _subinfo *subinfo, *newsub;
+	char *id, *newid;
+	int idlen, newidlen;
+	int cmp;
+	int cmplen;
+
+	subinfo = g_malloc0(sizeof(*subinfo));
+	subinfo->rowsort = rowsort;
+	subinfo->start = start;
+	lastinsert = -1;
+	id = groupinfo[start].group;
+	newid = strrchr(id, '/');
+	idlen = strlen(id);
+	if (newid)
+		cmplen = newid-id;
+	else
+		cmplen = idlen;
+	d(printf("%d scanning level %s\n", start, id));
+	for (i=start;i<rows;i++) {
+		newid = groupinfo[i].group;
+		newidlen = strlen(newid);
+		d(printf("%d checking group %s\n", start, newid));
+		cmp = strncmp(id, newid, cmplen);
+		/* check for common parent */
+		if (idlen == newidlen && cmp == 0) {
+			struct _rowinfo rowinfo;
+
+			d(printf("%d Same parent\n", start));
+			rowinfo.row = groupinfo[i].row;
+			rowinfo.subinfo = NULL;
+			rowinfo.groupinfo = &groupinfo[i];
+			lastinsert = rowsort->len;
+			g_array_append_val(rowsort, rowinfo);
+#ifdef DEBUG
+			total++;
+#endif
+		} else if (newidlen > idlen) {
+			/* must be a new subtree */
+			d(printf("%d checking subtree instead\n", start));
+			newsub = etsu_sort_build_subset(rows, groupinfo, i, &i);
+			d(printf("found %d nodes in subtree\n", newsub->rowsort->len));
+			g_array_index(rowsort, struct _rowinfo, lastinsert).subinfo = newsub;
+		} else {
+			i--;
+			break;
+		}
+	}
+	if (end)
+		*end = i;
+	d(printf("finished level %s start was %d end was %d\n", id, start, i));
+	return subinfo;
+}
+
+/* sort each level, and then sort each level below that level (once we know
+   where the sublevel will fit in the overall list) */
+static int
+etsu_sort_subset(int *map_table, struct _subinfo *subinfo, int startoffset)
+{
+	GArray *rowsort = subinfo->rowsort;
+	int offset, i;
+
+	d(printf("sorting subset start %d rows %d\n", startoffset, rowsort->len));
+
+	/* first, sort the actual data */
+	qsort(rowsort->data, rowsort->len, sizeof(struct _rowinfo), qsort_callback_complex);
+
+	/* then put it back in the map table, where appropriate */
+	offset = startoffset;
+	for (i=0;i<rowsort->len;i++) {
+		struct _rowinfo *rowinfo;
+
+		d(printf("setting offset %d\n", offset));
+
+		rowinfo = &g_array_index(rowsort, struct _rowinfo, i);
+		map_table[offset] = rowinfo->row;
+		if (rowinfo->subinfo) {
+			offset = etsu_sort_subset(map_table, rowinfo->subinfo, offset+1);
+		} else
+			offset += 1;
+	}
+	d(printf("end sort subset start %d\n", startoffset));
+
+	return offset;
+}
+
+static void
+etsu_sort_free_subset(struct _subinfo *subinfo)
+{
+	int i;
+
+	for (i=0;i<subinfo->rowsort->len;i++) {
+		struct _rowinfo *rowinfo;
+
+		rowinfo = &g_array_index(subinfo->rowsort, struct _rowinfo, i);
+		if (rowinfo->subinfo)
+			etsu_sort_free_subset(rowinfo->subinfo);
+	}
+	g_array_free(subinfo->rowsort, TRUE);
+	g_free(subinfo);
+}
+
+static int
+sort_groups_compare(const void *ap, const void *bp)
+{
+	struct _group_info *a = (struct _group_info *)ap;
+	struct _group_info *b = (struct _group_info *)bp;
+
+	return strcmp(a->group, b->group);
+}
+
+#ifdef DEBUG
+static void
+print_id(int key, int val, void *data)
+{
+	printf("gained id %d\n", key);
+}
+#endif
+
+/* use the sort group to select subsorts */
+static void
+etsu_sort_by_group(ETableModel *source, int *map_table, int rows)
+{
+	struct _group_info *groups;
+	struct _subinfo *subinfo;
+	int i;
+#ifdef DEBUG
+	GHashTable *members = g_hash_table_new(0, 0);
+
+	total = 0;
+	total_sorted = 0;
+#endif
+
+	d(printf("sorting %d rows\n", rows));
+
+	if (rows == 0)
+		return;
+
+	/* get the subset rows */
+	groups = g_malloc(sizeof(struct _group_info) * rows);
+	for (i=0;i<rows;i++) {
+		groups[i].row = map_table[i];
+		groups[i].group = g_strdup(e_table_model_row_sort_group(source, groups[i].row));
+#ifdef DEBUG
+		g_hash_table_insert(members, map_table[i], 1);
+		map_table[i] = 0;
+#endif
+	}
+
+	/* sort the group info */
+	qsort(groups, rows, sizeof(struct _group_info), sort_groups_compare);
+
+	d(printf("sorted groups:\n");
+	for (i=0;i<rows;i++) {
+		printf(" %s\n", groups[i].group);
+	});
+	
+	/* now sort based on the group info */
+	subinfo = etsu_sort_build_subset(rows, groups, 0, NULL);
+	for (i=0;i<rows;i++) {
+		g_free(groups[i].group);
+	}
+	g_free(groups);
+	etsu_sort_subset(map_table, subinfo, 0);
+	etsu_sort_free_subset(subinfo);
+#ifdef DEBUG
+	for (i=0;i<rows;i++) {
+		if (g_hash_table_lookup(members, map_table[i]) == 0) {
+			printf("lost id %d\n", map_table[i]);
+		}
+		g_hash_table_remove(members, map_table[i]);
+	}
+	g_hash_table_foreach(members, print_id, 0);
+
+	printf("total rows = %d, total processed = %d, total sorted = %d\n", rows, total, total_sorted);
+#endif
+
+}
+
+void
+e_table_sorting_utils_sort(ETableModel *source, ETableSortInfo *sort_info, ETableHeader *full_header, int *map_table, int rows)
+{
+	int total_rows;
+	int i;
+	int j;
+	int cols;
+
+	g_return_if_fail(source != NULL);
+	g_return_if_fail(E_IS_TABLE_MODEL(source));
+	g_return_if_fail(sort_info != NULL);
+	g_return_if_fail(E_IS_TABLE_SORT_INFO(sort_info));
+	g_return_if_fail(full_header != NULL);
+	g_return_if_fail(E_IS_TABLE_HEADER(full_header));
+
+	total_rows = e_table_model_row_count(source);
+	cols = e_table_sort_info_sorting_get_count(sort_info);
+	cols_closure = cols;
+	vals_closure = g_new(void *, total_rows * cols);
+	sort_info_closure = sort_info;
+	ascending_closure = g_new(int, cols);
+	compare_closure = g_new(GCompareFunc, cols);
+	for (j = 0; j < cols; j++) {
+		ETableSortColumn column = e_table_sort_info_sorting_get_nth(sort_info, j);
+		ETableCol *col;
+		col = e_table_header_get_column_by_col_idx(full_header, column.column);
+		if (col == NULL)
+			col = e_table_header_get_column (full_header, e_table_header_count (full_header) - 1);
+		for (i = 0; i < rows; i++) {
+#if 0
+			if( !(i & 0xff) ) {
+				while(gtk_events_pending())
+					gtk_main_iteration();
+			}
+#endif
+			vals_closure[map_table[i] * cols + j] = e_table_model_value_at (source, col->col_idx, map_table[i]);
+		}
+		compare_closure[j] = col->compare;
+		ascending_closure[j] = column.ascending;
+	}
+
+	if (e_table_model_has_sort_group(source)) {
+		etsu_sort_by_group(source, map_table, rows);
+	} else {
+		qsort(map_table, rows, sizeof(int), qsort_callback);
+	}
+	g_free(vals_closure);
+	g_free(ascending_closure);
+	g_free(compare_closure);
+}
+
+gboolean
+e_table_sorting_utils_affects_sort  (ETableModel    *source,
+				     ETableSortInfo *sort_info,
+				     ETableHeader   *full_header,
+				     int             col)
+{
+	int j;
+	int cols;
+
+	g_return_val_if_fail(source != NULL, TRUE);
+	g_return_val_if_fail(E_IS_TABLE_MODEL(source), TRUE);
+	g_return_val_if_fail(sort_info != NULL, TRUE);
+	g_return_val_if_fail(E_IS_TABLE_SORT_INFO(sort_info), TRUE);
+	g_return_val_if_fail(full_header != NULL, TRUE);
+	g_return_val_if_fail(E_IS_TABLE_HEADER(full_header), TRUE);
+
+	cols = e_table_sort_info_sorting_get_count(sort_info);
+
+	for (j = 0; j < cols; j++) {
+		ETableSortColumn column = e_table_sort_info_sorting_get_nth(sort_info, j);
+		ETableCol *tablecol;
+		tablecol = e_table_header_get_column_by_col_idx(full_header, column.column);
+		if (tablecol == NULL)
+			tablecol = e_table_header_get_column (full_header, e_table_header_count (full_header) - 1);
+		if (col == tablecol->col_idx)
+			return TRUE;
+	}
+	return FALSE;
+}