Use uft8_collate_key() to sort utf8 strings.

Modify file_data_set_path() to sync collate keys and file_data_pool
on path change.

Partially fix bug 1959854.

diff --git a/src/collect.c b/src/collect.c
index 5908f84..62af952 100644 (file)
--- a/src/collect.c
+++ b/src/collect.c
@@ -132,6 +132,8 @@ static gint collection_list_sort_cb(gconstpointer a, gconstpointer b)
 
        switch(collection_list_sort_method)
                {
+               case SORT_NAME:
+                       break;
                case SORT_NONE:
                        return 0;
                        break;
@@ -146,20 +148,21 @@ static gint collection_list_sort_cb(gconstpointer a, gconstpointer b)
                        return 0;
                        break;
                case SORT_PATH:
-                       return CASE_SORT(cia->fd->path, cib->fd->path);
+                       return CASE_SORT(cia->fd->path, cib->fd->path); /* FIXME: utf8_collate */
                        break;
 #ifdef HAVE_STRVERSCMP
                case SORT_NUMBER:
                        return strverscmp(cia->fd->name, cib->fd->name);
                        break;
 #endif
-               case SORT_NAME:
                default:
-                       return CASE_SORT(cia->fd->name, cib->fd->name);
                        break;
                }
 
-       return 0;
+       if (options->file_sort.case_sensitive)
+               return strcmp(cia->fd->collate_key_name, cib->fd->collate_key_name);
+       else
+               return strcmp(cia->fd->collate_key_name_nocase, cib->fd->collate_key_name_nocase);
 }
 
 GList *collection_list_sort(GList *list, SortType method)

diff --git a/src/dupe.c b/src/dupe.c
index d6dd295..85c266e 100644 (file)
--- a/src/dupe.c
+++ b/src/dupe.c
@@ -1132,11 +1132,11 @@ static gint dupe_match(DupeItem *a, DupeItem *b, DupeMatchType mask, gdouble *ra
                }
        if (mask & DUPE_MATCH_NAME)
                {
-               if (strcmp(a->fd->name, b->fd->name) != 0) return FALSE;
+               if (strcmp(a->fd->collate_key_name, b->fd->collate_key_name) != 0) return FALSE;
                }
        if (mask & DUPE_MATCH_NAME_CI)
                {
-               if (strcasecmp(a->fd->name, b->fd->name) != 0) return FALSE;
+               if (strcmp(a->fd->collate_key_name_nocase, b->fd->collate_key_name_nocase) != 0) return FALSE;
                }
        if (mask & DUPE_MATCH_SIZE)
                {

diff --git a/src/filedata.c b/src/filedata.c
index f72efd4..cf004af 100644 (file)
--- a/src/filedata.c
+++ b/src/filedata.c
@@ -22,6 +22,8 @@
 #include "ui_fileops.h"
 
 
+static GHashTable *file_data_pool = NULL;
+
 static gint sidecar_file_priority(const gchar *path);
 
 
@@ -144,15 +146,48 @@ void file_data_increment_version(FileData *fd)
        file_data_send_notification(fd); /* FIXME there are probably situations when we don't want to call this  */
 }
 
+static void file_data_set_collate_keys(FileData *fd)
+{
+       gchar *caseless_name;
+
+       g_assert(g_utf8_validate(fd->name, -1, NULL));
+
+       caseless_name = g_utf8_casefold(fd->name, -1);
+
+       g_free(fd->collate_key_name);
+       g_free(fd->collate_key_name_nocase);
+
+#if GLIB_CHECK_VERSION(2, 8, 0)
+       fd->collate_key_name = g_utf8_collate_key_for_filename(fd->name, -1);
+       fd->collate_key_name_nocase = g_utf8_collate_key_for_filename(caseless_name, -1);
+#else
+       fd->collate_key_name = g_utf8_collate_key(fd->name, -1);
+       fd->collate_key_name_nocase = g_utf8_collate_key(caseless_name, -1);
+#endif
+       g_free(caseless_name);
+}
 
 static void file_data_set_path(FileData *fd, const gchar *path)
 {
+       g_assert(path && *path);
+       g_assert(file_data_pool);
+
+       g_free(fd->path);
+
+       if (fd->original_path)
+               {
+               g_hash_table_remove(file_data_pool, fd->original_path);
+               g_free(fd->original_path);
+               }
+       fd->original_path = g_strdup(path);
+       g_hash_table_insert(file_data_pool, fd->original_path, fd);
 
        if (strcmp(path, G_DIR_SEPARATOR_S) == 0)
                {
                fd->path = g_strdup(path);
                fd->name = fd->path;
                fd->extension = fd->name + 1;
+               file_data_set_collate_keys(fd);
                return;
                }
 
@@ -167,6 +202,7 @@ static void file_data_set_path(FileData *fd, const gchar *path)
                g_free(dir);
                fd->name = "..";
                fd->extension = fd->name + 2;
+               file_data_set_collate_keys(fd);
                return;
                }
        else if (strcmp(fd->name, ".") == 0)
@@ -175,12 +211,15 @@ static void file_data_set_path(FileData *fd, const gchar *path)
                fd->path = remove_level_from_path(path);
                fd->name = ".";
                fd->extension = fd->name + 1;
+               file_data_set_collate_keys(fd);
                return;
                }
 
        fd->extension = extension_from_path(fd->path);
        if (fd->extension == NULL)
                fd->extension = fd->name + strlen(fd->name);
+
+       file_data_set_collate_keys(fd);
 }
 
 static void file_data_check_changed_files(FileData *fd, struct stat *st)
@@ -212,8 +251,6 @@ static void file_data_check_changed_files(FileData *fd, struct stat *st)
                }
 }
 
-static GHashTable *file_data_pool = NULL;
-
 static FileData *file_data_new(const gchar *path_utf8, struct stat *st, gboolean check_sidecars)
 {
        FileData *fd;
@@ -232,10 +269,13 @@ static FileData *file_data_new(const gchar *path_utf8, struct stat *st, gboolean
                }
 
        fd = g_new0(FileData, 1);
+       
+       fd->path = NULL;
+       fd->name = NULL;
+       fd->collate_key_name = NULL;
+       fd->collate_key_name_nocase = NULL;
+       fd->original_path = NULL;
 
-       file_data_set_path(fd, path_utf8);
-
-       fd->original_path = g_strdup(path_utf8);
        fd->size = st->st_size;
        fd->date = st->st_mtime;
        fd->pixbuf = NULL;
@@ -243,10 +283,11 @@ static FileData *file_data_new(const gchar *path_utf8, struct stat *st, gboolean
        fd->ref = 1;
        fd->magick = 0x12345678;
 
-       g_hash_table_insert(file_data_pool, fd->original_path, fd);
+       file_data_set_path(fd, path_utf8); /* set path, name, collate_key_*, original_path */
 
        if (check_sidecars && sidecar_file_priority(fd->extension))
                file_data_check_sidecars(fd);
+
        return fd;
 }
 
@@ -366,6 +407,8 @@ static void file_data_free(FileData *fd)
 
        g_free(fd->path);
        g_free(fd->original_path);
+       g_free(fd->collate_key_name);
+       g_free(fd->collate_key_name_nocase);
        if (fd->pixbuf) g_object_unref(fd->pixbuf);
 
 
@@ -446,7 +489,7 @@ gint file_data_compare_name_without_ext(FileData *fd1, FileData *fd2)
        if (len1 < len2) return -1;
        if (len1 > len2) return 1;
 
-       return strncmp(fd1->name, fd2->name, len1);
+       return strncmp(fd1->name, fd2->name, len1); /* FIXME: utf8 */
 }
 
 gboolean file_data_add_change_info(FileData *fd, FileDataChangeType type, const gchar *src, const gchar *dest)
@@ -541,26 +584,31 @@ gint filelist_sort_compare_filedata(FileData *fa, FileData *fb)
 
        switch (filelist_sort_method)
                {
+               case SORT_NAME:
+                       break;
                case SORT_SIZE:
                        if (fa->size < fb->size) return -1;
                        if (fa->size > fb->size) return 1;
-                       return CASE_SORT(fa->name, fb->name); /* fall back to name */
+                       /* fall back to name */
                        break;
                case SORT_TIME:
                        if (fa->date < fb->date) return -1;
                        if (fa->date > fb->date) return 1;
-                       return CASE_SORT(fa->name, fb->name); /* fall back to name */
+                       /* fall back to name */
                        break;
 #ifdef HAVE_STRVERSCMP
                case SORT_NUMBER:
                        return strverscmp(fa->name, fb->name);
                        break;
 #endif
-               case SORT_NAME:
                default:
-                       return CASE_SORT(fa->name, fb->name);
                        break;
                }
+
+       if (options->file_sort.case_sensitive)
+               return strcmp(fa->collate_key_name, fb->collate_key_name);
+       else
+               return strcmp(fa->collate_key_name_nocase, fb->collate_key_name_nocase);
 }
 
 gint filelist_sort_compare_filedata_full(FileData *fa, FileData *fb, SortType method, gint ascend)
@@ -1356,12 +1404,7 @@ static void file_data_apply_ci(FileData *fd)
        /* FIXME delete ?*/
        if (type == FILEDATA_CHANGE_MOVE || type == FILEDATA_CHANGE_RENAME)
                {
-               g_free(fd->path);
-               g_hash_table_remove(file_data_pool, fd->original_path);
-               g_free(fd->original_path);
                file_data_set_path(fd, fd->change->dest);
-               fd->original_path = g_strdup(fd->change->dest);
-               g_hash_table_insert(file_data_pool, fd->original_path, fd);
                }
        file_data_increment_version(fd);
 }

diff --git a/src/search.c b/src/search.c
index c507c44..304fc45 100644 (file)
--- a/src/search.c
+++ b/src/search.c
@@ -2234,7 +2234,10 @@ static gint search_result_sort_cb(GtkTreeModel *model, GtkTreeIter *a, GtkTreeIt
                        return 0;
                        break;
                case SEARCH_COLUMN_NAME:
-                       return CASE_SORT(fda->fd->name, fdb->fd->name);
+                       if (options->file_sort.case_sensitive)
+                               return strcmp(fda->fd->collate_key_name, fdb->fd->collate_key_name);
+                       else
+                               return strcmp(fda->fd->collate_key_name_nocase, fdb->fd->collate_key_name_nocase);
                        break;
                case SEARCH_COLUMN_SIZE:
                        if (fda->fd->size > fdb->fd->size) return 1;
@@ -2250,7 +2253,7 @@ static gint search_result_sort_cb(GtkTreeModel *model, GtkTreeIter *a, GtkTreeIt
                        return sort_matchdata_dimensions(fda, fdb);
                        break;
                case SEARCH_COLUMN_PATH:
-                       return CASE_SORT(fda->fd->path, fdb->fd->path);
+                       return CASE_SORT(fda->fd->path, fdb->fd->path); /* FIXME: utf8_collate */
                        break;
                default:
                        break;

diff --git a/src/typedefs.h b/src/typedefs.h
index 5a5fefd..3801a0e 100644 (file)
--- a/src/typedefs.h
+++ b/src/typedefs.h
@@ -425,6 +425,8 @@ struct _FileData {
        gchar *path;
        const gchar *name;
        const gchar *extension;
+       gchar *collate_key_name;
+       gchar *collate_key_name_nocase;
        gint64 size;
        time_t date;
        gboolean marks[FILEDATA_MARKS_SIZE];

diff --git a/src/view_dir_tree.c b/src/view_dir_tree.c
index 9c57a1f..4644f2c 100644 (file)
--- a/src/view_dir_tree.c
+++ b/src/view_dir_tree.c
@@ -883,7 +883,10 @@ static gint vdtree_sort_cb(GtkTreeModel *store, GtkTreeIter *a, GtkTreeIter *b,
        gtk_tree_model_get(store, a, DIR_COLUMN_POINTER, &nda, -1);
        gtk_tree_model_get(store, b, DIR_COLUMN_POINTER, &ndb, -1);
 
-       return CASE_SORT(nda->fd->name, ndb->fd->name);
+       if (options->file_sort.case_sensitive)
+               return strcmp(nda->fd->collate_key_name, nda->fd->collate_key_name);
+       else
+               return strcmp(nda->fd->collate_key_name_nocase, nda->fd->collate_key_name_nocase);
 }
 
 /*