Branch: refs/heads/master Author: Jonathan Michalon jonathan@michalon.eu Committer: Colomban Wendling ban@herbesfolles.org Date: Sat, 04 Aug 2012 15:08:40 Commit: cb1f5acfe4d8e1d2c26f62d36f66b7fbc7ee7813 https://github.com/geany/geany-plugins/commit/cb1f5acfe4d8e1d2c26f62d36f66b7...

Log Message: ----------- commander: Switch to a better sort algorithm

Modified Paths: -------------- commander/src/commander-plugin.c

Modified: commander/src/commander-plugin.c 97 files changed, 48 insertions(+), 49 deletions(-) =================================================================== @@ -78,56 +78,55 @@ enum { };

-/* TODO: improve this algorithm for better matches. - * what's needed is probably not to simply eat a letter when it matches, but - * rather give it a chance to provide an heavier match... whatever, be smart. */ +#define SEPARATORS " -_/\"'" +#define IS_SEPARATOR(c) (strchr (SEPARATORS, (c))) +#define next_separator(p) (strpbrk (p, SEPARATORS)) + +/* TODO: be more tolerant regarding unmatched character in the needle. + * Right now, we implicitly accept unmatched characters at the end of the + * needle but absolutely not at the start. e.g. "xpy" won't match "python" at + * all, though "pyx" will. */ +static inline gint +get_score (const gchar *needle, + const gchar *haystack) +{ + if (needle == NULL || haystack == NULL || + *needle == '\0' || *haystack == '\0') { + return 0; + } + + if (IS_SEPARATOR (*haystack)) { + return get_score (needle, haystack + 1); + } + + if (IS_SEPARATOR (*needle)) { + return get_score (needle + 1, next_separator (haystack)); + } + + if (*needle == *haystack) { + gint a = get_score (needle + 1, haystack + 1) + 1; + gint b = get_score (needle, next_separator (haystack)); + + return MAX (a, b); + } else { + return get_score (needle, next_separator (haystack)); + } +} + static gint -key_dist (const gchar *key_, - const gchar *text_) +key_score (const gchar *key_, + const gchar *text_) { gchar *text = g_utf8_casefold (text_, -1); gchar *key = g_utf8_casefold (key_, -1); - gint dist = 0; - -#if 0 - const gchar *text_p = text; - const gchar *key_p = key; - gint weight = 0; - - while (*key_p && *text_p) { - if (*text_p == *key_p) { - text_p ++; - key_p ++; - weight ++; - dist -= weight * 2; - } else { - text_p ++; - dist ++; - weight = 0; - } -#else /* rather search end to start because end of paths are more interesting */ - const gchar *text_p = text + strlen (text) - 1; - const gchar *key_p = key + strlen (key) - 1; - gint weight = 0; - - while (key_p >= key && text_p >= text) { - if (*text_p == *key_p) { - text_p --; - key_p --; - weight ++; - dist -= weight * 2; - } else { - text_p --; - dist ++; - weight = 0; - } - } -#endif + gint score; + + score = get_score (key, text);

g_free (text); g_free (key);

- return dist; + return score; }

static const gchar * @@ -364,8 +363,8 @@ enum { GtkTreeIter *b, gpointer dummy) { - gint dista; - gint distb; + gint scorea; + gint scoreb; gchar *patha; gchar *pathb; gint typea; @@ -376,20 +375,20 @@ enum { gtk_tree_model_get (model, a, COL_PATH, &patha, COL_TYPE, &typea, -1); gtk_tree_model_get (model, b, COL_PATH, &pathb, COL_TYPE, &typeb, -1);

- dista = key_dist (key, patha); - distb = key_dist (key, pathb); + scorea = key_score (key, patha); + scoreb = key_score (key, pathb);

if (! (typea & type)) { - dista += 0xf000; + scorea -= 0xf000; } if (! (typeb & type)) { - distb += 0xf000; + scoreb -= 0xf000; }

g_free (patha); g_free (pathb);

- return dista - distb; + return scoreb - scorea; }

static gboolean

@@ Diff output truncated at 100000 characters. @@

-------------- This E-Mail was brought to you by github_commit_mail.py (Source: TBD).