feat: add smartcase support like vim (#2060)

config/config.c

@@ -104,6 +104,8 @@ Settings config = {
     .sorting_method = "normal",
     /** Case sensitivity of the search */
     .case_sensitive = FALSE,
+    /** Case smart of the search */
+    .case_smart = FALSE,
     /** Cycle through in the element list */
     .cycle = TRUE,
     /** Height of an element in #chars */

doc/rofi.1.markdown

@@ -246,6 +246,12 @@ exec command. For that case, `#` can be used as a separator.
 Start in case-sensitive mode. This option can be changed at run-time using the
 `-kb-toggle-case-sensitivity` key binding.
 
+`-case-smart`
+
+Start in case-smart mode behave like vim's `smartcase`, which determines
+case-sensitivity by input.  When enabled, this will suppress `-case-sensitive`
+config.
+
 `-cycle`
 
 Cycle through the result list. Default is 'true'.

include/helper.h

@@ -200,13 +200,15 @@ char *rofi_expand_path(const char *input);
  * @param needlelen The length of the needle
  * @param haystack The string to match against
  * @param haystacklen The length of the haystack
+ * @param case_sensitive Whether case is significant.
  *
  * UTF-8 aware levenshtein distance calculation
  *
  * @returns the levenshtein distance between needle and haystack
  */
 unsigned int levenshtein(const char *needle, const glong needlelen,
-                         const char *haystack, const glong haystacklen);
+                         const char *haystack, const glong haystacklen,
+                         const int case_sensitive);
 
 /**
  * @param data the unvalidated character array holding possible UTF-8 data
@@ -234,6 +236,7 @@ char *rofi_latin_to_utf8_strdup(const char *input, gssize length);
  * @param plen      Pattern length.
  * @param str       The input to match against pattern.
  * @param slen      Length of str.
+ * @param case_sensitive Whether case is significant.
  *
  *  rofi_scorer_fuzzy_evaluate implements a global sequence alignment algorithm
  * to find the maximum accumulated score by aligning `pattern` to `str`. It
@@ -263,7 +266,7 @@ char *rofi_latin_to_utf8_strdup(const char *input, gssize length);
  * @returns the sorting weight.
  */
 int rofi_scorer_fuzzy_evaluate(const char *pattern, glong plen, const char *str,
-                               glong slen);
+                               glong slen, const int case_sensitive);
 /*@}*/
 
 /**
@@ -353,6 +356,13 @@ cairo_surface_t *cairo_image_surface_create_from_svg(const gchar *file,
  */
 void parse_ranges(char *input, rofi_range_pair **list, unsigned int *length);
 
+/**
+ * @param input String to parse
+ *
+ * @returns String matching should be case sensitive or insensitive
+ */
+int parse_case_sensitivity(const char *input);
+
 /**
  * @param format The format string used. See below for possible syntax.
  * @param string The selected entry.

include/settings.h

@@ -127,6 +127,8 @@ typedef struct {
 
   /** Search case sensitivity */
   unsigned int case_sensitive;
+  /** Smart case sensitivity like vim */
+  unsigned int case_smart;
   /** Cycle through in the element list */
   unsigned int cycle;
   /** Height of an element in number of rows */

include/view-internal.h

@@ -144,6 +144,8 @@ struct RofiViewState {
 
   /** Regexs used for matching */
   rofi_int_matcher **tokens;
+  /** For case-sensitivity */
+  gboolean case_sensitive;
 };
 /** @} */
 #endif

source/helper.c

@@ -768,7 +768,8 @@ char *rofi_expand_path(const char *input) {
   ((a) < (b) ? ((a) < (c) ? (a) : (c)) : ((b) < (c) ? (b) : (c)))
 
 unsigned int levenshtein(const char *needle, const glong needlelen,
-                         const char *haystack, const glong haystacklen) {
+                         const char *haystack, const glong haystacklen,
+                         int case_sensitive) {
   if (needlelen == G_MAXLONG) {
     // String to long, we cannot handle this.
     return UINT_MAX;
@@ -784,12 +785,12 @@ unsigned int levenshtein(const char *needle, const glong needlelen,
     const char *needles = needle;
     column[0] = x;
     gunichar haystackc = g_utf8_get_char(haystack);
-    if (!config.case_sensitive) {
+    if (!case_sensitive) {
       haystackc = g_unichar_tolower(haystackc);
     }
     for (glong y = 1, lastdiag = x - 1; y <= needlelen; y++) {
       gunichar needlec = g_utf8_get_char(needles);
-      if (!config.case_sensitive) {
+      if (!case_sensitive) {
         needlec = g_unichar_tolower(needlec);
       }
       unsigned int olddiag = column[y];
@@ -916,7 +917,7 @@ static int rofi_scorer_get_score_for(enum CharClass prev, enum CharClass curr) {
 }
 
 int rofi_scorer_fuzzy_evaluate(const char *pattern, glong plen, const char *str,
-                               glong slen) {
+                               glong slen, int case_sensitive) {
   if (slen > FUZZY_SCORER_MAX_LENGTH) {
     return -MIN_SCORE;
   }
@@ -951,9 +952,8 @@ int rofi_scorer_fuzzy_evaluate(const char *pattern, glong plen, const char *str,
       left = dp[si];
       lefts = MAX(lefts + GAP_SCORE, left);
       sc = g_utf8_get_char(sit);
-      if (config.case_sensitive
-              ? pc == sc
-              : g_unichar_tolower(pc) == g_unichar_tolower(sc)) {
+      if (case_sensitive ? pc == sc
+                         : g_unichar_tolower(pc) == g_unichar_tolower(sc)) {
         int t = score[si] * (pstart ? PATTERN_START_MULTIPLIER
                                     : PATTERN_NON_START_MULTIPLIER);
         dp[si] = pfirst ? LEADING_GAP_SCORE * si + t
@@ -1247,6 +1247,28 @@ void parse_ranges(char *input, rofi_range_pair **list, unsigned int *length) {
     }
   }
 }
+
+int parse_case_sensitivity(const char *input) {
+  int case_sensitive = config.case_sensitive;
+  if (config.case_smart) {
+    // By default case is false, unless the search query has a
+    // uppercase in it?
+    case_sensitive = FALSE;
+    const char *end;
+    if (g_utf8_validate(input, -1, &end)) {
+      for (const char *c = (input); !case_sensitive && c != NULL && *c;
+           c = g_utf8_next_char(c)) {
+        gunichar uc = g_utf8_get_char(c);
+        if (g_unichar_isupper(uc)) {
+          case_sensitive = TRUE;
+        }
+      }
+    }
+  }
+
+  return case_sensitive;
+}
+
 void rofi_output_formatted_line(const char *format, const char *string,
                                 int selected_line, const char *filter) {
   for (int i = 0; format && format[i]; i++) {

source/modes/dmenu.c

@@ -954,7 +954,8 @@ int dmenu_mode_dialog(void) {
   char *select = NULL;
   find_arg_str("-select", &select);
   if (select != NULL) {
-    rofi_int_matcher **tokens = helper_tokenize(select, config.case_sensitive);
+    rofi_int_matcher **tokens =
+        helper_tokenize(select, parse_case_sensitivity(select));
     unsigned int i = 0;
     for (i = 0; i < cmd_list_length; i++) {
       if (helper_token_match(tokens, cmd_list[i].entry)) {
@@ -965,8 +966,9 @@ int dmenu_mode_dialog(void) {
     helper_tokenize_free(tokens);
   }
   if (find_arg("-dump") >= 0) {
-    rofi_int_matcher **tokens = helper_tokenize(
-        config.filter ? config.filter : "", config.case_sensitive);
+    char *filter = config.filter ? config.filter : "";
+    rofi_int_matcher **tokens =
+        helper_tokenize(filter, parse_case_sensitivity(filter));
     unsigned int i = 0;
     for (i = 0; i < cmd_list_length; i++) {
       if (tokens == NULL || helper_token_match(tokens, cmd_list[i].entry)) {

source/view.c

@@ -188,8 +188,8 @@ void rofi_view_get_current_monitor(int *width, int *height) {
     *height = CacheState.mon.h;
   }
 }
-static char *get_matching_state(void) {
-  if (config.case_sensitive) {
+static char *get_matching_state(RofiViewState *state) {
+  if (state->case_sensitive) {
     if (config.sort) {
       return "±";
     }
@@ -775,12 +775,13 @@ static void filter_elements(thread_state *ts,
         glong slen = g_utf8_strlen(str, -1);
         switch (config.sorting_method_enum) {
         case SORT_FZF:
-          t->state->distance[i] =
-              rofi_scorer_fuzzy_evaluate(t->pattern, t->plen, str, slen);
+          t->state->distance[i] = rofi_scorer_fuzzy_evaluate(
+              t->pattern, t->plen, str, slen, t->state->case_sensitive);
           break;
         case SORT_NORMAL:
         default:
-          t->state->distance[i] = levenshtein(t->pattern, t->plen, str, slen);
+          t->state->distance[i] = levenshtein(t->pattern, t->plen, str, slen,
+                                              t->state->case_sensitive);
           break;
         }
         g_free(str);
@@ -1482,7 +1483,12 @@ static gboolean rofi_view_refilter_real(RofiViewState *state) {
     unsigned int j = 0;
     gchar *pattern = mode_preprocess_input(state->sw, state->text->text);
     glong plen = pattern ? g_utf8_strlen(pattern, -1) : 0;
-    state->tokens = helper_tokenize(pattern, config.case_sensitive);
+    state->case_sensitive = parse_case_sensitivity(state->text->text);
+    state->tokens = helper_tokenize(pattern, state->case_sensitive);
+
+    if ( config.case_smart && state->case_indicator ) {
+      textbox_text(state->case_indicator, get_matching_state(state));
+    }
     /**
      * On long lists it can be beneficial to parallelize.
      * If number of threads is 1, no thread is spawn.
@@ -1709,7 +1715,7 @@ static void rofi_view_trigger_global_action(KeyBindingAction action) {
     if (state->case_indicator != NULL) {
       config.sort = !config.sort;
       state->refilter = TRUE;
-      textbox_text(state->case_indicator, get_matching_state());
+      textbox_text(state->case_indicator, get_matching_state(state));
     }
     break;
   case MODE_PREVIOUS:
@@ -1739,7 +1745,7 @@ static void rofi_view_trigger_global_action(KeyBindingAction action) {
       config.case_sensitive = !config.case_sensitive;
       (state->selected_line) = 0;
       state->refilter = TRUE;
-      textbox_text(state->case_indicator, get_matching_state());
+      textbox_text(state->case_indicator, get_matching_state(state));
     }
     break;
   // Special delete entry command.
@@ -2397,7 +2403,7 @@ static void rofi_view_add_widget(RofiViewState *state, widget *parent_widget,
                        TB_AUTOWIDTH | TB_AUTOHEIGHT, NORMAL, "*", 0, 0);
     // Add small separator between case indicator and text box.
     box_add((box *)parent_widget, WIDGET(state->case_indicator), FALSE);
-    textbox_text(state->case_indicator, get_matching_state());
+    textbox_text(state->case_indicator, get_matching_state(state));
   }
   /**
    * ENTRY BOX

source/xrmoptions.c

@@ -290,6 +290,12 @@ static XrmOption xrmOptions[] = {
      NULL,
      "Set case-sensitivity",
      CONFIG_DEFAULT},
+    {xrm_Boolean,
+     "case-smart",
+     {.num = &config.case_smart},
+     NULL,
+     "Set smartcase like vim (determine case-sensitivity by input)",
+     CONFIG_DEFAULT},
     {xrm_Boolean,
      "cycle",
      {.num = &config.cycle},

test/helper-test.c

@@ -142,25 +142,25 @@ int main(int argc, char **argv) {
    */
 
   TASSERT(levenshtein("aap", g_utf8_strlen("aap", -1), "aap",
-                      g_utf8_strlen("aap", -1)) == 0);
+                      g_utf8_strlen("aap", -1), 0) == 0);
   TASSERT(levenshtein("aap", g_utf8_strlen("aap", -1), "aap ",
-                      g_utf8_strlen("aap ", -1)) == 1);
+                      g_utf8_strlen("aap ", -1), 0) == 1);
   TASSERT(levenshtein("aap ", g_utf8_strlen("aap ", -1), "aap",
-                      g_utf8_strlen("aap", -1)) == 1);
+                      g_utf8_strlen("aap", -1), 0) == 1);
   TASSERTE(levenshtein("aap", g_utf8_strlen("aap", -1), "aap noot",
-                       g_utf8_strlen("aap noot", -1)),
+                       g_utf8_strlen("aap noot", -1), 0),
            5u);
   TASSERTE(levenshtein("aap", g_utf8_strlen("aap", -1), "noot aap",
-                       g_utf8_strlen("noot aap", -1)),
+                       g_utf8_strlen("noot aap", -1), 0),
            5u);
   TASSERTE(levenshtein("aap", g_utf8_strlen("aap", -1), "noot aap mies",
-                       g_utf8_strlen("noot aap mies", -1)),
+                       g_utf8_strlen("noot aap mies", -1), 0),
            10u);
   TASSERTE(levenshtein("noot aap mies", g_utf8_strlen("noot aap mies", -1),
-                       "aap", g_utf8_strlen("aap", -1)),
+                       "aap", g_utf8_strlen("aap", -1), 0),
            10u);
   TASSERTE(levenshtein("otp", g_utf8_strlen("otp", -1), "noot aap",
-                       g_utf8_strlen("noot aap", -1)),
+                       g_utf8_strlen("noot aap", -1), 0),
            5u);
   /**
    * Quick converision check.
@@ -192,20 +192,48 @@ int main(int argc, char **argv) {
   }
   {
     TASSERTL(
-        rofi_scorer_fuzzy_evaluate("aap noot mies", 12, "aap noot mies", 12),
+        rofi_scorer_fuzzy_evaluate("aap noot mies", 12, "aap noot mies", 12, 0),
         -605);
-    TASSERTL(rofi_scorer_fuzzy_evaluate("anm", 3, "aap noot mies", 12), -155);
-    TASSERTL(rofi_scorer_fuzzy_evaluate("blu", 3, "aap noot mies", 12),
+    TASSERTL(rofi_scorer_fuzzy_evaluate("anm", 3, "aap noot mies", 12, 0),
+             -155);
+    TASSERTL(rofi_scorer_fuzzy_evaluate("blu", 3, "aap noot mies", 12, 0),
              1073741824);
-    config.case_sensitive = TRUE;
-    TASSERTL(rofi_scorer_fuzzy_evaluate("Anm", 3, "aap noot mies", 12),
+    TASSERTL(rofi_scorer_fuzzy_evaluate("Anm", 3, "aap noot mies", 12, 1),
              1073741754);
-    config.case_sensitive = FALSE;
-    TASSERTL(rofi_scorer_fuzzy_evaluate("Anm", 3, "aap noot mies", 12), -155);
-    TASSERTL(rofi_scorer_fuzzy_evaluate("aap noot mies", 12, "Anm", 3),
+    TASSERTL(rofi_scorer_fuzzy_evaluate("Anm", 3, "aap noot mies", 12, 0),
+             -155);
+    TASSERTL(rofi_scorer_fuzzy_evaluate("aap noot mies", 12, "Anm", 3, 0),
              1073741824);
   }
 
+  /**
+   * Case sensitivity
+   */
+  {
+    int case_smart = config.case_smart;
+    int case_sensitive = config.case_sensitive;
+    {
+      config.case_smart = FALSE;
+      config.case_sensitive = FALSE;
+      TASSERT(parse_case_sensitivity("all lower case 你好") == 0);
+      TASSERT(parse_case_sensitivity("not All lowEr Case 你好") == 0);
+      config.case_sensitive = TRUE;
+      TASSERT(parse_case_sensitivity("all lower case 你好") == 1);
+      TASSERT(parse_case_sensitivity("not All lowEr Case 你好") == 1);
+    }
+    {
+      config.case_smart = TRUE;
+      config.case_sensitive = TRUE;
+      TASSERT(parse_case_sensitivity("all lower case") == 0);
+      TASSERT(parse_case_sensitivity("AAAAAAAAAAAA") == 1);
+      config.case_sensitive = FALSE;
+      TASSERT(parse_case_sensitivity("all lower case 你好") == 0);
+      TASSERT(parse_case_sensitivity("not All lowEr Case 你好") == 1);
+    }
+    config.case_smart = case_smart;
+    config.case_sensitive = case_sensitive;
+  }
+
   char *a;
   a = helper_string_replace_if_exists(
       "{terminal} [-t {title} blub ]-e {cmd}", "{cmd}", "aap", "{title}",