filterx: Redesigned regexp_search functionality

Introduced two new optional flags: keep_zero and list_mode.
The result type no longer switches between dict and list based on the presence of named groups.
The default result type is now dict.
list_mode can force a list result, either via the flag or the type of the fillable (the fillable's type takes precedence).
Match group zero is now excluded by default unless no other groups are present. This behavior can be overridden using the keep_zero flag.

Additionally, the FilterXRematchState structure, shared across other regexp functions, has been updated with a generic flags field. This allows functions to pass custom options to the state without incurring additional overhead.

Signed-off-by: shifter <shifter@axoflow.com>

lib/filterx/expr-regexp.c

@@ -41,7 +41,14 @@
   FILTERX_FUNC_REGEXP_SUBST_FLAG_NEWLINE_NAME"=(boolean)" \
   FILTERX_FUNC_REGEXP_SUBST_FLAG_GROUPS_NAME"=(boolean))" \
 
-#define FILTERX_FUNC_REGEXP_SEARCH_USAGE "Usage: regexp_search(string, pattern)"
+DEFINE_FUNC_FLAG_NAMES(FilterXRegexpSearchFlags,
+                       FILTERX_REGEXP_SEARCH_KEEP_GRP_ZERO_NAME,
+                       FILTERX_REGEXP_SEARCH_LIST_MODE_NAME
+                      );
+
+#define FILTERX_FUNC_REGEXP_SEARCH_USAGE "Usage: regexp_search(string, pattern, " \
+FILTERX_REGEXP_SEARCH_KEEP_GRP_ZERO_NAME"=(boolean), "\
+FILTERX_REGEXP_SEARCH_LIST_MODE_NAME"=(boolean))"
 
 typedef struct FilterXReMatchState_
 {
@@ -50,6 +57,7 @@ typedef struct FilterXReMatchState_
   const gchar *lhs_str;
   gsize lhs_str_len;
   gint rc;
+  FLAGSET flags;
 } FilterXReMatchState;
 
 static void
@@ -164,14 +172,6 @@ _match(FilterXExpr *lhs_expr, pcre2_code_8 *pattern, FilterXReMatchState *state)
   return FALSE;
 }
 
-static gboolean
-_has_named_capture_groups(pcre2_code_8 *pattern)
-{
-  guint32 namecount = 0;
-  pcre2_pattern_info(pattern, PCRE2_INFO_NAMECOUNT, &namecount);
-  return namecount > 0;
-}
-
 static gboolean
 _store_matches_to_list(pcre2_code_8 *pattern, const FilterXReMatchState *state, FilterXObject *fillable)
 {
@@ -180,6 +180,8 @@ _store_matches_to_list(pcre2_code_8 *pattern, const FilterXReMatchState *state,
 
   for (gint i = 0; i < num_matches; i++)
     {
+      if (num_matches > 1 && i==0 && !check_flag(state->flags, FILTERX_REGEXP_SEARCH_KEEP_GRP_ZERO))
+        continue;
       gint begin_index = matches[2 * i];
       gint end_index = matches[2 * i + 1];
       if (begin_index < 0 || end_index < 0)
@@ -209,6 +211,9 @@ _store_matches_to_dict(pcre2_code_8 *pattern, const FilterXReMatchState *state,
   /* First store all matches with string formatted indexes as keys. */
   for (guint32 i = 0; i < num_matches; i++)
     {
+      if (num_matches > 1 && i==0 && !check_flag(state->flags, FILTERX_REGEXP_SEARCH_KEEP_GRP_ZERO))
+        continue;
+
       PCRE2_SIZE begin_index = matches[2 * i];
       PCRE2_SIZE end_index = matches[2 * i + 1];
       if (begin_index < 0 || end_index < 0)
@@ -385,6 +390,7 @@ typedef struct FilterXExprRegexpSearchGenerator_
   FilterXGeneratorFunction super;
   FilterXExpr *lhs;
   pcre2_code_8 *pattern;
+  FLAGSET flags;
 } FilterXExprRegexpSearchGenerator;
 
 static gboolean
@@ -395,6 +401,7 @@ _regexp_search_generator_generate(FilterXExprGenerator *s, FilterXObject *fillab
   gboolean result;
   FilterXReMatchState state;
   _state_init(&state);
+  state.flags = self->flags;
 
   gboolean matched = _match(self->lhs, self->pattern, &state);
   if (!matched)
@@ -422,10 +429,10 @@ _regexp_search_generator_create_container(FilterXExprGenerator *s, FilterXExpr *
 {
   FilterXExprRegexpSearchGenerator *self = (FilterXExprRegexpSearchGenerator *) s;
 
-  if (_has_named_capture_groups(self->pattern))
-    return filterx_generator_create_dict_container(s, fillable_parent);
+  if (check_flag(self->flags, FILTERX_REGEXP_SEARCH_LIST_MODE))
+    return filterx_generator_create_list_container(s, fillable_parent);
 
-  return filterx_generator_create_list_container(s, fillable_parent);
+  return filterx_generator_create_dict_container(s, fillable_parent);
 }
 
 static gboolean
@@ -459,6 +466,29 @@ _regexp_search_generator_free(FilterXExpr *s)
   filterx_generator_function_free_method(&self->super);
 }
 
+static gboolean
+_extract_optional_arg_flag(FilterXExprRegexpSearchGenerator *self, FilterXRegexpSearchFlags flag,
+                           FilterXFunctionArgs *args, GError **error)
+{
+  gboolean exists, eval_error;
+  g_assert(flag < FilterXRegexpSearchFlags_MAX);
+  const gchar *arg_name = FilterXRegexpSearchFlags_NAMES[flag];
+  gboolean value = filterx_function_args_get_named_literal_boolean(args, arg_name, &exists, &eval_error);
+  if (!exists)
+    return TRUE;
+
+  if (eval_error)
+    {
+      g_set_error(error, FILTERX_FUNCTION_ERROR, FILTERX_FUNCTION_ERROR_CTOR_FAIL,
+                  "%s argument must be boolean literal. " FILTERX_FUNC_REGEXP_SEARCH_USAGE, arg_name);
+      return FALSE;
+    }
+
+  set_flag(&self->flags, flag, value);
+
+  return TRUE;
+}
+
 static gboolean
 _extract_search_args(FilterXExprRegexpSearchGenerator *self, FilterXFunctionArgs *args, GError **error)
 {
@@ -504,6 +534,12 @@ filterx_generator_function_regexp_search_new(FilterXFunctionArgs *args, GError *
   self->super.super.super.free_fn = _regexp_search_generator_free;
   self->super.super.create_container = _regexp_search_generator_create_container;
 
+  if (!_extract_optional_arg_flag(self, FILTERX_REGEXP_SEARCH_KEEP_GRP_ZERO, args, error))
+    goto error;
+
+  if (!_extract_optional_arg_flag(self, FILTERX_REGEXP_SEARCH_LIST_MODE, args, error))
+    goto error;
+
   if (!_extract_search_args(self, args, error) ||
       !filterx_function_args_check(args, error))
     goto error;

lib/filterx/expr-regexp.h

@@ -27,6 +27,7 @@
 #include "filterx/filterx-expr.h"
 #include "filterx/expr-generator.h"
 #include "filterx/expr-function.h"
+#include "filterx/func-flags.h"
 
 #define FILTERX_FUNC_REGEXP_SUBST_FLAG_JIT_NAME "jit"
 #define FILTERX_FUNC_REGEXP_SUBST_FLAG_GLOBAL_NAME "global"
@@ -35,6 +36,16 @@
 #define FILTERX_FUNC_REGEXP_SUBST_FLAG_NEWLINE_NAME "newline"
 #define FILTERX_FUNC_REGEXP_SUBST_FLAG_GROUPS_NAME "groups"
 
+DEFINE_FUNC_FLAGS(FilterXRegexpSearchFlags,
+                  FILTERX_REGEXP_SEARCH_KEEP_GRP_ZERO,
+                  FILTERX_REGEXP_SEARCH_LIST_MODE
+                 );
+
+#define FILTERX_REGEXP_SEARCH_KEEP_GRP_ZERO_NAME "keep_zero"
+#define FILTERX_REGEXP_SEARCH_LIST_MODE_NAME "list_mode"
+
+extern const char *FilterXRegexpSearchFlags_NAMES[];
+
 typedef struct FilterXFuncRegexpSubstOpts_
 {
   gboolean global;