Modification to fix failing test

jonclayden · Nov 10, 2023 · c856df1 · c856df1
1 parent 1228b4d
commit c856df1
Show file tree

Hide file tree

Showing 3 changed files with 25 additions and 3 deletions.
diff --git a/src/compile.c b/src/compile.c
@@ -150,7 +150,8 @@ static char * ore_build_pattern (SEXP pattern_)
     return pattern;
 }
 
-static Rboolean ore_group_name_vector (SEXP vec, regex_t *regex)
+// Insert group names into an R character vector of appropriate size
+Rboolean ore_group_name_vector (SEXP vec, regex_t *regex)
 {
     const int n_groups = onig_number_of_captures(regex);
 

diff --git a/src/compile.h b/src/compile.h
@@ -10,6 +10,8 @@ regex_t * ore_retrieve (SEXP regex_, encoding_t *encoding);
 
 void ore_free (regex_t *regex, SEXP source);
 
+Rboolean ore_group_name_vector (SEXP vec, regex_t *regex);
+
 SEXP ore_build (SEXP pattern_, SEXP options_, SEXP encoding_name_, SEXP syntax_name_);
 
 #endif
diff --git a/src/match.c b/src/match.c
@@ -252,7 +252,6 @@ void ore_char_matrix (SEXP mat, const char **data, const int n_regions, const in
 SEXP ore_search_all (SEXP regex_, SEXP text_, SEXP all_, SEXP start_, SEXP simplify_, SEXP incremental_)
 {
     // Convert R objects to C types
-    SEXP group_names = getAttrib(regex_, install("groupNames"));
     const Rboolean all = asLogical(all_) == TRUE;
     const Rboolean simplify = asLogical(simplify_) == TRUE;
     const Rboolean incremental = (asLogical(incremental_) == TRUE) && !all;
@@ -271,6 +270,26 @@ SEXP ore_search_all (SEXP regex_, SEXP text_, SEXP all_, SEXP start_, SEXP simpl
     text_t *text = ore_text(text_);
     regex_t *regex = ore_retrieve(regex_, text->encoding);
 
+    SEXP group_names = R_NilValue;
+    Rboolean group_names_protected = FALSE;
+    if (inherits(regex_, "ore"))
+        group_names = getAttrib(regex_, install("groupNames"));
+    else
+    {
+        const int n_groups = onig_number_of_captures(regex);
+        if (n_groups > 0)
+        {
+            PROTECT(group_names = NEW_CHARACTER(n_groups));
+            if (ore_group_name_vector(group_names, regex))
+                group_names_protected = TRUE;
+            else
+            {
+                UNPROTECT(1);
+                group_names = R_NilValue;
+            }
+        }
+    }
+
     // Obtain the length of the start vector (which will be recycled if necessary)
     const int start_len = length(start_);
 
@@ -440,7 +459,7 @@ SEXP ore_search_all (SEXP regex_, SEXP text_, SEXP all_, SEXP start_, SEXP simpl
     ore_free(regex, regex_);
     ore_text_done(text);
 
-    UNPROTECT(using_file ? 1 : 2);
+    UNPROTECT(2 + group_names_protected - using_file);
 
     // Return just the first (and only) element of the full list, if requested
     if (simplify && text->length == 1)