Merge pull request #876 from jjnicola/regex-master

[master] Add `find_all` to eregmatch() nasl function. Backport #875

CHANGELOG.md

@@ -18,6 +18,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 - Add nasl function sftp_enabled_check() to check if sftp subsystem is enabled in the target.
   - [#853](https://github.com/greenbone/openvas/pull/853)
   - [#862](https://github.com/greenbone/openvas/pull/862)
+- Add `find_all` to eregmatch() nasl function. Backport PR #875. [#876](https://github.com/greenbone/openvas/pull/876)
 
 ### Changed
 - function script_bugtraq_id getting skipped, linter warns. [#724](https://github.com/greenbone/openvas/pull/724)

nasl/nasl_text_utils.c

@@ -786,16 +786,30 @@ nasl_egrep (lex_ctxt *lexic)
 
 /**
  * @brief Does extended regular expression pattern matching.
+ * @naslfn{eregmatch}
+ *
+ * @nasluparam
+ *
+ * - @a pattern An regex pattern
+ * - @a string A string
+ * - @a icase Boolean, for case sensitve
+ * - @a find_all Boolean, to find all matches
+ *
+ * @naslret An array with the first match (find_all: False)
+ *          or an array with all matches (find_all: TRUE).
+ *          NULL or empty if no match was found.
+ *
+ * @param[in] lexic Lexical context of NASL interpreter.
  *
- * In NASL, this function returns an array.
  */
 tree_cell *
 nasl_eregmatch (lex_ctxt *lexic)
 {
   char *pattern = get_str_var_by_name (lexic, "pattern");
   char *string = get_str_var_by_name (lexic, "string");
   int icase = get_int_var_by_name (lexic, "icase", 0);
-  int copt = 0, i;
+  int find_all = get_int_var_by_name (lexic, "find_all", 0);
+  int copt = 0;
   tree_cell *retc;
   regex_t re;
   regmatch_t subs[NS];
@@ -814,24 +828,63 @@ nasl_eregmatch (lex_ctxt *lexic)
                    pattern);
       return NULL;
     }
+  retc = alloc_typed_cell (DYN_ARRAY);
+  retc->x.ref_val = a = g_malloc0 (sizeof (nasl_array));
 
-  if (regexec (&re, string, (size_t) NS, subs, 0) != 0)
+  if (find_all == 0)
     {
-      regfree (&re);
-      return NULL;
+      if (regexec (&re, string, (size_t) NS, subs, 0) != 0)
+        {
+          regfree (&re);
+          return NULL;
+        }
+
+      int i;
+      for (i = 0; i < NS; i++)
+        if (subs[i].rm_so != -1)
+          {
+            v.var_type = VAR2_DATA;
+            v.v.v_str.s_siz = subs[i].rm_eo - subs[i].rm_so;
+            v.v.v_str.s_val = (unsigned char *) string + subs[i].rm_so;
+            (void) add_var_to_list (a, i, &v);
+          }
     }
+  else
+    {
+      int index = 0;
+      char *current_pos;
+      current_pos = string;
+      while (1)
+        {
+          if (regexec (&re, current_pos, (size_t) NS, subs, 0) != 0)
+            {
+              regfree (&re);
+              break;
+            }
 
-  retc = alloc_typed_cell (DYN_ARRAY);
-  retc->x.ref_val = a = g_malloc0 (sizeof (nasl_array));
+          unsigned int offset = 0, i = 0;
+          for (i = 0; i < NS; i++)
+            {
+              char current_pos_cp[strlen (current_pos) + 1];
 
-  for (i = 0; i < NS; i++)
-    if (subs[i].rm_so != -1)
-      {
-        v.var_type = VAR2_DATA;
-        v.v.v_str.s_siz = subs[i].rm_eo - subs[i].rm_so;
-        v.v.v_str.s_val = (unsigned char *) string + subs[i].rm_so;
-        (void) add_var_to_list (a, i, &v);
-      }
+              if (subs[i].rm_so == -1)
+                break;
+
+              if (i == 0)
+                offset = subs[i].rm_eo;
+
+              strcpy (current_pos_cp, current_pos);
+              current_pos_cp[subs[i].rm_eo] = 0;
+              v.var_type = VAR2_DATA;
+              v.v.v_str.s_siz = subs[i].rm_eo - subs[i].rm_so;
+              v.v.v_str.s_val =
+                (unsigned char *) current_pos_cp + subs[i].rm_so;
+              (void) add_var_to_list (a, index, &v);
+              index++;
+            }
+          current_pos += offset;
+        }
+    }
 
   regfree (&re);
   return retc;