[GR-54974] Escape possible future wildcards in globs

PullRequest: graal/18164

Author: dnestoro

substratevm/src/com.oracle.svm.configure/src/com/oracle/svm/configure/config/ResourceConfiguration.java

@@ -48,6 +48,7 @@
 import com.oracle.svm.core.configure.ConfigurationParser;
 import com.oracle.svm.core.configure.ResourceConfigurationParser;
 import com.oracle.svm.core.configure.ResourcesRegistry;
+import com.oracle.svm.core.jdk.resources.CompressedGlobTrie.GlobUtils;
 import com.oracle.svm.core.util.VMError;
 
 import jdk.graal.compiler.util.json.JsonPrinter;
@@ -209,12 +210,21 @@ public void addResourcePattern(UnresolvedConfigurationCondition condition, Strin
     }
 
     public void addGlobPattern(UnresolvedConfigurationCondition condition, String pattern, String module) {
-        ResourceEntry element = new ResourceEntry(pattern, module);
+        ResourceEntry element = new ResourceEntry(escapePossibleGlobWildcards(pattern), module);
         addedGlobs.add(new ConditionalElement<>(condition, element));
     }
 
-    private static String unquotePattern(String pattern) {
-        return pattern.replace("\\Q", "").replace("\\E", "");
+    private static String escapePossibleGlobWildcards(String pattern) {
+        String escapedPattern = pattern;
+        for (char wildcard : GlobUtils.ALWAYS_ESCAPED_GLOB_WILDCARDS) {
+            escapedPattern = escapedPattern.replace(String.valueOf(wildcard), "\\" + wildcard);
+        }
+
+        return escapedPattern;
+    }
+
+    private static String convertSimpleRegexToGlob(String pattern) {
+        return escapePossibleGlobWildcards(pattern.replace("\\Q", "").replace("\\E", ""));
     }
 
     private static boolean isModuleIdentifierChar(int c) {
@@ -255,7 +265,7 @@ private static boolean isSimpleQuotedPattern(String pattern) {
     private void classifyAndAddPattern(ConditionalElement<String> entry) {
         String pattern = entry.element();
         if (isSimpleQuotedPattern(pattern)) {
-            String unquotedPattern = unquotePattern(pattern);
+            String unquotedPattern = convertSimpleRegexToGlob(pattern);
             String module = null;
             int moduleSplitter = pattern.indexOf(':');
             if (moduleSplitter != -1) {

substratevm/src/com.oracle.svm.core/src/com/oracle/svm/core/jdk/resources/CompressedGlobTrie/CompressedGlobTrie.java

@@ -127,8 +127,22 @@ public static GlobTrieNode build(List<GlobWithInfo> patterns) {
             return root;
         }
 
+        /*
+         * this function should be called only when we are certain that all possible wildcards are
+         * actually escaped (i.e. only after validatePattern function)
+         */
+        private static String unescapePossibleWildcards(String pattern) {
+            String unescapedPattern = pattern;
+            for (char esc : GlobUtils.ALWAYS_ESCAPED_GLOB_WILDCARDS) {
+                unescapedPattern = unescapedPattern.replace("\\" + esc, String.valueOf(esc));
+            }
+
+            return unescapedPattern;
+        }
+
         private static void addPattern(GlobTrieNode root, GlobWithInfo pattern) {
-            List<GlobTrieNode> parts = getPatternParts(pattern.pattern());
+            String unescapedPattern = unescapePossibleWildcards(pattern.pattern());
+            List<GlobTrieNode> parts = getPatternParts(unescapedPattern);
 
             /*
              * if this pattern can be matched with some other existing pattern, we should just
@@ -384,6 +398,17 @@ public static String validatePattern(String pattern) {
             sb.append("Pattern contains invalid sequence **/**. Valid pattern should have ** followed by something other than **. ");
         }
 
+        /* check if there are unescaped wildcards */
+        boolean escapeMode = false;
+        for (int i = 0; i < pattern.length(); i++) {
+            char current = pattern.charAt(i);
+            if (GlobUtils.ALWAYS_ESCAPED_GLOB_WILDCARDS.contains(current) && !escapeMode) {
+                sb.append("Pattern contains unescaped character ").append(current).append(". ");
+            }
+
+            escapeMode = current == '\\';
+        }
+
         // check if pattern contains ** without previous Literal parent. Example: */**/... or **/...
         List<GlobTrieNode> patternParts = getPatternParts(pattern);
         for (GlobTrieNode part : patternParts) {

substratevm/src/com.oracle.svm.core/src/com/oracle/svm/core/jdk/resources/CompressedGlobTrie/GlobUtils.java

@@ -25,10 +25,15 @@
 
 package com.oracle.svm.core.jdk.resources.CompressedGlobTrie;
 
+import java.util.List;
+
 import com.oracle.svm.util.StringUtil;
 
 public class GlobUtils {
 
+    /* list of glob wildcards we are always escaping because they are not supported yet */
+    public static final List<Character> ALWAYS_ESCAPED_GLOB_WILDCARDS = List.of('?', '[', ']', '{', '}');
+
     public static String transformToTriePath(String resource, String module) {
         String resolvedModuleName;
         if (module == null || module.isEmpty()) {