fix: expand_absolute_path() function and utils::patternMatch()

Author: aritosteles

src/common/file_helper/filesystem/src/filesystem_wrapper.cpp

@@ -121,21 +121,28 @@ namespace filesystem_wrapper
 
             if (exists(baseDir))
             {
-                for (const auto& entry : list_directory(baseDir))
+                if(is_directory(baseDir))
                 {
-                    const auto entryName {entry.filename().string()};
-
-                    if (Utils::patternMatch(entryName, pattern))
+                    for (const auto& entry : list_directory(baseDir))
                     {
-                        std::string nextPath;
-                        nextPath += baseDir;
-                        nextPath += std::filesystem::path::preferred_separator;
-                        nextPath += entryName;
-                        nextPath += nextDirectoryPos == std::string::npos ? "" : path.substr(nextDirectoryPos);
-
-                        expand_absolute_path(nextPath, output);
+                        const auto entryName {entry.filename().string()};
+
+                        if (Utils::patternMatch(entryName, pattern))
+                        {
+                            std::string nextPath;
+                            nextPath += baseDir;
+                            nextPath += std::filesystem::path::preferred_separator;
+                            nextPath += entryName;
+                            nextPath += nextDirectoryPos == std::string::npos ? "" : path.substr(nextDirectoryPos);
+
+                            expand_absolute_path(nextPath, output);
+                        }
                     }
                 }
+                else if (Utils::patternMatch(baseDir, pattern))
+                {
+                    output.push_back(baseDir);
+                }
             }
         }
         else

src/common/file_helper/filesystem/tests/filesystem_test.cpp

@@ -22,6 +22,7 @@ constexpr auto PATH_TO_EXPAND_3 {"/tmp/wazuh_test/prefix_*_data/prefix*"};
 constexpr auto PATH_TO_EXPAND_4 {"/tmp/wazuh_test/prefix_*_data/*_1"};
 constexpr auto PATH_TO_EXPAND_5 {"/tmp/wazuh_test/prefix_?_data/*_1"};
 constexpr auto PATH_TO_EXPAND_6 {"/tmp/wazuh_test/prefix_*_data/prefix?*1"};
+constexpr auto PATH_TO_EXPAND_7 {"/tmp/wazuh_test/*/*1"};
 constexpr auto TMP_PATH {"/tmp"};
 constexpr auto ROOT_PATH {"/tmp/wazuh_test"};
 constexpr auto ROOT_PATH_1 {"/tmp/wazuh_test/prefix_1_data"};
@@ -41,6 +42,7 @@ constexpr auto PATH_TO_EXPAND_3 {"C:\\tmp\\wazuh_test\\prefix_*_data\\prefix*"};
 constexpr auto PATH_TO_EXPAND_4 {"C:\\tmp\\wazuh_test\\prefix_*_data\\*_1"};
 constexpr auto PATH_TO_EXPAND_5 {"C:\\tmp\\wazuh_test\\prefix_?_data\\*_1"};
 constexpr auto PATH_TO_EXPAND_6 {"C:\\tmp\\wazuh_test\\prefix_*_data\\prefix?*1"};
+constexpr auto PATH_TO_EXPAND_7 {"C:\\tmp\\wazuh_test\\*\\*1"};
 constexpr auto TMP_PATH {"C:\\tmp"};
 constexpr auto ROOT_PATH {"C:\\tmp\\wazuh_test"};
 constexpr auto ROOT_PATH_1 {"C:\\tmp\\wazuh_test\\prefix_1_data"};
@@ -218,6 +220,25 @@ TEST_F(FileSystemTest, FilesystemExpandWildcardWithPrefix)
     EXPECT_EQ(output.size(), PATH_MATCH_SIZE);
 }
 
+TEST_F(FileSystemTest, FilesystemExpandWildcardWithPrefixFull)
+{
+    constexpr auto PATH_MATCH_SIZE {2u};
+    std::deque<std::string> output;
+    std::unordered_map<std::string, uint32_t> outputMap;
+
+    const auto fsWrapper = std::make_unique<filesystem_wrapper::FileSystemWrapper>();
+    fsWrapper->expand_absolute_path(PATH_TO_EXPAND_7, output);
+
+    for (const auto& item : output)
+    {
+        outputMap[item]++;
+    }
+
+    EXPECT_TRUE(outputMap.at(EXPAND_PATH_1) == 1);
+    EXPECT_TRUE(outputMap.at(EXPAND_PATH_3) == 1);
+    EXPECT_EQ(output.size(), PATH_MATCH_SIZE);
+}
+
 TEST_F(FileSystemTest, FilesystemExpandWildcardWithSuffix)
 {
     constexpr auto PATH_MATCH_SIZE {2u};

src/common/globHelper/include/globHelper.h

@@ -26,74 +26,49 @@ namespace Utils
 #pragma warning(disable: 4505)
 #endif
 
-    static bool patternMatch(const std::string& entryName, const std::string& pattern)
-    {
-        auto match { true };
-        // Match the glob pattern without regex
-        auto patternPos { 0u };
+static bool patternMatch(const std::string& entryName, const std::string& pattern)
+{
+    std::string_view entry(entryName);
+    std::string_view pat(pattern);
+
+    // Dynamic programming approach - create a table for memoization
+    // dp[i][j] is true if the first i characters of entry match the first j characters of pattern
+    std::vector<std::vector<bool>> dp(entry.size() + 1, std::vector<bool>(pat.size() + 1, false));
+
+    // Empty pattern matches empty string
+    dp[0][0] = true;
 
-        for (auto i { 0u }; i < entryName.size(); ++i)
+    // Handle patterns like "*", "*a", "a*b*" etc. where '*' can match empty strings
+    for (size_t j = 1; j <= pat.size(); ++j)
+    {
+        if (pat[j - 1] == '*')
         {
-            if (patternPos < pattern.size())
-            {
-                // 'x' matches 'x'
-                if (entryName.at(i) == pattern.at(patternPos))
-                {
-                    ++patternPos;
-                }
-                // '*' matches any number of characters
-                else if (pattern.at(patternPos) == '*')
-                {
-                    // '*' matches zero characters
-                    if (patternPos + 1 < pattern.size() && pattern.at(patternPos + 1) == entryName.at(i))
-                    {
-                        ++patternPos;
-                        --i;
-                    }
-                    // '*' matches one or more characters
-                    else if (patternPos + 1 == pattern.size())
-                    {
-                        break;
-                    }
-                }
-                // '?' matches any single character
-                else if (pattern.at(patternPos) == '?')
-                {
-                    ++patternPos;
-                }
-                // No match
-                else
-                {
-                    match = false;
-                    break;
-                }
-            }
-            else
-            {
-                match = false;
-                break;
-            }
+            dp[0][j] = dp[0][j - 1];
         }
+    }
 
-        // if the pattern is not fully matched, check if the remaining characters are '*'
-        // and if so, the match is successful.
-        while (match && patternPos < pattern.size())
+    // Fill the dp table
+    for (size_t i = 1; i <= entry.size(); ++i)
+    {
+        for (size_t j = 1; j <= pat.size(); ++j)
         {
-            // '*' matches zero characters
-            if (pattern.at(patternPos) == '*')
+            if (pat[j - 1] == '*')
             {
-                ++patternPos;
+                // '*' can match 0 or more characters
+                // Either ignore '*' (dp[i][j-1]) or use '*' to match current character (dp[i-1][j])
+                dp[i][j] = dp[i][j - 1] || dp[i - 1][j];
             }
-            // No match
-            else
+            else if (pat[j - 1] == '?' || entry[i - 1] == pat[j - 1])
             {
-                match = false;
+                // '?' matches any single character, or exact character match
+                dp[i][j] = dp[i - 1][j - 1];
             }
         }
-
-        return match;
     }
 
+    return dp[entry.size()][pat.size()];
+}
+
 #ifdef __GNUC__
 #pragma GCC diagnostic pop
 #endif