MGM: Added a new Tokenize function for quoted values. Fixes EOS-5783

Author: tavisit

common/StringConversion.cc

@@ -78,6 +78,53 @@ StringConversion::Tokenize(const std::string& str,
   }
 }
 
+//------------------------------------------------------------------------------
+// Tokenize a string respecting quoted strings and escapes
+//------------------------------------------------------------------------------
+void
+StringConversion::TokenizeQuoted(const std::string& str, std::vector<std::string>& tokens,
+                                 const std::string& delimiters)
+{
+  if (str.empty()) {
+    return;
+  }
+
+  size_t pos = 0;
+  const size_t len = str.length();
+  bool in_quotes = false;
+  bool escaped = false;
+  std::string current_token;
+
+  while (pos < len) {
+    char ch = str[pos++];
+
+    if (escaped) {
+      // Previous char was '\', add this char literally (e.g., \" becomes ")
+      current_token += ch;
+      escaped = false;
+    } else if (ch == '\\') {
+      // escape next character
+      escaped = true;
+    } else if (ch == '"') {
+      // Toggle quote state
+      in_quotes = !in_quotes;
+    } else if (!in_quotes && delimiters.find(ch) != std::string::npos) {
+      // Found delimiter outside quotes - end current token
+      if (!current_token.empty()) {
+        tokens.push_back(std::move(current_token));
+        current_token.clear();
+      }
+    } else {
+      // Regular character or delimiter inside quotes
+      current_token += ch;
+    }
+  }
+
+  // Add final token if any
+  if (!current_token.empty()) {
+    tokens.push_back(std::move(current_token));
+  }
+}
 
 //------------------------------------------------------------------------------
 // Tokenize a string seperated by one single charactor or multichar string
@@ -614,7 +661,7 @@ StringConversion::GetKeyValueMap(const char* mapstring,
   std::string is = mapstring;
   std::string delimiter = sdelimiter;
   std::vector<std::string> slist;
-  Tokenize(is, slist, delimiter);
+  TokenizeQuoted(is, slist, delimiter);
 
   if (!slist.size()) {
     return false;

common/StringConversion.hh

@@ -82,6 +82,13 @@ public:
                        std::vector<std::string>& tokens,
                        const std::string& delimiters = " ");
 
+  /**
+   * Tokenize a string respecting quoted strings
+   * Quotes and escape sequences are handled properly
+   */
+  static void TokenizeQuoted(const std::string& str, std::vector<std::string>& tokens,
+                             const std::string& delimiters);
+
   // ---------------------------------------------------------------------------
   /**
    * Tokenize a string

mgm/proc/IProcCommand.cc

@@ -357,10 +357,6 @@ IProcCommand::ConvertOutputToJsonFormat(const std::string& stdOut)
       for (int i = 1; i < (int)token.size(); i++) {
         jep = &((*jep)[token[i]]);
       }
-
-      // Unquote value
-      std::stringstream quoted_ss(value);
-      quoted_ss >> std::quoted(value);
       // Seal value
       XrdOucString svalue = value.c_str();
       eos::common::StringConversion::Seal(svalue);

unit_tests/common/StringConversionTests.cc

@@ -196,4 +196,55 @@ TEST(StringConversion, ReplaceStringInPlace)
   ASSERT_STREQ("aabbccddxyeeffggxyhhiijjxy", input.c_str());
 }
 
+TEST(StringConversion, TokenizeQuoted)
+{
+  std::vector<std::string> tokens;
+
+  StringConversion::TokenizeQuoted("", tokens, " ");
+  ASSERT_EQ(tokens.size(), 0u);
+  tokens.clear();
+
+  StringConversion::TokenizeQuoted("hello world test", tokens, " ");
+  ASSERT_EQ(tokens.size(), 3u);
+  ASSERT_STREQ(tokens[0].c_str(), "hello");
+  ASSERT_STREQ(tokens[1].c_str(), "world");
+  ASSERT_STREQ(tokens[2].c_str(), "test");
+  tokens.clear();
+
+  StringConversion::TokenizeQuoted("user.reva.overleaf.name=\"my first project\"", tokens,
+                                   " ");
+  ASSERT_EQ(tokens.size(), 1u);
+  ASSERT_STREQ(tokens[0].c_str(), "user.reva.overleaf.name=my first project");
+
+  std::vector<std::string> kv;
+  StringConversion::TokenizeQuoted(tokens[0], kv, "=");
+  ASSERT_EQ(kv.size(), 2u);
+  ASSERT_STREQ(kv[0].c_str(), "user.reva.overleaf.name");
+  ASSERT_STREQ(kv[1].c_str(), "my first project");
+  tokens.clear();
+  kv.clear();
+
+  std::string attr_line = "sys.acl=\"u:12345:rwxm+dq\" sys.versioning=\"10\" "
+                          "user.reva.overleaf.name=\"my first project\"";
+  StringConversion::TokenizeQuoted(attr_line, tokens, " ");
+  ASSERT_EQ(tokens.size(), 3u);
+  ASSERT_STREQ(tokens[0].c_str(), "sys.acl=u:12345:rwxm+dq");
+  ASSERT_STREQ(tokens[1].c_str(), "sys.versioning=10");
+  ASSERT_STREQ(tokens[2].c_str(), "user.reva.overleaf.name=my first project");
+  tokens.clear();
+
+  StringConversion::TokenizeQuoted("key=\"value with \\\"quoted\\\" text\"", tokens,
+                                   " =");
+  ASSERT_EQ(tokens.size(), 2u);
+  ASSERT_STREQ(tokens[0].c_str(), "key");
+  ASSERT_STREQ(tokens[1].c_str(), "value with \"quoted\" text");
+  tokens.clear();
+
+  StringConversion::TokenizeQuoted("hello   world", tokens, " ");
+  ASSERT_EQ(tokens.size(), 2u);
+  ASSERT_STREQ(tokens[0].c_str(), "hello");
+  ASSERT_STREQ(tokens[1].c_str(), "world");
+  tokens.clear();
+}
+
 EOSCOMMONTESTING_END