In JSONParser::getToken reject illegal control characters

m-holger
1 parent 1b89e768
Showing 11 changed files with 30 additions and 5 deletions
libqpdf/JSON.cc
libtests/qtest/json_parse.test
libtests/qtest/json_parse/bad-18.out
libtests/qtest/json_parse/bad-45.json
libtests/qtest/json_parse/bad-45.out
libtests/qtest/json_parse/bad-46.json
libtests/qtest/json_parse/bad-46.out
libtests/qtest/json_parse/bad-47.json
libtests/qtest/json_parse/bad-47.out
libtests/qtest/json_parse/bad-48.json
libtests/qtest/json_parse/bad-48.out
@@ -780,10 +780,22 @@ JSONParser::getToken()
             }
         }
  
-        if (*p == 0) {
-            QTC::TC("libtests", "JSON parse null character");
-            throw std::runtime_error(
-                "JSON: null character at offset " + std::to_string(offset));
+        if ((*p < 32 && *p >= 0)) {
+            if (*p == '\t' || *p == '\n' || *p == '\r') {
+                // Legal white space not permitted in strings. This will always
+                // end the current token (unless we are still before the start
+                // of the token).
+                if (lex_state == ls_top) {
+                    // Continue with token
+                } else {
+                    // done
+                }
+            } else {
+                QTC::TC("libtests", "JSON parse null character");
+                throw std::runtime_error(
+                    "JSON: control or null character at offset " +
+                    std::to_string(offset));
+            }
         }
         action = append;
         switch (lex_state) {
@@ -125,6 +125,10 @@ my @bad = (
     "e after minus",            # 42
     "missing digit after e",    # 43
     "missing digit after e+/-", # 44
+    # "tab char in string",       # 45
+    # "cr char in string",        # 46
+    # "lf char in string",        # 47
+    # "bs char in string",        # 48
     );
  
 my $i = 0;
-exception: bad-18.json: JSON: null character at offset 5
+exception: bad-18.json: JSON: control or null character at offset 5
+"Tab in str	ing"
+"Tab in str\ting"
+"cr in str
ing"
+"cr in str\ring"
+"lf in str
+ing"
+"lf in str\ning"
+"bs in string"
 \ No newline at end of file
+exception: bad-48.json: JSON: control or null character at offset 10