Merge pull request github#15706 from erik-krogh/pol-reg

ReDoS: Restrict some edges related to upper/lower-case when constructing possible attack strings for polynomial-redos.

Author: erik-krogh

javascript/ql/test/query-tests/Security/CWE-400/ReDoS/PolynomialBackTracking.expected

@@ -10,7 +10,7 @@
 | highlight.js:19:56:19:61 | [^\\]]+ | Strings starting with '[' and with many repetitions of '.[' can start matching anywhere after the start of the preceeding (\\.\|\\.\\/\|\\/)?(""\|"[^"]+"\|''\|'[^']+'\|\\[\\]\|\\[[^\\]]+\\]\|[^\\s!"#%&'()*+,.\\/;<=>@\\[\\\\\\]^`{\|}~]+)((\\.\|\\/)(""\|"[^"]+"\|''\|'[^']+'\|\\[\\]\|\\[[^\\]]+\\]\|[^\\s!"#%&'()*+,.\\/;<=>@\\[\\\\\\]^`{\|}~]+))* |
 | highlight.js:22:12:22:82 | ((decltype\\(auto\\)\|(?:[a-zA-Z_]\\w*::)?[a-zA-Z_]\\w*(?:<.*?>)?)[\\*&\\s]+)+ | Strings with many repetitions of 'A\\t' can start matching anywhere after the start of the preceeding .*? |
 | highlight.js:22:43:22:45 | \\w* | Strings starting with 'A' and with many repetitions of 'A' can start matching anywhere after the start of the preceeding .*? |
-| highlight.js:22:66:22:68 | .*? | Strings starting with 'A<' and with many repetitions of 'A<' can start matching anywhere after the start of the preceeding \\w* |
+| highlight.js:22:66:22:68 | .*? | Strings starting with 'A<' and with many repetitions of 'a<' can start matching anywhere after the start of the preceeding \\w* |
 | highlight.js:22:73:22:80 | [\\*&\\s]+ | Strings starting with 'A' and with many repetitions of '\\tA\\t' can start matching anywhere after the start of the preceeding .*? |
 | highlight.js:23:13:23:82 | ((decltype\\(auto\\)\|([a-zA-Z_]\\w*::)?[a-zA-Z_]\\w*(<[^<>]+>)?)[\\*&\\s]+)+ | Strings with many repetitions of 'A\\t' can start matching anywhere after the start of the preceeding ((decltype\\(auto\\)\|([a-zA-Z_]\\w*::)?[a-zA-Z_]\\w*(<[^<>]+>)?)[\\*&\\s]+)+([a-zA-Z_]\\w*::)?[a-zA-Z]\\w*\\s*\\( |
 | highlight.js:23:42:23:44 | \\w* | Strings starting with 'A' and with many repetitions of 'A' can start matching anywhere after the start of the preceeding ((decltype\\(auto\\)\|([a-zA-Z_]\\w*::)?[a-zA-Z_]\\w*(<[^<>]+>)?)[\\*&\\s]+)+([a-zA-Z_]\\w*::)?[a-zA-Z]\\w*\\s*\\( |
@@ -279,7 +279,7 @@
 | regexplib/misc.js:117:25:117:26 | .+ | Strings starting with '(a}' and with many repetitions of 'a)' can start matching anywhere after the start of the preceeding .+ |
 | regexplib/misc.js:119:20:119:22 | \\w+ | Strings with many repetitions of '0' can start matching anywhere after the start of the preceeding (NOT)?(\\s*\\(*)\\s*(\\w+)\\s*(=\|<>\|<\|>\|LIKE\|IN)\\s*(\\(([^\\)]*)\\)\|'([^']*)'\|(-?\\d*\\.?\\d+))(\\s*\\)*\\s*)(AND\|OR)? |
 | regexplib/misc.js:119:52:119:57 | [^\\)]* | Strings starting with '0=(' and with many repetitions of '0<((' can start matching anywhere after the start of the preceeding (NOT)?(\\s*\\(*)\\s*(\\w+)\\s*(=\|<>\|<\|>\|LIKE\|IN)\\s*(\\(([^\\)]*)\\)\|'([^']*)'\|(-?\\d*\\.?\\d+))(\\s*\\)*\\s*)(AND\|OR)? |
-| regexplib/misc.js:123:36:123:38 | .*? | Strings starting with '?se[A' and with many repetitions of '?se[Aa' can start matching anywhere after the start of the preceeding (?s)(?:\\e\\[(?:(\\d+);?)*([A-Za-z])(.*?))(?=\\e\\[\|\\z) |
+| regexplib/misc.js:123:36:123:38 | .*? | Strings starting with '?se[A' and with many repetitions of '?se[aa' can start matching anywhere after the start of the preceeding (?s)(?:\\e\\[(?:(\\d+);?)*([A-Za-z])(.*?))(?=\\e\\[\|\\z) |
 | regexplib/misc.js:126:15:126:20 | [a-z]+ | Strings starting with 'a' and with many repetitions of 'aa' can start matching anywhere after the start of the preceeding [a-z]+ |
 | regexplib/misc.js:141:15:141:19 | [^;]+ | Strings starting with '{\\\\f\\\\' and with many repetitions of '{\\\\f\\\\:' can start matching anywhere after the start of the preceeding (\\{\\\\f\\d*)\\\\([^;]+;) |
 | regexplib/misc.js:144:52:144:70 | [a-z0-9\\/\\.\\?\\=\\&]* | Strings starting with '".htm' and with many repetitions of '.asp' can start matching anywhere after the start of the preceeding [a-z0-9\\/\\.\\?\\=\\&]* |
@@ -334,7 +334,7 @@
 | regexplib/strings.js:54:20:54:22 | \\w+ | Strings with many repetitions of '0' can start matching anywhere after the start of the preceeding (NOT)?(\\s*\\(*)\\s*(\\w+)\\s*(=\|<>\|<\|>\|LIKE\|IN)\\s*(\\(([^\\)]*)\\)\|'([^']*)'\|(-?\\d*\\.?\\d+))(\\s*\\)*\\s*)(AND\|OR)? |
 | regexplib/strings.js:54:52:54:57 | [^\\)]* | Strings starting with '0=(' and with many repetitions of '0<((' can start matching anywhere after the start of the preceeding (NOT)?(\\s*\\(*)\\s*(\\w+)\\s*(=\|<>\|<\|>\|LIKE\|IN)\\s*(\\(([^\\)]*)\\)\|'([^']*)'\|(-?\\d*\\.?\\d+))(\\s*\\)*\\s*)(AND\|OR)? |
 | regexplib/strings.js:56:52:56:53 | .+ | Strings starting with 'AUX.' and with many repetitions of '.' can start matching anywhere after the start of the preceeding .* |
-| regexplib/strings.js:57:36:57:38 | .*? | Strings starting with '?se[A' and with many repetitions of '?se[Aa' can start matching anywhere after the start of the preceeding (?s)(?:\\e\\[(?:(\\d+);?)*([A-Za-z])(.*?))(?=\\e\\[\|\\z) |
+| regexplib/strings.js:57:36:57:38 | .*? | Strings starting with '?se[A' and with many repetitions of '?se[aa' can start matching anywhere after the start of the preceeding (?s)(?:\\e\\[(?:(\\d+);?)*([A-Za-z])(.*?))(?=\\e\\[\|\\z) |
 | regexplib/strings.js:64:3:64:5 | \\w+ | Strings with many repetitions of '0' can start matching anywhere after the start of the preceeding (\\w+)\\s+\\1 |
 | regexplib/strings.js:70:6:70:17 | [a-zA-Z,\\s]+ | Strings with many repetitions of '\\t' can start matching anywhere after the start of the preceeding \\s* |
 | regexplib/strings.js:70:18:70:20 | \\s* | Strings starting with '\\t' and with many repetitions of '\\t' can start matching anywhere after the start of the preceeding \\s* |
@@ -345,7 +345,7 @@
 | regexplib/strings.js:74:2:74:3 | .* | Strings with many repetitions of 'a' can start matching anywhere after the start of the preceeding .*[Pp]re[Ss\\$]cr[iI1]pt.* |
 | regexplib/strings.js:75:2:75:3 | .* | Strings with many repetitions of 'a' can start matching anywhere after the start of the preceeding .*[Vv][Ii1]agr.* |
 | regexplib/strings.js:76:2:76:3 | .* | Strings with many repetitions of 'a' can start matching anywhere after the start of the preceeding .*[Oo0][Ee][Mm].* |
-| regexplib/strings.js:81:36:81:38 | .*? | Strings starting with '?se[A' and with many repetitions of '?se[Aa' can start matching anywhere after the start of the preceeding (?s)(?:\\e\\[(?:(\\d+);?)*([A-Za-z])(.*?))(?=\\e\\[\|\\z) |
+| regexplib/strings.js:81:36:81:38 | .*? | Strings starting with '?se[A' and with many repetitions of '?se[aa' can start matching anywhere after the start of the preceeding (?s)(?:\\e\\[(?:(\\d+);?)*([A-Za-z])(.*?))(?=\\e\\[\|\\z) |
 | regexplib/strings.js:82:20:82:22 | \\w+ | Strings with many repetitions of '0' can start matching anywhere after the start of the preceeding (NOT)?(\\s*\\(*)\\s*(\\w+)\\s*(=\|<>\|<\|>\|LIKE\|IN)\\s*(\\(([^\\)]*)\\)\|'([^']*)'\|(-?\\d*\\.?\\d+))(\\s*\\)*\\s*)(AND\|OR)? |
 | regexplib/strings.js:82:52:82:57 | [^\\)]* | Strings starting with '0=(' and with many repetitions of '0<((' can start matching anywhere after the start of the preceeding (NOT)?(\\s*\\(*)\\s*(\\w+)\\s*(=\|<>\|<\|>\|LIKE\|IN)\\s*(\\(([^\\)]*)\\)\|'([^']*)'\|(-?\\d*\\.?\\d+))(\\s*\\)*\\s*)(AND\|OR)? |
 | regexplib/strings.js:88:3:88:12 | [^\\.\\?\\!]* | Strings with many repetitions of ' ' can start matching anywhere after the start of the preceeding ([^\\.\\?\\!]*)[\\.\\?\\!] |

shared/regex/codeql/regex/nfa/SuperlinearBackTracking.qll

@@ -365,11 +365,19 @@ module Make<RegexTreeViewSig TreeImpl> {
       )
     }
 
-    string getChar(CharNode t) {
+    private string getCharInternal(CharNode t) {
       exists(InputSymbol s1, InputSymbol s2, InputSymbol s3 | t = Step(s1, s2, s3, _) |
         result = getAThreewayIntersect(s1, s2, s3)
       )
     }
+
+    string getChar(CharNode t) {
+      result = getCharInternal(t) and
+      not (
+        // skip the upper-case char if we have the lower-case version.
+        result.toLowerCase() != result and result.toLowerCase() = getCharInternal(t)
+      )
+    }
   }
 
   /**