Merge pull request #7097 from erik-krogh/railsReDoS

JS/PY/RB: support a limited number of ranges for ReDoS analysis

Author: erik-krogh

javascript/ql/lib/semmle/javascript/security/performance/ReDoSUtil.qll

@@ -539,6 +539,55 @@ private class EdgeLabel extends TInputSymbol {
   }
 }
 
+/**
+ * A RegExp term that acts like a plus.
+ * Either it's a RegExpPlus, or it is a range {1,X} where X is >= 30.
+ * 30 has been chosen as a threshold because for exponential blowup 2^30 is enough to get a decent DOS attack.
+ */
+private class EffectivelyPlus extends RegExpTerm {
+  EffectivelyPlus() {
+    this instanceof RegExpPlus
+    or
+    exists(RegExpRange range |
+      range.getLowerBound() = 1 and
+      (range.getUpperBound() >= 30 or not exists(range.getUpperBound()))
+    |
+      this = range
+    )
+  }
+}
+
+/**
+ * A RegExp term that acts like a star.
+ * Either it's a RegExpStar, or it is a range {0,X} where X is >= 30.
+ */
+private class EffectivelyStar extends RegExpTerm {
+  EffectivelyStar() {
+    this instanceof RegExpStar
+    or
+    exists(RegExpRange range |
+      range.getLowerBound() = 0 and
+      (range.getUpperBound() >= 30 or not exists(range.getUpperBound()))
+    |
+      this = range
+    )
+  }
+}
+
+/**
+ * A RegExp term that acts like a question mark.
+ * Either it's a RegExpQuestion, or it is a range {0,1}.
+ */
+private class EffectivelyQuestion extends RegExpTerm {
+  EffectivelyQuestion() {
+    this instanceof RegExpOpt
+    or
+    exists(RegExpRange range | range.getLowerBound() = 0 and range.getUpperBound() = 1 |
+      this = range
+    )
+  }
+}
+
 /**
  * Gets the state before matching `t`.
  */
@@ -559,14 +608,14 @@ State after(RegExpTerm t) {
   or
   exists(RegExpGroup grp | t = grp.getAChild() | result = after(grp))
   or
-  exists(RegExpStar star | t = star.getAChild() | result = before(star))
+  exists(EffectivelyStar star | t = star.getAChild() | result = before(star))
   or
-  exists(RegExpPlus plus | t = plus.getAChild() |
+  exists(EffectivelyPlus plus | t = plus.getAChild() |
     result = before(plus) or
     result = after(plus)
   )
   or
-  exists(RegExpOpt opt | t = opt.getAChild() | result = after(opt))
+  exists(EffectivelyQuestion opt | t = opt.getAChild() | result = after(opt))
   or
   exists(RegExpRoot root | t = root | result = AcceptAnySuffix(root))
 }
@@ -617,15 +666,17 @@ predicate delta(State q1, EdgeLabel lbl, State q2) {
   or
   exists(RegExpGroup grp | lbl = Epsilon() | q1 = before(grp) and q2 = before(grp.getChild(0)))
   or
-  exists(RegExpStar star | lbl = Epsilon() |
+  exists(EffectivelyStar star | lbl = Epsilon() |
     q1 = before(star) and q2 = before(star.getChild(0))
     or
     q1 = before(star) and q2 = after(star)
   )
   or
-  exists(RegExpPlus plus | lbl = Epsilon() | q1 = before(plus) and q2 = before(plus.getChild(0)))
+  exists(EffectivelyPlus plus | lbl = Epsilon() |
+    q1 = before(plus) and q2 = before(plus.getChild(0))
+  )
   or
-  exists(RegExpOpt opt | lbl = Epsilon() |
+  exists(EffectivelyQuestion opt | lbl = Epsilon() |
     q1 = before(opt) and q2 = before(opt.getChild(0))
     or
     q1 = before(opt) and q2 = after(opt)

javascript/ql/test/query-tests/Performance/ReDoS/PolynomialBackTracking.expected

@@ -201,6 +201,7 @@
 | regexplib/markup.js:13:14:13:16 | .+? | Strings starting with '<' and with many repetitions of '!' can start matching anywhere after the start of the preceeding .*? |
 | regexplib/markup.js:14:13:14:14 | .* | Strings starting with '<' and with many repetitions of 'a' can start matching anywhere after the start of the preceeding .* |
 | regexplib/markup.js:14:24:14:25 | .* | Strings starting with '<>' and with many repetitions of '>a' can start matching anywhere after the start of the preceeding .* |
+| regexplib/markup.js:15:16:15:18 | .*? | Strings starting with '<img' and with many repetitions of '<imga' can start matching anywhere after the start of the preceeding <(\\/{0,1})img(.*?)(\\/{0,1})\\> |
 | regexplib/markup.js:16:5:16:9 | [^>]* | Strings starting with 'src' and with many repetitions of 'src' can start matching anywhere after the start of the preceeding src[^>]*[^/].(?:jpg\|bmp\|gif)(?:\\"\|\\') |
 | regexplib/markup.js:17:8:17:24 | (\\s(\\w*=".*?")?)* | Strings starting with '<a' and with many repetitions of ' =""' can start matching anywhere after the start of the preceeding .*? |
 | regexplib/markup.js:17:12:17:14 | \\w* | Strings starting with '<a ' and with many repetitions of '="" a' can start matching anywhere after the start of the preceeding .*? |
@@ -213,6 +214,10 @@
 | regexplib/markup.js:20:197:20:198 | "+ | Strings with many repetitions of '""' can start matching anywhere after the start of the preceeding "+ |
 | regexplib/markup.js:20:245:20:247 | .*? | Strings with many repetitions of 'color: # IF found THEN move ahead "" # single or double  # or no quotes\\t' can start matching anywhere after the start of the preceeding .*? |
 | regexplib/markup.js:20:274:20:276 | .*? | Strings starting with '<font # Match start of Font Tag ' and with many repetitions of '<font # Match start of Font Tag a' can start matching anywhere after the start of the preceeding <\\*?font # Match start of Font Tag (?(?=[^>]+color.*>) #IF\\/THEN lookahead color in tag (.*?color\\s*?[=\|:]\\s*?) # IF found THEN move ahead ('+\\#*?[\\w\\s]*'+ # CAPTURE ColorName\\/Hex \|"+\\#*?[\\w\\s]*"+ # single or double \|\\#*\\w*\\b) # or no quotes\t.*?> # & move to end of tag \|.*?> # ELSE move to end of Tag ) # Close the If\\/Then lookahead # Use Multiline and IgnoreCase # Replace the matches from RE with MatchEvaluator below: # if m.Groups(1).Value<>"" then # Return "<font color=" & m.Groups(1).Value & ">" # else # Return "<font>" # end if |
+| regexplib/markup.js:24:39:24:41 | \\s+ | Strings starting with '&lt;A' and with many repetitions of ' - != ' can start matching anywhere after the start of the preceeding \\s* |
+| regexplib/markup.js:24:43:24:45 | \\S+ | Strings starting with '&lt;A ' and with many repetitions of '- !=' can start matching anywhere after the start of the preceeding \\s* |
+| regexplib/markup.js:24:48:24:50 | \\s* | Strings starting with '&lt;A !' and with many repetitions of ' =- ! ' can start matching anywhere after the start of the preceeding \\s+ |
+| regexplib/markup.js:24:52:24:54 | \\s* | Strings starting with '&lt;A !=' and with many repetitions of '- !=' can start matching anywhere after the start of the preceeding \\S+ |
 | regexplib/markup.js:25:11:25:15 | [^>]* | Strings starting with '<A' and with many repetitions of '<A' can start matching anywhere after the start of the preceeding <[a-zA-Z][^>]*\\son\\w+=(\\w+\|'[^']*'\|"[^"]*")[^>]*> |
 | regexplib/markup.js:25:45:25:49 | [^>]* | Strings starting with '<A ona=a' and with many repetitions of '0' can start matching anywhere after the start of the preceeding \\w+ |
 | regexplib/markup.js:27:3:27:7 | [^>]* | Strings starting with '<' and with many repetitions of '<' can start matching anywhere after the start of the preceeding <[^>]*name[\\s]*=[\\s]*"?[^\\w_]*"?[^>]*> |
@@ -228,6 +233,10 @@
 | regexplib/markup.js:44:3:44:7 | [^>]* | Strings starting with '<' and with many repetitions of '<' can start matching anywhere after the start of the preceeding <[^>]*name[\\s]*=[\\s]*"?[^\\w_]*"?[^>]*> |
 | regexplib/markup.js:44:34:44:38 | [^>]* | Strings starting with '<name=' and with many repetitions of '\\t' can start matching anywhere after the start of the preceeding [\\s]* |
 | regexplib/markup.js:45:6:45:13 | [\\d\\D]*? | Strings starting with '/*' and with many repetitions of 'a/*' can start matching anywhere after the start of the preceeding \\/\\*[\\d\\D]*?\\*\\/ |
+| regexplib/markup.js:47:39:47:41 | \\s+ | Strings starting with '&lt;A' and with many repetitions of ' - != ' can start matching anywhere after the start of the preceeding \\s* |
+| regexplib/markup.js:47:43:47:45 | \\S+ | Strings starting with '&lt;A ' and with many repetitions of '- !=' can start matching anywhere after the start of the preceeding \\s* |
+| regexplib/markup.js:47:48:47:50 | \\s* | Strings starting with '&lt;A !' and with many repetitions of ' =- ! ' can start matching anywhere after the start of the preceeding \\s+ |
+| regexplib/markup.js:47:52:47:54 | \\s* | Strings starting with '&lt;A !=' and with many repetitions of '- !=' can start matching anywhere after the start of the preceeding \\S+ |
 | regexplib/markup.js:48:6:48:13 | [\\s\\S]*? | Strings starting with '<!--' and with many repetitions of '<!--' can start matching anywhere after the start of the preceeding <!--[\\s\\S]*?--> |
 | regexplib/markup.js:53:15:53:19 | [\\w]* | Strings starting with '[a' and with many repetitions of '0' can start matching anywhere after the start of the preceeding \\w+ |
 | regexplib/markup.js:56:23:56:25 | \\w+ | Strings with many repetitions of 'a' can start matching anywhere after the start of the preceeding (\\/?(?<step>\\w+))+ |
@@ -300,6 +309,7 @@
 | regexplib/strings.js:14:61:14:63 | \\w* | Strings starting with 'AA' and with many repetitions of 'A' can start matching anywhere after the start of the preceeding \\w* |
 | regexplib/strings.js:14:107:14:109 | \\w* | Strings starting with 'AAA' and with many repetitions of 'A' can start matching anywhere after the start of the preceeding \\w* |
 | regexplib/strings.js:19:31:19:57 | [a-z&#230;&#248;&#229;0-9]+ | Strings starting with '#@' and with many repetitions of '##' can start matching anywhere after the start of the preceeding [a-z&#230;&#248;&#229;0-9]+ |
+| regexplib/strings.js:19:69:19:95 | [a-z&#230;&#248;&#229;0-9]+ | Strings starting with '#@#' and with many repetitions of '##' can start matching anywhere after the start of the preceeding [a-z&#230;&#248;&#229;0-9]+ |
 | regexplib/strings.js:20:3:20:20 | ((\\\\")\|[^"(\\\\")])+ | Strings starting with '"' and with many repetitions of '\\\\"' can start matching anywhere after the start of the preceeding "((\\\\")\|[^"(\\\\")])+" |
 | regexplib/strings.js:21:3:21:7 | [^>]+ | Strings starting with '<' and with many repetitions of '<' can start matching anywhere after the start of the preceeding <[^>]+> |
 | regexplib/strings.js:23:3:23:20 | ((\\\\")\|[^"(\\\\")])+ | Strings starting with '"' and with many repetitions of '\\\\"' can start matching anywhere after the start of the preceeding "((\\\\")\|[^"(\\\\")])+" |
@@ -313,8 +323,10 @@
 | regexplib/strings.js:40:3:40:5 | \\w+ | Strings with many repetitions of 'a' can start matching anywhere after the start of the preceeding (\\w+)\\s+\\1 |
 | regexplib/strings.js:48:3:48:12 | [^\\.\\?\\!]* | Strings with many repetitions of ' ' can start matching anywhere after the start of the preceeding ([^\\.\\?\\!]*)[\\.\\?\\!] |
 | regexplib/strings.js:49:3:49:5 | \\S+ | Strings with many repetitions of '!' can start matching anywhere after the start of the preceeding (\\S+)\\x20{2,}(?=\\S+) |
-| regexplib/strings.js:53:25:53:33 | [a-z0-9]+ | Strings with many repetitions of '0' can start matching anywhere after the start of the preceeding [a-z0-9]+ |
-| regexplib/strings.js:53:65:53:73 | [a-z0-9]+ | Strings with many repetitions of '0' can start matching anywhere after the start of the preceeding [a-z0-9]+ |
+| regexplib/strings.js:53:4:53:12 | [a-z0-9]+ | Strings with many repetitions of '0.00' can start matching anywhere after the start of the preceeding [a-z0-9]+ |
+| regexplib/strings.js:53:25:53:33 | [a-z0-9]+ | Strings starting with '0' and with many repetitions of '0' can start matching anywhere after the start of the preceeding [a-z0-9]+ |
+| regexplib/strings.js:53:44:53:52 | [a-z0-9]+ | Strings with many repetitions of '00' can start matching anywhere after the start of the preceeding [a-z0-9]+ |
+| regexplib/strings.js:53:65:53:73 | [a-z0-9]+ | Strings starting with '0' and with many repetitions of '0' can start matching anywhere after the start of the preceeding [a-z0-9]+ |
 | regexplib/strings.js:54:20:54:22 | \\w+ | Strings with many repetitions of 'a' 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 'PRN.' 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) |
@@ -519,3 +531,6 @@
 | tst.js:399:6:399:12 | (d\|dd)* | Strings with many repetitions of 'd' can start matching anywhere after the start of the preceeding ((c\|cc)*\|(d\|dd)*\|(e\|ee)*)f$ |
 | tst.js:400:6:401:1 | (e\|ee)* | Strings with many repetitions of 'e' can start matching anywhere after the start of the preceeding ((c\|cc)*\|(d\|dd)*\|(e\|ee)*)f$ |
 | tst.js:404:6:405:7 | (g\|gg)* | Strings with many repetitions of 'g' can start matching anywhere after the start of the preceeding (g\|gg)*h$ |
+| tst.js:407:128:407:129 |  * | Strings starting with '0/*' and with many repetitions of ' ' can start matching anywhere after the start of the preceeding \\s* |
+| tst.js:409:23:409:29 | [\\w.-]* | Strings starting with '//' and with many repetitions of '//' can start matching anywhere after the start of the preceeding (\\/(?:\\/[\\w.-]*)*){0,1}:([\\w.-]+) |
+| tst.js:411:15:411:19 | a{1,} | Strings with many repetitions of 'a' can start matching anywhere after the start of the preceeding (a{1,})* |