Merge pull request #6693 from yoff/python/promote-regex-injection

Python: Promote `py/regex-injection`
This commit is contained in:
Rasmus Wriedt Larsen
2021-10-14 14:49:05 +02:00
committed by GitHub
20 changed files with 318 additions and 377 deletions

View File

@@ -355,6 +355,53 @@ module SqlExecution {
}
}
/**
* A data-flow node that executes a regular expression.
*
* Extend this class to refine existing API models. If you want to model new APIs,
* extend `RegexExecution::Range` instead.
*/
class RegexExecution extends DataFlow::Node {
RegexExecution::Range range;
RegexExecution() { this = range }
/** Gets the data flow node for the regex being executed by this node. */
DataFlow::Node getRegex() { result = range.getRegex() }
/** Gets a dataflow node for the string to be searched or matched against. */
DataFlow::Node getString() { result = range.getString() }
/**
* Gets the name of this regex execution, typically the name of an executing method.
* This is used for nice alert messages and should include the module if possible.
*/
string getName() { result = range.getName() }
}
/** Provides classes for modeling new regular-expression execution APIs. */
module RegexExecution {
/**
* A data-flow node that executes a regular expression.
*
* Extend this class to model new APIs. If you want to refine existing API models,
* extend `RegexExecution` instead.
*/
abstract class Range extends DataFlow::Node {
/** Gets the data flow node for the regex being executed by this node. */
abstract DataFlow::Node getRegex();
/** Gets a dataflow node for the string to be searched or matched against. */
abstract DataFlow::Node getString();
/**
* Gets the name of this regex execution, typically the name of an executing method.
* This is used for nice alert messages and should include the module if possible.
*/
abstract string getName();
}
}
/**
* A data-flow node that escapes meta-characters, which could be used to prevent
* injection attacks.
@@ -411,6 +458,9 @@ module Escaping {
/** Gets the escape-kind for escaping a string so it can safely be included in HTML. */
string getHtmlKind() { result = "html" }
/** Gets the escape-kind for escaping a string so it can safely be included in HTML. */
string getRegexKind() { result = "regex" }
// TODO: If adding an XML kind, update the modeling of the `MarkupSafe` PyPI package.
//
// Technically it claims to escape for both HTML and XML, but for now we don't have
@@ -427,6 +477,14 @@ class HtmlEscaping extends Escaping {
HtmlEscaping() { range.getKind() = Escaping::getHtmlKind() }
}
/**
* An escape of a string so it can be safely included in
* the body of a regex.
*/
class RegexEscaping extends Escaping {
RegexEscaping() { range.getKind() = Escaping::getRegexKind() }
}
/** Provides classes for modeling HTTP-related APIs. */
module HTTP {
import semmle.python.web.HttpConstants

View File

@@ -49,6 +49,7 @@ newtype TRegExpParent =
* or another regular expression term.
*/
class RegExpParent extends TRegExpParent {
/** Gets a textual representation of this element. */
string toString() { result = "RegExpParent" }
/** Gets the `i`th child term. */
@@ -72,14 +73,18 @@ class RegExpLiteral extends TRegExpLiteral, RegExpParent {
override RegExpTerm getChild(int i) { i = 0 and result.getRegex() = re and result.isRootTerm() }
/** Holds if dot, `.`, matches all characters, including newlines. */
predicate isDotAll() { re.getAMode() = "DOTALL" }
/** Holds if this regex matching is case-insensitive for this regex. */
predicate isIgnoreCase() { re.getAMode() = "IGNORECASE" }
/** Get a string representing all modes for this regex. */
string getFlags() { result = concat(string mode | mode = re.getAMode() | mode, " | ") }
override Regex getRegex() { result = re }
/** Gets the primary QL class for this regex. */
string getPrimaryQLClass() { result = "RegExpLiteral" }
}
@@ -246,8 +251,10 @@ class RegExpQuantifier extends RegExpTerm, TRegExpQuantifier {
result.getEnd() = part_end
}
/** Hols if this term may match an unlimited number of times. */
predicate mayRepeatForever() { may_repeat_forever = true }
/** Gets the qualifier for this term. That is e.g "?" for "a?". */
string getQualifier() { result = re.getText().substring(part_end, end) }
override string getPrimaryQLClass() { result = "RegExpQuantifier" }
@@ -322,8 +329,10 @@ class RegExpRange extends RegExpQuantifier {
RegExpRange() { re.multiples(part_end, end, lower, upper) }
/** Gets the string defining the upper bound of this range, if any. */
string getUpper() { result = upper }
/** Gets the string defining the lower bound of this range, if any. */
string getLower() { result = lower }
/**
@@ -465,11 +474,13 @@ class RegExpEscape extends RegExpNormalChar {
result = this.getUnicode()
}
/** Holds if this terms name is given by the part following the escape character. */
predicate isIdentityEscape() { not this.getUnescaped() in ["n", "r", "t", "f"] }
override string getPrimaryQLClass() { result = "RegExpEscape" }
string getUnescaped() { result = this.getText().suffix(1) }
/** Gets the part of the term following the escape character. That is e.g. "w" if the term is "\w". */
private string getUnescaped() { result = this.getText().suffix(1) }
/**
* Gets the text for this escape. That is e.g. "\w".
@@ -536,15 +547,8 @@ private int toHex(string hex) {
* ```
*/
class RegExpCharacterClassEscape extends RegExpEscape {
// string value;
RegExpCharacterClassEscape() {
// value = re.getText().substring(start + 1, end) and
// value in ["d", "D", "s", "S", "w", "W"]
this.getValue() in ["d", "D", "s", "S", "w", "W"]
}
RegExpCharacterClassEscape() { this.getValue() in ["d", "D", "s", "S", "w", "W"] }
/** Gets the name of the character class; for example, `w` for `\w`. */
// override string getValue() { result = value }
override RegExpTerm getChild(int i) { none() }
override string getPrimaryQLClass() { result = "RegExpCharacterClassEscape" }
@@ -563,10 +567,13 @@ class RegExpCharacterClassEscape extends RegExpEscape {
class RegExpCharacterClass extends RegExpTerm, TRegExpCharacterClass {
RegExpCharacterClass() { this = TRegExpCharacterClass(re, start, end) }
/** Holds if this character class is inverted, matching the opposite of its content. */
predicate isInverted() { re.getChar(start + 1) = "^" }
/** Gets the `i`th char inside this charater class. */
string getCharThing(int i) { result = re.getChar(i + start) }
/** Holds if this character class can match anything. */
predicate isUniversalClass() {
// [^]
this.isInverted() and not exists(this.getAChild())
@@ -620,6 +627,7 @@ class RegExpCharacterRange extends RegExpTerm, TRegExpCharacterRange {
re.charRange(_, start, lower_end, upper_start, end)
}
/** Holds if this range goes from `lo` to `hi`, in effect is `lo-hi`. */
predicate isRange(string lo, string hi) {
lo = re.getText().substring(start, lower_end) and
hi = re.getText().substring(upper_start, end)
@@ -653,8 +661,13 @@ class RegExpCharacterRange extends RegExpTerm, TRegExpCharacterRange {
class RegExpNormalChar extends RegExpTerm, TRegExpNormalChar {
RegExpNormalChar() { this = TRegExpNormalChar(re, start, end) }
/**
* Holds if this constant represents a valid Unicode character (as opposed
* to a surrogate code point that does not correspond to a character by itself.)
*/
predicate isCharacter() { any() }
/** Gets the string representation of the char matched by this term. */
string getValue() { result = re.getText().substring(start, end) }
override RegExpTerm getChild(int i) { none() }
@@ -684,15 +697,15 @@ class RegExpConstant extends RegExpTerm {
qstart <= start and end <= qend
) and
value = this.(RegExpNormalChar).getValue()
// This will never hold
// or
// this = TRegExpSpecialChar(re, start, end) and
// re.inCharSet(start) and
// value = this.(RegExpSpecialChar).getChar()
}
/**
* Holds if this constant represents a valid Unicode character (as opposed
* to a surrogate code point that does not correspond to a character by itself.)
*/
predicate isCharacter() { any() }
/** Gets the string matched by this constant term. */
string getValue() { result = value }
override RegExpTerm getChild(int i) { none() }
@@ -731,10 +744,6 @@ class RegExpGroup extends RegExpTerm, TRegExpGroup {
/** Gets the name of this capture group, if any. */
string getName() { result = re.getGroupName(start, end) }
predicate isCharacter() { any() }
string getValue() { result = re.getText().substring(start, end) }
override RegExpTerm getChild(int i) {
result.getRegex() = re and
i = 0 and
@@ -762,8 +771,13 @@ class RegExpSpecialChar extends RegExpTerm, TRegExpSpecialChar {
re.specialCharacter(start, end, char)
}
/**
* Holds if this constant represents a valid Unicode character (as opposed
* to a surrogate code point that does not correspond to a character by itself.)
*/
predicate isCharacter() { any() }
/** Gets the char for this term. */
string getChar() { result = char }
override RegExpTerm getChild(int i) { none() }
@@ -828,8 +842,6 @@ class RegExpCaret extends RegExpSpecialChar {
class RegExpZeroWidthMatch extends RegExpGroup {
RegExpZeroWidthMatch() { re.zeroWidthMatch(start, end) }
override predicate isCharacter() { any() }
override RegExpTerm getChild(int i) { none() }
override string getPrimaryQLClass() { result = "RegExpZeroWidthMatch" }

View File

@@ -1636,6 +1636,119 @@ private module StdlibPrivate {
result = this.getArg(any(int i | i >= msgIndex))
}
}
// ---------------------------------------------------------------------------
// re
// ---------------------------------------------------------------------------
/**
* List of methods in the `re` module immediately executing a regular expression.
*
* See https://docs.python.org/3/library/re.html#module-contents
*/
private class RegexExecutionMethod extends string {
RegexExecutionMethod() {
this in ["match", "fullmatch", "search", "split", "findall", "finditer", "sub", "subn"]
}
/** Gets the index of the argument representing the string to be searched by a regex. */
int getStringArgIndex() {
this in ["match", "fullmatch", "search", "split", "findall", "finditer"] and
result = 1
or
this in ["sub", "subn"] and
result = 2
}
}
/**
* A a call to a method from the `re` module immediately executing a regular expression.
*
* See `RegexExecutionMethods`
*/
private class DirectRegexExecution extends DataFlow::CallCfgNode, RegexExecution::Range {
RegexExecutionMethod method;
DirectRegexExecution() { this = API::moduleImport("re").getMember(method).getACall() }
override DataFlow::Node getRegex() { result in [this.getArg(0), this.getArgByName("pattern")] }
override DataFlow::Node getString() {
result in [this.getArg(method.getStringArgIndex()), this.getArgByName("string")]
}
override string getName() { result = "re." + method }
}
/** Helper module for tracking compiled regexes. */
private module CompiledRegexes {
private DataFlow::TypeTrackingNode compiledRegex(DataFlow::TypeTracker t, DataFlow::Node regex) {
t.start() and
result = API::moduleImport("re").getMember("compile").getACall() and
regex in [
result.(DataFlow::CallCfgNode).getArg(0),
result.(DataFlow::CallCfgNode).getArgByName("pattern")
]
or
exists(DataFlow::TypeTracker t2 | result = compiledRegex(t2, regex).track(t2, t))
}
DataFlow::Node compiledRegex(DataFlow::Node regex) {
compiledRegex(DataFlow::TypeTracker::end(), regex).flowsTo(result)
}
}
private import CompiledRegexes
/**
* A call on compiled regular expression (obtained via `re.compile`) executing a
* regular expression.
*
* Given the following example:
*
* ```py
* pattern = re.compile(input)
* pattern.match(s)
* ```
*
* This class will identify that `re.compile` compiles `input` and afterwards
* executes `re`'s `match`. As a result, `this` will refer to `pattern.match(s)`
* and `this.getRegexNode()` will return the node for `input` (`re.compile`'s first argument).
*
*
* See `RegexExecutionMethods`
*
* See https://docs.python.org/3/library/re.html#regular-expression-objects
*/
private class CompiledRegexExecution extends DataFlow::MethodCallNode, RegexExecution::Range {
DataFlow::Node regexNode;
RegexExecutionMethod method;
CompiledRegexExecution() { this.calls(compiledRegex(regexNode), method) }
override DataFlow::Node getRegex() { result = regexNode }
override DataFlow::Node getString() {
result in [this.getArg(method.getStringArgIndex() - 1), this.getArgByName("string")]
}
override string getName() { result = "re." + method }
}
/**
* A call to 're.escape'.
* See https://docs.python.org/3/library/re.html#re.escape
*/
private class ReEscapeCall extends Escaping::Range, DataFlow::CallCfgNode {
ReEscapeCall() { this = API::moduleImport("re").getMember("escape").getACall() }
override DataFlow::Node getAnInput() {
result in [this.getArg(0), this.getArgByName("pattern")]
}
override DataFlow::Node getOutput() { result = this }
override string getKind() { result = Escaping::getRegexKind() }
}
}
// ---------------------------------------------------------------------------

View File

@@ -60,8 +60,8 @@ module PolynomialReDoS {
RegExpTerm t;
RegexExecutionAsSink() {
exists(CompiledRegexes::RegexExecution re |
re.getRegexNode().asExpr() = t.getRegex() and
exists(RegexExecution re |
re.getRegex().asExpr() = t.getRegex() and
this = re.getString()
) and
t.isRootTerm()
@@ -76,137 +76,3 @@ module PolynomialReDoS {
*/
class StringConstCompareAsSanitizerGuard extends SanitizerGuard, StringConstCompare { }
}
/** Helper module for tracking compiled regexes. */
private module CompiledRegexes {
// TODO: This module should be refactored and merged with the experimental work done on detecting
// regex injections, such that this can be expressed from just using a concept.
/** A configuration for finding uses of compiled regexes. */
class RegexDefinitionConfiguration extends DataFlow2::Configuration {
RegexDefinitionConfiguration() { this = "RegexDefinitionConfiguration" }
override predicate isSource(DataFlow::Node source) { source instanceof RegexDefinitonSource }
override predicate isSink(DataFlow::Node sink) { sink instanceof RegexDefinitionSink }
}
/** A regex compilation. */
class RegexDefinitonSource extends DataFlow::CallCfgNode {
DataFlow::Node regexNode;
RegexDefinitonSource() {
this = API::moduleImport("re").getMember("compile").getACall() and
regexNode in [this.getArg(0), this.getArgByName("pattern")]
}
/** Gets the regex that is being compiled by this node. */
RegExpTerm getRegExp() { result.getRegex() = regexNode.asExpr() and result.isRootTerm() }
/** Gets the data flow node for the regex being compiled by this node. */
DataFlow::Node getRegexNode() { result = regexNode }
}
/** A use of a compiled regex. */
class RegexDefinitionSink extends DataFlow::Node {
RegexExecutionMethod method;
DataFlow::CallCfgNode executingCall;
RegexDefinitionSink() {
exists(DataFlow::AttrRead reMethod |
executingCall.getFunction() = reMethod and
reMethod.getAttributeName() = method and
this = reMethod.getObject()
)
}
/** Gets the method used to execute the regex. */
RegexExecutionMethod getMethod() { result = method }
/** Gets the data flow node for the executing call. */
DataFlow::CallCfgNode getExecutingCall() { result = executingCall }
}
/** A data flow node executing a regex. */
abstract class RegexExecution extends DataFlow::Node {
/** Gets the data flow node for the regex being compiled by this node. */
abstract DataFlow::Node getRegexNode();
/** Gets a dataflow node for the string to be searched or matched against. */
abstract DataFlow::Node getString();
}
private class RegexExecutionMethod extends string {
RegexExecutionMethod() {
this in ["match", "fullmatch", "search", "split", "findall", "finditer", "sub", "subn"]
}
}
/** Gets the index of the argument representing the string to be searched by a regex. */
int stringArg(RegexExecutionMethod method) {
method in ["match", "fullmatch", "search", "split", "findall", "finditer"] and
result = 1
or
method in ["sub", "subn"] and
result = 2
}
/**
* A class to find `re` methods immediately executing an expression.
*
* See `RegexExecutionMethods`
*/
class DirectRegex extends DataFlow::CallCfgNode, RegexExecution {
RegexExecutionMethod method;
DirectRegex() { this = API::moduleImport("re").getMember(method).getACall() }
override DataFlow::Node getRegexNode() {
result in [this.getArg(0), this.getArgByName("pattern")]
}
override DataFlow::Node getString() {
result in [this.getArg(stringArg(method)), this.getArgByName("string")]
}
}
/**
* A class to find `re` methods immediately executing a compiled expression by `re.compile`.
*
* Given the following example:
*
* ```py
* pattern = re.compile(input)
* pattern.match(s)
* ```
*
* This class will identify that `re.compile` compiles `input` and afterwards
* executes `re`'s `match`. As a result, `this` will refer to `pattern.match(s)`
* and `this.getRegexNode()` will return the node for `input` (`re.compile`'s first argument)
*
*
* See `RegexExecutionMethods`
*
* See https://docs.python.org/3/library/re.html#regular-expression-objects
*/
private class CompiledRegex extends DataFlow::CallCfgNode, RegexExecution {
DataFlow::Node regexNode;
RegexExecutionMethod method;
CompiledRegex() {
exists(
RegexDefinitionConfiguration conf, RegexDefinitonSource source, RegexDefinitionSink sink
|
conf.hasFlow(source, sink) and
regexNode = source.getRegexNode() and
method = sink.getMethod() and
this = sink.getExecutingCall()
)
}
override DataFlow::Node getRegexNode() { result = regexNode }
override DataFlow::Node getString() {
result in [this.getArg(stringArg(method) - 1), this.getArgByName("string")]
}
}
}

View File

@@ -59,7 +59,7 @@ module ReflectedXSS {
class HtmlEscapingAsSanitizer extends Sanitizer {
HtmlEscapingAsSanitizer() {
// TODO: For now, since there is not an `isSanitizingStep` member-predicate part of a
// `TaintTracking::Configuration`, we use treat the output is a taint-sanitizer. This
// `TaintTracking::Configuration`, we treat the output as a taint-sanitizer. This
// is slightly imprecise, which you can see in the `m_unsafe + SAFE` test-case in
// python/ql/test/library-tests/frameworks/markupsafe/taint_test.py
//

View File

@@ -0,0 +1,37 @@
/**
* Provides a taint-tracking configuration for detecting regular expression injection
* vulnerabilities.
*
* Note, for performance reasons: only import this file if
* `RegexInjection::Configuration` is needed, otherwise
* `RegexInjectionCustomizations` should be imported instead.
*/
private import python
import semmle.python.dataflow.new.DataFlow
import semmle.python.dataflow.new.TaintTracking
/**
* Provides a taint-tracking configuration for detecting regular expression injection
* vulnerabilities.
*/
module RegexInjection {
import RegexInjectionCustomizations::RegexInjection
/**
* A taint-tracking configuration for detecting "reflected server-side cross-site scripting" vulnerabilities.
*/
class Configuration extends TaintTracking::Configuration {
Configuration() { this = "RegexInjection" }
override predicate isSource(DataFlow::Node source) { source instanceof Source }
override predicate isSink(DataFlow::Node sink) { sink instanceof Sink }
override predicate isSanitizer(DataFlow::Node node) { node instanceof Sanitizer }
override predicate isSanitizerGuard(DataFlow::BarrierGuard guard) {
guard instanceof SanitizerGuard
}
}
}

View File

@@ -0,0 +1,62 @@
/**
* Provides default sources, sinks and sanitizers for detecting
* "regular expression injection"
* vulnerabilities, as well as extension points for adding your own.
*/
private import python
private import semmle.python.Concepts
private import semmle.python.dataflow.new.DataFlow
private import semmle.python.dataflow.new.TaintTracking
private import semmle.python.dataflow.new.RemoteFlowSources
/**
* Provides default sources, sinks and sanitizers for detecting
* "regular expression injection"
* vulnerabilities, as well as extension points for adding your own.
*/
module RegexInjection {
/**
* A data flow source for "regular expression injection" vulnerabilities.
*/
abstract class Source extends DataFlow::Node { }
/**
* A sink for "regular expression injection" vulnerabilities is the execution of a regular expression.
* If you have a custom way to execute regular expressions, you can extend `RegexExecution::Range`.
*/
class Sink extends DataFlow::Node {
RegexExecution regexExecution;
Sink() { this = regexExecution.getRegex() }
/** Gets the call that executes the regular expression marked by this sink. */
RegexExecution getRegexExecution() { result = regexExecution }
}
/**
* A sanitizer for "regular expression injection" vulnerabilities.
*/
abstract class Sanitizer extends DataFlow::Node { }
/**
* A sanitizer guard for "regular expression injection" vulnerabilities.
*/
abstract class SanitizerGuard extends DataFlow::BarrierGuard { }
/**
* A source of remote user input, considered as a flow source.
*/
class RemoteFlowSourceAsSource extends Source, RemoteFlowSource { }
/**
* A regex escaping, considered as a sanitizer.
*/
class RegexEscapingAsSanitizer extends Sanitizer {
RegexEscapingAsSanitizer() {
// Due to use-use flow, we want the output rather than an input
// (so the input can still flow to other sinks).
this = any(RegexEscaping esc).getOutput()
}
}
}