Merge pull request #136 from microsoft/powershell-first-query-preps

PS: Add AST and CFG classes for operator `&` and add environment variable reads as local flow sources

Author: MathiasVP

powershell/ql/lib/semmle/code/powershell/Command.qll

@@ -1,4 +1,5 @@
 import powershell
+
 private predicate parseCommandName(Cmd cmd, string namespace, string name) {
   exists(string qualified | command(cmd, qualified, _, _, _) |
     namespace = qualified.regexpCapture("([^\\\\]+)\\\\([^\\\\]+)", 1) and
@@ -11,6 +12,7 @@ private predicate parseCommandName(Cmd cmd, string namespace, string name) {
   )
 }
 
+/** A call to a command. */
 class Cmd extends @command, CmdBase {
   override string toString() { result = "call to " + this.getQualifiedCommandName() }
 
@@ -88,3 +90,8 @@ class Cmd extends @command, CmdBase {
 
   Redirection getARedirection() { result = this.getRedirection(_) }
 }
+
+/** A call to operator `&`. */
+class CallOperator extends Cmd {
+  CallOperator() { this.getKind() = 28 }
+}

powershell/ql/lib/semmle/code/powershell/Variable.qll

@@ -273,6 +273,21 @@ class LocalVariable extends AbstractLocalScopeVariable, TLocalVariable {
   final override Scope getDeclaringScope() { result = scope }
 }
 
+/**
+ * A variable of the form `$Env:HOME`.
+ */
+class EnvVariable extends Variable {
+  string var;
+
+  EnvVariable() { this.getName() = ["env:", "Env:"] + var }
+
+  /**
+   * Gets the part of the variable name that represens which environment
+   * variable.
+   */
+  string getEnvironmentVariable() { result = var }
+}
+
 class Parameter extends AbstractLocalScopeVariable, TParameter {
   ParameterImpl p;
 

powershell/ql/lib/semmle/code/powershell/VariableExpression.qll

@@ -48,3 +48,12 @@ class VarWriteAccess extends VarAccess {
 
   predicate isImplicit() { isImplicitVariableWriteAccess(this) }
 }
+
+/** An access to an environment variable such as `$Env:PATH` */
+class EnvVarAccess extends VarAccess {
+  EnvVarAccess() { super.getVariable() instanceof EnvVariable }
+
+  override EnvVariable getVariable() { result = super.getVariable() }
+
+  string getEnvironmentVariable() { result = this.getVariable().getEnvironmentVariable() }
+}

powershell/ql/lib/semmle/code/powershell/controlflow/CfgNodes.qll

@@ -597,6 +597,11 @@ module StmtNodes {
     final override string getName() { result = s.getCmdName().getValue().getValue() }
   }
 
+  /** A control-flow node that wraps a call to operator `&` */
+  class CallOperatorCfgNode extends CmdCfgNode {
+    CallOperatorCfgNode() { this.getStmt() instanceof CallOperator }
+  }
+
   private class AssignStmtChildMapping extends PipelineBaseChildMapping, AssignStmt {
     override predicate relevantChild(Ast n) {
       n = this.getLeftHandSide() or n = this.getRightHandSide()

powershell/ql/lib/semmle/code/powershell/dataflow/flowsources/FlowSources.qll

@@ -1,5 +1,5 @@
 /** Provides classes representing various flow sources for taint tracking. */
-import semmle.code.powershell.dataflow.internal.DataFlowPublic as DataFlow
+private import semmle.code.powershell.dataflow.internal.DataFlowPublic as DataFlow
 import semmle.code.powershell.dataflow.flowsources.Remote
 import semmle.code.powershell.dataflow.flowsources.Local
 import semmle.code.powershell.frameworks.data.internal.ApiGraphModels

powershell/ql/lib/semmle/code/powershell/dataflow/flowsources/Local.qll

@@ -30,6 +30,12 @@ abstract class EnvironmentVariableSource extends LocalFlowSource {
   override string getSourceType() { result = "environment variable" }
 }
 
+private class EnvironmentVariableEnv extends EnvironmentVariableSource {
+  EnvironmentVariableEnv() {
+    this.asExpr().getExpr().(VarReadAccess).getVariable() instanceof EnvVariable
+  }
+}
+
 private class ExternalEnvironmentVariableSource extends EnvironmentVariableSource {
   ExternalEnvironmentVariableSource() {
     this = ModelOutput::getASourceNode("environment", _).asSource()

powershell/ql/lib/semmle/code/powershell/dataflow/internal/DataFlowPublic.qll

@@ -448,9 +448,30 @@ class CallNode extends AstNode {
 
   Node getQualifier() { result.asExpr() = call.getQualifier() }
 
+  /** Gets the i'th argument to this call. */
+  Node getArgument(int i) { result.asExpr() = call.getArgument(i) }
+
+  /** Gets the i'th positional argument to this call. */
+  Node getPositionalArgument(int i) { result.asExpr() = call.getPositionalArgument(i) }
+
+  /** Gets the argument with the name `name`, if any. */
+  Node getNamedArgument(string name) { result.asExpr() = call.getNamedArgument(name) }
+
+  /**
+   * Gets any argument of this call.
+   *
+   * Note that this predicate doesn't get the pipeline argument, if any.
+   */
+  Node getAnArgument() { result.asExpr() = call.getAnArgument() }
+
   int getNumberOfArguments() { result = call.getNumberOfArguments() }
 }
 
+/** A call to operator `&`, viwed as a node in a data flow graph. */
+class CallOperatorNode extends CallNode {
+  CallOperatorNode() { this.getCallNode() instanceof CfgNodes::StmtNodes::CallOperatorCfgNode }
+}
+
 /** A use of a type name, viewed as a node in a data flow graph. */
 class TypeNameNode extends ExprNode {
   override CfgNodes::ExprNodes::TypeNameCfgNode n;

powershell/ql/lib/semmle/code/powershell/dataflow/internal/TaintTrackingPrivate.qll

@@ -65,3 +65,15 @@ private module Cached {
 }
 
 import Cached
+import SpeculativeTaintFlow
+
+private module SpeculativeTaintFlow {
+  private import semmle.code.powershell.dataflow.internal.DataFlowDispatch as DataFlowDispatch
+  private import semmle.code.powershell.dataflow.internal.DataFlowPublic as DataFlowPublic
+
+  /**
+   * Holds if the additional step from `src` to `sink` should be considered in
+   * speculative taint flow exploration.
+   */
+  predicate speculativeTaintStep(DataFlow::Node src, DataFlow::Node sink) { none() }
+}