Merge pull request #13324 from jcogs33/jcogs33/shared-sink-kind-validation

Shared: share MaD kind validation across languages

csharp/ql/lib/qlpack.yml

@@ -6,6 +6,7 @@ extractor: csharp
 library: true
 upgrades: upgrades
 dependencies:
+  codeql/mad: ${workspace}
   codeql/ssa: ${workspace}
   codeql/tutorial: ${workspace}
   codeql/util: ${workspace}

csharp/ql/lib/semmle/code/csharp/dataflow/ExternalFlow.qll

@@ -95,6 +95,7 @@ private import internal.DataFlowPublic
 private import internal.FlowSummaryImpl::Public
 private import internal.FlowSummaryImpl::Private::External
 private import internal.FlowSummaryImplSpecific
+private import codeql.mad.ModelValidation as SharedModelVal
 
 /** Holds if a source model exists for the given parameters. */
 predicate sourceModel = Extensions::sourceModel/9;
@@ -204,30 +205,18 @@ module ModelValidation {
     )
   }
 
-  private string getInvalidModelKind() {
-    exists(string kind | summaryModel(_, _, _, _, _, _, _, _, kind, _) |
-      not kind = ["taint", "value"] and
-      result = "Invalid kind \"" + kind + "\" in summary model."
-    )
-    or
-    exists(string kind | sinkModel(_, _, _, _, _, _, _, kind, _) |
-      not kind =
-        ["code-injection", "sql-injection", "js-injection", "html-injection", "file-content-store"] and
-      not kind.matches("encryption-%") and
-      result = "Invalid kind \"" + kind + "\" in sink model."
-    )
-    or
-    exists(string kind | sourceModel(_, _, _, _, _, _, _, kind, _) |
-      not kind = ["local", "remote", "file", "file-write"] and
-      result = "Invalid kind \"" + kind + "\" in source model."
-    )
-    or
-    exists(string kind | neutralModel(_, _, _, _, kind, _) |
-      not kind = ["summary", "source", "sink"] and
-      result = "Invalid kind \"" + kind + "\" in neutral model."
-    )
+  private module KindValConfig implements SharedModelVal::KindValidationConfigSig {
+    predicate summaryKind(string kind) { summaryModel(_, _, _, _, _, _, _, _, kind, _) }
+
+    predicate sinkKind(string kind) { sinkModel(_, _, _, _, _, _, _, kind, _) }
+
+    predicate sourceKind(string kind) { sourceModel(_, _, _, _, _, _, _, kind, _) }
+
+    predicate neutralKind(string kind) { neutralModel(_, _, _, _, kind, _) }
   }
 
+  private module KindVal = SharedModelVal::KindValidation<KindValConfig>;
+
   private string getInvalidModelSignature() {
     exists(
       string pred, string namespace, string type, string name, string signature, string ext,
@@ -269,7 +258,7 @@ module ModelValidation {
     msg =
       [
         getInvalidModelSignature(), getInvalidModelInput(), getInvalidModelOutput(),
-        getInvalidModelKind()
+        KindVal::getInvalidModelKind()
       ]
   }
 }

go/ql/lib/qlpack.yml

@@ -6,6 +6,7 @@ extractor: go
 library: true
 upgrades: upgrades
 dependencies:
+  codeql/mad: ${workspace}
   codeql/tutorial: ${workspace}
   codeql/util: ${workspace}
 dataExtensions:

go/ql/lib/semmle/go/dataflow/ExternalFlow.qll

@@ -80,6 +80,7 @@ private import internal.FlowSummaryImpl::Private::External
 private import internal.FlowSummaryImplSpecific
 private import internal.AccessPathSyntax
 private import FlowSummary
+private import codeql.mad.ModelValidation as SharedModelVal
 
 /**
  * A module importing the frameworks that provide external flow data,
@@ -200,13 +201,16 @@ module ModelValidation {
     )
   }
 
-  private string getInvalidModelKind() {
-    exists(string kind | summaryModel(_, _, _, _, _, _, _, _, kind, _) |
-      not kind = ["taint", "value"] and
-      result = "Invalid kind \"" + kind + "\" in summary model."
-    )
+  private module KindValConfig implements SharedModelVal::KindValidationConfigSig {
+    predicate summaryKind(string kind) { summaryModel(_, _, _, _, _, _, _, _, kind, _) }
+
+    predicate sinkKind(string kind) { sinkModel(_, _, _, _, _, _, _, kind, _) }
+
+    predicate sourceKind(string kind) { sourceModel(_, _, _, _, _, _, _, kind, _) }
   }
 
+  private module KindVal = SharedModelVal::KindValidation<KindValConfig>;
+
   private string getInvalidModelSignature() {
     exists(
       string pred, string package, string type, string name, string signature, string ext,
@@ -243,7 +247,7 @@ module ModelValidation {
     msg =
       [
         getInvalidModelSignature(), getInvalidModelInput(), getInvalidModelOutput(),
-        getInvalidModelKind()
+        KindVal::getInvalidModelKind()
       ]
   }
 }

java/ql/lib/qlpack.yml

@@ -6,6 +6,7 @@ extractor: java
 library: true
 upgrades: upgrades
 dependencies:
+  codeql/mad: ${workspace}
   codeql/regex: ${workspace}
   codeql/tutorial: ${workspace}
   codeql/typetracking: ${workspace}

java/ql/lib/semmle/code/java/dataflow/ExternalFlow.qll

@@ -87,6 +87,7 @@ private import internal.FlowSummaryImplSpecific as FlowSummaryImplSpecific
 private import internal.AccessPathSyntax
 private import ExternalFlowExtensions as Extensions
 private import FlowSummary
+private import codeql.mad.ModelValidation as SharedModelVal
 
 /**
  * A class for activating additional model rows.
@@ -265,86 +266,17 @@ module ModelValidation {
     )
   }
 
-  private class OutdatedSinkKind extends string {
-    OutdatedSinkKind() {
-      this =
-        [
-          "sql", "url-redirect", "xpath", "ssti", "logging", "groovy", "jexl", "mvel", "xslt",
-          "ldap", "pending-intent-sent", "intent-start", "set-hostname-verifier",
-          "header-splitting", "xss", "write-file", "create-file", "read-file", "open-url",
-          "jdbc-url"
-        ]
-    }
+  private module KindValConfig implements SharedModelVal::KindValidationConfigSig {
+    predicate summaryKind(string kind) { summaryModel(_, _, _, _, _, _, _, _, kind, _) }
 
-    private string replacementKind() {
-      this = ["sql", "xpath", "groovy", "jexl", "mvel", "xslt", "ldap"] and
-      result = this + "-injection"
-      or
-      this = "url-redirect" and result = "url-redirection"
-      or
-      this = "ssti" and result = "template-injection"
-      or
-      this = "logging" and result = "log-injection"
-      or
-      this = "pending-intent-sent" and result = "pending-intents"
-      or
-      this = "intent-start" and result = "intent-redirection"
-      or
-      this = "set-hostname-verifier" and result = "hostname-verification"
-      or
-      this = "header-splitting" and result = "response-splitting"
-      or
-      this = "xss" and result = "html-injection\" or \"js-injection"
-      or
-      this = "write-file" and result = "file-content-store"
-      or
-      this = ["create-file", "read-file"] and result = "path-injection"
-      or
-      this = ["open-url", "jdbc-url"] and result = "request-forgery"
-    }
+    predicate sinkKind(string kind) { sinkModel(_, _, _, _, _, _, _, kind, _) }
 
-    string outdatedMessage() {
-      result =
-        "The kind \"" + this + "\" is outdated. Use \"" + this.replacementKind() + "\" instead."
-    }
+    predicate sourceKind(string kind) { sourceModel(_, _, _, _, _, _, _, kind, _) }
+
+    predicate neutralKind(string kind) { neutralModel(_, _, _, _, kind, _) }
   }
 
-  private string getInvalidModelKind() {
-    exists(string kind | summaryModel(_, _, _, _, _, _, _, _, kind, _) |
-      not kind = ["taint", "value"] and
-      result = "Invalid kind \"" + kind + "\" in summary model."
-    )
-    or
-    exists(string kind, string msg | sinkModel(_, _, _, _, _, _, _, kind, _) |
-      not kind =
-        [
-          "request-forgery", "jndi-injection", "ldap-injection", "sql-injection", "log-injection",
-          "mvel-injection", "xpath-injection", "groovy-injection", "html-injection", "js-injection",
-          "ognl-injection", "intent-redirection", "pending-intents", "url-redirection",
-          "path-injection", "file-content-store", "hostname-verification", "response-splitting",
-          "information-leak", "xslt-injection", "jexl-injection", "bean-validation",
-          "template-injection", "fragment-injection", "command-injection"
-        ] and
-      not kind.matches("regex-use%") and
-      not kind.matches("qltest%") and
-      msg = "Invalid kind \"" + kind + "\" in sink model." and
-      // The part of this message that refers to outdated sink kinds can be deleted after June 1st, 2024.
-      if kind instanceof OutdatedSinkKind
-      then result = msg + " " + kind.(OutdatedSinkKind).outdatedMessage()
-      else result = msg
-    )
-    or
-    exists(string kind | sourceModel(_, _, _, _, _, _, _, kind, _) |
-      not kind = ["remote", "contentprovider", "android-external-storage-dir"] and
-      not kind.matches("qltest%") and
-      result = "Invalid kind \"" + kind + "\" in source model."
-    )
-    or
-    exists(string kind | neutralModel(_, _, _, _, kind, _) |
-      not kind = ["summary", "source", "sink"] and
-      result = "Invalid kind \"" + kind + "\" in neutral model."
-    )
-  }
+  private module KindVal = SharedModelVal::KindValidation<KindValConfig>;
 
   private string getInvalidModelSignature() {
     exists(
@@ -387,7 +319,7 @@ module ModelValidation {
     msg =
       [
         getInvalidModelSignature(), getInvalidModelInput(), getInvalidModelOutput(),
-        getInvalidModelKind()
+        KindVal::getInvalidModelKind()
       ]
   }
 }

javascript/ql/lib/qlpack.yml

@@ -6,6 +6,7 @@ extractor: javascript
 library: true
 upgrades: upgrades
 dependencies:
+  codeql/mad: ${workspace}
   codeql/regex: ${workspace}
   codeql/tutorial: ${workspace}
   codeql/util: ${workspace}

javascript/ql/lib/semmle/javascript/frameworks/data/internal/ApiGraphModels.qll

@@ -653,6 +653,17 @@ module ModelOutput {
 
   import Cached
   import Specific::ModelOutputSpecific
+  private import codeql.mad.ModelValidation as SharedModelVal
+
+  private module KindValConfig implements SharedModelVal::KindValidationConfigSig {
+    predicate summaryKind(string kind) { summaryModel(_, _, _, _, kind) }
+
+    predicate sinkKind(string kind) { sinkModel(_, _, kind) }
+
+    predicate sourceKind(string kind) { sourceModel(_, _, kind) }
+  }
+
+  private module KindVal = SharedModelVal::KindValidation<KindValConfig>;
 
   /**
    * Gets an error message relating to an invalid CSV row in a model.
@@ -698,5 +709,8 @@ module ModelOutput {
       not isValidNoArgumentTokenInIdentifyingAccessPath(token.getName()) and
       result = "Invalid token '" + token + "' is missing its arguments, in access path: " + path
     )
+    or
+    // Check for invalid model kinds
+    result = KindVal::getInvalidModelKind()
   }
 }

python/ql/lib/qlpack.yml

@@ -6,6 +6,7 @@ extractor: python
 library: true
 upgrades: upgrades
 dependencies:
+  codeql/mad: ${workspace}
   codeql/regex: ${workspace}
   codeql/tutorial: ${workspace}
   codeql/util: ${workspace}

python/ql/lib/semmle/python/frameworks/data/internal/ApiGraphModels.qll

@@ -653,6 +653,17 @@ module ModelOutput {
 
   import Cached
   import Specific::ModelOutputSpecific
+  private import codeql.mad.ModelValidation as SharedModelVal
+
+  private module KindValConfig implements SharedModelVal::KindValidationConfigSig {
+    predicate summaryKind(string kind) { summaryModel(_, _, _, _, kind) }
+
+    predicate sinkKind(string kind) { sinkModel(_, _, kind) }
+
+    predicate sourceKind(string kind) { sourceModel(_, _, kind) }
+  }
+
+  private module KindVal = SharedModelVal::KindValidation<KindValConfig>;
 
   /**
    * Gets an error message relating to an invalid CSV row in a model.
@@ -698,5 +709,8 @@ module ModelOutput {
       not isValidNoArgumentTokenInIdentifyingAccessPath(token.getName()) and
       result = "Invalid token '" + token + "' is missing its arguments, in access path: " + path
     )
+    or
+    // Check for invalid model kinds
+    result = KindVal::getInvalidModelKind()
   }
 }

ruby/ql/lib/codeql/ruby/frameworks/data/internal/ApiGraphModels.qll

@@ -653,6 +653,17 @@ module ModelOutput {
 
   import Cached
   import Specific::ModelOutputSpecific
+  private import codeql.mad.ModelValidation as SharedModelVal
+
+  private module KindValConfig implements SharedModelVal::KindValidationConfigSig {
+    predicate summaryKind(string kind) { summaryModel(_, _, _, _, kind) }
+
+    predicate sinkKind(string kind) { sinkModel(_, _, kind) }
+
+    predicate sourceKind(string kind) { sourceModel(_, _, kind) }
+  }
+
+  private module KindVal = SharedModelVal::KindValidation<KindValConfig>;
 
   /**
    * Gets an error message relating to an invalid CSV row in a model.
@@ -698,5 +709,8 @@ module ModelOutput {
       not isValidNoArgumentTokenInIdentifyingAccessPath(token.getName()) and
       result = "Invalid token '" + token + "' is missing its arguments, in access path: " + path
     )
+    or
+    // Check for invalid model kinds
+    result = KindVal::getInvalidModelKind()
   }
 }

ruby/ql/lib/qlpack.yml

@@ -6,6 +6,7 @@ dbscheme: ruby.dbscheme
 upgrades: upgrades
 library: true
 dependencies:
+  codeql/mad: ${workspace}
   codeql/regex: ${workspace}
   codeql/ssa: ${workspace}
   codeql/tutorial: ${workspace}

shared/mad/codeql/mad/ModelValidation.qll

@@ -0,0 +1,186 @@
+/**
+ * Provides classes and predicates related to validating models-as-data rows.
+ */
+
+/** Provides predicates for determining if a model exists for a given `kind`. */
+signature module KindValidationConfigSig {
+  /** Holds if a summary model exists for the given `kind`. */
+  predicate summaryKind(string kind);
+
+  /** Holds if a sink model exists for the given `kind`. */
+  predicate sinkKind(string kind);
+
+  /** Holds if a source model exists for the given `kind`. */
+  predicate sourceKind(string kind);
+
+  /** Holds if a neutral model exists for the given `kind`. */
+  default predicate neutralKind(string kind) { none() }
+}
+
+/** Provides validation for models-as-data summary, sink, source, and neutral kinds. */
+module KindValidation<KindValidationConfigSig Config> {
+  /** A valid models-as-data sink kind. */
+  private class ValidSinkKind extends string {
+    bindingset[this]
+    ValidSinkKind() {
+      this =
+        [
+          // shared
+          "code-injection", "command-injection", "file-content-store", "html-injection",
+          "js-injection", "ldap-injection", "log-injection", "path-injection", "request-forgery",
+          "sql-injection", "url-redirection",
+          // Java-only currently, but may be shared in the future
+          "bean-validation", "fragment-injection", "groovy-injection", "hostname-verification",
+          "information-leak", "intent-redirection", "jexl-injection", "jndi-injection",
+          "mvel-injection", "ognl-injection", "pending-intents", "response-splitting",
+          "template-injection", "xpath-injection", "xslt-injection",
+          // JavaScript-only currently, but may be shared in the future
+          "mongodb.sink", "nosql-injection", "unsafe-deserialization",
+          // Swift-only currently, but may be shared in the future
+          "database-store", "format-string", "hash-iteration-count", "predicate-injection",
+          "preferences-store", "tls-protocol-version", "transmission", "webview-fetch", "xxe"
+        ]
+      or
+      this.matches([
+          // shared
+          "encryption-%", "qltest%", "test-%",
+          // Java-only currently, but may be shared in the future
+          "regex-use%",
+          // JavaScript-only currently, but may be shared in the future
+          "credentials-%",
+          // Swift-only currently, but may be shared in the future
+          "%string-%length", "weak-hash-input-%"
+        ])
+    }
+  }
+
+  /** An outdated models-as-data sink kind. */
+  private class OutdatedSinkKind extends string {
+    OutdatedSinkKind() {
+      this =
+        [
+          "sql", "url-redirect", "xpath", "ssti", "logging", "groovy", "jexl", "mvel", "xslt",
+          "ldap", "pending-intent-sent", "intent-start", "set-hostname-verifier",
+          "header-splitting", "xss", "write-file", "create-file", "read-file", "open-url",
+          "jdbc-url", "command-line-injection", "code", "html", "remote",
+          "uncontrolled-format-string", "js-eval"
+        ]
+    }
+
+    /** Gets a replacement kind for an outdated sink kind. */
+    private string replacementKind() {
+      this = ["sql", "xpath", "groovy", "jexl", "mvel", "xslt", "ldap", "code", "html"] and
+      result = this + "-injection"
+      or
+      this = "js-eval" and result = "code-injection"
+      or
+      this = "url-redirect" and result = "url-redirection"
+      or
+      this = "ssti" and result = "template-injection"
+      or
+      this = "logging" and result = "log-injection"
+      or
+      this = "pending-intent-sent" and result = "pending-intents"
+      or
+      this = "intent-start" and result = "intent-redirection"
+      or
+      this = "set-hostname-verifier" and result = "hostname-verification"
+      or
+      this = "header-splitting" and result = "response-splitting"
+      or
+      this = "xss" and result = "html-injection\" or \"js-injection"
+      or
+      this = ["write-file", "remote"] and result = "file-content-store"
+      or
+      this = ["create-file", "read-file"] and result = "path-injection"
+      or
+      this = ["open-url", "jdbc-url"] and result = "request-forgery"
+      or
+      this = "command-line-injection" and result = "command-injection"
+      or
+      this = "uncontrolled-format-string" and result = "format-string"
+    }
+
+    /** Gets an error message for an outdated sink kind. */
+    string outdatedMessage() {
+      result =
+        "The kind \"" + this + "\" is outdated. Use \"" + this.replacementKind() + "\" instead."
+    }
+  }
+
+  /** A valid models-as-data source kind. */
+  private class ValidSourceKind extends string {
+    bindingset[this]
+    ValidSourceKind() {
+      this =
+        [
+          // shared
+          "local", "remote",
+          // Java
+          "android-external-storage-dir", "contentprovider",
+          // C#
+          "file", "file-write",
+          // JavaScript
+          "database-access-result"
+        ]
+      or
+      this.matches([
+          // shared
+          "qltest%", "test-%",
+          // Swift
+          "%string-%length"
+        ])
+    }
+  }
+
+  /** A valid models-as-data summary kind. */
+  private class ValidSummaryKind extends string {
+    ValidSummaryKind() {
+      this =
+        [
+          // shared
+          "taint", "value",
+          // Dynamic languages
+          "type"
+        ]
+    }
+  }
+
+  /** A valid models-as-data neutral kind. */
+  private class ValidNeutralKind extends string {
+    ValidNeutralKind() {
+      this =
+        [
+          // Java/C# currently
+          "sink", "source", "summary"
+        ]
+    }
+  }
+
+  /** Gets an error message relating to an invalid kind in a model. */
+  string getInvalidModelKind() {
+    exists(string kind | Config::summaryKind(kind) |
+      not kind instanceof ValidSummaryKind and
+      result = "Invalid kind \"" + kind + "\" in summary model."
+    )
+    or
+    exists(string kind, string msg | Config::sinkKind(kind) |
+      not kind instanceof ValidSinkKind and
+      msg = "Invalid kind \"" + kind + "\" in sink model." and
+      // The part of this message that refers to outdated sink kinds can be deleted after June 1st, 2024.
+      if kind instanceof OutdatedSinkKind
+      then result = msg + " " + kind.(OutdatedSinkKind).outdatedMessage()
+      else result = msg
+    )
+    or
+    exists(string kind | Config::sourceKind(kind) |
+      not kind instanceof ValidSourceKind and
+      result = "Invalid kind \"" + kind + "\" in source model."
+    )
+    or
+    exists(string kind | Config::neutralKind(kind) |
+      not kind instanceof ValidNeutralKind and
+      result = "Invalid kind \"" + kind + "\" in neutral model."
+    )
+  }
+}

shared/mad/qlpack.yml

@@ -0,0 +1,6 @@
+name: codeql/mad
+version: 0.0.1-dev
+groups: shared
+library: true
+dependencies:
+warnOnImplicitThis: true

swift/ql/lib/codeql/swift/dataflow/ExternalFlow.qll

@@ -74,6 +74,7 @@ private import internal.FlowSummaryImpl::Public
 private import internal.FlowSummaryImpl::Private::External
 private import internal.FlowSummaryImplSpecific
 private import FlowSummary as FlowSummary
+private import codeql.mad.ModelValidation as SharedModelVal
 
 /**
  * A unit class for adding additional source model rows.
@@ -263,14 +264,16 @@ module CsvValidation {
     )
   }
 
-  private string getInvalidModelKind() {
-    exists(string row, string kind | summaryModel(row) |
-      kind = row.splitAt(";", 8) and
-      not kind = ["taint", "value"] and
-      result = "Invalid kind \"" + kind + "\" in summary model."
-    )
+  private module KindValConfig implements SharedModelVal::KindValidationConfigSig {
+    predicate summaryKind(string kind) { summaryModel(_, _, _, _, _, _, _, _, kind, _) }
+
+    predicate sinkKind(string kind) { sinkModel(_, _, _, _, _, _, _, kind, _) }
+
+    predicate sourceKind(string kind) { sourceModel(_, _, _, _, _, _, _, kind, _) }
   }
 
+  private module KindVal = SharedModelVal::KindValidation<KindValConfig>;
+
   private string getInvalidModelSubtype() {
     exists(string pred, string row |
       sourceModel(row) and pred = "source"
@@ -335,7 +338,7 @@ module CsvValidation {
     msg =
       [
         getInvalidModelSignature(), getInvalidModelInput(), getInvalidModelOutput(),
-        getInvalidModelSubtype(), getInvalidModelColumnCount(), getInvalidModelKind()
+        getInvalidModelSubtype(), getInvalidModelColumnCount(), KindVal::getInvalidModelKind()
       ]
   }
 }

swift/ql/lib/qlpack.yml

@@ -6,6 +6,7 @@ dbscheme: swift.dbscheme
 upgrades: upgrades
 library: true
 dependencies:
+  codeql/mad: ${workspace}
   codeql/ssa: ${workspace}
   codeql/tutorial: ${workspace}
   codeql/util: ${workspace}