Azure python sanitizer upstream2

python/ql/lib/semmle/python/Concepts.qll

@@ -5,7 +5,7 @@
  */
 
 private import python
 private import semmle.python.dataflow.new.DataFlow
 private import semmle.python.dataflow.new.internal.DataFlowImplSpecific
 private import semmle.python.dataflow.new.RemoteFlowSources
 private import semmle.python.dataflow.new.TaintTracking
@@ -15,6 +15,8 @@
 private import semmle.python.dataflow.new.SensitiveDataSources
 private import codeql.threatmodels.ThreatModels
 private import codeql.concepts.ConceptsShared
+private import semmle.python.ApiGraphs
+private import semmle.python.frameworks.data.ModelsAsData
 
 private module ConceptsShared = ConceptsMake<Location, PythonDataFlow>;
 
@@ -1656,8 +1658,35 @@
   }
 
   import ConceptsShared::Http::Client as Client
+
   // TODO: investigate whether we should treat responses to client requests as
   // remote-flow-sources in general.
+  /**
+   * An HTTP request modeled from `request-forgery` sinks, modeled using MaD.
+   */
+  class HttpClientRequestFromModel extends Http::Client::Request::Range instanceof API::CallNode {
+    DataFlow::Node urlArg;
+
+    HttpClientRequestFromModel() {
+      (
+        this.getArg(_) = urlArg
+        or
+        this.getArgByName(_) = urlArg
+      ) and
+      ModelOutput::sinkNode(urlArg, "request-forgery")
+    }
+
+    override DataFlow::Node getAUrlPart() { result = urlArg }
+
+    override string getFramework() { result = "MaD" }
+
+    override predicate disablesCertificateValidation(
+      DataFlow::Node disablingNode, DataFlow::Node argumentOrigin
+    ) {
+      // NOTE: if you need to define this, you have to special case it for every possible API in MaD
+      none()
+    }
+  }
 }
 
 /**

python/ql/lib/semmle/python/frameworks/Azure.Keyvault.model.yml

@@ -0,0 +1,9 @@
+extensions:
+  - addsTo:
+      pack: codeql/python-all
+      extensible: sinkModel
+    data:
+      - ['azure.keyvault.certificates.CertificateClient!', 'Call.Argument[0,vault_url:]', 'request-forgery']
+      - ['azure.keyvault.certificates.DeletedCertificate!', 'Call.Argument[recovery_id:]', 'request-forgery']
+      - ['azure.keyvault.keys.KeyClient!', 'Call.Argument[0,vault_url:]', 'request-forgery']
+      - ['azure.keyvault.secrets.SecretClient!', 'Call.Argument[0,vault_url:]', 'request-forgery']

python/ql/lib/semmle/python/frameworks/Azure.Storage.model.yml

@@ -0,0 +1,34 @@
+extensions:
+  - addsTo:
+      pack: codeql/python-all
+      extensible: sinkModel
+    data:
+      - ['azure.storage.blob.BlobClient!', 'Call.Argument[0,account_url:]', 'request-forgery']
+      - ['azure.storage.blob.BlobClient', 'Member[append_block_from_url].Argument[0,copy_source_url:]', 'request-forgery']
+      - ['azure.storage.blob.BlobClient', 'Member[get_page_range_diff_for_managed_disk].Argument[0,previous_snapshot_url:]', 'request-forgery']
+      - ['azure.storage.blob.BlobClient', 'Member[stage_block_from_url].Argument[1,source_url:]', 'request-forgery']
+      - ['azure.storage.blob.BlobClient', 'Member[start_copy_from_url].Argument[0,source_url:]', 'request-forgery']
+      - ['azure.storage.blob.BlobClient', 'Member[upload_blob_from_url].Argument[0,source_url:]', 'request-forgery']
+      - ['azure.storage.blob.BlobClient', 'Member[upload_pages_from_url].Argument[0,source_url:]', 'request-forgery']
+      - ['azure.storage.blob.BlobClient!', 'Member[from_blob_url].Argument[0,blob_url:]', 'request-forgery']
+      - ['azure.storage.blob.BlobServiceClient!', 'Call.Argument[0,account_url:]', 'request-forgery']
+      - ['azure.storage.blob.ContainerClient!', 'Call.Argument[0,account_url:]', 'request-forgery']
+      - ['azure.storage.blob.ContainerClient!', 'Member[from_container_url].Argument[0,container_url:]', 'request-forgery']
+      - ['azure', 'Member[storage].Member[blob].Member[download_blob_from_url].Argument[0,blob_url:]', 'request-forgery']
+      - ['azure', 'Member[storage].Member[blob].Member[upload_blob_to_url].Argument[0,blob_url:]', 'request-forgery']
+      - ['azure.storage.filedatalake.DataLakeDirectoryClient!', 'Call.Argument[0,account_url:]', 'request-forgery']
+      - ['azure.storage.filedatalake.DataLakeFileClient!', 'Call.Argument[0,account_url:]', 'request-forgery']
+      - ['azure.storage.filedatalake.DataLakeServiceClient!', 'Call.Argument[0,account_url:]', 'request-forgery']
+      - ['azure.storage.filedatalake.FileSystemClient!', 'Call.Argument[0,account_url:]', 'request-forgery']
+      - ['azure.storage.fileshare.ShareClient!', 'Call.Argument[0,account_url:]', 'request-forgery']
+      - ['azure.storage.fileshare.ShareClient!', 'Member[from_share_url].Argument[0,share_url:]', 'request-forgery']
+      - ['azure.storage.fileshare.ShareDirectoryClient!', 'Call.Argument[0,account_url:]', 'request-forgery']
+      - ['azure.storage.fileshare.ShareDirectoryClient!', 'Member[from_directory_url].Argument[0,directory_url:]', 'request-forgery']
+      - ['azure.storage.fileshare.ShareFileClient!', 'Call.Argument[0,account_url:]', 'request-forgery']
+      - ['azure.storage.fileshare.ShareFileClient!', 'Member[from_file_url].Argument[0,file_url:]', 'request-forgery']
+      - ['azure.storage.fileshare.ShareFileClient', 'Member[start_copy_from_url].Argument[0,source_url:]', 'request-forgery']
+      - ['azure.storage.fileshare.ShareFileClient', 'Member[upload_range_from_url].Argument[0,source_url:]', 'request-forgery']
+      - ['azure.storage.fileshare.ShareServiceClient!', 'Call.Argument[0,account_url:]', 'request-forgery']
+      - ['azure.storage.queue.QueueClient!', 'Call.Argument[0,account_url:]', 'request-forgery']
+      - ['azure.storage.queue.QueueClient', 'Member[from_queue_url].Argument[0,queue_url:]', 'request-forgery']
+      - ['azure.storage.queue.QueueServiceClient!', 'Call.Argument[0,account_url:]', 'request-forgery']

python/ql/lib/semmle/python/security/dataflow/ServerSideRequestForgeryCustomizations.qll

@@ -176,4 +176,48 @@ module ServerSideRequestForgery {
       strNode = [call.getArg(0), call.getArgByName("string")]
     )
   }
+
+  /** A validation that a string does not contain certain characters, considered as a sanitizer. */
+  private class UriValidator extends FullUrlControlSanitizer {
+    UriValidator() { this = DataFlow::BarrierGuard<uri_validator/3>::getABarrierNode() }
+  }
+
+  import semmle.python.dataflow.new.internal.DataFlowPublic
+
+  private predicate uri_validator(DataFlow::GuardNode g, ControlFlowNode node, boolean branch) {
+    exists(DataFlow::CallCfgNode call, string funcs |
+      funcs in ["in_domain", "in_azure_keyvault_domain", "in_azure_storage_domain"]
+    |
+      call = API::moduleImport("AntiSSRF").getMember("URIValidator").getMember(funcs).getACall() and
+      call.getArg(0).asCfgNode() = node and
+      (
+        // validator used in a comparison
+        exists(CompareNode cn, Cmpop op, Node n | cn = g and n.getALocalSource() = call |
+          (
+            // validator == true or validator == false or validator is True or validator is False
+            (op instanceof Eq or op instanceof Is) and
+            exists(ControlFlowNode l, boolean bool |
+              l.getNode().(BooleanLiteral).booleanValue() = bool and
+              bool in [true, false] and
+              branch = bool and
+              cn.operands(n.asCfgNode(), op, l)
+            )
+            or
+            // validator != false or validator != true or validator is not True or validator is not False
+            (op instanceof NotEq or op instanceof IsNot) and
+            exists(ControlFlowNode l, boolean bool |
+              l.getNode().(BooleanLiteral).booleanValue() = bool and
+              bool in [true, false] and
+              branch = bool.booleanNot() and
+              cn.operands(n.asCfgNode(), op, l)
+            )
+          )
+        )
+        or
+        // validator call directly (e.g., if URIValidator.in_domain(...) )
+        g = call.asCfgNode() and
+        branch = true
+      )
+    )
+  }
 }