Merge pull request #3028 from github/koesie10/bqrs-decode-languages

Use MaD definition when decoding BQRS
This commit is contained in:
Koen Vlaswinkel
2023-10-27 11:58:48 +02:00
committed by GitHub
11 changed files with 189 additions and 148 deletions

View File

@@ -4,13 +4,18 @@ import { ModeledMethodType } from "./modeled-method";
import { parseLibraryFilename } from "./library";
import { Mode } from "./shared/mode";
import { ApplicationModeTuple, FrameworkModeTuple } from "./queries/query";
import { QueryLanguage } from "../common/query-language";
import { getModelsAsDataLanguage } from "./languages";
export function decodeBqrsToMethods(
chunk: DecodedBqrsChunk,
mode: Mode,
language: QueryLanguage,
): Method[] {
const methodsByApiName = new Map<string, Method>();
const definition = getModelsAsDataLanguage(language);
chunk?.tuples.forEach((tuple) => {
let usage: Call;
let packageName: string;
@@ -51,7 +56,12 @@ export function decodeBqrsToMethods(
classification = CallClassification.Unknown;
}
const signature = `${packageName}.${typeName}#${methodName}${methodParameters}`;
const signature = definition.createMethodSignature({
packageName,
typeName,
methodName,
methodParameters,
});
// For Java, we'll always get back a .jar file, and the library version may be bad because not all library authors
// properly specify the version. Therefore, we'll always try to parse the name and version from the library filename

View File

@@ -156,7 +156,7 @@ async function runSingleFlowQuery(
// Interpret the results
const modelsAsDataLanguage = getModelsAsDataLanguage(language);
const definition = modelsAsDataLanguage[type];
const definition = modelsAsDataLanguage.predicates[type];
const bqrsPath = completedQuery.outputDir.bqrsPath;

View File

@@ -1,3 +1,4 @@
import { MethodDefinition } from "../method";
import { ModeledMethod, ModeledMethodType } from "../modeled-method";
import { DataTuple } from "../model-extension-file";
@@ -6,14 +7,19 @@ type ReadModeledMethod = (row: DataTuple[]) => ModeledMethod;
export type ModelsAsDataLanguageModelType = Exclude<ModeledMethodType, "none">;
export type ModelsAsDataLanguageModel = {
export type ModelsAsDataLanguagePredicate = {
extensiblePredicate: string;
supportedKinds: string[];
generateMethodDefinition: GenerateMethodDefinition;
readModeledMethod: ReadModeledMethod;
};
export type ModelsAsDataLanguage = Record<
export type ModelsAsDataLanguagePredicates = Record<
ModelsAsDataLanguageModelType,
ModelsAsDataLanguageModel
ModelsAsDataLanguagePredicate
>;
export type ModelsAsDataLanguage = {
createMethodSignature: (method: MethodDefinition) => string;
predicates: ModelsAsDataLanguagePredicates;
};

View File

@@ -8,125 +8,133 @@ function readRowToMethod(row: DataTuple[]): string {
}
export const staticLanguage: ModelsAsDataLanguage = {
source: {
extensiblePredicate: sharedExtensiblePredicates.source,
supportedKinds: sharedKinds.source,
// extensible predicate sourceModel(
// string package, string type, boolean subtypes, string name, string signature, string ext,
// string output, string kind, string provenance
// );
generateMethodDefinition: (method) => [
method.packageName,
method.typeName,
true,
method.methodName,
method.methodParameters,
"",
method.output,
method.kind,
method.provenance,
],
readModeledMethod: (row) => ({
type: "source" as ModeledMethodType,
input: "",
output: row[6] as string,
kind: row[7] as string,
provenance: row[8] as Provenance,
signature: readRowToMethod(row),
packageName: row[0] as string,
typeName: row[1] as string,
methodName: row[3] as string,
methodParameters: row[4] as string,
}),
},
sink: {
extensiblePredicate: sharedExtensiblePredicates.sink,
supportedKinds: sharedKinds.sink,
// extensible predicate sinkModel(
// string package, string type, boolean subtypes, string name, string signature, string ext,
// string input, string kind, string provenance
// );
generateMethodDefinition: (method) => [
method.packageName,
method.typeName,
true,
method.methodName,
method.methodParameters,
"",
method.input,
method.kind,
method.provenance,
],
readModeledMethod: (row) => ({
type: "sink",
input: row[6] as string,
output: "",
kind: row[7] as string,
provenance: row[8] as Provenance,
signature: readRowToMethod(row),
packageName: row[0] as string,
typeName: row[1] as string,
methodName: row[3] as string,
methodParameters: row[4] as string,
}),
},
summary: {
extensiblePredicate: sharedExtensiblePredicates.summary,
supportedKinds: sharedKinds.summary,
// extensible predicate summaryModel(
// string package, string type, boolean subtypes, string name, string signature, string ext,
// string input, string output, string kind, string provenance
// );
generateMethodDefinition: (method) => [
method.packageName,
method.typeName,
true,
method.methodName,
method.methodParameters,
"",
method.input,
method.output,
method.kind,
method.provenance,
],
readModeledMethod: (row) => ({
type: "summary",
input: row[6] as string,
output: row[7] as string,
kind: row[8] as string,
provenance: row[9] as Provenance,
signature: readRowToMethod(row),
packageName: row[0] as string,
typeName: row[1] as string,
methodName: row[3] as string,
methodParameters: row[4] as string,
}),
},
neutral: {
extensiblePredicate: sharedExtensiblePredicates.neutral,
supportedKinds: sharedKinds.neutral,
// extensible predicate neutralModel(
// string package, string type, string name, string signature, string kind, string provenance
// );
generateMethodDefinition: (method) => [
method.packageName,
method.typeName,
method.methodName,
method.methodParameters,
method.kind,
method.provenance,
],
readModeledMethod: (row) => ({
type: "neutral",
input: "",
output: "",
kind: row[4] as string,
provenance: row[5] as Provenance,
signature: `${row[0]}.${row[1]}#${row[2]}${row[3]}`,
packageName: row[0] as string,
typeName: row[1] as string,
methodName: row[2] as string,
methodParameters: row[3] as string,
}),
createMethodSignature: ({
packageName,
typeName,
methodName,
methodParameters,
}) => `${packageName}.${typeName}#${methodName}${methodParameters}`,
predicates: {
source: {
extensiblePredicate: sharedExtensiblePredicates.source,
supportedKinds: sharedKinds.source,
// extensible predicate sourceModel(
// string package, string type, boolean subtypes, string name, string signature, string ext,
// string output, string kind, string provenance
// );
generateMethodDefinition: (method) => [
method.packageName,
method.typeName,
true,
method.methodName,
method.methodParameters,
"",
method.output,
method.kind,
method.provenance,
],
readModeledMethod: (row) => ({
type: "source" as ModeledMethodType,
input: "",
output: row[6] as string,
kind: row[7] as string,
provenance: row[8] as Provenance,
signature: readRowToMethod(row),
packageName: row[0] as string,
typeName: row[1] as string,
methodName: row[3] as string,
methodParameters: row[4] as string,
}),
},
sink: {
extensiblePredicate: sharedExtensiblePredicates.sink,
supportedKinds: sharedKinds.sink,
// extensible predicate sinkModel(
// string package, string type, boolean subtypes, string name, string signature, string ext,
// string input, string kind, string provenance
// );
generateMethodDefinition: (method) => [
method.packageName,
method.typeName,
true,
method.methodName,
method.methodParameters,
"",
method.input,
method.kind,
method.provenance,
],
readModeledMethod: (row) => ({
type: "sink",
input: row[6] as string,
output: "",
kind: row[7] as string,
provenance: row[8] as Provenance,
signature: readRowToMethod(row),
packageName: row[0] as string,
typeName: row[1] as string,
methodName: row[3] as string,
methodParameters: row[4] as string,
}),
},
summary: {
extensiblePredicate: sharedExtensiblePredicates.summary,
supportedKinds: sharedKinds.summary,
// extensible predicate summaryModel(
// string package, string type, boolean subtypes, string name, string signature, string ext,
// string input, string output, string kind, string provenance
// );
generateMethodDefinition: (method) => [
method.packageName,
method.typeName,
true,
method.methodName,
method.methodParameters,
"",
method.input,
method.output,
method.kind,
method.provenance,
],
readModeledMethod: (row) => ({
type: "summary",
input: row[6] as string,
output: row[7] as string,
kind: row[8] as string,
provenance: row[9] as Provenance,
signature: readRowToMethod(row),
packageName: row[0] as string,
typeName: row[1] as string,
methodName: row[3] as string,
methodParameters: row[4] as string,
}),
},
neutral: {
extensiblePredicate: sharedExtensiblePredicates.neutral,
supportedKinds: sharedKinds.neutral,
// extensible predicate neutralModel(
// string package, string type, string name, string signature, string kind, string provenance
// );
generateMethodDefinition: (method) => [
method.packageName,
method.typeName,
method.methodName,
method.methodParameters,
method.kind,
method.provenance,
],
readModeledMethod: (row) => ({
type: "neutral",
input: "",
output: "",
kind: row[4] as string,
provenance: row[5] as Provenance,
signature: `${row[0]}.${row[1]}#${row[2]}${row[3]}`,
packageName: row[0] as string,
typeName: row[1] as string,
methodName: row[2] as string,
methodParameters: row[3] as string,
}),
},
},
};

View File

@@ -17,19 +17,7 @@ export type Usage = Call & {
readonly classification: CallClassification;
};
export interface MethodSignature {
/**
* Contains the version of the library if it can be determined by CodeQL, e.g. `4.2.2.2`
*/
readonly libraryVersion?: string;
/**
* A unique signature that can be used to identify this external API usage.
*
* The signature contains the package name, type name, method name, and method parameters
* in the form "packageName.typeName#methodName(methodParameters)".
* e.g. `org.sql2o.Connection#createQuery(String)`
*/
readonly signature: string;
export interface MethodDefinition {
/**
* The package name in Java, or the namespace in C#, e.g. `org.sql2o` or `System.Net.Http.Headers`.
*
@@ -44,6 +32,21 @@ export interface MethodSignature {
readonly methodParameters: string;
}
export interface MethodSignature extends MethodDefinition {
/**
* Contains the version of the library if it can be determined by CodeQL, e.g. `4.2.2.2`
*/
readonly libraryVersion?: string;
/**
* A unique signature that can be used to identify this external API usage.
*
* The signature contains the package name, type name, method name, and method parameters
* in the form "packageName.typeName#methodName(methodParameters)".
* e.g. `org.sql2o.Connection#createQuery(String)`
*/
readonly signature: string;
}
export interface Method extends MethodSignature {
/**
* Contains the name of the library containing the method declaration, e.g. `sql2o-1.6.0.jar` or `System.Runtime.dll`

View File

@@ -25,6 +25,7 @@ type RunQueryOptions = {
cliServer: CodeQLCliServer;
queryRunner: QueryRunner;
databaseItem: DatabaseItem;
language: QueryLanguage;
queryStorageDir: string;
queryDir: string;
@@ -70,6 +71,7 @@ export async function runModelEditorQueries(
cliServer,
queryRunner,
databaseItem,
language,
queryStorageDir,
queryDir,
progress,
@@ -157,7 +159,7 @@ export async function runModelEditorQueries(
maxStep: externalApiQueriesProgressMaxStep,
});
return decodeBqrsToMethods(bqrsChunk, mode);
return decodeBqrsToMethods(bqrsChunk, mode, language);
}
type GetResultsOptions = {

View File

@@ -430,6 +430,7 @@ export class ModelEditorView extends AbstractWebview<
cliServer: this.cliServer,
queryRunner: this.queryRunner,
databaseItem: this.databaseItem,
language: this.language,
queryStorageDir: this.queryStorageDir,
queryDir: this.queryDir,
progress: (update) =>

View File

@@ -4,7 +4,7 @@ import { Method } from "./method";
import { ModeledMethod, ModeledMethodType } from "./modeled-method";
import {
getModelsAsDataLanguage,
ModelsAsDataLanguageModel,
ModelsAsDataLanguagePredicate,
} from "./languages";
import * as modelExtensionFileSchema from "./model-extension-file.schema.json";
@@ -18,7 +18,7 @@ const modelExtensionFileSchemaValidate = ajv.compile(modelExtensionFileSchema);
function createDataProperty(
methods: readonly ModeledMethod[],
definition: ModelsAsDataLanguageModel,
definition: ModelsAsDataLanguagePredicate,
) {
if (methods.length === 0) {
return " []";
@@ -56,7 +56,7 @@ export function createDataExtensionYaml(
}
}
const extensions = Object.entries(modelsAsDataLanguage).map(
const extensions = Object.entries(modelsAsDataLanguage.predicates).map(
([type, definition]) => ` - addsTo:
pack: codeql/${language}-all
extensible: ${definition.extensiblePredicate}
@@ -260,7 +260,7 @@ export function loadDataExtensionYaml(
const extensible = addsTo.extensible;
const data = extension.data;
const definition = Object.values(modelsAsDataLanguage).find(
const definition = Object.values(modelsAsDataLanguage.predicates).find(
(definition) => definition.extensiblePredicate === extensible,
);
if (!definition) {

View File

@@ -26,7 +26,7 @@ export const ModelKindDropdown = ({
const modelsAsDataLanguage = getModelsAsDataLanguage(language);
return modeledMethod?.type && modeledMethod.type !== "none"
? modelsAsDataLanguage[modeledMethod.type]
? modelsAsDataLanguage.predicates[modeledMethod.type]
: undefined;
}, [language, modeledMethod?.type]);

View File

@@ -2,6 +2,7 @@ import { decodeBqrsToMethods } from "../../../src/model-editor/bqrs";
import { DecodedBqrsChunk } from "../../../src/common/bqrs-cli-types";
import { CallClassification } from "../../../src/model-editor/method";
import { Mode } from "../../../src/model-editor/shared/mode";
import { QueryLanguage } from "../../../src/common/query-language";
describe("decodeBqrsToMethods", () => {
describe("Java queries", () => {
@@ -237,7 +238,9 @@ describe("decodeBqrsToMethods", () => {
// Even though there are a number of methods with the same number of usages, the order returned should be stable:
// - Iterating over a map (as done by .values()) is guaranteed to be in insertion order
// - Sorting the array of methods is guaranteed to be a stable sort
expect(decodeBqrsToMethods(chunk, Mode.Application)).toEqual([
expect(
decodeBqrsToMethods(chunk, Mode.Application, QueryLanguage.Java),
).toEqual([
{
library: "rt",
libraryVersion: undefined,
@@ -499,7 +502,9 @@ describe("decodeBqrsToMethods", () => {
};
it("extracts methods", () => {
expect(decodeBqrsToMethods(chunk, Mode.Framework)).toEqual([
expect(
decodeBqrsToMethods(chunk, Mode.Framework, QueryLanguage.Java),
).toEqual([
{
library: "",
libraryVersion: undefined,
@@ -594,7 +599,9 @@ describe("decodeBqrsToMethods", () => {
};
it("extracts methods", () => {
expect(decodeBqrsToMethods(chunk, Mode.Application)).toEqual([
expect(
decodeBqrsToMethods(chunk, Mode.Application, QueryLanguage.Java),
).toEqual([
{
library: "mscorlib",
libraryVersion: "4.0.0.0",
@@ -660,7 +667,9 @@ describe("decodeBqrsToMethods", () => {
};
it("extracts methods", () => {
expect(decodeBqrsToMethods(chunk, Mode.Framework)).toEqual([
expect(
decodeBqrsToMethods(chunk, Mode.Framework, QueryLanguage.Java),
).toEqual([
{
library: "Times",
libraryVersion: undefined,

View File

@@ -74,6 +74,7 @@ describe("runModelEditorQueries", () => {
},
language,
}),
language,
queryStorageDir: "/tmp/queries",
queryDir,
progress: jest.fn(),
@@ -139,6 +140,7 @@ describe("runModelEditorQueries", () => {
},
language,
}),
language,
queryStorageDir: "/tmp/queries",
queryDir,
progress: jest.fn(),