Populate path filters env variables using the CLI

This commit is contained in:
Henry Mercer 2024-01-04 15:17:20 +00:00
parent b30262ccb4
commit f934b28e51
12 changed files with 26 additions and 407 deletions

View file

@ -1,104 +0,0 @@
import * as path from "path";
import test from "ava";
import * as analysisPaths from "./analysis-paths";
import { setupTests } from "./testing-utils";
import * as util from "./util";
setupTests(test);
test("emptyPaths", async (t) => {
return await util.withTmpDir(async (tmpDir) => {
const config = {
languages: [],
queries: {},
pathsIgnore: [],
paths: [],
originalUserInput: {},
tempDir: tmpDir,
codeQLCmd: "",
gitHubVersion: { type: util.GitHubVariant.DOTCOM } as util.GitHubVersion,
dbLocation: path.resolve(tmpDir, "codeql_databases"),
packs: {},
debugMode: false,
debugArtifactName: util.DEFAULT_DEBUG_ARTIFACT_NAME,
debugDatabaseName: util.DEFAULT_DEBUG_DATABASE_NAME,
augmentationProperties: {
injectedMlQueries: false,
packsInputCombines: false,
queriesInputCombines: false,
},
trapCaches: {},
trapCacheDownloadTime: 0,
};
analysisPaths.includeAndExcludeAnalysisPaths(config);
t.is(process.env["LGTM_INDEX_INCLUDE"], undefined);
t.is(process.env["LGTM_INDEX_EXCLUDE"], undefined);
t.is(process.env["LGTM_INDEX_FILTERS"], undefined);
});
});
test("nonEmptyPaths", async (t) => {
return await util.withTmpDir(async (tmpDir) => {
const config = {
languages: [],
queries: {},
paths: ["path1", "path2", "**/path3"],
pathsIgnore: ["path4", "path5", "path6/**"],
originalUserInput: {},
tempDir: tmpDir,
codeQLCmd: "",
gitHubVersion: { type: util.GitHubVariant.DOTCOM } as util.GitHubVersion,
dbLocation: path.resolve(tmpDir, "codeql_databases"),
packs: {},
debugMode: false,
debugArtifactName: util.DEFAULT_DEBUG_ARTIFACT_NAME,
debugDatabaseName: util.DEFAULT_DEBUG_DATABASE_NAME,
augmentationProperties: {
injectedMlQueries: false,
packsInputCombines: false,
queriesInputCombines: false,
},
trapCaches: {},
trapCacheDownloadTime: 0,
};
analysisPaths.includeAndExcludeAnalysisPaths(config);
t.is(process.env["LGTM_INDEX_INCLUDE"], "path1\npath2");
t.is(process.env["LGTM_INDEX_EXCLUDE"], "path4\npath5");
t.is(
process.env["LGTM_INDEX_FILTERS"],
"include:path1\ninclude:path2\ninclude:**/path3\nexclude:path4\nexclude:path5\nexclude:path6/**",
);
});
});
test("exclude temp dir", async (t) => {
const tempDir = path.join(process.cwd(), "codeql-runner-temp");
const config = {
languages: [],
queries: {},
pathsIgnore: [],
paths: [],
originalUserInput: {},
tempDir,
codeQLCmd: "",
gitHubVersion: { type: util.GitHubVariant.DOTCOM } as util.GitHubVersion,
dbLocation: path.resolve(tempDir, "codeql_databases"),
packs: {},
debugMode: false,
debugArtifactName: util.DEFAULT_DEBUG_ARTIFACT_NAME,
debugDatabaseName: util.DEFAULT_DEBUG_DATABASE_NAME,
augmentationProperties: {
injectedMlQueries: false,
packsInputCombines: false,
queriesInputCombines: false,
},
trapCaches: {},
trapCacheDownloadTime: 0,
};
analysisPaths.includeAndExcludeAnalysisPaths(config);
t.is(process.env["LGTM_INDEX_INCLUDE"], undefined);
t.is(process.env["LGTM_INDEX_EXCLUDE"], "codeql-runner-temp");
t.is(process.env["LGTM_INDEX_FILTERS"], undefined);
});

View file

@ -1,77 +0,0 @@
import * as path from "path";
import * as configUtils from "./config-utils";
import { Logger } from "./logging";
function isInterpretedLanguage(language): boolean {
return (
language === "javascript" || language === "python" || language === "ruby"
);
}
// Matches a string containing only characters that are legal to include in paths on windows.
export const legalWindowsPathCharactersRegex = /^[^<>:"|?]*$/;
// Builds an environment variable suitable for LGTM_INDEX_INCLUDE or LGTM_INDEX_EXCLUDE
function buildIncludeExcludeEnvVar(paths: string[]): string {
// Ignore anything containing a *
paths = paths.filter((p) => p.indexOf("*") === -1);
// Some characters are illegal in path names in windows
if (process.platform === "win32") {
paths = paths.filter((p) => p.match(legalWindowsPathCharactersRegex));
}
return paths.join("\n");
}
export function printPathFiltersWarning(
config: configUtils.Config,
logger: Logger,
) {
// Index include/exclude/filters only work in javascript/python/ruby.
// If any other languages are detected/configured then show a warning.
if (
(config.paths.length !== 0 || config.pathsIgnore.length !== 0) &&
!config.languages.every(isInterpretedLanguage)
) {
logger.warning(
'The "paths"/"paths-ignore" fields of the config only have effect for JavaScript, Python, and Ruby',
);
}
}
export function includeAndExcludeAnalysisPaths(config: configUtils.Config) {
// The 'LGTM_INDEX_INCLUDE' and 'LGTM_INDEX_EXCLUDE' environment variables
// control which files/directories are traversed when scanning.
// This allows including files that otherwise would not be scanned, or
// excluding and not traversing entire file subtrees.
// It does not understand globs or double-globs because that would require it to
// traverse the entire file tree to determine which files are matched.
// Any paths containing "*" are not included in these.
if (config.paths.length !== 0) {
process.env["LGTM_INDEX_INCLUDE"] = buildIncludeExcludeEnvVar(config.paths);
}
// If the temporary or tools directory is in the working directory ignore that too.
const tempRelativeToWorking = path.relative(process.cwd(), config.tempDir);
let pathsIgnore = config.pathsIgnore;
if (
!tempRelativeToWorking.startsWith("..") &&
!path.isAbsolute(tempRelativeToWorking)
) {
pathsIgnore = pathsIgnore.concat(tempRelativeToWorking);
}
if (pathsIgnore.length !== 0) {
process.env["LGTM_INDEX_EXCLUDE"] = buildIncludeExcludeEnvVar(pathsIgnore);
}
// The 'LGTM_INDEX_FILTERS' environment variable controls which files are
// extracted or ignored. It does not control which directories are traversed.
// This does understand the glob and double-glob syntax.
const filters: string[] = [];
filters.push(...config.paths.map((p) => `include:${p}`));
filters.push(...config.pathsIgnore.map((p) => `exclude:${p}`));
if (filters.length !== 0) {
process.env["LGTM_INDEX_FILTERS"] = filters.join("\n");
}
}

View file

@ -6,7 +6,6 @@ import * as toolrunner from "@actions/exec/lib/toolrunner";
import del from "del";
import * as yaml from "js-yaml";
import * as analysisPaths from "./analysis-paths";
import {
CODEQL_VERSION_ANALYSIS_SUMMARY_V2,
CodeQL,
@ -147,10 +146,6 @@ export async function createdDBForScannedLanguages(
logger: Logger,
features: FeatureEnablement,
) {
// Insert the LGTM_INDEX_X env vars at this point so they are set when
// we extract any scanned languages.
analysisPaths.includeAndExcludeAnalysisPaths(config);
for (const language of config.languages) {
if (
isScannedLanguage(language) &&

View file

@ -4,12 +4,11 @@ import * as path from "path";
import * as toolrunner from "@actions/exec/lib/toolrunner";
import * as safeWhich from "@chrisgavin/safe-which";
import * as analysisPaths from "./analysis-paths";
import { GitHubApiCombinedDetails, GitHubApiDetails } from "./api-client";
import { CodeQL, setupCodeQL } from "./codeql";
import * as configUtils from "./config-utils";
import { CodeQLDefaultVersionInfo } from "./feature-flags";
import { Language } from "./languages";
import { Language, isScannedLanguage } from "./languages";
import { Logger } from "./logging";
import { RepositoryNwo } from "./repository";
import { ToolsSource } from "./setup-codeql";
@ -84,7 +83,7 @@ export async function initConfig(
apiDetails,
logger,
);
analysisPaths.printPathFiltersWarning(config, logger);
printPathFiltersWarning(config, logger);
logger.endGroup();
return config;
}
@ -128,6 +127,19 @@ export async function runInit(
return await getCombinedTracerConfig(config);
}
function printPathFiltersWarning(config: configUtils.Config, logger: Logger) {
// Index include/exclude/filters only work in javascript/python/ruby.
// If any other languages are detected/configured then show a warning.
if (
(config.paths.length !== 0 || config.pathsIgnore.length !== 0) &&
!config.languages.every(isScannedLanguage)
) {
logger.warning(
'The "paths"/"paths-ignore" fields of the config only have effect for JavaScript, Python, and Ruby',
);
}
}
/**
* Possibly convert this error into a UserError in order to avoid
* counting this error towards our internal error budget.