lion/packages/providence-analytics/src/program/analyzers/helpers/Analyzer.js
2020-10-07 15:23:26 +02:00

331 lines
11 KiB
JavaScript

/* eslint-disable no-param-reassign */
const fs = require('fs');
const semver = require('semver');
const pathLib = require('path');
const { LogService } = require('../../services/LogService.js');
const { QueryService } = require('../../services/QueryService.js');
const { ReportService } = require('../../services/ReportService.js');
const { InputDataService } = require('../../services/InputDataService.js');
const { aForEach } = require('../../utils/async-array-utils.js');
const { toPosixPath } = require('../../utils/to-posix-path.js');
const { getFilePathRelativeFromRoot } = require('../../utils/get-file-path-relative-from-root.js');
/**
* @desc analyzes one entry: the callback can traverse a given ast for each entry
* @param {AstDataProject[]} astDataProjects
* @param {function} astAnalysis
*/
async function analyzePerAstEntry(projectData, astAnalysis) {
const entries = [];
await aForEach(projectData.entries, async ({ file, ast, context: astContext }) => {
const relativePath = getFilePathRelativeFromRoot(file, projectData.project.path);
const context = { code: astContext.code, relativePath, projectData };
LogService.debug(`${pathLib.resolve(projectData.project.path, file)}`);
const { result, meta } = await astAnalysis(ast, context);
entries.push({ file: relativePath, meta, result });
});
const filteredEntries = entries.filter(({ result }) => Boolean(result.length));
return filteredEntries;
}
/**
* Transforms QueryResult entries to posix path notations on Windows
* @param {array|object} data
*/
function posixify(data) {
if (!data) {
return;
}
if (Array.isArray(data)) {
data.forEach(posixify);
} else if (typeof data === 'object') {
Object.entries(data).forEach(([k, v]) => {
if (Array.isArray(v) || typeof v === 'object') {
posixify(v);
}
// TODO: detect whether filePath instead of restricting by key name?
else if (typeof v === 'string' && k === 'file') {
data[k] = toPosixPath(v);
}
});
}
}
/**
* @desc This method ensures that the result returned by an analyzer always has a consistent format.
* By returning the configuration for the queryOutput, it will be possible to run later queries
* under the same circumstances
* @param {array} queryOutput
* @param {object} configuration
* @param {object} analyzer
*/
function ensureAnalyzerResultFormat(queryOutput, configuration, analyzer) {
const { targetProjectMeta, identifier, referenceProjectMeta } = analyzer;
const optional = {};
if (targetProjectMeta) {
optional.targetProject = targetProjectMeta;
delete optional.targetProject.path; // get rid of machine specific info
}
if (referenceProjectMeta) {
optional.referenceProject = referenceProjectMeta;
delete optional.referenceProject.path; // get rid of machine specific info
}
/** @type {AnalyzerResult} */
const aResult = {
queryOutput,
analyzerMeta: {
name: analyzer.name,
requiredAst: analyzer.requiredAst,
identifier,
...optional,
configuration,
},
};
// For now, delete data relatable to local machine + path data that will recognize
// projX#v1 (via rootA/projX#v1, rootB/projX#v2) as identical entities.
// Cleaning up local data paths will make sure their hashes will be similar
// across different machines
delete aResult.analyzerMeta.configuration.referenceProjectPath;
delete aResult.analyzerMeta.configuration.targetProjectPath;
const { referenceProjectResult, targetProjectResult } = aResult.analyzerMeta.configuration;
if (referenceProjectResult) {
delete aResult.analyzerMeta.configuration.referenceProjectResult;
} else if (targetProjectResult) {
delete aResult.analyzerMeta.configuration.targetProjectResult;
}
if (Array.isArray(aResult.queryOutput)) {
aResult.queryOutput.forEach(projectOutput => {
if (projectOutput.project) {
delete projectOutput.project.path;
}
});
}
if (process.platform === 'win32') {
posixify(aResult);
}
return aResult;
}
/**
* @desc Before running the analyzer, we need two conditions for a 'compatible match':
* 1. referenceProject is imported by targetProject at all
* 2. referenceProject and targetProject have compatible major versions
* @param {string} referencePath
* @param {string} targetPath
*/
function checkForMatchCompatibility(referencePath, targetPath) {
const refFile = pathLib.resolve(referencePath, 'package.json');
const referencePkg = JSON.parse(fs.readFileSync(refFile, 'utf8'));
const targetFile = pathLib.resolve(targetPath, 'package.json');
const targetPkg = JSON.parse(fs.readFileSync(targetFile, 'utf8'));
const allTargetDeps = [
...Object.entries(targetPkg.devDependencies || {}),
...Object.entries(targetPkg.dependencies || {}),
];
const importEntry = allTargetDeps.find(([name]) => referencePkg.name === name);
if (!importEntry) {
return { compatible: false, reason: 'no-dependency' };
}
if (!semver.satisfies(referencePkg.version, importEntry[1])) {
return { compatible: false, reason: 'no-matched-version' };
}
return { compatible: true };
}
/**
* If in json format, 'unwind' to be compatible for analysis...
* @param {AnalyzerResult} targetOrReferenceProjectResult
*/
function unwindJsonResult(targetOrReferenceProjectResult) {
const { queryOutput } = targetOrReferenceProjectResult;
const { analyzerMeta } = targetOrReferenceProjectResult.meta;
return { queryOutput, analyzerMeta };
}
class Analyzer {
constructor() {
this.requiredAst = 'babel';
}
static get requiresReference() {
return false;
}
/**
* In a MatchAnalyzer, two Analyzers (a reference and targer) are run.
* For instance, in a MatchImportsAnalyzer, a FindExportsAnalyzer and FinImportsAnalyzer are run.
* Their results can be provided as config params.
* If they are stored in json format, 'unwind' them to be compatible for analysis...
* @param {MatchAnalyzerConfig} cfg
*/
static __unwindProvidedResults(cfg) {
if (cfg.targetProjectResult && !cfg.targetProjectResult.analyzerMeta) {
cfg.targetProjectResult = unwindJsonResult(cfg.targetProjectResult);
}
if (cfg.referenceProjectResult && !cfg.referenceProjectResult.analyzerMeta) {
cfg.referenceProjectResult = unwindJsonResult(cfg.referenceProjectResult);
}
}
/**
* @param {AnalyzerConfig} cfg
* @returns {CachedAnalyzerResult|undefined}
*/
_prepare(cfg) {
this.constructor.__unwindProvidedResults(cfg);
if (!cfg.targetProjectResult) {
this.targetProjectMeta = InputDataService.getProjectMeta(cfg.targetProjectPath, true);
} else {
this.targetProjectMeta = cfg.targetProjectResult.analyzerMeta.targetProject;
}
if (cfg.referenceProjectPath && !cfg.referenceProjectResult) {
this.referenceProjectMeta = InputDataService.getProjectMeta(cfg.referenceProjectPath, true);
} else if (cfg.referenceProjectResult) {
this.referenceProjectMeta = cfg.referenceProjectResult.analyzerMeta.targetProject;
}
/**
* Create a unique hash based on target, reference and configuration
*/
this.identifier = ReportService.createIdentifier({
targetProject: this.targetProjectMeta,
referenceProject: this.referenceProjectMeta,
analyzerConfig: cfg,
});
// If we have a provided result cfg.referenceProjectResult, we assume the providing
// party provides compatible results for now...
if (cfg.referenceProjectPath) {
const { compatible, reason } = checkForMatchCompatibility(
cfg.referenceProjectPath,
cfg.targetProjectPath,
);
if (!compatible) {
LogService.info(
`skipping ${LogService.pad(this.name, 16)} for ${
this.identifier
}: (${reason})\n${cfg.targetProjectPath.replace(
`${process.cwd()}/providence-input-data/search-targets/`,
'',
)}`,
);
return ensureAnalyzerResultFormat(`[${reason}]`, cfg, this);
}
}
/**
* See if we maybe already have our result in cache in the file-system.
*/
const cachedResult = Analyzer._getCachedAnalyzerResult({
analyzerName: this.name,
identifier: this.identifier,
});
if (cachedResult) {
return cachedResult;
}
LogService.info(`starting ${LogService.pad(this.name, 16)} for ${this.identifier}`);
/**
* Get reference and search-target data
*/
if (!cfg.targetProjectResult) {
this.targetData = InputDataService.createDataObject(
[cfg.targetProjectPath],
cfg.gatherFilesConfig,
);
}
if (cfg.referenceProjectPath) {
this.referenceData = InputDataService.createDataObject(
[cfg.referenceProjectPath],
cfg.gatherFilesConfigReference || cfg.gatherFilesConfig,
);
}
return undefined;
}
/**
* @param {QueryOutput} queryOutput
* @param {AnalyzerConfig} cfg
* @returns {AnalyzerResult}
*/
_finalize(queryOutput, cfg) {
const analyzerResult = ensureAnalyzerResultFormat(queryOutput, cfg, this);
LogService.success(`finished ${LogService.pad(this.name, 16)} for ${this.identifier}`);
return analyzerResult;
}
/**
* @param {function} traverseEntry
*/
async _traverse(traverseEntry) {
/**
* Create ASTs for our inputData
*/
const astDataProjects = await QueryService.addAstToProjectsData(this.targetData, 'babel');
return analyzePerAstEntry(astDataProjects[0], traverseEntry);
}
async execute(customConfig = {}) {
const cfg = {
targetProjectPath: null,
referenceProjectPath: null,
...customConfig,
};
/**
* Prepare
*/
const analyzerResult = this._prepare(cfg);
if (analyzerResult) {
return analyzerResult;
}
/**
* Traverse
*/
const queryOutput = await this._traverse(() => {});
/**
* Finalize
*/
return this._finalize(queryOutput, cfg);
}
/**
* @desc Gets a cached result from ReportService. Since ReportService slightly modifies analyzer
* output, we 'unwind' before we return...
* @param {object} config
* @param {string} config.analyzerName
* @param {string} config.identifier
* @returns {AnalyzerResult|undefined}
*/
static _getCachedAnalyzerResult({ analyzerName, identifier }) {
const cachedResult = ReportService.getCachedResult({ analyzerName, identifier });
if (!cachedResult) {
return undefined;
}
LogService.success(`cached version found for ${identifier}`);
/** @type {AnalyzerResult} */
const result = unwindJsonResult(cachedResult);
result.analyzerMeta.__fromCache = true;
return result;
}
}
module.exports = { Analyzer };