react-native/scripts/circleci/code-analysis-bot.js
Héctor Ramos 6967a4fa56 Flag code analysis issues once per converter used (#21503)
Summary:
Fixes issue where the bot would leave multiple lines in the top review comment if, say, eslint found 30 issues, there would be 30 mentions of eslint having found issues.

See https://github.com/facebook/react-native/pull/21492 for an example of such a case, where `analysis-bot` left a spammy review at https://github.com/facebook/react-native/pull/21492#pullrequestreview-161837975.
Pull Request resolved: https://github.com/facebook/react-native/pull/21503

Differential Revision: D10219439

Pulled By: hramos

fbshipit-source-id: 75d32ef3bfeaa91ab614763a19494659ad1be0dd
2018-10-05 11:33:38 -07:00

325 lines
8.3 KiB
JavaScript

/**
* Copyright (c) Facebook, Inc. and its affiliates.
*
* This source code is licensed under the MIT license found in the
* LICENSE file in the root directory of this source tree.
*
* @format
*/
'use strict';
if (!process.env.GITHUB_OWNER) {
console.error('Missing GITHUB_OWNER. Example: facebook');
process.exit(1);
}
if (!process.env.GITHUB_REPO) {
console.error('Missing GITHUB_REPO. Example: react-native');
process.exit(1);
}
// https://octokit.github.io/rest.js/
const octokit = require('@octokit/rest')();
const path = require('path');
function push(arr, key, value) {
if (!arr[key]) {
arr[key] = [];
}
arr[key].push(value);
}
/**
* There is unfortunately no standard format to report an error, so we have
* to write a specific converter for each tool we want to support.
*
* Those functions take a json object as input and fill the output with the
* following format:
*
* { [ path: string ]: Array< { message: string, line: number }> }
*
* This is an object where the keys are the path of the files and values
* is an array of objects of the shape message and line.
*/
const converters = {
raw: function(output, input) {
for (let key in input) {
input[key].forEach(function(message) {
push(output, key, message);
});
}
},
flow: function(output, input) {
if (!input || !input.errors) {
return;
}
input.errors.forEach(function(error) {
push(output, error.message[0].path, {
message: error.message.map(message => message.descr).join(' '),
line: error.message[0].line,
converter: 'flow',
});
});
},
eslint: function(output, input) {
if (!input) {
return;
}
input.forEach(function(file) {
file.messages.forEach(function(message) {
push(output, file.filePath, {
message: message.ruleId + ': ' + message.message,
line: message.line,
converter: 'eslint',
});
});
});
},
shellcheck: function(output, input) {
if (!input) {
return;
}
input.forEach(function(report) {
push(output, report.file, {
message:
'**[SC' +
report.code +
'](https://github.com/koalaman/shellcheck/wiki/SC' +
report.code +
'):** (' +
report.level +
') ' +
report.message,
line: report.line,
endLine: report.endLine,
column: report.column,
endColumn: report.endColumn,
converter: 'shellcheck',
});
});
},
};
function getShaFromPullRequest(owner, repo, number, callback) {
octokit.pullRequests.get({owner, repo, number}, (error, res) => {
if (error) {
console.error(error);
return;
}
callback(res.data.head.sha);
});
}
function getFilesFromCommit(owner, repo, sha, callback) {
octokit.repos.getCommit({owner, repo, sha}, (error, res) => {
if (error) {
console.error(error);
return;
}
// A merge commit should not have any new changes to report
if (res.parents && res.parents.length > 1) {
return;
}
callback(res.data.files);
});
}
/**
* Sadly we can't just give the line number to github, we have to give the
* line number relative to the patch file which is super annoying. This
* little function builds a map of line number in the file to line number
* in the patch file
*/
function getLineMapFromPatch(patchString) {
let diffLineIndex = 0;
let fileLineIndex = 0;
let lineMap = {};
patchString.split('\n').forEach(line => {
if (line.match(/^@@/)) {
fileLineIndex = line.match(/\+([0-9]+)/)[1] - 1;
return;
}
diffLineIndex++;
if (line[0] !== '-') {
fileLineIndex++;
if (line[0] === '+') {
lineMap[fileLineIndex] = diffLineIndex;
}
}
});
return lineMap;
}
function sendReview(owner, repo, number, commit_id, body, comments) {
if (process.env.GITHUB_TOKEN) {
if (comments.length === 0) {
// Do not leave an empty review.
return;
}
const event = 'REQUEST_CHANGES';
const opts = {
owner,
repo,
number,
commit_id,
body,
event,
comments,
};
octokit.pullRequests.createReview(opts, function(error, res) {
if (error) {
console.error(error);
return;
}
});
} else {
if (comments.length === 0) {
console.log('No issues found.');
return;
}
if (process.env.CIRCLE_CI) {
console.error(
'Code analysis found issues, but the review cannot be posted to GitHub without an access token.',
);
process.exit(1);
}
let results = body + '\n';
comments.forEach(comment => {
results +=
comment.path + ':' + comment.position + ': ' + comment.body + '\n';
});
console.log(results);
}
}
function main(messages, owner, repo, number) {
// No message, we don't need to do anything :)
if (Object.keys(messages).length === 0) {
return;
}
if (process.env.GITHUB_TOKEN) {
octokit.authenticate({
type: 'oauth',
token: process.env.GITHUB_TOKEN,
});
} else {
console.log(
'Missing GITHUB_TOKEN. Example: 5fd88b964fa214c4be2b144dc5af5d486a2f8c1e. Review feedback with code analysis results will not be provided on GitHub.',
);
}
getShaFromPullRequest(owner, repo, number, sha => {
getFilesFromCommit(owner, repo, sha, files => {
let comments = [];
let convertersUsed = [];
files.filter(file => messages[file.filename]).forEach(file => {
// github api sometimes does not return a patch on large commits
if (!file.patch) {
return;
}
const lineMap = getLineMapFromPatch(file.patch);
messages[file.filename].forEach(message => {
if (lineMap[message.line]) {
const comment = {
path: file.filename,
position: lineMap[message.line],
body: message.message,
};
convertersUsed.push(message.converter);
comments.push(comment);
}
}); // forEach
}); // filter
let body = '**Code analysis results:**\n\n';
const uniqueconvertersUsed = [...new Set(convertersUsed)];
uniqueconvertersUsed.forEach(converter => {
body += '* `' + converter + '` found some issues.\n';
});
sendReview(owner, repo, number, sha, body, comments);
}); // getFilesFromCommit
}); // getShaFromPullRequest
}
let content = '';
process.stdin.resume();
process.stdin.on('data', function(buf) {
content += buf.toString();
});
process.stdin.on('end', function() {
let messages = {};
// Since we send a few http requests to setup the process, we don't want
// to run this file one time per code analysis tool. Instead, we write all
// the results in the same stdin stream.
// The format of this stream is
//
// name-of-the-converter
// {"json":"payload"}
// name-of-the-other-converter
// {"other": ["json", "payload"]}
//
// In order to generate such stream, here is a sample bash command:
//
// cat <(echo eslint; npm run lint --silent -- --format=json; echo flow; flow --json) | node code-analysis-bot.js
const lines = content.trim().split('\n');
for (let i = 0; i < Math.ceil(lines.length / 2); ++i) {
const converter = converters[lines[i * 2]];
if (!converter) {
throw new Error('Unknown converter ' + lines[i * 2]);
}
let json;
try {
json = JSON.parse(lines[i * 2 + 1]);
} catch (e) {}
converter(messages, json);
}
// The paths are returned in absolute from code analysis tools but github works
// on paths relative from the root of the project. Doing the normalization here.
const pwd = path.resolve('.');
for (let absolutePath in messages) {
const relativePath = path.relative(pwd, absolutePath);
if (relativePath === absolutePath) {
continue;
}
messages[relativePath] = messages[absolutePath];
delete messages[absolutePath];
}
const owner = process.env.GITHUB_OWNER;
const repo = process.env.GITHUB_REPO;
if (!process.env.GITHUB_PR_NUMBER) {
console.error('Missing GITHUB_PR_NUMBER. Example: 4687');
// for master branch, don't throw an error
process.exit(0);
}
const number = process.env.GITHUB_PR_NUMBER;
// intentional lint warning to make sure that the bot is working :)
main(messages, owner, repo, number);
});