tools: Rate limit operation and fix strange buffer issues

Slightly relaxes the necessary amount of memory, as we are no longer loading everything all at once. Also for unknown reasons git interferes with other git processes running in the same repository only on Linux. This causes Linux runs of this tool to have strange issues detecting the authors, as git just quits with a success error code. Fun.
This commit is contained in:
Michael Fabian 'Xaymar' Dirks 2023-03-01 11:29:57 +01:00
parent a091b08259
commit c9ff7093d4
1 changed files with 401 additions and 369 deletions

View File

@ -11,416 +11,448 @@ const OS = require("os");
const SECTION_START = "AUTOGENERATED COPYRIGHT HEADER START";
const SECTION_END = "AUTOGENERATED COPYRIGHT HEADER END";
const IGNORED = [
".git",
"cmake/clang",
"cmake/version",
"third-party",
/^\.git$/gi,
/^cmake\/clang$/gi,
/^cmake\/version$/gi,
/^third-party$/gi,
]
let abortAllWork = false;
class RateLimiter {
constructor(limit = undefined) {
const OS = require("node:os");
this._limit = limit;
if (!this._limit) {
this._limit = Math.ceil(Math.max(2, OS.cpus().length / 3 * 2));
}
this._cur = this._limit;
this._pend = 0;
this._locks = [];
}
async run(runner) {
// Use Promises to spin-lock this execution path until there is a free slot.
this._pend += 1;
while (true) {
if (this._cur > 0) {
this._cur -= 1;
break;
} else {
await Promise.race(this._locks);
}
}
this._pend -= 1;
let data = {};
data.pri = new Promise((resolve, reject) => {
try {
if (runner.constructor.name == "AsyncFunction") {
runner().then((res) => {
resolve(res);
}, (err) => {
reject(err);
})
} else {
resolve(runner());
}
} catch (ex) {
reject(ex);
}
});
data.sec = data.pri.finally(() => {
// Remove this promise from the locks list.
let idx = this._locks.indexOf(data.pri);
if (idx >= 0) {
this._locks.splice(idx, 1);
}
let idx2 = this._locks.indexOf(data.sec);
if (idx2 >= 0) {
this._locks.splice(idx2, 1);
}
this._cur += 1;
//console.log(`Avail: ${this._cur} / ${this._limit}; Pending: ${this._pend}`)
});
this._locks.push(data.sec);
return await data.sec;
}
}
let gitRL = new RateLimiter(1);
let workRL = new RateLimiter();
async function isIgnored(path) {
let rpath = PATH.relative(process.cwd(), path).replaceAll(PATH.sep, PATH.posix.sep);
for (let ignore of IGNORED) {
if (ignore instanceof RegExp) {
if (ignore.global) {
if (!rpath.matchAll(ignore).done) {
return true;
}
} else {
if (rpath.match(ignore) !== null) {
return true;
}
}
} else if (rpath.startsWith(ignore)) {
return true;
}
}
let rpath = PATH.relative(process.cwd(), path).replaceAll(PATH.sep, PATH.posix.sep);
for (let ignore of IGNORED) {
if (ignore instanceof RegExp) {
if (ignore.global) {
let matches = rpath.matchAll(ignore);
for (let match of matches) {
return true;
}
} else {
if (rpath.match(ignore) !== null) {
return true;
}
}
} else if (rpath.startsWith(ignore)) {
return true;
}
}
return await new Promise((resolve, reject) => {
try {
let proc = CHILD_PROCESS.spawn("git", [
"check-ignore",
path
], {
"cwd": PROCESS.cwd(),
"encoding": "utf8",
});
proc.stdout.on('data', (data) => {
})
proc.on('close', (code) => {
resolve(code == 0);
});
proc.on('exit', (code) => {
resolve(code == 0);
});
} catch (ex) {
reject(ex);
}
});
/* Sync alternative
try {
return CHILD_PROCESS.spawnSync("git", [
"check-ignore",
path
], {
"cwd": PROCESS.cwd(),
"encoding": "utf8"
}).status == 0;
} catch (ex) {
return true;
}
*/
return await gitRL.run(async () => {
return await new Promise((resolve, reject) => {
try {
let proc = CHILD_PROCESS.spawn("git", [
"check-ignore",
path
], {
"cwd": PROCESS.cwd(),
"encoding": "utf8",
});
proc.stdout.on('data', (data) => {
})
proc.on('close', (code) => {
resolve(code == 0);
});
proc.on('exit', (code) => {
resolve(code == 0);
});
} catch (ex) {
reject(ex);
}
});
});
}
async function git_retrieveAuthors(file) {
// git --no-pager log --date-order --reverse "--format=format:%aI|%aN <%aE>" -- file
let lines = await new Promise((resolve, reject) => {
try {
let lines = "";
let proc = CHILD_PROCESS.spawn("git", [
"--no-pager",
"log",
"--date-order",
"--reverse",
"--format=format:%aI|%aN <%aE>",
"--",
file
], {
"cwd": PROCESS.cwd(),
"encoding": "utf8",
});
proc.stdout.on('data', (data) => {
lines += data.toString();
})
proc.on('close', (code) => {
resolve(lines);
});
proc.on('exit', (code) => {
resolve(lines);
});
} catch (ex) {
reject(ex);
}
});
// git --no-pager log --date-order --reverse "--format=format:%aI|%aN <%aE>" -- file
let lines = await gitRL.run(async () => {
return await new Promise((resolve, reject) => {
try {
let chunks = [];
let proc = CHILD_PROCESS.spawn("git", [
"--no-pager",
"log",
"--date-order",
"--reverse",
"--format=format:%aI|%aN <%aE>",
"--",
file
], {
"cwd": PROCESS.cwd(),
"encoding": "utf8",
});
proc.stdout.on('data', (chunk) => {
chunks.push(chunk);
});
proc.stdout.on('close', () => {
let chunk = proc.stdout.read();
if (chunk) {
chunks.push(chunk);
}
});
proc.on('exit', (code) => {
// Merge all data into one buffer.
let length = 0;
for (let chunk of chunks) {
length += chunk.byteLength;
}
let buf = Buffer.alloc(length);
length = 0;
for (let chunk of chunks) {
if (!(chunk instanceof Buffer)) {
chunk = Buffer.from(chunk);
}
chunk.copy(buf, length, 0);
length += chunk.byteLength;
}
lines = lines.split(lines.indexOf("\r\n") >= 0 ? "\r\n" : "\n");
let authors = new Map();
for (let line of lines) {
let [date, name] = line.split("|");
if (code == 0) {
if (buf) {
resolve(buf.toString());
} else {
reject(code);
}
} else {
reject(code);
}
});
} catch (ex) {
reject(ex);
}
});
});
let author = authors.get(name);
if (author) {
author.to = new Date(date)
} else {
authors.set(name, {
from: new Date(date),
to: new Date(date),
})
}
}
return authors;
lines = lines.split(lines.indexOf("\r\n") >= 0 ? "\r\n" : "\n");
let authors = new Map();
for (let line of lines) {
let [date, name] = line.split("|");
/* Sync Variant
try {
let data = await CHILD_PROCESS
let lines = data.stdout.toString().split("\n");
let authors = new Map();
for (let line of lines) {
let [date, name] = line.split("|");
let author = authors.get(name);
if (author) {
author.to = new Date(date)
} else {
authors.set(name, {
from: new Date(date),
to: new Date(date),
})
}
}
return authors;
} catch (ex) {
console.error(ex);
throw ex;
}
*/
let author = authors.get(name);
if (author) {
author.to = new Date(date)
} else {
authors.set(name, {
from: new Date(date),
to: new Date(date),
})
}
}
return authors;
}
async function generateCopyright(file) {
let authors = await git_retrieveAuthors(file)
let lines = [];
for (let entry of authors) {
let from = entry[1].from.getUTCFullYear();
let to = entry[1].to.getUTCFullYear();
lines.push(`Copyright (C) ${from != to ? `${from}-${to}` : to} ${entry[0]}`);
}
return lines;
let authors = await git_retrieveAuthors(file)
let lines = [];
for (let entry of authors) {
let from = entry[1].from.getUTCFullYear();
let to = entry[1].to.getUTCFullYear();
lines.push(`Copyright (C) ${from != to ? `${from}-${to}` : to} ${entry[0]}`);
}
return lines;
}
function makeHeader(file, copyright) {
let file_name = PATH.basename(file).toLocaleLowerCase();
let file_exts = file_name.substring(file_name.indexOf("."));
let file_name = PATH.basename(file).toLocaleLowerCase();
let file_exts = file_name.substring(file_name.indexOf("."));
let styles = {
"#": {
files: [
"cmakelists.txt"
], exts: [
".clang-tidy",
".clang-format",
".cmake",
".editorconfig",
".gitignore",
".gitmodules",
".yml",
],
prepend: [
`# ${SECTION_START}`,
],
append: [
`# ${SECTION_END}`,
],
prefix: "# ",
suffix: "",
},
";": {
files: [
""
], exts: [
".iss",
".iss.in",
],
prepend: [
`; ${SECTION_START}`,
],
append: [
`; ${SECTION_END}`,
],
prefix: "; ",
suffix: "",
},
"//": {
files: [
], exts: [
".c",
".c.in",
".cpp",
".cpp.in",
".h",
".h.in",
".hpp",
".hpp.in",
".js",
".rc",
".rc.in",
".effect"
],
prepend: [
`// ${SECTION_START}`,
],
append: [
`// ${SECTION_END}`,
],
prefix: "// ",
suffix: "",
},
"<!---->": {
files: [
], exts: [
".htm",
".htm.in",
".html",
".html.in",
".xml",
".xml.in",
".plist",
".plist.in",
".pkgproj",
".pkgproj.in",
],
prepend: [
`<!-- ${SECTION_START} -->`,
],
append: [
`<!-- ${SECTION_END} -->`,
],
prefix: "<!-- ",
suffix: " -->",
}
};
let styles = {
"#": {
files: [
"cmakelists.txt"
], exts: [
".clang-tidy",
".clang-format",
".cmake",
".editorconfig",
".gitignore",
".gitmodules",
".yml",
],
prepend: [
`# ${SECTION_START}`,
],
append: [
`# ${SECTION_END}`,
],
prefix: "# ",
suffix: "",
},
";": {
files: [
""
], exts: [
".iss",
".iss.in",
],
prepend: [
`; ${SECTION_START}`,
],
append: [
`; ${SECTION_END}`,
],
prefix: "; ",
suffix: "",
},
"//": {
files: [
], exts: [
".c",
".c.in",
".cpp",
".cpp.in",
".h",
".h.in",
".hpp",
".hpp.in",
".js",
".rc",
".rc.in",
".effect"
],
prepend: [
`// ${SECTION_START}`,
],
append: [
`// ${SECTION_END}`,
],
prefix: "// ",
suffix: "",
},
"<!---->": {
files: [
], exts: [
".htm",
".htm.in",
".html",
".html.in",
".xml",
".xml.in",
".plist",
".plist.in",
".pkgproj",
".pkgproj.in",
],
prepend: [
`<!-- ${SECTION_START} -->`,
],
append: [
`<!-- ${SECTION_END} -->`,
],
prefix: "<!-- ",
suffix: " -->",
}
};
for (let key in styles) {
let style = [key, styles[key]];
if (style[1].files.includes(file_name)
|| style[1].files.includes(file)
|| style[1].exts.includes(file_exts)) {
let header = [];
header.push(...style[1].prepend);
for (let line of copyright) {
header.push(`${style[1].prefix}${line}${style[1].suffix}`);
}
header.push(...style[1].append);
return header;
}
}
for (let key in styles) {
let style = [key, styles[key]];
if (style[1].files.includes(file_name)
|| style[1].files.includes(file)
|| style[1].exts.includes(file_exts)) {
let header = [];
header.push(...style[1].prepend);
for (let line of copyright) {
header.push(`${style[1].prefix}${line}${style[1].suffix}`);
}
header.push(...style[1].append);
return header;
}
}
throw new Error("Unrecognized file format.")
throw new Error("Unrecognized file format.")
}
async function addCopyright(file) {
try {
if (abortAllWork) {
return;
}
async function updateFile(file) {
await workRL.run(async () => {
try {
if (abortAllWork) {
return;
}
// Async/Promises
// Copyright information.
let copyright = await generateCopyright(file);
let header = undefined;
try {
header = makeHeader(file, copyright);
} catch (ex) {
return;
}
console.log(`Updating file '${file}'...`);
// Copyright information.
let copyright = await generateCopyright(file);
let header = undefined;
try {
header = makeHeader(file, copyright);
} catch (ex) {
console.log(`Skipping file '${file}'...`);
return;
}
console.log(`Updating file '${file}'...`);
// File contents.
let content = await FSPROMISES.readFile(file);
let eol = (content.indexOf("\r\n") != -1 ? OS.EOL : "\n");
let insert = Buffer.from(header.join(eol) + eol);
// File contents.
let content = await FSPROMISES.readFile(file);
let eol = (content.indexOf("\r\n") != -1 ? OS.EOL : "\n");
let insert = Buffer.from(header.join(eol) + eol);
// Find the starting point.
let startHeader = content.indexOf(SECTION_START);
startHeader = content.lastIndexOf(eol, startHeader);
startHeader += Buffer.from(eol).byteLength;
// Find the starting point.
let startHeader = content.indexOf(SECTION_START);
startHeader = content.lastIndexOf(eol, startHeader);
startHeader += Buffer.from(eol).byteLength;
// Find the ending point.
let endHeader = content.indexOf(SECTION_END);
endHeader = content.indexOf(eol, endHeader);
endHeader += Buffer.from(eol).byteLength;
// Find the ending point.
let endHeader = content.indexOf(SECTION_END);
endHeader = content.indexOf(eol, endHeader);
endHeader += Buffer.from(eol).byteLength;
if (abortAllWork) {
return;
}
if (abortAllWork) {
return;
}
let fd = await FSPROMISES.open(file, "w");
let fp = [];
if ((startHeader >= 0) && (endHeader >= 0)) {
let pos = 0;
if (startHeader > 0) {
fd.write(content, 0, startHeader, 0);
pos += startHeader;
}
fd.write(insert, 0, undefined, pos);
pos += insert.byteLength;
fd.write(content, endHeader, undefined, pos);
} else {
fd.write(insert, 0, undefined, 0);
fd.write(content, 0, undefined, insert.byteLength);
}
await fd.close();
/* Sync variant (slow!)
let content = FS.readFileSync(file);
let eol = (content.indexOf("\r\n") != -1 ? OS.EOL : "\n");
let copyright = await generateCopyright(file);
let header = makeHeader(file, copyright);
let insert = Buffer.from(header.join(eol) + eol);
let startHeader = content.indexOf(header[0]);
let endHeader = content.indexOf(header[header.length - 1], startHeader + 1);
endHeader += header[header.length - 1].length + eol.length;
let fd = FS.openSync(file, "w+");
if ((startHeader >= 0) && (endHeader >= 0)) {
let pos = 0;
if (startHeader > 0) {
FS.writeSync(fd, content, 0, startHeader, 0);
pos += startHeader;
}
FS.writeSync(fd, insert, 0, undefined, pos);
pos += insert.byteLength;
FS.writeSync(fd, content, endHeader, undefined, pos);
} else {
FS.writeSync(fd, insert, 0, undefined, 0);
FS.writeSync(fd, content, 0, undefined, insert.byteLength);
}
FS.close(fd, (err) => {
if (err)
throw err;
})*/
} catch (ex) {
console.error(`Error processing '${file}'!: ${ex}`);
return;
}
let fd = await FSPROMISES.open(file, "w");
let fp = [];
if ((startHeader >= 0) && (endHeader >= 0)) {
let pos = 0;
if (startHeader > 0) {
fd.write(content, 0, startHeader, 0);
pos += startHeader;
}
fd.write(insert, 0, undefined, pos);
pos += insert.byteLength;
fd.write(content, endHeader, undefined, pos);
} else {
fd.write(insert, 0, undefined, 0);
fd.write(content, 0, undefined, insert.byteLength);
}
await fd.close();
} catch (ex) {
console.error(`Error processing '${file}'!: ${ex}`);
abortAllWork = true;
PROCESS.exitCode = 1;
return;
}
});
}
async function addCopyrights(path) {
if (abortAllWork) {
return;
}
if (await isIgnored(path)) {
return;
}
async function scanPath(path) {
// Abort here if the user aborted the process, or if the path is ignored.
if (abortAllWork) {
return;
}
let promises = [];
let promises = [];
let files = await FSPROMISES.readdir(path, { "withFileTypes": true });
for (let file of files) {
if (abortAllWork) {
break;
}
await workRL.run(async () => {
let files = await FSPROMISES.readdir(path, { "withFileTypes": true });
for (let file of files) {
if (abortAllWork) {
break;
}
let fullname = PATH.join(path, file.name);
if (await isIgnored(fullname)) {
console.log(`Ignoring path '${fullname}'...`);
continue;
}
let fullname = PATH.join(path, file.name);
if (await isIgnored(fullname)) {
console.log(`Ignoring path '${fullname}'...`);
continue;
}
if (file.isDirectory()) {
//console.log(`Scanning path '${fullname}'...`);
promises.push(addCopyrights(fullname));
} else {
promises.push(addCopyright(fullname));
}
}
if (file.isDirectory()) {
console.log(`Scanning path '${fullname}'...`);
promises.push(scanPath(fullname));
} else {
promises.push(updateFile(fullname));
}
}
});
await Promise.all(promises);
await Promise.all(promises);
}
(async function () {
PROCESS.on("SIGINT", (ev) => {
abortAllWork = true;
console.log("Sanely aborting all pending work...");
})
PROCESS.on("SIGINT", () => {
abortAllWork = true;
PROCESS.exitCode = 1;
console.log("Sanely aborting all pending work...");
})
let path = PATH.resolve(PROCESS.argv[2]);
let path = PATH.resolve(PROCESS.argv[2]);
{ // Bootstrap to actually be in the directory where '.git' is.
let is_git_directory = false;
while (!is_git_directory) {
if (abortAllWork) {
return;
}
{ // Bootstrap to actually be in the directory where '.git' is.
let is_git_directory = false;
while (!is_git_directory) {
if (abortAllWork) {
return;
}
let entries = await FSPROMISES.readdir(PROCESS.cwd());
if (entries.includes(".git")) {
console.log(`Found .git at '${process.cwd()}'.`);
is_git_directory = true;
} else {
PROCESS.chdir(PATH.resolve(PATH.join(PROCESS.cwd(), "..")));
}
}
path = PATH.normalize(PATH.relative(process.cwd(), path));
}
let entries = await FSPROMISES.readdir(PROCESS.cwd());
if (entries.includes(".git")) {
console.log(`Found .git at '${process.cwd()}'.`);
is_git_directory = true;
} else {
PROCESS.chdir(PATH.resolve(PATH.join(PROCESS.cwd(), "..")));
}
}
path = PATH.normalize(PATH.relative(process.cwd(), path));
}
let pathStat = await FSPROMISES.stat(path);
if (pathStat.isDirectory()) {
await addCopyrights(path);
} else {
await addCopyright(path);
}
console.log("Done");
if (!await isIgnored(path)) {
if ((await FSPROMISES.stat(path)).isDirectory()) {
console.log(`Scanning path '${path}'...`);
await scanPath(path);
} else {
await updateFile(path);
}
} else {
console.log(`Ignoring path '${path}'...`);
}
console.log("Done");
})();