github/overleaf
1
0
Fork 0
mirror of https://github.com/overleaf/overleaf.git synced 2024-09-16 02:52:31 -04:00
Code Issues Projects Releases Packages Wiki Activity

prettier: convert app/js decaffeinated files to Prettier format

Browse source
This commit is contained in:
mserranom 2020-02-19 12:14:37 +01:00
parent 3af6bdd588
commit 8729acd48c
26 changed files with 3881 additions and 2639 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

18
services/clsi/app/js/CommandRunner.js
View file

@ -5,16 +5,16 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let commandRunnerPath;
const Settings = require("settings-sharelatex");
const logger = require("logger-sharelatex");
let commandRunnerPath
const Settings = require('settings-sharelatex')
const logger = require('logger-sharelatex')
if ((Settings.clsi != null ? Settings.clsi.dockerRunner : undefined) === true) {
commandRunnerPath = "./DockerRunner";
} else {
commandRunnerPath = "./LocalCommandRunner";
commandRunnerPath = './DockerRunner'
} else {
commandRunnerPath = './LocalCommandRunner'
}
logger.info({commandRunnerPath}, "selecting command runner for clsi");
const CommandRunner = require(commandRunnerPath);
logger.info({ commandRunnerPath }, 'selecting command runner for clsi')
const CommandRunner = require(commandRunnerPath)
module.exports = CommandRunner;
module.exports = CommandRunner

356
services/clsi/app/js/CompileController.js
View file

@ -12,159 +12,227 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let CompileController;
const RequestParser = require("./RequestParser");
const CompileManager = require("./CompileManager");
const Settings = require("settings-sharelatex");
const Metrics = require("./Metrics");
const ProjectPersistenceManager = require("./ProjectPersistenceManager");
const logger = require("logger-sharelatex");
const Errors = require("./Errors");
let CompileController
const RequestParser = require('./RequestParser')
const CompileManager = require('./CompileManager')
const Settings = require('settings-sharelatex')
const Metrics = require('./Metrics')
const ProjectPersistenceManager = require('./ProjectPersistenceManager')
const logger = require('logger-sharelatex')
const Errors = require('./Errors')
module.exports = (CompileController = {
compile(req, res, next) {
if (next == null) { next = function(error) {}; }
const timer = new Metrics.Timer("compile-request");
return RequestParser.parse(req.body, function(error, request) {
if (error != null) { return next(error); }
request.project_id = req.params.project_id;
if (req.params.user_id != null) { request.user_id = req.params.user_id; }
return ProjectPersistenceManager.markProjectAsJustAccessed(request.project_id, function(error) {
if (error != null) { return next(error); }
return CompileManager.doCompileWithLock(request, function(error, outputFiles) {
let code, status;
if (outputFiles == null) { outputFiles = []; }
if (error instanceof Errors.AlreadyCompilingError) {
code = 423; // Http 423 Locked
status = "compile-in-progress";
} else if (error instanceof Errors.FilesOutOfSyncError) {
code = 409; // Http 409 Conflict
status = "retry";
} else if (error != null ? error.terminated : undefined) {
status = "terminated";
} else if (error != null ? error.validate : undefined) {
status = `validation-${error.validate}`;
} else if (error != null ? error.timedout : undefined) {
status = "timedout";
logger.log({err: error, project_id: request.project_id}, "timeout running compile");
} else if (error != null) {
status = "error";
code = 500;
logger.warn({err: error, project_id: request.project_id}, "error running compile");
} else {
let file;
status = "failure";
for (file of Array.from(outputFiles)) {
if (file.path != null ? file.path.match(/output\.pdf$/) : undefined) {
status = "success";
}
}
module.exports = CompileController = {
compile(req, res, next) {
if (next == null) {
next = function(error) {}
}
const timer = new Metrics.Timer('compile-request')
return RequestParser.parse(req.body, function(error, request) {
if (error != null) {
return next(error)
}
request.project_id = req.params.project_id
if (req.params.user_id != null) {
request.user_id = req.params.user_id
}
return ProjectPersistenceManager.markProjectAsJustAccessed(
request.project_id,
function(error) {
if (error != null) {
return next(error)
}
return CompileManager.doCompileWithLock(request, function(
error,
outputFiles
) {
let code, status
if (outputFiles == null) {
outputFiles = []
}
if (error instanceof Errors.AlreadyCompilingError) {
code = 423 // Http 423 Locked
status = 'compile-in-progress'
} else if (error instanceof Errors.FilesOutOfSyncError) {
code = 409 // Http 409 Conflict
status = 'retry'
} else if (error != null ? error.terminated : undefined) {
status = 'terminated'
} else if (error != null ? error.validate : undefined) {
status = `validation-${error.validate}`
} else if (error != null ? error.timedout : undefined) {
status = 'timedout'
logger.log(
{ err: error, project_id: request.project_id },
'timeout running compile'
)
} else if (error != null) {
status = 'error'
code = 500
logger.warn(
{ err: error, project_id: request.project_id },
'error running compile'
)
} else {
let file
status = 'failure'
for (file of Array.from(outputFiles)) {
if (
file.path != null
? file.path.match(/output\.pdf$/)
: undefined
) {
status = 'success'
}
}
if (status === "failure") {
logger.warn({project_id: request.project_id, outputFiles}, "project failed to compile successfully, no output.pdf generated");
}
if (status === 'failure') {
logger.warn(
{ project_id: request.project_id, outputFiles },
'project failed to compile successfully, no output.pdf generated'
)
}
// log an error if any core files are found
for (file of Array.from(outputFiles)) {
if (file.path === "core") {
logger.error({project_id:request.project_id, req, outputFiles}, "core file found in output");
}
}
}
// log an error if any core files are found
for (file of Array.from(outputFiles)) {
if (file.path === 'core') {
logger.error(
{ project_id: request.project_id, req, outputFiles },
'core file found in output'
)
}
}
}
if (error != null) {
outputFiles = error.outputFiles || [];
}
if (error != null) {
outputFiles = error.outputFiles || []
}
timer.done();
return res.status(code || 200).send({
compile: {
status,
error: (error != null ? error.message : undefined) || error,
outputFiles: outputFiles.map(file =>
({
url:
`${Settings.apis.clsi.url}/project/${request.project_id}` +
((request.user_id != null) ? `/user/${request.user_id}` : "") +
((file.build != null) ? `/build/${file.build}` : "") +
`/output/${file.path}`,
path: file.path,
type: file.type,
build: file.build
})
)
}
});
});
});
});
},
timer.done()
return res.status(code || 200).send({
compile: {
status,
error: (error != null ? error.message : undefined) || error,
outputFiles: outputFiles.map(file => ({
url:
`${Settings.apis.clsi.url}/project/${request.project_id}` +
(request.user_id != null
? `/user/${request.user_id}`
: '') +
(file.build != null ? `/build/${file.build}` : '') +
`/output/${file.path}`,
path: file.path,
type: file.type,
build: file.build
}))
}
})
})
}
)
})
},
stopCompile(req, res, next) {
const {project_id, user_id} = req.params;
return CompileManager.stopCompile(project_id, user_id, function(error) {
if (error != null) { return next(error); }
return res.sendStatus(204);
});
},
stopCompile(req, res, next) {
const { project_id, user_id } = req.params
return CompileManager.stopCompile(project_id, user_id, function(error) {
if (error != null) {
return next(error)
}
return res.sendStatus(204)
})
},
clearCache(req, res, next) {
if (next == null) { next = function(error) {}; }
return ProjectPersistenceManager.clearProject(req.params.project_id, req.params.user_id, function(error) {
if (error != null) { return next(error); }
return res.sendStatus(204);
});
}, // No content
clearCache(req, res, next) {
if (next == null) {
next = function(error) {}
}
return ProjectPersistenceManager.clearProject(
req.params.project_id,
req.params.user_id,
function(error) {
if (error != null) {
return next(error)
}
return res.sendStatus(204)
}
)
}, // No content
syncFromCode(req, res, next) {
if (next == null) { next = function(error) {}; }
const { file } = req.query;
const line = parseInt(req.query.line, 10);
const column = parseInt(req.query.column, 10);
const { project_id } = req.params;
const { user_id } = req.params;
return CompileManager.syncFromCode(project_id, user_id, file, line, column, function(error, pdfPositions) {
if (error != null) { return next(error); }
return res.json({
pdf: pdfPositions
});
});
},
syncFromCode(req, res, next) {
if (next == null) {
next = function(error) {}
}
const { file } = req.query
const line = parseInt(req.query.line, 10)
const column = parseInt(req.query.column, 10)
const { project_id } = req.params
const { user_id } = req.params
return CompileManager.syncFromCode(
project_id,
user_id,
file,
line,
column,
function(error, pdfPositions) {
if (error != null) {
return next(error)
}
return res.json({
pdf: pdfPositions
})
}
)
},
syncFromPdf(req, res, next) {
if (next == null) { next = function(error) {}; }
const page = parseInt(req.query.page, 10);
const h = parseFloat(req.query.h);
const v = parseFloat(req.query.v);
const { project_id } = req.params;
const { user_id } = req.params;
return CompileManager.syncFromPdf(project_id, user_id, page, h, v, function(error, codePositions) {
if (error != null) { return next(error); }
return res.json({
code: codePositions
});
});
},
syncFromPdf(req, res, next) {
if (next == null) {
next = function(error) {}
}
const page = parseInt(req.query.page, 10)
const h = parseFloat(req.query.h)
const v = parseFloat(req.query.v)
const { project_id } = req.params
const { user_id } = req.params
return CompileManager.syncFromPdf(project_id, user_id, page, h, v, function(
error,
codePositions
) {
if (error != null) {
return next(error)
}
return res.json({
code: codePositions
})
})
},
wordcount(req, res, next) {
if (next == null) { next = function(error) {}; }
const file = req.query.file || "main.tex";
const { project_id } = req.params;
const { user_id } = req.params;
const { image } = req.query;
logger.log({image, file, project_id}, "word count request");
wordcount(req, res, next) {
if (next == null) {
next = function(error) {}
}
const file = req.query.file || 'main.tex'
const { project_id } = req.params
const { user_id } = req.params
const { image } = req.query
logger.log({ image, file, project_id }, 'word count request')
return CompileManager.wordcount(project_id, user_id, file, image, function(error, result) {
if (error != null) { return next(error); }
return res.json({
texcount: result
});
});
},
status(req, res, next ){
if (next == null) { next = function(error){}; }
return res.send("OK");
}
});
return CompileManager.wordcount(project_id, user_id, file, image, function(
error,
result
) {
if (error != null) {
return next(error)
}
return res.json({
texcount: result
})
})
},
status(req, res, next) {
if (next == null) {
next = function(error) {}
}
return res.send('OK')
}
}

1070
services/clsi/app/js/CompileManager.js
View file

File diff suppressed because it is too large Load diff

53
services/clsi/app/js/ContentTypeMapper.js
View file

@ -3,31 +3,36 @@
*/
// TODO: This file was created by bulk-decaffeinate.
// Fix any style issues and re-enable lint.
let ContentTypeMapper;
const Path = require('path');
let ContentTypeMapper
const Path = require('path')
// here we coerce html, css and js to text/plain,
// otherwise choose correct mime type based on file extension,
// falling back to octet-stream
module.exports = (ContentTypeMapper = {
map(path) {
switch (Path.extname(path)) {
case '.txt': case '.html': case '.js': case '.css': case '.svg':
return 'text/plain';
case '.csv':
return 'text/csv';
case '.pdf':
return 'application/pdf';
case '.png':
return 'image/png';
case '.jpg': case '.jpeg':
return 'image/jpeg';
case '.tiff':
return 'image/tiff';
case '.gif':
return 'image/gif';
default:
return 'application/octet-stream';
}
}
});
module.exports = ContentTypeMapper = {
map(path) {
switch (Path.extname(path)) {
case '.txt':
case '.html':
case '.js':
case '.css':
case '.svg':
return 'text/plain'
case '.csv':
return 'text/csv'
case '.pdf':
return 'application/pdf'
case '.png':
return 'image/png'
case '.jpg':
case '.jpeg':
return 'image/jpeg'
case '.tiff':
return 'image/tiff'
case '.gif':
return 'image/gif'
default:
return 'application/octet-stream'
}
}
}

18
services/clsi/app/js/DbQueue.js
View file

@ -5,14 +5,14 @@
* DS102: Remove unnecessary code created because of implicit returns
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
const async = require("async");
const Settings = require("settings-sharelatex");
const logger = require("logger-sharelatex");
const queue = async.queue((task, cb)=> task(cb)
, Settings.parallelSqlQueryLimit);
const async = require('async')
const Settings = require('settings-sharelatex')
const logger = require('logger-sharelatex')
const queue = async.queue(
(task, cb) => task(cb),
Settings.parallelSqlQueryLimit
)
queue.drain = ()=> logger.debug('all items have been processed');
module.exports =
{queue};
queue.drain = () => logger.debug('all items have been processed')
module.exports = { queue }

166
services/clsi/app/js/DockerLockManager.js
View file

@ -10,80 +10,104 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let LockManager;
const logger = require("logger-sharelatex");
let LockManager
const logger = require('logger-sharelatex')
const LockState = {}; // locks for docker container operations, by container name
const LockState = {} // locks for docker container operations, by container name
module.exports = (LockManager = {
module.exports = LockManager = {
MAX_LOCK_HOLD_TIME: 15000, // how long we can keep a lock
MAX_LOCK_WAIT_TIME: 10000, // how long we wait for a lock
LOCK_TEST_INTERVAL: 1000, // retry time
MAX_LOCK_HOLD_TIME: 15000, // how long we can keep a lock
MAX_LOCK_WAIT_TIME: 10000, // how long we wait for a lock
LOCK_TEST_INTERVAL: 1000, // retry time
tryLock(key, callback) {
let lockValue
if (callback == null) {
callback = function(err, gotLock) {}
}
const existingLock = LockState[key]
if (existingLock != null) {
// the lock is already taken, check how old it is
const lockAge = Date.now() - existingLock.created
if (lockAge < LockManager.MAX_LOCK_HOLD_TIME) {
return callback(null, false) // we didn't get the lock, bail out
} else {
logger.error(
{ key, lock: existingLock, age: lockAge },
'taking old lock by force'
)
}
}
// take the lock
LockState[key] = lockValue = { created: Date.now() }
return callback(null, true, lockValue)
},
tryLock(key, callback) {
let lockValue;
if (callback == null) { callback = function(err, gotLock) {}; }
const existingLock = LockState[key];
if (existingLock != null) { // the lock is already taken, check how old it is
const lockAge = Date.now() - existingLock.created;
if (lockAge < LockManager.MAX_LOCK_HOLD_TIME) {
return callback(null, false); // we didn't get the lock, bail out
} else {
logger.error({key, lock: existingLock, age:lockAge}, "taking old lock by force");
}
}
// take the lock
LockState[key] = (lockValue = {created: Date.now()});
return callback(null, true, lockValue);
},
getLock(key, callback) {
let attempt
if (callback == null) {
callback = function(error, lockValue) {}
}
const startTime = Date.now()
return (attempt = () =>
LockManager.tryLock(key, function(error, gotLock, lockValue) {
if (error != null) {
return callback(error)
}
if (gotLock) {
return callback(null, lockValue)
} else if (Date.now() - startTime > LockManager.MAX_LOCK_WAIT_TIME) {
const e = new Error('Lock timeout')
e.key = key
return callback(e)
} else {
return setTimeout(attempt, LockManager.LOCK_TEST_INTERVAL)
}
}))()
},
getLock(key, callback) {
let attempt;
if (callback == null) { callback = function(error, lockValue) {}; }
const startTime = Date.now();
return (attempt = () =>
LockManager.tryLock(key, function(error, gotLock, lockValue) {
if (error != null) { return callback(error); }
if (gotLock) {
return callback(null, lockValue);
} else if ((Date.now() - startTime) > LockManager.MAX_LOCK_WAIT_TIME) {
const e = new Error("Lock timeout");
e.key = key;
return callback(e);
} else {
return setTimeout(attempt, LockManager.LOCK_TEST_INTERVAL);
}
})
)();
},
releaseLock(key, lockValue, callback) {
if (callback == null) {
callback = function(error) {}
}
const existingLock = LockState[key]
if (existingLock === lockValue) {
// lockValue is an object, so we can test by reference
delete LockState[key] // our lock, so we can free it
return callback()
} else if (existingLock != null) {
// lock exists but doesn't match ours
logger.error(
{ key, lock: existingLock },
'tried to release lock taken by force'
)
return callback()
} else {
logger.error(
{ key, lock: existingLock },
'tried to release lock that has gone'
)
return callback()
}
},
releaseLock(key, lockValue, callback) {
if (callback == null) { callback = function(error) {}; }
const existingLock = LockState[key];
if (existingLock === lockValue) { // lockValue is an object, so we can test by reference
delete LockState[key]; // our lock, so we can free it
return callback();
} else if (existingLock != null) { // lock exists but doesn't match ours
logger.error({key, lock: existingLock}, "tried to release lock taken by force");
return callback();
} else {
logger.error({key, lock: existingLock}, "tried to release lock that has gone");
return callback();
}
},
runWithLock(key, runner, callback) {
if (callback == null) { callback = function(error) {}; }
return LockManager.getLock(key, function(error, lockValue) {
if (error != null) { return callback(error); }
return runner((error1, ...args) =>
LockManager.releaseLock(key, lockValue, function(error2) {
error = error1 || error2;
if (error != null) { return callback(error); }
return callback(null, ...Array.from(args));
})
);
});
}
});
runWithLock(key, runner, callback) {
if (callback == null) {
callback = function(error) {}
}
return LockManager.getLock(key, function(error, lockValue) {
if (error != null) {
return callback(error)
}
return runner((error1, ...args) =>
LockManager.releaseLock(key, lockValue, function(error2) {
error = error1 || error2
if (error != null) {
return callback(error)
}
return callback(null, ...Array.from(args))
})
)
})
}
}

1011
services/clsi/app/js/DockerRunner.js
View file

File diff suppressed because it is too large Load diff

73
services/clsi/app/js/DraftModeManager.js
View file

@ -11,34 +11,47 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let DraftModeManager;
const fs = require("fs");
const logger = require("logger-sharelatex");
let DraftModeManager
const fs = require('fs')
const logger = require('logger-sharelatex')
module.exports = (DraftModeManager = {
injectDraftMode(filename, callback) {
if (callback == null) { callback = function(error) {}; }
return fs.readFile(filename, "utf8", function(error, content) {
if (error != null) { return callback(error); }
// avoid adding draft mode more than once
if ((content != null ? content.indexOf("\\documentclass\[draft") : undefined) >= 0) {
return callback();
}
const modified_content = DraftModeManager._injectDraftOption(content);
logger.log({
content: content.slice(0,1024), // \documentclass is normally v near the top
modified_content: modified_content.slice(0,1024),
filename
}, "injected draft class");
return fs.writeFile(filename, modified_content, callback);
});
},
_injectDraftOption(content) {
return content
// With existing options (must be first, otherwise both are applied)
.replace(/\\documentclass\[/g, "\\documentclass[draft,")
// Without existing options
.replace(/\\documentclass\{/g, "\\documentclass[draft]{");
}
});
module.exports = DraftModeManager = {
injectDraftMode(filename, callback) {
if (callback == null) {
callback = function(error) {}
}
return fs.readFile(filename, 'utf8', function(error, content) {
if (error != null) {
return callback(error)
}
// avoid adding draft mode more than once
if (
(content != null
? content.indexOf('\\documentclass[draft')
: undefined) >= 0
) {
return callback()
}
const modified_content = DraftModeManager._injectDraftOption(content)
logger.log(
{
content: content.slice(0, 1024), // \documentclass is normally v near the top
modified_content: modified_content.slice(0, 1024),
filename
},
'injected draft class'
)
return fs.writeFile(filename, modified_content, callback)
})
},
_injectDraftOption(content) {
return (
content
// With existing options (must be first, otherwise both are applied)
.replace(/\\documentclass\[/g, '\\documentclass[draft,')
// Without existing options
.replace(/\\documentclass\{/g, '\\documentclass[draft]{')
)
}
}

48
services/clsi/app/js/Errors.js
View file

@ -4,33 +4,33 @@
*/
// TODO: This file was created by bulk-decaffeinate.
// Fix any style issues and re-enable lint.
let Errors;
let Errors
var NotFoundError = function(message) {
const error = new Error(message);
error.name = "NotFoundError";
error.__proto__ = NotFoundError.prototype;
return error;
};
NotFoundError.prototype.__proto__ = Error.prototype;
const error = new Error(message)
error.name = 'NotFoundError'
error.__proto__ = NotFoundError.prototype
return error
}
NotFoundError.prototype.__proto__ = Error.prototype
var FilesOutOfSyncError = function(message) {
const error = new Error(message);
error.name = "FilesOutOfSyncError";
error.__proto__ = FilesOutOfSyncError.prototype;
return error;
};
FilesOutOfSyncError.prototype.__proto__ = Error.prototype;
const error = new Error(message)
error.name = 'FilesOutOfSyncError'
error.__proto__ = FilesOutOfSyncError.prototype
return error
}
FilesOutOfSyncError.prototype.__proto__ = Error.prototype
var AlreadyCompilingError = function(message) {
const error = new Error(message);
error.name = "AlreadyCompilingError";
error.__proto__ = AlreadyCompilingError.prototype;
return error;
};
AlreadyCompilingError.prototype.__proto__ = Error.prototype;
const error = new Error(message)
error.name = 'AlreadyCompilingError'
error.__proto__ = AlreadyCompilingError.prototype
return error
}
AlreadyCompilingError.prototype.__proto__ = Error.prototype
module.exports = (Errors = {
NotFoundError,
FilesOutOfSyncError,
AlreadyCompilingError
});
module.exports = Errors = {
NotFoundError,
FilesOutOfSyncError,
AlreadyCompilingError
}

273
services/clsi/app/js/LatexRunner.js
View file

@ -13,119 +13,192 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let LatexRunner;
const Path = require("path");
const Settings = require("settings-sharelatex");
const logger = require("logger-sharelatex");
const Metrics = require("./Metrics");
const CommandRunner = require("./CommandRunner");
let LatexRunner
const Path = require('path')
const Settings = require('settings-sharelatex')
const logger = require('logger-sharelatex')
const Metrics = require('./Metrics')
const CommandRunner = require('./CommandRunner')
const ProcessTable = {}; // table of currently running jobs (pids or docker container names)
const ProcessTable = {} // table of currently running jobs (pids or docker container names)
module.exports = (LatexRunner = {
runLatex(project_id, options, callback) {
let command;
if (callback == null) { callback = function(error) {}; }
let {directory, mainFile, compiler, timeout, image, environment, flags} = options;
if (!compiler) { compiler = "pdflatex"; }
if (!timeout) { timeout = 60000; } // milliseconds
module.exports = LatexRunner = {
runLatex(project_id, options, callback) {
let command
if (callback == null) {
callback = function(error) {}
}
let {
directory,
mainFile,
compiler,
timeout,
image,
environment,
flags
} = options
if (!compiler) {
compiler = 'pdflatex'
}
if (!timeout) {
timeout = 60000
} // milliseconds
logger.log({directory, compiler, timeout, mainFile, environment, flags}, "starting compile");
logger.log(
{ directory, compiler, timeout, mainFile, environment, flags },
'starting compile'
)
// We want to run latexmk on the tex file which we will automatically
// generate from the Rtex/Rmd/md file.
mainFile = mainFile.replace(/\.(Rtex|md|Rmd)$/, ".tex");
// We want to run latexmk on the tex file which we will automatically
// generate from the Rtex/Rmd/md file.
mainFile = mainFile.replace(/\.(Rtex|md|Rmd)$/, '.tex')
if (compiler === "pdflatex") {
command = LatexRunner._pdflatexCommand(mainFile, flags);
} else if (compiler === "latex") {
command = LatexRunner._latexCommand(mainFile, flags);
} else if (compiler === "xelatex") {
command = LatexRunner._xelatexCommand(mainFile, flags);
} else if (compiler === "lualatex") {
command = LatexRunner._lualatexCommand(mainFile, flags);
} else {
return callback(new Error(`unknown compiler: ${compiler}`));
}
if (compiler === 'pdflatex') {
command = LatexRunner._pdflatexCommand(mainFile, flags)
} else if (compiler === 'latex') {
command = LatexRunner._latexCommand(mainFile, flags)
} else if (compiler === 'xelatex') {
command = LatexRunner._xelatexCommand(mainFile, flags)
} else if (compiler === 'lualatex') {
command = LatexRunner._lualatexCommand(mainFile, flags)
} else {
return callback(new Error(`unknown compiler: ${compiler}`))
}
if (Settings.clsi != null ? Settings.clsi.strace : undefined) {
command = ["strace", "-o", "strace", "-ff"].concat(command);
}
if (Settings.clsi != null ? Settings.clsi.strace : undefined) {
command = ['strace', '-o', 'strace', '-ff'].concat(command)
}
const id = `${project_id}`; // record running project under this id
const id = `${project_id}` // record running project under this id
return ProcessTable[id] = CommandRunner.run(project_id, command, directory, image, timeout, environment, function(error, output) {
delete ProcessTable[id];
if (error != null) { return callback(error); }
const runs = __guard__(__guard__(output != null ? output.stderr : undefined, x1 => x1.match(/^Run number \d+ of .*latex/mg)), x => x.length) || 0;
const failed = (__guard__(output != null ? output.stdout : undefined, x2 => x2.match(/^Latexmk: Errors/m)) != null) ? 1 : 0;
// counters from latexmk output
const stats = {};
stats["latexmk-errors"] = failed;
stats["latex-runs"] = runs;
stats["latex-runs-with-errors"] = failed ? runs : 0;
stats[`latex-runs-${runs}`] = 1;
stats[`latex-runs-with-errors-${runs}`] = failed ? 1 : 0;
// timing information from /usr/bin/time
const timings = {};
const stderr = output != null ? output.stderr : undefined;
timings["cpu-percent"] = __guard__(stderr != null ? stderr.match(/Percent of CPU this job got: (\d+)/m) : undefined, x3 => x3[1]) || 0;
timings["cpu-time"] = __guard__(stderr != null ? stderr.match(/User time.*: (\d+.\d+)/m) : undefined, x4 => x4[1]) || 0;
timings["sys-time"] = __guard__(stderr != null ? stderr.match(/System time.*: (\d+.\d+)/m) : undefined, x5 => x5[1]) || 0;
return callback(error, output, stats, timings);
});
},
return (ProcessTable[id] = CommandRunner.run(
project_id,
command,
directory,
image,
timeout,
environment,
function(error, output) {
delete ProcessTable[id]
if (error != null) {
return callback(error)
}
const runs =
__guard__(
__guard__(output != null ? output.stderr : undefined, x1 =>
x1.match(/^Run number \d+ of .*latex/gm)
),
x => x.length
) || 0
const failed =
__guard__(output != null ? output.stdout : undefined, x2 =>
x2.match(/^Latexmk: Errors/m)
) != null
? 1
: 0
// counters from latexmk output
const stats = {}
stats['latexmk-errors'] = failed
stats['latex-runs'] = runs
stats['latex-runs-with-errors'] = failed ? runs : 0
stats[`latex-runs-${runs}`] = 1
stats[`latex-runs-with-errors-${runs}`] = failed ? 1 : 0
// timing information from /usr/bin/time
const timings = {}
const stderr = output != null ? output.stderr : undefined
timings['cpu-percent'] =
__guard__(
stderr != null
? stderr.match(/Percent of CPU this job got: (\d+)/m)
: undefined,
x3 => x3[1]
) || 0
timings['cpu-time'] =
__guard__(
stderr != null
? stderr.match(/User time.*: (\d+.\d+)/m)
: undefined,
x4 => x4[1]
) || 0
timings['sys-time'] =
__guard__(
stderr != null
? stderr.match(/System time.*: (\d+.\d+)/m)
: undefined,
x5 => x5[1]
) || 0
return callback(error, output, stats, timings)
}
))
},
killLatex(project_id, callback) {
if (callback == null) { callback = function(error) {}; }
const id = `${project_id}`;
logger.log({id}, "killing running compile");
if ((ProcessTable[id] == null)) {
logger.warn({id}, "no such project to kill");
return callback(null);
} else {
return CommandRunner.kill(ProcessTable[id], callback);
}
},
killLatex(project_id, callback) {
if (callback == null) {
callback = function(error) {}
}
const id = `${project_id}`
logger.log({ id }, 'killing running compile')
if (ProcessTable[id] == null) {
logger.warn({ id }, 'no such project to kill')
return callback(null)
} else {
return CommandRunner.kill(ProcessTable[id], callback)
}
},
_latexmkBaseCommand(flags) {
let args = ["latexmk", "-cd", "-f", "-jobname=output", "-auxdir=$COMPILE_DIR", "-outdir=$COMPILE_DIR", "-synctex=1","-interaction=batchmode"];
if (flags) {
args = args.concat(flags);
}
return (__guard__(Settings != null ? Settings.clsi : undefined, x => x.latexmkCommandPrefix) || []).concat(args);
},
_latexmkBaseCommand(flags) {
let args = [
'latexmk',
'-cd',
'-f',
'-jobname=output',
'-auxdir=$COMPILE_DIR',
'-outdir=$COMPILE_DIR',
'-synctex=1',
'-interaction=batchmode'
]
if (flags) {
args = args.concat(flags)
}
return (
__guard__(
Settings != null ? Settings.clsi : undefined,
x => x.latexmkCommandPrefix
) || []
).concat(args)
},
_pdflatexCommand(mainFile, flags) {
return LatexRunner._latexmkBaseCommand(flags).concat([
"-pdf",
Path.join("$COMPILE_DIR", mainFile)
]);
},
_pdflatexCommand(mainFile, flags) {
return LatexRunner._latexmkBaseCommand(flags).concat([
'-pdf',
Path.join('$COMPILE_DIR', mainFile)
])
},
_latexCommand(mainFile, flags) {
return LatexRunner._latexmkBaseCommand(flags).concat([
"-pdfdvi",
Path.join("$COMPILE_DIR", mainFile)
]);
},
_latexCommand(mainFile, flags) {
return LatexRunner._latexmkBaseCommand(flags).concat([
'-pdfdvi',
Path.join('$COMPILE_DIR', mainFile)
])
},
_xelatexCommand(mainFile, flags) {
return LatexRunner._latexmkBaseCommand(flags).concat([
"-xelatex",
Path.join("$COMPILE_DIR", mainFile)
]);
},
_lualatexCommand(mainFile, flags) {
return LatexRunner._latexmkBaseCommand(flags).concat([
"-lualatex",
Path.join("$COMPILE_DIR", mainFile)
]);
}
});
_xelatexCommand(mainFile, flags) {
return LatexRunner._latexmkBaseCommand(flags).concat([
'-xelatex',
Path.join('$COMPILE_DIR', mainFile)
])
},
_lualatexCommand(mainFile, flags) {
return LatexRunner._latexmkBaseCommand(flags).concat([
'-lualatex',
Path.join('$COMPILE_DIR', mainFile)
])
}
}
function __guard__(value, transform) {
return (typeof value !== 'undefined' && value !== null) ? transform(value) : undefined;
}
return typeof value !== 'undefined' && value !== null
? transform(value)
: undefined
}

117
services/clsi/app/js/LocalCommandRunner.js
View file

@ -13,62 +13,79 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let CommandRunner;
const { spawn } = require("child_process");
const logger = require("logger-sharelatex");
let CommandRunner
const { spawn } = require('child_process')
const logger = require('logger-sharelatex')
logger.info("using standard command runner");
logger.info('using standard command runner')
module.exports = (CommandRunner = {
run(project_id, command, directory, image, timeout, environment, callback) {
let key, value;
if (callback == null) { callback = function(error) {}; }
command = (Array.from(command).map((arg) => arg.toString().replace('$COMPILE_DIR', directory)));
logger.log({project_id, command, directory}, "running command");
logger.warn("timeouts and sandboxing are not enabled with CommandRunner");
module.exports = CommandRunner = {
run(project_id, command, directory, image, timeout, environment, callback) {
let key, value
if (callback == null) {
callback = function(error) {}
}
command = Array.from(command).map(arg =>
arg.toString().replace('$COMPILE_DIR', directory)
)
logger.log({ project_id, command, directory }, 'running command')
logger.warn('timeouts and sandboxing are not enabled with CommandRunner')
// merge environment settings
const env = {};
for (key in process.env) { value = process.env[key]; env[key] = value; }
for (key in environment) { value = environment[key]; env[key] = value; }
// merge environment settings
const env = {}
for (key in process.env) {
value = process.env[key]
env[key] = value
}
for (key in environment) {
value = environment[key]
env[key] = value
}
// run command as detached process so it has its own process group (which can be killed if needed)
const proc = spawn(command[0], command.slice(1), {cwd: directory, env});
// run command as detached process so it has its own process group (which can be killed if needed)
const proc = spawn(command[0], command.slice(1), { cwd: directory, env })
let stdout = "";
proc.stdout.on("data", data=> stdout += data);
let stdout = ''
proc.stdout.on('data', data => (stdout += data))
proc.on("error", function(err){
logger.err({err, project_id, command, directory}, "error running command");
return callback(err);
});
proc.on('error', function(err) {
logger.err(
{ err, project_id, command, directory },
'error running command'
)
return callback(err)
})
proc.on("close", function(code, signal) {
let err;
logger.info({code, signal, project_id}, "command exited");
if (signal === 'SIGTERM') { // signal from kill method below
err = new Error("terminated");
err.terminated = true;
return callback(err);
} else if (code === 1) { // exit status from chktex
err = new Error("exited");
err.code = code;
return callback(err);
} else {
return callback(null, {"stdout": stdout});
}
});
proc.on('close', function(code, signal) {
let err
logger.info({ code, signal, project_id }, 'command exited')
if (signal === 'SIGTERM') {
// signal from kill method below
err = new Error('terminated')
err.terminated = true
return callback(err)
} else if (code === 1) {
// exit status from chktex
err = new Error('exited')
err.code = code
return callback(err)
} else {
return callback(null, { stdout: stdout })
}
})
return proc.pid;
}, // return process id to allow job to be killed if necessary
return proc.pid
}, // return process id to allow job to be killed if necessary
kill(pid, callback) {
if (callback == null) { callback = function(error) {}; }
try {
process.kill(-pid); // kill all processes in group
} catch (err) {
return callback(err);
}
return callback();
}
});
kill(pid, callback) {
if (callback == null) {
callback = function(error) {}
}
try {
process.kill(-pid) // kill all processes in group
} catch (err) {
return callback(err)
}
return callback()
}
}

100
services/clsi/app/js/LockManager.js
View file

@ -11,46 +11,62 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let LockManager;
const Settings = require('settings-sharelatex');
const logger = require("logger-sharelatex");
const Lockfile = require('lockfile'); // from https://github.com/npm/lockfile
const Errors = require("./Errors");
const fs = require("fs");
const Path = require("path");
module.exports = (LockManager = {
LOCK_TEST_INTERVAL: 1000, // 50ms between each test of the lock
MAX_LOCK_WAIT_TIME: 15000, // 10s maximum time to spend trying to get the lock
LOCK_STALE: 5*60*1000, // 5 mins time until lock auto expires
let LockManager
const Settings = require('settings-sharelatex')
const logger = require('logger-sharelatex')
const Lockfile = require('lockfile') // from https://github.com/npm/lockfile
const Errors = require('./Errors')
const fs = require('fs')
const Path = require('path')
module.exports = LockManager = {
LOCK_TEST_INTERVAL: 1000, // 50ms between each test of the lock
MAX_LOCK_WAIT_TIME: 15000, // 10s maximum time to spend trying to get the lock
LOCK_STALE: 5 * 60 * 1000, // 5 mins time until lock auto expires
runWithLock(path, runner, callback) {
if (callback == null) { callback = function(error) {}; }
const lockOpts = {
wait: this.MAX_LOCK_WAIT_TIME,
pollPeriod: this.LOCK_TEST_INTERVAL,
stale: this.LOCK_STALE
};
return Lockfile.lock(path, lockOpts, function(error) {
if ((error != null ? error.code : undefined) === 'EEXIST') {
return callback(new Errors.AlreadyCompilingError("compile in progress"));
} else if (error != null) {
return fs.lstat(path, (statLockErr, statLock)=>
fs.lstat(Path.dirname(path), (statDirErr, statDir)=>
fs.readdir(Path.dirname(path), function(readdirErr, readdirDir){
logger.err({error, path, statLock, statLockErr, statDir, statDirErr, readdirErr, readdirDir}, "unable to get lock");
return callback(error);
})
)
);
} else {
return runner((error1, ...args) =>
Lockfile.unlock(path, function(error2) {
error = error1 || error2;
if (error != null) { return callback(error); }
return callback(null, ...Array.from(args));
})
);
}
});
}
});
runWithLock(path, runner, callback) {
if (callback == null) {
callback = function(error) {}
}
const lockOpts = {
wait: this.MAX_LOCK_WAIT_TIME,
pollPeriod: this.LOCK_TEST_INTERVAL,
stale: this.LOCK_STALE
}
return Lockfile.lock(path, lockOpts, function(error) {
if ((error != null ? error.code : undefined) === 'EEXIST') {
return callback(new Errors.AlreadyCompilingError('compile in progress'))
} else if (error != null) {
return fs.lstat(path, (statLockErr, statLock) =>
fs.lstat(Path.dirname(path), (statDirErr, statDir) =>
fs.readdir(Path.dirname(path), function(readdirErr, readdirDir) {
logger.err(
{
error,
path,
statLock,
statLockErr,
statDir,
statDirErr,
readdirErr,
readdirDir
},
'unable to get lock'
)
return callback(error)
})
)
)
} else {
return runner((error1, ...args) =>
Lockfile.unlock(path, function(error2) {
error = error1 || error2
if (error != null) {
return callback(error)
}
return callback(null, ...Array.from(args))
})
)
}
})
}
}

3
services/clsi/app/js/Metrics.js
View file

@ -1,4 +1,3 @@
// TODO: This file was created by bulk-decaffeinate.
// Sanity-check the conversion and remove this comment.
module.exports = require("metrics-sharelatex");
module.exports = require('metrics-sharelatex')

602
services/clsi/app/js/OutputCacheManager.js
View file

@ -13,263 +13,387 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let OutputCacheManager;
const async = require("async");
const fs = require("fs");
const fse = require("fs-extra");
const Path = require("path");
const logger = require("logger-sharelatex");
const _ = require("underscore");
const Settings = require("settings-sharelatex");
const crypto = require("crypto");
let OutputCacheManager
const async = require('async')
const fs = require('fs')
const fse = require('fs-extra')
const Path = require('path')
const logger = require('logger-sharelatex')
const _ = require('underscore')
const Settings = require('settings-sharelatex')
const crypto = require('crypto')
const OutputFileOptimiser = require("./OutputFileOptimiser");
const OutputFileOptimiser = require('./OutputFileOptimiser')
module.exports = (OutputCacheManager = {
CACHE_SUBDIR: '.cache/clsi',
ARCHIVE_SUBDIR: '.archive/clsi',
// build id is HEXDATE-HEXRANDOM from Date.now()and RandomBytes
// for backwards compatibility, make the randombytes part optional
BUILD_REGEX: /^[0-9a-f]+(-[0-9a-f]+)?$/,
CACHE_LIMIT: 2, // maximum number of cache directories
CACHE_AGE: 60*60*1000, // up to one hour old
module.exports = OutputCacheManager = {
CACHE_SUBDIR: '.cache/clsi',
ARCHIVE_SUBDIR: '.archive/clsi',
// build id is HEXDATE-HEXRANDOM from Date.now()and RandomBytes
// for backwards compatibility, make the randombytes part optional
BUILD_REGEX: /^[0-9a-f]+(-[0-9a-f]+)?$/,
CACHE_LIMIT: 2, // maximum number of cache directories
CACHE_AGE: 60 * 60 * 1000, // up to one hour old
path(buildId, file) {
// used by static server, given build id return '.cache/clsi/buildId'
if (buildId.match(OutputCacheManager.BUILD_REGEX)) {
return Path.join(OutputCacheManager.CACHE_SUBDIR, buildId, file);
} else {
// for invalid build id, return top level
return file;
}
},
path(buildId, file) {
// used by static server, given build id return '.cache/clsi/buildId'
if (buildId.match(OutputCacheManager.BUILD_REGEX)) {
return Path.join(OutputCacheManager.CACHE_SUBDIR, buildId, file)
} else {
// for invalid build id, return top level
return file
}
},
generateBuildId(callback) {
// generate a secure build id from Date.now() and 8 random bytes in hex
if (callback == null) { callback = function(error, buildId) {}; }
return crypto.randomBytes(8, function(err, buf) {
if (err != null) { return callback(err); }
const random = buf.toString('hex');
const date = Date.now().toString(16);
return callback(err, `${date}-${random}`);
});
},
generateBuildId(callback) {
// generate a secure build id from Date.now() and 8 random bytes in hex
if (callback == null) {
callback = function(error, buildId) {}
}
return crypto.randomBytes(8, function(err, buf) {
if (err != null) {
return callback(err)
}
const random = buf.toString('hex')
const date = Date.now().toString(16)
return callback(err, `${date}-${random}`)
})
},
saveOutputFiles(outputFiles, compileDir, callback) {
if (callback == null) { callback = function(error) {}; }
return OutputCacheManager.generateBuildId(function(err, buildId) {
if (err != null) { return callback(err); }
return OutputCacheManager.saveOutputFilesInBuildDir(outputFiles, compileDir, buildId, callback);
});
},
saveOutputFiles(outputFiles, compileDir, callback) {
if (callback == null) {
callback = function(error) {}
}
return OutputCacheManager.generateBuildId(function(err, buildId) {
if (err != null) {
return callback(err)
}
return OutputCacheManager.saveOutputFilesInBuildDir(
outputFiles,
compileDir,
buildId,
callback
)
})
},
saveOutputFilesInBuildDir(outputFiles, compileDir, buildId, callback) {
// make a compileDir/CACHE_SUBDIR/build_id directory and
// copy all the output files into it
if (callback == null) { callback = function(error) {}; }
const cacheRoot = Path.join(compileDir, OutputCacheManager.CACHE_SUBDIR);
// Put the files into a new cache subdirectory
const cacheDir = Path.join(compileDir, OutputCacheManager.CACHE_SUBDIR, buildId);
// Is it a per-user compile? check if compile directory is PROJECTID-USERID
const perUser = Path.basename(compileDir).match(/^[0-9a-f]{24}-[0-9a-f]{24}$/);
saveOutputFilesInBuildDir(outputFiles, compileDir, buildId, callback) {
// make a compileDir/CACHE_SUBDIR/build_id directory and
// copy all the output files into it
if (callback == null) {
callback = function(error) {}
}
const cacheRoot = Path.join(compileDir, OutputCacheManager.CACHE_SUBDIR)
// Put the files into a new cache subdirectory
const cacheDir = Path.join(
compileDir,
OutputCacheManager.CACHE_SUBDIR,
buildId
)
// Is it a per-user compile? check if compile directory is PROJECTID-USERID
const perUser = Path.basename(compileDir).match(
/^[0-9a-f]{24}-[0-9a-f]{24}$/
)
// Archive logs in background
if ((Settings.clsi != null ? Settings.clsi.archive_logs : undefined) || (Settings.clsi != null ? Settings.clsi.strace : undefined)) {
OutputCacheManager.archiveLogs(outputFiles, compileDir, buildId, function(err) {
if (err != null) {
return logger.warn({err}, "erroring archiving log files");
}
});
}
// Archive logs in background
if (
(Settings.clsi != null ? Settings.clsi.archive_logs : undefined) ||
(Settings.clsi != null ? Settings.clsi.strace : undefined)
) {
OutputCacheManager.archiveLogs(outputFiles, compileDir, buildId, function(
err
) {
if (err != null) {
return logger.warn({ err }, 'erroring archiving log files')
}
})
}
// make the new cache directory
return fse.ensureDir(cacheDir, function(err) {
if (err != null) {
logger.error({err, directory: cacheDir}, "error creating cache directory");
return callback(err, outputFiles);
} else {
// copy all the output files into the new cache directory
const results = [];
return async.mapSeries(outputFiles, function(file, cb) {
// don't send dot files as output, express doesn't serve them
if (OutputCacheManager._fileIsHidden(file.path)) {
logger.debug({compileDir, path: file.path}, "ignoring dotfile in output");
return cb();
}
// copy other files into cache directory if valid
const newFile = _.clone(file);
const [src, dst] = Array.from([Path.join(compileDir, file.path), Path.join(cacheDir, file.path)]);
return OutputCacheManager._checkFileIsSafe(src, function(err, isSafe) {
if (err != null) { return cb(err); }
if (!isSafe) {
return cb();
}
return OutputCacheManager._checkIfShouldCopy(src, function(err, shouldCopy) {
if (err != null) { return cb(err); }
if (!shouldCopy) {
return cb();
}
return OutputCacheManager._copyFile(src, dst, function(err) {
if (err != null) { return cb(err); }
newFile.build = buildId; // attach a build id if we cached the file
results.push(newFile);
return cb();
});
});
});
}
, function(err) {
if (err != null) {
// pass back the original files if we encountered *any* error
callback(err, outputFiles);
// clean up the directory we just created
return fse.remove(cacheDir, function(err) {
if (err != null) {
return logger.error({err, dir: cacheDir}, "error removing cache dir after failure");
}
});
} else {
// pass back the list of new files in the cache
callback(err, results);
// let file expiry run in the background, expire all previous files if per-user
return OutputCacheManager.expireOutputFiles(cacheRoot, {keep: buildId, limit: perUser ? 1 : null});
}
});
}
});
},
// make the new cache directory
return fse.ensureDir(cacheDir, function(err) {
if (err != null) {
logger.error(
{ err, directory: cacheDir },
'error creating cache directory'
)
return callback(err, outputFiles)
} else {
// copy all the output files into the new cache directory
const results = []
return async.mapSeries(
outputFiles,
function(file, cb) {
// don't send dot files as output, express doesn't serve them
if (OutputCacheManager._fileIsHidden(file.path)) {
logger.debug(
{ compileDir, path: file.path },
'ignoring dotfile in output'
)
return cb()
}
// copy other files into cache directory if valid
const newFile = _.clone(file)
const [src, dst] = Array.from([
Path.join(compileDir, file.path),
Path.join(cacheDir, file.path)
])
return OutputCacheManager._checkFileIsSafe(src, function(
err,
isSafe
) {
if (err != null) {
return cb(err)
}
if (!isSafe) {
return cb()
}
return OutputCacheManager._checkIfShouldCopy(src, function(
err,
shouldCopy
) {
if (err != null) {
return cb(err)
}
if (!shouldCopy) {
return cb()
}
return OutputCacheManager._copyFile(src, dst, function(err) {
if (err != null) {
return cb(err)
}
newFile.build = buildId // attach a build id if we cached the file
results.push(newFile)
return cb()
})
})
})
},
function(err) {
if (err != null) {
// pass back the original files if we encountered *any* error
callback(err, outputFiles)
// clean up the directory we just created
return fse.remove(cacheDir, function(err) {
if (err != null) {
return logger.error(
{ err, dir: cacheDir },
'error removing cache dir after failure'
)
}
})
} else {
// pass back the list of new files in the cache
callback(err, results)
// let file expiry run in the background, expire all previous files if per-user
return OutputCacheManager.expireOutputFiles(cacheRoot, {
keep: buildId,
limit: perUser ? 1 : null
})
}
}
)
}
})
},
archiveLogs(outputFiles, compileDir, buildId, callback) {
if (callback == null) { callback = function(error) {}; }
const archiveDir = Path.join(compileDir, OutputCacheManager.ARCHIVE_SUBDIR, buildId);
logger.log({dir: archiveDir}, "archiving log files for project");
return fse.ensureDir(archiveDir, function(err) {
if (err != null) { return callback(err); }
return async.mapSeries(outputFiles, function(file, cb) {
const [src, dst] = Array.from([Path.join(compileDir, file.path), Path.join(archiveDir, file.path)]);
return OutputCacheManager._checkFileIsSafe(src, function(err, isSafe) {
if (err != null) { return cb(err); }
if (!isSafe) { return cb(); }
return OutputCacheManager._checkIfShouldArchive(src, function(err, shouldArchive) {
if (err != null) { return cb(err); }
if (!shouldArchive) { return cb(); }
return OutputCacheManager._copyFile(src, dst, cb);
});
});
}
, callback);
});
},
archiveLogs(outputFiles, compileDir, buildId, callback) {
if (callback == null) {
callback = function(error) {}
}
const archiveDir = Path.join(
compileDir,
OutputCacheManager.ARCHIVE_SUBDIR,
buildId
)
logger.log({ dir: archiveDir }, 'archiving log files for project')
return fse.ensureDir(archiveDir, function(err) {
if (err != null) {
return callback(err)
}
return async.mapSeries(
outputFiles,
function(file, cb) {
const [src, dst] = Array.from([
Path.join(compileDir, file.path),
Path.join(archiveDir, file.path)
])
return OutputCacheManager._checkFileIsSafe(src, function(
err,
isSafe
) {
if (err != null) {
return cb(err)
}
if (!isSafe) {
return cb()
}
return OutputCacheManager._checkIfShouldArchive(src, function(
err,
shouldArchive
) {
if (err != null) {
return cb(err)
}
if (!shouldArchive) {
return cb()
}
return OutputCacheManager._copyFile(src, dst, cb)
})
})
},
callback
)
})
},
expireOutputFiles(cacheRoot, options, callback) {
// look in compileDir for build dirs and delete if > N or age of mod time > T
if (callback == null) { callback = function(error) {}; }
return fs.readdir(cacheRoot, function(err, results) {
if (err != null) {
if (err.code === 'ENOENT') { return callback(null); } // cache directory is empty
logger.error({err, project_id: cacheRoot}, "error clearing cache");
return callback(err);
}
expireOutputFiles(cacheRoot, options, callback) {
// look in compileDir for build dirs and delete if > N or age of mod time > T
if (callback == null) {
callback = function(error) {}
}
return fs.readdir(cacheRoot, function(err, results) {
if (err != null) {
if (err.code === 'ENOENT') {
return callback(null)
} // cache directory is empty
logger.error({ err, project_id: cacheRoot }, 'error clearing cache')
return callback(err)
}
const dirs = results.sort().reverse();
const currentTime = Date.now();
const dirs = results.sort().reverse()
const currentTime = Date.now()
const isExpired = function(dir, index) {
if ((options != null ? options.keep : undefined) === dir) { return false; }
// remove any directories over the requested (non-null) limit
if (((options != null ? options.limit : undefined) != null) && (index > options.limit)) { return true; }
// remove any directories over the hard limit
if (index > OutputCacheManager.CACHE_LIMIT) { return true; }
// we can get the build time from the first part of the directory name DDDD-RRRR
// DDDD is date and RRRR is random bytes
const dirTime = parseInt(__guard__(dir.split('-'), x => x[0]), 16);
const age = currentTime - dirTime;
return age > OutputCacheManager.CACHE_AGE;
};
const isExpired = function(dir, index) {
if ((options != null ? options.keep : undefined) === dir) {
return false
}
// remove any directories over the requested (non-null) limit
if (
(options != null ? options.limit : undefined) != null &&
index > options.limit
) {
return true
}
// remove any directories over the hard limit
if (index > OutputCacheManager.CACHE_LIMIT) {
return true
}
// we can get the build time from the first part of the directory name DDDD-RRRR
// DDDD is date and RRRR is random bytes
const dirTime = parseInt(
__guard__(dir.split('-'), x => x[0]),
16
)
const age = currentTime - dirTime
return age > OutputCacheManager.CACHE_AGE
}
const toRemove = _.filter(dirs, isExpired);
const toRemove = _.filter(dirs, isExpired)
const removeDir = (dir, cb) =>
fse.remove(Path.join(cacheRoot, dir), function(err, result) {
logger.log({cache: cacheRoot, dir}, "removed expired cache dir");
if (err != null) {
logger.error({err, dir}, "cache remove error");
}
return cb(err, result);
})
;
const removeDir = (dir, cb) =>
fse.remove(Path.join(cacheRoot, dir), function(err, result) {
logger.log({ cache: cacheRoot, dir }, 'removed expired cache dir')
if (err != null) {
logger.error({ err, dir }, 'cache remove error')
}
return cb(err, result)
})
return async.eachSeries(
toRemove,
(dir, cb) => removeDir(dir, cb),
callback
)
})
},
return async.eachSeries(toRemove, (dir, cb) => removeDir(dir, cb)
, callback);
});
},
_fileIsHidden(path) {
return (path != null ? path.match(/^\.|\/\./) : undefined) != null
},
_fileIsHidden(path) {
return ((path != null ? path.match(/^\.|\/\./) : undefined) != null);
},
_checkFileIsSafe(src, callback) {
// check if we have a valid file to copy into the cache
if (callback == null) {
callback = function(error, isSafe) {}
}
return fs.stat(src, function(err, stats) {
if ((err != null ? err.code : undefined) === 'ENOENT') {
logger.warn(
{ err, file: src },
'file has disappeared before copying to build cache'
)
return callback(err, false)
} else if (err != null) {
// some other problem reading the file
logger.error({ err, file: src }, 'stat error for file in cache')
return callback(err, false)
} else if (!stats.isFile()) {
// other filetype - reject it
logger.warn(
{ src, stat: stats },
'nonfile output - refusing to copy to cache'
)
return callback(null, false)
} else {
// it's a plain file, ok to copy
return callback(null, true)
}
})
},
_checkFileIsSafe(src, callback) {
// check if we have a valid file to copy into the cache
if (callback == null) { callback = function(error, isSafe) {}; }
return fs.stat(src, function(err, stats) {
if ((err != null ? err.code : undefined) === 'ENOENT') {
logger.warn({err, file: src}, "file has disappeared before copying to build cache");
return callback(err, false);
} else if (err != null) {
// some other problem reading the file
logger.error({err, file: src}, "stat error for file in cache");
return callback(err, false);
} else if (!stats.isFile()) {
// other filetype - reject it
logger.warn({src, stat: stats}, "nonfile output - refusing to copy to cache");
return callback(null, false);
} else {
// it's a plain file, ok to copy
return callback(null, true);
}
});
},
_copyFile(src, dst, callback) {
// copy output file into the cache
return fse.copy(src, dst, function(err) {
if ((err != null ? err.code : undefined) === 'ENOENT') {
logger.warn(
{ err, file: src },
'file has disappeared when copying to build cache'
)
return callback(err, false)
} else if (err != null) {
logger.error({ err, src, dst }, 'copy error for file in cache')
return callback(err)
} else {
if (
Settings.clsi != null ? Settings.clsi.optimiseInDocker : undefined
) {
// don't run any optimisations on the pdf when they are done
// in the docker container
return callback()
} else {
// call the optimiser for the file too
return OutputFileOptimiser.optimiseFile(src, dst, callback)
}
}
})
},
_copyFile(src, dst, callback) {
// copy output file into the cache
return fse.copy(src, dst, function(err) {
if ((err != null ? err.code : undefined) === 'ENOENT') {
logger.warn({err, file: src}, "file has disappeared when copying to build cache");
return callback(err, false);
} else if (err != null) {
logger.error({err, src, dst}, "copy error for file in cache");
return callback(err);
} else {
if ((Settings.clsi != null ? Settings.clsi.optimiseInDocker : undefined)) {
// don't run any optimisations on the pdf when they are done
// in the docker container
return callback();
} else {
// call the optimiser for the file too
return OutputFileOptimiser.optimiseFile(src, dst, callback);
}
}
});
},
_checkIfShouldCopy(src, callback) {
if (callback == null) {
callback = function(err, shouldCopy) {}
}
return callback(null, !Path.basename(src).match(/^strace/))
},
_checkIfShouldCopy(src, callback) {
if (callback == null) { callback = function(err, shouldCopy) {}; }
return callback(null, !Path.basename(src).match(/^strace/));
},
_checkIfShouldArchive(src, callback) {
let needle;
if (callback == null) { callback = function(err, shouldCopy) {}; }
if (Path.basename(src).match(/^strace/)) {
return callback(null, true);
}
if ((Settings.clsi != null ? Settings.clsi.archive_logs : undefined) && (needle = Path.basename(src), ["output.log", "output.blg"].includes(needle))) {
return callback(null, true);
}
return callback(null, false);
}
});
_checkIfShouldArchive(src, callback) {
let needle
if (callback == null) {
callback = function(err, shouldCopy) {}
}
if (Path.basename(src).match(/^strace/)) {
return callback(null, true)
}
if (
(Settings.clsi != null ? Settings.clsi.archive_logs : undefined) &&
((needle = Path.basename(src)),
['output.log', 'output.blg'].includes(needle))
) {
return callback(null, true)
}
return callback(null, false)
}
}
function __guard__(value, transform) {
return (typeof value !== 'undefined' && value !== null) ? transform(value) : undefined;
}
return typeof value !== 'undefined' && value !== null
? transform(value)
: undefined
}

155
services/clsi/app/js/OutputFileFinder.js
View file

@ -14,73 +14,102 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let OutputFileFinder;
const async = require("async");
const fs = require("fs");
const Path = require("path");
const { spawn } = require("child_process");
const logger = require("logger-sharelatex");
let OutputFileFinder
const async = require('async')
const fs = require('fs')
const Path = require('path')
const { spawn } = require('child_process')
const logger = require('logger-sharelatex')
module.exports = (OutputFileFinder = {
findOutputFiles(resources, directory, callback) {
if (callback == null) { callback = function(error, outputFiles, allFiles) {}; }
const incomingResources = {};
for (const resource of Array.from(resources)) {
incomingResources[resource.path] = true;
}
return OutputFileFinder._getAllFiles(directory, function(error, allFiles) {
if (allFiles == null) { allFiles = []; }
if (error != null) {
logger.err({err:error}, "error finding all output files");
return callback(error);
}
const outputFiles = [];
for (const file of Array.from(allFiles)) {
if (!incomingResources[file]) {
outputFiles.push({
path: file,
type: __guard__(file.match(/\.([^\.]+)$/), x => x[1])
});
}
}
return callback(null, outputFiles, allFiles);
});
},
module.exports = OutputFileFinder = {
findOutputFiles(resources, directory, callback) {
if (callback == null) {
callback = function(error, outputFiles, allFiles) {}
}
const incomingResources = {}
for (const resource of Array.from(resources)) {
incomingResources[resource.path] = true
}
_getAllFiles(directory, _callback) {
if (_callback == null) { _callback = function(error, fileList) {}; }
const callback = function(error, fileList) {
_callback(error, fileList);
return _callback = function() {};
};
return OutputFileFinder._getAllFiles(directory, function(error, allFiles) {
if (allFiles == null) {
allFiles = []
}
if (error != null) {
logger.err({ err: error }, 'error finding all output files')
return callback(error)
}
const outputFiles = []
for (const file of Array.from(allFiles)) {
if (!incomingResources[file]) {
outputFiles.push({
path: file,
type: __guard__(file.match(/\.([^\.]+)$/), x => x[1])
})
}
}
return callback(null, outputFiles, allFiles)
})
},
// don't include clsi-specific files/directories in the output list
const EXCLUDE_DIRS = ["-name", ".cache", "-o", "-name", ".archive","-o", "-name", ".project-*"];
const args = [directory, "(", ...Array.from(EXCLUDE_DIRS), ")", "-prune", "-o", "-type", "f", "-print"];
logger.log({args}, "running find command");
_getAllFiles(directory, _callback) {
if (_callback == null) {
_callback = function(error, fileList) {}
}
const callback = function(error, fileList) {
_callback(error, fileList)
return (_callback = function() {})
}
const proc = spawn("find", args);
let stdout = "";
proc.stdout.on("data", chunk => stdout += chunk.toString());
proc.on("error", callback);
return proc.on("close", function(code) {
if (code !== 0) {
logger.warn({directory, code}, "find returned error, directory likely doesn't exist");
return callback(null, []);
}
let fileList = stdout.trim().split("\n");
fileList = fileList.map(function(file) {
// Strip leading directory
let path;
return path = Path.relative(directory, file);
});
return callback(null, fileList);
});
}
});
// don't include clsi-specific files/directories in the output list
const EXCLUDE_DIRS = [
'-name',
'.cache',
'-o',
'-name',
'.archive',
'-o',
'-name',
'.project-*'
]
const args = [
directory,
'(',
...Array.from(EXCLUDE_DIRS),
')',
'-prune',
'-o',
'-type',
'f',
'-print'
]
logger.log({ args }, 'running find command')
const proc = spawn('find', args)
let stdout = ''
proc.stdout.on('data', chunk => (stdout += chunk.toString()))
proc.on('error', callback)
return proc.on('close', function(code) {
if (code !== 0) {
logger.warn(
{ directory, code },
"find returned error, directory likely doesn't exist"
)
return callback(null, [])
}
let fileList = stdout.trim().split('\n')
fileList = fileList.map(function(file) {
// Strip leading directory
let path
return (path = Path.relative(directory, file))
})
return callback(null, fileList)
})
}
}
function __guard__(value, transform) {
return (typeof value !== 'undefined' && value !== null) ? transform(value) : undefined;
}
return typeof value !== 'undefined' && value !== null
? transform(value)
: undefined
}

152
services/clsi/app/js/OutputFileOptimiser.js
View file

@ -13,74 +13,92 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let OutputFileOptimiser;
const fs = require("fs");
const Path = require("path");
const { spawn } = require("child_process");
const logger = require("logger-sharelatex");
const Metrics = require("./Metrics");
const _ = require("underscore");
let OutputFileOptimiser
const fs = require('fs')
const Path = require('path')
const { spawn } = require('child_process')
const logger = require('logger-sharelatex')
const Metrics = require('./Metrics')
const _ = require('underscore')
module.exports = (OutputFileOptimiser = {
module.exports = OutputFileOptimiser = {
optimiseFile(src, dst, callback) {
// check output file (src) and see if we can optimise it, storing
// the result in the build directory (dst)
if (callback == null) {
callback = function(error) {}
}
if (src.match(/\/output\.pdf$/)) {
return OutputFileOptimiser.checkIfPDFIsOptimised(src, function(
err,
isOptimised
) {
if (err != null || isOptimised) {
return callback(null)
}
return OutputFileOptimiser.optimisePDF(src, dst, callback)
})
} else {
return callback(null)
}
},
optimiseFile(src, dst, callback) {
// check output file (src) and see if we can optimise it, storing
// the result in the build directory (dst)
if (callback == null) { callback = function(error) {}; }
if (src.match(/\/output\.pdf$/)) {
return OutputFileOptimiser.checkIfPDFIsOptimised(src, function(err, isOptimised) {
if ((err != null) || isOptimised) { return callback(null); }
return OutputFileOptimiser.optimisePDF(src, dst, callback);
});
} else {
return callback((null));
}
},
checkIfPDFIsOptimised(file, callback) {
const SIZE = 16 * 1024 // check the header of the pdf
const result = new Buffer(SIZE)
result.fill(0) // prevent leakage of uninitialised buffer
return fs.open(file, 'r', function(err, fd) {
if (err != null) {
return callback(err)
}
return fs.read(fd, result, 0, SIZE, 0, (errRead, bytesRead, buffer) =>
fs.close(fd, function(errClose) {
if (errRead != null) {
return callback(errRead)
}
if (typeof errReadClose !== 'undefined' && errReadClose !== null) {
return callback(errClose)
}
const isOptimised =
buffer.toString('ascii').indexOf('/Linearized 1') >= 0
return callback(null, isOptimised)
})
)
})
},
checkIfPDFIsOptimised(file, callback) {
const SIZE = 16*1024; // check the header of the pdf
const result = new Buffer(SIZE);
result.fill(0); // prevent leakage of uninitialised buffer
return fs.open(file, "r", function(err, fd) {
if (err != null) { return callback(err); }
return fs.read(fd, result, 0, SIZE, 0, (errRead, bytesRead, buffer) =>
fs.close(fd, function(errClose) {
if (errRead != null) { return callback(errRead); }
if (typeof errReadClose !== 'undefined' && errReadClose !== null) { return callback(errClose); }
const isOptimised = buffer.toString('ascii').indexOf("/Linearized 1") >= 0;
return callback(null, isOptimised);
})
);
});
},
optimisePDF(src, dst, callback) {
if (callback == null) {
callback = function(error) {}
}
const tmpOutput = dst + '.opt'
const args = ['--linearize', src, tmpOutput]
logger.log({ args }, 'running qpdf command')
optimisePDF(src, dst, callback) {
if (callback == null) { callback = function(error) {}; }
const tmpOutput = dst + '.opt';
const args = ["--linearize", src, tmpOutput];
logger.log({args}, "running qpdf command");
const timer = new Metrics.Timer("qpdf");
const proc = spawn("qpdf", args);
let stdout = "";
proc.stdout.on("data", chunk => stdout += chunk.toString());
callback = _.once(callback); // avoid double call back for error and close event
proc.on("error", function(err) {
logger.warn({err, args}, "qpdf failed");
return callback(null);
}); // ignore the error
return proc.on("close", function(code) {
timer.done();
if (code !== 0) {
logger.warn({code, args}, "qpdf returned error");
return callback(null); // ignore the error
}
return fs.rename(tmpOutput, dst, function(err) {
if (err != null) {
logger.warn({tmpOutput, dst}, "failed to rename output of qpdf command");
}
return callback(null);
});
});
} // ignore the error
});
const timer = new Metrics.Timer('qpdf')
const proc = spawn('qpdf', args)
let stdout = ''
proc.stdout.on('data', chunk => (stdout += chunk.toString()))
callback = _.once(callback) // avoid double call back for error and close event
proc.on('error', function(err) {
logger.warn({ err, args }, 'qpdf failed')
return callback(null)
}) // ignore the error
return proc.on('close', function(code) {
timer.done()
if (code !== 0) {
logger.warn({ code, args }, 'qpdf returned error')
return callback(null) // ignore the error
}
return fs.rename(tmpOutput, dst, function(err) {
if (err != null) {
logger.warn(
{ tmpOutput, dst },
'failed to rename output of qpdf command'
)
}
return callback(null)
})
})
} // ignore the error
}

242
services/clsi/app/js/ProjectPersistenceManager.js
View file

@ -11,113 +11,153 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let ProjectPersistenceManager;
const UrlCache = require("./UrlCache");
const CompileManager = require("./CompileManager");
const db = require("./db");
const dbQueue = require("./DbQueue");
const async = require("async");
const logger = require("logger-sharelatex");
const oneDay = 24 * 60 * 60 * 1000;
const Settings = require("settings-sharelatex");
let ProjectPersistenceManager
const UrlCache = require('./UrlCache')
const CompileManager = require('./CompileManager')
const db = require('./db')
const dbQueue = require('./DbQueue')
const async = require('async')
const logger = require('logger-sharelatex')
const oneDay = 24 * 60 * 60 * 1000
const Settings = require('settings-sharelatex')
module.exports = (ProjectPersistenceManager = {
module.exports = ProjectPersistenceManager = {
EXPIRY_TIMEOUT: Settings.project_cache_length_ms || oneDay * 2.5,
EXPIRY_TIMEOUT: Settings.project_cache_length_ms || (oneDay * 2.5),
markProjectAsJustAccessed(project_id, callback) {
if (callback == null) {
callback = function(error) {}
}
const job = cb =>
db.Project.findOrCreate({ where: { project_id } })
.spread((project, created) =>
project
.updateAttributes({ lastAccessed: new Date() })
.then(() => cb())
.error(cb)
)
.error(cb)
return dbQueue.queue.push(job, callback)
},
markProjectAsJustAccessed(project_id, callback) {
if (callback == null) { callback = function(error) {}; }
const job = cb=>
db.Project.findOrCreate({where: {project_id}})
.spread(
(project, created) =>
project.updateAttributes({lastAccessed: new Date()})
.then(() => cb())
.error(cb)
)
.error(cb)
;
return dbQueue.queue.push(job, callback);
},
clearExpiredProjects(callback) {
if (callback == null) {
callback = function(error) {}
}
return ProjectPersistenceManager._findExpiredProjectIds(function(
error,
project_ids
) {
if (error != null) {
return callback(error)
}
logger.log({ project_ids }, 'clearing expired projects')
const jobs = Array.from(project_ids || []).map(project_id =>
(project_id => callback =>
ProjectPersistenceManager.clearProjectFromCache(project_id, function(
err
) {
if (err != null) {
logger.error({ err, project_id }, 'error clearing project')
}
return callback()
}))(project_id)
)
return async.series(jobs, function(error) {
if (error != null) {
return callback(error)
}
return CompileManager.clearExpiredProjects(
ProjectPersistenceManager.EXPIRY_TIMEOUT,
error => callback()
)
})
})
}, // ignore any errors from deleting directories
clearProject(project_id, user_id, callback) {
if (callback == null) {
callback = function(error) {}
}
logger.log({ project_id, user_id }, 'clearing project for user')
return CompileManager.clearProject(project_id, user_id, function(error) {
if (error != null) {
return callback(error)
}
return ProjectPersistenceManager.clearProjectFromCache(
project_id,
function(error) {
if (error != null) {
return callback(error)
}
return callback()
}
)
})
},
clearExpiredProjects(callback) {
if (callback == null) { callback = function(error) {}; }
return ProjectPersistenceManager._findExpiredProjectIds(function(error, project_ids) {
if (error != null) { return callback(error); }
logger.log({project_ids}, "clearing expired projects");
const jobs = (Array.from(project_ids || [])).map((project_id) =>
(project_id =>
callback =>
ProjectPersistenceManager.clearProjectFromCache(project_id, function(err) {
if (err != null) {
logger.error({err, project_id}, "error clearing project");
}
return callback();
})
)(project_id));
return async.series(jobs, function(error) {
if (error != null) { return callback(error); }
return CompileManager.clearExpiredProjects(ProjectPersistenceManager.EXPIRY_TIMEOUT, error => callback());
});
});
}, // ignore any errors from deleting directories
clearProjectFromCache(project_id, callback) {
if (callback == null) {
callback = function(error) {}
}
logger.log({ project_id }, 'clearing project from cache')
return UrlCache.clearProject(project_id, function(error) {
if (error != null) {
logger.err({ error, project_id }, 'error clearing project from cache')
return callback(error)
}
return ProjectPersistenceManager._clearProjectFromDatabase(
project_id,
function(error) {
if (error != null) {
logger.err(
{ error, project_id },
'error clearing project from database'
)
}
return callback(error)
}
)
})
},
clearProject(project_id, user_id, callback) {
if (callback == null) { callback = function(error) {}; }
logger.log({project_id, user_id}, "clearing project for user");
return CompileManager.clearProject(project_id, user_id, function(error) {
if (error != null) { return callback(error); }
return ProjectPersistenceManager.clearProjectFromCache(project_id, function(error) {
if (error != null) { return callback(error); }
return callback();
});
});
},
_clearProjectFromDatabase(project_id, callback) {
if (callback == null) {
callback = function(error) {}
}
logger.log({ project_id }, 'clearing project from database')
const job = cb =>
db.Project.destroy({ where: { project_id } })
.then(() => cb())
.error(cb)
return dbQueue.queue.push(job, callback)
},
clearProjectFromCache(project_id, callback) {
if (callback == null) { callback = function(error) {}; }
logger.log({project_id}, "clearing project from cache");
return UrlCache.clearProject(project_id, function(error) {
if (error != null) {
logger.err({error, project_id}, "error clearing project from cache");
return callback(error);
}
return ProjectPersistenceManager._clearProjectFromDatabase(project_id, function(error) {
if (error != null) {
logger.err({error, project_id}, "error clearing project from database");
}
return callback(error);
});
});
},
_findExpiredProjectIds(callback) {
if (callback == null) {
callback = function(error, project_ids) {}
}
const job = function(cb) {
const keepProjectsFrom = new Date(
Date.now() - ProjectPersistenceManager.EXPIRY_TIMEOUT
)
const q = {}
q[db.op.lt] = keepProjectsFrom
return db.Project.findAll({ where: { lastAccessed: q } })
.then(projects =>
cb(
null,
projects.map(project => project.project_id)
)
)
.error(cb)
}
_clearProjectFromDatabase(project_id, callback) {
if (callback == null) { callback = function(error) {}; }
logger.log({project_id}, "clearing project from database");
const job = cb=>
db.Project.destroy({where: {project_id}})
.then(() => cb())
.error(cb)
;
return dbQueue.queue.push(job, callback);
},
return dbQueue.queue.push(job, callback)
}
}
_findExpiredProjectIds(callback) {
if (callback == null) { callback = function(error, project_ids) {}; }
const job = function(cb){
const keepProjectsFrom = new Date(Date.now() - ProjectPersistenceManager.EXPIRY_TIMEOUT);
const q = {};
q[db.op.lt] = keepProjectsFrom;
return db.Project.findAll({where:{lastAccessed:q}})
.then(projects => cb(null, projects.map(project => project.project_id))).error(cb);
};
return dbQueue.queue.push(job, callback);
}
});
logger.log({EXPIRY_TIMEOUT: ProjectPersistenceManager.EXPIRY_TIMEOUT}, "project assets kept timeout");
logger.log(
{ EXPIRY_TIMEOUT: ProjectPersistenceManager.EXPIRY_TIMEOUT },
'project assets kept timeout'
)

334
services/clsi/app/js/RequestParser.js
View file

@ -17,177 +17,201 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let RequestParser;
const settings = require("settings-sharelatex");
let RequestParser
const settings = require('settings-sharelatex')
module.exports = (RequestParser = {
VALID_COMPILERS: ["pdflatex", "latex", "xelatex", "lualatex"],
MAX_TIMEOUT: 600,
module.exports = RequestParser = {
VALID_COMPILERS: ['pdflatex', 'latex', 'xelatex', 'lualatex'],
MAX_TIMEOUT: 600,
parse(body, callback) {
let resource;
if (callback == null) { callback = function(error, data) {}; }
const response = {};
parse(body, callback) {
let resource
if (callback == null) {
callback = function(error, data) {}
}
const response = {}
if ((body.compile == null)) {
return callback("top level object should have a compile attribute");
}
if (body.compile == null) {
return callback('top level object should have a compile attribute')
}
const { compile } = body;
if (!compile.options) { compile.options = {}; }
const { compile } = body
if (!compile.options) {
compile.options = {}
}
try {
response.compiler = this._parseAttribute("compiler",
compile.options.compiler, {
validValues: this.VALID_COMPILERS,
default: "pdflatex",
type: "string"
}
);
response.timeout = this._parseAttribute("timeout",
compile.options.timeout, {
default: RequestParser.MAX_TIMEOUT,
type: "number"
}
);
response.imageName = this._parseAttribute("imageName",
compile.options.imageName,
{type: "string"});
response.draft = this._parseAttribute("draft",
compile.options.draft, {
default: false,
type: "boolean"
}
);
response.check = this._parseAttribute("check",
compile.options.check,
{type: "string"});
response.flags = this._parseAttribute("flags",
compile.options.flags, {
default: [],
type: "object"
}
);
try {
response.compiler = this._parseAttribute(
'compiler',
compile.options.compiler,
{
validValues: this.VALID_COMPILERS,
default: 'pdflatex',
type: 'string'
}
)
response.timeout = this._parseAttribute(
'timeout',
compile.options.timeout,
{
default: RequestParser.MAX_TIMEOUT,
type: 'number'
}
)
response.imageName = this._parseAttribute(
'imageName',
compile.options.imageName,
{ type: 'string' }
)
response.draft = this._parseAttribute('draft', compile.options.draft, {
default: false,
type: 'boolean'
})
response.check = this._parseAttribute('check', compile.options.check, {
type: 'string'
})
response.flags = this._parseAttribute('flags', compile.options.flags, {
default: [],
type: 'object'
})
// The syncType specifies whether the request contains all
// resources (full) or only those resources to be updated
// in-place (incremental).
response.syncType = this._parseAttribute("syncType",
compile.options.syncType, {
validValues: ["full", "incremental"],
type: "string"
}
);
// The syncType specifies whether the request contains all
// resources (full) or only those resources to be updated
// in-place (incremental).
response.syncType = this._parseAttribute(
'syncType',
compile.options.syncType,
{
validValues: ['full', 'incremental'],
type: 'string'
}
)
// The syncState is an identifier passed in with the request
// which has the property that it changes when any resource is
// added, deleted, moved or renamed.
//
// on syncType full the syncState identifier is passed in and
// stored
//
// on syncType incremental the syncState identifier must match
// the stored value
response.syncState = this._parseAttribute("syncState",
compile.options.syncState,
{type: "string"});
// The syncState is an identifier passed in with the request
// which has the property that it changes when any resource is
// added, deleted, moved or renamed.
//
// on syncType full the syncState identifier is passed in and
// stored
//
// on syncType incremental the syncState identifier must match
// the stored value
response.syncState = this._parseAttribute(
'syncState',
compile.options.syncState,
{ type: 'string' }
)
if (response.timeout > RequestParser.MAX_TIMEOUT) {
response.timeout = RequestParser.MAX_TIMEOUT;
}
response.timeout = response.timeout * 1000; // milliseconds
if (response.timeout > RequestParser.MAX_TIMEOUT) {
response.timeout = RequestParser.MAX_TIMEOUT
}
response.timeout = response.timeout * 1000 // milliseconds
response.resources = ((() => {
const result = [];
for (resource of Array.from((compile.resources || []))) { result.push(this._parseResource(resource));
}
return result;
})());
response.resources = (() => {
const result = []
for (resource of Array.from(compile.resources || [])) {
result.push(this._parseResource(resource))
}
return result
})()
const rootResourcePath = this._parseAttribute("rootResourcePath",
compile.rootResourcePath, {
default: "main.tex",
type: "string"
}
);
const originalRootResourcePath = rootResourcePath;
const sanitizedRootResourcePath = RequestParser._sanitizePath(rootResourcePath);
response.rootResourcePath = RequestParser._checkPath(sanitizedRootResourcePath);
const rootResourcePath = this._parseAttribute(
'rootResourcePath',
compile.rootResourcePath,
{
default: 'main.tex',
type: 'string'
}
)
const originalRootResourcePath = rootResourcePath
const sanitizedRootResourcePath = RequestParser._sanitizePath(
rootResourcePath
)
response.rootResourcePath = RequestParser._checkPath(
sanitizedRootResourcePath
)
for (resource of Array.from(response.resources)) {
if (resource.path === originalRootResourcePath) {
resource.path = sanitizedRootResourcePath;
}
}
} catch (error1) {
const error = error1;
return callback(error);
}
for (resource of Array.from(response.resources)) {
if (resource.path === originalRootResourcePath) {
resource.path = sanitizedRootResourcePath
}
}
} catch (error1) {
const error = error1
return callback(error)
}
return callback(null, response);
},
return callback(null, response)
},
_parseResource(resource) {
let modified;
if ((resource.path == null) || (typeof resource.path !== "string")) {
throw "all resources should have a path attribute";
}
_parseResource(resource) {
let modified
if (resource.path == null || typeof resource.path !== 'string') {
throw 'all resources should have a path attribute'
}
if (resource.modified != null) {
modified = new Date(resource.modified);
if (isNaN(modified.getTime())) {
throw `resource modified date could not be understood: ${resource.modified}`;
}
}
if (resource.modified != null) {
modified = new Date(resource.modified)
if (isNaN(modified.getTime())) {
throw `resource modified date could not be understood: ${resource.modified}`
}
}
if ((resource.url == null) && (resource.content == null)) {
throw "all resources should have either a url or content attribute";
}
if ((resource.content != null) && (typeof resource.content !== "string")) {
throw "content attribute should be a string";
}
if ((resource.url != null) && (typeof resource.url !== "string")) {
throw "url attribute should be a string";
}
if (resource.url == null && resource.content == null) {
throw 'all resources should have either a url or content attribute'
}
if (resource.content != null && typeof resource.content !== 'string') {
throw 'content attribute should be a string'
}
if (resource.url != null && typeof resource.url !== 'string') {
throw 'url attribute should be a string'
}
return {
path: resource.path,
modified,
url: resource.url,
content: resource.content
};
},
return {
path: resource.path,
modified,
url: resource.url,
content: resource.content
}
},
_parseAttribute(name, attribute, options) {
if (attribute != null) {
if (options.validValues != null) {
if (options.validValues.indexOf(attribute) === -1) {
throw `${name} attribute should be one of: ${options.validValues.join(", ")}`;
}
}
if (options.type != null) {
if (typeof attribute !== options.type) {
throw `${name} attribute should be a ${options.type}`;
}
}
} else {
if (options.default != null) { return options.default; }
}
return attribute;
},
_parseAttribute(name, attribute, options) {
if (attribute != null) {
if (options.validValues != null) {
if (options.validValues.indexOf(attribute) === -1) {
throw `${name} attribute should be one of: ${options.validValues.join(
', '
)}`
}
}
if (options.type != null) {
if (typeof attribute !== options.type) {
throw `${name} attribute should be a ${options.type}`
}
}
} else {
if (options.default != null) {
return options.default
}
}
return attribute
},
_sanitizePath(path) {
// See http://php.net/manual/en/function.escapeshellcmd.php
return path.replace(/[\#\&\;\`\|\*\?\~\<\>\^\(\)\[\]\{\}\$\\\x0A\xFF\x00]/g, "");
},
_sanitizePath(path) {
// See http://php.net/manual/en/function.escapeshellcmd.php
return path.replace(
/[\#\&\;\`\|\*\?\~\<\>\^\(\)\[\]\{\}\$\\\x0A\xFF\x00]/g,
''
)
},
_checkPath(path) {
// check that the request does not use a relative path
for (const dir of Array.from(path.split('/'))) {
if (dir === '..') {
throw "relative path in root resource";
}
}
return path;
}
});
_checkPath(path) {
// check that the request does not use a relative path
for (const dir of Array.from(path.split('/'))) {
if (dir === '..') {
throw 'relative path in root resource'
}
}
return path
}
}

224
services/clsi/app/js/ResourceStateManager.js
View file

@ -13,102 +13,142 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let ResourceStateManager;
const Path = require("path");
const fs = require("fs");
const logger = require("logger-sharelatex");
const settings = require("settings-sharelatex");
const Errors = require("./Errors");
const SafeReader = require("./SafeReader");
let ResourceStateManager
const Path = require('path')
const fs = require('fs')
const logger = require('logger-sharelatex')
const settings = require('settings-sharelatex')
const Errors = require('./Errors')
const SafeReader = require('./SafeReader')
module.exports = (ResourceStateManager = {
module.exports = ResourceStateManager = {
// The sync state is an identifier which must match for an
// incremental update to be allowed.
//
// The initial value is passed in and stored on a full
// compile, along with the list of resources..
//
// Subsequent incremental compiles must come with the same value - if
// not they will be rejected with a 409 Conflict response. The
// previous list of resources is returned.
//
// An incremental compile can only update existing files with new
// content. The sync state identifier must change if any docs or
// files are moved, added, deleted or renamed.
// The sync state is an identifier which must match for an
// incremental update to be allowed.
//
// The initial value is passed in and stored on a full
// compile, along with the list of resources..
//
// Subsequent incremental compiles must come with the same value - if
// not they will be rejected with a 409 Conflict response. The
// previous list of resources is returned.
//
// An incremental compile can only update existing files with new
// content. The sync state identifier must change if any docs or
// files are moved, added, deleted or renamed.
SYNC_STATE_FILE: '.project-sync-state',
SYNC_STATE_MAX_SIZE: 128 * 1024,
SYNC_STATE_FILE: ".project-sync-state",
SYNC_STATE_MAX_SIZE: 128*1024,
saveProjectState(state, resources, basePath, callback) {
if (callback == null) {
callback = function(error) {}
}
const stateFile = Path.join(basePath, this.SYNC_STATE_FILE)
if (state == null) {
// remove the file if no state passed in
logger.log({ state, basePath }, 'clearing sync state')
return fs.unlink(stateFile, function(err) {
if (err != null && err.code !== 'ENOENT') {
return callback(err)
} else {
return callback()
}
})
} else {
logger.log({ state, basePath }, 'writing sync state')
const resourceList = Array.from(resources).map(resource => resource.path)
return fs.writeFile(
stateFile,
[...Array.from(resourceList), `stateHash:${state}`].join('\n'),
callback
)
}
},
saveProjectState(state, resources, basePath, callback) {
if (callback == null) { callback = function(error) {}; }
const stateFile = Path.join(basePath, this.SYNC_STATE_FILE);
if ((state == null)) { // remove the file if no state passed in
logger.log({state, basePath}, "clearing sync state");
return fs.unlink(stateFile, function(err) {
if ((err != null) && (err.code !== 'ENOENT')) {
return callback(err);
} else {
return callback();
}
});
} else {
logger.log({state, basePath}, "writing sync state");
const resourceList = (Array.from(resources).map((resource) => resource.path));
return fs.writeFile(stateFile, [...Array.from(resourceList), `stateHash:${state}`].join("\n"), callback);
}
},
checkProjectStateMatches(state, basePath, callback) {
if (callback == null) {
callback = function(error, resources) {}
}
const stateFile = Path.join(basePath, this.SYNC_STATE_FILE)
const size = this.SYNC_STATE_MAX_SIZE
return SafeReader.readFile(stateFile, size, 'utf8', function(
err,
result,
bytesRead
) {
if (err != null) {
return callback(err)
}
if (bytesRead === size) {
logger.error(
{ file: stateFile, size, bytesRead },
'project state file truncated'
)
}
const array =
__guard__(result != null ? result.toString() : undefined, x =>
x.split('\n')
) || []
const adjustedLength = Math.max(array.length, 1)
const resourceList = array.slice(0, adjustedLength - 1)
const oldState = array[adjustedLength - 1]
const newState = `stateHash:${state}`
logger.log(
{ state, oldState, basePath, stateMatches: newState === oldState },
'checking sync state'
)
if (newState !== oldState) {
return callback(
new Errors.FilesOutOfSyncError('invalid state for incremental update')
)
} else {
const resources = Array.from(resourceList).map(path => ({ path }))
return callback(null, resources)
}
})
},
checkProjectStateMatches(state, basePath, callback) {
if (callback == null) { callback = function(error, resources) {}; }
const stateFile = Path.join(basePath, this.SYNC_STATE_FILE);
const size = this.SYNC_STATE_MAX_SIZE;
return SafeReader.readFile(stateFile, size, 'utf8', function(err, result, bytesRead) {
if (err != null) { return callback(err); }
if (bytesRead === size) {
logger.error({file:stateFile, size, bytesRead}, "project state file truncated");
}
const array = __guard__(result != null ? result.toString() : undefined, x => x.split("\n")) || [];
const adjustedLength = Math.max(array.length, 1);
const resourceList = array.slice(0, adjustedLength - 1);
const oldState = array[adjustedLength - 1];
const newState = `stateHash:${state}`;
logger.log({state, oldState, basePath, stateMatches: (newState === oldState)}, "checking sync state");
if (newState !== oldState) {
return callback(new Errors.FilesOutOfSyncError("invalid state for incremental update"));
} else {
const resources = (Array.from(resourceList).map((path) => ({path})));
return callback(null, resources);
}
});
},
checkResourceFiles(resources, allFiles, basePath, callback) {
// check the paths are all relative to current directory
let file;
if (callback == null) { callback = function(error) {}; }
for (file of Array.from(resources || [])) {
for (const dir of Array.from(__guard__(file != null ? file.path : undefined, x => x.split('/')))) {
if (dir === '..') {
return callback(new Error("relative path in resource file list"));
}
}
}
// check if any of the input files are not present in list of files
const seenFile = {};
for (file of Array.from(allFiles)) {
seenFile[file] = true;
}
const missingFiles = (Array.from(resources).filter((resource) => !seenFile[resource.path]).map((resource) => resource.path));
if ((missingFiles != null ? missingFiles.length : undefined) > 0) {
logger.err({missingFiles, basePath, allFiles, resources}, "missing input files for project");
return callback(new Errors.FilesOutOfSyncError("resource files missing in incremental update"));
} else {
return callback();
}
}
});
checkResourceFiles(resources, allFiles, basePath, callback) {
// check the paths are all relative to current directory
let file
if (callback == null) {
callback = function(error) {}
}
for (file of Array.from(resources || [])) {
for (const dir of Array.from(
__guard__(file != null ? file.path : undefined, x => x.split('/'))
)) {
if (dir === '..') {
return callback(new Error('relative path in resource file list'))
}
}
}
// check if any of the input files are not present in list of files
const seenFile = {}
for (file of Array.from(allFiles)) {
seenFile[file] = true
}
const missingFiles = Array.from(resources)
.filter(resource => !seenFile[resource.path])
.map(resource => resource.path)
if ((missingFiles != null ? missingFiles.length : undefined) > 0) {
logger.err(
{ missingFiles, basePath, allFiles, resources },
'missing input files for project'
)
return callback(
new Errors.FilesOutOfSyncError(
'resource files missing in incremental update'
)
)
} else {
return callback()
}
}
}
function __guard__(value, transform) {
return (typeof value !== 'undefined' && value !== null) ? transform(value) : undefined;
}
return typeof value !== 'undefined' && value !== null
? transform(value)
: undefined
}

511
services/clsi/app/js/ResourceWriter.js
View file

@ -14,202 +14,339 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let ResourceWriter;
const UrlCache = require("./UrlCache");
const Path = require("path");
const fs = require("fs");
const async = require("async");
const mkdirp = require("mkdirp");
const OutputFileFinder = require("./OutputFileFinder");
const ResourceStateManager = require("./ResourceStateManager");
const Metrics = require("./Metrics");
const logger = require("logger-sharelatex");
const settings = require("settings-sharelatex");
let ResourceWriter
const UrlCache = require('./UrlCache')
const Path = require('path')
const fs = require('fs')
const async = require('async')
const mkdirp = require('mkdirp')
const OutputFileFinder = require('./OutputFileFinder')
const ResourceStateManager = require('./ResourceStateManager')
const Metrics = require('./Metrics')
const logger = require('logger-sharelatex')
const settings = require('settings-sharelatex')
const parallelFileDownloads = settings.parallelFileDownloads || 1;
const parallelFileDownloads = settings.parallelFileDownloads || 1
module.exports = (ResourceWriter = {
module.exports = ResourceWriter = {
syncResourcesToDisk(request, basePath, callback) {
if (callback == null) {
callback = function(error, resourceList) {}
}
if (request.syncType === 'incremental') {
logger.log(
{ project_id: request.project_id, user_id: request.user_id },
'incremental sync'
)
return ResourceStateManager.checkProjectStateMatches(
request.syncState,
basePath,
function(error, resourceList) {
if (error != null) {
return callback(error)
}
return ResourceWriter._removeExtraneousFiles(
resourceList,
basePath,
function(error, outputFiles, allFiles) {
if (error != null) {
return callback(error)
}
return ResourceStateManager.checkResourceFiles(
resourceList,
allFiles,
basePath,
function(error) {
if (error != null) {
return callback(error)
}
return ResourceWriter.saveIncrementalResourcesToDisk(
request.project_id,
request.resources,
basePath,
function(error) {
if (error != null) {
return callback(error)
}
return callback(null, resourceList)
}
)
}
)
}
)
}
)
} else {
logger.log(
{ project_id: request.project_id, user_id: request.user_id },
'full sync'
)
return this.saveAllResourcesToDisk(
request.project_id,
request.resources,
basePath,
function(error) {
if (error != null) {
return callback(error)
}
return ResourceStateManager.saveProjectState(
request.syncState,
request.resources,
basePath,
function(error) {
if (error != null) {
return callback(error)
}
return callback(null, request.resources)
}
)
}
)
}
},
syncResourcesToDisk(request, basePath, callback) {
if (callback == null) { callback = function(error, resourceList) {}; }
if (request.syncType === "incremental") {
logger.log({project_id: request.project_id, user_id: request.user_id}, "incremental sync");
return ResourceStateManager.checkProjectStateMatches(request.syncState, basePath, function(error, resourceList) {
if (error != null) { return callback(error); }
return ResourceWriter._removeExtraneousFiles(resourceList, basePath, function(error, outputFiles, allFiles) {
if (error != null) { return callback(error); }
return ResourceStateManager.checkResourceFiles(resourceList, allFiles, basePath, function(error) {
if (error != null) { return callback(error); }
return ResourceWriter.saveIncrementalResourcesToDisk(request.project_id, request.resources, basePath, function(error) {
if (error != null) { return callback(error); }
return callback(null, resourceList);
});
});
});
});
} else {
logger.log({project_id: request.project_id, user_id: request.user_id}, "full sync");
return this.saveAllResourcesToDisk(request.project_id, request.resources, basePath, function(error) {
if (error != null) { return callback(error); }
return ResourceStateManager.saveProjectState(request.syncState, request.resources, basePath, function(error) {
if (error != null) { return callback(error); }
return callback(null, request.resources);
});
});
}
},
saveIncrementalResourcesToDisk(project_id, resources, basePath, callback) {
if (callback == null) {
callback = function(error) {}
}
return this._createDirectory(basePath, error => {
if (error != null) {
return callback(error)
}
const jobs = Array.from(resources).map(resource =>
(resource => {
return callback =>
this._writeResourceToDisk(project_id, resource, basePath, callback)
})(resource)
)
return async.parallelLimit(jobs, parallelFileDownloads, callback)
})
},
saveIncrementalResourcesToDisk(project_id, resources, basePath, callback) {
if (callback == null) { callback = function(error) {}; }
return this._createDirectory(basePath, error => {
if (error != null) { return callback(error); }
const jobs = Array.from(resources).map((resource) =>
(resource => {
return callback => this._writeResourceToDisk(project_id, resource, basePath, callback);
})(resource));
return async.parallelLimit(jobs, parallelFileDownloads, callback);
});
},
saveAllResourcesToDisk(project_id, resources, basePath, callback) {
if (callback == null) {
callback = function(error) {}
}
return this._createDirectory(basePath, error => {
if (error != null) {
return callback(error)
}
return this._removeExtraneousFiles(resources, basePath, error => {
if (error != null) {
return callback(error)
}
const jobs = Array.from(resources).map(resource =>
(resource => {
return callback =>
this._writeResourceToDisk(
project_id,
resource,
basePath,
callback
)
})(resource)
)
return async.parallelLimit(jobs, parallelFileDownloads, callback)
})
})
},
saveAllResourcesToDisk(project_id, resources, basePath, callback) {
if (callback == null) { callback = function(error) {}; }
return this._createDirectory(basePath, error => {
if (error != null) { return callback(error); }
return this._removeExtraneousFiles(resources, basePath, error => {
if (error != null) { return callback(error); }
const jobs = Array.from(resources).map((resource) =>
(resource => {
return callback => this._writeResourceToDisk(project_id, resource, basePath, callback);
})(resource));
return async.parallelLimit(jobs, parallelFileDownloads, callback);
});
});
},
_createDirectory(basePath, callback) {
if (callback == null) {
callback = function(error) {}
}
return fs.mkdir(basePath, function(err) {
if (err != null) {
if (err.code === 'EEXIST') {
return callback()
} else {
logger.log({ err, dir: basePath }, 'error creating directory')
return callback(err)
}
} else {
return callback()
}
})
},
_createDirectory(basePath, callback) {
if (callback == null) { callback = function(error) {}; }
return fs.mkdir(basePath, function(err) {
if (err != null) {
if (err.code === 'EEXIST') {
return callback();
} else {
logger.log({err, dir:basePath}, "error creating directory");
return callback(err);
}
} else {
return callback();
}
});
},
_removeExtraneousFiles(resources, basePath, _callback) {
if (_callback == null) {
_callback = function(error, outputFiles, allFiles) {}
}
const timer = new Metrics.Timer('unlink-output-files')
const callback = function(error, ...result) {
timer.done()
return _callback(error, ...Array.from(result))
}
_removeExtraneousFiles(resources, basePath, _callback) {
if (_callback == null) { _callback = function(error, outputFiles, allFiles) {}; }
const timer = new Metrics.Timer("unlink-output-files");
const callback = function(error, ...result) {
timer.done();
return _callback(error, ...Array.from(result));
};
return OutputFileFinder.findOutputFiles(resources, basePath, function(
error,
outputFiles,
allFiles
) {
if (error != null) {
return callback(error)
}
return OutputFileFinder.findOutputFiles(resources, basePath, function(error, outputFiles, allFiles) {
if (error != null) { return callback(error); }
const jobs = []
for (const file of Array.from(outputFiles || [])) {
;(function(file) {
const { path } = file
let should_delete = true
if (
path.match(/^output\./) ||
path.match(/\.aux$/) ||
path.match(/^cache\//)
) {
// knitr cache
should_delete = false
}
if (path.match(/^output-.*/)) {
// Tikz cached figures (default case)
should_delete = false
}
if (path.match(/\.(pdf|dpth|md5)$/)) {
// Tikz cached figures (by extension)
should_delete = false
}
if (
path.match(/\.(pygtex|pygstyle)$/) ||
path.match(/(^|\/)_minted-[^\/]+\//)
) {
// minted files/directory
should_delete = false
}
if (
path.match(/\.md\.tex$/) ||
path.match(/(^|\/)_markdown_[^\/]+\//)
) {
// markdown files/directory
should_delete = false
}
if (path.match(/-eps-converted-to\.pdf$/)) {
// Epstopdf generated files
should_delete = false
}
if (
path === 'output.pdf' ||
path === 'output.dvi' ||
path === 'output.log' ||
path === 'output.xdv'
) {
should_delete = true
}
if (path === 'output.tex') {
// created by TikzManager if present in output files
should_delete = true
}
if (should_delete) {
return jobs.push(callback =>
ResourceWriter._deleteFileIfNotDirectory(
Path.join(basePath, path),
callback
)
)
}
})(file)
}
const jobs = [];
for (const file of Array.from(outputFiles || [])) {
(function(file) {
const { path } = file;
let should_delete = true;
if (path.match(/^output\./) || path.match(/\.aux$/) || path.match(/^cache\//)) { // knitr cache
should_delete = false;
}
if (path.match(/^output-.*/)) { // Tikz cached figures (default case)
should_delete = false;
}
if (path.match(/\.(pdf|dpth|md5)$/)) { // Tikz cached figures (by extension)
should_delete = false;
}
if (path.match(/\.(pygtex|pygstyle)$/) || path.match(/(^|\/)_minted-[^\/]+\//)) { // minted files/directory
should_delete = false;
}
if (path.match(/\.md\.tex$/) || path.match(/(^|\/)_markdown_[^\/]+\//)) { // markdown files/directory
should_delete = false;
}
if (path.match(/-eps-converted-to\.pdf$/)) { // Epstopdf generated files
should_delete = false;
}
if ((path === "output.pdf") || (path === "output.dvi") || (path === "output.log") || (path === "output.xdv")) {
should_delete = true;
}
if (path === "output.tex") { // created by TikzManager if present in output files
should_delete = true;
}
if (should_delete) {
return jobs.push(callback => ResourceWriter._deleteFileIfNotDirectory(Path.join(basePath, path), callback));
}
})(file);
}
return async.series(jobs, function(error) {
if (error != null) {
return callback(error)
}
return callback(null, outputFiles, allFiles)
})
})
},
return async.series(jobs, function(error) {
if (error != null) { return callback(error); }
return callback(null, outputFiles, allFiles);
});
});
},
_deleteFileIfNotDirectory(path, callback) {
if (callback == null) {
callback = function(error) {}
}
return fs.stat(path, function(error, stat) {
if (error != null && error.code === 'ENOENT') {
return callback()
} else if (error != null) {
logger.err(
{ err: error, path },
'error stating file in deleteFileIfNotDirectory'
)
return callback(error)
} else if (stat.isFile()) {
return fs.unlink(path, function(error) {
if (error != null) {
logger.err(
{ err: error, path },
'error removing file in deleteFileIfNotDirectory'
)
return callback(error)
} else {
return callback()
}
})
} else {
return callback()
}
})
},
_deleteFileIfNotDirectory(path, callback) {
if (callback == null) { callback = function(error) {}; }
return fs.stat(path, function(error, stat) {
if ((error != null) && (error.code === 'ENOENT')) {
return callback();
} else if (error != null) {
logger.err({err: error, path}, "error stating file in deleteFileIfNotDirectory");
return callback(error);
} else if (stat.isFile()) {
return fs.unlink(path, function(error) {
if (error != null) {
logger.err({err: error, path}, "error removing file in deleteFileIfNotDirectory");
return callback(error);
} else {
return callback();
}
});
} else {
return callback();
}
});
},
_writeResourceToDisk(project_id, resource, basePath, callback) {
if (callback == null) {
callback = function(error) {}
}
return ResourceWriter.checkPath(basePath, resource.path, function(
error,
path
) {
if (error != null) {
return callback(error)
}
return mkdirp(Path.dirname(path), function(error) {
if (error != null) {
return callback(error)
}
// TODO: Don't overwrite file if it hasn't been modified
if (resource.url != null) {
return UrlCache.downloadUrlToFile(
project_id,
resource.url,
path,
resource.modified,
function(err) {
if (err != null) {
logger.err(
{
err,
project_id,
path,
resource_url: resource.url,
modified: resource.modified
},
'error downloading file for resources'
)
}
return callback()
}
) // try and continue compiling even if http resource can not be downloaded at this time
} else {
const process = require('process')
fs.writeFile(path, resource.content, callback)
try {
let result
return (result = fs.lstatSync(path))
} catch (e) {}
}
})
})
},
_writeResourceToDisk(project_id, resource, basePath, callback) {
if (callback == null) { callback = function(error) {}; }
return ResourceWriter.checkPath(basePath, resource.path, function(error, path) {
if (error != null) { return callback(error); }
return mkdirp(Path.dirname(path), function(error) {
if (error != null) { return callback(error); }
// TODO: Don't overwrite file if it hasn't been modified
if (resource.url != null) {
return UrlCache.downloadUrlToFile(project_id, resource.url, path, resource.modified, function(err){
if (err != null) {
logger.err({err, project_id, path, resource_url:resource.url, modified:resource.modified}, "error downloading file for resources");
}
return callback();
}); // try and continue compiling even if http resource can not be downloaded at this time
} else {
const process = require("process");
fs.writeFile(path, resource.content, callback);
try {
let result;
return result = fs.lstatSync(path);
} catch (e) {}
}
});
});
},
checkPath(basePath, resourcePath, callback) {
const path = Path.normalize(Path.join(basePath, resourcePath));
if (path.slice(0, basePath.length + 1) !== (basePath + "/")) {
return callback(new Error("resource path is outside root directory"));
} else {
return callback(null, path);
}
}
});
checkPath(basePath, resourcePath, callback) {
const path = Path.normalize(Path.join(basePath, resourcePath))
if (path.slice(0, basePath.length + 1) !== basePath + '/') {
return callback(new Error('resource path is outside root directory'))
} else {
return callback(null, path)
}
}
}

73
services/clsi/app/js/SafeReader.js
View file

@ -12,36 +12,49 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let SafeReader;
const fs = require("fs");
const logger = require("logger-sharelatex");
let SafeReader
const fs = require('fs')
const logger = require('logger-sharelatex')
module.exports = (SafeReader = {
module.exports = SafeReader = {
// safely read up to size bytes from a file and return result as a
// string
// safely read up to size bytes from a file and return result as a
// string
readFile(file, size, encoding, callback) {
if (callback == null) {
callback = function(error, result) {}
}
return fs.open(file, 'r', function(err, fd) {
if (err != null && err.code === 'ENOENT') {
return callback()
}
if (err != null) {
return callback(err)
}
readFile(file, size, encoding, callback) {
if (callback == null) { callback = function(error, result) {}; }
return fs.open(file, 'r', function(err, fd) {
if ((err != null) && (err.code === 'ENOENT')) { return callback(); }
if (err != null) { return callback(err); }
// safely return always closing the file
const callbackWithClose = (err, ...result) =>
fs.close(fd, function(err1) {
if (err != null) { return callback(err); }
if (err1 != null) { return callback(err1); }
return callback(null, ...Array.from(result));
})
;
const buff = new Buffer(size, 0); // fill with zeros
return fs.read(fd, buff, 0, buff.length, 0, function(err, bytesRead, buffer) {
if (err != null) { return callbackWithClose(err); }
const result = buffer.toString(encoding, 0, bytesRead);
return callbackWithClose(null, result, bytesRead);
});
});
}
});
// safely return always closing the file
const callbackWithClose = (err, ...result) =>
fs.close(fd, function(err1) {
if (err != null) {
return callback(err)
}
if (err1 != null) {
return callback(err1)
}
return callback(null, ...Array.from(result))
})
const buff = new Buffer(size, 0) // fill with zeros
return fs.read(fd, buff, 0, buff.length, 0, function(
err,
bytesRead,
buffer
) {
if (err != null) {
return callbackWithClose(err)
}
const result = buffer.toString(encoding, 0, bytesRead)
return callbackWithClose(null, result, bytesRead)
})
})
}
}

128
services/clsi/app/js/StaticServerForbidSymlinks.js
View file

@ -14,59 +14,81 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let ForbidSymlinks;
const Path = require("path");
const fs = require("fs");
const Settings = require("settings-sharelatex");
const logger = require("logger-sharelatex");
const url = require("url");
let ForbidSymlinks
const Path = require('path')
const fs = require('fs')
const Settings = require('settings-sharelatex')
const logger = require('logger-sharelatex')
const url = require('url')
module.exports = (ForbidSymlinks = function(staticFn, root, options) {
const expressStatic = staticFn(root, options);
const basePath = Path.resolve(root);
return function(req, res, next) {
let file, project_id, result;
const path = __guard__(url.parse(req.url), x => x.pathname);
// check that the path is of the form /project_id_or_name/path/to/file.log
if (result = path.match(/^\/?([a-zA-Z0-9_-]+)\/(.*)/)) {
project_id = result[1];
file = result[2];
} else {
logger.warn({path}, "unrecognized file request");
return res.sendStatus(404);
}
// check that the file does not use a relative path
for (const dir of Array.from(file.split('/'))) {
if (dir === '..') {
logger.warn({path}, "attempt to use a relative path");
return res.sendStatus(404);
}
}
// check that the requested path is normalized
const requestedFsPath = `${basePath}/${project_id}/${file}`;
if (requestedFsPath !== Path.normalize(requestedFsPath)) {
logger.error({path: requestedFsPath}, "requestedFsPath is not normalized");
return res.sendStatus(404);
}
// check that the requested path is not a symlink
return fs.realpath(requestedFsPath, function(err, realFsPath){
if (err != null) {
if (err.code === 'ENOENT') {
return res.sendStatus(404);
} else {
logger.error({err, requestedFsPath, realFsPath, path: req.params[0], project_id: req.params.project_id}, "error checking file access");
return res.sendStatus(500);
}
} else if (requestedFsPath !== realFsPath) {
logger.warn({requestedFsPath, realFsPath, path: req.params[0], project_id: req.params.project_id}, "trying to access a different file (symlink), aborting");
return res.sendStatus(404);
} else {
return expressStatic(req, res, next);
}
});
};
});
module.exports = ForbidSymlinks = function(staticFn, root, options) {
const expressStatic = staticFn(root, options)
const basePath = Path.resolve(root)
return function(req, res, next) {
let file, project_id, result
const path = __guard__(url.parse(req.url), x => x.pathname)
// check that the path is of the form /project_id_or_name/path/to/file.log
if ((result = path.match(/^\/?([a-zA-Z0-9_-]+)\/(.*)/))) {
project_id = result[1]
file = result[2]
} else {
logger.warn({ path }, 'unrecognized file request')
return res.sendStatus(404)
}
// check that the file does not use a relative path
for (const dir of Array.from(file.split('/'))) {
if (dir === '..') {
logger.warn({ path }, 'attempt to use a relative path')
return res.sendStatus(404)
}
}
// check that the requested path is normalized
const requestedFsPath = `${basePath}/${project_id}/${file}`
if (requestedFsPath !== Path.normalize(requestedFsPath)) {
logger.error(
{ path: requestedFsPath },
'requestedFsPath is not normalized'
)
return res.sendStatus(404)
}
// check that the requested path is not a symlink
return fs.realpath(requestedFsPath, function(err, realFsPath) {
if (err != null) {
if (err.code === 'ENOENT') {
return res.sendStatus(404)
} else {
logger.error(
{
err,
requestedFsPath,
realFsPath,
path: req.params[0],
project_id: req.params.project_id
},
'error checking file access'
)
return res.sendStatus(500)
}
} else if (requestedFsPath !== realFsPath) {
logger.warn(
{
requestedFsPath,
realFsPath,
path: req.params[0],
project_id: req.params.project_id
},
'trying to access a different file (symlink), aborting'
)
return res.sendStatus(404)
} else {
return expressStatic(req, res, next)
}
})
}
}
function __guard__(value, transform) {
return (typeof value !== 'undefined' && value !== null) ? transform(value) : undefined;
}
return typeof value !== 'undefined' && value !== null
? transform(value)
: undefined
}

118
services/clsi/app/js/TikzManager.js
View file

@ -11,52 +11,84 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let TikzManager;
const fs = require("fs");
const Path = require("path");
const ResourceWriter = require("./ResourceWriter");
const SafeReader = require("./SafeReader");
const logger = require("logger-sharelatex");
let TikzManager
const fs = require('fs')
const Path = require('path')
const ResourceWriter = require('./ResourceWriter')
const SafeReader = require('./SafeReader')
const logger = require('logger-sharelatex')
// for \tikzexternalize or pstool to work the main file needs to match the
// jobname. Since we set the -jobname to output, we have to create a
// copy of the main file as 'output.tex'.
module.exports = (TikzManager = {
module.exports = TikzManager = {
checkMainFile(compileDir, mainFile, resources, callback) {
// if there's already an output.tex file, we don't want to touch it
if (callback == null) {
callback = function(error, needsMainFile) {}
}
for (const resource of Array.from(resources)) {
if (resource.path === 'output.tex') {
logger.log({ compileDir, mainFile }, 'output.tex already in resources')
return callback(null, false)
}
}
// if there's no output.tex, see if we are using tikz/pgf or pstool in the main file
return ResourceWriter.checkPath(compileDir, mainFile, function(
error,
path
) {
if (error != null) {
return callback(error)
}
return SafeReader.readFile(path, 65536, 'utf8', function(error, content) {
if (error != null) {
return callback(error)
}
const usesTikzExternalize =
(content != null
? content.indexOf('\\tikzexternalize')
: undefined) >= 0
const usesPsTool =
(content != null ? content.indexOf('{pstool}') : undefined) >= 0
logger.log(
{ compileDir, mainFile, usesTikzExternalize, usesPsTool },
'checked for packages needing main file as output.tex'
)
const needsMainFile = usesTikzExternalize || usesPsTool
return callback(null, needsMainFile)
})
})
},
checkMainFile(compileDir, mainFile, resources, callback) {
// if there's already an output.tex file, we don't want to touch it
if (callback == null) { callback = function(error, needsMainFile) {}; }
for (const resource of Array.from(resources)) {
if (resource.path === "output.tex") {
logger.log({compileDir, mainFile}, "output.tex already in resources");
return callback(null, false);
}
}
// if there's no output.tex, see if we are using tikz/pgf or pstool in the main file
return ResourceWriter.checkPath(compileDir, mainFile, function(error, path) {
if (error != null) { return callback(error); }
return SafeReader.readFile(path, 65536, "utf8", function(error, content) {
if (error != null) { return callback(error); }
const usesTikzExternalize = (content != null ? content.indexOf("\\tikzexternalize") : undefined) >= 0;
const usesPsTool = (content != null ? content.indexOf("{pstool}") : undefined) >= 0;
logger.log({compileDir, mainFile, usesTikzExternalize, usesPsTool}, "checked for packages needing main file as output.tex");
const needsMainFile = (usesTikzExternalize || usesPsTool);
return callback(null, needsMainFile);
});
});
},
injectOutputFile(compileDir, mainFile, callback) {
if (callback == null) { callback = function(error) {}; }
return ResourceWriter.checkPath(compileDir, mainFile, function(error, path) {
if (error != null) { return callback(error); }
return fs.readFile(path, "utf8", function(error, content) {
if (error != null) { return callback(error); }
logger.log({compileDir, mainFile}, "copied file to output.tex as project uses packages which require it");
// use wx flag to ensure that output file does not already exist
return fs.writeFile(Path.join(compileDir, "output.tex"), content, {flag:'wx'}, callback);
});
});
}
});
injectOutputFile(compileDir, mainFile, callback) {
if (callback == null) {
callback = function(error) {}
}
return ResourceWriter.checkPath(compileDir, mainFile, function(
error,
path
) {
if (error != null) {
return callback(error)
}
return fs.readFile(path, 'utf8', function(error, content) {
if (error != null) {
return callback(error)
}
logger.log(
{ compileDir, mainFile },
'copied file to output.tex as project uses packages which require it'
)
// use wx flag to ensure that output file does not already exist
return fs.writeFile(
Path.join(compileDir, 'output.tex'),
content,
{ flag: 'wx' },
callback
)
})
})
}
}

420
services/clsi/app/js/UrlCache.js
View file

@ -12,185 +12,267 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let UrlCache;
const db = require("./db");
const dbQueue = require("./DbQueue");
const UrlFetcher = require("./UrlFetcher");
const Settings = require("settings-sharelatex");
const crypto = require("crypto");
const fs = require("fs");
const logger = require("logger-sharelatex");
const async = require("async");
let UrlCache
const db = require('./db')
const dbQueue = require('./DbQueue')
const UrlFetcher = require('./UrlFetcher')
const Settings = require('settings-sharelatex')
const crypto = require('crypto')
const fs = require('fs')
const logger = require('logger-sharelatex')
const async = require('async')
module.exports = (UrlCache = {
downloadUrlToFile(project_id, url, destPath, lastModified, callback) {
if (callback == null) { callback = function(error) {}; }
return UrlCache._ensureUrlIsInCache(project_id, url, lastModified, (error, pathToCachedUrl) => {
if (error != null) { return callback(error); }
return UrlCache._copyFile(pathToCachedUrl, destPath, function(error) {
if (error != null) {
return UrlCache._clearUrlDetails(project_id, url, () => callback(error));
} else {
return callback(error);
}
});
});
},
module.exports = UrlCache = {
downloadUrlToFile(project_id, url, destPath, lastModified, callback) {
if (callback == null) {
callback = function(error) {}
}
return UrlCache._ensureUrlIsInCache(
project_id,
url,
lastModified,
(error, pathToCachedUrl) => {
if (error != null) {
return callback(error)
}
return UrlCache._copyFile(pathToCachedUrl, destPath, function(error) {
if (error != null) {
return UrlCache._clearUrlDetails(project_id, url, () =>
callback(error)
)
} else {
return callback(error)
}
})
}
)
},
clearProject(project_id, callback) {
if (callback == null) { callback = function(error) {}; }
return UrlCache._findAllUrlsInProject(project_id, function(error, urls) {
logger.log({project_id, url_count: urls.length}, "clearing project URLs");
if (error != null) { return callback(error); }
const jobs = (Array.from(urls || [])).map((url) =>
(url =>
callback =>
UrlCache._clearUrlFromCache(project_id, url, function(error) {
if (error != null) {
logger.error({err: error, project_id, url}, "error clearing project URL");
}
return callback();
})
)(url));
return async.series(jobs, callback);
});
},
clearProject(project_id, callback) {
if (callback == null) {
callback = function(error) {}
}
return UrlCache._findAllUrlsInProject(project_id, function(error, urls) {
logger.log(
{ project_id, url_count: urls.length },
'clearing project URLs'
)
if (error != null) {
return callback(error)
}
const jobs = Array.from(urls || []).map(url =>
(url => callback =>
UrlCache._clearUrlFromCache(project_id, url, function(error) {
if (error != null) {
logger.error(
{ err: error, project_id, url },
'error clearing project URL'
)
}
return callback()
}))(url)
)
return async.series(jobs, callback)
})
},
_ensureUrlIsInCache(project_id, url, lastModified, callback) {
if (callback == null) { callback = function(error, pathOnDisk) {}; }
if (lastModified != null) {
// MYSQL only stores dates to an accuracy of a second but the incoming lastModified might have milliseconds.
// So round down to seconds
lastModified = new Date(Math.floor(lastModified.getTime() / 1000) * 1000);
}
return UrlCache._doesUrlNeedDownloading(project_id, url, lastModified, (error, needsDownloading) => {
if (error != null) { return callback(error); }
if (needsDownloading) {
logger.log({url, lastModified}, "downloading URL");
return UrlFetcher.pipeUrlToFile(url, UrlCache._cacheFilePathForUrl(project_id, url), error => {
if (error != null) { return callback(error); }
return UrlCache._updateOrCreateUrlDetails(project_id, url, lastModified, error => {
if (error != null) { return callback(error); }
return callback(null, UrlCache._cacheFilePathForUrl(project_id, url));
});
});
} else {
logger.log({url, lastModified}, "URL is up to date in cache");
return callback(null, UrlCache._cacheFilePathForUrl(project_id, url));
}
});
},
_doesUrlNeedDownloading(project_id, url, lastModified, callback) {
if (callback == null) { callback = function(error, needsDownloading) {}; }
if ((lastModified == null)) {
return callback(null, true);
}
return UrlCache._findUrlDetails(project_id, url, function(error, urlDetails) {
if (error != null) { return callback(error); }
if ((urlDetails == null) || (urlDetails.lastModified == null) || (urlDetails.lastModified.getTime() < lastModified.getTime())) {
return callback(null, true);
} else {
return callback(null, false);
}
});
},
_ensureUrlIsInCache(project_id, url, lastModified, callback) {
if (callback == null) {
callback = function(error, pathOnDisk) {}
}
if (lastModified != null) {
// MYSQL only stores dates to an accuracy of a second but the incoming lastModified might have milliseconds.
// So round down to seconds
lastModified = new Date(Math.floor(lastModified.getTime() / 1000) * 1000)
}
return UrlCache._doesUrlNeedDownloading(
project_id,
url,
lastModified,
(error, needsDownloading) => {
if (error != null) {
return callback(error)
}
if (needsDownloading) {
logger.log({ url, lastModified }, 'downloading URL')
return UrlFetcher.pipeUrlToFile(
url,
UrlCache._cacheFilePathForUrl(project_id, url),
error => {
if (error != null) {
return callback(error)
}
return UrlCache._updateOrCreateUrlDetails(
project_id,
url,
lastModified,
error => {
if (error != null) {
return callback(error)
}
return callback(
null,
UrlCache._cacheFilePathForUrl(project_id, url)
)
}
)
}
)
} else {
logger.log({ url, lastModified }, 'URL is up to date in cache')
return callback(null, UrlCache._cacheFilePathForUrl(project_id, url))
}
}
)
},
_cacheFileNameForUrl(project_id, url) {
return project_id + ":" + crypto.createHash("md5").update(url).digest("hex");
},
_doesUrlNeedDownloading(project_id, url, lastModified, callback) {
if (callback == null) {
callback = function(error, needsDownloading) {}
}
if (lastModified == null) {
return callback(null, true)
}
return UrlCache._findUrlDetails(project_id, url, function(
error,
urlDetails
) {
if (error != null) {
return callback(error)
}
if (
urlDetails == null ||
urlDetails.lastModified == null ||
urlDetails.lastModified.getTime() < lastModified.getTime()
) {
return callback(null, true)
} else {
return callback(null, false)
}
})
},
_cacheFilePathForUrl(project_id, url) {
return `${Settings.path.clsiCacheDir}/${UrlCache._cacheFileNameForUrl(project_id, url)}`;
},
_cacheFileNameForUrl(project_id, url) {
return (
project_id +
':' +
crypto
.createHash('md5')
.update(url)
.digest('hex')
)
},
_copyFile(from, to, _callback) {
if (_callback == null) { _callback = function(error) {}; }
const callbackOnce = function(error) {
if (error != null) {
logger.error({err: error, from, to}, "error copying file from cache");
}
_callback(error);
return _callback = function() {};
};
const writeStream = fs.createWriteStream(to);
const readStream = fs.createReadStream(from);
writeStream.on("error", callbackOnce);
readStream.on("error", callbackOnce);
writeStream.on("close", callbackOnce);
return writeStream.on("open", () => readStream.pipe(writeStream));
},
_cacheFilePathForUrl(project_id, url) {
return `${Settings.path.clsiCacheDir}/${UrlCache._cacheFileNameForUrl(
project_id,
url
)}`
},
_clearUrlFromCache(project_id, url, callback) {
if (callback == null) { callback = function(error) {}; }
return UrlCache._clearUrlDetails(project_id, url, function(error) {
if (error != null) { return callback(error); }
return UrlCache._deleteUrlCacheFromDisk(project_id, url, function(error) {
if (error != null) { return callback(error); }
return callback(null);
});
});
},
_copyFile(from, to, _callback) {
if (_callback == null) {
_callback = function(error) {}
}
const callbackOnce = function(error) {
if (error != null) {
logger.error({ err: error, from, to }, 'error copying file from cache')
}
_callback(error)
return (_callback = function() {})
}
const writeStream = fs.createWriteStream(to)
const readStream = fs.createReadStream(from)
writeStream.on('error', callbackOnce)
readStream.on('error', callbackOnce)
writeStream.on('close', callbackOnce)
return writeStream.on('open', () => readStream.pipe(writeStream))
},
_deleteUrlCacheFromDisk(project_id, url, callback) {
if (callback == null) { callback = function(error) {}; }
return fs.unlink(UrlCache._cacheFilePathForUrl(project_id, url), function(error) {
if ((error != null) && (error.code !== 'ENOENT')) { // no error if the file isn't present
return callback(error);
} else {
return callback();
}
});
},
_clearUrlFromCache(project_id, url, callback) {
if (callback == null) {
callback = function(error) {}
}
return UrlCache._clearUrlDetails(project_id, url, function(error) {
if (error != null) {
return callback(error)
}
return UrlCache._deleteUrlCacheFromDisk(project_id, url, function(error) {
if (error != null) {
return callback(error)
}
return callback(null)
})
})
},
_findUrlDetails(project_id, url, callback) {
if (callback == null) { callback = function(error, urlDetails) {}; }
const job = cb=>
db.UrlCache.find({where: { url, project_id }})
.then(urlDetails => cb(null, urlDetails))
.error(cb)
;
return dbQueue.queue.push(job, callback);
},
_deleteUrlCacheFromDisk(project_id, url, callback) {
if (callback == null) {
callback = function(error) {}
}
return fs.unlink(UrlCache._cacheFilePathForUrl(project_id, url), function(
error
) {
if (error != null && error.code !== 'ENOENT') {
// no error if the file isn't present
return callback(error)
} else {
return callback()
}
})
},
_updateOrCreateUrlDetails(project_id, url, lastModified, callback) {
if (callback == null) { callback = function(error) {}; }
const job = cb=>
db.UrlCache.findOrCreate({where: {url, project_id}})
.spread(
(urlDetails, created) =>
urlDetails.updateAttributes({lastModified})
.then(() => cb())
.error(cb)
)
.error(cb)
;
return dbQueue.queue.push(job, callback);
},
_findUrlDetails(project_id, url, callback) {
if (callback == null) {
callback = function(error, urlDetails) {}
}
const job = cb =>
db.UrlCache.find({ where: { url, project_id } })
.then(urlDetails => cb(null, urlDetails))
.error(cb)
return dbQueue.queue.push(job, callback)
},
_clearUrlDetails(project_id, url, callback) {
if (callback == null) { callback = function(error) {}; }
const job = cb=>
db.UrlCache.destroy({where: {url, project_id}})
.then(() => cb(null))
.error(cb)
;
return dbQueue.queue.push(job, callback);
},
_updateOrCreateUrlDetails(project_id, url, lastModified, callback) {
if (callback == null) {
callback = function(error) {}
}
const job = cb =>
db.UrlCache.findOrCreate({ where: { url, project_id } })
.spread((urlDetails, created) =>
urlDetails
.updateAttributes({ lastModified })
.then(() => cb())
.error(cb)
)
.error(cb)
return dbQueue.queue.push(job, callback)
},
_clearUrlDetails(project_id, url, callback) {
if (callback == null) {
callback = function(error) {}
}
const job = cb =>
db.UrlCache.destroy({ where: { url, project_id } })
.then(() => cb(null))
.error(cb)
return dbQueue.queue.push(job, callback)
},
_findAllUrlsInProject(project_id, callback) {
if (callback == null) { callback = function(error, urls) {}; }
const job = cb=>
db.UrlCache.findAll({where: { project_id }})
.then(
urlEntries => cb(null, urlEntries.map(entry => entry.url)))
.error(cb)
;
return dbQueue.queue.push(job, callback);
}
});
_findAllUrlsInProject(project_id, callback) {
if (callback == null) {
callback = function(error, urls) {}
}
const job = cb =>
db.UrlCache.findAll({ where: { project_id } })
.then(urlEntries =>
cb(
null,
urlEntries.map(entry => entry.url)
)
)
.error(cb)
return dbQueue.queue.push(job, callback)
}
}

166
services/clsi/app/js/UrlFetcher.js
View file

@ -12,85 +12,109 @@
* DS207: Consider shorter variations of null checks
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
let UrlFetcher;
const request = require("request").defaults({jar: false});
const fs = require("fs");
const logger = require("logger-sharelatex");
const settings = require("settings-sharelatex");
const URL = require('url');
let UrlFetcher
const request = require('request').defaults({ jar: false })
const fs = require('fs')
const logger = require('logger-sharelatex')
const settings = require('settings-sharelatex')
const URL = require('url')
const oneMinute = 60 * 1000;
const oneMinute = 60 * 1000
module.exports = (UrlFetcher = {
pipeUrlToFile(url, filePath, _callback) {
if (_callback == null) { _callback = function(error) {}; }
const callbackOnce = function(error) {
if (timeoutHandler != null) { clearTimeout(timeoutHandler); }
_callback(error);
return _callback = function() {};
};
module.exports = UrlFetcher = {
pipeUrlToFile(url, filePath, _callback) {
if (_callback == null) {
_callback = function(error) {}
}
const callbackOnce = function(error) {
if (timeoutHandler != null) {
clearTimeout(timeoutHandler)
}
_callback(error)
return (_callback = function() {})
}
if (settings.filestoreDomainOveride != null) {
const p = URL.parse(url).path;
url = `${settings.filestoreDomainOveride}${p}`;
}
var timeoutHandler = setTimeout(function() {
timeoutHandler = null;
logger.error({url, filePath}, "Timed out downloading file to cache");
return callbackOnce(new Error(`Timed out downloading file to cache ${url}`));
}
// FIXME: maybe need to close fileStream here
, 3 * oneMinute);
if (settings.filestoreDomainOveride != null) {
const p = URL.parse(url).path
url = `${settings.filestoreDomainOveride}${p}`
}
var timeoutHandler = setTimeout(
function() {
timeoutHandler = null
logger.error({ url, filePath }, 'Timed out downloading file to cache')
return callbackOnce(
new Error(`Timed out downloading file to cache ${url}`)
)
},
// FIXME: maybe need to close fileStream here
3 * oneMinute
)
logger.log({url, filePath}, "started downloading url to cache");
const urlStream = request.get({url, timeout: oneMinute});
urlStream.pause(); // stop data flowing until we are ready
logger.log({ url, filePath }, 'started downloading url to cache')
const urlStream = request.get({ url, timeout: oneMinute })
urlStream.pause() // stop data flowing until we are ready
// attach handlers before setting up pipes
urlStream.on("error", function(error) {
logger.error({err: error, url, filePath}, "error downloading url");
return callbackOnce(error || new Error(`Something went wrong downloading the URL ${url}`));
});
// attach handlers before setting up pipes
urlStream.on('error', function(error) {
logger.error({ err: error, url, filePath }, 'error downloading url')
return callbackOnce(
error || new Error(`Something went wrong downloading the URL ${url}`)
)
})
urlStream.on("end", () => logger.log({url, filePath}, "finished downloading file into cache"));
urlStream.on('end', () =>
logger.log({ url, filePath }, 'finished downloading file into cache')
)
return urlStream.on("response", function(res) {
if ((res.statusCode >= 200) && (res.statusCode < 300)) {
const fileStream = fs.createWriteStream(filePath);
return urlStream.on('response', function(res) {
if (res.statusCode >= 200 && res.statusCode < 300) {
const fileStream = fs.createWriteStream(filePath)
// attach handlers before setting up pipes
fileStream.on('error', function(error) {
logger.error({err: error, url, filePath}, "error writing file into cache");
return fs.unlink(filePath, function(err) {
if (err != null) {
logger.err({err, filePath}, "error deleting file from cache");
}
return callbackOnce(error);
});
});
// attach handlers before setting up pipes
fileStream.on('error', function(error) {
logger.error(
{ err: error, url, filePath },
'error writing file into cache'
)
return fs.unlink(filePath, function(err) {
if (err != null) {
logger.err({ err, filePath }, 'error deleting file from cache')
}
return callbackOnce(error)
})
})
fileStream.on('finish', function() {
logger.log({url, filePath}, "finished writing file into cache");
return callbackOnce();
});
fileStream.on('finish', function() {
logger.log({ url, filePath }, 'finished writing file into cache')
return callbackOnce()
})
fileStream.on('pipe', () => logger.log({url, filePath}, "piping into filestream"));
fileStream.on('pipe', () =>
logger.log({ url, filePath }, 'piping into filestream')
)
urlStream.pipe(fileStream);
return urlStream.resume(); // now we are ready to handle the data
} else {
logger.error({statusCode: res.statusCode, url, filePath}, "unexpected status code downloading url to cache");
// https://nodejs.org/api/http.html#http_class_http_clientrequest
// If you add a 'response' event handler, then you must consume
// the data from the response object, either by calling
// response.read() whenever there is a 'readable' event, or by
// adding a 'data' handler, or by calling the .resume()
// method. Until the data is consumed, the 'end' event will not
// fire. Also, until the data is read it will consume memory
// that can eventually lead to a 'process out of memory' error.
urlStream.resume(); // discard the data
return callbackOnce(new Error(`URL returned non-success status code: ${res.statusCode} ${url}`));
}
});
}
});
urlStream.pipe(fileStream)
return urlStream.resume() // now we are ready to handle the data
} else {
logger.error(
{ statusCode: res.statusCode, url, filePath },
'unexpected status code downloading url to cache'
)
// https://nodejs.org/api/http.html#http_class_http_clientrequest
// If you add a 'response' event handler, then you must consume
// the data from the response object, either by calling
// response.read() whenever there is a 'readable' event, or by
// adding a 'data' handler, or by calling the .resume()
// method. Until the data is consumed, the 'end' event will not
// fire. Also, until the data is read it will consume memory
// that can eventually lead to a 'process out of memory' error.
urlStream.resume() // discard the data
return callbackOnce(
new Error(
`URL returned non-success status code: ${res.statusCode} ${url}`
)
)
}
})
}
}

89
services/clsi/app/js/db.js
View file

@ -8,57 +8,60 @@
* DS102: Remove unnecessary code created because of implicit returns
* Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
*/
const Sequelize = require("sequelize");
const Settings = require("settings-sharelatex");
const _ = require("underscore");
const logger = require("logger-sharelatex");
const Sequelize = require('sequelize')
const Settings = require('settings-sharelatex')
const _ = require('underscore')
const logger = require('logger-sharelatex')
const options = _.extend({logging:false}, Settings.mysql.clsi);
const options = _.extend({ logging: false }, Settings.mysql.clsi)
logger.log({dbPath:Settings.mysql.clsi.storage}, "connecting to db");
logger.log({ dbPath: Settings.mysql.clsi.storage }, 'connecting to db')
const sequelize = new Sequelize(
Settings.mysql.clsi.database,
Settings.mysql.clsi.username,
Settings.mysql.clsi.password,
options
);
Settings.mysql.clsi.database,
Settings.mysql.clsi.username,
Settings.mysql.clsi.password,
options
)
if (Settings.mysql.clsi.dialect === "sqlite") {
logger.log("running PRAGMA journal_mode=WAL;");
sequelize.query("PRAGMA journal_mode=WAL;");
sequelize.query("PRAGMA synchronous=OFF;");
sequelize.query("PRAGMA read_uncommitted = true;");
if (Settings.mysql.clsi.dialect === 'sqlite') {
logger.log('running PRAGMA journal_mode=WAL;')
sequelize.query('PRAGMA journal_mode=WAL;')
sequelize.query('PRAGMA synchronous=OFF;')
sequelize.query('PRAGMA read_uncommitted = true;')
}
module.exports = {
UrlCache: sequelize.define("UrlCache", {
url: Sequelize.STRING,
project_id: Sequelize.STRING,
lastModified: Sequelize.DATE
}, {
indexes: [
{fields: ['url', 'project_id']},
{fields: ['project_id']}
]
}),
UrlCache: sequelize.define(
'UrlCache',
{
url: Sequelize.STRING,
project_id: Sequelize.STRING,
lastModified: Sequelize.DATE
},
{
indexes: [{ fields: ['url', 'project_id'] }, { fields: ['project_id'] }]
}
),
Project: sequelize.define("Project", {
project_id: {type: Sequelize.STRING, primaryKey: true},
lastAccessed: Sequelize.DATE
}, {
indexes: [
{fields: ['lastAccessed']}
]
}),
Project: sequelize.define(
'Project',
{
project_id: { type: Sequelize.STRING, primaryKey: true },
lastAccessed: Sequelize.DATE
},
{
indexes: [{ fields: ['lastAccessed'] }]
}
),
op: Sequelize.Op,
sync() {
logger.log({dbPath:Settings.mysql.clsi.storage}, "syncing db schema");
return sequelize.sync()
.then(() => logger.log("db sync complete")).catch(err=> console.log(err, "error syncing"));
}
};
op: Sequelize.Op,
sync() {
logger.log({ dbPath: Settings.mysql.clsi.storage }, 'syncing db schema')
return sequelize
.sync()
.then(() => logger.log('db sync complete'))
.catch(err => console.log(err, 'error syncing'))
}
}