|
@@ -2,33 +2,55 @@ request = require("request").defaults(jar: false)
|
|
|
fs = require("fs")
|
|
fs = require("fs")
|
|
|
logger = require "logger-sharelatex"
|
|
logger = require "logger-sharelatex"
|
|
|
|
|
|
|
|
|
|
+oneMinute = 60 * 1000
|
|
|
|
|
+
|
|
|
module.exports = UrlFetcher =
|
|
module.exports = UrlFetcher =
|
|
|
pipeUrlToFile: (url, filePath, _callback = (error) ->) ->
|
|
pipeUrlToFile: (url, filePath, _callback = (error) ->) ->
|
|
|
callbackOnce = (error) ->
|
|
callbackOnce = (error) ->
|
|
|
- cleanUp error, (error) ->
|
|
|
|
|
- _callback(error)
|
|
|
|
|
- _callback = () ->
|
|
|
|
|
-
|
|
|
|
|
- cleanUp = (error, callback) ->
|
|
|
|
|
- if error?
|
|
|
|
|
- logger.log filePath: filePath, "deleting file from cache due to error"
|
|
|
|
|
- fs.unlink filePath, (err) ->
|
|
|
|
|
- if err?
|
|
|
|
|
- logger.err err: err, filePath: filePath, "error deleting file from cache"
|
|
|
|
|
- callback(error)
|
|
|
|
|
- else
|
|
|
|
|
- callback()
|
|
|
|
|
|
|
+ clearTimeout timeoutHandler if timeoutHandler?
|
|
|
|
|
+ _callback(error)
|
|
|
|
|
+ _callback = () ->
|
|
|
|
|
+
|
|
|
|
|
+ timeoutHandler = setTimeout () ->
|
|
|
|
|
+ timeoutHandler = null
|
|
|
|
|
+ logger.error url:url, filePath: filePath, "Timed out downloading file to cache"
|
|
|
|
|
+ callbackOnce(new Error("Timed out downloading file to cache #{url}"))
|
|
|
|
|
+ # FIXME: maybe need to close fileStream here
|
|
|
|
|
+ , 3 * oneMinute
|
|
|
|
|
+
|
|
|
|
|
+ logger.log url:url, filePath: filePath, "started downloading url to cache"
|
|
|
|
|
+ urlStream = request.get({url: url, timeout: oneMinute})
|
|
|
|
|
+ urlStream.pause() # stop data flowing until we are ready
|
|
|
|
|
+
|
|
|
|
|
+ # attach handlers before setting up pipes
|
|
|
|
|
+ urlStream.on "error", (error) ->
|
|
|
|
|
+ logger.error err: error, url:url, filePath: filePath, "error downloading url"
|
|
|
|
|
+ callbackOnce(error or new Error("Something went wrong downloading the URL #{url}"))
|
|
|
|
|
|
|
|
- fileStream = fs.createWriteStream(filePath)
|
|
|
|
|
- fileStream.on 'error', (error) ->
|
|
|
|
|
- logger.error err: error, url:url, filePath: filePath, "error writing file into cache"
|
|
|
|
|
- callbackOnce(error)
|
|
|
|
|
|
|
+ urlStream.on "end", () ->
|
|
|
|
|
+ logger.log url:url, filePath: filePath, "finished downloading file into cache"
|
|
|
|
|
|
|
|
- logger.log url:url, filePath: filePath, "downloading url to cache"
|
|
|
|
|
- urlStream = request.get(url)
|
|
|
|
|
urlStream.on "response", (res) ->
|
|
urlStream.on "response", (res) ->
|
|
|
if res.statusCode >= 200 and res.statusCode < 300
|
|
if res.statusCode >= 200 and res.statusCode < 300
|
|
|
|
|
+ fileStream = fs.createWriteStream(filePath)
|
|
|
|
|
+
|
|
|
|
|
+ # attach handlers before setting up pipes
|
|
|
|
|
+ fileStream.on 'error', (error) ->
|
|
|
|
|
+ logger.error err: error, url:url, filePath: filePath, "error writing file into cache"
|
|
|
|
|
+ fs.unlink filePath, (err) ->
|
|
|
|
|
+ if err?
|
|
|
|
|
+ logger.err err: err, filePath: filePath, "error deleting file from cache"
|
|
|
|
|
+ callbackOnce(error)
|
|
|
|
|
+
|
|
|
|
|
+ fileStream.on 'finish', () ->
|
|
|
|
|
+ logger.log url:url, filePath: filePath, "finished writing file into cache"
|
|
|
|
|
+ callbackOnce()
|
|
|
|
|
+
|
|
|
|
|
+ fileStream.on 'pipe', () ->
|
|
|
|
|
+ logger.log url:url, filePath: filePath, "piping into filestream"
|
|
|
|
|
+
|
|
|
urlStream.pipe(fileStream)
|
|
urlStream.pipe(fileStream)
|
|
|
|
|
+ urlStream.resume() # now we are ready to handle the data
|
|
|
else
|
|
else
|
|
|
logger.error statusCode: res.statusCode, url:url, filePath: filePath, "unexpected status code downloading url to cache"
|
|
logger.error statusCode: res.statusCode, url:url, filePath: filePath, "unexpected status code downloading url to cache"
|
|
|
# https://nodejs.org/api/http.html#http_class_http_clientrequest
|
|
# https://nodejs.org/api/http.html#http_class_http_clientrequest
|
|
@@ -39,15 +61,5 @@ module.exports = UrlFetcher =
|
|
|
# method. Until the data is consumed, the 'end' event will not
|
|
# method. Until the data is consumed, the 'end' event will not
|
|
|
# fire. Also, until the data is read it will consume memory
|
|
# fire. Also, until the data is read it will consume memory
|
|
|
# that can eventually lead to a 'process out of memory' error.
|
|
# that can eventually lead to a 'process out of memory' error.
|
|
|
- urlStream.on 'data', () -> # discard the data
|
|
|
|
|
|
|
+ urlStream.resume() # discard the data
|
|
|
callbackOnce(new Error("URL returned non-success status code: #{res.statusCode} #{url}"))
|
|
callbackOnce(new Error("URL returned non-success status code: #{res.statusCode} #{url}"))
|
|
|
-
|
|
|
|
|
- urlStream.on "error", (error) ->
|
|
|
|
|
- logger.error err: error, url:url, filePath: filePath, "error downloading url"
|
|
|
|
|
- callbackOnce(error or new Error("Something went wrong downloading the URL #{url}"))
|
|
|
|
|
-
|
|
|
|
|
- urlStream.on "end", () ->
|
|
|
|
|
- # FIXME: what if we get an error writing the file? Maybe we
|
|
|
|
|
- # should be using the fileStream end event as the point of
|
|
|
|
|
- # callback.
|
|
|
|
|
- callbackOnce()
|
|
|