class UploadMongoData():
def __init__(self):
self.settings = Settings()
self.file = FileIOMiddleware()
self.doraemon = Doraemon()
self.log_path = self.settings.LOG_PATH
self.doraemon.createFilePath(self.log_path)
def startUpload(self):
fromFile = self.settings.LOCAL_MONGO_DATA_PATH
toFile = self.settings.REMOTE_MONGO_DATA_PATH
if not os.path.exists(fromFile):
print 'no mongo data file to upload'
return
while os.path.exists(fromFile):
try:
if self.doraemon.sshUpload(
self.settings.IP_WEBSERVER0,
self.settings.PORT_WEBSERVER0,
self.settings.USER_ROOT_WEBSERVER0,
self.settings.USER_ROOT_PASSWORD_WEBSERVER0, fromFile,
toFile):
self.doraemon.deleteFile(fromFile)
message1 = 'Success to upload mongo data file: {0}'.format(
fromFile)
print message1
self.file.logger(self.log_path, message1)
except Exception as e:
message2 = 'Exception {0} to upload mongo data file: {1}'.format(
e.message, fromFile)
print message2
self.file.logger(self.log_path, message2)
class NoNameBone():
def __init__(self, settingName, callback=callable):
self.settingName = settingName
self.callBack = callback
self.globalSettings = Settings()
self.getSettings()
self.file = FileIOMiddleware()
self.doraemon = Doraemon()
def getSettings(self):
self.settings = self.globalSettings.CreateSettings(self.settingName)
self.log_path = self.globalSettings.LOG_PATH_PRD2
self.author_path = self.settings.AUTHORS_PATH
self.name = self.settings.NAME
def store(self):
result = self.callBack()
if result == None:
return
print 'Start to store authors for page: {0}'.format(result.page_url)
if len(result.authors) == 0:
message1 = 'No author for page: {0}'.format(result.page_url)
self.file.logger(self.log_path, message1)
print message1
for item in result.authors:
is_title_empty = self.doraemon.isEmpty(item)
if (is_title_empty is False) and (self.doraemon.isDuplicated(
self.doraemon.bf_authors, item) is False):
message2 = 'Start to store author: {0} for page: {1}.'.format(
item, result.page_url)
self.file.logger(self.log_path, message2)
print message2
self.doraemon.storeTxtAdd(self.author_path, item,
self.settingName)
message3 = 'Success to store author: {0} for page: {1}.'.format(
item, result.page_url)
self.file.logger(self.log_path, message3)
print message3
else:
if is_title_empty is True:
message4 = 'Empty author for {0}'.format(result.page_url)
self.file.logger(self.log_path, message4)
print message4
else:
message5 = 'Duplicated author for {0}'.format(
result.page_url)
self.file.logger(self.log_path, message5)
print message5
print 'End to store author for page: {0}.'.format(result.page_url)
del result
gc.collect()
class StoreFiles():
def __init__(self,
htmlpath=None,
imagepath=None,
templatepath=None,
articleurl=None,
alidomain=None,
alidomaindeepinews=None,
alidomaindeepinewsimg=None,
ipwebserver0=None,
portwebserver0=None,
userrootwebserver0=None,
userrootpasswordwebserver0=None,
htmlwebserver0=None,
needselfimage=None,
needselfhtml=None,
localhtmlpath=None,
logpath=None):
self.doraemon = Doraemon()
self.file = FileIOMiddleware()
self.image_count = 0
self.htmlpath = htmlpath
self.imagepath = imagepath
self.templatepath = templatepath
self.articleurl = articleurl
self.alidomain = alidomain
self.alidomaindeepinews = alidomaindeepinews
self.alidomaindeepinewsimg = alidomaindeepinewsimg
self.ipwebserver0 = ipwebserver0
self.portwebserver0 = portwebserver0
self.userrootwebserver0 = userrootwebserver0
self.userrootpasswordwebserver0 = userrootpasswordwebserver0
self.htmlwebserver0 = htmlwebserver0
self.needselfimage = needselfimage
self.needselfhtml = needselfhtml
self.localhtmlpath = localhtmlpath
self.logpath = logpath
def parseContentRegxRule(self, content_regx_rule):
result = matchRules(None, None, None)
rules = [
matchRules(r'[\.][\/][\/](.*?)[[]', r'[\@](.*?)[\=]',
r'[\'](.*?)[\']'),
matchRules(r'[\.][\/][\/](.*?)[[]', r'[\@](.*?)[,]',
r'[\'](.*?)[\']'),
]
for rule in rules:
tag = re.findall(rule.tag, content_regx_rule)
key = re.findall(rule.key, content_regx_rule)
value = re.findall(rule.value, content_regx_rule)
if self.doraemon.isEmpty(tag) is False and \
self.doraemon.isEmpty(key) is False and \
self.doraemon.isEmpty(value) is False:
result.tag = tag[0]
result.key = key[0]
result.value = value[0]
break
return result
def addHighlightTextInner(self, content):
return '<strong class="article_paragraph_border">{0}</strong>'.format(content) + \
'<p class="article_paragraph">' + \
'<br class="article_paragraph_border"/>' + \
'</p>'
def addHighlightTextOuter(self, node, content):
return '{0}<p class="article_paragraph">'.format(node) + \
'<strong class="article_paragraph_border">{0}</strong>'.format(content) + \
'</p>' + \
'<p class="article_paragraph">' + \
'<br class="article_paragraph_border"/>' + \
'</p>'
def addImgNode(self, node, dataSrc, dataRef, width, dataRatio):
if width == None:
width = 1000
return '{0}<p class="article_paragraph_imag">'.format(node) + \
'<img data-ratio="{0}"'.format(dataRatio) + \
'data-src="{0}"'.format(dataSrc) + \
'data-ref="{0}"'.format(dataRef) + \
'data-type="jpeg"' + \
'data-w={0} '.format(width) + \
'class="article_paragraph_img"/>' + \
'</p>' + \
'<p class="article_paragraph">' + \
'<br class="article_paragraph_border"/>' + \
'</p>'
def addTextNodeOuter(self, node, content):
if self.doraemon.isEmpty(content):
return ''
return '{0}<p class="article_paragraph">{1}'.format(node, content) + \
'</p>' + \
'<p class="article_paragraph">' + \
'<br class="article_paragraph_border"/>' + \
'</p>'
def addParagraphGapNode(self, node):
return '{0}<p class="article_paragraph">'.format(node) + \
'<br class="article_paragraph_border"/>' + \
'</p>'
def addH1Node(self, node, content):
return '{0}<p label="h1" class="article_paragraph_h1">'.format(node) + \
'<span class="article_paragraph_h1_1">' + \
'<span class="article_paragraph_h1_1_1">' + \
'<span class="article_paragraph_h1_1_1_1">{0}'.format(content) + \
'</span>' + \
'</span>' + \
'</span>' + \
'</p>' + \
'<p class="article_paragraph">' + \
'<br class="article_paragraph_border"/>' + \
'</p>'
def extractImgSize(self, style, mode):
size = re.findall(r'{0}:(.*?)px;'.format(mode), style)
if len(size) == 1:
return size[0].strip()
return None
def extractImg(self, url, node):
result = imgInfo(None, None, None)
if isinstance(node, NavigableString):
return None
if node.name != 'img' and len(node.contents) == 0:
return result
if node.name == 'img':
if node.attrs.has_key('src') and self.doraemon.isEmpty(result.src):
if 'data:image/' not in node.attrs['src']:
result.src = node.attrs['src']
if node.attrs.has_key('_src') and self.doraemon.isEmpty(
result.src):
if 'data:image/' not in node.attrs['_src']:
result.src = node.attrs['_src']
if node.attrs.has_key('data-original') and self.doraemon.isEmpty(
result.src):
if 'data:image/' not in node.attrs['data-original']:
result.src = node.attrs['data-original']
if node.attrs.has_key('data-src') and self.doraemon.isEmpty(
result.src):
if 'data:image/' not in node.attrs['data-src']:
result.src = node.attrs['data-src']
if node.attrs.has_key('data-lazy-src') and self.doraemon.isEmpty(
result.src):
if 'data:image/' not in node.attrs['data-lazy-src']:
result.src = node.attrs['data-lazy-src']
if node.attrs.has_key('width') and result.width == None:
result.width = node.attrs['width']
if node.attrs.has_key('height') and result.height == None:
result.height = node.attrs['height']
if node.attrs.has_key('data-w') and result.width == None:
result.width = node.attrs['data-w']
if node.attrs.has_key('data-h') and result.height == None:
result.height = node.attrs['data-h']
if node.attrs.has_key('data-backh') and result.height == None:
result.height = node.attrs['data-backh']
if node.attrs.has_key('data-backw') and result.width == None:
result.width = node.attrs['data-backw']
if node.attrs.has_key('data-wscnh') and result.height == None:
result.height = node.attrs['data-wscnh']
if node.attrs.has_key('data-wscnw') and result.width == None:
result.width = node.attrs['data-wscnw']
if node.attrs.has_key('style') and (result.width == None
or result.height == None):
result.width = self.extractImgSize(node.attrs['style'],
'width')
result.height = self.extractImgSize(node.attrs['style'],
'height')
if isinstance(result.width, int) and isinstance(
result.height, int):
result.dataRatio = float(
float(result.height) / float(result.width))
if result.src != None:
result.src = urlparse.urljoin(url, result.src).strip()
return result
if len(node.contents) > 0:
for n in node.contents:
result = self.extractImg(url, n)
if result != None:
return result
return result
def nodeTraversal(self, url, node, newNode, articleId):
if node.name == 'strong' and \
node.parent.name == 'div' and \
self.doraemon.isEmpty(node.string) is False:
newNode = '{0}{1}'.format(
newNode, self.addHighlightTextOuter(newNode, node.string))
if (node.name == 'h1' or \
node.name == 'h2' or \
node.name == 'h3' or \
node.name == 'h4') and \
self.doraemon.isEmpty(node.string) is False:
newNode = '{0}{1}'.format(newNode,
self.addH1Node(newNode, node.string))
if isinstance(node, NavigableString) or \
node.name == 'a' or \
node.name == 'p' or \
node.name == 'span' or \
node.name == 'section':
if isinstance(node, NavigableString):
newNode = self.addTextNodeOuter(newNode, str(node))
else:
if self.doraemon.isEmpty(node.text) == False:
newNode = self.addTextNodeOuter(newNode, node.text)
img = self.extractImg(url, node)
updatedNode = updateNode(False, newNode, None, None)
if img != None and img.src != None:
updatedNode.isImageNode = True
updatedNode.imageOriginUrl = img.src
updatedNode.imageNewUrl = img.src
try:
imageType = self.doraemon.getImageTypeFromUrl(
updatedNode.imageOriginUrl)
imageId = '{0}_{1}'.format(articleId, self.image_count)
newImageName = '{0}.{1}'.format(imageId, imageType)
if self.doraemon.downloadImage(updatedNode.imageOriginUrl,
self.imagepath, newImageName):
imageInfo = Image.open('{0}/{1}'.format(
self.imagepath, newImageName))
if self.doraemon.isEmpty(imageInfo.width) is False:
img.width = imageInfo.width
if self.doraemon.isEmpty(imageInfo.height) is False:
img.height = imageInfo.height
if isinstance(img.width, int) and isinstance(
img.height, int):
img.dataRatio = float(
float(img.height) / float(img.width))
if self.needselfimage:
updatedNode.imageNewUrl = 'https://{0}.{1}/{2}/{3}'.format(
self.alidomaindeepinews, self.alidomain,
self.alidomaindeepinewsimg, newImageName)
imageUpload = AliUpload(
'{0}'.format(self.imagepath), newImageName,
'{0}'.format(self.alidomaindeepinews),
'{0}'.format(self.alidomaindeepinewsimg))
if imageUpload.start():
updatedNode.node = '{0}{1}'.format(
newNode,
self.addImgNode(newNode,
updatedNode.imageNewUrl,
updatedNode.imageNewUrl,
img.width, img.dataRatio))
self.image_count += 1
else:
updatedNode.node = '{0}{1}'.format(
newNode,
self.addImgNode(newNode, img.src, img.src,
img.width, img.dataRatio))
else:
updatedNode.node = '{0}{1}'.format(
newNode,
self.addImgNode(newNode, img.src, img.src, img.width,
img.dataRatio))
except Exception as e:
updatedNode.node = '{0}{1}'.format(
newNode,
self.addImgNode(newNode, img.src, img.src, img.width,
img.dataRatio))
print 'Exception {0} to download image: {1}'.format(
e.message, updatedNode.imageOriginUrl)
return updatedNode
def updateTemplate(self, template, articleHeadDescription,
articleHeadAuthor, articleHeadTitle,
articleHeadOriginUrl, articleBodyTitle,
articleBodyAuthor, articleBodyPublishTime,
articleBodyParagraph, articleBodyOriginUrl):
template = template.replace('ArticleHeadDescription',
articleHeadDescription)
template = template.replace('ArticleHeadAuthor', articleHeadAuthor)
template = template.replace('ArticleHeadTitle', articleHeadTitle)
template = template.replace('ArticleHeadOriginUrl',
articleHeadOriginUrl)
template = template.replace('ArticleBodyTitle', articleBodyTitle)
template = template.replace('ArticleBodyAuthor', articleBodyAuthor)
template = template.replace('ArticleBodyPublishTime',
articleBodyPublishTime)
template = template.replace('ArticleBodyParagraph',
articleBodyParagraph)
template = template.replace('ArticleBodyOriginUrl',
articleBodyOriginUrl)
return template
def hasText(self, nodes):
for node in nodes:
if isinstance(node, NavigableString):
continue
if node.name == 'img' or \
node.name == 'a' or \
node.name == 'p' or \
node.name == 'span' or \
node.name == 'section':
return True
return False
def goDeepToArticleBody(self, contents):
if isinstance(contents, NavigableString):
return contents
if len(contents) == 0:
return contents
if self.hasText(contents):
return contents
if len(contents) > 0:
for n in contents:
if isinstance(n, NavigableString):
continue
return self.goDeepToArticleBody(n.contents)
def storeFiles(self, data, page_source, content_regx_rule):
if self.needselfhtml == False:
return data
try:
self.image_count = 0
newData = copy.copy(data)
newArticleId = self.doraemon.getMD5('{0}_{1}'.format(
data.author_name, data.id))
newData.url = '{0}{1}.html'.format(self.articleurl, newArticleId)
template = self.file.readFromTxt(self.templatepath)
match = self.parseContentRegxRule(content_regx_rule)
if match.tag is None or \
match.key is None or \
match.value is None:
print 'No match rule available for html'
return data
soup = BeautifulSoup(page_source, 'lxml')
matchTags = soup.select('{0}[{1}="{2}"]'.format(
match.tag, match.key, match.value))
if len(matchTags) == 0:
print 'No tag matched for html'
return data
nodes = self.goDeepToArticleBody(matchTags[0].contents)
articleContent = ''
for node in nodes:
if isinstance(node, NavigableString):
continue
if self.doraemon.isEmpty(node):
continue
newNode = ''
updateNodeInfo = self.nodeTraversal(data.url, node, newNode,
newArticleId)
articleContent = '{0}{1}'.format(articleContent,
updateNodeInfo.node)
if updateNodeInfo.isImageNode:
if updateNodeInfo.imageOriginUrl in newData.images:
for i in newData.images:
if updateNodeInfo.imageOriginUrl in i or \
updateNodeInfo.imageOriginUrl == i:
newData.images[newData.images.index(
i)] = updateNodeInfo.imageNewUrl
else:
newData.images.append(updateNodeInfo.imageNewUrl)
template = self.updateTemplate(template, newData.title,
'深度资讯DeepINews', newData.title,
newData.url, newData.title,
newData.source, newData.public_time,
articleContent, data.url)
if self.doraemon.storeHtml(newArticleId, template, self.htmlpath):
htmlName = '{0}.html'.format(newArticleId)
fromFile = '{0}/{1}'.format(self.htmlpath, htmlName)
toFile = '{0}/{1}'.format(self.localhtmlpath, htmlName)
if self.doraemon.copyFile(fromFile, toFile):
print 'Copy file {0} done.'.format(fromFile)
return newData
else:
message1 = 'Copy file {0} fail.'.format(fromFile)
print message1
self.file.logger(self.logpath, message1)
return data
except Exception as e:
message2 = 'Exception {0} when update : {1}'.format(
e.message, data.url)
print message2
self.file.logger(self.logpath, message2)
return data
class CamelBone():
def __init__(self, siteinfo=None, callback=callable):
self.siteinfo = siteinfo
self.callBack = callback
self.globalSettings = Settings()
self.file = FileIOMiddleware()
self.doraemon = Doraemon()
self.getSettings()
def getSettings(self):
self.settings = self.globalSettings.CreateSettings(self.siteinfo)
self.log_path = self.globalSettings.LOG_PATH_PRD2
self.today = self.globalSettings.TODAY
self.source = self.settings.SOURCE_NAME
self.work_path_prd2 = self.settings.WORK_PATH_PRD2
self.mongo = self.settings.MONGO_URLS
self.name = self.settings.NAME
self.max_pool_size = self.settings.MAX_POOL_SIZE_URL
self.urls = self.settings.URLS
self.max_concurrency = self.globalSettings.MAX_CONCURRENCY
self.concurrency_file = self.globalSettings.CONCURRENCY_FILE
self.url_backup_folder_path = self.settings.URL_BACKUP_FOLDER_PATH
self.url_timeout = self.settings.URL_TIMEOUT
self.createPath()
def createPath(self):
self.doraemon.createFilePath(self.work_path_prd2)
self.doraemon.createFilePath(self.log_path)
self.doraemon.createFilePath(self.url_backup_folder_path)
def parse(self, response):
time.sleep(1)
current_url = response['response'].current_url.encode('gbk')
print 'Start to parse: {0}'.format(current_url)
html = etree.HTML(response['response'].page_source)
results = self.callBack(current_url, html)
if len(results) == 0:
message1 = 'No url for page: {0}'.format(current_url)
self.file.logger(self.log_path, message1)
print message1
for item in results:
is_title_empty = self.doraemon.isEmpty(item.title)
if (is_title_empty is False) and (self.doraemon.isDuplicated(
self.doraemon.bf_urls, item.title) is False):
message2 = 'Start to store mongo {0}'.format(item.url)
self.file.logger(self.log_path, message2)
print message2
self.doraemon.storeMongodb(
self.mongo, self.doraemon.createCamelMongoJson(item))
message3 = 'End to store mongo {0}'.format(item.url)
self.file.logger(self.log_path, message3)
print message3
self.file.logger(self.log_path,
'Done for {0}'.format(item.url))
else:
if is_title_empty is True:
message4 = 'Empty title for {0}'.format(item.url)
self.file.logger(self.log_path, message4)
print message4
else:
print 'Finished title for {0}'.format(item.url)
print 'End to parse {0}'.format(current_url)
del current_url, results, html
gc.collect()
def start(self, isdebug=False):
if self.doraemon.isCamelReadyToRun(
self.settings) is False and isdebug is False:
message5 = 'It is not ready to run for {0}'.format(self.name)
print message5
return
message6 = 'Start {0} requests'.format(self.name)
self.file.logger(self.log_path, message6)
print message6
new_urls = []
content = self.file.readFromTxt(self.urls)
url_list = content.split('\n')
for url in url_list:
if self.doraemon.isEmpty(url) is False:
new_urls.append([url, ''])
if len(new_urls) == 0:
print 'No url.'
return
request = BrowserRequest()
content = request.start_chrome(new_urls,
self.url_timeout,
self.max_pool_size,
self.log_path,
None,
callback=self.parse)
self.doraemon.recoveryConcurrency(self.concurrency_file,
self.max_concurrency)
message7 = 'End for {0} requests of {1}.'.format(
str(len(content)), self.name)
self.file.logger(self.log_path, message7)
print message7
del new_urls, content, url_list, request
gc.collect()
class SpiderBone():
def __init__(self, siteinfo=None, callback=callable):
self.siteinfo = siteinfo
self.callBack = callback
self.globalSettings = Settings()
self.file = FileIOMiddleware()
self.doraemon = Doraemon()
self.getSettings()
def getSettings(self):
self.settings = self.globalSettings.CreateSettings(self.siteinfo)
self.log_path = self.globalSettings.LOG_PATH
self.today = self.globalSettings.TODAY
self.source = self.settings.SOURCE_NAME
self.work_path_prd1 = self.settings.WORK_PATH_PRD1
self.finished_txt_path = self.settings.FINISHED_TXT_PATH
self.finished_html_path = self.settings.FINISHED_HTML_PATH
self.finished_image_path = self.settings.FINISHED_IMG_PATH
self.template_path = self.globalSettings.TEMPLATE_PATH
self.article_url = self.globalSettings.ARTICLE_URL
self.ali_domain = self.globalSettings.ALI_DOMAIN
self.ali_domain_deepinews = self.globalSettings.ALI_BUCKET_NAME_DEEPINEWS
self.ali_domain_deepinews_img = self.globalSettings.ALI_BUCKET_NAME_DEEPINEWS_IMG
self.ip_webserver0 = self.globalSettings.IP_WEBSERVER0
self.port_webserver0 = self.globalSettings.PORT_WEBSERVER0
self.user_root_webserver0 = self.globalSettings.USER_ROOT_WEBSERVER0
self.user_root_password_webserver0 = self.globalSettings.USER_ROOT_PASSWORD_WEBSERVER0
self.html_webserver0 = self.globalSettings.HTML_WEBSERVER0
self.mongo = self.settings.MONGO
self.name = self.settings.NAME
self.max_pool_size = self.settings.MAX_POOL_SIZE_CONTENT
self.url_path = self.settings.URL_PATH
self.is_open_cache = self.settings.IS_OPEN_CACHE
self.finished_backup_folder_path = self.settings.FINISHED_BACKUP_FOLDER_PATH
self.max_concurrency_spider = self.globalSettings.MAX_CONCURRENCY_SPIDER
self.concurrency_file_spider = self.globalSettings.CONCURRENCY_FILE_SPIDER
self.monitor_upload_local = self.globalSettings.MONITOR_UPLOAD_LOCAL
self.local_html_path = self.globalSettings.LOCAL_HTML_PATH
self.content_timeout = self.settings.CONTENT_TIMEOUT
self.createPath()
def createPath(self):
self.doraemon.createFilePath(self.work_path_prd1)
self.doraemon.createFilePath(self.log_path)
self.doraemon.createFilePath(self.finished_backup_folder_path)
self.doraemon.createFilePath(self.monitor_upload_local)
self.doraemon.createFilePath(self.local_html_path)
def parse(self, response):
time.sleep(1)
current_url = response['response'].current_url.encode('gbk')
request_title = response['request_title']
print 'Start to parse: {0}'.format(current_url)
page_source = response['response'].page_source
html = etree.HTML(page_source)
results = None
try:
results = self.callBack(current_url, html, page_source)
if results == None:
message1 = 'No content for: {0}'.format(current_url)
print message1
self.file.logger(self.log_path, message1)
return
dataToMongo = self.doraemon.createSpiderMongoJson(results)
except Exception as e:
message1 = 'Exception when parse: {0} for {1}'.format(current_url, e.message)
print message1
self.file.logger(self.log_path, message1)
print 'End to parse: {0}'.format(current_url)
if results == None:
self.doraemon.storeFinished(self.doraemon.bf_content, request_title)
print 'No data for {0}'.format(request_title)
else:
message2 = 'Start to store mongo {0}'.format(results.url)
self.file.logger(self.log_path, message2)
print message2
self.doraemon.storeMongodb(self.mongo, dataToMongo)
message3 = 'End to store mongo {0}'.format(results.url)
self.file.logger(self.log_path, message3)
print message3
self.doraemon.storeTxt(results.id, results.content, self.finished_txt_path, self.name)
self.doraemon.storeFinished(self.doraemon.bf_content, request_title)
def start(self):
if self.doraemon.isSpiderReadyToRun() is False:
message4 = 'It is not ready to run spider: {0}'.format(self.name)
print message4
return
message5 = 'Start {0} requests'.format(self.name)
self.file.logger(self.log_path, message5)
print message5
message6 = 'Start requests: {0} '.format(self.name)
self.file.logger(self.log_path, message6)
print message6
new_url_titles = self.doraemon.readNewUrls(self.doraemon.bf_content, self.url_path)
if len(new_url_titles) == 0:
self.doraemon.recoveryConcurrency(self.concurrency_file_spider, self.max_concurrency_spider)
message7 = 'No new url for {0}'.format(self.name)
self.file.logger(self.log_path, message7)
print message7
return
request = BrowserRequest()
content = request.start_chrome(new_url_titles, self.content_timeout, self.max_pool_size, self.log_path, None, callback=self.parse)
self.doraemon.recoveryConcurrency(self.concurrency_file_spider, self.max_concurrency_spider)
message8 = 'End requests for {0}'.format(str(len(content)))
self.file.logger(self.log_path, message8)
print message8
del content, new_url_titles, request
gc.collect()