def pipe_926a27580e74684fc63559b92bddaa30(context=None, _INPUT=None, conf=None, **kwargs):
# todo: insert pipeline description here
conf = conf or {}
if context and context.describe_input:
return []
forever = pipe_forever()
sw_68 = pipe_fetch(
context,
forever,
conf={
"URL": [
{"type": "url", "value": "http://www.guru.com/rss/jobs/c/web-software-it/"},
{
"type": "url",
"value": "https://www.elance.com/r/rss/jobs/cat-it-programming/fxd-true/o-1/bgt-gt500-ns1/sct-database-development-10217-data-analysis-14174-database-administration-14177-business-intelligence-14173-data-engineering-14175-system-administration-10219-other-data-science-14178-technical-support-10218-other-it-programming-12350-software-application-10216-website-design-10225-web-programming-10224/tls-1/s-timelistedSort",
},
]
},
)
sw_90 = pipe_uniq(context, sw_68, conf={"field": {"type": "text", "value": "link"}})
sw_87 = pipe_filter(
context,
sw_90,
conf={
"COMBINE": {"type": "text", "value": "or"},
"MODE": {"type": "text", "value": "block"},
"RULE": [
{
"field": {"type": "text", "value": "title"},
"value": {"type": "text", "value": "php"},
"op": {"type": "text", "value": "contains"},
}
],
},
)
sw_101 = pipe_sort(
context,
sw_87,
conf={"KEY": [{"field": {"type": "text", "value": "pubDate"}, "dir": {"type": "text", "value": "DESC"}}]},
)
_OUTPUT = pipe_output(context, sw_101, conf={})
return _OUTPUT
def pipe_6e30c269a69baf92cd420900b0645f88(context=None, _INPUT=None, conf=None, **kwargs):
# todo: insert pipeline description here
conf = conf or {}
if context and context.describe_input:
return []
if context and context.describe_dependencies:
return [u'pipefetch', u'pipefilter', u'pipeoutput', u'piperegex', u'piperename', u'pipesort', u'pipeunion', u'pipeuniq']
forever = pipe_forever()
sw_135 = pipe_fetch(
context, forever, conf={'URL': {'type': 'url', 'value': 'file://data/rss.sueddeutsche.de_rss_Topthemen.xml'}})
sw_233 = pipe_fetch(
context, forever, conf={'URL': {'type': 'url', 'value': 'file://data/rss.sueddeutsche.de_rss_Politik.xml'}})
sw_154 = pipe_union(
context, forever, _OTHER3=sw_233, conf={}, _OTHER=sw_135)
sw_173 = pipe_uniq(
context, sw_154, conf={'field': {'type': 'text', 'value': 'title'}})
sw_180 = pipe_filter(
context, sw_173, conf={'COMBINE': {'type': 'text', 'value': 'or'}, 'MODE': {'type': 'text', 'value': 'block'}, 'RULE': [{'field': {'type': 'text', 'value': 'link'}, 'value': {'type': 'text', 'value': '/sport/'}, 'op': {'type': 'text', 'value': 'contains'}}, {'field': {'type': 'text', 'value': 'title'}, 'value': {'type': 'text', 'value': 'Bildstrecke:'}, 'op': {'type': 'text', 'value': 'contains'}}]})
sw_210 = pipe_rename(
context, sw_180, conf={'RULE': [{'field': {'type': 'text', 'value': 'y:id.value'}, 'op': {'type': 'text', 'value': 'copy'}, 'newval': {'type': 'text', 'value': 'link'}}]})
sw_195 = pipe_regex(
context, sw_210, conf={'RULE': [{'singlelinematch': {'type': 'text', 'value': '2'}, 'globalmatch': {'type': 'text', 'value': '1'}, 'replace': {'type': 'text', 'value': ''}, 'field': {'type': 'text', 'value': 'description'}, 'casematch': {'type': 'text', 'value': '8'}, 'match': {'type': 'text', 'value': '</div>.*$'}}, {'field': {'type': 'text', 'value': 'link'}, 'match': {'type': 'text', 'value': '^(.*\\/.*)\\/'}, 'replace': {'type': 'text', 'value': '$1/2.220/'}}]})
sw_191 = pipe_sort(
context, sw_195, conf={'KEY': [{'field': {'type': 'text', 'value': 'pubDate'}, 'dir': {'type': 'text', 'value': 'DESC'}}]})
_OUTPUT = pipe_output(
context, sw_191, conf={})
return _OUTPUT
def pipe_926a27580e74684fc63559b92bddaa30(context=None,
_INPUT=None,
conf=None,
**kwargs):
# todo: insert pipeline description here
conf = conf or {}
if context and context.describe_input:
return []
forever = pipe_forever()
sw_68 = pipe_fetch(
context,
forever,
conf={
'URL': [{
'type': 'url',
'value': 'http://www.guru.com/rss/jobs/c/web-software-it/'
}, {
'type':
'url',
'value':
'https://www.elance.com/r/rss/jobs/cat-it-programming/fxd-true/o-1/bgt-gt500-ns1/sct-database-development-10217-data-analysis-14174-database-administration-14177-business-intelligence-14173-data-engineering-14175-system-administration-10219-other-data-science-14178-technical-support-10218-other-it-programming-12350-software-application-10216-website-design-10225-web-programming-10224/tls-1/s-timelistedSort'
}]
})
sw_90 = pipe_uniq(context,
sw_68,
conf={'field': {
'type': 'text',
'value': 'link'
}})
sw_87 = pipe_filter(context,
sw_90,
conf={
'COMBINE': {
'type': 'text',
'value': 'or'
},
'MODE': {
'type': 'text',
'value': 'block'
},
'RULE': [{
'field': {
'type': 'text',
'value': 'title'
},
'value': {
'type': 'text',
'value': 'php'
},
'op': {
'type': 'text',
'value': 'contains'
}
}]
})
sw_101 = pipe_sort(context,
sw_87,
conf={
'KEY': [{
'field': {
'type': 'text',
'value': 'pubDate'
},
'dir': {
'type': 'text',
'value': 'DESC'
}
}]
})
_OUTPUT = pipe_output(context, sw_101, conf={})
return _OUTPUT
def pipe_6e30c269a69baf92cd420900b0645f88(context=None, _INPUT=None, conf=None, **kwargs):
# todo: insert pipeline description here
conf = conf or {}
if context and context.describe_input:
return []
if context and context.describe_dependencies:
return [
u"pipefetch",
u"pipefilter",
u"pipeoutput",
u"piperegex",
u"piperename",
u"pipesort",
u"pipeunion",
u"pipeuniq",
]
forever = pipe_forever()
sw_135 = pipe_fetch(
context, forever, conf={"URL": {"type": "url", "value": "file://data/rss.sueddeutsche.de_rss_Topthemen.xml"}}
)
sw_233 = pipe_fetch(
context, forever, conf={"URL": {"type": "url", "value": "file://data/rss.sueddeutsche.de_rss_Politik.xml"}}
)
sw_154 = pipe_union(context, forever, _OTHER3=sw_233, conf={}, _OTHER=sw_135)
sw_173 = pipe_uniq(context, sw_154, conf={"field": {"type": "text", "value": "title"}})
sw_180 = pipe_filter(
context,
sw_173,
conf={
"COMBINE": {"type": "text", "value": "or"},
"MODE": {"type": "text", "value": "block"},
"RULE": [
{
"field": {"type": "text", "value": "link"},
"value": {"type": "text", "value": "/sport/"},
"op": {"type": "text", "value": "contains"},
},
{
"field": {"type": "text", "value": "title"},
"value": {"type": "text", "value": "Bildstrecke:"},
"op": {"type": "text", "value": "contains"},
},
],
},
)
sw_210 = pipe_rename(
context,
sw_180,
conf={
"RULE": [
{
"field": {"type": "text", "value": "y:id.value"},
"op": {"type": "text", "value": "copy"},
"newval": {"type": "text", "value": "link"},
}
]
},
)
sw_195 = pipe_regex(
context,
sw_210,
conf={
"RULE": [
{
"singlelinematch": {"type": "text", "value": "2"},
"globalmatch": {"type": "text", "value": "1"},
"replace": {"type": "text", "value": ""},
"field": {"type": "text", "value": "description"},
"casematch": {"type": "text", "value": "8"},
"match": {"type": "text", "value": "</div>.*$"},
},
{
"field": {"type": "text", "value": "link"},
"match": {"type": "text", "value": "^(.*\\/.*)\\/"},
"replace": {"type": "text", "value": "$1/2.220/"},
},
]
},
)
sw_191 = pipe_sort(
context,
sw_195,
conf={"KEY": [{"field": {"type": "text", "value": "pubDate"}, "dir": {"type": "text", "value": "DESC"}}]},
)
_OUTPUT = pipe_output(context, sw_191, conf={})
return _OUTPUT