def test_parse_item(self):
filename = os.path.join('data/defensa', '19-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'19/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'HORTENCIA VANESSA GONZALES VALDIVIA')
self.assertEqual(item.get('time_start'), u'08:48')
self.assertEqual(item.get('institution'), u'defensa')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'41795231')
self.assertEqual(item.get('entity'), u'CONIDA')
self.assertEqual(item.get('reason'), u'REUNION DE TRABAJO')
self.assertEqual(item.get('host_name'), u'DUPEYRAT LUQUE WOLFGANG CARLOS DOUGLAS')
self.assertEqual(item.get('time_end'), u'09:49')
self.assertEqual(item.get('date'), u'08/11/2017')
self.assertEqual(item.get('sha1'), u'd7d3fc2a9a0f123473817b201dac7e651aee445a')
item = next(items)
self.assertEqual(item.get('full_name'), u'RIGOBERTO SALAS ASENCIOS')
self.assertEqual(item.get('institution'), u'defensa')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'43847220')
self.assertEqual(item.get('entity'), u'AGENCIA DE COMPRAS DE LAS FF. AA.')
self.assertEqual(item.get('reason'), u'REUNION DE TRABAJO')
self.assertEqual(item.get('host_name'), u'DUPEYRAT LUQUE WOLFGANG CARLOS DOUGLAS')
self.assertEqual(item.get('time_start'), u'09:00')
self.assertEqual(item.get('time_end'), u'12:35')
self.assertEqual(item.get('date'), u'08/11/2017')
self.assertEqual(item.get('sha1'), u'eb0ba95644a1e97f93d2ca8332ecece7007627f2')
number_of_items = sum(1 for _ in items)
self.assertEqual(number_of_items, 8)
def test_parse_item(self):
filename = os.path.join('data/defensa', '19-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'19/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'AURELIO COREDOR MIRANO')
self.assertEqual(item.get('time_start'), u'08:38')
self.assertEqual(item.get('institution'), u'defensa')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'43447287')
self.assertEqual(item.get('entity'), None)
self.assertEqual(item.get('reason'), u'REUNIÓN DE TRABAJO')
self.assertEqual(item.get('host_name'), u'HUGO DAVID MEJIA HUAMAN')
self.assertEqual(item.get('time_end'), None)
self.assertEqual(item.get('date'), u'2015-08-19')
self.assertEqual(item.get('sha1'), u'd9f07e3a5effd7f0b9164dfc14822c5395ed3b58')
item = next(items)
self.assertEqual(item.get('full_name'), u'LUIS ANIBAL OLIVERA SANTA CRUZ')
self.assertEqual(item.get('institution'), u'defensa')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'09392580')
self.assertEqual(item.get('entity'), u'FAP')
self.assertEqual(item.get('reason'), u'REUNIÓN DE TRABAJO')
self.assertEqual(item.get('host_name'), u'JORGE RICARDO TORRES MONTEZA')
self.assertEqual(item.get('time_start'), u'08:44')
self.assertEqual(item.get('time_end'), u'11:49')
self.assertEqual(item.get('date'), u'2015-08-19')
self.assertEqual(item.get('sha1'), u'8d0e1ee7b60b8b2ee9e26d30e708d606a8d06a45')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 13)
def test_parse_item(self):
filename = os.path.join('data/justicia', '27-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'27/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'CERON GUTIERREZ, NANCY')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'07862529')
self.assertEqual(item.get('entity'), u'PARTICULAR')
self.assertEqual(
item.get('reason'),
u'CONSULTA DE EXPEDIENTE DEL REGISTRO UNICO DE VICTIMAS')
self.assertEqual(item.get('host_name'), u'SOTO PEREZ, ADRIEL EDUARDO')
self.assertEqual(item.get('office'), u'CONSEJO DE REPARACIONES')
# self.assertEqual(item.get('meeting_place'), u'POR DEFINIR')
self.assertEqual(item.get('time_start'), u'10:12')
self.assertEqual(item.get('time_end'), u'11:12')
self.assertEqual(item.get('institution'), u'minjus')
self.assertEqual(
item.get('location'),
u'CONSEJO DE REPARACIONES - SECRETARIA TECNICA, NEISER LLACZA ARCE 158, MIRAFLORES'
)
self.assertEqual(item.get('date'), u'2015-08-27')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 30)
def test_parse_item(self):
filename = os.path.join('data/minem', '19-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'19/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'),
u'CARMEN ALICIA GUTIERREZ VELASCO')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'08241251')
self.assertEqual(item.get('entity'), u'ESLOM')
self.assertEqual(item.get('reason'), u'REUNI\xc3\u201cN DE TRABAJO')
self.assertEqual(item.get('host_name'),
u'OMAR FRANCO CHAMBERGO RODRIGUEZ')
self.assertEqual(item.get('office'),
u'DIRECCION GENERAL DE HIDROCARBUROS-N')
self.assertEqual(item.get('time_start'), u'08:01')
self.assertEqual(item.get('time_end'), u'08:17')
self.assertEqual(item.get('meeting_place'), u'OFICINA DEL FUNCIONARIO')
self.assertEqual(item.get('institution'), u'minem')
self.assertEqual(item.get('date'), u'2015-08-19')
self.assertEqual(item.get('sha1'),
u'fa5238c796089a49ed8583ce36457c30e5e58e05')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 20)
def test_parse_item(self):
filename = os.path.join('data/pcm', '18-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('date'), u'2015-08-18')
self.assertEqual(item.get('full_name'),
u'CUEVA FRANCISCO GUERRA GARCIA')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'09179830')
self.assertEqual(item.get('entity'), u'COMISION DINI')
self.assertEqual(item.get('reason'), u'MOTIVO INSTITUCIONAL')
self.assertEqual(item.get('location'), u'PALACIO')
self.assertEqual(item.get('host_name'), u'Pedro Cateriano Bellido')
self.assertEqual(
item.get('office'),
u'DESPACHO MINISTERIAL [PRESIDENTE DEL CONSEJO DE MINISTROS]')
self.assertEqual(item.get('meeting_place'), u'SALA TELLO')
self.assertEqual(item.get('time_start'), u'19:00')
self.assertEqual(item.get('time_end'), u'19:40')
self.assertEqual(item.get('institution'), u'pcm')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 15)
def test_parse_item(self):
filename = os.path.join('data/produce', '20-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'20/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'),
u'MAGUIÑA ROBLES, WILFREDO HERIBERTO')
self.assertEqual(item.get('time_start'), u'09:08:34')
self.assertEqual(item.get('institution'), u'produce')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'32824731')
self.assertEqual(item.get('office'),
u'DESPACHO VICEMINISTERIAL DE PESQUERIA')
self.assertEqual(item.get('reason'), u'ENTREVISTA')
self.assertEqual(item.get('host_name'),
u'KASTNER URIBE, MONICA CARLOTA')
self.assertEqual(item.get('time_end'), u'09:16:28')
self.assertEqual(item.get('date'), u'2015-08-20')
self.assertEqual(item.get('sha1'),
u'1edbaca51007f25bd6bd07b0025bc94309544e3e')
item = next(items)
item = next(items)
self.assertEqual(item.get('full_name'),
u'REGALO QUIJANO, WALTER MANUEL')
self.assertEqual(item.get('time_start'), u'09:16:39')
self.assertEqual(item.get('institution'), u'produce')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'08182131')
self.assertEqual(item.get('office'),
u'DESPACHO VICEMINISTERIAL DE PESQUERIA')
self.assertEqual(item.get('reason'), u'ENTREVISTA')
self.assertEqual(item.get('host_name'),
u'KASTNER URIBE, MONICA CARLOTA')
self.assertEqual(item.get('time_end'), u'11:53:16')
self.assertEqual(item.get('date'), u'2015-08-20')
self.assertEqual(item.get('sha1'),
u'd435683995c845a4a947895d3197725e255753ef')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 28)
def test_parse_item(self):
filename = os.path.join('data/ambiente', '19-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'19/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'PATRICIA ITURREGUI BYRNE')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'07231335')
self.assertEqual(item.get('entity'), u'PERSONAL')
self.assertEqual(item.get('reason'), u'OTROS')
self.assertEqual(item.get('host_name'), u'RUPERTO ANDRES TABOADA DELGADO')
self.assertEqual(item.get('office'), u'POR DEFINIR')
self.assertEqual(item.get('meeting_place'), u'POR DEFINIR')
self.assertEqual(item.get('time_start'), u'04:01:20 p.m.')
self.assertEqual(item.get('time_end'), u'09:07:00 a.m.')
self.assertEqual(item.get('institution'), u'ambiente')
self.assertEqual(item.get('date'), u'2015-08-19')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 15)
def test_parse_item(self):
filename = os.path.join('data/minvi', '18-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'CESAR MAMANI ROMERO')
self.assertEqual(item.get('time_start'), u'18:07')
self.assertEqual(item.get('institution'), u'vivienda')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'10157944')
self.assertEqual(item.get('entity'), u'TRABAJADOR - PARH')
self.assertEqual(item.get('reason'), u'REUNI\xc3\u201cN DE TRABAJO')
self.assertEqual(item.get('host_name'), u'DAYANA FARRO .')
self.assertEqual(item.get('title'), u'P.A.H.R. [OTROS]')
self.assertEqual(item.get('office'), u'PNT')
self.assertEqual(item.get('time_end'), None)
self.assertEqual(item.get('date'), u'2015-08-18')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 15)
def test_parse_item(self):
filename = os.path.join('data/osce', '18-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'Silvia Sousa Cristofol')
self.assertEqual(item.get('time_start'), u'16:38')
self.assertEqual(item.get('institution'), u'osce')
self.assertEqual(item.get('id_document'), u'CARNET DE EXTRANJERIA')
self.assertEqual(item.get('id_number'), u'000904735')
self.assertEqual(item.get('entity'), u'everis')
self.assertEqual(item.get('reason'), u'REUNIÓN DE TRABAJO')
self.assertEqual(item.get('host_name'), u'Isabel Rosario Vega Palomino')
self.assertEqual(item.get('title'), u'[Ninguno]')
self.assertEqual(item.get('office'), u'Sala de Espera')
self.assertEqual(item.get('time_end'), None)
self.assertEqual(item.get('date'), u'2015-08-18')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 15)
def test_parse_item(self):
filename = os.path.join('data/mujer', '20-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'20/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'VIGO LOPEZ BETTY CATHERINE')
self.assertEqual(item.get('time_start'), u'07:56 AM')
self.assertEqual(item.get('institution'), u'min. mujer')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'43521527')
self.assertEqual(item.get('entity'), u'SAN MIGUEL')
self.assertEqual(item.get('reason'), None)
self.assertEqual(item.get('host_name'), u'PNCVFS ,')
self.assertEqual(item.get('office'), u'PROGRAMA NACIONAL CONTRA LA VIOLENCIA FAMILIAR Y SEXUAL')
self.assertEqual(item.get('time_end'), u'10:42 AM')
self.assertEqual(item.get('date'), u'2015-08-20')
self.assertEqual(item.get('sha1'), u'a00e952857d8c86ab3877ee3805bda686bd3a999')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 20)
def test_parse_item(self):
filename = os.path.join('data/minagr', '18-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'VICTOR HUGO SEVERINO VALLE')
self.assertEqual(item.get('time_start'), u'17:04')
self.assertEqual(item.get('institution'), u'minagr')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'25856019')
self.assertEqual(item.get('entity'), u'MAKA SAC')
self.assertEqual(item.get('reason'), u'DOCUMENTOS')
self.assertEqual(item.get('host_name'), u'CARLOS ANTONIO LARA PALACIOS')
self.assertEqual(item.get('title'), u'[CONTADOR P\xc3\u0161BLICO PARA ALMAC\xc3\u2030N Y PATRIMONIO]')
self.assertEqual(item.get('office'), u'ALMACEN CENTRAL')
self.assertEqual(item.get('time_end'), u'17:23')
self.assertEqual(item.get('date'), u'2015-08-18')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 15)
def test_parse_item(self):
filename = os.path.join('data/mincu', '18-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'INGRID BARRIONUEVO ECHEGARAY')
self.assertEqual(item.get('time_start'), u'16:40')
self.assertEqual(item.get('institution'), u'mincu')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'10085172')
self.assertEqual(item.get('entity'), u'PARTICULAR')
self.assertEqual(item.get('reason'), u'REUNIÓN DE TRABAJO')
self.assertEqual(item.get('host_name'), u'JOIZ ELIZABETH DOBLADILLO ORTIZ')
self.assertEqual(item.get('title'), u'[SERVICIOS DE UN ASISTENTE EN COMUNICACIONES]')
self.assertEqual(item.get('office'), u'QHAPAQ ÑAN')
self.assertEqual(item.get('time_end'), u'16:53')
self.assertEqual(item.get('date'), u'2015-08-18')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 15)
def test_parse_item(self):
filename = os.path.join('data/minvi', '18-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'CESAR MAMANI ROMERO')
self.assertEqual(item.get('time_start'), u'18:07')
self.assertEqual(item.get('institution'), u'vivienda')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'10157944')
self.assertEqual(item.get('entity'), u'TRABAJADOR - PARH')
self.assertEqual(item.get('reason'), u'REUNI\xc3\u201cN DE TRABAJO')
self.assertEqual(item.get('host_name'), u'DAYANA FARRO .')
self.assertEqual(item.get('title'), u'P.A.H.R. [OTROS]')
self.assertEqual(item.get('office'), u'PNT')
self.assertEqual(item.get('time_end'), None)
self.assertEqual(item.get('date'), u'2015-08-18')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 15)
def test_parse_item(self):
filename = os.path.join('data/presidencia', '31-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'31/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'PAJARES CRIADO SERGIO ADOLFO')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'09272448')
self.assertEqual(item.get('entity'), u'COSAPI DATA.S.A.')
self.assertEqual(item.get('reason'), u'OTROS: PROYECTO DE CABLEADO (SUPERVISAR)')
self.assertEqual(item.get('host_name'), u'CALDERON MONTOYA BLAS')
self.assertEqual(item.get('office'), u'DIRECCION DE OPERACIONES')
self.assertEqual(item.get('time_start'), u'19:04')
self.assertEqual(item.get('time_end'), u'19:53')
self.assertEqual(item.get('institution'), u'presidencia')
self.assertEqual(item.get('meeting_place'), u'DESAMPARADOS')
self.assertEqual(item.get('date'), u'2015-08-31')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 47)
def test_parse_item(self):
filename = os.path.join('data/minsa', '18-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'MELENDEZ ARISTA GREIDY')
self.assertEqual(item.get('time_start'), u'18:45:09')
self.assertEqual(item.get('institution'), u'minsa')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'41339966')
self.assertEqual(item.get('entity'), u'DIRESA AMAZONAS')
self.assertEqual(item.get('reason'), u'TRAMITE')
self.assertEqual(item.get('host_name'), u'VELASQUEZ VALDIVIA ANIBAL')
self.assertEqual(item.get('title'), u'MINISTRO DE SALUD')
self.assertEqual(item.get('office'), u'DESPACHO MINISTERIAL')
self.assertEqual(item.get('time_end'), None)
self.assertEqual(item.get('date'), u'2015-08-18')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 20)
def test_parse_item(self):
filename = os.path.join('data/congreso', '18-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'ZEVALLOS FLOREZ, CESAR')
self.assertEqual(item.get('id_document'), u'DNI/LE')
self.assertEqual(item.get('id_number'), u'07632139')
self.assertEqual(item.get('entity'), u'EL ROCOTO')
self.assertEqual(item.get('reason'), u'PERSONAL DEL CONCESIONARIO')
self.assertEqual(item.get('host_name'), u'PEREYRA SALAZAR, WALTER')
self.assertEqual(item.get('office'), u'G.F. DE SERVICIO SOCIAL')
self.assertEqual(item.get('time_start'), u'08:15')
self.assertEqual(item.get('time_end'), None)
self.assertEqual(item.get('institution'), u'congreso')
self.assertEqual(item.get('title'), u'TECNICO ADMINISTRATIVO')
self.assertEqual(item.get('date'), u'2015-08-18')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 10)
def test_parse_item(self):
filename = os.path.join('data/ambiente', '19-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'19/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'PATRICIA ITURREGUI BYRNE')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'07231335')
self.assertEqual(item.get('entity'), u'PERSONAL')
self.assertEqual(item.get('reason'), u'OTROS')
self.assertEqual(item.get('host_name'),
u'RUPERTO ANDRES TABOADA DELGADO')
self.assertEqual(item.get('office'), u'POR DEFINIR')
self.assertEqual(item.get('meeting_place'), u'POR DEFINIR')
self.assertEqual(item.get('time_start'), u'04:01:20 p.m.')
self.assertEqual(item.get('time_end'), u'09:07:00 a.m.')
self.assertEqual(item.get('institution'), u'ambiente')
self.assertEqual(item.get('date'), u'2015-08-19')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 15)
def test_parse_item(self):
filename = os.path.join('data/pcm', '18-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('date'), u'2015-08-18')
self.assertEqual(item.get('full_name'), u'CUEVA FRANCISCO GUERRA GARCIA')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'09179830')
self.assertEqual(item.get('entity'), u'COMISION DINI')
self.assertEqual(item.get('reason'), u'MOTIVO INSTITUCIONAL')
self.assertEqual(item.get('location'), u'PALACIO')
self.assertEqual(item.get('host_name'), u'Pedro Cateriano Bellido')
self.assertEqual(item.get('office'), u'DESPACHO MINISTERIAL [PRESIDENTE DEL CONSEJO DE MINISTROS]')
self.assertEqual(item.get('meeting_place'), u'SALA TELLO')
self.assertEqual(item.get('time_start'), u'19:00')
self.assertEqual(item.get('time_end'), u'19:40')
self.assertEqual(item.get('institution'), u'pcm')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 15)
def test_parse_item(self):
filename = os.path.join('data/inpe', '19-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'19/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'MARIA DEL PILAR LLUEN MACALOPU')
self.assertEqual(item.get('time_start'), u'08:39:00')
self.assertEqual(item.get('institution'), u'inpe')
self.assertEqual(item.get('id_document'), u'DNI/LE')
self.assertEqual(item.get('id_number'), u'17434996')
self.assertEqual(item.get('entity'), u'Particular')
self.assertEqual(item.get('reason'), u'Reunion')
self.assertEqual(item.get('host_name'), u'MILAGROS MAGDALENA MU\xc3\u2018OZ GONZALES')
self.assertEqual(item.get('title'), u'---')
self.assertEqual(item.get('office'), u'Unidad De Recursos Humanos')
self.assertEqual(item.get('time_end'), u'08:54:00')
self.assertEqual(item.get('date'), u'2015-08-19')
self.assertEqual(item.get('sha1'), u'b5e65b73ec827ec50734d748e09e641b2990d679')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 60)
def test_parse_item(self):
filename = os.path.join('data/osce', '18-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'Silvia Sousa Cristofol')
self.assertEqual(item.get('time_start'), u'16:38')
self.assertEqual(item.get('institution'), u'osce')
self.assertEqual(item.get('id_document'), u'CARNET DE EXTRANJERIA')
self.assertEqual(item.get('id_number'), u'000904735')
self.assertEqual(item.get('entity'), u'everis')
self.assertEqual(item.get('reason'), u'REUNIÓN DE TRABAJO')
self.assertEqual(item.get('host_name'),
u'Isabel Rosario Vega Palomino')
self.assertEqual(item.get('title'), u'[Ninguno]')
self.assertEqual(item.get('office'), u'Sala de Espera')
self.assertEqual(item.get('time_end'), None)
self.assertEqual(item.get('date'), u'2015-08-18')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 15)
def test_parse_item(self):
filename = os.path.join('data/defensa', '19-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'19/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'),
u'HORTENCIA VANESSA GONZALES VALDIVIA')
self.assertEqual(item.get('time_start'), u'08:48')
self.assertEqual(item.get('institution'), u'defensa')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'41795231')
self.assertEqual(item.get('entity'), u'CONIDA')
self.assertEqual(item.get('reason'), u'REUNION DE TRABAJO')
self.assertEqual(item.get('host_name'),
u'DUPEYRAT LUQUE WOLFGANG CARLOS DOUGLAS')
self.assertEqual(item.get('time_end'), u'09:49')
self.assertEqual(item.get('date'), u'08/11/2017')
self.assertEqual(item.get('sha1'),
u'd7d3fc2a9a0f123473817b201dac7e651aee445a')
item = next(items)
self.assertEqual(item.get('full_name'), u'RIGOBERTO SALAS ASENCIOS')
self.assertEqual(item.get('institution'), u'defensa')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'43847220')
self.assertEqual(item.get('entity'),
u'AGENCIA DE COMPRAS DE LAS FF. AA.')
self.assertEqual(item.get('reason'), u'REUNION DE TRABAJO')
self.assertEqual(item.get('host_name'),
u'DUPEYRAT LUQUE WOLFGANG CARLOS DOUGLAS')
self.assertEqual(item.get('time_start'), u'09:00')
self.assertEqual(item.get('time_end'), u'12:35')
self.assertEqual(item.get('date'), u'08/11/2017')
self.assertEqual(item.get('sha1'),
u'eb0ba95644a1e97f93d2ca8332ecece7007627f2')
number_of_items = sum(1 for _ in items)
self.assertEqual(number_of_items, 8)
def test_parse_item(self):
filename = os.path.join('data/minem', '19-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'19/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'CARMEN ALICIA GUTIERREZ VELASCO')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'08241251')
self.assertEqual(item.get('entity'), u'ESLOM')
self.assertEqual(item.get('reason'), u'REUNI\xc3\u201cN DE TRABAJO')
self.assertEqual(item.get('host_name'), u'OMAR FRANCO CHAMBERGO RODRIGUEZ')
self.assertEqual(item.get('office'), u'DIRECCION GENERAL DE HIDROCARBUROS-N')
self.assertEqual(item.get('time_start'), u'08:01')
self.assertEqual(item.get('time_end'), u'08:17')
self.assertEqual(item.get('meeting_place'), u'OFICINA DEL FUNCIONARIO')
self.assertEqual(item.get('institution'), u'minem')
self.assertEqual(item.get('date'), u'2015-08-19')
self.assertEqual(item.get('sha1'), u'fa5238c796089a49ed8583ce36457c30e5e58e05')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 20)
def test_parse_item(self):
filename = os.path.join('data/minedu', '18-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'),
u'CARLOS MANUEL RIVERA BARDALES')
self.assertEqual(item.get('time_start'), u'17:19')
self.assertEqual(item.get('institution'), u'minedu')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'43531636')
self.assertEqual(item.get('entity'), u'PARTICULAR')
self.assertEqual(item.get('reason'), u'MOTIVO INSTITUCIONAL')
self.assertEqual(item.get('host_name'), u'BERTHA ANGELA BANICH ALLEON')
self.assertEqual(item.get('title'), u'[SECRETARIA / O IV]')
self.assertEqual(item.get('office'), u'EDIFICIO L PISO 02')
self.assertEqual(item.get('time_end'), None)
self.assertEqual(item.get('date'), u'2015-08-18')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 15)
def test_parse_item(self):
filename = os.path.join('data/minsa', '18-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'MELENDEZ ARISTA GREIDY')
self.assertEqual(item.get('time_start'), u'18:45:09')
self.assertEqual(item.get('institution'), u'minsa')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'41339966')
self.assertEqual(item.get('entity'), u'DIRESA AMAZONAS')
self.assertEqual(item.get('reason'), u'TRAMITE')
self.assertEqual(item.get('host_name'), u'VELASQUEZ VALDIVIA ANIBAL')
self.assertEqual(item.get('title'), u'MINISTRO DE SALUD')
self.assertEqual(item.get('office'), u'DESPACHO MINISTERIAL')
self.assertEqual(item.get('time_end'), None)
self.assertEqual(item.get('date'), u'2015-08-18')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 20)
def test_parse_item(self):
filename = os.path.join('data/presidencia', '31-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'31/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'RICCE CHUMBE WALTER HUMBERTO')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'25424360')
self.assertEqual(item.get('entity'), u'MINISTERIO DE AMBIENTE')
self.assertEqual(item.get('reason'), u'REUNION DE TRABAJO')
self.assertEqual(item.get('host_name'), u'MONTALVA DE FALLA JOSE')
self.assertEqual(item.get('office'), u'SUBSECRETARIA GENERAL')
self.assertEqual(item.get('title'), u'DIRECTOR GENERAL')
self.assertEqual(item.get('time_start'), u'10:50')
self.assertEqual(item.get('time_end'), u'10:54')
self.assertEqual(item.get('institution'), u'presidencia')
self.assertEqual(item.get('meeting_place'), u'EDIFICIO PALACIO')
self.assertEqual(item.get('date'), u'2015-08-31')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 12)
def test_parse_item(self):
filename = os.path.join('data/justicia', '27-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'27/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'CERON GUTIERREZ, NANCY')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'07862529')
self.assertEqual(item.get('entity'), u'PARTICULAR')
self.assertEqual(item.get('reason'), u'CONSULTA DE EXPEDIENTE DEL REGISTRO UNICO DE VICTIMAS')
self.assertEqual(item.get('host_name'), u'SOTO PEREZ, ADRIEL EDUARDO')
self.assertEqual(item.get('office'), u'CONSEJO DE REPARACIONES')
# self.assertEqual(item.get('meeting_place'), u'POR DEFINIR')
self.assertEqual(item.get('time_start'), u'10:12')
self.assertEqual(item.get('time_end'), u'11:12')
self.assertEqual(item.get('institution'), u'minjus')
self.assertEqual(item.get('location'), u'CONSEJO DE REPARACIONES - SECRETARIA TECNICA, NEISER LLACZA ARCE 158, MIRAFLORES')
self.assertEqual(item.get('date'), u'2015-08-27')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 30)
def test_parse_item(self):
filename = os.path.join('data/congreso', '18-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'ZEVALLOS FLOREZ, CESAR')
self.assertEqual(item.get('id_document'), u'DNI/LE')
self.assertEqual(item.get('id_number'), u'07632139')
self.assertEqual(item.get('entity'), u'EL ROCOTO')
self.assertEqual(item.get('reason'), u'PERSONAL DEL CONCESIONARIO')
self.assertEqual(item.get('host_name'), u'PEREYRA SALAZAR, WALTER')
self.assertEqual(item.get('office'), u'G.F. DE SERVICIO SOCIAL')
self.assertEqual(item.get('time_start'), u'08:15')
self.assertEqual(item.get('time_end'), None)
self.assertEqual(item.get('institution'), u'congreso')
self.assertEqual(item.get('title'), u'TECNICO ADMINISTRATIVO')
self.assertEqual(item.get('date'), u'2015-08-18')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 10)
def test_parse_item(self):
filename = os.path.join('data/defensa', '19-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'19/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'AURELIO COREDOR MIRANO')
self.assertEqual(item.get('time_start'), u'08:38')
self.assertEqual(item.get('institution'), u'defensa')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'43447287')
self.assertEqual(item.get('entity'), None)
self.assertEqual(item.get('reason'), u'REUNIÓN DE TRABAJO')
self.assertEqual(item.get('host_name'), u'HUGO DAVID MEJIA HUAMAN')
self.assertEqual(item.get('time_end'), None)
self.assertEqual(item.get('date'), u'2015-08-19')
self.assertEqual(item.get('sha1'),
u'd9f07e3a5effd7f0b9164dfc14822c5395ed3b58')
item = next(items)
self.assertEqual(item.get('full_name'),
u'LUIS ANIBAL OLIVERA SANTA CRUZ')
self.assertEqual(item.get('institution'), u'defensa')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'09392580')
self.assertEqual(item.get('entity'), u'FAP')
self.assertEqual(item.get('reason'), u'REUNIÓN DE TRABAJO')
self.assertEqual(item.get('host_name'),
u'JORGE RICARDO TORRES MONTEZA')
self.assertEqual(item.get('time_start'), u'08:44')
self.assertEqual(item.get('time_end'), u'11:49')
self.assertEqual(item.get('date'), u'2015-08-19')
self.assertEqual(item.get('sha1'),
u'8d0e1ee7b60b8b2ee9e26d30e708d606a8d06a45')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 13)
def test_parse_item(self):
filename = os.path.join('data/mincu', '18-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'),
u'INGRID BARRIONUEVO ECHEGARAY')
self.assertEqual(item.get('time_start'), u'16:40')
self.assertEqual(item.get('institution'), u'mincu')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'10085172')
self.assertEqual(item.get('entity'), u'PARTICULAR')
self.assertEqual(item.get('reason'), u'REUNIÓN DE TRABAJO')
self.assertEqual(item.get('host_name'),
u'JOIZ ELIZABETH DOBLADILLO ORTIZ')
self.assertEqual(item.get('title'),
u'[SERVICIOS DE UN ASISTENTE EN COMUNICACIONES]')
self.assertEqual(item.get('office'), u'QHAPAQ ÑAN')
self.assertEqual(item.get('time_end'), u'16:53')
self.assertEqual(item.get('date'), u'2015-08-18')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 15)
def test_parse_item(self):
filename = os.path.join('data/produce', '20-08-2015.html')
items = self.spider.parse(fake_response_from_file(filename, meta={'date': u'20/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'MAGUIÑA ROBLES, WILFREDO HERIBERTO')
self.assertEqual(item.get('time_start'), u'09:08:34')
self.assertEqual(item.get('institution'), u'produce')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'32824731')
self.assertEqual(item.get('office'), u'DESPACHO VICEMINISTERIAL DE PESQUERIA')
self.assertEqual(item.get('reason'), u'ENTREVISTA')
self.assertEqual(item.get('host_name'), u'KASTNER URIBE, MONICA CARLOTA')
self.assertEqual(item.get('time_end'), u'09:16:28')
self.assertEqual(item.get('date'), u'2015-08-20')
self.assertEqual(item.get('sha1'), u'1edbaca51007f25bd6bd07b0025bc94309544e3e')
item = next(items)
item = next(items)
self.assertEqual(item.get('full_name'), u'REGALO QUIJANO, WALTER MANUEL')
self.assertEqual(item.get('time_start'), u'09:16:39')
self.assertEqual(item.get('institution'), u'produce')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'08182131')
self.assertEqual(item.get('office'), u'DESPACHO VICEMINISTERIAL DE PESQUERIA')
self.assertEqual(item.get('reason'), u'ENTREVISTA')
self.assertEqual(item.get('host_name'), u'KASTNER URIBE, MONICA CARLOTA')
self.assertEqual(item.get('time_end'), u'11:53:16')
self.assertEqual(item.get('date'), u'2015-08-20')
self.assertEqual(item.get('sha1'), u'd435683995c845a4a947895d3197725e255753ef')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 28)
def test_parse_item(self):
filename = os.path.join('data/minagr', '18-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'18/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'VICTOR HUGO SEVERINO VALLE')
self.assertEqual(item.get('time_start'), u'17:04')
self.assertEqual(item.get('institution'), u'minagr')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'25856019')
self.assertEqual(item.get('entity'), u'MAKA SAC')
self.assertEqual(item.get('reason'), u'DOCUMENTOS')
self.assertEqual(item.get('host_name'),
u'CARLOS ANTONIO LARA PALACIOS')
self.assertEqual(
item.get('title'),
u'[CONTADOR P\xc3\u0161BLICO PARA ALMAC\xc3\u2030N Y PATRIMONIO]')
self.assertEqual(item.get('office'), u'ALMACEN CENTRAL')
self.assertEqual(item.get('time_end'), u'17:23')
self.assertEqual(item.get('date'), u'2015-08-18')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 15)
def test_parse_item(self):
filename = os.path.join('data/presidencia', '31-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'31/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'),
u'RICCE CHUMBE WALTER HUMBERTO')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'25424360')
self.assertEqual(item.get('entity'), u'MINISTERIO DE AMBIENTE')
self.assertEqual(item.get('reason'), u'REUNION DE TRABAJO')
self.assertEqual(item.get('host_name'), u'MONTALVA DE FALLA JOSE')
self.assertEqual(item.get('office'), u'SUBSECRETARIA GENERAL')
self.assertEqual(item.get('title'), u'DIRECTOR GENERAL')
self.assertEqual(item.get('time_start'), u'10:50')
self.assertEqual(item.get('time_end'), u'10:54')
self.assertEqual(item.get('institution'), u'presidencia')
self.assertEqual(item.get('meeting_place'), u'EDIFICIO PALACIO')
self.assertEqual(item.get('date'), u'2015-08-31')
number_of_items = 1 + sum(1 for _ in items)
self.assertEqual(number_of_items, 12)
def test_parse_item(self):
filename = os.path.join('data/mujer', '20-08-2015.html')
items = self.spider.parse(
fake_response_from_file(filename, meta={'date': u'20/08/2015'}))
item = next(items)
self.assertEqual(item.get('full_name'), u'VIGO LOPEZ BETTY CATHERINE')
self.assertEqual(item.get('time_start'), u'07:56 AM')
self.assertEqual(item.get('institution'), u'min. mujer')
self.assertEqual(item.get('id_document'), u'DNI')
self.assertEqual(item.get('id_number'), u'43521527')
self.assertEqual(item.get('entity'), u'SAN MIGUEL')
self.assertEqual(item.get('reason'), None)
self.assertEqual(item.get('host_name'), u'PNCVFS ,')
self.assertEqual(
item.get('office'),
u'PROGRAMA NACIONAL CONTRA LA VIOLENCIA FAMILIAR Y SEXUAL')
self.assertEqual(item.get('time_end'), u'10:42 AM')
self.assertEqual(item.get('date'), u'2015-08-20')
self.assertEqual(item.get('sha1'),
u'a00e952857d8c86ab3877ee3805bda686bd3a999')
number_of_items = 1 + sum(1 for x in items)
self.assertEqual(number_of_items, 20)
