def test_generator(self):
from ambry_sources.sources import GeneratorSource, SourceSpec
cache_fs = fsopendir(self.setup_temp_dir())
def gen():
yield list('abcde')
for i in range(10):
yield [i, i + 1, i + 2, i + 3, i + 4]
f = MPRowsFile(cache_fs, 'foobar').load_rows(
GeneratorSource(SourceSpec('foobar'), gen()))
self.assertEqual(1, f.info['data_start_row'])
self.assertEqual(11, f.info['data_end_row'])
self.assertEqual([0], f.info['header_rows'])
self.assertEqual(f.headers, list('abcde'))
rows = list(f.select())
self.assertEqual(len(rows), 10)
self.assertEqual(sorted(rows[0].keys()), sorted(list('abcde')))
self.assertTrue(f.is_finalized)
def test_creates_virtual_tables_for_partition_with_segment_without_errors(
self):
fs = fsopendir('temp://')
def gen():
# generate header
yield ['col1', 'col2']
# generate rows
yield [0, 0]
yield [1, 1]
mprows = MPRowsFile(fs, 'example.com/simple-0.1.3/1.mpr')
mprows.load_rows(GeneratorSource(SourceSpec('foobar'), gen()))
# create virtual tables. This should not raise an error.
#
connection = apsw.Connection(':memory:')
try:
add_partition(connection, mprows, 'vid1')
except Exception as exc:
raise AssertionError(
'partition adding unexpectadly failed with {} error.'.format(
exc))
# check selected rows
#
cursor = connection.cursor()
result = cursor.execute('SELECT * FROM {}'.format('vid1')).fetchall()
self.assertEqual(result, [(0, 0), (1, 1)])
def test_creates_virtual_tables_for_partition_with_segment_without_errors(self):
fs = fsopendir('temp://')
def gen():
# generate header
yield ['col1', 'col2']
# generate rows
yield [0, 0]
yield [1, 1]
mprows = MPRowsFile(fs, 'example.com/simple-0.1.3/1.mpr')
mprows.load_rows(GeneratorSource(SourceSpec('foobar'), gen()))
# create virtual tables. This should not raise an error.
#
connection = apsw.Connection(':memory:')
try:
add_partition(connection, mprows, 'vid1')
except Exception as exc:
raise AssertionError('partition adding unexpectadly failed with {} error.'.format(exc))
# check selected rows
#
cursor = connection.cursor()
result = cursor.execute('SELECT * FROM {}'.format('vid1')).fetchall()
self.assertEqual(result, [(0, 0), (1, 1)])
def test_spec_load(self):
"""Test that setting a SourceSpec propertly sets the header_lines data start position"""
from ambry_sources.sources import SourceSpec
import string
rs = string.ascii_letters
n = 500
rows, headers = self.generate_rows(n)
blank = ['' for e in rows[0]]
# Append a complex header, to give the RowIntuiter something to do.
rows = [
['Dataset Title'] + blank[1:],
blank,
blank,
[rs[i] for i, e in enumerate(rows[0])],
[rs[i + 1] for i, e in enumerate(rows[0])],
[rs[i + 2] for i, e in enumerate(rows[0])],
] + rows
f = MPRowsFile('mem://frh').load_rows(rows)
d = f.info
self.assertEqual(6, d['data_start_row'])
self.assertEqual(506, d['data_end_row'])
self.assertEqual([3, 4, 5], d['header_rows'])
self.assertEqual([
u('a_b_c'),
u('b_c_d'),
u('c_d_e'),
u('d_e_f'),
u('e_f_g'),
u('f_g_h')
], d['headers'])
class Rows(object):
spec = SourceSpec(None, header_lines=(3, 4), start_line=5)
def __iter__(self):
return iter(rows)
f = MPRowsFile('mem://frh').load_rows(Rows())
d = f.info
self.assertEqual(5, d['data_start_row'])
self.assertEqual(506, d['data_end_row'])
self.assertEqual([3, 4], d['header_rows'])
self.assertEqual(
[u('a_b'),
u('b_c'),
u('c_d'),
u('d_e'),
u('e_f'),
u('f_g')], d['headers'])
def write_large_blocks():
df = MPRowsFile(fs, 'foobar')
if df.exists:
df.remove()
with Timer() as t, df.writer as w:
w.headers = headers
w.insert_rows(rows)
print('MSGPack write L', float(N) / t.elapsed, w.n_rows)
def write_large_blocks():
df = MPRowsFile(fs, 'foobar')
if df.exists:
df.remove()
with Timer() as t, df.writer as w:
w.headers = headers
w.insert_rows(rows)
print('MSGPack write L', float(N) / t.elapsed, w.n_rows)
def write_small_blocks():
df = MPRowsFile(fs, 'foobar')
if df.exists:
df.remove()
with Timer() as t, df.writer as w:
for i in range(N):
w.headers = headers
w.insert_row(rows[i])
print('MSGPack write S', float(N) / t.elapsed, w.n_rows)
def write_small_blocks():
df = MPRowsFile(fs, 'foobar')
if df.exists:
df.remove()
with Timer() as t, df.writer as w:
for i in range(N):
w.headers = headers
w.insert_row(rows[i])
print('MSGPack write S', float(N) / t.elapsed, w.n_rows)
def test_load_check_headers(self):
"""Just check that all of the sources can be loaded without exceptions"""
cache_fs = fsopendir('temp://')
headers = {
'mz_with_zip_xl': [
u('id'), u('gvid'), u('renter_cost_gt_30'), u('renter_cost_gt_30_cv'),
u('owner_cost_gt_30_pct'), u('owner_cost_gt_30_pct_cv')],
'mz_no_zip': [u('id'), u('uuid'), u('int'), u('float')],
'namesu8': [u('origin_english'), u('name_english'), u('origin_native'), u('name_native')],
'sf_zip': [u('id'), u('uuid'), u('int'), u('float')],
'simple': [u('id'), u('uuid'), u('int'), u('float')],
'csv_no_csv': [u('id'), u('uuid'), u('int'), u('float')],
'mz_with_zip': [u('id'), u('uuid'), u('int'), u('float')],
'rpeople': [u('name'), u('size')],
'rent07': [
u('id'), u('gvid'), u('renter_cost_gt_30'), u('renter_cost_gt_30_cv'),
u('owner_cost_gt_30_pct'), u('owner_cost_gt_30_pct_cv')],
'simple_fixed': [u('id'), u('uuid'), u('int'), u('float')],
'altname': [u('id'), u('foo'), u('bar'), u('baz')],
'rentcsv': [
u('id'), u('gvid'), u('renter_cost_gt_30'), u('renter_cost_gt_30_cv'),
u('owner_cost_gt_30_pct'), u('owner_cost_gt_30_pct_cv')],
'renttab': [
u('id'), u('gvid'), u('renter_cost_gt_30'), u('renter_cost_gt_30_cv'),
u('owner_cost_gt_30_pct'), u('owner_cost_gt_30_pct_cv')],
'multiexcel': [
u('id'), u('gvid'), u('renter_cost_gt_30'), u('renter_cost_gt_30_cv'),
u('owner_cost_gt_30_pct'), u('owner_cost_gt_30_pct_cv')],
'rent97': [
u('id'), u('gvid'), u('renter_cost_gt_30'), u('renter_cost_gt_30_cv'),
u('owner_cost_gt_30_pct'), u('owner_cost_gt_30_pct_cv')]
}
for source_name, spec in self.sources.items():
print(source_name)
s = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
f = MPRowsFile(cache_fs, spec.name)
if f.exists:
f.remove()
f.load_rows(s)
with f.reader as r:
if spec.name in headers:
self.assertEqual(headers[spec.name], r.headers)
def test_creates_virtual_table_for_simple_fixed_mpr(self):
# build rows reader
cache_fs = fsopendir(self.setup_temp_dir())
sources = self.load_sources()
spec = sources['simple_fixed']
s = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
mprows = MPRowsFile(cache_fs, spec.name).load_rows(s)
# first make sure file not changed.
expected_names = ['id', 'uuid', 'int', 'float']
expected_types = ['int', binary_type.__name__, 'int', 'float']
self.assertEqual([x['name'] for x in mprows.reader.columns],
expected_names)
self.assertEqual([x['type'] for x in mprows.reader.columns],
expected_types)
connection = apsw.Connection(':memory:')
table = 'table1'
add_partition(connection, mprows, table)
# check all columns and some rows.
cursor = connection.cursor()
query = 'SELECT count(*) FROM {};'.format(table)
result = cursor.execute(query).fetchall()
self.assertEqual(result, [(10000, )])
with mprows.reader as r:
expected_first_row = next(iter(r)).row
# query by columns.
query = 'SELECT id, uuid, int, float FROM {} LIMIT 1;'.format(table)
result = cursor.execute(query).fetchall()
self.assertEqual(len(result), 1)
self.assertEqual(result[0], expected_first_row)
def Create(
self,
db,
modulename,
dbname,
tablename, # These argare are required by APSW
mpr_url,
*args): # These are our args.
mprows = MPRowsFile(mpr_url)
columns_types = []
column_names = []
for column in sorted(mprows.reader.columns,
key=lambda x: x['pos']):
sqlite_type = TYPE_MAP.get(column['type'])
if not sqlite_type:
raise Exception(
'Do not know how to convert {} to sql column.'.format(
column['type']))
columns_types.append('"{}" {}'.format(column['name'],
sqlite_type))
column_names.append(column['name'])
columns_types_str = ',\n'.join(columns_types)
schema = 'CREATE TABLE {}({});'.format(tablename,
columns_types_str)
return schema, Table(column_names, mprows)
def first_row_header(data_start_row=None, data_end_row=None):
# Normal Headers
f = MPRowsFile('mem://frh')
w = f.writer
w.columns = headers
for row in rows:
w.insert_row(row)
if data_start_row is not None:
w.data_start_row = data_start_row
if data_end_row is not None:
w.data_end_row = data_end_row
w.close()
self.assertEqual((u('a'), u('b'), u('c'), u('d'), u('e'), u('f')),
tuple(w.parent.reader.headers))
w.parent.reader.close()
return f
def test_ctor(self):
d = '/tmp/socrata'
from os import makedirs
from os.path import exists
from shutil import rmtree
if exists(d):
print "Make", d
rmtree(d)
makedirs(d)
cache_fs = fsopendir(d) # fsopendir(self.setup_temp_dir())
sources = self.load_sources(file_name='sources.csv')
spec = sources['facilities']
source = get_source(spec, cache_fs)
def cb(*args):
print args
mpr = MPRowsFile(cache_fs, spec.name).load_rows(source,
callback=cb,
limit=10)
def test_type_intuit(self):
from ambry_sources.intuit import TypeIntuiter
cache_fs = fsopendir(self.setup_temp_dir())
spec = self.sources['simple_fixed']
s = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
f = MPRowsFile(cache_fs, spec.name)
with f.writer as w:
w.load_rows(s)
with f.reader as r:
ti = TypeIntuiter().process_header(r.headers).run(r.rows, r.n_rows)
with f.writer as w:
w.set_types(ti)
columns = []
with f.reader as w:
for col in w.columns:
columns.append((col.pos, col.name, col.type))
expected_columns = [(1, u'id', u'int'), (2, u'uuid', u'str'),
(3, u'int', u'int'), (4, u'float', u'float')]
self.assertEqual(columns, expected_columns)
def schema_header(data_start_row=None, data_end_row=None):
# Set the schema
f = MPRowsFile('mem://sh')
w = f.writer
w.headers = ['x' + str(e) for e in range(len(headers))]
for row in rows:
w.insert_row(row)
if data_start_row is not None:
w.data_start_row = data_start_row
if data_end_row is not None:
w.data_end_row = data_end_row
w.close()
self.assertEqual(
(u('x0'), u('x1'), u('x2'), u('x3'), u('x4'), u('x5')),
tuple(w.parent.reader.headers))
w.parent.reader.close()
return f
def test_all(self):
""" Test all sources from geo_sources.csv """
cache_fs = fsopendir(self.setup_temp_dir())
sources = self.load_sources(file_name='geo_sources.csv')
for name, spec in sources.items():
if name == 'highways':
# it is already tested. Skip.
continue
source = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
# now check its load to MPRows
mpr = MPRowsFile(cache_fs, spec.name).load_rows(source)
first_row = next(iter(mpr.reader))
# Are columns recognized properly?
NAME_INDEX = 1 # which element of the column description contains name.
# Collect all names from column descriptors. Skip first elem of the schema because
# it's descriptor of column descriptor elements.
columns = [x[NAME_INDEX] for x in mpr.meta['schema'][1:]]
self.assertIn('id', columns)
self.assertIn('geometry', columns)
# Is first row valid?
self.assertEqual(len(columns), len(first_row))
def test_row_load_intuit(self):
"""Check that the sources can be loaded and analyzed without exceptions and that the
guesses for headers and start are as expected"""
from itertools import islice
cache_fs = fsopendir('temp://')
cache_fs.makedir('/mpr')
# cache_fs = fsopendir('/tmp/ritest/')
sources = self.load_sources('sources-non-std-headers.csv')
for source_name, spec in sources.items():
s = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
f = MPRowsFile(cache_fs, '/mpr/'+source_name)
if f.exists:
f.remove()
f.load_rows(s, intuit_type=False, run_stats=False, limit=500)
self.assertEqual(f.info['data_start_row'], spec.expect_start)
with f.reader as r:
# First row, marked with metadata, that is marked as a data row
m1, row1 = next(six.moves.filter(lambda e: e[0][2] == 'D', r.meta_raw))
with f.reader as r:
# First row
row2 = next(r.rows)
with f.reader as r:
# First row proxy
row3 = next(iter(r)).row
self.assertEqual(row1, row2)
self.assertEqual(row1, row3)
with f.reader as r:
raw_rows = list(islice(r.raw, None, 40))
self.assertEqual(row2, raw_rows[f.info['data_start_row']])
def test_selects_correct_rows_from_many_mprows(self):
fs = fsopendir('temp://')
header = ['col1', 'col2']
# create 3 mprows files.
#
rows1 = [(0, 0), (1, 1)]
mprows1 = MPRowsFile(fs, 'vid1')
mprows1.load_rows(self._get_generator_source(header, rows1))
rows2 = [(2, 2), (3, 3)]
mprows2 = MPRowsFile(fs, 'vid2')
mprows2.load_rows(self._get_generator_source(header, rows2))
rows3 = [(4, 4), (5, 5)]
mprows3 = MPRowsFile(fs, 'vid3')
mprows3.load_rows(self._get_generator_source(header, rows3))
# create virtual tables for all mprows
#
connection = apsw.Connection(':memory:')
add_partition(connection, mprows1, 'vid1')
add_partition(connection, mprows2, 'vid2')
add_partition(connection, mprows3, 'vid3')
# check rows of all added mprows.
#
cursor = connection.cursor()
query_tmpl = 'SELECT * FROM {};'
# check rows of the first file.
#
query = query_tmpl.format('vid1')
result = cursor.execute(query).fetchall()
self.assertEqual(result, rows1)
# check rows of the second mprows file.
#
query = query_tmpl.format('vid2')
result = cursor.execute(query).fetchall()
self.assertEqual(result, rows2)
# check rows of the third mprows file.
#
query = query_tmpl.format('vid3')
result = cursor.execute(query).fetchall()
self.assertEqual(result, rows3)
def test_full_load(self):
"""Just check that all of the sources can be loaded without exceptions"""
cache_fs = fsopendir('temp://')
for source_name, spec in self.sources.items():
s = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
f = MPRowsFile(cache_fs, spec.name)
if f.exists:
f.remove()
f.load_rows(s)
with f.reader as r:
self.assertTrue(len(r.headers) > 0)
def test_creates_foreign_data_table_for_simple_fixed_mpr(
self, fake_shares):
fake_shares.return_value = True
# build rows reader
cache_fs = fsopendir(self.setup_temp_dir())
sources = self.load_sources()
spec = sources['simple_fixed']
s = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
mprows = MPRowsFile(cache_fs, spec.name).load_rows(s)
# first make sure file was not changed.
expected_names = ['id', 'uuid', 'int', 'float']
expected_types = ['int', binary_type.__name__, 'int', 'float']
self.assertEqual(sorted([x['name'] for x in mprows.reader.columns]),
sorted(expected_names))
self.assertEqual(sorted([x['type'] for x in mprows.reader.columns]),
sorted(expected_types))
try:
# create foreign data table
PostgreSQLTestBase._create_postgres_test_db()
conn = psycopg2.connect(**PostgreSQLTestBase.pg_test_db_data)
try:
with conn.cursor() as cursor:
# we have to close opened transaction.
cursor.execute('COMMIT;')
add_partition(cursor, mprows, 'table1')
# try to query just added partition foreign data table.
with conn.cursor() as cursor:
table = 'table1'
# count all rows
query = 'SELECT count(*) FROM {}.{};'.format(
POSTGRES_PARTITION_SCHEMA_NAME, table)
cursor.execute(query)
result = cursor.fetchall()
self.assertEqual(result, [(10000, )])
# check first row
cursor.execute(
'SELECT id, uuid, int, float FROM {}.{} LIMIT 1;'.
format(POSTGRES_PARTITION_SCHEMA_NAME, table))
result = cursor.fetchall()
self.assertEqual(len(result), 1)
expected_first_row = (1,
'eb385c36-9298-4427-8925-fe09294dbd',
30, Decimal('99.734691532'))
self.assertEqual(result[0], expected_first_row)
finally:
conn.close()
finally:
PostgreSQLTestBase._drop_postgres_test_db()
def test_headers(self):
fs = fsopendir('mem://')
df = MPRowsFile(fs, 'foobar')
with df.writer as w:
schema = lambda row, col: w.meta['schema'][row][col]
w.headers = list('abcdefghi')
self.assertEqual('a', schema(1, 1))
self.assertEqual('e', schema(5, 1))
self.assertEqual('i', schema(9, 1))
for h in w.columns:
h.description = "{}-{}".format(h.pos, h.name)
self.assertEqual('1-a', schema(1, 3))
self.assertEqual('5-e', schema(5, 3))
self.assertEqual('9-i', schema(9, 3))
w.column(1).description = 'one'
w.column(2).description = 'two'
w.column('c').description = 'C'
w.column('d')['description'] = 'D'
self.assertEqual('one', schema(1, 3))
self.assertEqual('two', schema(2, 3))
self.assertEqual('C', schema(3, 3))
self.assertEqual('D', schema(4, 3))
with df.reader as r:
schema = lambda row, col: r.meta['schema'][row][col]
self.assertEqual([
u('a'),
u('b'),
u('c'),
u('d'),
u('e'),
u('f'),
u('g'),
u('h'),
u('i')
], r.headers)
self.assertEqual('one', schema(1, 3))
self.assertEqual('two', schema(2, 3))
self.assertEqual('C', schema(3, 3))
self.assertEqual('D', schema(4, 3))
def test_intuit_headers(self):
sources = self.load_sources(file_name='sources.csv')
for source_name in ['headers4', 'headers3', 'headers2', 'headers1']:
cache_fs = fsopendir(self.setup_temp_dir())
spec = sources[source_name]
f = MPRowsFile(cache_fs, spec.name) \
.load_rows(get_source(spec, cache_fs, callback=lambda x, y: (x, y)))
self.assertEqual(spec.expect_start, f.info['data_start_row'])
self.assertEqual([int(e) for e in spec.expect_headers.split(',')],
f.info['header_rows'])
def test_generator(self):
from ambry_sources.sources import GeneratorSource, SourceSpec
cache_fs = fsopendir(self.setup_temp_dir())
def gen():
yield list('abcde')
for i in range(10):
yield [i, i+1, i+2, i+3, i+4]
f = MPRowsFile(cache_fs, 'foobar').load_rows(GeneratorSource(SourceSpec('foobar'), gen()))
self.assertEqual(1, f.info['data_start_row'])
self.assertEqual(11, f.info['data_end_row'])
self.assertEqual([0], f.info['header_rows'])
self.assertEqual(f.headers, list('abcde'))
rows = list(f.select())
self.assertEqual(len(rows), 10)
self.assertEqual(sorted(rows[0].keys()), sorted(list('abcde')))
self.assertTrue(f.is_finalized)
def test_intuit_footer(self):
sources = self.load_sources(file_name='sources.csv')
for source_name in ['headers4', 'headers3', 'headers2', 'headers1']:
cache_fs = fsopendir(self.setup_temp_dir())
spec = sources[source_name]
f = MPRowsFile(cache_fs, spec.name) \
.load_rows(get_source(spec, cache_fs, callback=lambda x, y: (x, y)))
with f.reader as r:
last = list(r.rows)[-1] # islice isn't working on the reader.
print source_name, last
self.assertEqual(11999, int(last[0]))
self.assertEqual('2q080z003Cg2', last[1])
def test_highways(self):
# FIXME: Optimize to use local file instead of downloading it all the time.
cache_fs = fsopendir(self.setup_temp_dir())
sources = self.load_sources(file_name='geo_sources.csv')
spec = sources['highways']
source = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
# first check is it converted properly.
row_gen = source._get_row_gen()
first_row = next(row_gen)
# generates valid first row
self.assertEqual(len(first_row), 68)
self.assertEqual(first_row[0], 0)
# last element is wkt.
self.assertIn('LINESTRING', first_row[-1])
# header is valid
self.assertEqual(len(source._headers), 68)
self.assertEqual(source._headers[0], 'id')
self.assertEqual(source._headers[-1], 'geometry')
# now check its load to MPRows
mpr = MPRowsFile(cache_fs, spec.name).load_rows(source)
# Are columns recognized properly?
NAME_INDEX = 1 # which element of the column description contains name.
# Collect all names from column descriptors. Skip first elem of the schema because
# it's descriptor of column descriptor elements.
columns = [x[NAME_INDEX] for x in mpr.meta['schema'][1:]]
self.assertIn('id', columns)
self.assertIn('geometry', columns)
self.assertIn('length', columns) # column from shape file.
# Is first row valid?
first_row = next(iter(mpr.reader))
self.assertEqual(len(first_row), 68)
self.assertEqual(first_row['id'], 0)
self.assertIn('LINESTRING', first_row['geometry'])
return
# spec columns are properly populated
self.assertEqual(len(spec.columns), 68)
self.assertEqual(spec.columns[0]['name'], 'id')
self.assertEqual(spec.columns[-1]['name'], 'geometry')
def __init__(self, options, columns):
"""
Args:
options (dict): filesystem and path, filesystem is root directory str, path is relative
name of the file.
Example: {
'filesystem': '/tmp/my-root',
'path': '/dir1/file1.mpr'
}
"""
super(MPRForeignDataWrapper, self).__init__(options, columns)
self.columns = columns
if 'path' not in options:
log_to_postgres(
'Filename is required option of the partition msgpack fdw.',
ERROR,
hint=
'Try to add the `path` option to the table creation statement')
raise RuntimeError(
'`path` is required option of the MPR (Message Pack Rows) fdw.'
)
if 'filesystem' not in options:
log_to_postgres(
'filesystem is required option of the partition msgpack fdw.',
ERROR,
hint=
'Try to add the `filesystem` option to the table creation statement'
)
raise RuntimeError(
'`filesystem` is required option of the MPR (Message Pack Rows) fdw.'
)
self.filesystem = fsopendir(options['filesystem'])
self.path = options['path']
if logger.level == logging.DEBUG:
current_user = getpass.getuser()
log_to_postgres(
'Initializing Foreign Data Wrapper: user: {}, filesystem: {}, path: {}'
.format(current_user, options['filesystem'],
options['path']), DEBUG)
self._mp_rows = MPRowsFile(self.filesystem, self.path)
def import_source(spec, cache_fs, file_path=None, account_accessor=None):
"""Download a source and load it into an MPR file. """
s = get_source(spec, cache_fs, account_accessor)
if not file_path:
file_path = spec.name
f = MPRowsFile(cache_fs, file_path)
w = f.writer
w.set_spec(spec)
for row in s:
w.insert_row(row)
w.close()
return f
def test_stats(self):
"""Check that the sources can be loaded and analyzed without exceptions and that the
guesses for headers and start are as expected"""
#cache_fs = fsopendir('temp://')
from shutil import rmtree
from os import makedirs
tp = '/tmp/mpr-test'
rmtree(tp, ignore_errors=True)
makedirs(tp)
cache_fs = fsopendir(tp)
s = get_source(self.sources['simple_stats'],
cache_fs,
callback=lambda x, y: (x, y))
f = MPRowsFile(cache_fs, s.spec.name).load_rows(s, run_stats=True)
stat_names = ('count', 'min', 'mean', 'max', 'nuniques')
vals = {
u('str_a'): (30, None, None, None, 10),
u('str_b'): (30, None, None, None, 10),
u('float_a'): (30, 1.0, 5.5, 10.0, 10),
u('float_b'): (30, 1.1, 5.5, 9.9, 10),
u('float_c'): (30, None, None, None, 10),
u('int_b'): (30, None, None, None, 10),
u('int_a'): (30, 1.0, 5.5, 10.0, 10)
}
with f.reader as r:
for col in r.columns:
stats = (col.stat_count, col.min,
round(col.mean, 1) if col.mean else None, col.max,
col.nuniques)
for a, b, stat_name in zip(vals[col.name], stats, stat_names):
self.assertEqual(
a, b, "{} failed for stat {}: {} != {}".format(
col.name, stat_name, a, b))
def test_creates_virtual_table_for_source_with_header_containing_sql_reserved_words(
self):
# build rows reader
cache_fs = fsopendir(self.setup_temp_dir())
spec = SourceSpec('foobar')
def gen():
# yield header
yield ['create', 'index', 'where', 'select', 'distinct']
# yield rows
for i in range(10):
yield [i, i + 1, i + 2, i + 3, i + 4]
s = GeneratorSource(spec, gen())
mprows = MPRowsFile(cache_fs, spec.name).load_rows(s)
connection = apsw.Connection(':memory:')
table = 'table1'
add_partition(connection, mprows, table)
# check all columns and some rows.
cursor = connection.cursor()
query = 'SELECT count(*) FROM {};'.format(table)
result = cursor.execute(query).fetchall()
self.assertEqual(result, [(10, )])
with mprows.reader as r:
expected_first_row = next(iter(r)).row
# query by columns.
query = 'SELECT "create", "index", "where", "select", "distinct" FROM {} LIMIT 1;'.format(
table)
result = cursor.execute(query).fetchall()
self.assertEqual(len(result), 1)
self.assertEqual(result[0], expected_first_row)
def test_row_load_intuit(self):
"""Check that the sources can be loaded and analyzed without exceptions and that the
guesses for headers and start are as expected"""
from itertools import islice
cache_fs = fsopendir('temp://')
cache_fs.makedir('/mpr')
# cache_fs = fsopendir('/tmp/ritest/')
sources = self.load_sources('sources-non-std-headers.csv')
for source_name, spec in sources.items():
s = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
f = MPRowsFile(cache_fs, '/mpr/' + source_name)
if f.exists:
f.remove()
f.load_rows(s, intuit_type=False, run_stats=False, limit=500)
self.assertEqual(f.info['data_start_row'], spec.expect_start)
with f.reader as r:
# First row, marked with metadata, that is marked as a data row
m1, row1 = next(
six.moves.filter(lambda e: e[0][2] == 'D', r.meta_raw))
with f.reader as r:
# First row
row2 = next(r.rows)
with f.reader as r:
# First row proxy
row3 = next(iter(r)).row
self.assertEqual(row1, row2)
self.assertEqual(row1, row3)
with f.reader as r:
raw_rows = list(islice(r.raw, None, 40))
self.assertEqual(row2, raw_rows[f.info['data_start_row']])
def no_header(data_start_row=None, data_end_row=None):
# No header, column labels.
f = MPRowsFile('mem://nh')
w = f.writer
for row in rows:
w.insert_row(row)
if data_start_row is not None:
w.data_start_row = data_start_row
if data_end_row is not None:
w.data_end_row = data_end_row
w.close()
self.assertEqual(['col1', 'col2', 'col3', 'col4', 'col5', 'col6'],
w.parent.reader.headers)
w.parent.reader.close()
return f
def test_full_load(self):
"""Just check that all of the sources can be loaded without exceptions"""
cache_fs = fsopendir('temp://')
for source_name, spec in self.sources.items():
s = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
f = MPRowsFile(cache_fs, spec.name)
if f.exists:
f.remove()
f.load_rows(s)
with f.reader as r:
self.assertTrue(len(r.headers) > 0)
def test_bad_row_intuition(self):
from ambry_sources.mpf import MPRowsFile
from ambry_sources.sources.spec import SourceSpec
cache_fs = fsopendir('temp://')
spec = SourceSpec('http://public.source.civicknowledge.com/example.com/sources/simple-example.csv',
name='simple')
s = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
f = MPRowsFile(cache_fs, spec.name)
if f.exists:
f.remove()
f.load_rows(s)
self.assertEqual(10001, f.reader.info['data_end_row'])
def test_selects_correct_rows_from_many_mprows(self):
fs = fsopendir('temp://')
header = ['col1', 'col2']
# create 3 mprows files.
#
rows1 = [(0, 0), (1, 1)]
mprows1 = MPRowsFile(fs, 'vid1')
mprows1.load_rows(self._get_generator_source(header, rows1))
rows2 = [(2, 2), (3, 3)]
mprows2 = MPRowsFile(fs, 'vid2')
mprows2.load_rows(self._get_generator_source(header, rows2))
rows3 = [(4, 4), (5, 5)]
mprows3 = MPRowsFile(fs, 'vid3')
mprows3.load_rows(self._get_generator_source(header, rows3))
# create virtual tables for all mprows
#
connection = apsw.Connection(':memory:')
add_partition(connection, mprows1, 'vid1')
add_partition(connection, mprows2, 'vid2')
add_partition(connection, mprows3, 'vid3')
# check rows of all added mprows.
#
cursor = connection.cursor()
query_tmpl = 'SELECT * FROM {};'
# check rows of the first file.
#
query = query_tmpl.format('vid1')
result = cursor.execute(query).fetchall()
self.assertEqual(result, rows1)
# check rows of the second mprows file.
#
query = query_tmpl.format('vid2')
result = cursor.execute(query).fetchall()
self.assertEqual(result, rows2)
# check rows of the third mprows file.
#
query = query_tmpl.format('vid3')
result = cursor.execute(query).fetchall()
self.assertEqual(result, rows3)
def test_bad_row_intuition(self):
from ambry_sources.mpf import MPRowsFile
from ambry_sources.sources.spec import SourceSpec, ColumnSpec
cache_fs = fsopendir('temp://')
spec = SourceSpec('http://www2.census.gov/acs2009_1yr/summaryfile/Entire_States/Arizona.zip',
file='g2009.*\.txt',
filetype='fixed',
name='geofile',
encoding='latin1',
)
spec.columns = [ColumnSpec(position=1, width=6, name='fileid', start=1),
ColumnSpec(position=2, width=2, name='stusab', start=7),
ColumnSpec(position=3, width=3, name='sumlevel', start=9),
ColumnSpec(position=4, width=2, name='component', start=12),
ColumnSpec(position=5, width=7, name='logrecno', start=14),
ColumnSpec(position=6, width=1, name='us', start=21),
ColumnSpec(position=7, width=1, name='region', start=22),
ColumnSpec(position=8, width=1, name='division', start=23),
ColumnSpec(position=9, width=2, name='statece', start=24),
ColumnSpec(position=10, width=2, name='state', start=26),
ColumnSpec(position=11, width=3, name='county', start=28),
ColumnSpec(position=12, width=5, name='cousub', start=31),
ColumnSpec(position=13, width=5, name='place', start=36),
ColumnSpec(position=14, width=6, name='tract', start=41),
ColumnSpec(position=15, width=1, name='blkgrp', start=47),
ColumnSpec(position=16, width=5, name='concit', start=48),
ColumnSpec(position=17, width=4, name='aianhh', start=53),
ColumnSpec(position=18, width=5, name='aianhhfp', start=57),
ColumnSpec(position=19, width=1, name='aihhtli', start=62),
ColumnSpec(position=20, width=3, name='aitsce', start=63),
ColumnSpec(position=21, width=5, name='aits', start=66),
ColumnSpec(position=22, width=5, name='anrc', start=71),
ColumnSpec(position=23, width=5, name='cbsa', start=76),
ColumnSpec(position=24, width=3, name='csa', start=81),
ColumnSpec(position=25, width=5, name='metdiv', start=84),
ColumnSpec(position=26, width=1, name='macc', start=89),
ColumnSpec(position=27, width=1, name='memi', start=90),
ColumnSpec(position=28, width=5, name='necta', start=91),
ColumnSpec(position=29, width=3, name='cnecta', start=96),
ColumnSpec(position=30, width=5, name='nectadiv', start=99),
ColumnSpec(position=31, width=5, name='ua', start=104),
ColumnSpec(position=33, width=2, name='cdcurr', start=114),
ColumnSpec(position=34, width=3, name='sldu', start=116),
ColumnSpec(position=35, width=3, name='sldl', start=119),
ColumnSpec(position=39, width=5, name='submcd', start=136),
ColumnSpec(position=40, width=5, name='sdelm', start=141),
ColumnSpec(position=41, width=5, name='sdsec', start=146),
ColumnSpec(position=42, width=5, name='sduni', start=151),
ColumnSpec(position=43, width=1, name='ur', start=156),
ColumnSpec(position=44, width=1, name='pci', start=157),
ColumnSpec(position=47, width=5, name='puma5', start=169),
ColumnSpec(position=49, width=40, name='geoid', start=179),
ColumnSpec(position=50, width=200, name='name', start=219)]
s = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
f = MPRowsFile(cache_fs, spec.name)
if f.exists:
f.remove()
f.load_rows(s)
self.assertEqual(119, f.reader.info['data_end_row'])
def test_fixed(self):
cache_fs = fsopendir(self.setup_temp_dir())
spec = self.sources['simple_fixed']
s = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
f = MPRowsFile(cache_fs, spec.name).load_rows(s)
self.assertEqual(f.headers, ['id', 'uuid', 'int', 'float'])
def test_bad_row_intuition(self):
from ambry_sources.mpf import MPRowsFile
from ambry_sources.sources.spec import SourceSpec, ColumnSpec
cache_fs = fsopendir('temp://')
spec = SourceSpec(
'http://www2.census.gov/acs2009_1yr/summaryfile/Entire_States/Arizona.zip',
file='g2009.*\.txt',
filetype='fixed',
name='geofile',
encoding='latin1',
)
spec.columns = [
ColumnSpec(position=1, width=6, name='fileid', start=1),
ColumnSpec(position=2, width=2, name='stusab', start=7),
ColumnSpec(position=3, width=3, name='sumlevel', start=9),
ColumnSpec(position=4, width=2, name='component', start=12),
ColumnSpec(position=5, width=7, name='logrecno', start=14),
ColumnSpec(position=6, width=1, name='us', start=21),
ColumnSpec(position=7, width=1, name='region', start=22),
ColumnSpec(position=8, width=1, name='division', start=23),
ColumnSpec(position=9, width=2, name='statece', start=24),
ColumnSpec(position=10, width=2, name='state', start=26),
ColumnSpec(position=11, width=3, name='county', start=28),
ColumnSpec(position=12, width=5, name='cousub', start=31),
ColumnSpec(position=13, width=5, name='place', start=36),
ColumnSpec(position=14, width=6, name='tract', start=41),
ColumnSpec(position=15, width=1, name='blkgrp', start=47),
ColumnSpec(position=16, width=5, name='concit', start=48),
ColumnSpec(position=17, width=4, name='aianhh', start=53),
ColumnSpec(position=18, width=5, name='aianhhfp', start=57),
ColumnSpec(position=19, width=1, name='aihhtli', start=62),
ColumnSpec(position=20, width=3, name='aitsce', start=63),
ColumnSpec(position=21, width=5, name='aits', start=66),
ColumnSpec(position=22, width=5, name='anrc', start=71),
ColumnSpec(position=23, width=5, name='cbsa', start=76),
ColumnSpec(position=24, width=3, name='csa', start=81),
ColumnSpec(position=25, width=5, name='metdiv', start=84),
ColumnSpec(position=26, width=1, name='macc', start=89),
ColumnSpec(position=27, width=1, name='memi', start=90),
ColumnSpec(position=28, width=5, name='necta', start=91),
ColumnSpec(position=29, width=3, name='cnecta', start=96),
ColumnSpec(position=30, width=5, name='nectadiv', start=99),
ColumnSpec(position=31, width=5, name='ua', start=104),
ColumnSpec(position=33, width=2, name='cdcurr', start=114),
ColumnSpec(position=34, width=3, name='sldu', start=116),
ColumnSpec(position=35, width=3, name='sldl', start=119),
ColumnSpec(position=39, width=5, name='submcd', start=136),
ColumnSpec(position=40, width=5, name='sdelm', start=141),
ColumnSpec(position=41, width=5, name='sdsec', start=146),
ColumnSpec(position=42, width=5, name='sduni', start=151),
ColumnSpec(position=43, width=1, name='ur', start=156),
ColumnSpec(position=44, width=1, name='pci', start=157),
ColumnSpec(position=47, width=5, name='puma5', start=169),
ColumnSpec(position=49, width=40, name='geoid', start=179),
ColumnSpec(position=50, width=200, name='name', start=219)
]
s = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
f = MPRowsFile(cache_fs, spec.name)
if f.exists:
f.remove()
f.load_rows(s)
self.assertEqual(119, f.reader.info['data_end_row'])
def test_datafile_read_write(self):
from fs.opener import fsopendir
import datetime
from random import randint, random
from contexttimer import Timer
from uuid import uuid4
fs = fsopendir('mem://')
# fs = fsopendir('/tmp/pmpf')
N = 50000
# Basic read/ write tests.
def rand_date():
return datetime.date(randint(2000, 2015), randint(1, 12), 10)
def rand_datetime():
return datetime.datetime(randint(2000, 2015), randint(1, 12), 10)
def rand_time():
return datetime.time(randint(0, 23), randint(0, 59), 10)
row = lambda: (None, 1, random(), str(uuid4()), rand_date(),
rand_datetime(), rand_time())
headers = list('abcdefghi')[:len(row())]
rows = [row() for i in range(N)]
def write_large_blocks():
df = MPRowsFile(fs, 'foobar')
if df.exists:
df.remove()
with Timer() as t, df.writer as w:
w.headers = headers
w.insert_rows(rows)
print('MSGPack write L', float(N) / t.elapsed, w.n_rows)
def write_small_blocks():
df = MPRowsFile(fs, 'foobar')
if df.exists:
df.remove()
with Timer() as t, df.writer as w:
for i in range(N):
w.headers = headers
w.insert_row(rows[i])
print('MSGPack write S', float(N) / t.elapsed, w.n_rows)
print()
# Write the whole file with insert_rows() which writes all of the rows at once.
write_large_blocks()
# Write the file in blocks, with insert_rows collecting rows into a cache, then writting the
# cached blocks.
write_small_blocks()
df = MPRowsFile(fs, 'foobar')
with Timer() as t:
count = 0
i = 0
s = 0
r = df.reader
for i, row in enumerate(r):
count += 1
r.close()
print('MSGPack read ', float(N) / t.elapsed, i, count, s)
with Timer() as t:
count = 0
r = df.reader
for row in r.rows:
count += 1
r.close()
print('MSGPack rows ', float(N) / t.elapsed)
with Timer() as t:
count = 0
r = df.reader
for row in r.raw:
count += 1
r.close()
print('MSGPack raw ', float(N) / t.elapsed)
def test_load_check_headers(self):
"""Just check that all of the sources can be loaded without exceptions"""
cache_fs = fsopendir('temp://')
headers = {
'mz_with_zip_xl': [
u('id'),
u('gvid'),
u('renter_cost_gt_30'),
u('renter_cost_gt_30_cv'),
u('owner_cost_gt_30_pct'),
u('owner_cost_gt_30_pct_cv')
],
'mz_no_zip': [u('id'), u('uuid'),
u('int'), u('float')],
'namesu8': [
u('origin_english'),
u('name_english'),
u('origin_native'),
u('name_native')
],
'sf_zip': [u('id'), u('uuid'),
u('int'), u('float')],
'simple': [u('id'), u('uuid'),
u('int'), u('float')],
'csv_no_csv': [u('id'), u('uuid'),
u('int'), u('float')],
'mz_with_zip': [u('id'), u('uuid'),
u('int'), u('float')],
'rpeople': [u('name'), u('size')],
'rent07': [
u('id'),
u('gvid'),
u('renter_cost_gt_30'),
u('renter_cost_gt_30_cv'),
u('owner_cost_gt_30_pct'),
u('owner_cost_gt_30_pct_cv')
],
'simple_fixed': [u('id'), u('uuid'),
u('int'),
u('float')],
'altname': [u('id'), u('foo'),
u('bar'), u('baz')],
'rentcsv': [
u('id'),
u('gvid'),
u('renter_cost_gt_30'),
u('renter_cost_gt_30_cv'),
u('owner_cost_gt_30_pct'),
u('owner_cost_gt_30_pct_cv')
],
'renttab': [
u('id'),
u('gvid'),
u('renter_cost_gt_30'),
u('renter_cost_gt_30_cv'),
u('owner_cost_gt_30_pct'),
u('owner_cost_gt_30_pct_cv')
],
'multiexcel': [
u('id'),
u('gvid'),
u('renter_cost_gt_30'),
u('renter_cost_gt_30_cv'),
u('owner_cost_gt_30_pct'),
u('owner_cost_gt_30_pct_cv')
],
'rent97': [
u('id'),
u('gvid'),
u('renter_cost_gt_30'),
u('renter_cost_gt_30_cv'),
u('owner_cost_gt_30_pct'),
u('owner_cost_gt_30_pct_cv')
]
}
for source_name, spec in self.sources.items():
print(source_name)
s = get_source(spec, cache_fs, callback=lambda x, y: (x, y))
f = MPRowsFile(cache_fs, spec.name)
if f.exists:
f.remove()
f.load_rows(s)
with f.reader as r:
if spec.name in headers:
self.assertEqual(headers[spec.name], r.headers)
