def push_first_part_to_sql(streets, table_name, drop_prev):
"""
SOME OF THE ADJUSTMENT MADE WERE IN REGARD TO:
--WEST 49 STREET
--avenue b vs. b avenue
--WEST 160 STREET
--BENNETT AVENUE
--WADSWORTH TERRACE
--75 PARK TERRACE EAST
--MARGINAL STREET
--AVENUE OF THE AMER
--8 LITTLE WEST 12 ST
--74 PIKE SLIP
Do anything about?
CENTRAL PARK WEST -- Central Park W or Central Pk W
NORTH END AVE -- N End Ave. or North End Ave.
NEED 'cust_snd' FOR THE FOLLOWING
PLCE --> PLACE "WASHINGTON PLCE"
S --> SOUTH ST
W --> WEST ST
FREDERICK DOUGLASS B --> F.D. BLVD
"""
if drop_prev:
conn.set_isolation_level(0)
cur.execute('drop table if exists %s;' % table_name)
grps = streets.groupby('sc5')
df_cols = ['primary_name', 'variation', 'full_variation']
df = pd.DataFrame(columns=df_cols)
for k, v in grps.groups.iteritems():
t = grps.get_group(k)
non_primary_idx = t[t.primary_flag != 'P'].index.tolist()
primary_idx = t[t.index.isin(non_primary_idx) ==
False].index.tolist()
tdf = pd.DataFrame()
tdf['variation'] = t.ix[non_primary_idx, 'stname'].tolist()
tdf['full_variation'] = t.ix[non_primary_idx,
'full_stname'].tolist()
tdf['primary_name'] = t.ix[primary_idx,
'full_stname'].tolist()[0]
tdf['sc5'] = t.ix[primary_idx, 'sc5'].tolist()[0]
assert t.ix[v, 'sc5'].unique().tolist()
df = df.append(tdf, ignore_index=True)
df.to_sql(table_name, routing_eng, index=False)
return True
def push_first_part_to_sql(streets,table_name,drop_prev):
"""
SOME OF THE ADJUSTMENT MADE WERE IN REGARD TO:
--WEST 49 STREET
--avenue b vs. b avenue
--WEST 160 STREET
--BENNETT AVENUE
--WADSWORTH TERRACE
--75 PARK TERRACE EAST
--MARGINAL STREET
--AVENUE OF THE AMER
--8 LITTLE WEST 12 ST
--74 PIKE SLIP
Do anything about?
CENTRAL PARK WEST -- Central Park W or Central Pk W
NORTH END AVE -- N End Ave. or North End Ave.
NEED 'cust_snd' FOR THE FOLLOWING
PLCE --> PLACE "WASHINGTON PLCE"
S --> SOUTH ST
W --> WEST ST
FREDERICK DOUGLASS B --> F.D. BLVD
"""
if drop_prev:
conn.set_isolation_level( 0)
cur.execute( 'drop table if exists %s;' % table_name)
grps = streets.groupby('sc5')
df_cols = ['primary_name','variation','full_variation']
df = pd.DataFrame(columns=df_cols)
for k,v in grps.groups.iteritems():
t = grps.get_group(k)
non_primary_idx = t[t.primary_flag!='P'].index.tolist()
primary_idx = t[t.index.isin(non_primary_idx)==False].index.tolist()
tdf = pd.DataFrame()
tdf['variation'] = t.ix[non_primary_idx,'stname'].tolist()
tdf['full_variation'] = t.ix[non_primary_idx,'full_stname'].tolist()
tdf['primary_name'] = t.ix[primary_idx,'full_stname'].tolist()[0]
tdf['sc5'] = t.ix[primary_idx,'sc5'].tolist()[0]
assert t.ix[v,'sc5'].unique().tolist()
df = df.append(tdf,ignore_index=True)
df.to_sql(table_name,routing_eng,index=False)
return True
def load_parsed_snd_datafile_into_db(self,
table_name='snd',
drop_prev=True):
py_path.append(os_path.join(os_environ['BD'], 'html'))
from scrape_vendors import Scrape_Vendors, conn, cur, routing_eng
SV = Scrape_Vendors()
T = SV.T
def push_first_part_to_sql(streets, table_name, drop_prev):
"""
SOME OF THE ADJUSTMENT MADE WERE IN REGARD TO:
--WEST 49 STREET
--avenue b vs. b avenue
--WEST 160 STREET
--BENNETT AVENUE
--WADSWORTH TERRACE
--75 PARK TERRACE EAST
--MARGINAL STREET
--AVENUE OF THE AMER
--8 LITTLE WEST 12 ST
--74 PIKE SLIP
Do anything about?
CENTRAL PARK WEST -- Central Park W or Central Pk W
NORTH END AVE -- N End Ave. or North End Ave.
NEED 'cust_snd' FOR THE FOLLOWING
PLCE --> PLACE "WASHINGTON PLCE"
S --> SOUTH ST
W --> WEST ST
FREDERICK DOUGLASS B --> F.D. BLVD
"""
if drop_prev:
conn.set_isolation_level(0)
cur.execute('drop table if exists %s;' % table_name)
grps = streets.groupby('sc5')
df_cols = ['primary_name', 'variation', 'full_variation']
df = pd.DataFrame(columns=df_cols)
for k, v in grps.groups.iteritems():
t = grps.get_group(k)
non_primary_idx = t[t.primary_flag != 'P'].index.tolist()
primary_idx = t[t.index.isin(non_primary_idx) ==
False].index.tolist()
tdf = pd.DataFrame()
tdf['variation'] = t.ix[non_primary_idx, 'stname'].tolist()
tdf['full_variation'] = t.ix[non_primary_idx,
'full_stname'].tolist()
tdf['primary_name'] = t.ix[primary_idx,
'full_stname'].tolist()[0]
tdf['sc5'] = t.ix[primary_idx, 'sc5'].tolist()[0]
assert t.ix[v, 'sc5'].unique().tolist()
df = df.append(tdf, ignore_index=True)
df.to_sql(table_name, routing_eng, index=False)
return True
# I. NON S-TYPE RECORDS
d = pd.read_csv(SND_NON_S_PATH, index_col=0)
drop_idx = d[d.boro != 1].index.tolist()
d = d.drop(drop_idx, axis=0)
# 1. PROVE ONLY MN STREETS ARE CONSIDERED
assert len(d.boro.unique().tolist()) == 1
assert d.boro.unique().tolist()[0] == 1
# 2. Remove non-essential Geographic Feature Types (GFT)
remove_gft_features = ['B', 'C', 'J', 'O', 'R']
rem_idx = d[d.GFT.isin(remove_gft_features)].index.tolist()
d = d.drop(rem_idx, axis=0)
assert len(d[d.GFT.isin(remove_gft_features)]) == 0
# 3. PROVE ALL STREET NAMES ARE UPPER CASE
d['stname'] = d['stname'].map(lambda s: s.upper())
assert len(d[d.stname.str.match('[a-z]+')]) == 0
# 4. Remove Roadbeds (Horizontal Typology Type Code (ht_name_type_code='R')
rem_idx = d[d.ht_name_type_code == 'R'].index.tolist()
d = d.drop(rem_idx, axis=0)
assert len(d[d.ht_name_type_code == 'R']) == 0
# II. S-TYPE RECORDS
dd = pd.read_csv(SND_S_PATH, index_col=0)
drop_idx = dd[dd.boro != 1].index.tolist()
dd = dd.drop(drop_idx, axis=0)
# 1. PROVE ONLY MN STREETS ARE CONSIDERED
assert len(dd.boro.unique().tolist()) == 1
assert dd.boro.unique().tolist()[0] == 1
# 2. Remove non-essential Geographic Feature Types (GFT)
remove_features = ['B', 'C', 'J', 'O', 'R']
rem_idx = dd[dd.GFT.isin(['B', 'C', 'J', 'O', 'R'])].index.tolist()
dd = dd.drop(rem_idx, axis=0)
assert len(dd[dd.GFT.isin(['B', 'C', 'J', 'O', 'R'])]) == 0
# 3. PROVE ALL STREET NAMES ARE UPPER CASE
dd['stname'] = dd['stname'].map(lambda s: s.upper())
assert len(dd[dd.stname.str.match('[a-z]+')]) == 0
# 4. Remove non-essential Geographic Feature Types (GFT) from progenitors [progen_gft_1=='Z']
remove_gft_features = ['Z']
rem_idx = dd[dd.progen_gft_1 == 'Z'].index.tolist()
dd = dd.drop(rem_idx, axis=0)
assert len(dd[dd.progen_gft_1 == 'Z']) == 0
##
# START STREET DATAFRAME
##
# 1. Take First Part of Data from non-type-S records
streets = d.copy()
# PROVE ALL NAP'S WERE REMOVED
rem_idx = streets[streets.GFT.isin(['N', 'X'])].index.tolist()
streets = streets.drop(rem_idx, axis=0)
assert len(streets[streets.GFT.isin(['N', 'X'])]) == 0
# print len(dd),'initial rows from S-Type records'
# 2. Supplement with Data from type-S records
uniq_street_sc5 = streets.sc5.unique().tolist()
nd = dd[(dd.sc5_1.isin(uniq_street_sc5)) |
(dd.sc5_1.isin(uniq_street_sc5))].index.tolist()
ndf = dd.ix[nd, :].copy()
# print len(ndf),'remaining rows from S-Type records after taking only matching sc5'
# - Remove Blank Columns from Supplemental Data
remove_cols = []
# ---- PROVE THAT OK TO REMOVE 'progen_gft_1' b/c NO VALUES EXIST
test_col = 'progen_gft_1'
t = ndf[test_col].unique().tolist()
assert True == (len(t) == 1) == (np.float(t[0]).is_integer() == False)
remove_cols.append(test_col)
# --
ndf = ndf.drop(remove_cols, axis=1)
# print len(ndf),'remaining rows before push'
##
# PUSH TO SQL
##
push_first_part_to_sql(streets, table_name, drop_prev)
if drop_prev:
conn.set_isolation_level(0)
cur.execute('drop table if exists %(tmp_tbl)s;' %
{'tmp_tbl': table_name + '_tmp'})
ndf.to_sql(table_name + '_tmp', routing_eng, index=False)
return
# PG SQL CMDS...
cmd = """
alter table snd
add column east boolean default false,
add column south boolean default false,
add column west boolean default false,
add column north boolean default false,
add column sc5_2 bigint,
add column stname_grp text[];
-- 276 distinct sc5_1 in _tmp
-- 221 rows for below (276 without regex exclusions)
update snd _orig set stname_grp = name_grp
from
(select array_agg(t.stname) name_grp,f2.s_sc5 s_sc5
from
(select array_agg(distinct a.variation) orig_variations from snd a) as f3,
snd_tmp t,
(select distinct s.sc5_1 s_sc5 from snd_tmp s) as f2
where t.sc5_1 = f2.s_sc5
and not (orig_variations && array[t.stname] )
and not (t.stname ilike '%roadbed%' or t.stname ilike '%EXTENSION%'
or t.stname ilike '%PEDESTRIAN%' or t.stname ilike '%SIDE HW%' )
group by f2.s_sc5) as f1
where s_sc5 = _orig.sc5::bigint;
-- 454 rows in snd with non-null stname_grp
insert into snd (variation,primary_name,sc5)
select variation,primary_name,sc5
from
(select distinct unnest(n.stname_grp) variation,
n.primary_name primary_name,
n.sc5 sc5 from snd n) as f1,
(select array_agg(full_variation) all_full_varies,
array_agg(variation) all_varies,
array_agg(primary_name) all_primaries from snd t) as f2
where not (all_full_varies && array[variation] OR all_varies && array[variation] OR all_primaries && array[variation]);
-- ASSERT -- res==True
--select all_vars=uniq_vars res
--from
-- (select count(n1.variation) all_vars from snd n1 where n1.variation is not null or n1.variation !='') as f1,
-- (select count(distinct n2.variation) uniq_vars from snd n2 where n2.variation is not null or n2.variation !='') as f2;
update snd n set east = true
from
(select t.progen_word_1 t_progen_word_1,
t.progen_word_2 t_progen_word_2,
t.sc5_1 t_sc5_1,
t.sc5_2 t_sc5_2
from snd_tmp t) as f1
where ( (n.sc5 = t_sc5_1 or n.sc5 = t_sc5_2) OR (n.sc5_2 = t_sc5_1 or n.sc5_2 = t_sc5_2) )
and (t_progen_word_1 = 'E' or t_progen_word_2 = 'E');
update snd n set west = true
from
(select t.progen_word_1 t_progen_word_1,
t.progen_word_2 t_progen_word_2,
t.sc5_1 t_sc5_1,
t.sc5_2 t_sc5_2
from snd_tmp t) as f1
where ( (n.sc5 = t_sc5_1 or n.sc5 = t_sc5_2) OR (n.sc5_2 = t_sc5_1 or n.sc5_2 = t_sc5_2) )
and (t_progen_word_1 = 'W' or t_progen_word_2 = 'W');
-- ASSERT -- len({below}) == 0
--select progen_word_1 from snd_tmp where progen_word_1 ilike 'w';
-- ASSERT -- len({below}) == 0
--select progen_word_2 from snd_tmp where progen_word_2 ilike 'e';
update snd set primary_name = regexp_replace(primary_name,
'^(EAST|WEST)([\s]+)([0-9]+)\s(.*)$','\\1 \\3 \\4','g');
update snd set full_variation = regexp_replace(full_variation,
'^(EAST|WEST)([\s]+)([0-9]+)\s(.*)$','\\1 \\3 \\4','g');
update snd set variation = regexp_replace(variation,
'^(EAST|WEST)([\s]+)([0-9]+)\s(.*)$','\\1 \\3 \\4','g');
update snd set full_variation = regexp_replace(full_variation,
'^(TRANSVRS|CPE|CPW|DOUGLASS|RIIS|RISS|NY|PATH|VLADECK|NEW)([\s]+)([0-9]+)\s(.*)$','\\1 \\3 \\4','g');
update snd set variation = regexp_replace(variation,
'^(TRANSVRS|CPE|CPW|DOUGLASS|RIIS|RISS|NY|PATH|VLADECK|NEW)([\s]+)([0-9]+)\s(.*)$','\\1 \\3 \\4','g');
update snd set primary_name = 'FDR DRIVE' where primary_name = 'F D R DRIVE';
-- ASSERT -- len({below}) == 0
--select count(*)=0 res from snd where variation!=full_variation;
alter table snd drop column if exists full_variation;
--PROVE ALL STREETNAMES AND STREET IDS ARE FOUND IN 'snd'
-- ASSERT -- res==True
--select count(*)=0 res from
-- snd_tmp t,
-- (select array_agg(n.sc5) all_sc5 from snd n) as f1,
-- (select array_agg(n2.sc5_2) all_sc5_2 from snd n2) as f2,
-- (select array_agg(n3.variation) all_names from snd n3) as f3
--where NOT ( t.sc5_1::bigint = ANY (all_sc5) OR t.sc5_1::bigint = ANY (all_sc5_2) )
--and NOT ( t.sc5_2::bigint = ANY (all_sc5) OR t.sc5_2::bigint = ANY (all_sc5_2) )
--and NOT ( t.stname = ANY (all_names) );
drop table if exists snd_tmp;
-- ASSERT
--select count(distinct sc5_2)=0 res from snd
alter table snd drop column sc5_2;
update snd set last_updated = 'now'::timestamp with time zone;
update snd s set primary_name = regexp_replace(s.primary_name,
'([a-zA-Z0-9])[\s\s]+([a-zA-Z0-9]*)','\\1 \\2','g')
drop table if exists tmp_snd;
insert into snd (variation,sc5,west,east)
select distinct on (s1.primary_name) s1.primary_name variation,s1.sc5 sc5,s1.west west,s1.east east
from
snd s1,
(select array_agg(s2.variation) all_variations from snd s2) as f1
where not s1.primary_name = ANY (all_variations);
--PROVE THAT ALL stname_grp NAMES ARE IN VARIATION COLUMN
--select count(*)=0 res from
-- (select unnest(s.stname_grp) all_grp_names from snd s where s.stname_grp is not null) as f1,
-- (select array_agg(n3.variation) all_names from snd n3) as f3
--where NOT ( all_grp_names = ANY (all_names) );
alter table snd drop column stname_grp;
alter table snd add column tmp bigint;
update snd set tmp = sc5::bigint;
alter table snd drop column sc5;
alter table snd rename column tmp to sc5;
update snd s set
primary_name = variation,
last_updated = 'now'::timestamp with time zone
where last_updated is null;
"""
conn.set_isolation_level(0)
cur.execute(cmd)
d = pd.read_sql("select * from snd", routing_eng)
d = d.sort('primary_name').reset_index(drop=True)
l_funct = lambda s: 0 if len(str(s).strip()) == 0 else int(s)
d['sc5'] = d.sc5.map(l_funct)
cols = ['pattern', 'repl', '_flags']
ndf = pd.DataFrame(columns=cols)
grp = d.groupby('primary_name')
for k, v in grp.groups.iteritems():
patt = '(' + ' | '.join(d.ix[v, 'variation'].tolist()) + ')'
ndf = ndf.append(dict(zip(cols, [patt, k, 'g'])),
ignore_index=True)
ndf['repl'] = ndf.repl.str.replace(
r'([a-zA-Z0-9]*)([\s\s]+)([a-zA-Z0-9]*)', r'\g<1> \g<3>')
ndf.to_sql('tmp_snd', routing_eng, index=False)
a = """
select s.variation,s.primary_name
from snd s,(select array_agg(t.address) all_addr from pluto p where p.geom is null) as f1
where s.variation = ANY (all_addr);
select sl_addr
from
(select array_agg(s.variation) all_variations from snd s) as f1,
(select sl.address sl_addr from seamless sl where geom is null) as f2
where sl_addr = ANY (all_variations);
"""
# LOTS OF COMMANDS SHOULD BE HERE (FROM TAIL PART OF LONGSTRING ABOVE)
# PROVE THAT TABLE IS IN ORIGINAL CONDITION
saved_col_type_d_snd = {
u'east': u'boolean',
u'last_updated': u'timestamp with time zone',
u'primary_name': u'text',
u'sc5': u'bigint',
u'uid': u'integer',
u'variation': u'text',
u'west': u'boolean'
}
x = pd.read_sql(
""" select column_name, data_type
from INFORMATION_SCHEMA.COLUMNS
where table_name = 'snd'""", routing_eng)
col_type_d = dict(zip(x.column_name.tolist(), x.data_type.tolist()))
assert col_type_d == saved_col_type_d_snd
return
def load_parsed_snd_datafile_into_db(self,table_name='snd',drop_prev=True):
py_path.append(os_path.join(os_environ['BD'],'html'))
from scrape_vendors import Scrape_Vendors,conn,cur,routing_eng
SV = Scrape_Vendors()
T = SV.T
def push_first_part_to_sql(streets,table_name,drop_prev):
"""
SOME OF THE ADJUSTMENT MADE WERE IN REGARD TO:
--WEST 49 STREET
--avenue b vs. b avenue
--WEST 160 STREET
--BENNETT AVENUE
--WADSWORTH TERRACE
--75 PARK TERRACE EAST
--MARGINAL STREET
--AVENUE OF THE AMER
--8 LITTLE WEST 12 ST
--74 PIKE SLIP
Do anything about?
CENTRAL PARK WEST -- Central Park W or Central Pk W
NORTH END AVE -- N End Ave. or North End Ave.
NEED 'cust_snd' FOR THE FOLLOWING
PLCE --> PLACE "WASHINGTON PLCE"
S --> SOUTH ST
W --> WEST ST
FREDERICK DOUGLASS B --> F.D. BLVD
"""
if drop_prev:
conn.set_isolation_level( 0)
cur.execute( 'drop table if exists %s;' % table_name)
grps = streets.groupby('sc5')
df_cols = ['primary_name','variation','full_variation']
df = pd.DataFrame(columns=df_cols)
for k,v in grps.groups.iteritems():
t = grps.get_group(k)
non_primary_idx = t[t.primary_flag!='P'].index.tolist()
primary_idx = t[t.index.isin(non_primary_idx)==False].index.tolist()
tdf = pd.DataFrame()
tdf['variation'] = t.ix[non_primary_idx,'stname'].tolist()
tdf['full_variation'] = t.ix[non_primary_idx,'full_stname'].tolist()
tdf['primary_name'] = t.ix[primary_idx,'full_stname'].tolist()[0]
tdf['sc5'] = t.ix[primary_idx,'sc5'].tolist()[0]
assert t.ix[v,'sc5'].unique().tolist()
df = df.append(tdf,ignore_index=True)
df.to_sql(table_name,routing_eng,index=False)
return True
# I. NON S-TYPE RECORDS
d = pd.read_csv(SND_NON_S_PATH,index_col=0)
drop_idx = d[d.boro!=1].index.tolist()
d = d.drop(drop_idx,axis=0)
# 1. PROVE ONLY MN STREETS ARE CONSIDERED
assert len(d.boro.unique().tolist())==1
assert d.boro.unique().tolist()[0]==1
# 2. Remove non-essential Geographic Feature Types (GFT)
remove_gft_features = ['B','C','J','O','R']
rem_idx = d[d.GFT.isin(remove_gft_features)].index.tolist()
d = d.drop(rem_idx,axis=0)
assert len(d[d.GFT.isin(remove_gft_features)])==0
# 3. PROVE ALL STREET NAMES ARE UPPER CASE
d['stname'] = d['stname'].map(lambda s: s.upper())
assert len(d[d.stname.str.match('[a-z]+')])==0
# 4. Remove Roadbeds (Horizontal Typology Type Code (ht_name_type_code='R')
rem_idx = d[d.ht_name_type_code=='R'].index.tolist()
d = d.drop(rem_idx,axis=0)
assert len(d[d.ht_name_type_code=='R'])==0
# II. S-TYPE RECORDS
dd = pd.read_csv(SND_S_PATH,index_col=0)
drop_idx = dd[dd.boro!=1].index.tolist()
dd = dd.drop(drop_idx,axis=0)
# 1. PROVE ONLY MN STREETS ARE CONSIDERED
assert len(dd.boro.unique().tolist())==1
assert dd.boro.unique().tolist()[0]==1
# 2. Remove non-essential Geographic Feature Types (GFT)
remove_features = ['B','C','J','O','R']
rem_idx = dd[dd.GFT.isin(['B','C','J','O','R'])].index.tolist()
dd = dd.drop(rem_idx,axis=0)
assert len(dd[dd.GFT.isin(['B','C','J','O','R'])])==0
# 3. PROVE ALL STREET NAMES ARE UPPER CASE
dd['stname'] = dd['stname'].map(lambda s: s.upper())
assert len(dd[dd.stname.str.match('[a-z]+')])==0
# 4. Remove non-essential Geographic Feature Types (GFT) from progenitors [progen_gft_1=='Z']
remove_gft_features = ['Z']
rem_idx = dd[dd.progen_gft_1=='Z'].index.tolist()
dd = dd.drop(rem_idx,axis=0)
assert len(dd[dd.progen_gft_1=='Z'])==0
##
# START STREET DATAFRAME
##
# 1. Take First Part of Data from non-type-S records
streets = d.copy()
# PROVE ALL NAP'S WERE REMOVED
rem_idx = streets[streets.GFT.isin(['N','X'])].index.tolist()
streets = streets.drop(rem_idx,axis=0)
assert len(streets[streets.GFT.isin(['N','X'])])==0
# print len(dd),'initial rows from S-Type records'
# 2. Supplement with Data from type-S records
uniq_street_sc5 = streets.sc5.unique().tolist()
nd = dd[(dd.sc5_1.isin(uniq_street_sc5))|(dd.sc5_1.isin(uniq_street_sc5))].index.tolist()
ndf = dd.ix[nd,:].copy()
# print len(ndf),'remaining rows from S-Type records after taking only matching sc5'
# - Remove Blank Columns from Supplemental Data
remove_cols = []
# ---- PROVE THAT OK TO REMOVE 'progen_gft_1' b/c NO VALUES EXIST
test_col='progen_gft_1'
t=ndf[test_col].unique().tolist()
assert True == (len(t)==1) == (np.float(t[0]).is_integer()==False)
remove_cols.append(test_col)
# --
ndf = ndf.drop(remove_cols,axis=1)
# print len(ndf),'remaining rows before push'
##
# PUSH TO SQL
##
push_first_part_to_sql(streets,table_name,drop_prev)
if drop_prev:
conn.set_isolation_level( 0)
cur.execute( 'drop table if exists %(tmp_tbl)s;' % {'tmp_tbl' : table_name+'_tmp'})
ndf.to_sql( table_name+'_tmp',routing_eng,index=False)
return
# PG SQL CMDS...
cmd = """
alter table snd
add column east boolean default false,
add column south boolean default false,
add column west boolean default false,
add column north boolean default false,
add column sc5_2 bigint,
add column stname_grp text[];
-- 276 distinct sc5_1 in _tmp
-- 221 rows for below (276 without regex exclusions)
update snd _orig set stname_grp = name_grp
from
(select array_agg(t.stname) name_grp,f2.s_sc5 s_sc5
from
(select array_agg(distinct a.variation) orig_variations from snd a) as f3,
snd_tmp t,
(select distinct s.sc5_1 s_sc5 from snd_tmp s) as f2
where t.sc5_1 = f2.s_sc5
and not (orig_variations && array[t.stname] )
and not (t.stname ilike '%roadbed%' or t.stname ilike '%EXTENSION%'
or t.stname ilike '%PEDESTRIAN%' or t.stname ilike '%SIDE HW%' )
group by f2.s_sc5) as f1
where s_sc5 = _orig.sc5::bigint;
-- 454 rows in snd with non-null stname_grp
insert into snd (variation,primary_name,sc5)
select variation,primary_name,sc5
from
(select distinct unnest(n.stname_grp) variation,
n.primary_name primary_name,
n.sc5 sc5 from snd n) as f1,
(select array_agg(full_variation) all_full_varies,
array_agg(variation) all_varies,
array_agg(primary_name) all_primaries from snd t) as f2
where not (all_full_varies && array[variation] OR all_varies && array[variation] OR all_primaries && array[variation]);
-- ASSERT -- res==True
--select all_vars=uniq_vars res
--from
-- (select count(n1.variation) all_vars from snd n1 where n1.variation is not null or n1.variation !='') as f1,
-- (select count(distinct n2.variation) uniq_vars from snd n2 where n2.variation is not null or n2.variation !='') as f2;
update snd n set east = true
from
(select t.progen_word_1 t_progen_word_1,
t.progen_word_2 t_progen_word_2,
t.sc5_1 t_sc5_1,
t.sc5_2 t_sc5_2
from snd_tmp t) as f1
where ( (n.sc5 = t_sc5_1 or n.sc5 = t_sc5_2) OR (n.sc5_2 = t_sc5_1 or n.sc5_2 = t_sc5_2) )
and (t_progen_word_1 = 'E' or t_progen_word_2 = 'E');
update snd n set west = true
from
(select t.progen_word_1 t_progen_word_1,
t.progen_word_2 t_progen_word_2,
t.sc5_1 t_sc5_1,
t.sc5_2 t_sc5_2
from snd_tmp t) as f1
where ( (n.sc5 = t_sc5_1 or n.sc5 = t_sc5_2) OR (n.sc5_2 = t_sc5_1 or n.sc5_2 = t_sc5_2) )
and (t_progen_word_1 = 'W' or t_progen_word_2 = 'W');
-- ASSERT -- len({below}) == 0
--select progen_word_1 from snd_tmp where progen_word_1 ilike 'w';
-- ASSERT -- len({below}) == 0
--select progen_word_2 from snd_tmp where progen_word_2 ilike 'e';
update snd set primary_name = regexp_replace(primary_name,
'^(EAST|WEST)([\s]+)([0-9]+)\s(.*)$','\\1 \\3 \\4','g');
update snd set full_variation = regexp_replace(full_variation,
'^(EAST|WEST)([\s]+)([0-9]+)\s(.*)$','\\1 \\3 \\4','g');
update snd set variation = regexp_replace(variation,
'^(EAST|WEST)([\s]+)([0-9]+)\s(.*)$','\\1 \\3 \\4','g');
update snd set full_variation = regexp_replace(full_variation,
'^(TRANSVRS|CPE|CPW|DOUGLASS|RIIS|RISS|NY|PATH|VLADECK|NEW)([\s]+)([0-9]+)\s(.*)$','\\1 \\3 \\4','g');
update snd set variation = regexp_replace(variation,
'^(TRANSVRS|CPE|CPW|DOUGLASS|RIIS|RISS|NY|PATH|VLADECK|NEW)([\s]+)([0-9]+)\s(.*)$','\\1 \\3 \\4','g');
update snd set primary_name = 'FDR DRIVE' where primary_name = 'F D R DRIVE';
-- ASSERT -- len({below}) == 0
--select count(*)=0 res from snd where variation!=full_variation;
alter table snd drop column if exists full_variation;
--PROVE ALL STREETNAMES AND STREET IDS ARE FOUND IN 'snd'
-- ASSERT -- res==True
--select count(*)=0 res from
-- snd_tmp t,
-- (select array_agg(n.sc5) all_sc5 from snd n) as f1,
-- (select array_agg(n2.sc5_2) all_sc5_2 from snd n2) as f2,
-- (select array_agg(n3.variation) all_names from snd n3) as f3
--where NOT ( t.sc5_1::bigint = ANY (all_sc5) OR t.sc5_1::bigint = ANY (all_sc5_2) )
--and NOT ( t.sc5_2::bigint = ANY (all_sc5) OR t.sc5_2::bigint = ANY (all_sc5_2) )
--and NOT ( t.stname = ANY (all_names) );
drop table if exists snd_tmp;
-- ASSERT
--select count(distinct sc5_2)=0 res from snd
alter table snd drop column sc5_2;
update snd set last_updated = 'now'::timestamp with time zone;
update snd s set primary_name = regexp_replace(s.primary_name,
'([a-zA-Z0-9])[\s\s]+([a-zA-Z0-9]*)','\\1 \\2','g')
drop table if exists tmp_snd;
insert into snd (variation,sc5,west,east)
select distinct on (s1.primary_name) s1.primary_name variation,s1.sc5 sc5,s1.west west,s1.east east
from
snd s1,
(select array_agg(s2.variation) all_variations from snd s2) as f1
where not s1.primary_name = ANY (all_variations);
--PROVE THAT ALL stname_grp NAMES ARE IN VARIATION COLUMN
--select count(*)=0 res from
-- (select unnest(s.stname_grp) all_grp_names from snd s where s.stname_grp is not null) as f1,
-- (select array_agg(n3.variation) all_names from snd n3) as f3
--where NOT ( all_grp_names = ANY (all_names) );
alter table snd drop column stname_grp;
alter table snd add column tmp bigint;
update snd set tmp = sc5::bigint;
alter table snd drop column sc5;
alter table snd rename column tmp to sc5;
update snd s set
primary_name = variation,
last_updated = 'now'::timestamp with time zone
where last_updated is null;
"""
conn.set_isolation_level( 0)
cur.execute( cmd)
d = pd.read_sql("select * from snd",routing_eng)
d = d.sort('primary_name').reset_index(drop=True)
l_funct = lambda s: 0 if len(str(s).strip())==0 else int(s)
d['sc5']=d.sc5.map(l_funct)
cols = ['pattern','repl','_flags']
ndf = pd.DataFrame(columns=cols)
grp = d.groupby('primary_name')
for k,v in grp.groups.iteritems():
patt = '('+' | '.join(d.ix[v,'variation'].tolist()) + ')'
ndf = ndf.append(dict(zip(cols,[patt,k,'g'])),ignore_index=True)
ndf['repl'] = ndf.repl.str.replace(r'([a-zA-Z0-9]*)([\s\s]+)([a-zA-Z0-9]*)',r'\g<1> \g<3>')
ndf.to_sql('tmp_snd',routing_eng,index=False)
a="""
select s.variation,s.primary_name
from snd s,(select array_agg(t.address) all_addr from pluto p where p.geom is null) as f1
where s.variation = ANY (all_addr);
select sl_addr
from
(select array_agg(s.variation) all_variations from snd s) as f1,
(select sl.address sl_addr from seamless sl where geom is null) as f2
where sl_addr = ANY (all_variations);
"""
# LOTS OF COMMANDS SHOULD BE HERE (FROM TAIL PART OF LONGSTRING ABOVE)
# PROVE THAT TABLE IS IN ORIGINAL CONDITION
saved_col_type_d_snd = {u'east': u'boolean',
u'last_updated': u'timestamp with time zone',
u'primary_name': u'text',
u'sc5': u'bigint',
u'uid': u'integer',
u'variation': u'text',
u'west': u'boolean'}
x=pd.read_sql(""" select column_name, data_type
from INFORMATION_SCHEMA.COLUMNS
where table_name = 'snd'""",routing_eng)
col_type_d = dict(zip(x.column_name.tolist(),x.data_type.tolist()))
assert col_type_d==saved_col_type_d_snd
return
