-
Notifications
You must be signed in to change notification settings - Fork 66
/
load-gnaf.py
1002 lines (793 loc) · 44.1 KB
/
load-gnaf.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
#!/usr/bin/env python
# *********************************************************************************************************************
# load-gnaf.py
# *********************************************************************************************************************
#
# A script for loading raw GNAF & Geoscape Admin boundaries and creating flattened, complete, easy to use versions
#
# Author: Hugh Saalmans
# GitHub: minus34
# Twitter: @minus34
#
# Copyright:
# - Code is licensed under an Apache License, version 2.0
# - Data is copyright Geoscape - licensed under a Creative Commons (By Attribution) license.
# See http://data.gov.au for the correct attribution to use
#
# Process:
# 1. Loads raw GNAF into Postgres from PSV files, using COPY
# 2. Loads raw Geoscape Admin Boundaries from Shapefiles into Postgres using shp2pgsql (part of PostGIS)
# 3. Creates flattened and simplified GNAF tables containing all relevant data
# 4. Creates a ready to use Locality Boundaries table containing a number of fixes to overcome known data issues
# 5. Splits the locality boundary for Melbourne into 2, one for each of its postcodes (3000 & 3004)
# 6. Creates final principal & alias address tables containing fixes based on the above locality customisations
# 7. Creates an almost correct Postcode Boundary table from locality boundary aggregates with address based postcodes
# 8. Adds primary and foreign keys to confirm data integrity across the output tables
#
# *********************************************************************************************************************
import os
import psycopg
import logging.config
import geoscape
import settings # gets global vars and runtime arguments
from datetime import datetime
def main():
full_start_time = datetime.now()
# log Python and OS versions
logger.info(f"\t- running Python {settings.python_version} with psycopg {settings.psycopg_version}")
logger.info(f"\t- on {settings.os_version}")
# get Postgres connection & cursor
pg_conn = psycopg.connect(settings.pg_connect_string)
pg_conn.autocommit = True
pg_cur = pg_conn.cursor()
# add postgis to database (in the public schema) - run this in a try to confirm db user has privileges
try:
pg_cur.execute("SET search_path = public, pg_catalog; CREATE EXTENSION IF NOT EXISTS postgis")
except psycopg.Error:
logger.fatal("Unable to add PostGIS extension\nACTION: Check your Postgres user privileges or PostGIS install")
return False
# test if ST_Subdivide exists (only in PostGIS 2.2+). It's used to split boundaries for faster processing
logger.info(f"\t- using Postgres {settings.pg_version} and PostGIS {settings.postgis_version} "
f"(with GEOS {settings.geos_version})")
# log the user's input parameters
logger.info("")
logger.info("Arguments")
for arg in vars(settings.args):
value = getattr(settings.args, arg)
if value is not None:
if arg != "pgpassword":
logger.info(f"\t- {arg} : {value}")
else:
logger.info(f"\t- {arg} : ************")
# START LOADING DATA
# PART 1 - create new schemas
logger.info("")
start_time = datetime.now()
logger.info(f"Part 1 of 6 : Create schemas : {start_time}")
if settings.raw_gnaf_schema != "public":
pg_cur.execute(f"CREATE SCHEMA IF NOT EXISTS {settings.raw_gnaf_schema} AUTHORIZATION {settings.pg_user}")
if settings.raw_admin_bdys_schema != "public":
pg_cur.execute(f"CREATE SCHEMA IF NOT EXISTS {settings.raw_admin_bdys_schema} AUTHORIZATION {settings.pg_user}")
if settings.admin_bdys_schema != "public":
pg_cur.execute(f"CREATE SCHEMA IF NOT EXISTS {settings.admin_bdys_schema} AUTHORIZATION {settings.pg_user}")
if settings.gnaf_schema != "public":
pg_cur.execute(f"CREATE SCHEMA IF NOT EXISTS {settings.gnaf_schema} AUTHORIZATION {settings.pg_user}")
logger.info(f"Part 1 of 6 : Schemas created! : {datetime.now() - start_time}")
# PART 2 - load gnaf from PSV files
logger.info("")
start_time = datetime.now()
logger.info(f"Part 2 of 6 : Start raw GNAF load : {start_time}")
drop_tables_and_vacuum_db(pg_cur)
create_raw_gnaf_tables(pg_cur)
populate_raw_gnaf(pg_cur)
clean_authority_files(pg_cur, settings.raw_gnaf_schema, False)
index_raw_gnaf()
if settings.primary_foreign_keys:
create_primary_foreign_keys()
else:
logger.info("\t- Step 6 of 7 : primary & foreign keys NOT created")
analyse_raw_gnaf_tables(pg_cur)
# set postgres search path back to the default
pg_cur.execute("SET search_path = public, pg_catalog")
logger.info(f"Part 2 of 6 : Raw GNAF loaded! : {datetime.now() - start_time}")
# PART 3 - load raw admin boundaries from Shapefiles
logger.info("")
start_time = datetime.now()
logger.info(f"Part 3 of 6 : Start raw admin boundary load : {start_time}")
load_raw_admin_boundaries(pg_cur)
clean_authority_files(pg_cur, settings.raw_admin_bdys_schema, True)
prep_admin_bdys(pg_cur)
create_admin_bdys_for_analysis()
logger.info(f"Part 3 of 6 : Raw admin boundaries loaded! : {datetime.now() - start_time}")
# PART 4 - create flattened and standardised GNAF and Administrative Boundary reference tables
logger.info("")
start_time = datetime.now()
logger.info(f"Part 4 of 6 : Start create reference tables : {start_time}")
create_reference_tables(pg_cur)
logger.info(f"Part 4 of 6 : Reference tables created! : {datetime.now() - start_time}")
# PART 5 - boundary tag GNAF addresses
logger.info("")
if settings.no_boundary_tag:
logger.warning("Part 5 of 6 : Addresses NOT boundary tagged")
else:
start_time = datetime.now()
logger.info(f"Part 5 of 6 : Start boundary tagging addresses : {start_time}")
boundary_tag_gnaf(pg_cur)
logger.info(f"Part 5 of 6 : Addresses boundary tagged : {datetime.now() - start_time}")
# PART 6 - get record counts for QA
logger.info("")
start_time = datetime.now()
logger.info(f"Part 6 of 6 : Start row counts : {start_time}")
create_qa_tables(pg_cur)
logger.info(f"Part 6 of 6 : Got row counts : {datetime.now() - start_time}")
# close Postgres connection
pg_cur.close()
pg_conn.close()
logger.info("")
logger.info(f"Total time : {datetime.now() - full_start_time}")
return True
def drop_tables_and_vacuum_db(pg_cur):
# Step 1 of 7 : drop tables
start_time = datetime.now()
pg_cur.execute(geoscape.open_sql_file("01-01-drop-tables.sql"))
logger.info(f"\t- Step 1 of 7 : tables dropped : {datetime.now() - start_time}")
# Step 2 of 7 : vacuum database (if requested)
start_time = datetime.now()
if settings.vacuum_db:
pg_cur.execute("VACUUM")
logger.info(f"\t- Step 2 of 7 : database vacuumed : {datetime.now() - start_time}")
else:
logger.info("\t- Step 2 of 7 : database NOT vacuumed")
def create_raw_gnaf_tables(pg_cur):
# Step 3 of 7 : create tables
start_time = datetime.now()
# prep create table sql scripts (note: file doesn't contain any schema prefixes on table names)
sql = geoscape.open_sql_file("01-03-raw-gnaf-create-tables.sql")
# set search path
if settings.raw_gnaf_schema != "public":
pg_cur.execute(f"SET search_path = {settings.raw_gnaf_schema}")
# alter create table script to run on chosen schema
sql = sql.replace("SET search_path = public", f"SET search_path = {settings.raw_gnaf_schema}")
# set tables to unlogged to speed up the load? (if requested)
# -- they'll have to be rebuilt using this script again after a system crash --
if settings.unlogged_tables:
sql = sql.replace("CREATE TABLE ", "CREATE UNLOGGED TABLE ")
unlogged_string = "UNLOGGED "
else:
unlogged_string = ""
# create raw gnaf tables
pg_cur.execute(sql)
logger.info(f"\t- Step 3 of 7 : {unlogged_string}tables created : {datetime.now() - start_time}")
# load raw gnaf authority & state tables using multiprocessing
def populate_raw_gnaf(pg_cur):
# Step 4 of 7 : load raw gnaf authority & state tables
start_time = datetime.now()
# authority code file list
sql_list = get_raw_gnaf_files("authority_code")
# add state file lists
for state in settings.states_to_load:
logger.info(f"\t\t- Loading state {state}")
sql_list.extend(get_raw_gnaf_files(state))
# are there any files to load?
if len(sql_list) == 0:
logger.fatal("No raw GNAF PSV files found\nACTION: Check your 'gnaf_network_directory' path")
logger.fatal("\t- Step 4 of 7 : table populate FAILED!")
else:
# load all PSV files using multiprocessing
geoscape.multiprocess_list("sql", sql_list, logger)
# fix missing geocodes (added due to missing data in 202408 release)
sql = geoscape.open_sql_file("01-04-raw-gnaf-fix-missing-geocodes.sql")
pg_cur.execute(sql)
logger.info(f"\t- Step 4 of 7 : tables populated : {datetime.now() - start_time}")
logger.info("\t\t- fixed missing geocodes")
def get_raw_gnaf_files(prefix):
sql_list = []
prefix = prefix.lower()
# get a dictionary of all files matching the filename prefix
for root, dirs, files in os.walk(settings.gnaf_network_directory):
for file_name in files:
if file_name.lower().startswith(prefix + "_"):
if file_name.lower().endswith(".psv"):
file_path = os.path.join(root, file_name)\
.replace(settings.gnaf_network_directory, settings.gnaf_pg_server_local_directory)
table = file_name.lower().replace(prefix + "_", "", 1).replace("_psv", "").replace(".psv", "")
# if a non-Windows Postgres server OS - fix file path
if settings.gnaf_pg_server_local_directory[0:1] == "/":
file_path = file_path.replace("\\", "/")
# logger.info(file_path
sql = f"COPY {settings.raw_gnaf_schema}.{table} FROM '{file_path}' DELIMITER '|' CSV HEADER;"
sql_list.append(sql)
return sql_list
# index raw gnaf using multiprocessing
def index_raw_gnaf():
# Step 5 of 7 : create indexes
start_time = datetime.now()
raw_sql_list = geoscape.open_sql_file("01-05-raw-gnaf-create-indexes.sql").split("\n")
sql_list = []
for sql in raw_sql_list:
if sql[0:2] != "--" and sql[0:2] != "":
sql_list.append(sql)
geoscape.multiprocess_list("sql", sql_list, logger)
# # create distinct new & old locality pid lookup table
# pg_cur.execute(geoscape.open_sql_file("01-05b-create-distinct-locality-pid-linkage-table.sql"))
logger.info(f"\t- Step 5 of 7 : indexes created : {datetime.now() - start_time}")
# create raw gnaf primary & foreign keys (for data integrity) using multiprocessing
def create_primary_foreign_keys():
start_time = datetime.now()
key_sql = geoscape.open_sql_file("01-06-raw-gnaf-create-primary-foreign-keys.sql")
key_sql_list = key_sql.split("--")
sql_list = []
for sql in key_sql_list:
sql = sql.strip()
if sql[0:6] == "ALTER ":
# add schema to tables names, in case raw gnaf schema not the default
sql = sql.replace(f"ALTER TABLE ONLY ", f"ALTER TABLE ONLY {settings.raw_gnaf_schema}.")
sql_list.append(sql)
# run queries in separate processes
geoscape.multiprocess_list("sql", sql_list, logger)
logger.info(f"\t- Step 6 of 7 : primary & foreign keys created : {datetime.now() - start_time}")
# analyse raw GNAF tables that have not stats - need actual row counts for QA at the end
def analyse_raw_gnaf_tables(pg_cur):
start_time = datetime.now()
# get list of tables that haven't been analysed (i.e. that have no real row count)
sql = f"""SELECT nspname|| '.' || relname AS table_name
FROM pg_class C LEFT JOIN pg_namespace N ON (N.oid = C.relnamespace)
WHERE nspname = '{settings.raw_gnaf_schema}' AND relkind='r' AND reltuples = 0"""
pg_cur.execute(sql)
sql_list = []
for pg_row in pg_cur:
sql_list.append(f"ANALYZE {pg_row[0]}")
# run queries in separate processes
geoscape.multiprocess_list("sql", sql_list, logger)
logger.info(f"\t- Step 7 of 7 : tables analysed : {datetime.now() - start_time}")
# loads the admin bdy shapefiles using the shp2pgsql command line tool (part of PostGIS), using multiprocessing
def load_raw_admin_boundaries(pg_cur):
start_time = datetime.now()
# drop existing views
pg_cur.execute(geoscape.open_sql_file("02-01-drop-admin-bdy-views.sql"))
# add authority code tables
settings.states_to_load.extend(["authority_code"])
# get file list
table_list = list()
create_list = list()
append_list = list()
for state in settings.states_to_load:
state = state.lower()
# get a dictionary of Shapefiles and DBFs matching the state
for root, dirs, files in os.walk(settings.admin_bdys_local_directory):
for file_name in files:
if file_name.lower().startswith(state + "_"):
if file_name.lower().endswith(".shp") or file_name.lower().endswith("_shp.dbf"):
file_dict = dict()
# list .shp files and standalone .dbf files - ignore the rest
if file_name.lower().endswith(".shp"):
file_dict["spatial"] = True
file_dict["file_path"] = os.path.join(root, file_name)
elif file_name.lower().endswith(".dbf") and not file_name.lower().endswith("_polygon_shp.dbf") \
and not file_name.lower().endswith("_point_shp.dbf"):
file_dict["spatial"] = False
file_dict["file_path"] = os.path.join(root, file_name)
if file_dict.get("file_path") is not None:
file_dict["pg_table"] = file_name.lower().replace(state + "_", "aus_", 1)\
.replace(".dbf", "").replace(".shp", "").replace("_shp", "")
file_dict["pg_schema"] = settings.raw_admin_bdys_schema
# set command line parameters depending on whether this is the 1st state
table_list_add = False
if file_dict["pg_table"] not in table_list:
table_list_add = True
file_dict["delete_table"] = True
else:
file_dict["delete_table"] = False
# if locality file from Towns folder: don't add - it's a duplicate
if "town points" not in file_dict["file_path"].lower():
if table_list_add:
table_list.append(file_dict["pg_table"])
create_list.append(file_dict)
else:
# # don't add duplicates if more than one Authority Code file per boundary type
# if "_aut_" not in file_name.lower():
append_list.append(file_dict)
else:
if not file_dict["file_path"].lower().endswith("_locality_shp.dbf"):
if table_list_add:
table_list.append(file_dict["pg_table"])
create_list.append(file_dict)
else:
# # don't add duplicates if more than one Authority Code file per boundary type
# if "_aut_" not in file_name.lower():
append_list.append(file_dict)
# [print(table) for table in create_list]
# print("---------------------------------------------------------------------------------------")
# [print(table) for table in append_list]
# are there any files to load?
if len(create_list) == 0:
logger.fatal("No admin boundary files found\nACTION: Check your 'admin-bdys-path' argument")
pg_cur.close()
quit()
else:
# load files in separate processes
geoscape.multiprocess_shapefile_load(create_list, logger)
# Run the appends one at a time (Can't multiprocess as sets of parallel INSERTs can cause database deadlocks)
for shp in append_list:
result = geoscape.import_shapefile_to_postgres(shp["file_path"], shp["pg_table"], shp["pg_schema"],
shp["delete_table"], shp["spatial"])
if result != "SUCCESS":
logger.warning(result)
logger.info(f"\t- Step 1 of 3 : raw admin boundaries loaded : {datetime.now() - start_time}")
def clean_authority_files(pg_cur, schema_name, create_indexes=False):
# ensure authority tables have unique values - admin bdys now have duplicates
error_count = 0
# get table list for schema
sql = f"""SELECT table_name
FROM information_schema.tables
WHERE table_schema='{schema_name}'
AND table_type='BASE TABLE'
AND table_name LIKE '%_aut'"""
pg_cur.execute(sql)
tables = pg_cur.fetchall()
for table in tables:
table_name = table[0]
# print(table_name)
# fix inconsistent field names with brute force method (issue loading Shapefile/DBF data)
try:
pg_cur.execute(f"ALTER TABLE {schema_name}.{table_name} RENAME COLUMN code_aut TO code")
except psycopg.Error:
pass
try:
pg_cur.execute(f"ALTER TABLE {schema_name}.{table_name} RENAME COLUMN name_aut TO name")
except psycopg.Error:
pass
try:
pg_cur.execute(f"ALTER TABLE {schema_name}.{table_name} RENAME COLUMN dscpn_aut TO description")
except psycopg.Error:
pass
try:
pg_cur.execute(f"ALTER TABLE {schema_name}.{table_name} RENAME COLUMN desc_aut TO description")
except psycopg.Error:
pass
try:
pg_cur.execute(f"ALTER TABLE {schema_name}.{table_name} RENAME COLUMN descriptio TO description")
except psycopg.Error:
pass
# fix inconsistent descriptions in meshblock authority table by setting them to null
if table_name == "aus_mb_category_class_aut":
pg_cur.execute(f"UPDATE {schema_name}.{table_name} SET description = NULL")
# get original row count
pg_cur.execute(f"SELECT count(*) FROM {schema_name}.{table_name}")
old_row_count = int(pg_cur.fetchone()[0])
# get distinct records
sql = f"""DROP TABLE IF EXISTS temp_aut;
CREATE TABLE temp_aut AS
SELECT DISTINCT code, name, description FROM {schema_name}.{table_name};"""
pg_cur.execute(sql)
# get new row count
pg_cur.execute("SELECT count(*) FROM temp_aut")
new_row_count = int(pg_cur.fetchone()[0])
# only delete and replace if duplicates found
duplicate_row_count = old_row_count - new_row_count
if duplicate_row_count > 0:
# delete all rows
pg_cur.execute(f"TRUNCATE TABLE {schema_name}.{table_name}")
# insert distinct rows
pg_cur.execute(f"INSERT INTO {schema_name}.{table_name} (code, name, description) SELECT * FROM temp_aut")
logger.info(f"\t\t- {duplicate_row_count} duplicates removed from {schema_name}.{table_name}")
# This is required due to complexities introduced by mix of authority
# and non-authority table admin bdy layers in the Aug 2021 release
if create_indexes:
# drop primary key on gid field
try:
pg_cur.execute(f"ALTER TABLE ONLY {schema_name}.{table_name} DROP CONSTRAINT {table_name}_pkey")
except psycopg.Error:
pass
# attempt to create a primary key on the authority code - failure will imply a raw data error from Geoscape
try:
pg_cur.execute(f"ALTER TABLE ONLY {schema_name}.{table_name} "
f"ADD CONSTRAINT {table_name}_pkey PRIMARY KEY (code)")
except psycopg.Error as ex:
error_count += 1
logger.warning(f"CAN'T CREATE PRIMARY KEY ON {schema_name}.{table_name} "
f"DUE TO DUPLICATE AUTHORITY CODE(S) : {ex}")
# clean up
pg_cur.execute("DROP TABLE IF EXISTS temp_aut")
pg_cur.execute(f"VACUUM ANALYZE {schema_name}.{table_name}")
# kill gnaf-loader if duplicates couldn't be fixed - significant data integrity issue
if error_count > 0:
exit()
logger.info("\t\t- authority tables deduplicated")
def prep_admin_bdys(pg_cur):
# Step 3 of 4 : create admin bdy tables read to be used
start_time = datetime.now()
# create tables using multiprocessing - using flag in file to split file up into sets of statements
sql_list = geoscape.open_sql_file("02-02a-prep-admin-bdys-tables.sql").format(settings.srid).split("-- # --")
sql_list = sql_list + geoscape.open_sql_file("02-02c-prep-census-2016-bdys-tables.sql").split("-- # --")
sql_list = sql_list + geoscape.open_sql_file("02-02d-prep-census-2021-bdys-tables.sql").split("-- # --")
# # Account for bdys that are not in states to load - not yet working
# for sql in sql_list:
# if settings.states_to_load == ["OT"] and ".commonwealth_electorates " in sql:
# sql_list.remove(sql)
#
# if settings.states_to_load == ["ACT"] and ".local_government_areas " in sql:
# sql_list.remove(sql)
#
# logger.info(settings.states_to_load
#
# if not ("NT" in settings.states_to_load or "SA" in settings.states_to_load
# or "VIC" in settings.states_to_load or "WA" in settings.states_to_load) \
# and ".local_government_wards " in sql:
# sql_list.remove(sql)
#
# if settings.states_to_load == ["OT"] and ".state_lower_house_electorates " in sql:
# sql_list.remove(sql)
#
# if not ("TAS" in settings.states_to_load or "VIC" in settings.states_to_load
# or "WA" in settings.states_to_load) and ".state_upper_house_electorates " in sql:
# sql_list.remove(sql)
geoscape.multiprocess_list("sql", sql_list, logger)
# Special case - remove custom outback bdy if South Australia not requested
if "SA" not in settings.states_to_load:
pg_cur.execute(geoscape.prep_sql("DELETE FROM admin_bdys.locality_bdys WHERE locality_pid = 'SA999999'"))
pg_cur.execute(geoscape.prep_sql("VACUUM ANALYZE admin_bdys.locality_bdys"))
logger.info(f"\t- Step 2 of 3 : admin boundaries prepped : {datetime.now() - start_time}")
def create_admin_bdys_for_analysis():
# Step 3 of 3 : create admin bdy tables optimised for spatial analysis
start_time = datetime.now()
if settings.st_subdivide_supported:
template_sql = geoscape.open_sql_file("02-03-create-admin-bdy-analysis-tables_template.sql")
sql_list = list()
for table in settings.admin_bdy_list:
sql = template_sql.format(table[0], table[1], settings.srid)
if table[0] == "locality_bdys": # special case, need to change schema name
# sql = sql.replace(settings.raw_admin_bdys_schema, settings.admin_bdys_schema)
sql = sql.replace("name", "locality_name")
# add postcodes
sql = sql.replace("locality_name text NOT NULL,",
"locality_name text NOT NULL, postcode text NULL,")
sql = sql.replace("locality_name,", "locality_name, postcode,")
sql_list.append(sql)
geoscape.multiprocess_list("sql", sql_list, logger)
logger.info(f"\t- Step 3 of 3 : admin boundaries for analysis created : {datetime.now() - start_time}")
else:
logger.warning("\t- Step 3 of 3 : admin boundaries for analysis NOT created - "
"requires PostGIS 2.2+ with GEOS 3.5.0+")
# create gnaf reference tables by flattening raw gnaf address, streets & localities into a usable form
# also creates all supporting lookup tables and usable admin bdy tables
def create_reference_tables(pg_cur):
# set postgres search path back to the default
pg_cur.execute("SET search_path = public, pg_catalog")
# Step 1 of 14 : create reference tables
start_time = datetime.now()
pg_cur.execute(geoscape.open_sql_file("03-01-reference-create-tables.sql").format(settings.srid))
logger.info(f"\t- Step 1 of 14 : create reference tables : {datetime.now() - start_time}")
# Step 2 of 14 : populate localities
start_time = datetime.now()
pg_cur.execute(geoscape.open_sql_file("03-02-reference-populate-localities.sql").format(settings.srid))
logger.info(f"\t- Step 2 of 14 : localities populated : {datetime.now() - start_time}")
# Step 3 of 14 : populate locality aliases
start_time = datetime.now()
pg_cur.execute(geoscape.open_sql_file("03-03-reference-populate-locality-aliases.sql"))
logger.info(f"\t- Step 3 of 14 : locality aliases populated : {datetime.now() - start_time}")
# Step 4 of 14 : populate locality neighbours
start_time = datetime.now()
pg_cur.execute(geoscape.open_sql_file("03-04-reference-populate-locality-neighbours.sql"))
logger.info(f"\t- Step 4 of 14 : locality neighbours populated : {datetime.now() - start_time}")
# Step 5 of 14 : populate streets
start_time = datetime.now()
pg_cur.execute(geoscape.open_sql_file("03-05-reference-populate-streets.sql").format(settings.srid))
logger.info(f"\t- Step 5 of 14 : streets populated : {datetime.now() - start_time}")
# Step 6 of 14 : populate street aliases
start_time = datetime.now()
pg_cur.execute(geoscape.open_sql_file("03-06-reference-populate-street-aliases.sql"))
logger.info(f"\t- Step 6 of 14 : street aliases populated : {datetime.now() - start_time}")
# Step 7 of 14 : populate addresses, using multiprocessing
start_time = datetime.now()
sql = geoscape.open_sql_file("03-07-reference-populate-addresses-1.sql").format(settings.srid)
sql_list = geoscape.split_sql_into_list(pg_cur, sql, settings.gnaf_schema, "streets", "str", "gid", logger)
if sql_list is not None:
geoscape.multiprocess_list("sql", sql_list, logger)
pg_cur.execute(geoscape.prep_sql("ANALYZE gnaf.temp_addresses;"))
logger.info(f"\t- Step 7 of 14 : addresses populated : {datetime.now() - start_time}")
# Step 8 of 14 : populate principal alias lookup
start_time = datetime.now()
pg_cur.execute(geoscape.open_sql_file("03-08-reference-populate-address-alias-lookup.sql"))
logger.info(f"\t- Step 8 of 14 : principal alias lookup populated : {datetime.now() - start_time}")
# Step 9 of 14 : populate primary secondary lookup
start_time = datetime.now()
pg_cur.execute(geoscape.open_sql_file("03-09-reference-populate-address-secondary-lookup.sql"))
pg_cur.execute(geoscape.prep_sql("VACUUM ANALYSE gnaf.address_secondary_lookup"))
logger.info(f"\t- Step 9 of 14 : primary secondary lookup populated : {datetime.now() - start_time}")
# Step 10 of 14 : split the Melbourne locality into its 2 postcodes (3000, 3004)
start_time = datetime.now()
pg_cur.execute(geoscape.open_sql_file("03-10-reference-split-melbourne.sql").format(settings.srid))
logger.info(f"\t- Step 10 of 14 : Melbourne split : {datetime.now() - start_time}")
# Step 11 of 14 : finalise localities assigned to streets and addresses
start_time = datetime.now()
pg_cur.execute(geoscape.open_sql_file("03-11-reference-finalise-localities.sql"))
logger.info(f"\t- Step 11 of 14 : localities finalised : {datetime.now() - start_time}")
# Step 12 of 14 : finalise addresses, using multiprocessing
start_time = datetime.now()
sql = geoscape.open_sql_file("03-12-reference-populate-addresses-2.sql")
sql_list = geoscape.split_sql_into_list(pg_cur, sql, settings.gnaf_schema, "localities", "loc", "gid", logger)
if sql_list is not None:
geoscape.multiprocess_list("sql", sql_list, logger)
# turf the temp address table
pg_cur.execute(geoscape.prep_sql("DROP TABLE IF EXISTS gnaf.temp_addresses"))
logger.info(f"\t- Step 12 of 14 : addresses finalised : {datetime.now() - start_time}")
# Step 13 of 14 : create almost correct postcode boundaries by aggregating localities, using multiprocessing
start_time = datetime.now()
sql = geoscape.open_sql_file("03-13-reference-derived-postcode-bdys.sql")
sql_list = []
for state in settings.states_to_load:
state_sql = sql.replace("GROUP BY ", f"WHERE state = '{state}' GROUP BY ")
sql_list.append(state_sql)
geoscape.multiprocess_list("sql", sql_list, logger)
# create analysis table?
if settings.st_subdivide_supported:
pg_cur.execute(geoscape.open_sql_file("03-13a-create-postcode-analysis-table.sql").format(settings.srid))
logger.info(f"\t- Step 13 of 14 : postcode boundaries created : {datetime.now() - start_time}")
# Step 14 of 14 : create indexes, primary and foreign keys, using multiprocessing
start_time = datetime.now()
raw_sql_list = geoscape.open_sql_file("03-14-reference-create-indexes.sql").split("\n")
sql_list = []
for sql in raw_sql_list:
if sql[0:2] != "--" and sql[0:2] != "":
sql_list.append(sql)
geoscape.multiprocess_list("sql", sql_list, logger)
logger.info(f"\t- Step 14 of 14 : create primary & foreign keys and indexes : {datetime.now() - start_time}")
def boundary_tag_gnaf(pg_cur):
# create bdy table list
# remove localities, postcodes and states as these IDs are already assigned to GNAF addresses
table_list = list()
for table in settings.admin_bdy_list:
if table[0] not in ["locality_bdys", "postcode_bdys", "state_bdys"]:
# if no analysis tables created - use the full tables instead of the subdivided ones
# WARNING: this can add hours to the processing
if settings.st_subdivide_supported:
table_name = f"{table[0]}_analysis"
else:
table_name = table[0]
table_list.append([table_name, table[1]])
# create bdy tagged address tables
for address_table in ["address_principal", "address_alias"]:
pg_cur.execute(f"DROP TABLE IF EXISTS {settings.gnaf_schema}.{address_table}_admin_boundaries CASCADE")
create_table_list = list()
create_table_list.append(f"""CREATE TABLE {settings.gnaf_schema}.{address_table}_admin_boundaries (
gid serial NOT NULL,
gnaf_pid text NOT NULL,
-- alias_principal character(1) NOT NULL,
locality_pid text NOT NULL,
-- old_locality_pid text NULL,
locality_name text NOT NULL,
postcode text,
state text NOT NULL""")
for table in table_list:
pid_field = table[1]
name_field = pid_field.replace("_pid", "_name")
create_table_list.append(f", {pid_field} text, {name_field} text")
create_table_list.append(f""") WITH (OIDS=FALSE);
ALTER TABLE {settings.gnaf_schema}.{address_table}_admin_boundaries
OWNER TO {settings.pg_user}""")
pg_cur.execute("".join(create_table_list))
# Step 1 of 7 : tag gnaf addresses with admin boundary IDs, using multiprocessing
start_time = datetime.now()
# create temp tables
template_sql = geoscape.open_sql_file("04-01a-bdy-tag-create-table-template.sql")
for table in table_list:
pg_cur.execute(template_sql.format(table[0],))
# create temp tables of bdy tagged gnaf_pids
template_sql = geoscape.open_sql_file("04-01b-bdy-tag-template.sql")
sql_list = list()
for table in table_list:
sql = template_sql.format(table[0], table[1])
short_sql_list = geoscape.split_sql_into_list(pg_cur, sql, settings.admin_bdys_schema, table[0],
"bdys", "gid", logger)
if short_sql_list is not None:
sql_list.extend(short_sql_list)
# logger.info("\n".join(sql_list))
if sql_list is not None:
geoscape.multiprocess_list("sql", sql_list, logger)
logger.info(f"\t- Step 1 of 7 : principal addresses tagged with admin boundary IDs: {datetime.now() - start_time}")
start_time = datetime.now()
# Step 2 of 7 : delete invalid matches, create indexes and analyse tables
sql_list = list()
for table in table_list:
sql = f"""DELETE FROM {settings.gnaf_schema}.temp_{table[0]}_tags
WHERE gnaf_state <> bdy_state AND gnaf_state <> 'OT';
CREATE INDEX temp_{table[0]}_tags_gnaf_pid_idx
ON {settings.gnaf_schema}.temp_{table[0]}_tags USING btree(gnaf_pid);
ANALYZE {settings.gnaf_schema}.temp_{table[0]}_tags"""""
sql_list.append(sql)
geoscape.multiprocess_list("sql", sql_list, logger)
logger.info(f"\t- Step 2 of 7 : principal addresses - invalid matches deleted & bdy tag indexes created : "
f"{datetime.now() - start_time}")
start_time = datetime.now()
# Step 3 of 7 : insert boundary tagged addresses
# create insert statement for multiprocessing
insert_field_list = list()
insert_field_list.append("(gnaf_pid, locality_pid, locality_name, postcode, state")
insert_join_list = list()
insert_join_list.append(f"FROM {settings.gnaf_schema}.address_principals AS pnts ")
select_field_list = list()
select_field_list.append("SELECT pnts.gnaf_pid, pnts.locality_pid, pnts.locality_name, pnts.postcode, pnts.state")
drop_table_list = list()
for table in table_list:
pid_field = table[1]
name_field = pid_field. replace("_pid", "_name")
insert_field_list.append(f", {pid_field}, {name_field}")
select_field_list.append(f", temp_{table[0]}_tags.bdy_pid, temp_{table[0]}_tags.bdy_name ")
insert_join_list.append(f"LEFT OUTER JOIN {settings.gnaf_schema}.temp_{table[0]}_tags "
f"ON pnts.gnaf_pid = temp_{table[0]}_tags.gnaf_pid ")
drop_table_list.append(f"DROP TABLE IF EXISTS {settings.gnaf_schema}.temp_{table[0]}_tags;")
insert_field_list.append(") ")
insert_statement_list = list()
insert_statement_list.append(f"INSERT INTO {settings.gnaf_schema}.address_principal_admin_boundaries ")
insert_statement_list.append("".join(insert_field_list))
insert_statement_list.append("".join(select_field_list))
insert_statement_list.append("".join(insert_join_list))
sql = "".join(insert_statement_list) + ";"
sql_list = geoscape.split_sql_into_list(pg_cur, sql, settings.gnaf_schema, "address_principals",
"pnts", "gid", logger)
# logger.info("\n".join(sql_list)
if sql_list is not None:
geoscape.multiprocess_list("sql", sql_list, logger)
# drop temp tables
pg_cur.execute("".join(drop_table_list))
# get stats
pg_cur.execute(f"ANALYZE {settings.gnaf_schema}.address_principal_admin_boundaries ")
logger.info(f"\t- Step 3 of 7 : principal addresses - bdy tags added to output table : "
f"{datetime.now() - start_time}")
start_time = datetime.now()
# Step 4 of 7 : add index to output table
sql = f"CREATE INDEX address_principal_admin_boundaries_gnaf_pid_idx " \
f"ON {settings.gnaf_schema}.address_principal_admin_boundaries USING btree (gnaf_pid)"
pg_cur.execute(sql)
logger.info(f"\t- Step 4 of 7 : created index on bdy tagged address table : {datetime.now() - start_time}")
start_time = datetime.now()
# Step 5 of 7 : log duplicates - happens when 2 boundaries overlap by a very small amount
# (can be ignored if there's a small number of records affected)
sql = f"""SELECT gnaf_pid FROM (
SELECT Count(*) AS cnt, gnaf_pid
FROM {settings.gnaf_schema}.address_principal_admin_boundaries
GROUP BY gnaf_pid
) AS sqt WHERE cnt > 1"""
pg_cur.execute(sql)
# get cursor description to test if any rows returned safely
columns = pg_cur.description
# log gnaf_pids that got duplicate results
if columns is not None:
duplicates = pg_cur.fetchall()
gnaf_pids = list()
for duplicate in duplicates:
gnaf_pids.append("\t\t" + duplicate[0])
if len(gnaf_pids) > 0:
logger.warning(f"\t- Step 5 of 7 : found boundary tag duplicates : {datetime.now() - start_time}")
logger.warning("\n".join(gnaf_pids))
else:
logger.info(f"\t- Step 5 of 7 : no boundary tag duplicates : {datetime.now() - start_time}")
else:
logger.info(f"\t- Step 5 of 7 : no boundary tag duplicates : {datetime.now() - start_time}")
# Step 6 of 7 : Copy principal boundary tags to alias addresses
pg_cur.execute(geoscape.open_sql_file("04-06-bdy-tags-for-alias-addresses.sql"))
logger.info(f"\t- Step 6 of 7 : alias addresses boundary tagged : {datetime.now() - start_time}")
start_time = datetime.now()
# Step 7 of 7 : Create view of all bdy tags
pg_cur.execute(geoscape.open_sql_file("04-07-create-bdy-tag-view.sql"))
logger.info(f"\t- Step 7 of 7 : boundary tagged address view created : {datetime.now() - start_time}")
def create_qa_tables(pg_cur):
start_time = datetime.now()
i = 0
for schema in [settings.gnaf_schema, settings.admin_bdys_schema]:
i += 1
# STEP 1 - get row counts of tables in each schema, by state, for visual QA
# create qa table of rows counts
sql = f"""DROP TABLE IF EXISTS {schema}.qa;
CREATE TABLE {schema}.qa (table_name text, aus integer, act integer, nsw integer, nt integer,
ot integer, qld integer, sa integer, tas integer, vic integer, wa integer)
WITH (OIDS=FALSE);
ALTER TABLE {schema}.qa OWNER TO {settings.pg_user}"""
pg_cur.execute(sql)
# get table names in schema
sql = f"""SELECT table_name
FROM information_schema.tables
WHERE table_schema = '{schema}'
AND table_name <> 'qa'
ORDER BY table_name"""
pg_cur.execute(sql)
table_names = []
for pg_row in pg_cur:
table_names.append(pg_row[0])
# get row counts by state
for table_name in table_names:
sql = f"""INSERT INTO {schema}.qa
SELECT '{table_name}', SUM(AUS), SUM(ACT), SUM(NSW), SUM(NT), SUM(OT), SUM(QLD), SUM(SA),
SUM(TAS), SUM(VIC), SUM(WA)
FROM (
SELECT 1 AS AUS,
CASE WHEN state = 'ACT' THEN 1 ELSE 0 END AS ACT,
CASE WHEN state = 'NSW' THEN 1 ELSE 0 END AS NSW,
CASE WHEN state = 'NT' THEN 1 ELSE 0 END AS NT,
CASE WHEN state = 'OT' THEN 1 ELSE 0 END AS OT,
CASE WHEN state = 'QLD' THEN 1 ELSE 0 END AS QLD,
CASE WHEN state = 'SA' THEN 1 ELSE 0 END AS SA,
CASE WHEN state = 'TAS' THEN 1 ELSE 0 END AS TAS,
CASE WHEN state = 'VIC' THEN 1 ELSE 0 END AS VIC,
CASE WHEN state = 'WA' THEN 1 ELSE 0 END AS WA
FROM {schema}.{table_name}
) AS sqt"""
try:
pg_cur.execute(sql)
except psycopg.Error: # triggers when there is no state field in the table
# change the query for an Australia count only
sql = f"INSERT INTO {schema}.qa (table_name, aus) " \
f"SELECT '{table_name}', Count(*) FROM {schema}.{table_name}"
try:
pg_cur.execute(sql)
except Exception as ex:
# if no state field - change the query for an Australia count only
logger.warning(f"Couldn't get row count for {schema}.{table_name} : {ex}")
pg_cur.execute(f"ANALYZE {schema}.qa")
# STEP 2 - compare row counts with previous Geoscape release
# get previous schema name
if "gnaf_" in schema:
previous_schema = settings.previous_gnaf_schema
else:
previous_schema = settings.previous_admin_bdys_schema
# check if previous schema exists in database
pg_cur.execute(f"SELECT schema_name FROM information_schema.schemata WHERE schema_name = '{previous_schema}'")
test_schema_row = pg_cur.fetchone()
if test_schema_row is not None:
# create qa table of rows counts
sql = f"""DROP TABLE IF EXISTS {schema}.qa_comparison;
CREATE TABLE {schema}.qa_comparison (
table_name text,
difference integer,
new_count integer,
old_count integer
) WITH (OIDS=FALSE);
ALTER TABLE {schema}.qa_comparison OWNER TO {settings.pg_user}"""
pg_cur.execute(sql)
# into get counts into qa_comparison table
sql = f"""INSERT INTO {schema}.qa_comparison
SELECT new.table_name,
new.aus - old.aus as difference,
new.aus as new_count,
old.aus as old_count
FROM {schema}.qa as new
INNER JOIN {previous_schema}.qa as old ON new.table_name = old.table_name"""
pg_cur.execute(sql)
pg_cur.execute(f"ANALYZE {schema}.qa_comparison")
# pretty print row counts to screen
pg_cur.execute(f"SELECT * FROM {schema}.qa_comparison ORDER BY table_name")
rows = pg_cur.fetchall()
logger.info("\t\t------------------------------------------------------------------------")
logger.info("\t\t|{:39}|{:10}|{:9}|{:9}|".format("table_name", "difference", "new_count", "old_count"))
logger.info("\t\t------------------------------------------------------------------------")
for row in rows:
logger.info("\t\t|{:39}|{:10}|{:9}|{:9}|".format(row[0], row[1], row[2], row[3]))
logger.info("\t\t------------------------------------------------------------------------")
else:
logger.warning(f"\t\t- Previous schema ({previous_schema}) doesn't exist - row count comparison not done")
logger.info(f"\t- Step {i} of 2 : got row counts for {schema} schema : {datetime.now() - start_time}")
if __name__ == "__main__":
logger = logging.getLogger()
# set logger
log_file = os.path.abspath(__file__).replace(".py", ".log")
logging.basicConfig(filename=log_file, level=logging.DEBUG, format="%(asctime)s %(message)s",
datefmt="%m/%d/%Y %I:%M:%S %p")
# setup logger to write to screen as well as writing to log file
# define a Handler which writes INFO messages or higher to the sys.stderr
console = logging.StreamHandler()
console.setLevel(logging.INFO)
# set a format which is simpler for console use
formatter = logging.Formatter("%(name)-12s: %(levelname)-8s %(message)s")
# tell the handler to use this format
console.setFormatter(formatter)
# add the handler to the root logger
logging.getLogger("").addHandler(console)
logger.info("")
logger.info("Start gnaf-loader")
if main():
logger.info("Finished successfully!")
else:
logger.fatal("Something bad happened!")