Project

General

Profile

1
selfDir_uZPPqC := $(dir $(lastword $(MAKEFILE_LIST)))
2
root := $(selfDir_uZPPqC)..
3
include $(root)/lib/common.Makefile
4

    
5

    
6
##### Configuration
7

    
8
# Command line
9
log ?= $(if $(test),,1)
10
profile ?=
11
quiet ?=
12
reverify ?= 1
13
schema_only ?=
14
use_staged ?= $(by_col)
15

    
16
# Makefile
17
exts ?= csv tsv txt dmp xml
18
test_n ?= 2
19

    
20
##### Vars/functions
21

    
22
# Paths
23
datasrc := $(patsubst .%,%,$(notdir $(realpath .)))
24
bin := $(root)/bin
25
mappings := $(root)/mappings
26

    
27
# Make
28
SHELL := /bin/bash
29
selfMake = $(MAKE) --makefile=../input.Makefile
30
subMake = $(MAKE) $(@:$(root)/%=%) --directory=$(root)
31
+_ = $(+:_%=)
32
addBeforeExt = $(basename $(2))$(1)$(suffix $(2))
33

    
34
# Terminal
35
termCols := $(shell tput cols)
36
esc := '['
37
reset := $(esc)'0m'
38
emph := $(esc)'7m '
39
endEmph := ' '$(reset)
40

    
41
# User interaction
42

    
43
confirm = $(if $(shell read -p $(emph)"$(1)"$(endEmph)$$'$(if\
44
$(2),\n$(2))\nContinue? (y/n) ' REPLY; test "$$REPLY" = y && echo t),,\
45
$(error Aborting))
46

    
47
# Commands
48
MKDIR = mkdir -p
49
mkdir = $(MKDIR) $(@D)
50
CP = cp -p
51
diff = diff --unified=2
52
diffIgnoreSpace = $(diff) --ignore-space-change
53
diffVerbose = $(if $(verbose),diff --side-by-side --left-column\
54
--width=$(termCols),$(diff))
55

    
56
# BIEN commands
57
sortFilenames = $(shell $(bin)/sort_filenames $(1))
58
selfMap = $(bin)/cols 0 0
59
psqlOpts := --set ON_ERROR_STOP=1 --quiet
60
psqlAsBien := $(bin)/psql_vegbien $(psqlOpts)
61
searchPath := $(datasrc),$(shell prefix=; . $(bin)/vegbien_dest;\
62
echo "$$schemas")
63
searchPath := "$(subst $(comma),"$(comma)",$(searchPath))"
64
# Usage: ($(inDatasrc); cat $(file))|$(psqlCmd)
65
inDatasrc := echo 'SET search_path TO $(searchPath);'
66

    
67
# SVN
68
addDir = $(if $(wildcard $(1)/),svn add --depth=empty $(1),svn mkdir $(1))
69
setSvnIgnore = svn propset svn:ignore $(2) $(1)
70
define addDirWithIgnore
71
$(addDir)
72
$(setSvnIgnore)
73
endef
74

    
75
##### Environment
76

    
77
export PATH := $(bin):$(PATH)
78

    
79
##### General targets
80

    
81
all: _always maps ;
82

    
83
clean: _always
84
	$(RM) $(all)
85

    
86
remake: _always clean
87
	+$(selfMake)
88
# re-run make so that cache of existing files is reset
89

    
90
# Only remake if doesn't exist. This prevents unintentional remaking when the
91
# make script is newly checked out from svn (which sets the mod time to now) but
92
# the output is synced externally.
93
# Can't remove prereq to do this, because it determines when the rule applies.
94
make_script = $(if $(wildcard $@),,"time" ./$< >$@)
95

    
96
%/: % _always ;
97

    
98
%: %.make
99
	$(make_script)
100
.PRECIOUS: % # save partial outputs of aborted src make scripts
101

    
102
##### SVN
103

    
104
svnFilesGlob:= */{,{,.}{map,*terms,VegBIEN}.csv{,.*},*header.*,*.sql,test*.xml*}
105
svnFilesGlob := {map.csv,*{schema,~}*.sql,{,*/}*.make,$(svnFilesGlob)}
106
_svnFilesGlob := {_MySQL/{,*schema*.sql,*.make},_src/*.{url,pdf}}
107
svnFiles = $(filter-out _% logs/%,$(call wildcard/,$(svnFilesGlob)))\
108
$(call wildcard/,$(_svnFilesGlob))
109

    
110
add: _always
111
	$(call setSvnIgnore,.,'*')
112
	$(call addDirWithIgnore,logs,$$'*.log.sql\n*.trace')
113
	$(call addDirWithIgnore,verify,'*.out')
114
	$(call addFile,import_order.txt)
115
	$(if $(wildcard _MySQL/),$(call addDirWithIgnore,_MySQL,'*'))
116
	$(if $(wildcard _src/),$(call addDirWithIgnore,_src,'*'))
117
	$(call add*,$(svnFiles))
118

    
119
# Adds a new table subdir
120
%/add: _always
121
	$(call addDirWithIgnore,$*,'*')
122
	$(call addDirWithIgnore,$*/logs,$$'*.log.sql\n*.trace')
123

    
124
##### Existing maps discovery
125

    
126
sortFile := import_order.txt
127

    
128
tables := $(if $(wildcard $(sortFile)),$(shell cat $(sortFile)))
129
    # $(shell) replaces "\n" with " "
130
allSubdirs := $(call wildcard/,*/)
131
allTables := $(call sortFilenames,$(filter-out _% verify logs,$(allSubdirs:%/=%)))
132
joinedTables := $(filter-out $(tables),$(allTables))
133
allTables := $(joinedTables) $(tables)# move joined tables to beginning
134
ifeq ($(tables),)# none specified in sort file
135
tables := $(allTables)
136
endif
137

    
138
anyMap := %/map.csv %/VegBIEN.csv %/unmapped_terms.csv %/new_terms.csv
139

    
140
extsFilter := $(addprefix %.,$(exts))
141
dataOnly = $(filter $(extsFilter),$(1))
142

    
143
anyTest = $*/test.%
144
srcsOnly = $(filter-out $(anyMap) $(anyTest) %/logs,$(call dataOnly,$(1)))
145

    
146
srcDict := map.csv
147

    
148
vocab := $(mappings)/VegCore.csv
149
coreMap := $(mappings)/VegCore-VegBIEN.csv
150
dict := $(mappings)/Veg+-VegCore.csv
151

    
152
viaMaps := $(tables:%=%/map.csv)
153

    
154
autogenMaps := $(subst map.,VegBIEN.,$(viaMaps))
155
directMaps := $(autogenMaps) $(filter-out $(autogenMaps),\
156
$(wildcard */VegBIEN.csv))
157

    
158
##### Sources
159

    
160
srcs = $(call sortFilenames,$(call srcsOnly,$(wildcard $*/*)))
161
nonHeaderSrcs = $(filter-out %/header.csv,$(srcs))
162
isRef = $(if $(nonHeaderSrcs),,1)
163
    # empty subdir, so references an already-installed staging table
164
isXml = $(filter %.xml,$(nonHeaderSrcs))
165
nonXml = $(if $(isXml),,1)
166
isCsv = $(if $(nonHeaderSrcs),$(if $(isXml),,1))
167
    # true if $(srcs) non-empty and contains no *.xml
168
catSrcs = $(bin)/cat$(if $(nonXml),_csv) $(srcs)
169
withCatSrcs = $(catSrcs:$(bin)/%=$(bin)/with_%) --
170

    
171
# Usage: `make {--silent|-s} inputs/<datasrc>/cat` (don't echo make commands)
172
cat: $(tables:%=%/cat) _always ;
173

    
174
%/cat: _always
175
	$(catSrcs)
176

    
177
##### Staging tables installation
178

    
179
srcTable := %.src
180

    
181
dbExports := $(sort $(wildcard *schema*.sql))# schemas first
182
ifeq ($(schema_only),)
183
dbExports += $(sort $(filter-out $(dbExports),$(wildcard *.sql)))# all others
184
endif
185
dbExports := $(strip $(dbExports))# += adds extra whitespace
186
allInstalls := $(if $(dbExports),sql) $(allTables)
187

    
188
install: _always schema $(allInstalls:%=%/install) ;
189

    
190
uninstall: _always confirm_rm_schema rm_schema ;
191
# rm_schema will also drop all staging tables
192

    
193
reinstall: _always uninstall install ;
194

    
195
confirm_rm_schema: _always
196
	$(if $(filter TNRS,$(datasrc)),$(call confirm,WARNING: This will delete the\
197
TNRS cache!,To save it: make backups/TNRS.backup-remake))
198

    
199
schema: _always
200
	-echo 'CREATE SCHEMA "$(datasrc)";'|$(psqlAsBien)
201
# ignore errors if schema exists
202

    
203
rm_schema: _always
204
	echo 'DROP SCHEMA IF EXISTS "$(datasrc)" CASCADE;'|$(psqlAsBien)
205

    
206
installLog := logs/install.log.sql
207

    
208
logInstall = $(if $(log),$(if $(quiet),$(2)$(1)$(installLog) 2>&1,2>&1|tee $(3)\
209
$(1)$(installLog)))
210
logInstallRoot = $(call logInstall,,>)
211
logInstall* = $(call logInstall,$*/,>)
212
logInstall*Add = $(call logInstall,$*/,>>,-a)# append to log
213

    
214
# Must come before %/install to override it
215
sql/install: $(dbExports)
216
	($(inDatasrc); cat $+|pg_dump_limit)|"time" $(psqlAsBien) \
217
--set=schema='"$(datasrc)"' $(logInstallRoot)
218

    
219
# Must come before `%.sql: _MySQL/%.sql` to override it
220
%.sql: %.sql.make
221
	$(make_script)
222

    
223
# The export must be created with:
224
# `--compatible=postgresql --add-locks=false --set-charset --no-create-info`
225
# Must come before `%.sql: _MySQL/%.sql` to override it
226
%.data.sql: _MySQL/%.data.sql
227
	$(if $(wildcard $@),,$(bin)/my2pg.data <$< >$@)
228

    
229
# The export must be created with:
230
# `--compatible=postgresql --add-locks=false --set-charset`
231
# Add `--no-data` to create a schema-only export.
232
%.sql: _MySQL/%.sql
233
	$(if $(wildcard $@),,$(bin)/my2pg <$< >$@)
234

    
235
cleanup = $(if $(wildcard $*/cleanup.sql),($(inDatasrc); cat $*/cleanup.sql)\
236
|"time" $(psqlAsBien) --echo-all --set=table='"$*"' $(logInstall*Add),\
237
(prefix=; . $(bin)/vegbien_dest; env schema=$(datasrc) table=$* $(bin)/csv2db)\
238
$(logInstall*Add))
239

    
240
define exportHeader
241
$(cleanup)
242
echo 'SELECT * FROM "$(datasrc)"."$*" LIMIT 0;'|$(psqlAsBien) \
243
--no-align --field-separator=, --pset=footer=off >$*/header.csv
244
endef
245

    
246
# For staging tables which are derived by joining together other staging tables.
247
%/install %/header.csv: %/create.sql _always
248
	($(inDatasrc); echo 'CREATE TABLE "$*" AS'; cat $<; echo ';')|"time" \
249
$(psqlAsBien) --echo-all --set=schema='"$(datasrc)"' --set=table='"$*"' \
250
$(logInstall*)
251
	$(exportHeader)
252
.PRECIOUS: %/header.csv
253

    
254
%/install: _always
255
	$(if $(isRef),$(exportHeader),$(if $(nonXml),$(import_install_)))
256
	$(if $(wildcard $*/postprocess.sql),($(inDatasrc); cat $*/postprocess.sql;)\
257
|"time" $(psqlAsBien) --echo-all --set=table='"$*"' $(logInstall*Add))
258
define import_install_
259
(prefix=; . $(bin)/vegbien_dest; "time" nice -n +5\
260
env schema=$(datasrc) table=$* $(bin)/csv2db $(catSrcs) $(logInstall*))
261
$(if $(filter $(srcTable),$*),($(inDatasrc);\
262
echo 'ALTER TABLE "$(datasrc)"."$*" RENAME row_num TO "$*.row_num";')|"time"\
263
$(psqlAsBien) --echo-all --set=table='"$*"' $(logInstall*Add))
264
endef
265
# table-scope src table's row_num col to allow joining it with other tables
266

    
267
%/uninstall: _always
268
	echo 'DROP TABLE IF EXISTS "$(datasrc)"."$*" CASCADE;'|$(psqlAsBien)
269

    
270
%/reinstall: _always %/uninstall %/install ;
271

    
272
cleanup: _always $(tables:%=%/cleanup) ;
273

    
274
# WARNING: This removes any index comments, due to a PostgreSQL bug.
275
# This occurs because ALTER TABLE recreates the index but not its comment.
276
%/cleanup: _always
277
	$(cleanup)
278

    
279
##### Maps building
280

    
281
# WARNING: You CANNOT make a subdir using `make inputs/<datasrc>/<subdir>/`.
282
# You must instead make the entire datasource dir: `make inputs/<datasrc>/`
283

    
284
# Maps to (try to) build are added to this
285
maps :=
286

    
287
srcRoot = $(mappings)/root.sh
288
mkSrcMap = $(catSrcs)|(. $(srcRoot); env datasrc=$(datasrc) $(bin)/src_map >$@)
289

    
290
define translate
291
$(bin)/in_place $< $(bin)/canon 1 $(1)
292
$(bin)/in_place $< $(bin)/translate 1 $(1)
293
endef
294
translate? = $(if $(wildcard $(1)),$(translate))
295

    
296
$(srcDict):# empty target in case it doesn't exist
297

    
298
# Via maps cleanup
299
ifneq ($(filter %/.map.csv.last_cleanup,$(MAKECMDGOALS)),)
300
%/.map.csv.last_cleanup: %/map.csv $(vocab) $(dict) $(coreMap) $(srcDict)
301
	$(call translate?,$(srcDict))
302
	$(bin)/in_place $< $(bin)/canon 1 $(vocab)
303
	$(call translate?,$(dict))
304
	touch $@
305
	+$(selfMake) $(<:%/map.csv=%/unmapped_terms.csv)
306
	+$(selfMake) $(<:%/map.csv=%/new_terms.csv)
307
.PRECIOUS: %/.map.csv.last_cleanup
308
else
309
%/map.csv: _always
310
	$(if $(wildcard $@),,$(if $(nonXml),$(mkSrcMap)))
311
	+$(selfMake) $(@:%/map.csv=%/.map.csv.last_cleanup)
312
.PRECIOUS: %/map.csv
313
endif
314

    
315
%/VegBIEN.csv: %/map.csv $(coreMap)
316
	<$< $(bin)/cat_cols 1 2|$(bin)/join $(coreMap)|$(bin)/sort_map >$@
317
maps += $(autogenMaps)
318

    
319
maps: $(maps) _always ;
320

    
321
all += $(maps)
322

    
323
##### Maps validation
324

    
325
# `tail -n +2`: Remove header before running filter_out_ci because filter_out_ci
326
# only removes the header if it matches the vocabulary's header.
327

    
328
%/unmapped_terms.csv: %/map.csv $(coreMap)
329
	tail -n +2 $<|$(bin)/cols 1|$(bin)/filter_out_ci 0 $(coreMap) >$@
330
	$(bin)/autoremove $@
331

    
332
%/new_terms.csv: %/map.csv $(vocab) $(dict) %/unmapped_terms.csv
333
	$(newTerms)
334
	$(bin)/autoremove $@
335
newTerms = tail -n +2 $<|$(bin)/filter_out_ci 0 $(vocab)|$(bin)/filter_out_ci 0\
336
$(dict) $(if $(wildcard $(word 4,$+)),|$(bin)/filter_out_ci 0 $(word 4,$+))\
337
|grep -vE '^"?:' >$@; exit 0# because grep exits nonzero if no match
338

    
339
termsSubdirs := $(tables)
340

    
341
include $(root)/lib/mappings.Makefile
342

    
343
##### External dependencies
344

    
345
$(root)/%: _always
346
	+$(subMake)
347
.PRECIOUS: $(root)/% # let ext. dir's Makefile decide whether to delete on error
348

    
349
##### Mapping
350

    
351
+maps = $(filter %/map.csv %/VegBIEN.csv $(mappings)/%,$(+_))
352
map2db = env in_database=vegbien in_schema=$(datasrc) in_table=$*\
353
out_database=vegbien $(root)/map $(+maps)
354

    
355
##### Import to VegBIEN
356

    
357
profileTest = $(if $(profile),$(if $(test),1))
358
profileOnly = -env profile_to=/dev/fd/3 $(map2db) 3>&1 1>&2|\
359
$(bin)/profile_stats /dev/fd/0
360

    
361
log_ = $*/logs/$(if $(n),n=$(n).,)$(version).log.sql
362
trace = $(log_:.log.sql=.trace)
363
import = -$(if $(profileTest),$(profileOnly),(set -x; "time" env commit=1\
364
$(if $(profile),profile_to=$(trace)) $(map2db)) $(if $(log),\
365
$(if $(n),,&>$(log_)))$(if $(log),$(if $(n), 2>&1|tee -a $(log_))))
366
# don't abort on import errors, which often relate to invalid input data
367

    
368
import: $(tables:%=%/import) _always ;
369

    
370
%/import: %/VegBIEN.csv _always
371
	$(import)
372
# default:
373
%/import: _always ;
374

    
375
##### Log files from import
376

    
377
logs := $(wildcard */logs/*.log.sql */logs/*.trace)
378

    
379
rm_logs: _always
380
	$(RM) $(logs)
381

    
382
##### Verification of import
383

    
384
verifyTables := $(patsubst verify/%.ref,%,$(wildcard verify/*.ref))
385

    
386
verify: $(verifyTables:%=%/verify) _always ;
387

    
388
%/verify: verify/%.ref verify/%.out _always
389
	-$(diffVerbose) $(+_)
390
# don't abort on verification errors, which are expected during development
391
# default:
392
%/verify: verify/%.out _always
393
	$(if $(shell test -e $< && echo t),cat $<)
394
# don't run if verify/%.out's default do-nothing action was used
395
# can't use $(wildcard) because it won't recheck file after verify/%.out is run
396

    
397
psqlExport := "time" $(psqlAsBien) --no-align --field-separator=$$'\t'\
398
--pset=footer=off --pset=null=NULL
399
# Note that using $(inDatasrc) will not work with datasources whose tables are
400
# the same name as VegBIEN tables (likely only VegBank), because the datasource
401
# is first in the search_path.
402
verify = $(if $(reverify),($(inDatasrc); cat $<)|$(psqlExport)\
403
--set=datasource="'$(datasrc)'" >$@)
404

    
405
verify/%.out: $(mappings)/verify.%.sql _always
406
	$(verify)
407
.PRECIOUS: verify/%.out # save partial output in case of error to help debugging
408
# default:
409
verify/%.out: _always ;
410

    
411
all += $(wildcard verify/*.out)
412

    
413
%.ref: %.ref.sql
414
	($(inDatasrc); cat $<)|$(psqlExport) >$@
415
.PRECIOUS: %.ref # there must always be a .ref for the make rules to work
416

    
417
##### Editing import
418

    
419
rotate: _always
420
	echo "UPDATE party SET organizationname = organizationname||'.$(version)'\
421
WHERE organizationname = '$(datasrc)';"|$(psqlAsBien)
422

    
423
rm: _always
424
	echo "DELETE FROM party WHERE organizationname = '$(datasrc)';"|\
425
$(psqlAsBien)
426

    
427
##### Testing
428

    
429
testRefOutput = $(subst .by_col,,$(1))
430
testRef = $(testRefOutput).ref
431
hasOwnRef = $(filter $@,$(call testRefOutput,$@))
432
# filter returns non-empty if they are equal
433

    
434
# `rm $@`: Remove outputs of successful tests to reduce clutter
435
# `$(foreach use_staged...)`: Run with use_staged=1
436
define runTest
437
@echo "Testing $(abspath $@)..."
438
>$@ env test=1 n=$(test_n) $(1) $(foreach use_staged,1,$(map2db))
439
@(set -x; $(diffIgnoreSpace) $(call testRef,$@) $@) 2>&1 && rm $@ || { e=$$?;\
440
$(if $(wildcard $(call testRef,$@)),,cat $@;)\
441
$(if $(hasOwnRef),\
442
{\
443
read -p $(emph)'Accept new test output? (y/n)'$(endEmph) REPLY;\
444
if test "$$REPLY" = y; then\
445
(set -x; $(MAKE) $@-ok --directory=$(realpath .) --makefile=../input.Makefile);\
446
exit 0;\
447
fi;\
448
};,\
449
echo $(emph)"Note: The preceding failed test is compared to another test's\
450
output"$(endEmph);\
451
echo $(emph)"When it fails, this always indicates a bug"$(endEmph);\
452
)\
453
exit $$e;}
454
endef
455

    
456
tests :=
457

    
458
# Requires staging tables. To create them, run `make inputs/<datasrc>/install`.
459
# Non-flat-file inputs fall back to mimicking a successful test
460
%/test.xml: %/VegBIEN.csv _always
461
	$(if $(nonXml),$(call runTest,by_col=))
462
tests += %/test.xml
463

    
464
%/test.by_col.xml: %/VegBIEN.csv _always
465
	$(if $(nonXml),$(call runTest,by_col=1))
466

    
467
# Only run column-based tests if column-based mode enabled, because these tests
468
# are much slower than the row-based tests for small numbers of rows
469
ifneq ($(by_col),)
470
tests += %/test.by_col.xml
471
endif
472

    
473
testOutputs := $(foreach test,$(tests),$(tables:%=$(test)))
474

    
475
.PRECIOUS: $(testOutputs) # save outputs of failed tests so they can be accepted
476

    
477
test: _always $(testOutputs) ;
478

    
479
all += $(wildcard %/test*.xml)
480

    
481
# Accepts a test output: make <test_output_path>-ok
482
%-ok: _always
483
	mv $* $(call testRef,$*)
484

    
485
accept-all: _always
486
	+yes|$(selfMake) test
487

    
488
##### Documentation
489

    
490
steps = $(selfMake) -s $*/import test=1 by_col=1 verbosity=2 n=100\
491
2>&1|$(bin)/debug2redmine >$@
492

    
493
%/logs/steps.by_col.log.sql: _always
494
	+$(steps)
(4-4/4)