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
use_staged ?= $(by_col)
14

    
15
# Makefile
16
exts ?= csv tsv txt xml
17
test_n ?= 2
18

    
19
##### Vars/functions
20

    
21
# Paths
22
datasrc := $(notdir $(realpath .))
23
bin := $(root)/bin
24
mappings := $(root)/mappings
25

    
26
# Make
27
SHELL := /bin/bash
28
selfMake = $(MAKE) --makefile=../input.Makefile
29
subMake = $(MAKE) $(@:$(root)/%=%) --directory=$(root)
30
+_ = $(+:_%=)
31
addBeforeExt = $(basename $(2))$(1)$(suffix $(2))
32

    
33
# OS
34
os := $(shell uname)
35
forOs = $(patsubst %,%-$(filter Linux Darwin,$(os)),$(1))
36

    
37
# Formatting
38
SED = sed -$(if $(filter Darwin,$(os)),E,r)
39

    
40
# System
41
date = $(shell date +"%Y-%m-%d-%H-%M-%S")
42

    
43
# Terminal
44
termCols := $(shell tput cols)
45
esc := '['
46
reset := $(esc)'0m'
47
emph := $(esc)'7m '
48
endEmph := ' '$(reset)
49

    
50
# Commands
51
MKDIR = mkdir -p
52
mkdir = $(MKDIR) $(@D)
53
CP = cp -p
54
diff = diff --unified=2
55
diffIgnoreSpace = $(diff) --ignore-space-change
56
diffVerbose = $(if $(verbose),diff --side-by-side --left-column\
57
--width=$(termCols),$(diff))
58

    
59
# BIEN commands
60
sortFilenames = $(shell $(bin)/sort_filenames $(1))
61
selfMap = $(bin)/cols 0 0
62
psqlOpts := --set ON_ERROR_STOP=1 --quiet
63
psqlAsBien := $(bin)/psql_vegbien $(psqlOpts)
64
# Usage: ($(inDatasrc); cat $(file))|$(psqlCmd)
65
inDatasrc = echo 'SET search_path TO "$(datasrc)";'
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
make_script = ./$< >$@
91

    
92
%/: % _always ;
93

    
94
%: %.make
95
	$(if $(wildcard $@),,"time" $(make_script))
96
# Only remake if doesn't exist. This prevents unintentional remaking when the
97
# make script is newly checked out from svn (which sets the mod time to now) but
98
# the output is synced externally.
99
# Can't remove prereq to do this, because it determines when the rule applies.
100
.PRECIOUS: % # save partial outputs of aborted src make scripts
101

    
102
##### SVN
103

    
104
add: _always
105
	$(call setSvnIgnore,.,'*')
106
	$(call addDirWithIgnore,logs,$$'*.log.sql\n*.trace')
107
	$(call addDirWithIgnore,verify,'*.out')
108
	$(call addFile,import_order.txt)
109

    
110
# Adds a new table subdir
111
%/add: _always
112
	$(call addDirWithIgnore,$*,'*')
113
	$(call addDirWithIgnore,$*/logs,$$'*.log.sql\n*.trace')
114

    
115
##### Existing maps discovery
116

    
117
sortFile := import_order.txt
118

    
119
tables := $(if $(wildcard $(sortFile)),$(shell cat $(sortFile)))
120
    # $(shell) replaces "\n" with " "
121
allSubdirs := $(call wildcard/,*/)
122
allTables := $(call sortFilenames,$(filter-out _% verify logs,$(allSubdirs:%/=%)))
123
joinedTables := $(filter-out $(tables),$(allTables))
124
allTables := $(joinedTables) $(tables)# move joined tables to beginning
125
ifeq ($(tables),)# none specified in sort file
126
tables := $(allTables)
127
endif
128

    
129
viaMap := %/map.csv
130
directMap := %/VegBIEN.csv
131
anyMap := $(viaMap) $(directMap) %/unmapped_terms.csv\
132
%/new_terms.csv
133

    
134
extsFilter := $(addprefix %.,$(exts))
135
dataOnly = $(filter $(extsFilter),$(1))
136

    
137
anyTest = $*/test.%
138
srcsOnly = $(filter-out $(anyMap) $(anyTest) %/logs,$(call dataOnly,$(1)))
139

    
140
via := Veg+
141

    
142
coreMap := $(mappings)/VegCore-VegBIEN.csv
143
vocab := $(mappings)/$(via).vocab.csv
144
dict := $(mappings)/$(via)-VegCore.csv
145

    
146
viaMaps := $(wildcard $(tables:%=%/map.csv))
147
viaMaps += $(filter-out $(viaMaps),$(wildcard */map.csv))
148

    
149
autogenMaps := $(subst map.,VegBIEN.,$(viaMaps))
150
directMaps := $(autogenMaps) $(filter-out $(autogenMaps),\
151
$(wildcard */VegBIEN.csv))
152

    
153
##### Sources
154

    
155
srcs = $(call sortFilenames,$(call srcsOnly,$(wildcard $*/*)))
156
nonHeaderSrcs = $(filter-out %/header.csv,$(srcs))
157
isRef = $(if $(nonHeaderSrcs),,1)
158
    # empty subdir, so references an already-installed staging table
159
isXml = $(filter %.xml,$(nonHeaderSrcs))
160
nonXml = $(if $(isXml),,1)
161
isCsv = $(if $(nonHeaderSrcs),$(if $(isXml),,1))
162
    # true if $(srcs) non-empty and contains no *.xml
163
catSrcs = $(bin)/cat$(if $(nonXml),_csv) $(srcs)
164
withCatSrcs = $(catSrcs:$(bin)/%=$(bin)/with_%) --
165

    
166
# Usage: `make {--silent|-s} inputs/<datasrc>/cat` (don't echo make commands)
167
cat: $(tables:%=%/cat) _always ;
168

    
169
%/cat: _always
170
	$(catSrcs)
171

    
172
##### Staging tables installation
173

    
174
dbExports := $(sort $(wildcard *.schema.sql))# schemas first
175
dbExports += $(sort $(filter-out $(dbExports),$(wildcard *.sql)))# all others
176
dbExports := $(strip $(dbExports))# += adds extra whitespace
177
allInstalls := $(if $(dbExports),sql) $(allTables)
178

    
179
install: _always schema $(allInstalls:%=%/install) ;
180

    
181
uninstall: _always rm_schema ;
182
# rm_schema will also drop all staging tables
183

    
184
reinstall: _always uninstall install ;
185

    
186
schema: _always
187
	-echo 'CREATE SCHEMA "$(datasrc)";'|$(psqlAsBien)
188
# ignore errors if schema exists
189

    
190
rm_schema: _always
191
	echo 'DROP SCHEMA IF EXISTS "$(datasrc)" CASCADE;'|$(psqlAsBien)
192

    
193
logInstall = $(if $(log),$(if $(quiet),&>,2>&1|tee $(2)\
194
)$(1)logs/install.log.sql)
195
logInstall* = $(call logInstall,$*/)
196
logInstall*Add = $(call logInstall,$*/,-a)# append to log
197

    
198
# Must come before %/install to override it
199
sql/install: $(dbExports)
200
	($(inDatasrc); cat $+|grep -vF 'SET search_path')|"time" $(psqlAsBien) \
201
$(logInstall)
202

    
203
# The export must be created with:
204
# `--compatible=postgresql --add-locks=false --set-charset --no-create-info`
205
# Must come before `%.sql: _MySQL/%.sql` to override it
206
%.data.sql: _MySQL/%.data.sql
207
	$(bin)/my2pg.data <$< >$@
208

    
209
# The export must be created with:
210
# `--compatible=postgresql --add-locks=false --set-charset`
211
# Add `--no-data` to create a schema-only export.
212
%.sql: _MySQL/%.sql
213
	$(bin)/my2pg <$< >$@
214

    
215
cleanup = (prefix=; . $(bin)/vegbien_dest; env schema=$(datasrc) table=$*\
216
$(bin)/csv2db) $(logInstall*Add)
217

    
218
define exportHeader
219
$(cleanup)
220
echo 'SELECT * FROM "$(datasrc)"."$*" LIMIT 0;'|$(psqlAsBien) \
221
--no-align --field-separator=, --pset=footer=off >$*/header.csv
222
endef
223

    
224
# For staging tables which are derived by joining together other staging tables.
225
%/install %/header.csv: %/create.sql _always
226
	($(inDatasrc); echo 'CREATE TABLE "$*" AS'; cat $<; echo ';')|"time" \
227
$(psqlAsBien) --echo-all --set=table='"$*"' $(logInstall*)
228
	-$(addRowNum)
229
	$(exportHeader)
230
# ignore errors if create.sql already added a primary key
231
addRowNum = echo 'ALTER TABLE "$(datasrc)"."$*"\
232
ADD COLUMN row_num serial NOT NULL PRIMARY KEY;'|$(psqlAsBien) $(logInstall*Add)
233

    
234
# The joined tables must be suffixed with ".src" to prevent the creation of a
235
# row_num column, which collides during joins.
236
isJoinedTable = $(filter %.src,$*)
237
hasRowNum = $(if $(isJoinedTable),,1)
238

    
239
%/install: _always
240
	$(if $(isRef),$(exportHeader),$(if $(nonXml),$(import_install_)))
241
import_install_ = (prefix=; . $(bin)/vegbien_dest; "time" nice -n +5\
242
env schema=$(datasrc) table=$* has_row_num=$(hasRowNum) $(bin)/csv2db\
243
$(catSrcs) $(logInstall*))
244

    
245
##### Maps building
246

    
247
# WARNING: You CANNOT make a subdir using `make inputs/<datasrc>/<subdir>/`.
248
# You must instead make the entire datasource dir: `make inputs/<datasrc>/`
249

    
250
# Maps to (try to) build are added to this
251
maps :=
252

    
253
srcRoot = $(mappings)/root.sh
254
mkSrcMap = $(catSrcs)|(. $(srcRoot); env datasrc=$(datasrc) $(bin)/src_map >$@)
255

    
256
# Via maps cleanup
257
ifneq ($(filter %/.map.csv.last_cleanup,$(MAKECMDGOALS)),)
258
%/.map.csv.last_cleanup: %/map.csv $(coreMap) $(dict)
259
	$(bin)/in_place $< $(bin)/canon 1 $(coreMap)
260
	$(bin)/in_place $< $(bin)/canon 1 $(dict)
261
	$(bin)/in_place $< $(bin)/translate 1 $(dict)
262
	touch $@
263
	+$(selfMake) $(<:%/map.csv=%/unmapped_terms.csv)
264
	+$(selfMake) $(<:%/map.csv=%/new_terms.csv)
265
# Include comment column so commented mappings are never removed
266
else
267
%/map.csv: _always
268
	$(if $(wildcard $@),,$(if $(nonXml),$(mkSrcMap)))
269
	+$(selfMake) $(@:%/map.csv=%/.map.csv.last_cleanup)
270
endif
271

    
272
%/VegBIEN.csv: %/map.csv $(coreMap)
273
	<$< $(bin)/cat_cols 1 2|$(bin)/join $(coreMap)|$(bin)/sort_map >$@
274
maps += $(autogenMaps)
275

    
276
maps: $(maps) _always ;
277

    
278
all += $(maps)
279

    
280
##### Maps validation
281

    
282
# `tail -n +2`: Remove header before running filter_out_ci because filter_out_ci
283
# only removes the header if it matches the vocabulary's header.
284
define newTerms
285
tail -n +2 $<|$(bin)/filter_out_ci $(1) $(word 2,$+)|$(bin)/cols $(1)$(2) >$@
286
$(bin)/autoremove $@
287
endef
288

    
289
%/unmapped_terms.csv: %/map.csv $(coreMap)
290
	$(call newTerms,1)
291

    
292
%/new_terms.csv: %/map.csv $(vocab) %/unmapped_terms.csv
293
	$(call newTerms,0,$(rmNewTerms))
294
rmNewTerms = $(if $(wildcard $(word 3,$+)),|$(bin)/filter_out_ci 0 $(word 3,$+))
295

    
296
termsSubdirs := $(tables)
297

    
298
include $(root)/lib/mappings.Makefile
299

    
300
##### External dependencies
301

    
302
$(root)/%: _always
303
	+$(subMake)
304
.PRECIOUS: $(root)/% # let ext. dir's Makefile decide whether to delete on error
305

    
306
##### Mapping
307

    
308
+maps = $(filter %/map.csv %/VegBIEN.csv $(mappings)/%,$(+_))
309
map = $(if $(srcs),$(withCatSrcs) $(root)/map $(+maps),\
310
$(shell echo Warning: No input file $*/*.{$(exts)} >&2)false)
311
# need false to run some command, when prefixed by `env ...` below
312
map2db = $(if $(use_staged),env in_database=vegbien in_schema=$(datasrc)\
313
in_table=$*) env out_database=vegbien $(map)
314

    
315
##### Import to VegBIEN
316

    
317
profileTest = $(if $(profile),$(if $(test),1))
318
profileOnly = -env profile_to=/dev/fd/3 $(map2db) 3>&1 1>&2|\
319
$(bin)/profile_stats /dev/fd/0
320

    
321
log_ = $*/logs/$(if $(n),n=$(n).,)$(date).log.sql
322
trace = $(log_:.log.sql=.trace)
323
import = -$(if $(profileTest),$(profileOnly),(set -x; "time" env commit=1\
324
$(if $(profile),profile_to=$(trace)) $(map2db)) $(if $(log),\
325
$(if $(n),,&>$(log_)))$(if $(log),$(if $(n), 2>&1|tee -a $(log_))))
326
# don't abort on import errors, which often relate to invalid input data
327

    
328
import: $(tables:%=%/import) _always ;
329

    
330
%/import: %/VegBIEN.csv _always
331
	$(import)
332
# default:
333
%/import: _always ;
334

    
335
##### Log files from import
336

    
337
logs := $(wildcard */logs/*.log.sql */logs/*.trace)
338

    
339
rm_logs: _always
340
	$(RM) $(logs)
341

    
342
##### Verification of import
343

    
344
verify: $(tables:%=%/verify) _always ;
345

    
346
%/verify: verify/%.ref verify/%.out _always
347
	-$(diffVerbose) $(+_)
348
# don't abort on verification errors, which are expected during development
349
# default:
350
%/verify: verify/%.out _always
351
	$(if $(shell test -e $< && echo t),cat $<)
352
# don't run if verify/%.out's default do-nothing action was used
353
# can't use $(wildcard) because it won't recheck file after verify/%.out is run
354

    
355
psqlExport := "time" $(psqlAsBien) --no-align --field-separator=$$'\t'\
356
--pset=footer=off --pset=null=NULL
357
verify = $(if $(reverify),$(psqlExport) --set=datasource="'$(datasrc)'" <$< >$@)
358

    
359
verify/%.out: $(mappings)/verify.%.sql _always
360
	$(verify)
361
# default:
362
verify/%.out: _always ;
363

    
364
all += $(wildcard verify/*.out)
365

    
366
%.ref: %.ref.sql
367
	($(inDatasrc); cat $<)|$(psqlExport) >$@
368

    
369
##### Editing import
370

    
371
rotate: _always
372
	echo "UPDATE party SET organizationname = organizationname||'.$(date)'\
373
WHERE organizationname = '$(datasrc)';"|$(psqlAsBien)
374

    
375
rm: _always
376
	echo "DELETE FROM party WHERE organizationname = '$(datasrc)';"|\
377
$(psqlAsBien)
378

    
379
##### Testing
380

    
381
testRefOutput = $(subst .by_col,,$(1))
382
testRef = $(testRefOutput).ref
383
hasOwnRef = $(filter $@,$(call testRefOutput,$@))
384
# filter returns non-empty if they are equal
385

    
386
# `rm $@`: Remove outputs of successful tests to reduce clutter
387
# `$(foreach use_staged...)`: Run with use_staged=1
388
define runTest
389
@echo "Testing $(abspath $@)..."
390
>$@ env test=1 n=$(test_n) $(1) $(foreach use_staged,1,$(map2db))
391
@(set -x; $(diffIgnoreSpace) $(call testRef,$@) $@) 2>&1 && rm $@ || { e=$$?;\
392
$(if $(wildcard $(call testRef,$@)),,cat $@;)\
393
$(if $(hasOwnRef),\
394
{\
395
read -p $(emph)'Accept new test output? (y/n)'$(endEmph) REPLY;\
396
if test "$$REPLY" = y; then\
397
(set -x; $(MAKE) $@-ok --directory=$(realpath .) --makefile=../input.Makefile);\
398
exit 0;\
399
fi;\
400
};,\
401
echo $(emph)"Note: The preceding failed test is compared to another test's\
402
output"$(endEmph);\
403
echo $(emph)"When it fails, this always indicates a bug"$(endEmph);\
404
)\
405
exit $$e;}
406
endef
407

    
408
tests :=
409

    
410
# Requires staging tables. To create them, run `make inputs/<datasrc>/install`.
411
# Non-flat-file inputs fall back to mimicking a successful test
412
%/test.xml: %/VegBIEN.csv _always
413
	$(if $(nonXml),$(call runTest,by_col=))
414
tests += %/test.xml
415

    
416
%/test.by_col.xml: %/VegBIEN.csv _always
417
	-$(if $(nonXml),$(call runTest,by_col=1))
418
# Don't abort tester if only by-column test fails
419

    
420
# Only run column-based tests if column-based mode enabled, because these tests
421
# are much slower than the row-based tests for small numbers of rows
422
ifneq ($(by_col),)
423
tests += %/test.by_col.xml
424
endif
425

    
426
testOutputs := $(foreach test,$(tests),$(tables:%=$(test)))
427

    
428
.PRECIOUS: $(testOutputs) # save outputs of failed tests so they can be accepted
429

    
430
test: _always $(testOutputs) ;
431

    
432
all += $(wildcard %/test*.xml)
433

    
434
# Accepts a test output: make <test_output_path>-ok
435
%-ok: _always
436
	mv $* $(call testRef,$*)
437

    
438
accept-all: _always
439
	+yes|$(selfMake) test
440

    
441
##### Documentation
442

    
443
steps = $(selfMake) -s $*/import test=1 by_col=1 verbosity=2 n=100\
444
2>&1|$(bin)/debug2redmine >$@
445

    
446
%/logs/steps.by_col.log.sql: _always
447
	+$(steps)
(4-4/4)