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
dict := $(mappings)/$(via)-VegCore.csv
144

    
145
viaMaps := $(tables:%=%/map.csv)
146

    
147
autogenMaps := $(subst map.,VegBIEN.,$(viaMaps))
148
directMaps := $(autogenMaps) $(filter-out $(autogenMaps),\
149
$(wildcard */VegBIEN.csv))
150

    
151
##### Sources
152

    
153
srcs = $(call sortFilenames,$(call srcsOnly,$(wildcard $*/*)))
154
nonHeaderSrcs = $(filter-out %/header.csv,$(srcs))
155
isRef = $(if $(nonHeaderSrcs),,1)
156
    # empty subdir, so references an already-installed staging table
157
isXml = $(filter %.xml,$(nonHeaderSrcs))
158
nonXml = $(if $(isXml),,1)
159
isCsv = $(if $(nonHeaderSrcs),$(if $(isXml),,1))
160
    # true if $(srcs) non-empty and contains no *.xml
161
catSrcs = $(bin)/cat$(if $(nonXml),_csv) $(srcs)
162
withCatSrcs = $(catSrcs:$(bin)/%=$(bin)/with_%) --
163

    
164
# Usage: `make {--silent|-s} inputs/<datasrc>/cat` (don't echo make commands)
165
cat: $(tables:%=%/cat) _always ;
166

    
167
%/cat: _always
168
	$(catSrcs)
169

    
170
##### Staging tables installation
171

    
172
dbExports := $(sort $(wildcard *.schema.sql))# schemas first
173
dbExports += $(sort $(filter-out $(dbExports),$(wildcard *.sql)))# all others
174
dbExports := $(strip $(dbExports))# += adds extra whitespace
175
allInstalls := $(if $(dbExports),sql) $(allTables)
176

    
177
install: _always schema $(allInstalls:%=%/install) ;
178

    
179
uninstall: _always rm_schema ;
180
# rm_schema will also drop all staging tables
181

    
182
reinstall: _always uninstall install ;
183

    
184
schema: _always
185
	-echo 'CREATE SCHEMA "$(datasrc)";'|$(psqlAsBien)
186
# ignore errors if schema exists
187

    
188
rm_schema: _always
189
	echo 'DROP SCHEMA IF EXISTS "$(datasrc)" CASCADE;'|$(psqlAsBien)
190

    
191
logInstall = $(if $(log),$(if $(quiet),&>,2>&1|tee $(2)\
192
)$(1)logs/install.log.sql)
193
logInstall* = $(call logInstall,$*/)
194
logInstall*Add = $(call logInstall,$*/,-a)# append to log
195

    
196
# Must come before %/install to override it
197
sql/install: $(dbExports)
198
	($(inDatasrc); cat $+|grep -vF 'SET search_path')|"time" $(psqlAsBien) \
199
$(logInstall)
200

    
201
# The export must be created with:
202
# `--compatible=postgresql --add-locks=false --set-charset --no-create-info`
203
# Must come before `%.sql: _MySQL/%.sql` to override it
204
%.data.sql: _MySQL/%.data.sql
205
	$(bin)/my2pg.data <$< >$@
206

    
207
# The export must be created with:
208
# `--compatible=postgresql --add-locks=false --set-charset`
209
# Add `--no-data` to create a schema-only export.
210
%.sql: _MySQL/%.sql
211
	$(bin)/my2pg <$< >$@
212

    
213
cleanup = (prefix=; . $(bin)/vegbien_dest; env schema=$(datasrc) table=$*\
214
$(bin)/csv2db) $(logInstall*Add)
215

    
216
define exportHeader
217
$(cleanup)
218
echo 'SELECT * FROM "$(datasrc)"."$*" LIMIT 0;'|$(psqlAsBien) \
219
--no-align --field-separator=, --pset=footer=off >$*/header.csv
220
endef
221

    
222
# For staging tables which are derived by joining together other staging tables.
223
%/install %/header.csv: %/create.sql _always
224
	($(inDatasrc); echo 'CREATE TABLE "$*" AS'; cat $<; echo ';')|"time" \
225
$(psqlAsBien) --echo-all --set=table='"$*"' $(logInstall*)
226
	-$(addRowNum)
227
	$(exportHeader)
228
# ignore errors if create.sql already added a primary key
229
addRowNum = echo 'ALTER TABLE "$(datasrc)"."$*"\
230
ADD COLUMN row_num serial NOT NULL PRIMARY KEY;'|$(psqlAsBien) $(logInstall*Add)
231

    
232
# The joined tables must be suffixed with ".src" to prevent the creation of a
233
# row_num column, which collides during joins.
234
isJoinedTable = $(filter %.src,$*)
235
hasRowNum = $(if $(isJoinedTable),,1)
236

    
237
%/install: _always
238
	$(if $(isRef),$(exportHeader),$(if $(nonXml),$(import_install_)))
239
import_install_ = (prefix=; . $(bin)/vegbien_dest; "time" nice -n +5\
240
env schema=$(datasrc) table=$* has_row_num=$(hasRowNum) $(bin)/csv2db\
241
$(catSrcs) $(logInstall*))
242

    
243
##### Maps building
244

    
245
# WARNING: You CANNOT make a subdir using `make inputs/<datasrc>/<subdir>/`.
246
# You must instead make the entire datasource dir: `make inputs/<datasrc>/`
247

    
248
# Maps to (try to) build are added to this
249
maps :=
250

    
251
srcRoot = $(mappings)/root.sh
252
mkSrcMap = $(catSrcs)|(. $(srcRoot); env datasrc=$(datasrc) $(bin)/src_map >$@)
253

    
254
# Via maps cleanup
255
ifneq ($(filter %/.map.csv.last_cleanup,$(MAKECMDGOALS)),)
256
%/.map.csv.last_cleanup: %/map.csv $(coreMap) $(dict)
257
	$(bin)/in_place $< $(bin)/canon 1 $(coreMap)
258
	$(bin)/in_place $< $(bin)/canon 1 $(dict)
259
	$(bin)/in_place $< $(bin)/translate 1 $(dict)
260
	touch $@
261
	+$(selfMake) $(<:%/map.csv=%/unmapped_terms.csv)
262
	+$(selfMake) $(<:%/map.csv=%/new_terms.csv)
263
# Include comment column so commented mappings are never removed
264
else
265
%/map.csv: _always
266
	$(if $(wildcard $@),,$(if $(nonXml),$(mkSrcMap)))
267
	+$(selfMake) $(@:%/map.csv=%/.map.csv.last_cleanup)
268
endif
269

    
270
%/VegBIEN.csv: %/map.csv $(coreMap)
271
	<$< $(bin)/cat_cols 1 2|$(bin)/join $(coreMap)|$(bin)/sort_map >$@
272
maps += $(autogenMaps)
273

    
274
maps: $(maps) _always ;
275

    
276
all += $(maps)
277

    
278
##### Maps validation
279

    
280
# `tail -n +2`: Remove header before running filter_out_ci because filter_out_ci
281
# only removes the header if it matches the vocabulary's header.
282
define newTerms
283
tail -n +2 $<|$(bin)/cols $(1)$(2) >$@
284
$(bin)/autoremove $@
285
endef
286

    
287
%/unmapped_terms.csv: %/map.csv $(coreMap)
288
	$(call newTerms,1,|$(bin)/filter_out_ci 0 $(coreMap))
289

    
290
%/new_terms.csv: %/map.csv $(coreMap) $(dict) %/unmapped_terms.csv
291
	$(call newTerms,0,$(rmNewTerms))
292
rmNewTerms = |$(bin)/filter_out_ci 0 $(coreMap)|$(bin)/filter_out_ci 0 $(dict)\
293
$(if $(wildcard $(word 4,$+)),|$(bin)/filter_out_ci 0 $(word 4,$+))
294

    
295
termsSubdirs := $(tables)
296

    
297
include $(root)/lib/mappings.Makefile
298

    
299
##### External dependencies
300

    
301
$(root)/%: _always
302
	+$(subMake)
303
.PRECIOUS: $(root)/% # let ext. dir's Makefile decide whether to delete on error
304

    
305
##### Mapping
306

    
307
+maps = $(filter %/map.csv %/VegBIEN.csv $(mappings)/%,$(+_))
308
map = $(if $(srcs),$(withCatSrcs) $(root)/map $(+maps),\
309
$(shell echo Warning: No input file $*/*.{$(exts)} >&2)false)
310
# need false to run some command, when prefixed by `env ...` below
311
map2db = $(if $(use_staged),env in_database=vegbien in_schema=$(datasrc)\
312
in_table=$*) env out_database=vegbien $(map)
313

    
314
##### Import to VegBIEN
315

    
316
profileTest = $(if $(profile),$(if $(test),1))
317
profileOnly = -env profile_to=/dev/fd/3 $(map2db) 3>&1 1>&2|\
318
$(bin)/profile_stats /dev/fd/0
319

    
320
log_ = $*/logs/$(if $(n),n=$(n).,)$(date).log.sql
321
trace = $(log_:.log.sql=.trace)
322
import = -$(if $(profileTest),$(profileOnly),(set -x; "time" env commit=1\
323
$(if $(profile),profile_to=$(trace)) $(map2db)) $(if $(log),\
324
$(if $(n),,&>$(log_)))$(if $(log),$(if $(n), 2>&1|tee -a $(log_))))
325
# don't abort on import errors, which often relate to invalid input data
326

    
327
import: $(tables:%=%/import) _always ;
328

    
329
%/import: %/VegBIEN.csv _always
330
	$(import)
331
# default:
332
%/import: _always ;
333

    
334
##### Log files from import
335

    
336
logs := $(wildcard */logs/*.log.sql */logs/*.trace)
337

    
338
rm_logs: _always
339
	$(RM) $(logs)
340

    
341
##### Verification of import
342

    
343
verify: $(tables:%=%/verify) _always ;
344

    
345
%/verify: verify/%.ref verify/%.out _always
346
	-$(diffVerbose) $(+_)
347
# don't abort on verification errors, which are expected during development
348
# default:
349
%/verify: verify/%.out _always
350
	$(if $(shell test -e $< && echo t),cat $<)
351
# don't run if verify/%.out's default do-nothing action was used
352
# can't use $(wildcard) because it won't recheck file after verify/%.out is run
353

    
354
psqlExport := "time" $(psqlAsBien) --no-align --field-separator=$$'\t'\
355
--pset=footer=off --pset=null=NULL
356
verify = $(if $(reverify),$(psqlExport) --set=datasource="'$(datasrc)'" <$< >$@)
357

    
358
verify/%.out: $(mappings)/verify.%.sql _always
359
	$(verify)
360
# default:
361
verify/%.out: _always ;
362

    
363
all += $(wildcard verify/*.out)
364

    
365
%.ref: %.ref.sql
366
	($(inDatasrc); cat $<)|$(psqlExport) >$@
367

    
368
##### Editing import
369

    
370
rotate: _always
371
	echo "UPDATE party SET organizationname = organizationname||'.$(date)'\
372
WHERE organizationname = '$(datasrc)';"|$(psqlAsBien)
373

    
374
rm: _always
375
	echo "DELETE FROM party WHERE organizationname = '$(datasrc)';"|\
376
$(psqlAsBien)
377

    
378
##### Testing
379

    
380
testRefOutput = $(subst .by_col,,$(1))
381
testRef = $(testRefOutput).ref
382
hasOwnRef = $(filter $@,$(call testRefOutput,$@))
383
# filter returns non-empty if they are equal
384

    
385
# `rm $@`: Remove outputs of successful tests to reduce clutter
386
# `$(foreach use_staged...)`: Run with use_staged=1
387
define runTest
388
@echo "Testing $(abspath $@)..."
389
>$@ env test=1 n=$(test_n) $(1) $(foreach use_staged,1,$(map2db))
390
@(set -x; $(diffIgnoreSpace) $(call testRef,$@) $@) 2>&1 && rm $@ || { e=$$?;\
391
$(if $(wildcard $(call testRef,$@)),,cat $@;)\
392
$(if $(hasOwnRef),\
393
{\
394
read -p $(emph)'Accept new test output? (y/n)'$(endEmph) REPLY;\
395
if test "$$REPLY" = y; then\
396
(set -x; $(MAKE) $@-ok --directory=$(realpath .) --makefile=../input.Makefile);\
397
exit 0;\
398
fi;\
399
};,\
400
echo $(emph)"Note: The preceding failed test is compared to another test's\
401
output"$(endEmph);\
402
echo $(emph)"When it fails, this always indicates a bug"$(endEmph);\
403
)\
404
exit $$e;}
405
endef
406

    
407
tests :=
408

    
409
# Requires staging tables. To create them, run `make inputs/<datasrc>/install`.
410
# Non-flat-file inputs fall back to mimicking a successful test
411
%/test.xml: %/VegBIEN.csv _always
412
	$(if $(nonXml),$(call runTest,by_col=))
413
tests += %/test.xml
414

    
415
%/test.by_col.xml: %/VegBIEN.csv _always
416
	-$(if $(nonXml),$(call runTest,by_col=1))
417
# Don't abort tester if only by-column test fails
418

    
419
# Only run column-based tests if column-based mode enabled, because these tests
420
# are much slower than the row-based tests for small numbers of rows
421
ifneq ($(by_col),)
422
tests += %/test.by_col.xml
423
endif
424

    
425
testOutputs := $(foreach test,$(tests),$(tables:%=$(test)))
426

    
427
.PRECIOUS: $(testOutputs) # save outputs of failed tests so they can be accepted
428

    
429
test: _always $(testOutputs) ;
430

    
431
all += $(wildcard %/test*.xml)
432

    
433
# Accepts a test output: make <test_output_path>-ok
434
%-ok: _always
435
	mv $* $(call testRef,$*)
436

    
437
accept-all: _always
438
	+yes|$(selfMake) test
439

    
440
##### Documentation
441

    
442
steps = $(selfMake) -s $*/import test=1 by_col=1 verbosity=2 n=100\
443
2>&1|$(bin)/debug2redmine >$@
444

    
445
%/logs/steps.by_col.log.sql: _always
446
	+$(steps)
(4-4/4)