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 dmp xml
17
test_n ?= 2
18

    
19
##### Vars/functions
20

    
21
# Paths
22
datasrc := $(patsubst .%,%,$(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
# Terminal
34
termCols := $(shell tput cols)
35
esc := '['
36
reset := $(esc)'0m'
37
emph := $(esc)'7m '
38
endEmph := ' '$(reset)
39

    
40
# User interaction
41

    
42
confirm = $(if $(shell read -p $(emph)"$(1)"$(endEmph)$$'$(if\
43
$(2),\n$(2))\nContinue? (y/n) ' REPLY; test "$$REPLY" = y && echo t),,\
44
$(error Aborting))
45

    
46
# Commands
47
MKDIR = mkdir -p
48
mkdir = $(MKDIR) $(@D)
49
CP = cp -p
50
diff = diff --unified=2
51
diffIgnoreSpace = $(diff) --ignore-space-change
52
diffVerbose = $(if $(verbose),diff --side-by-side --left-column\
53
--width=$(termCols),$(diff))
54

    
55
# BIEN commands
56
sortFilenames = $(shell $(bin)/sort_filenames $(1))
57
selfMap = $(bin)/cols 0 0
58
psqlOpts := --set ON_ERROR_STOP=1 --quiet
59
psqlAsBien := $(bin)/psql_vegbien $(psqlOpts)
60
searchPath := $(datasrc),$(shell prefix=; . $(bin)/vegbien_dest;\
61
echo "$$schemas")
62
searchPath := "$(subst $(comma),"$(comma)",$(searchPath))"
63
# Usage: ($(inDatasrc); cat $(file))|$(psqlCmd)
64
inDatasrc := echo 'SET search_path TO $(searchPath);'
65

    
66
# SVN
67
addDir = $(if $(wildcard $(1)/),svn add --depth=empty $(1),svn mkdir $(1))
68
setSvnIgnore = svn propset svn:ignore $(2) $(1)
69
define addDirWithIgnore
70
$(addDir)
71
$(setSvnIgnore)
72
endef
73

    
74
##### Environment
75

    
76
export PATH := $(bin):$(PATH)
77

    
78
##### General targets
79

    
80
all: _always maps ;
81

    
82
clean: _always
83
	$(RM) $(all)
84

    
85
remake: _always clean
86
	+$(selfMake)
87
# re-run make so that cache of existing files is reset
88

    
89
make_script = ./$< >$@
90

    
91
%/: % _always ;
92

    
93
%: %.make
94
	$(if $(wildcard $@),,"time" $(make_script))
95
# Only remake if doesn't exist. This prevents unintentional remaking when the
96
# make script is newly checked out from svn (which sets the mod time to now) but
97
# the output is synced externally.
98
# Can't remove prereq to do this, because it determines when the rule applies.
99
.PRECIOUS: % # save partial outputs of aborted src make scripts
100

    
101
##### SVN
102

    
103
add: _always
104
	$(call setSvnIgnore,.,'*')
105
	$(call addDirWithIgnore,logs,$$'*.log.sql\n*.trace')
106
	$(call addDirWithIgnore,verify,'*.out')
107
	$(call addFile,import_order.txt)
108

    
109
# Adds a new table subdir
110
%/add: _always
111
	$(call addDirWithIgnore,$*,'*')
112
	$(call addDirWithIgnore,$*/logs,$$'*.log.sql\n*.trace')
113

    
114
##### Existing maps discovery
115

    
116
sortFile := import_order.txt
117

    
118
tables := $(if $(wildcard $(sortFile)),$(shell cat $(sortFile)))
119
    # $(shell) replaces "\n" with " "
120
allSubdirs := $(call wildcard/,*/)
121
allTables := $(call sortFilenames,$(filter-out _% verify logs,$(allSubdirs:%/=%)))
122
joinedTables := $(filter-out $(tables),$(allTables))
123
allTables := $(joinedTables) $(tables)# move joined tables to beginning
124
ifeq ($(tables),)# none specified in sort file
125
tables := $(allTables)
126
endif
127

    
128
anyMap := %/map.csv %/VegBIEN.csv %/unmapped_terms.csv %/new_terms.csv
129

    
130
extsFilter := $(addprefix %.,$(exts))
131
dataOnly = $(filter $(extsFilter),$(1))
132

    
133
anyTest = $*/test.%
134
srcsOnly = $(filter-out $(anyMap) $(anyTest) %/logs,$(call dataOnly,$(1)))
135

    
136
vocab := $(mappings)/VegCore.csv
137
coreMap := $(mappings)/VegCore-VegBIEN.csv
138
dict := $(mappings)/Veg+-VegCore.csv
139

    
140
viaMaps := $(tables:%=%/map.csv)
141

    
142
autogenMaps := $(subst map.,VegBIEN.,$(viaMaps))
143
directMaps := $(autogenMaps) $(filter-out $(autogenMaps),\
144
$(wildcard */VegBIEN.csv))
145

    
146
##### Sources
147

    
148
srcs = $(call sortFilenames,$(call srcsOnly,$(wildcard $*/*)))
149
nonHeaderSrcs = $(filter-out %/header.csv,$(srcs))
150
isRef = $(if $(nonHeaderSrcs),,1)
151
    # empty subdir, so references an already-installed staging table
152
isXml = $(filter %.xml,$(nonHeaderSrcs))
153
nonXml = $(if $(isXml),,1)
154
isCsv = $(if $(nonHeaderSrcs),$(if $(isXml),,1))
155
    # true if $(srcs) non-empty and contains no *.xml
156
catSrcs = $(bin)/cat$(if $(nonXml),_csv) $(srcs)
157
withCatSrcs = $(catSrcs:$(bin)/%=$(bin)/with_%) --
158

    
159
# Usage: `make {--silent|-s} inputs/<datasrc>/cat` (don't echo make commands)
160
cat: $(tables:%=%/cat) _always ;
161

    
162
%/cat: _always
163
	$(catSrcs)
164

    
165
##### Staging tables installation
166

    
167
srcTable := %.src
168

    
169
dbExports := $(sort $(wildcard *.schema.sql))# schemas first
170
dbExports += $(sort $(filter-out $(dbExports),$(wildcard *.sql)))# all others
171
dbExports := $(strip $(dbExports))# += adds extra whitespace
172
allInstalls := $(if $(dbExports),sql) $(allTables)
173

    
174
install: _always schema $(allInstalls:%=%/install) ;
175

    
176
uninstall: _always confirm_rm_schema rm_schema ;
177
# rm_schema will also drop all staging tables
178

    
179
reinstall: _always uninstall install ;
180

    
181
confirm_rm_schema: _always
182
	$(if $(filter TNRS,$(datasrc)),$(call confirm,WARNING: This will delete the\
183
TNRS cache!,To save it: make backups/TNRS.backup-remake))
184

    
185
schema: _always
186
	-echo 'CREATE SCHEMA "$(datasrc)";'|$(psqlAsBien)
187
# ignore errors if schema exists
188

    
189
rm_schema: _always
190
	echo 'DROP SCHEMA IF EXISTS "$(datasrc)" CASCADE;'|$(psqlAsBien)
191

    
192
installLog := logs/install.log.sql
193

    
194
logInstall = $(if $(log),$(if $(quiet),$(2)$(1)$(installLog) 2>&1,2>&1|tee $(3)\
195
$(1)$(installLog)))
196
logInstallRoot = $(call logInstall,,>)
197
logInstall* = $(call logInstall,$*/,>)
198
logInstall*Add = $(call logInstall,$*/,>>,-a)# append to log
199

    
200
# Must come before %/install to override it
201
sql/install: $(dbExports)
202
	($(inDatasrc); cat $+|grep -vF 'SET search_path')|"time" $(psqlAsBien) \
203
$(logInstallRoot)
204

    
205
# The export must be created with:
206
# `--compatible=postgresql --add-locks=false --set-charset --no-create-info`
207
# Must come before `%.sql: _MySQL/%.sql` to override it
208
%.data.sql: _MySQL/%.data.sql
209
	$(bin)/my2pg.data <$< >$@
210

    
211
# The export must be created with:
212
# `--compatible=postgresql --add-locks=false --set-charset`
213
# Add `--no-data` to create a schema-only export.
214
%.sql: _MySQL/%.sql
215
	$(bin)/my2pg <$< >$@
216

    
217
cleanup = (prefix=; . $(bin)/vegbien_dest; env schema=$(datasrc) table=$*\
218
$(bin)/csv2db) $(logInstall*Add)
219

    
220
define exportHeader
221
$(cleanup)
222
echo 'SELECT * FROM "$(datasrc)"."$*" LIMIT 0;'|$(psqlAsBien) \
223
--no-align --field-separator=, --pset=footer=off >$*/header.csv
224
endef
225

    
226
# For staging tables which are derived by joining together other staging tables.
227
%/install %/header.csv: %/create.sql _always
228
	($(inDatasrc); echo 'CREATE TABLE "$*" AS'; cat $<; echo ';')|"time" \
229
$(psqlAsBien) --echo-all --set=table='"$*"' $(logInstall*)
230
	$(exportHeader)
231
.PRECIOUS: %/header.csv
232

    
233
%/install: _always
234
	$(if $(isRef),$(exportHeader),$(if $(nonXml),$(import_install_)))
235
define import_install_
236
(prefix=; . $(bin)/vegbien_dest; "time" nice -n +5\
237
env schema=$(datasrc) table=$* $(bin)/csv2db $(catSrcs) $(logInstall*))
238
$(if $(filter $(srcTable),$*),($(inDatasrc);\
239
echo 'ALTER TABLE "$(datasrc)"."$*" RENAME row_num TO "$*.row_num";')|"time"\
240
$(psqlAsBien) --echo-all --set=table='"$*"' $(logInstall*Add))
241
endef
242
# table-scope src table's row_num col to allow joining it with other tables
243

    
244
cleanup: _always $(tables:%=%/cleanup) ;
245

    
246
# WARNING: This removes any index comments, due to a PostgreSQL bug.
247
# This occurs because ALTER TABLE recreates the index but not its comment.
248
%/cleanup: _always
249
	$(cleanup)
250
	$(if $(wildcard $*/cleanup.sql),($(inDatasrc); cat $*/cleanup.sql;)\
251
|"time" $(psqlAsBien) --echo-all --set=table='"$*"' $(logInstall*Add))
252

    
253
##### Maps building
254

    
255
# WARNING: You CANNOT make a subdir using `make inputs/<datasrc>/<subdir>/`.
256
# You must instead make the entire datasource dir: `make inputs/<datasrc>/`
257

    
258
# Maps to (try to) build are added to this
259
maps :=
260

    
261
srcRoot = $(mappings)/root.sh
262
mkSrcMap = $(catSrcs)|(. $(srcRoot); env datasrc=$(datasrc) $(bin)/src_map >$@)
263

    
264
# Via maps cleanup
265
ifneq ($(filter %/.map.csv.last_cleanup,$(MAKECMDGOALS)),)
266
%/.map.csv.last_cleanup: %/map.csv $(vocab) $(dict) $(coreMap)
267
	$(bin)/in_place $< $(bin)/canon 1 $(vocab)
268
	$(bin)/in_place $< $(bin)/canon 1 $(dict)
269
	$(bin)/in_place $< $(bin)/translate 1 $(dict)
270
	touch $@
271
	+$(selfMake) $(<:%/map.csv=%/unmapped_terms.csv)
272
	+$(selfMake) $(<:%/map.csv=%/new_terms.csv)
273
.PRECIOUS: %/.map.csv.last_cleanup
274
else
275
%/map.csv: _always
276
	$(if $(wildcard $@),,$(if $(nonXml),$(mkSrcMap)))
277
	+$(selfMake) $(@:%/map.csv=%/.map.csv.last_cleanup)
278
.PRECIOUS: %/map.csv
279
endif
280

    
281
%/VegBIEN.csv: %/map.csv $(coreMap)
282
	<$< $(bin)/cat_cols 1 2|$(bin)/join $(coreMap)|$(bin)/sort_map >$@
283
maps += $(autogenMaps)
284

    
285
maps: $(maps) _always ;
286

    
287
all += $(maps)
288

    
289
##### Maps validation
290

    
291
# `tail -n +2`: Remove header before running filter_out_ci because filter_out_ci
292
# only removes the header if it matches the vocabulary's header.
293

    
294
%/unmapped_terms.csv: %/map.csv $(coreMap)
295
	tail -n +2 $<|$(bin)/cols 1|$(bin)/filter_out_ci 0 $(coreMap) >$@
296
	$(bin)/autoremove $@
297

    
298
%/new_terms.csv: %/map.csv $(vocab) $(dict) %/unmapped_terms.csv
299
	$(newTerms)
300
	$(bin)/autoremove $@
301
newTerms = tail -n +2 $<|$(bin)/filter_out_ci 0 $(vocab)|$(bin)/filter_out_ci 0\
302
$(dict) $(if $(wildcard $(word 4,$+)),|$(bin)/filter_out_ci 0 $(word 4,$+)) >$@
303

    
304
termsSubdirs := $(tables)
305

    
306
include $(root)/lib/mappings.Makefile
307

    
308
##### External dependencies
309

    
310
$(root)/%: _always
311
	+$(subMake)
312
.PRECIOUS: $(root)/% # let ext. dir's Makefile decide whether to delete on error
313

    
314
##### Mapping
315

    
316
+maps = $(filter %/map.csv %/VegBIEN.csv $(mappings)/%,$(+_))
317
map2db = env in_database=vegbien in_schema=$(datasrc) in_table=$*\
318
out_database=vegbien $(root)/map $(+maps)
319

    
320
##### Import to VegBIEN
321

    
322
profileTest = $(if $(profile),$(if $(test),1))
323
profileOnly = -env profile_to=/dev/fd/3 $(map2db) 3>&1 1>&2|\
324
$(bin)/profile_stats /dev/fd/0
325

    
326
log_ = $*/logs/$(if $(n),n=$(n).,)$(version).log.sql
327
trace = $(log_:.log.sql=.trace)
328
import = -$(if $(profileTest),$(profileOnly),(set -x; "time" env commit=1\
329
$(if $(profile),profile_to=$(trace)) $(map2db)) $(if $(log),\
330
$(if $(n),,&>$(log_)))$(if $(log),$(if $(n), 2>&1|tee -a $(log_))))
331
# don't abort on import errors, which often relate to invalid input data
332

    
333
import: $(tables:%=%/import) _always ;
334

    
335
%/import: %/VegBIEN.csv _always
336
	$(import)
337
# default:
338
%/import: _always ;
339

    
340
##### Log files from import
341

    
342
logs := $(wildcard */logs/*.log.sql */logs/*.trace)
343

    
344
rm_logs: _always
345
	$(RM) $(logs)
346

    
347
##### Verification of import
348

    
349
verify: $(tables:%=%/verify) _always ;
350

    
351
%/verify: verify/%.ref verify/%.out _always
352
	-$(diffVerbose) $(+_)
353
# don't abort on verification errors, which are expected during development
354
# default:
355
%/verify: verify/%.out _always
356
	$(if $(shell test -e $< && echo t),cat $<)
357
# don't run if verify/%.out's default do-nothing action was used
358
# can't use $(wildcard) because it won't recheck file after verify/%.out is run
359

    
360
psqlExport := "time" $(psqlAsBien) --no-align --field-separator=$$'\t'\
361
--pset=footer=off --pset=null=NULL
362
verify = $(if $(reverify),$(psqlExport) --set=datasource="'$(datasrc)'" <$< >$@)
363

    
364
verify/%.out: $(mappings)/verify.%.sql _always
365
	$(verify)
366
# default:
367
verify/%.out: _always ;
368

    
369
all += $(wildcard verify/*.out)
370

    
371
%.ref: %.ref.sql
372
	($(inDatasrc); cat $<)|$(psqlExport) >$@
373

    
374
##### Editing import
375

    
376
rotate: _always
377
	echo "UPDATE party SET organizationname = organizationname||'.$(version)'\
378
WHERE organizationname = '$(datasrc)';"|$(psqlAsBien)
379

    
380
rm: _always
381
	echo "DELETE FROM party WHERE organizationname = '$(datasrc)';"|\
382
$(psqlAsBien)
383

    
384
##### Testing
385

    
386
testRefOutput = $(subst .by_col,,$(1))
387
testRef = $(testRefOutput).ref
388
hasOwnRef = $(filter $@,$(call testRefOutput,$@))
389
# filter returns non-empty if they are equal
390

    
391
# `rm $@`: Remove outputs of successful tests to reduce clutter
392
# `$(foreach use_staged...)`: Run with use_staged=1
393
define runTest
394
@echo "Testing $(abspath $@)..."
395
>$@ env test=1 n=$(test_n) $(1) $(foreach use_staged,1,$(map2db))
396
@(set -x; $(diffIgnoreSpace) $(call testRef,$@) $@) 2>&1 && rm $@ || { e=$$?;\
397
$(if $(wildcard $(call testRef,$@)),,cat $@;)\
398
$(if $(hasOwnRef),\
399
{\
400
read -p $(emph)'Accept new test output? (y/n)'$(endEmph) REPLY;\
401
if test "$$REPLY" = y; then\
402
(set -x; $(MAKE) $@-ok --directory=$(realpath .) --makefile=../input.Makefile);\
403
exit 0;\
404
fi;\
405
};,\
406
echo $(emph)"Note: The preceding failed test is compared to another test's\
407
output"$(endEmph);\
408
echo $(emph)"When it fails, this always indicates a bug"$(endEmph);\
409
)\
410
exit $$e;}
411
endef
412

    
413
tests :=
414

    
415
# Requires staging tables. To create them, run `make inputs/<datasrc>/install`.
416
# Non-flat-file inputs fall back to mimicking a successful test
417
%/test.xml: %/VegBIEN.csv _always
418
	$(if $(nonXml),$(call runTest,by_col=))
419
tests += %/test.xml
420

    
421
%/test.by_col.xml: %/VegBIEN.csv _always
422
	$(if $(nonXml),$(call runTest,by_col=1))
423

    
424
# Only run column-based tests if column-based mode enabled, because these tests
425
# are much slower than the row-based tests for small numbers of rows
426
ifneq ($(by_col),)
427
tests += %/test.by_col.xml
428
endif
429

    
430
testOutputs := $(foreach test,$(tests),$(tables:%=$(test)))
431

    
432
.PRECIOUS: $(testOutputs) # save outputs of failed tests so they can be accepted
433

    
434
test: _always $(testOutputs) ;
435

    
436
all += $(wildcard %/test*.xml)
437

    
438
# Accepts a test output: make <test_output_path>-ok
439
%-ok: _always
440
	mv $* $(call testRef,$*)
441

    
442
accept-all: _always
443
	+yes|$(selfMake) test
444

    
445
##### Documentation
446

    
447
steps = $(selfMake) -s $*/import test=1 by_col=1 verbosity=2 n=100\
448
2>&1|$(bin)/debug2redmine >$@
449

    
450
%/logs/steps.by_col.log.sql: _always
451
	+$(steps)
(4-4/4)