Project

General

Profile

1 3761 aaronmk
selfDir_uZPPqC := $(dir $(lastword $(MAKEFILE_LIST)))
2
root := $(selfDir_uZPPqC)..
3
include $(root)/lib/common.Makefile
4
5
6 783 aaronmk
##### Configuration
7
8 1921 aaronmk
# Command line
9 1190 aaronmk
log ?= $(if $(test),,1)
10 1257 aaronmk
profile ?=
11 1983 aaronmk
quiet ?=
12 1620 aaronmk
reverify ?= 1
13 1990 aaronmk
use_staged ?= $(by_col)
14 1921 aaronmk
15
# Makefile
16 1386 aaronmk
exts ?= csv tsv txt xml
17 783 aaronmk
test_n ?= 2
18
19
##### Vars/functions
20
21 1821 aaronmk
# Paths
22
datasrc := $(notdir $(realpath .))
23
bin := $(root)/bin
24
mappings := $(root)/mappings
25
26 368 aaronmk
# Make
27 640 aaronmk
SHELL := /bin/bash
28 1509 aaronmk
selfMake = $(MAKE) --makefile=../input.Makefile
29 1821 aaronmk
subMake = $(MAKE) $(@:$(root)/%=%) --directory=$(root)
30 404 aaronmk
+_ = $(+:_%=)
31 368 aaronmk
addBeforeExt = $(basename $(2))$(1)$(suffix $(2))
32
33 1742 aaronmk
# 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
40 790 aaronmk
# System
41
date = $(shell date +"%Y-%m-%d-%H-%M-%S")
42
43 640 aaronmk
# Terminal
44 1184 aaronmk
termCols := $(shell tput cols)
45 640 aaronmk
esc := '['
46
reset := $(esc)'0m'
47
emph := $(esc)'7m '
48
endEmph := ' '$(reset)
49
50 368 aaronmk
# Commands
51 1245 aaronmk
MKDIR = mkdir -p
52
mkdir = $(MKDIR) $(@D)
53 395 aaronmk
CP = cp -p
54 1184 aaronmk
diff = diff --unified=2
55 3721 aaronmk
diffIgnoreSpace = $(diff) --ignore-space-change
56 1184 aaronmk
diffVerbose = $(if $(verbose),diff --side-by-side --left-column\
57
--width=$(termCols),$(diff))
58 368 aaronmk
59 1918 aaronmk
# BIEN commands
60 4137 aaronmk
sortFilenames = $(shell $(bin)/sort_filenames $(1))
61 1524 aaronmk
selfMap = $(bin)/cols 0 0
62 1081 aaronmk
psqlOpts := --set ON_ERROR_STOP=1 --quiet
63 1289 aaronmk
psqlAsBien := $(bin)/psql_vegbien $(psqlOpts)
64 4408 aaronmk
# Usage: ($(inDatasrc); cat $(file))|$(psqlCmd)
65
inDatasrc = echo 'SET search_path TO "$(datasrc)";'
66 353 aaronmk
67 1594 aaronmk
# SVN
68
addDir = $(if $(wildcard $(1)/),svn add --depth=empty $(1),svn mkdir $(1))
69 1771 aaronmk
setSvnIgnore = svn propset svn:ignore $(2) $(1)
70 1594 aaronmk
define addDirWithIgnore
71
$(addDir)
72 1771 aaronmk
$(setSvnIgnore)
73 1594 aaronmk
endef
74
75 4414 aaronmk
##### Environment
76
77
export PATH := $(bin):$(PATH)
78
79 783 aaronmk
##### General targets
80 250 aaronmk
81 418 aaronmk
all: _always maps ;
82 247 aaronmk
83 383 aaronmk
clean: _always
84 256 aaronmk
	$(RM) $(all)
85
86 1742 aaronmk
remake: _always clean
87
	+$(selfMake)
88
# re-run make so that cache of existing files is reset
89
90 1627 aaronmk
make_script = ./$< >$@
91 368 aaronmk
92 1604 aaronmk
%/: % _always ;
93
94 4379 aaronmk
%: %.make
95 4420 aaronmk
	$(if $(wildcard $@),,"time" $(make_script))
96 4411 aaronmk
# 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 4182 aaronmk
.PRECIOUS: % # save partial outputs of aborted src make scripts
101 1627 aaronmk
102 783 aaronmk
##### SVN
103 250 aaronmk
104 4221 aaronmk
add: _always
105 4182 aaronmk
	$(call setSvnIgnore,.,'*')
106 4459 aaronmk
	$(call addDirWithIgnore,logs,$$'*.log.sql\n*.trace')
107 4124 aaronmk
	$(call addDirWithIgnore,verify,'*.out')
108 4223 aaronmk
	$(call addFile,import_order.txt)
109 766 aaronmk
110 4217 aaronmk
# Adds a new table subdir
111
%/add: _always
112
	$(call addDirWithIgnore,$*,'*')
113
	$(call addDirWithIgnore,$*/logs,$$'*.log.sql\n*.trace')
114
115 1969 aaronmk
##### Existing maps discovery
116 1955 aaronmk
117 4182 aaronmk
sortFile := import_order.txt
118 4181 aaronmk
119 4253 aaronmk
tables := $(if $(wildcard $(sortFile)),$(shell cat $(sortFile)))
120
    # $(shell) replaces "\n" with " "
121 4182 aaronmk
allSubdirs := $(call wildcard/,*/)
122 4464 aaronmk
allTables := $(call sortFilenames,$(filter-out _% verify logs,$(allSubdirs:%/=%)))
123 4256 aaronmk
joinedTables := $(filter-out $(tables),$(allTables))
124
allTables := $(joinedTables) $(tables)# move joined tables to beginning
125 4253 aaronmk
ifeq ($(tables),)# none specified in sort file
126
tables := $(allTables)
127 4181 aaronmk
endif
128 4137 aaronmk
129 4182 aaronmk
viaMap := %/map.csv
130
directMap := %/VegBIEN.csv
131 4645 aaronmk
anyMap := $(viaMap) $(directMap) %/unmapped_terms.csv\
132 4604 aaronmk
%/new_terms.csv
133 3574 aaronmk
134 4214 aaronmk
extsFilter := $(addprefix %.,$(exts))
135
dataOnly = $(filter $(extsFilter),$(1))
136
137 4182 aaronmk
anyTest = $*/test.%
138 4214 aaronmk
srcsOnly = $(filter-out $(anyMap) $(anyTest) %/logs,$(call dataOnly,$(1)))
139 4120 aaronmk
140 4844 aaronmk
vocab := $(mappings)/VegCore.csv
141 4658 aaronmk
coreMap := $(mappings)/VegCore-VegBIEN.csv
142 4687 aaronmk
dict := $(mappings)/Veg+-VegCore.csv
143 783 aaronmk
144 4672 aaronmk
viaMaps := $(tables:%=%/map.csv)
145 783 aaronmk
146 4117 aaronmk
autogenMaps := $(subst map.,VegBIEN.,$(viaMaps))
147 728 aaronmk
directMaps := $(autogenMaps) $(filter-out $(autogenMaps),\
148 4182 aaronmk
$(wildcard */VegBIEN.csv))
149 728 aaronmk
150 1969 aaronmk
##### Sources
151
152 4182 aaronmk
srcs = $(call sortFilenames,$(call srcsOnly,$(wildcard $*/*)))
153 4450 aaronmk
nonHeaderSrcs = $(filter-out %/header.csv,$(srcs))
154
isRef = $(if $(nonHeaderSrcs),,1)
155
    # empty subdir, so references an already-installed staging table
156 4452 aaronmk
isXml = $(filter %.xml,$(nonHeaderSrcs))
157
nonXml = $(if $(isXml),,1)
158
isCsv = $(if $(nonHeaderSrcs),$(if $(isXml),,1))
159 1969 aaronmk
    # true if $(srcs) non-empty and contains no *.xml
160 4465 aaronmk
catSrcs = $(bin)/cat$(if $(nonXml),_csv) $(srcs)
161 1969 aaronmk
withCatSrcs = $(catSrcs:$(bin)/%=$(bin)/with_%) --
162
163 1986 aaronmk
# Usage: `make {--silent|-s} inputs/<datasrc>/cat` (don't echo make commands)
164 4252 aaronmk
cat: $(tables:%=%/cat) _always ;
165 1969 aaronmk
166 4252 aaronmk
%/cat: _always
167 1969 aaronmk
	$(catSrcs)
168
169 4121 aaronmk
##### Staging tables installation
170 1921 aaronmk
171 4534 aaronmk
dbExports := $(sort $(wildcard *.schema.sql))# schemas first
172
dbExports += $(sort $(filter-out $(dbExports),$(wildcard *.sql)))# all others
173 4438 aaronmk
dbExports := $(strip $(dbExports))# += adds extra whitespace
174 4425 aaronmk
allInstalls := $(if $(dbExports),sql) $(allTables)
175 1921 aaronmk
176 4413 aaronmk
install: _always schema $(allInstalls:%=%/install) ;
177
178 4121 aaronmk
uninstall: _always rm_schema ;
179
# rm_schema will also drop all staging tables
180 1921 aaronmk
181
reinstall: _always uninstall install ;
182
183 4121 aaronmk
schema: _always
184 1921 aaronmk
	-echo 'CREATE SCHEMA "$(datasrc)";'|$(psqlAsBien)
185
# ignore errors if schema exists
186
187 4121 aaronmk
rm_schema: _always
188 1921 aaronmk
	echo 'DROP SCHEMA IF EXISTS "$(datasrc)" CASCADE;'|$(psqlAsBien)
189
190 4546 aaronmk
logInstall = $(if $(log),$(if $(quiet),&>,2>&1|tee $(2)\
191
)$(1)logs/install.log.sql)
192 4543 aaronmk
logInstall* = $(call logInstall,$*/)
193 4546 aaronmk
logInstall*Add = $(call logInstall,$*/,-a)# append to log
194 4453 aaronmk
195 4413 aaronmk
# Must come before %/install to override it
196 4434 aaronmk
sql/install: $(dbExports)
197 4510 aaronmk
	($(inDatasrc); cat $+|grep -vF 'SET search_path')|"time" $(psqlAsBien) \
198 4454 aaronmk
$(logInstall)
199 4413 aaronmk
200 4423 aaronmk
# The export must be created with:
201 4429 aaronmk
# `--compatible=postgresql --add-locks=false --set-charset --no-create-info`
202 4431 aaronmk
# Must come before `%.sql: _MySQL/%.sql` to override it
203 4429 aaronmk
%.data.sql: _MySQL/%.data.sql
204
	$(bin)/my2pg.data <$< >$@
205
206 4431 aaronmk
# The export must be created with:
207
# `--compatible=postgresql --add-locks=false --set-charset`
208
# Add `--no-data` to create a schema-only export.
209
%.sql: _MySQL/%.sql
210
	$(bin)/my2pg <$< >$@
211
212 4449 aaronmk
cleanup = (prefix=; . $(bin)/vegbien_dest; env schema=$(datasrc) table=$*\
213 4550 aaronmk
$(bin)/csv2db) $(logInstall*Add)
214 4449 aaronmk
215
define exportHeader
216
$(cleanup)
217
echo 'SELECT * FROM "$(datasrc)"."$*" LIMIT 0;'|$(psqlAsBien) \
218
--no-align --field-separator=, --pset=footer=off >$*/header.csv
219
endef
220
221 4260 aaronmk
# For staging tables which are derived by joining together other staging tables.
222 4263 aaronmk
%/install %/header.csv: %/create.sql _always
223 4527 aaronmk
	($(inDatasrc); echo 'CREATE TABLE "$*" AS'; cat $<; echo ';')|"time" \
224 4543 aaronmk
$(psqlAsBien) --echo-all --set=table='"$*"' $(logInstall*)
225 4548 aaronmk
	-$(addRowNum)
226 4449 aaronmk
	$(exportHeader)
227 4527 aaronmk
# ignore errors if create.sql already added a primary key
228 4548 aaronmk
addRowNum = echo 'ALTER TABLE "$(datasrc)"."$*"\
229
ADD COLUMN row_num serial NOT NULL PRIMARY KEY;'|$(psqlAsBien) $(logInstall*Add)
230 4260 aaronmk
231 4259 aaronmk
# The joined tables must be suffixed with ".src" to prevent the creation of a
232
# row_num column, which collides during joins.
233
isJoinedTable = $(filter %.src,$*)
234
hasRowNum = $(if $(isJoinedTable),,1)
235
236 4252 aaronmk
%/install: _always
237 4456 aaronmk
	$(if $(isRef),$(exportHeader),$(if $(nonXml),$(import_install_)))
238 3476 aaronmk
import_install_ = (prefix=; . $(bin)/vegbien_dest; "time" nice -n +5\
239 4259 aaronmk
env schema=$(datasrc) table=$* has_row_num=$(hasRowNum) $(bin)/csv2db\
240 4543 aaronmk
$(catSrcs) $(logInstall*))
241 1921 aaronmk
242 1777 aaronmk
##### Maps building
243
244 4345 aaronmk
# WARNING: You CANNOT make a subdir using `make inputs/<datasrc>/<subdir>/`.
245
# You must instead make the entire datasource dir: `make inputs/<datasrc>/`
246
247 4646 aaronmk
# Maps to (try to) build are added to this
248 1779 aaronmk
maps :=
249
250 4208 aaronmk
srcRoot = $(mappings)/root.sh
251 3569 aaronmk
mkSrcMap = $(catSrcs)|(. $(srcRoot); env datasrc=$(datasrc) $(bin)/src_map >$@)
252 1779 aaronmk
253 3572 aaronmk
# Via maps cleanup
254 4182 aaronmk
ifneq ($(filter %/.map.csv.last_cleanup,$(MAKECMDGOALS)),)
255 4844 aaronmk
%/.map.csv.last_cleanup: %/map.csv $(vocab) $(dict)
256
	$(bin)/in_place $< $(bin)/canon 1 $(vocab)
257 4659 aaronmk
	$(bin)/in_place $< $(bin)/canon 1 $(dict)
258 4653 aaronmk
	$(bin)/in_place $< $(bin)/translate 1 $(dict)
259 3572 aaronmk
	touch $@
260 4601 aaronmk
	+$(selfMake) $(<:%/map.csv=%/unmapped_terms.csv)
261
	+$(selfMake) $(<:%/map.csv=%/new_terms.csv)
262 3572 aaronmk
# Include comment column so commented mappings are never removed
263
else
264 4644 aaronmk
%/map.csv: _always
265
	$(if $(wildcard $@),,$(if $(nonXml),$(mkSrcMap)))
266 4182 aaronmk
	+$(selfMake) $(@:%/map.csv=%/.map.csv.last_cleanup)
267 3572 aaronmk
endif
268
269 4641 aaronmk
%/VegBIEN.csv: %/map.csv $(coreMap)
270 4656 aaronmk
	<$< $(bin)/cat_cols 1 2|$(bin)/join $(coreMap)|$(bin)/sort_map >$@
271 1530 aaronmk
maps += $(autogenMaps)
272
273 4646 aaronmk
maps: $(maps) _always ;
274 1139 aaronmk
275
all += $(maps)
276
277 1742 aaronmk
##### Maps validation
278
279 4663 aaronmk
# `tail -n +2`: Remove header before running filter_out_ci because filter_out_ci
280
# only removes the header if it matches the vocabulary's header.
281 4601 aaronmk
282 4664 aaronmk
%/unmapped_terms.csv: %/map.csv $(coreMap)
283 4858 aaronmk
	tail -n +2 $<|$(bin)/cols 1|$(bin)/filter_out_ci 0 $(coreMap) >$@
284
	$(bin)/autoremove $@
285 4601 aaronmk
286 4844 aaronmk
%/new_terms.csv: %/map.csv $(vocab) $(dict) %/unmapped_terms.csv
287 4857 aaronmk
	$(newTerms)
288
	$(bin)/autoremove $@
289
newTerms = tail -n +2 $<|$(bin)/filter_out_ci 0 $(vocab)|$(bin)/filter_out_ci 0\
290
$(dict) $(if $(wildcard $(word 4,$+)),|$(bin)/filter_out_ci 0 $(word 4,$+)) >$@
291 4601 aaronmk
292 4600 aaronmk
termsSubdirs := $(tables)
293
294 3764 aaronmk
include $(root)/lib/mappings.Makefile
295 1742 aaronmk
296 1509 aaronmk
##### External dependencies
297
298
$(root)/%: _always
299
	+$(subMake)
300 1628 aaronmk
.PRECIOUS: $(root)/% # let ext. dir's Makefile decide whether to delete on error
301 1509 aaronmk
302 783 aaronmk
##### Mapping
303 368 aaronmk
304 4182 aaronmk
+maps = $(filter %/map.csv %/VegBIEN.csv $(mappings)/%,$(+_))
305 4746 aaronmk
map2db = env in_database=vegbien in_schema=$(datasrc) in_table=$*\
306 4748 aaronmk
out_database=vegbien $(root)/map $(+maps)
307 368 aaronmk
308 783 aaronmk
##### Import to VegBIEN
309 772 aaronmk
310 1853 aaronmk
profileTest = $(if $(profile),$(if $(test),1))
311
profileOnly = -env profile_to=/dev/fd/3 $(map2db) 3>&1 1>&2|\
312
$(bin)/profile_stats /dev/fd/0
313
314 4182 aaronmk
log_ = $*/logs/$(if $(n),n=$(n).,)$(date).log.sql
315 3317 aaronmk
trace = $(log_:.log.sql=.trace)
316 1853 aaronmk
import = -$(if $(profileTest),$(profileOnly),(set -x; "time" env commit=1\
317 1190 aaronmk
$(if $(profile),profile_to=$(trace)) $(map2db)) $(if $(log),\
318 3616 aaronmk
$(if $(n),,&>$(log_)))$(if $(log),$(if $(n), 2>&1|tee -a $(log_))))
319 1088 aaronmk
# don't abort on import errors, which often relate to invalid input data
320 624 aaronmk
321 4252 aaronmk
import: $(tables:%=%/import) _always ;
322 718 aaronmk
323 4252 aaronmk
%/import: %/VegBIEN.csv _always
324 718 aaronmk
	$(import)
325 772 aaronmk
# default:
326 4252 aaronmk
%/import: _always ;
327 718 aaronmk
328 790 aaronmk
##### Log files from import
329
330 4182 aaronmk
logs := $(wildcard */logs/*.log.sql */logs/*.trace)
331 339 aaronmk
332 383 aaronmk
rm_logs: _always
333 339 aaronmk
	$(RM) $(logs)
334
335 783 aaronmk
##### Verification of import
336 264 aaronmk
337 4252 aaronmk
verify: $(tables:%=%/verify) _always ;
338 1194 aaronmk
339 4252 aaronmk
%/verify: verify/%.ref verify/%.out _always
340 1184 aaronmk
	-$(diffVerbose) $(+_)
341 1088 aaronmk
# don't abort on verification errors, which are expected during development
342 1194 aaronmk
# default:
343 4252 aaronmk
%/verify: verify/%.out _always
344 1200 aaronmk
	$(if $(shell test -e $< && echo t),cat $<)
345 1199 aaronmk
# don't run if verify/%.out's default do-nothing action was used
346 1200 aaronmk
# can't use $(wildcard) because it won't recheck file after verify/%.out is run
347 368 aaronmk
348 4401 aaronmk
psqlExport := "time" $(psqlAsBien) --no-align --field-separator=$$'\t'\
349
--pset=footer=off --pset=null=NULL
350
verify = $(if $(reverify),$(psqlExport) --set=datasource="'$(datasrc)'" <$< >$@)
351 514 aaronmk
352 1199 aaronmk
verify/%.out: $(mappings)/verify.%.sql _always
353 1192 aaronmk
	$(verify)
354 1194 aaronmk
# default:
355 1199 aaronmk
verify/%.out: _always ;
356 369 aaronmk
357 1199 aaronmk
all += $(wildcard verify/*.out)
358 1192 aaronmk
359 4402 aaronmk
%.ref: %.ref.sql
360 4409 aaronmk
	($(inDatasrc); cat $<)|$(psqlExport) >$@
361 4402 aaronmk
362 1667 aaronmk
##### Editing import
363
364 4121 aaronmk
rotate: _always
365 1668 aaronmk
	echo "UPDATE party SET organizationname = organizationname||'.$(date)'\
366 1854 aaronmk
WHERE organizationname = '$(datasrc)';"|$(psqlAsBien)
367 1667 aaronmk
368 4121 aaronmk
rm: _always
369 1854 aaronmk
	echo "DELETE FROM party WHERE organizationname = '$(datasrc)';"|\
370
$(psqlAsBien)
371 1667 aaronmk
372 783 aaronmk
##### Testing
373 368 aaronmk
374 4274 aaronmk
testRefOutput = $(subst .by_col,,$(1))
375 1986 aaronmk
testRef = $(testRefOutput).ref
376
hasOwnRef = $(filter $@,$(call testRefOutput,$@))
377
# filter returns non-empty if they are equal
378 630 aaronmk
379 4246 aaronmk
# `rm $@`: Remove outputs of successful tests to reduce clutter
380 4272 aaronmk
# `$(foreach use_staged...)`: Run with use_staged=1
381 897 aaronmk
define runTest
382 776 aaronmk
@echo "Testing $(abspath $@)..."
383 4272 aaronmk
>$@ env test=1 n=$(test_n) $(1) $(foreach use_staged,1,$(map2db))
384 4246 aaronmk
@(set -x; $(diffIgnoreSpace) $(call testRef,$@) $@) 2>&1 && rm $@ || { e=$$?;\
385 780 aaronmk
$(if $(wildcard $(call testRef,$@)),,cat $@;)\
386 777 aaronmk
$(if $(hasOwnRef),\
387 3179 aaronmk
{\
388
read -p $(emph)'Accept new test output? (y/n)'$(endEmph) REPLY;\
389
if test "$$REPLY" = y; then\
390
(set -x; $(MAKE) $@-ok --directory=$(realpath .) --makefile=../input.Makefile);\
391
exit 0;\
392
fi;\
393
};,\
394 777 aaronmk
echo $(emph)"Note: The preceding failed test is compared to another test's\
395
output"$(endEmph);\
396
echo $(emph)"When it fails, this always indicates a bug"$(endEmph);\
397
)\
398 652 aaronmk
exit $$e;}
399 628 aaronmk
endef
400
401 876 aaronmk
tests :=
402
403 4269 aaronmk
# Requires staging tables. To create them, run `make inputs/<datasrc>/install`.
404
# Non-flat-file inputs fall back to mimicking a successful test
405 4182 aaronmk
%/test.xml: %/VegBIEN.csv _always
406 4452 aaronmk
	$(if $(nonXml),$(call runTest,by_col=))
407 4182 aaronmk
tests += %/test.xml
408 627 aaronmk
409 4274 aaronmk
%/test.by_col.xml: %/VegBIEN.csv _always
410 4831 aaronmk
	$(if $(nonXml),$(call runTest,by_col=1))
411 4275 aaronmk
412
# Only run column-based tests if column-based mode enabled, because these tests
413
# are much slower than the row-based tests for small numbers of rows
414
ifneq ($(by_col),)
415 4274 aaronmk
tests += %/test.by_col.xml
416 4275 aaronmk
endif
417 4274 aaronmk
418 876 aaronmk
testOutputs := $(foreach test,$(tests),$(tables:%=$(test)))
419 724 aaronmk
420 1361 aaronmk
.PRECIOUS: $(testOutputs) # save outputs of failed tests so they can be accepted
421
422 876 aaronmk
test: _always $(testOutputs) ;
423
424 4182 aaronmk
all += $(wildcard %/test*.xml)
425 876 aaronmk
426 634 aaronmk
# Accepts a test output: make <test_output_path>-ok
427 626 aaronmk
%-ok: _always
428 4383 aaronmk
	mv $* $(call testRef,$*)
429 502 aaronmk
430 4120 aaronmk
accept-all: _always
431 3579 aaronmk
	+yes|$(selfMake) test
432
433 3133 aaronmk
##### Documentation
434
435 4494 aaronmk
steps = $(selfMake) -s $*/import test=1 by_col=1 verbosity=2 n=100\
436 3385 aaronmk
2>&1|$(bin)/debug2redmine >$@
437 3188 aaronmk
438 4182 aaronmk
%/logs/steps.by_col.log.sql: _always
439 3188 aaronmk
	+$(steps)