Project

General

Profile

1 783 aaronmk
##### Configuration
2
3 1921 aaronmk
# Command line
4 1190 aaronmk
log ?= $(if $(test),,1)
5 1257 aaronmk
profile ?=
6 1983 aaronmk
quiet ?=
7 1620 aaronmk
reverify ?= 1
8 1990 aaronmk
use_staged ?= $(by_col)
9 1921 aaronmk
10
# Makefile
11
tablesSort ?= plots organisms stems specimens
12 1386 aaronmk
exts ?= csv tsv txt xml
13 783 aaronmk
test_n ?= 2
14
15
##### Vars/functions
16
17 640 aaronmk
selfDir_uZPPqC := $(dir $(lastword $(MAKEFILE_LIST)))
18 250 aaronmk
19 1821 aaronmk
# Paths
20
datasrc := $(notdir $(realpath .))
21
root := $(selfDir_uZPPqC)..
22
bin := $(root)/bin
23
mappings := $(root)/mappings
24
25 368 aaronmk
# Make
26 640 aaronmk
SHELL := /bin/bash
27 1509 aaronmk
selfMake = $(MAKE) --makefile=../input.Makefile
28 1821 aaronmk
subMake = $(MAKE) $(@:$(root)/%=%) --directory=$(root)
29 404 aaronmk
+_ = $(+:_%=)
30 368 aaronmk
addBeforeExt = $(basename $(2))$(1)$(suffix $(2))
31
32 1742 aaronmk
# OS
33
os := $(shell uname)
34
forOs = $(patsubst %,%-$(filter Linux Darwin,$(os)),$(1))
35
36
# Formatting
37
SED = sed -$(if $(filter Darwin,$(os)),E,r)
38
39 790 aaronmk
# System
40
date = $(shell date +"%Y-%m-%d-%H-%M-%S")
41
42 640 aaronmk
# Terminal
43 1184 aaronmk
termCols := $(shell tput cols)
44 640 aaronmk
esc := '['
45
reset := $(esc)'0m'
46
emph := $(esc)'7m '
47
endEmph := ' '$(reset)
48
49 368 aaronmk
# Commands
50 1245 aaronmk
MKDIR = mkdir -p
51
mkdir = $(MKDIR) $(@D)
52 395 aaronmk
CP = cp -p
53 1184 aaronmk
diff = diff --unified=2
54
diffVerbose = $(if $(verbose),diff --side-by-side --left-column\
55
--width=$(termCols),$(diff))
56 368 aaronmk
57 1918 aaronmk
# BIEN commands
58 1524 aaronmk
selfMap = $(bin)/cols 0 0
59 1081 aaronmk
psqlOpts := --set ON_ERROR_STOP=1 --quiet
60 1289 aaronmk
psqlAsBien := $(bin)/psql_vegbien $(psqlOpts)
61 353 aaronmk
62 1594 aaronmk
# SVN
63
addDir = $(if $(wildcard $(1)/),svn add --depth=empty $(1),svn mkdir $(1))
64 1771 aaronmk
setSvnIgnore = svn propset svn:ignore $(2) $(1)
65 1594 aaronmk
define addDirWithIgnore
66
$(addDir)
67 1771 aaronmk
$(setSvnIgnore)
68 1594 aaronmk
endef
69
70 783 aaronmk
##### General targets
71 250 aaronmk
72 418 aaronmk
all: _always maps ;
73 247 aaronmk
74 1260 aaronmk
.SUFFIXES: # turn off built-in suffix rules
75
.SECONDARY: # don't automatically delete intermediate files
76 1354 aaronmk
.DELETE_ON_ERROR: # delete target if recipe fails
77 247 aaronmk
78 383 aaronmk
_always:
79
.PHONY: _always
80 250 aaronmk
81 383 aaronmk
clean: _always
82 256 aaronmk
	$(RM) $(all)
83
84 1742 aaronmk
remake: _always clean
85
	+$(selfMake)
86
# re-run make so that cache of existing files is reset
87
88 1627 aaronmk
make_script = ./$< >$@
89 368 aaronmk
90 1604 aaronmk
%/: % _always ;
91
92 1627 aaronmk
# Must come before `%: %.make` to override it
93
src/%: src/%.make _always
94 3317 aaronmk
	(set -x; $(make_script)) 2>>$<.log.sql
95 1627 aaronmk
.PRECIOUS: src/% # save partial outputs of aborted src make scripts
96
97
%: %.make _always
98
	$(make_script)
99
100 783 aaronmk
##### SVN
101 250 aaronmk
102 1594 aaronmk
add: _always
103 1771 aaronmk
	$(call setSvnIgnore,.,$$'')
104 1594 aaronmk
	$(call addDirWithIgnore,src,$$'*')
105
	$(call addDirWithIgnore,maps,$$'.~*')
106
	$(call addDirWithIgnore,import,$$'*')
107
	$(call addDirWithIgnore,test,$$'*.out\n*.xml')
108
	$(call addDirWithIgnore,verify,$$'*.out')
109 766 aaronmk
110 1969 aaronmk
##### Existing maps discovery
111 1955 aaronmk
112
inputFiles := $(wildcard $(exts:%=src/*.%))
113
ifeq ($(inputFiles),)
114
dbExport := $(firstword $(wildcard src/db.*.sql))
115
endif
116
117 1289 aaronmk
srcMap := maps/src.%.csv
118 1139 aaronmk
fullViaMap := maps/%.full.csv
119 1289 aaronmk
directMap := maps/VegBIEN.%.csv
120
allViaMaps := $(filter-out $(srcMap) $(fullViaMap) $(directMap),\
121 1139 aaronmk
$(wildcard maps/*.csv))
122 728 aaronmk
via := $(firstword $(sort $(basename $(basename $(notdir $(allViaMaps))))))
123 783 aaronmk
124 775 aaronmk
coreMap := $(mappings)/$(via)-VegBIEN.%.csv
125 1524 aaronmk
coreSelfMap := $(mappings)/$(via).self.%.csv
126 783 aaronmk
127 749 aaronmk
viaMaps := $(wildcard $(tablesSort:%=maps/$(via).%.csv))
128 1139 aaronmk
viaMaps += $(filter-out $(viaMaps) $(fullViaMap),$(wildcard maps/$(via).*.csv))
129 783 aaronmk
130 728 aaronmk
autogenMaps := $(subst $(via).,VegBIEN.,$(viaMaps))
131
directMaps := $(autogenMaps) $(filter-out $(autogenMaps),\
132
$(wildcard maps/VegBIEN.*.csv))
133 772 aaronmk
tables := $(directMaps:maps/VegBIEN.%.csv=%)
134 728 aaronmk
135 1969 aaronmk
##### Sources
136
137
srcs = $(shell $(bin)/sort_filenames $(wildcard $(exts:%=src/*.$*.%)))
138
isCsv = $(if $(srcs),$(if $(filter %.xml,$(srcs)),,1))
139
    # true if $(srcs) non-empty and contains no *.xml
140
catSrcs = $(bin)/cat$(if $(isCsv),_csv) $(srcs)
141
withCatSrcs = $(catSrcs:$(bin)/%=$(bin)/with_%) --
142
143 1986 aaronmk
# Usage: `make {--silent|-s} inputs/<datasrc>/cat` (don't echo make commands)
144 1969 aaronmk
cat: $(addprefix cat-,$(tables)) _always ;
145
146
cat-%: _always
147
	$(catSrcs)
148
149 1921 aaronmk
##### Installation
150
151
install: _always src/install import/install ;
152
153
uninstall: _always import/uninstall src/uninstall ;
154
155
reinstall: _always uninstall install ;
156
157
##### Staging tables
158
159
import/install: _always import/schema $(addprefix import/install-,$(tables)) ;
160
161
import/uninstall: _always import/rm_schema ;
162
# rm_schema will also drop all staging tables
163
164
import/schema: _always
165
	-echo 'CREATE SCHEMA "$(datasrc)";'|$(psqlAsBien)
166
# ignore errors if schema exists
167
168
import/rm_schema: _always
169
	echo 'DROP SCHEMA IF EXISTS "$(datasrc)" CASCADE;'|$(psqlAsBien)
170
171
import/install-%: _always
172 1955 aaronmk
	$(if $(isCsv),$(import_install_))
173 3476 aaronmk
import_install_ = (prefix=; . $(bin)/vegbien_dest; "time" nice -n +5\
174
env schema=$(datasrc) table=$* $(bin)/csv2db $(catSrcs)\
175 3317 aaronmk
$(if $(log),$(if $(quiet),&>,2>&1|tee )import/install-$*.log.sql))
176 1921 aaronmk
177 1777 aaronmk
##### Maps building
178
179 1789 aaronmk
# Maps to (try to) build are added to one of these
180 1779 aaronmk
maps :=
181 1789 aaronmk
createOnlyMaps :=
182 1779 aaronmk
183 1775 aaronmk
srcMaps := $(wildcard maps/src.*.csv)
184 1289 aaronmk
185 1931 aaronmk
mkSrcMap = $(catSrcs)|env datasrc=$(datasrc) $(bin)/src_map >$@
186 1779 aaronmk
187
# Autogen src maps with known table names
188
maps/src.%.csv: # no prereqs so only built if doesn't exist
189 1955 aaronmk
	$(if $(isCsv),$(mkSrcMap))
190 1779 aaronmk
# only build if CSV srcs exist for that table name
191
# Try all table names
192 1789 aaronmk
createOnlyMaps += $(tablesSort:%=maps/src.%.csv)
193 1779 aaronmk
194 1273 aaronmk
# Must come before $(root)/% to override it
195 1524 aaronmk
$(coreSelfMap): _always
196 1273 aaronmk
	-+$(subMake)
197 1524 aaronmk
# ignore errors if $(coreSelfMap) does not exist
198 1273 aaronmk
199 1509 aaronmk
# Via maps cleanup
200
ifneq ($(filter maps/.%.last_cleanup,$(MAKECMDGOALS)),)
201 1787 aaronmk
maps/.$(via).%.csv.last_cleanup: maps/$(via).%.csv $(coreMap)
202 2022 aaronmk
	$(bin)/in_place $< env ignore=1 $(bin)/subtract $(word 2,$+) 0 1 2
203 1509 aaronmk
	touch $@
204 2022 aaronmk
# Include comment column so commented mappings are never removed
205 1509 aaronmk
else
206
$(viaMaps): _always
207
	$(selfMake) $(@:maps/%=maps/.%.last_cleanup)
208
endif
209 1273 aaronmk
210 1526 aaronmk
joinSrcMap = $(if $(wildcard maps/src.$*.csv),$(bin)/in_place $@\
211
$(bin)/intersect maps/src.$*.csv 0)
212
213 1273 aaronmk
makeFullCsv = $(if $(shell test -e $(word 2,$+) && echo t),\
214 1817 aaronmk
env ignore=1 $(bin)/union <$+|,<$< )$(bin)/sort_map >$@
215 1524 aaronmk
# can't use $(wildcard) because it won't recheck file after $(coreSelfMap) runs
216 1273 aaronmk
217 1524 aaronmk
maps/$(via).%.full.csv: maps/$(via).%.csv $(coreSelfMap)
218 1273 aaronmk
	$(makeFullCsv)
219 1526 aaronmk
	$(joinSrcMap)
220 1139 aaronmk
maps += $(patsubst maps/%.csv,maps/%.full.csv,$(viaMaps))
221
222 1530 aaronmk
maps/VegBIEN.%.csv: maps/$(via).%.full.csv $(coreMap)
223
	$(bin)/join <$+|$(bin)/sort_map >$@
224
maps += $(autogenMaps)
225
226 1789 aaronmk
maps: $(createOnlyMaps) $(maps) _always ;
227 1139 aaronmk
228
all += $(maps)
229
230 1742 aaronmk
##### Maps validation
231
232
missing_mappings: _always missing_join_mappings missing_input_mappings ;
233
234
missing_%_mappings: _always # stem is one of join|input
235
	@echo $(emph)"Missing $* mappings:"$(endEmph)
236
	@+$(selfMake) remake 2>&1\
237 1800 aaronmk
|$(SED) -n 's/^.*No.* $* mapping for ([0-9A-Za-z_-]+).*$$/\1/p'\
238 1742 aaronmk
$(if $(filter join,$*),|$(bin)/ucase_first 0)|sort|uniq
239
240 1509 aaronmk
##### External dependencies
241
242
$(root)/%: _always
243
	+$(subMake)
244 1628 aaronmk
.PRECIOUS: $(root)/% # let ext. dir's Makefile decide whether to delete on error
245 1509 aaronmk
246 783 aaronmk
##### Mapping
247 368 aaronmk
248 775 aaronmk
+maps = $(filter maps/% $(mappings)/%,$(+_))
249 1451 aaronmk
<in = $(firstword $(filter-out $(+maps),$(+_)))
250
map = $(if $(<in),<$(<in),\
251 1777 aaronmk
$(if $(srcs),$(withCatSrcs),\
252 1451 aaronmk
$(if $(mapEnv),env $(mapEnv),\
253 3387 aaronmk
$(error No input file src/*.$*.{$(exts)}))))\
254 1451 aaronmk
$(root)/map $(+maps)
255 3387 aaronmk
map2db = $(if $(use_staged),env in_database=vegbien in_schema=$(datasrc)\
256
in_table=$*) env out_database=vegbien $(map)
257 368 aaronmk
258 783 aaronmk
##### Import to VegBIEN
259 772 aaronmk
260 1378 aaronmk
ifneq ($(dbExport)$(inputFiles),)
261 772 aaronmk
262 1853 aaronmk
profileTest = $(if $(profile),$(if $(test),1))
263
profileOnly = -env profile_to=/dev/fd/3 $(map2db) 3>&1 1>&2|\
264
$(bin)/profile_stats /dev/fd/0
265
266 3317 aaronmk
log_ = import/$*$(if $(n),.n=$(n),).$(date).log.sql
267
trace = $(log_:.log.sql=.trace)
268 1853 aaronmk
import = -$(if $(profileTest),$(profileOnly),(set -x; "time" env commit=1\
269 1190 aaronmk
$(if $(profile),profile_to=$(trace)) $(map2db)) $(if $(log),\
270 3382 aaronmk
$(if $(n),,&>>$(log_)))$(if $(log),$(if $(n), 2>&1|tee -a $(log_))))
271 1088 aaronmk
# don't abort on import errors, which often relate to invalid input data
272 624 aaronmk
273 1193 aaronmk
import: $(addprefix import-,$(tables)) _always ;
274 718 aaronmk
275
import-%: maps/VegBIEN.%.csv _always
276
	$(import)
277 772 aaronmk
# default:
278
import-%: _always ;
279 718 aaronmk
280 772 aaronmk
else
281
import: _always ;
282
endif
283
284 790 aaronmk
##### Log files from import
285
286 3317 aaronmk
logs := $(wildcard import/*.log.sql import/*.trace)
287 339 aaronmk
288 383 aaronmk
rm_logs: _always
289 339 aaronmk
	$(RM) $(logs)
290
291 783 aaronmk
##### Verification of import
292 264 aaronmk
293 1194 aaronmk
verify: $(addprefix verify-,$(tables)) _always ;
294
295 1199 aaronmk
verify-%: verify/%.ref verify/%.out _always
296 1184 aaronmk
	-$(diffVerbose) $(+_)
297 1088 aaronmk
# don't abort on verification errors, which are expected during development
298 1194 aaronmk
# default:
299 1199 aaronmk
verify-%: verify/%.out _always
300 1200 aaronmk
	$(if $(shell test -e $< && echo t),cat $<)
301 1199 aaronmk
# don't run if verify/%.out's default do-nothing action was used
302 1200 aaronmk
# can't use $(wildcard) because it won't recheck file after verify/%.out is run
303 368 aaronmk
304 1620 aaronmk
verify = $(if $(reverify),"time" $(psqlAsBien) --set=datasource="'$(datasrc)'"\
305
--no-align --field-separator=$$'\t' --pset=footer=off --pset=null=NULL <$< >$@)
306 514 aaronmk
307 1199 aaronmk
verify/%.out: $(mappings)/verify.%.sql _always
308 1192 aaronmk
	$(verify)
309 1194 aaronmk
# default:
310 1199 aaronmk
verify/%.out: _always ;
311 369 aaronmk
312 1199 aaronmk
all += $(wildcard verify/*.out)
313 1192 aaronmk
314 1378 aaronmk
ifneq ($(dbExport),)
315 1082 aaronmk
%.ref: %.ref.sql
316
	$(dbAsBien) $(db) <$< >$@
317
endif
318
319 1667 aaronmk
##### Editing import
320
321
import/rotate: _always
322 1668 aaronmk
	echo "UPDATE party SET organizationname = organizationname||'.$(date)'\
323 1854 aaronmk
WHERE organizationname = '$(datasrc)';"|$(psqlAsBien)
324 1667 aaronmk
325
import/rm: _always
326 1854 aaronmk
	echo "DELETE FROM party WHERE organizationname = '$(datasrc)';"|\
327
$(psqlAsBien)
328 1667 aaronmk
329 783 aaronmk
##### Testing
330 368 aaronmk
331 1986 aaronmk
testRefOutput = $(subst .2-step,,$(subst .staging,,$(1)))
332
testRef = $(testRefOutput).ref
333
hasOwnRef = $(filter $@,$(call testRefOutput,$@))
334
# filter returns non-empty if they are equal
335 630 aaronmk
336 897 aaronmk
define runTest
337 776 aaronmk
@echo "Testing $(abspath $@)..."
338 991 aaronmk
>$@ env test=1 n=$(test_n) $(1)
339 1184 aaronmk
@(set -x; $(diff) $(call testRef,$@) $@) 2>&1 || { e=$$?;\
340 780 aaronmk
$(if $(wildcard $(call testRef,$@)),,cat $@;)\
341 777 aaronmk
$(if $(hasOwnRef),\
342 3179 aaronmk
{\
343
read -p $(emph)'Accept new test output? (y/n)'$(endEmph) REPLY;\
344
if test "$$REPLY" = y; then\
345
(set -x; $(MAKE) $@-ok --directory=$(realpath .) --makefile=../input.Makefile);\
346
exit 0;\
347
fi;\
348
};,\
349 777 aaronmk
echo $(emph)"Note: The preceding failed test is compared to another test's\
350
output"$(endEmph);\
351
echo $(emph)"When it fails, this always indicates a bug"$(endEmph);\
352
)\
353 652 aaronmk
exit $$e;}
354 628 aaronmk
endef
355
356 897 aaronmk
test2File = $(call runTest,$(map))
357 1986 aaronmk
test2Db = $(call runTest,$(map2db))
358 775 aaronmk
359 876 aaronmk
tests :=
360
361 1139 aaronmk
test/$(via).%.xml: maps/$(via).%.full.csv _always
362 775 aaronmk
	$(test2File)
363 876 aaronmk
tests += test/$(via).%.xml
364 629 aaronmk
365 876 aaronmk
test/VegBIEN.%.xml: maps/VegBIEN.%.csv _always
366 775 aaronmk
	$(test2File)
367 876 aaronmk
tests += test/VegBIEN.%.xml
368 628 aaronmk
369 884 aaronmk
test/VegBIEN.%.2-step.xml: test/$(via).%.xml $(coreMap) _always
370 775 aaronmk
	-$(test2File)
371 722 aaronmk
# Don't abort tester if only 2-step test fails, as it's often finicky
372 876 aaronmk
tests += test/VegBIEN.%.2-step.xml
373 630 aaronmk
374 876 aaronmk
test/import.%.out: maps/VegBIEN.%.csv _always
375 1986 aaronmk
	$(test2Db)
376 876 aaronmk
tests += test/import.%.out
377 627 aaronmk
378 1987 aaronmk
testStaged2Db = $(foreach use_staged,1,$(test2Db))
379
    # run with use_staged=1
380 1986 aaronmk
381
# Requires staging tables. To create them, run `make inputs/<datasrc>/install`.
382
test/import.%.staging.out: maps/VegBIEN.%.csv _always
383 1988 aaronmk
	-$(if $(isCsv),$(testStaged2Db),cp -p $(@:.staging.out=.out) $@)
384
# Don't abort tester if only staging test fails, in case staging table missing
385
# Non-flat-file inputs fall back to mimicking a successful test
386 1986 aaronmk
tests += test/import.%.staging.out
387
388 876 aaronmk
testOutputs := $(foreach test,$(tests),$(tables:%=$(test)))
389 724 aaronmk
390 1361 aaronmk
.PRECIOUS: $(testOutputs) # save outputs of failed tests so they can be accepted
391
392 876 aaronmk
test: _always $(testOutputs) ;
393
394 1987 aaronmk
all += $(filter-out %.ref,$(wildcard test/*))
395 876 aaronmk
396 634 aaronmk
# Accepts a test output: make <test_output_path>-ok
397 626 aaronmk
%-ok: _always
398 630 aaronmk
	$(CP) $* $(call testRef,$*)
399 502 aaronmk
400 3133 aaronmk
##### Documentation
401
402 3385 aaronmk
steps = $(selfMake) import test=1 by_col=1 verbosity=2 n=100\
403
2>&1|$(bin)/debug2redmine >$@
404 3188 aaronmk
405 3133 aaronmk
import/steps.by_col.sql: _always
406 3188 aaronmk
	+$(steps)
407 3133 aaronmk
408 783 aaronmk
##### Input-type-specific
409 368 aaronmk
410 1917 aaronmk
# Each input type needs var $(mapEnv) and targets src/install, src/uninstall
411 624 aaronmk
412 1379 aaronmk
#### DB export
413
414 1378 aaronmk
ifneq ($(dbExport),)
415 256 aaronmk
416 1378 aaronmk
dbEngineExt := $(subst .,,$(suffix $(basename $(notdir $(dbExport)))))
417 1196 aaronmk
db := $(datasrc)
418 622 aaronmk
419 1379 aaronmk
### Installation
420 256 aaronmk
421 1917 aaronmk
src/install: _always db ;
422 339 aaronmk
423 1917 aaronmk
src/uninstall: _always rm_db ;
424 368 aaronmk
425 1379 aaronmk
### DB-engine-specific
426 621 aaronmk
427 624 aaronmk
# Each DB engine needs vars $(dbEngine), $(dbAsBien) and targets db, rm_db
428 256 aaronmk
429 622 aaronmk
ifeq ($(dbEngineExt),my)
430 621 aaronmk
431 622 aaronmk
dbEngine := MySQL
432
433 621 aaronmk
bienPassword := $(shell cat $(root)/config/bien_password)
434
mysqlAs = mysql --user=$(1) --password='$(bienPassword)'
435
mysqlAsRoot := $(call mysqlAs,root)
436
dbAsBien := $(call mysqlAs,bien)
437
438 1097 aaronmk
dbExists = $(shell echo "SHOW DATABASES LIKE '$(db)';"|$(mysqlAsRoot))
439
440
define createDb
441 1098 aaronmk
echo "CREATE DATABASE $(db) DEFAULT CHARACTER SET latin1;"|$(mysqlAsRoot)
442 1097 aaronmk
-$(mysqlAsRoot) --database=$(db) <$<
443
endef
444
# ignore errors in db import so that GRANT will still be run
445
446 1378 aaronmk
db: $(dbExport) _always
447 1097 aaronmk
	$(if $(dbExists),,$(createDb))
448 368 aaronmk
	echo "GRANT SELECT ON $(db).* TO 'bien'@'localhost';"|$(mysqlAsRoot)
449 250 aaronmk
450 383 aaronmk
rm_db: _always
451 368 aaronmk
	-echo "REVOKE ALL ON $(db).* FROM 'bien'@'localhost';"|$(mysqlAsRoot)
452
	echo "DROP DATABASE IF EXISTS $(db);"|$(mysqlAsRoot)
453 365 aaronmk
# ignore errors if grant not defined
454 250 aaronmk
455 1379 aaronmk
## Unrecognized DB engine
456 368 aaronmk
457
else
458 1378 aaronmk
$(error The DB filename $(dbExport) must be db.my.sql)
459 368 aaronmk
endif
460
461 1379 aaronmk
### Other input types
462 622 aaronmk
463 1379 aaronmk
else
464 624 aaronmk
465 1917 aaronmk
src/install: _always ;
466
src/uninstall: _always ;
467 1379 aaronmk
468 256 aaronmk
endif
469 1379 aaronmk
470
#### DB connection info
471
472
ifneq ($(dbEngine),)
473
# Must come after dbEngine is set
474
mapEnv := in_engine=$(dbEngine) in_database=$(db)
475
endif