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
tablesSort ?= plots organisms stems specimens
17 1386 aaronmk
exts ?= csv tsv txt xml
18 783 aaronmk
test_n ?= 2
19
20
##### Vars/functions
21
22 1821 aaronmk
# Paths
23
datasrc := $(notdir $(realpath .))
24
bin := $(root)/bin
25
mappings := $(root)/mappings
26
27 368 aaronmk
# Make
28 640 aaronmk
SHELL := /bin/bash
29 1509 aaronmk
selfMake = $(MAKE) --makefile=../input.Makefile
30 1821 aaronmk
subMake = $(MAKE) $(@:$(root)/%=%) --directory=$(root)
31 404 aaronmk
+_ = $(+:_%=)
32 368 aaronmk
addBeforeExt = $(basename $(2))$(1)$(suffix $(2))
33
34 1742 aaronmk
# OS
35
os := $(shell uname)
36
forOs = $(patsubst %,%-$(filter Linux Darwin,$(os)),$(1))
37
38
# Formatting
39
SED = sed -$(if $(filter Darwin,$(os)),E,r)
40
41 790 aaronmk
# System
42
date = $(shell date +"%Y-%m-%d-%H-%M-%S")
43
44 640 aaronmk
# Terminal
45 1184 aaronmk
termCols := $(shell tput cols)
46 640 aaronmk
esc := '['
47
reset := $(esc)'0m'
48
emph := $(esc)'7m '
49
endEmph := ' '$(reset)
50
51 368 aaronmk
# Commands
52 1245 aaronmk
MKDIR = mkdir -p
53
mkdir = $(MKDIR) $(@D)
54 395 aaronmk
CP = cp -p
55 1184 aaronmk
diff = diff --unified=2
56 3721 aaronmk
diffIgnoreSpace = $(diff) --ignore-space-change
57 1184 aaronmk
diffVerbose = $(if $(verbose),diff --side-by-side --left-column\
58
--width=$(termCols),$(diff))
59 368 aaronmk
60 1918 aaronmk
# BIEN commands
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)
64 353 aaronmk
65 1594 aaronmk
# SVN
66
addDir = $(if $(wildcard $(1)/),svn add --depth=empty $(1),svn mkdir $(1))
67 1771 aaronmk
setSvnIgnore = svn propset svn:ignore $(2) $(1)
68 1594 aaronmk
define addDirWithIgnore
69
$(addDir)
70 1771 aaronmk
$(setSvnIgnore)
71 1594 aaronmk
endef
72
73 783 aaronmk
##### General targets
74 250 aaronmk
75 418 aaronmk
all: _always maps ;
76 247 aaronmk
77 383 aaronmk
clean: _always
78 256 aaronmk
	$(RM) $(all)
79
80 1742 aaronmk
remake: _always clean
81
	+$(selfMake)
82
# re-run make so that cache of existing files is reset
83
84 1627 aaronmk
make_script = ./$< >$@
85 368 aaronmk
86 1604 aaronmk
%/: % _always ;
87
88 1627 aaronmk
# Must come before `%: %.make` to override it
89
src/%: src/%.make _always
90 3317 aaronmk
	(set -x; $(make_script)) 2>>$<.log.sql
91 1627 aaronmk
.PRECIOUS: src/% # save partial outputs of aborted src make scripts
92
93
%: %.make _always
94
	$(make_script)
95
96 783 aaronmk
##### SVN
97 250 aaronmk
98 1594 aaronmk
add: _always
99 1771 aaronmk
	$(call setSvnIgnore,.,$$'')
100 1594 aaronmk
	$(call addDirWithIgnore,src,$$'*')
101
	$(call addDirWithIgnore,maps,$$'.~*')
102
	$(call addDirWithIgnore,import,$$'*')
103
	$(call addDirWithIgnore,test,$$'*.out\n*.xml')
104
	$(call addDirWithIgnore,verify,$$'*.out')
105 766 aaronmk
106 1969 aaronmk
##### Existing maps discovery
107 1955 aaronmk
108
inputFiles := $(wildcard $(exts:%=src/*.%))
109
110 3574 aaronmk
srcMaps := $(wildcard maps/src.*.csv)
111
srcRoots := $(srcMaps:maps/src.%.csv=$(mappings)/roots/%.sh)
112
113 1289 aaronmk
srcMap := maps/src.%.csv
114 1139 aaronmk
fullViaMap := maps/%.full.csv
115 1289 aaronmk
directMap := maps/VegBIEN.%.csv
116
allViaMaps := $(filter-out $(srcMap) $(fullViaMap) $(directMap),\
117 1139 aaronmk
$(wildcard maps/*.csv))
118 728 aaronmk
via := $(firstword $(sort $(basename $(basename $(notdir $(allViaMaps))))))
119 783 aaronmk
120 3574 aaronmk
# Look up via in src maps' roots
121
ifeq ($(via),)
122
ifneq ($(srcRoots),)
123
via := $(firstword $(sort\
124
$(shell $(SED) -n 's/^export out_root="([0-9A-Za-z_-]+).*$$/\1/p' $(srcRoots))))
125
endif
126
endif
127
128 775 aaronmk
coreMap := $(mappings)/$(via)-VegBIEN.%.csv
129 1524 aaronmk
coreSelfMap := $(mappings)/$(via).self.%.csv
130 783 aaronmk
131 749 aaronmk
viaMaps := $(wildcard $(tablesSort:%=maps/$(via).%.csv))
132 3574 aaronmk
viaMaps += $(filter-out $(viaMaps),$(srcMaps:maps/src.%=maps/$(via).%))
133 1139 aaronmk
viaMaps += $(filter-out $(viaMaps) $(fullViaMap),$(wildcard maps/$(via).*.csv))
134 783 aaronmk
135 728 aaronmk
autogenMaps := $(subst $(via).,VegBIEN.,$(viaMaps))
136
directMaps := $(autogenMaps) $(filter-out $(autogenMaps),\
137
$(wildcard maps/VegBIEN.*.csv))
138 772 aaronmk
tables := $(directMaps:maps/VegBIEN.%.csv=%)
139 728 aaronmk
140 1969 aaronmk
##### Sources
141
142
srcs = $(shell $(bin)/sort_filenames $(wildcard $(exts:%=src/*.$*.%)))
143
isCsv = $(if $(srcs),$(if $(filter %.xml,$(srcs)),,1))
144
    # true if $(srcs) non-empty and contains no *.xml
145
catSrcs = $(bin)/cat$(if $(isCsv),_csv) $(srcs)
146
withCatSrcs = $(catSrcs:$(bin)/%=$(bin)/with_%) --
147
148 1986 aaronmk
# Usage: `make {--silent|-s} inputs/<datasrc>/cat` (don't echo make commands)
149 1969 aaronmk
cat: $(addprefix cat-,$(tables)) _always ;
150
151
cat-%: _always
152
	$(catSrcs)
153
154 1921 aaronmk
##### Installation
155
156 4094 aaronmk
install: _always import/install ;
157 1921 aaronmk
158 4094 aaronmk
uninstall: _always import/uninstall ;
159 1921 aaronmk
160
reinstall: _always uninstall install ;
161
162
##### Staging tables
163
164
import/install: _always import/schema $(addprefix import/install-,$(tables)) ;
165
166
import/uninstall: _always import/rm_schema ;
167
# rm_schema will also drop all staging tables
168
169
import/schema: _always
170
	-echo 'CREATE SCHEMA "$(datasrc)";'|$(psqlAsBien)
171
# ignore errors if schema exists
172
173
import/rm_schema: _always
174
	echo 'DROP SCHEMA IF EXISTS "$(datasrc)" CASCADE;'|$(psqlAsBien)
175
176
import/install-%: _always
177 1955 aaronmk
	$(if $(isCsv),$(import_install_))
178 3476 aaronmk
import_install_ = (prefix=; . $(bin)/vegbien_dest; "time" nice -n +5\
179
env schema=$(datasrc) table=$* $(bin)/csv2db $(catSrcs)\
180 3317 aaronmk
$(if $(log),$(if $(quiet),&>,2>&1|tee )import/install-$*.log.sql))
181 1921 aaronmk
182 1777 aaronmk
##### Maps building
183
184 1789 aaronmk
# Maps to (try to) build are added to one of these
185 1779 aaronmk
maps :=
186 1789 aaronmk
createOnlyMaps :=
187 1779 aaronmk
188 3569 aaronmk
srcRoot = $(mappings)/roots/$*.sh
189
mkSrcMap = $(catSrcs)|(. $(srcRoot); env datasrc=$(datasrc) $(bin)/src_map >$@)
190 1779 aaronmk
191 3567 aaronmk
# Src maps cleanup
192
ifneq ($(filter maps/.%.last_cleanup,$(MAKECMDGOALS)),)
193
maps/.src.%.csv.last_cleanup: maps/src.%.csv
194 3814 aaronmk
	$(bin)/in_place $< $(bin)/cols +
195 3567 aaronmk
	touch $@
196
else
197 1779 aaronmk
# Autogen src maps with known table names
198 3567 aaronmk
maps/src.%.csv: _always
199
	$(if $(wildcard $@),,$(if $(isCsv),$(mkSrcMap)))
200
	+$(if $(isCsv),$(selfMake) $(@:maps/%=maps/.%.last_cleanup))
201
# only build if doesn't exist
202 1779 aaronmk
# only build if CSV srcs exist for that table name
203 3567 aaronmk
endif
204
205 1779 aaronmk
# Try all table names
206 1789 aaronmk
createOnlyMaps += $(tablesSort:%=maps/src.%.csv)
207 1779 aaronmk
208 1273 aaronmk
# Must come before $(root)/% to override it
209 1524 aaronmk
$(coreSelfMap): _always
210 1273 aaronmk
	-+$(subMake)
211 1524 aaronmk
# ignore errors if $(coreSelfMap) does not exist
212 1273 aaronmk
213 3561 aaronmk
joinSrcMap = $(if $(hasSrc),$(bin)/in_place $@ $(bin)/intersect $(word 3,$+) 0)
214
hasSrc = $(shell test -s $(word 3,$+) && echo t)
215 1526 aaronmk
216 1273 aaronmk
makeFullCsv = $(if $(shell test -e $(word 2,$+) && echo t),\
217 3561 aaronmk
env ignore=1 $(bin)/union <$(wordlist 1,2,$+)|,<$< )$(bin)/sort_map >$@
218 1524 aaronmk
# can't use $(wildcard) because it won't recheck file after $(coreSelfMap) runs
219 1273 aaronmk
220 3572 aaronmk
# Must come before maps/$(via).%.csv to override it
221 3561 aaronmk
maps/$(via).%.full.csv: maps/$(via).%.csv $(coreSelfMap) $(srcMap)
222 1273 aaronmk
	$(makeFullCsv)
223 1526 aaronmk
	$(joinSrcMap)
224 1139 aaronmk
maps += $(patsubst maps/%.csv,maps/%.full.csv,$(viaMaps))
225
226 3572 aaronmk
# Via maps cleanup
227
ifneq ($(filter maps/.%.last_cleanup,$(MAKECMDGOALS)),)
228
maps/.$(via).%.csv.last_cleanup: maps/$(via).%.csv $(coreMap)
229
	$(bin)/in_place $< env ignore=1 $(bin)/subtract $(word 2,$+) 0 1 2
230
	touch $@
231
# Include comment column so commented mappings are never removed
232
else
233 3574 aaronmk
maps/$(via).%.csv: maps/src.%.csv _always
234
	$(if $(wildcard $@),,$(if $(wildcard $<),$(CP) $< $@))
235 3572 aaronmk
	+$(selfMake) $(@:maps/%=maps/.%.last_cleanup)
236
endif
237
238 1530 aaronmk
maps/VegBIEN.%.csv: maps/$(via).%.full.csv $(coreMap)
239
	$(bin)/join <$+|$(bin)/sort_map >$@
240
maps += $(autogenMaps)
241
242 1789 aaronmk
maps: $(createOnlyMaps) $(maps) _always ;
243 1139 aaronmk
244
all += $(maps)
245
246 1742 aaronmk
##### Maps validation
247
248 3764 aaronmk
missingMappingsCmd = +$(selfMake) remake 2>&1\
249 3821 aaronmk
|$(SED) -n 's/^.*No $(*2Space) mapping for (.*)$$/\1/p'|$(SED) 's/\/_.*//'\
250 3764 aaronmk
$(if $(filter non-empty_join,$*),|$(bin)/ucase_first 0)|sort|uniq
251 1742 aaronmk
252 3764 aaronmk
include $(root)/lib/mappings.Makefile
253 1742 aaronmk
254 1509 aaronmk
##### External dependencies
255
256
$(root)/%: _always
257
	+$(subMake)
258 1628 aaronmk
.PRECIOUS: $(root)/% # let ext. dir's Makefile decide whether to delete on error
259 1509 aaronmk
260 783 aaronmk
##### Mapping
261 368 aaronmk
262 775 aaronmk
+maps = $(filter maps/% $(mappings)/%,$(+_))
263 1451 aaronmk
<in = $(firstword $(filter-out $(+maps),$(+_)))
264
map = $(if $(<in),<$(<in),\
265 1777 aaronmk
$(if $(srcs),$(withCatSrcs),\
266 1451 aaronmk
$(if $(mapEnv),env $(mapEnv),\
267 3387 aaronmk
$(error No input file src/*.$*.{$(exts)}))))\
268 1451 aaronmk
$(root)/map $(+maps)
269 3387 aaronmk
map2db = $(if $(use_staged),env in_database=vegbien in_schema=$(datasrc)\
270
in_table=$*) env out_database=vegbien $(map)
271 368 aaronmk
272 783 aaronmk
##### Import to VegBIEN
273 772 aaronmk
274 4094 aaronmk
ifneq ($(inputFiles),)
275 772 aaronmk
276 1853 aaronmk
profileTest = $(if $(profile),$(if $(test),1))
277
profileOnly = -env profile_to=/dev/fd/3 $(map2db) 3>&1 1>&2|\
278
$(bin)/profile_stats /dev/fd/0
279
280 3317 aaronmk
log_ = import/$*$(if $(n),.n=$(n),).$(date).log.sql
281
trace = $(log_:.log.sql=.trace)
282 1853 aaronmk
import = -$(if $(profileTest),$(profileOnly),(set -x; "time" env commit=1\
283 1190 aaronmk
$(if $(profile),profile_to=$(trace)) $(map2db)) $(if $(log),\
284 3616 aaronmk
$(if $(n),,&>$(log_)))$(if $(log),$(if $(n), 2>&1|tee -a $(log_))))
285 1088 aaronmk
# don't abort on import errors, which often relate to invalid input data
286 624 aaronmk
287 1193 aaronmk
import: $(addprefix import-,$(tables)) _always ;
288 718 aaronmk
289
import-%: maps/VegBIEN.%.csv _always
290
	$(import)
291 772 aaronmk
# default:
292
import-%: _always ;
293 718 aaronmk
294 772 aaronmk
else
295
import: _always ;
296
endif
297
298 790 aaronmk
##### Log files from import
299
300 3317 aaronmk
logs := $(wildcard import/*.log.sql import/*.trace)
301 339 aaronmk
302 383 aaronmk
rm_logs: _always
303 339 aaronmk
	$(RM) $(logs)
304
305 783 aaronmk
##### Verification of import
306 264 aaronmk
307 1194 aaronmk
verify: $(addprefix verify-,$(tables)) _always ;
308
309 1199 aaronmk
verify-%: verify/%.ref verify/%.out _always
310 1184 aaronmk
	-$(diffVerbose) $(+_)
311 1088 aaronmk
# don't abort on verification errors, which are expected during development
312 1194 aaronmk
# default:
313 1199 aaronmk
verify-%: verify/%.out _always
314 1200 aaronmk
	$(if $(shell test -e $< && echo t),cat $<)
315 1199 aaronmk
# don't run if verify/%.out's default do-nothing action was used
316 1200 aaronmk
# can't use $(wildcard) because it won't recheck file after verify/%.out is run
317 368 aaronmk
318 1620 aaronmk
verify = $(if $(reverify),"time" $(psqlAsBien) --set=datasource="'$(datasrc)'"\
319
--no-align --field-separator=$$'\t' --pset=footer=off --pset=null=NULL <$< >$@)
320 514 aaronmk
321 1199 aaronmk
verify/%.out: $(mappings)/verify.%.sql _always
322 1192 aaronmk
	$(verify)
323 1194 aaronmk
# default:
324 1199 aaronmk
verify/%.out: _always ;
325 369 aaronmk
326 1199 aaronmk
all += $(wildcard verify/*.out)
327 1192 aaronmk
328 1667 aaronmk
##### Editing import
329
330
import/rotate: _always
331 1668 aaronmk
	echo "UPDATE party SET organizationname = organizationname||'.$(date)'\
332 1854 aaronmk
WHERE organizationname = '$(datasrc)';"|$(psqlAsBien)
333 1667 aaronmk
334
import/rm: _always
335 1854 aaronmk
	echo "DELETE FROM party WHERE organizationname = '$(datasrc)';"|\
336
$(psqlAsBien)
337 1667 aaronmk
338 783 aaronmk
##### Testing
339 368 aaronmk
340 1986 aaronmk
testRefOutput = $(subst .2-step,,$(subst .staging,,$(1)))
341
testRef = $(testRefOutput).ref
342
hasOwnRef = $(filter $@,$(call testRefOutput,$@))
343
# filter returns non-empty if they are equal
344 630 aaronmk
345 897 aaronmk
define runTest
346 776 aaronmk
@echo "Testing $(abspath $@)..."
347 991 aaronmk
>$@ env test=1 n=$(test_n) $(1)
348 3721 aaronmk
@(set -x; $(diffIgnoreSpace) $(call testRef,$@) $@) 2>&1 || { e=$$?;\
349 780 aaronmk
$(if $(wildcard $(call testRef,$@)),,cat $@;)\
350 777 aaronmk
$(if $(hasOwnRef),\
351 3179 aaronmk
{\
352
read -p $(emph)'Accept new test output? (y/n)'$(endEmph) REPLY;\
353
if test "$$REPLY" = y; then\
354
(set -x; $(MAKE) $@-ok --directory=$(realpath .) --makefile=../input.Makefile);\
355
exit 0;\
356
fi;\
357
};,\
358 777 aaronmk
echo $(emph)"Note: The preceding failed test is compared to another test's\
359
output"$(endEmph);\
360
echo $(emph)"When it fails, this always indicates a bug"$(endEmph);\
361
)\
362 652 aaronmk
exit $$e;}
363 628 aaronmk
endef
364
365 1986 aaronmk
test2Db = $(call runTest,$(map2db))
366 775 aaronmk
367 876 aaronmk
tests :=
368
369 3642 aaronmk
test/import.%.xml: maps/VegBIEN.%.csv _always
370 1986 aaronmk
	$(test2Db)
371 3642 aaronmk
tests += test/import.%.xml
372 627 aaronmk
373 1987 aaronmk
testStaged2Db = $(foreach use_staged,1,$(test2Db))
374
    # run with use_staged=1
375 1986 aaronmk
376
# Requires staging tables. To create them, run `make inputs/<datasrc>/install`.
377 3642 aaronmk
test/import.%.staging.xml: maps/VegBIEN.%.csv _always
378
	-$(if $(isCsv),$(testStaged2Db),cp -p $(@:.staging.xml=.xml) $@)
379 1988 aaronmk
# Don't abort tester if only staging test fails, in case staging table missing
380
# Non-flat-file inputs fall back to mimicking a successful test
381 3642 aaronmk
tests += test/import.%.staging.xml
382 1986 aaronmk
383 876 aaronmk
testOutputs := $(foreach test,$(tests),$(tables:%=$(test)))
384 724 aaronmk
385 1361 aaronmk
.PRECIOUS: $(testOutputs) # save outputs of failed tests so they can be accepted
386
387 876 aaronmk
test: _always $(testOutputs) ;
388
389 1987 aaronmk
all += $(filter-out %.ref,$(wildcard test/*))
390 876 aaronmk
391 634 aaronmk
# Accepts a test output: make <test_output_path>-ok
392 626 aaronmk
%-ok: _always
393 630 aaronmk
	$(CP) $* $(call testRef,$*)
394 502 aaronmk
395 3579 aaronmk
test/accept-all: _always
396
	+yes|$(selfMake) test
397
398 3133 aaronmk
##### Documentation
399
400 3590 aaronmk
steps = $(selfMake) -s import test=1 by_col=1 verbosity=2 n=100\
401 3385 aaronmk
2>&1|$(bin)/debug2redmine >$@
402 3188 aaronmk
403 3133 aaronmk
import/steps.by_col.sql: _always
404 3188 aaronmk
	+$(steps)