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
srcMaps := $(wildcard */src.csv)
130

    
131
srcMap := %/src.csv
132
viaMap := %/map.csv
133
fullViaMap := %/map.full.csv
134
directMap := %/VegBIEN.csv
135
anyMap := $(srcMap) $(viaMap) $(fullViaMap) $(directMap)
136

    
137
extsFilter := $(addprefix %.,$(exts))
138
dataOnly = $(filter $(extsFilter),$(1))
139

    
140
anyTest = $*/test.%
141
srcsOnly = $(filter-out $(anyMap) $(anyTest) %/logs,$(call dataOnly,$(1)))
142

    
143
via := Veg+
144

    
145
coreMap := $(mappings)/$(via)-VegBIEN.csv
146
coreSelfMap := $(mappings)/$(via).self.csv
147

    
148
viaMaps := $(wildcard $(tables:%=%/map.csv))
149
viaMaps += $(filter-out $(viaMaps),$(srcMaps:%/src.csv=%/map.csv))
150
viaMaps += $(filter-out $(viaMaps) $(fullViaMap),$(wildcard */map.csv))
151

    
152
autogenMaps := $(subst map.,VegBIEN.,$(viaMaps))
153
directMaps := $(autogenMaps) $(filter-out $(autogenMaps),\
154
$(wildcard */VegBIEN.csv))
155

    
156
##### Sources
157

    
158
srcs = $(call sortFilenames,$(call srcsOnly,$(wildcard $*/*)))
159
nonHeaderSrcs = $(filter-out %/header.csv,$(srcs))
160
isRef = $(if $(nonHeaderSrcs),,1)
161
    # empty subdir, so references an already-installed staging table
162
isXml = $(filter %.xml,$(nonHeaderSrcs))
163
nonXml = $(if $(isXml),,1)
164
isCsv = $(if $(nonHeaderSrcs),$(if $(isXml),,1))
165
    # true if $(srcs) non-empty and contains no *.xml
166
catSrcs = $(bin)/cat$(if $(nonXml),_csv) $(srcs)
167
withCatSrcs = $(catSrcs:$(bin)/%=$(bin)/with_%) --
168

    
169
# Usage: `make {--silent|-s} inputs/<datasrc>/cat` (don't echo make commands)
170
cat: $(tables:%=%/cat) _always ;
171

    
172
%/cat: _always
173
	$(catSrcs)
174

    
175
##### Staging tables installation
176

    
177
dbExports := $(wildcard *.schema.sql)# schemas first
178
dbExports += $(filter-out $(dbExports),$(wildcard *.sql))# all others
179
dbExports := $(strip $(dbExports))# += adds extra whitespace
180
allInstalls := $(if $(dbExports),sql) $(allTables)
181

    
182
install: _always schema $(allInstalls:%=%/install) ;
183

    
184
uninstall: _always rm_schema ;
185
# rm_schema will also drop all staging tables
186

    
187
reinstall: _always uninstall install ;
188

    
189
schema: _always
190
	-echo 'CREATE SCHEMA "$(datasrc)";'|$(psqlAsBien)
191
# ignore errors if schema exists
192

    
193
rm_schema: _always
194
	echo 'DROP SCHEMA IF EXISTS "$(datasrc)" CASCADE;'|$(psqlAsBien)
195

    
196
logInstall = $(if $(log),$(if $(quiet),&>,2>&1|tee )$(1)logs/install.log.sql)
197

    
198
# Must come before %/install to override it
199
sql/install: $(dbExports)
200
	($(inDatasrc); cat $+|grep -vF 'SET search_path')|"time" $(psqlAsBien)\
201
$(logInstall)
202

    
203
# The export must be created with:
204
# `--compatible=postgresql --add-locks=false --set-charset --no-create-info`
205
# Must come before `%.sql: _MySQL/%.sql` to override it
206
%.data.sql: _MySQL/%.data.sql
207
	$(bin)/my2pg.data <$< >$@
208

    
209
# The export must be created with:
210
# `--compatible=postgresql --add-locks=false --set-charset`
211
# Add `--no-data` to create a schema-only export.
212
%.sql: _MySQL/%.sql
213
	$(bin)/my2pg <$< >$@
214

    
215
cleanup = (prefix=; . $(bin)/vegbien_dest; env schema=$(datasrc) table=$*\
216
$(bin)/csv2db)
217

    
218
define exportHeader
219
$(cleanup)
220
echo 'SELECT * FROM "$(datasrc)"."$*" LIMIT 0;'|$(psqlAsBien) \
221
--no-align --field-separator=, --pset=footer=off >$*/header.csv
222
endef
223

    
224
# For staging tables which are derived by joining together other staging tables.
225
%/install %/header.csv: %/create.sql _always
226
	($(inDatasrc); echo 'CREATE TABLE "$*" AS'; cat $<; echo ';'; \
227
echo 'ALTER TABLE "$*" ADD COLUMN row_num serial NOT NULL PRIMARY KEY;'; \
228
echo 'VACUUM ANALYZE "$*";')|$(psqlAsBien) --echo-all
229
	$(exportHeader)
230

    
231
# 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
%/install: _always
237
	$(if $(isRef),$(exportHeader),$(if $(nonXml),$(import_install_)))
238
import_install_ = (prefix=; . $(bin)/vegbien_dest; "time" nice -n +5\
239
env schema=$(datasrc) table=$* has_row_num=$(hasRowNum) $(bin)/csv2db\
240
$(catSrcs) $(call logInstall,$*/))
241

    
242
##### Maps building
243

    
244
# 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
# Maps to (try to) build are added to one of these
248
maps :=
249
createOnlyMaps :=
250

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

    
254
# Src maps cleanup
255
ifneq ($(filter %/.src.csv.last_cleanup,$(MAKECMDGOALS)),)
256
%/.src.csv.last_cleanup: %/src.csv
257
	$(bin)/in_place $< $(bin)/cols +
258
	touch $@
259
else
260
# Autogen src maps with known table names
261
%/src.csv: _always
262
	$(if $(wildcard $@),,$(if $(nonXml),$(mkSrcMap)))
263
	+$(if $(nonXml),$(selfMake) $(@:%/src.csv=%/.src.csv.last_cleanup))
264
# only build if doesn't exist
265
# only build if CSV srcs exist for that table name
266
endif
267

    
268
createOnlyMaps += $(tables:%=%/src.csv)
269

    
270
# Must come before $(root)/% to override it
271
$(coreSelfMap): _always
272
	-+$(subMake)
273
# ignore errors if $(coreSelfMap) does not exist
274

    
275
# Via maps cleanup
276
ifneq ($(filter %/.map.csv.last_cleanup,$(MAKECMDGOALS)),)
277
%/.map.csv.last_cleanup: %/map.csv $(coreMap)
278
	$(bin)/in_place $< env ignore=1 $(bin)/subtract $(word 2,$+) 0 1 2
279
	touch $@
280
# Include comment column so commented mappings are never removed
281
else
282
%/map.csv: %/src.csv _always
283
	$(if $(wildcard $@),,$(if $(wildcard $<),$(CP) $< $@))
284
	+$(selfMake) $(@:%/map.csv=%/.map.csv.last_cleanup)
285
endif
286

    
287
srcMap* = $*/src.csv
288
joinSrcMap = $(if $(hasSrc),$(bin)/in_place $@ $(bin)/intersect $(srcMap*) 0)
289
hasSrc = $(shell test -s $(word 3,$+) && echo t)
290

    
291
makeFullCsv = env ignore=1 $(bin)/union <$(wordlist 1,2,$+)|$(bin)/sort_map >$@
292

    
293
%/map.full.csv: %/map.csv $(coreSelfMap)
294
	$(makeFullCsv)
295
	$(joinSrcMap)
296
maps += $(patsubst %/map.csv,%/map.full.csv,$(viaMaps))
297

    
298
%/VegBIEN.csv: %/map.full.csv $(coreMap)
299
	$(bin)/join <$+|$(bin)/sort_map >$@
300
maps += $(autogenMaps)
301

    
302
maps: $(createOnlyMaps) $(maps) _always ;
303

    
304
all += $(maps)
305

    
306
##### Maps validation
307

    
308
# `env -i sort`: use system's sort, not bin/sort
309
missingMappingsCmd = +$(selfMake) remake 2>&1\
310
|$(SED) -n 's/^.*No $(*2Space) mapping for (.*)$$/\1/p'|$(SED) 's/\/_.*//'\
311
$(if $(filter non-empty_join,$*),|$(bin)/ucase_first 0)|env -i sort|uniq
312

    
313
include $(root)/lib/mappings.Makefile
314

    
315
##### External dependencies
316

    
317
$(root)/%: _always
318
	+$(subMake)
319
.PRECIOUS: $(root)/% # let ext. dir's Makefile decide whether to delete on error
320

    
321
##### Mapping
322

    
323
+maps = $(filter %/map.csv %/VegBIEN.csv $(mappings)/%,$(+_))
324
map = $(if $(srcs),$(withCatSrcs) $(root)/map $(+maps),\
325
$(shell echo Warning: No input file $*/*.{$(exts)} >&2)false)
326
# need false to run some command, when prefixed by `env ...` below
327
map2db = $(if $(use_staged),env in_database=vegbien in_schema=$(datasrc)\
328
in_table=$*) env out_database=vegbien $(map)
329

    
330
##### Import to VegBIEN
331

    
332
profileTest = $(if $(profile),$(if $(test),1))
333
profileOnly = -env profile_to=/dev/fd/3 $(map2db) 3>&1 1>&2|\
334
$(bin)/profile_stats /dev/fd/0
335

    
336
log_ = $*/logs/$(if $(n),n=$(n).,)$(date).log.sql
337
trace = $(log_:.log.sql=.trace)
338
import = -$(if $(profileTest),$(profileOnly),(set -x; "time" env commit=1\
339
$(if $(profile),profile_to=$(trace)) $(map2db)) $(if $(log),\
340
$(if $(n),,&>$(log_)))$(if $(log),$(if $(n), 2>&1|tee -a $(log_))))
341
# don't abort on import errors, which often relate to invalid input data
342

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

    
345
%/import: %/VegBIEN.csv _always
346
	$(import)
347
# default:
348
%/import: _always ;
349

    
350
##### Log files from import
351

    
352
logs := $(wildcard */logs/*.log.sql */logs/*.trace)
353

    
354
rm_logs: _always
355
	$(RM) $(logs)
356

    
357
##### Verification of import
358

    
359
verify: $(tables:%=%/verify) _always ;
360

    
361
%/verify: verify/%.ref verify/%.out _always
362
	-$(diffVerbose) $(+_)
363
# don't abort on verification errors, which are expected during development
364
# default:
365
%/verify: verify/%.out _always
366
	$(if $(shell test -e $< && echo t),cat $<)
367
# don't run if verify/%.out's default do-nothing action was used
368
# can't use $(wildcard) because it won't recheck file after verify/%.out is run
369

    
370
psqlExport := "time" $(psqlAsBien) --no-align --field-separator=$$'\t'\
371
--pset=footer=off --pset=null=NULL
372
verify = $(if $(reverify),$(psqlExport) --set=datasource="'$(datasrc)'" <$< >$@)
373

    
374
verify/%.out: $(mappings)/verify.%.sql _always
375
	$(verify)
376
# default:
377
verify/%.out: _always ;
378

    
379
all += $(wildcard verify/*.out)
380

    
381
%.ref: %.ref.sql
382
	($(inDatasrc); cat $<)|$(psqlExport) >$@
383

    
384
##### Editing import
385

    
386
rotate: _always
387
	echo "UPDATE party SET organizationname = organizationname||'.$(date)'\
388
WHERE organizationname = '$(datasrc)';"|$(psqlAsBien)
389

    
390
rm: _always
391
	echo "DELETE FROM party WHERE organizationname = '$(datasrc)';"|\
392
$(psqlAsBien)
393

    
394
##### Testing
395

    
396
testRefOutput = $(subst .by_col,,$(1))
397
testRef = $(testRefOutput).ref
398
hasOwnRef = $(filter $@,$(call testRefOutput,$@))
399
# filter returns non-empty if they are equal
400

    
401
# `rm $@`: Remove outputs of successful tests to reduce clutter
402
# `$(foreach use_staged...)`: Run with use_staged=1
403
define runTest
404
@echo "Testing $(abspath $@)..."
405
>$@ env test=1 n=$(test_n) $(1) $(foreach use_staged,1,$(map2db))
406
@(set -x; $(diffIgnoreSpace) $(call testRef,$@) $@) 2>&1 && rm $@ || { e=$$?;\
407
$(if $(wildcard $(call testRef,$@)),,cat $@;)\
408
$(if $(hasOwnRef),\
409
{\
410
read -p $(emph)'Accept new test output? (y/n)'$(endEmph) REPLY;\
411
if test "$$REPLY" = y; then\
412
(set -x; $(MAKE) $@-ok --directory=$(realpath .) --makefile=../input.Makefile);\
413
exit 0;\
414
fi;\
415
};,\
416
echo $(emph)"Note: The preceding failed test is compared to another test's\
417
output"$(endEmph);\
418
echo $(emph)"When it fails, this always indicates a bug"$(endEmph);\
419
)\
420
exit $$e;}
421
endef
422

    
423
tests :=
424

    
425
# Requires staging tables. To create them, run `make inputs/<datasrc>/install`.
426
# Non-flat-file inputs fall back to mimicking a successful test
427
%/test.xml: %/VegBIEN.csv _always
428
	$(if $(nonXml),$(call runTest,by_col=))
429
tests += %/test.xml
430

    
431
%/test.by_col.xml: %/VegBIEN.csv _always
432
	-$(if $(nonXml),$(call runTest,by_col=1))
433
# Don't abort tester if only by-column test fails
434

    
435
# Only run column-based tests if column-based mode enabled, because these tests
436
# are much slower than the row-based tests for small numbers of rows
437
ifneq ($(by_col),)
438
tests += %/test.by_col.xml
439
endif
440

    
441
testOutputs := $(foreach test,$(tests),$(tables:%=$(test)))
442

    
443
.PRECIOUS: $(testOutputs) # save outputs of failed tests so they can be accepted
444

    
445
test: _always $(testOutputs) ;
446

    
447
all += $(wildcard %/test*.xml)
448

    
449
# Accepts a test output: make <test_output_path>-ok
450
%-ok: _always
451
	mv $* $(call testRef,$*)
452

    
453
accept-all: _always
454
	+yes|$(selfMake) test
455

    
456
##### Documentation
457

    
458
steps = $(selfMake) -s $*/import test=1 by_col=1 verbosity=2 n=100\
459
2>&1|$(bin)/debug2redmine >$@
460

    
461
%/logs/steps.by_col.log.sql: _always
462
	+$(steps)
(4-4/4)