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

    
65
# SVN
66
addDir = $(if $(wildcard $(1)/),svn add --depth=empty $(1),svn mkdir $(1))
67
setSvnIgnore = svn propset svn:ignore $(2) $(1)
68
define addDirWithIgnore
69
$(addDir)
70
$(setSvnIgnore)
71
endef
72

    
73
##### General targets
74

    
75
all: _always maps ;
76

    
77
clean: _always
78
	$(RM) $(all)
79

    
80
remake: _always clean
81
	+$(selfMake)
82
# re-run make so that cache of existing files is reset
83

    
84
make_script = ./$< >$@
85

    
86
%/: % _always ;
87

    
88
# Must come before `%: %.make` to override it
89
%: %.make
90
	(set -x; "time" $(make_script)) 2>>$<.log
91
.PRECIOUS: % # save partial outputs of aborted src make scripts
92

    
93
##### SVN
94

    
95
add: _always
96
	$(call setSvnIgnore,.,'*')
97
	$(call addDirWithIgnore,verify,'*.out')
98
	$(call addFile,import_order.txt)
99

    
100
# Adds a new table subdir
101
%/add: _always
102
	$(call addDirWithIgnore,$*,'*')
103
	$(call addDirWithIgnore,$*/logs,$$'*.log.sql\n*.trace')
104

    
105
##### Existing maps discovery
106

    
107
sortFile := import_order.txt
108

    
109
tables := $(if $(wildcard $(sortFile)),$(shell cat $(sortFile)))
110
    # $(shell) replaces "\n" with " "
111
allSubdirs := $(call wildcard/,*/)
112
allTables := $(call sortFilenames,$(filter-out _% verify,$(allSubdirs:%/=%)))
113
joinedTables := $(filter-out $(tables),$(allTables))
114
allTables := $(joinedTables) $(tables)# move joined tables to beginning
115
ifeq ($(tables),)# none specified in sort file
116
tables := $(allTables)
117
endif
118

    
119
srcMaps := $(wildcard */src.csv)
120

    
121
srcMap := %/src.csv
122
viaMap := %/map.csv
123
fullViaMap := %/map.full.csv
124
directMap := %/VegBIEN.csv
125
anyMap := $(srcMap) $(viaMap) $(fullViaMap) $(directMap)
126

    
127
extsFilter := $(addprefix %.,$(exts))
128
dataOnly = $(filter $(extsFilter),$(1))
129

    
130
anyTest = $*/test.%
131
srcsOnly = $(filter-out $(anyMap) $(anyTest) %/logs,$(call dataOnly,$(1)))
132

    
133
via := Veg+
134

    
135
coreMap := $(mappings)/$(via)-VegBIEN.csv
136
coreSelfMap := $(mappings)/$(via).self.csv
137

    
138
viaMaps := $(wildcard $(tables:%=%/map.csv))
139
viaMaps += $(filter-out $(viaMaps),$(srcMaps:%/src.csv=%/map.csv))
140
viaMaps += $(filter-out $(viaMaps) $(fullViaMap),$(wildcard */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
isCsv = $(if $(srcs),$(if $(filter %.xml,$(srcs)),,1))
150
    # true if $(srcs) non-empty and contains no *.xml
151
catSrcs = $(bin)/cat$(if $(isCsv),_csv) $(srcs)
152
withCatSrcs = $(catSrcs:$(bin)/%=$(bin)/with_%) --
153

    
154
# Usage: `make {--silent|-s} inputs/<datasrc>/cat` (don't echo make commands)
155
cat: $(tables:%=%/cat) _always ;
156

    
157
%/cat: _always
158
	$(catSrcs)
159

    
160
##### Staging tables installation
161

    
162
install: _always schema $(allTables:%=%/install) ;
163

    
164
uninstall: _always rm_schema ;
165
# rm_schema will also drop all staging tables
166

    
167
reinstall: _always uninstall install ;
168

    
169
schema: _always
170
	-echo 'CREATE SCHEMA "$(datasrc)";'|$(psqlAsBien)
171
# ignore errors if schema exists
172

    
173
rm_schema: _always
174
	echo 'DROP SCHEMA IF EXISTS "$(datasrc)" CASCADE;'|$(psqlAsBien)
175

    
176
# For staging tables which are derived by joining together other staging tables.
177
%/install %/header.csv: %/create.sql _always
178
	(echo 'SET search_path TO "$(datasrc)";'; echo 'CREATE TABLE "$*" AS';\
179
cat $<; echo ';'; \
180
echo 'ALTER TABLE "$*" ADD COLUMN row_num serial NOT NULL PRIMARY KEY;'; \
181
echo 'VACUUM ANALYZE "$*";')|$(psqlAsBien) --echo-all
182
	(prefix=; . $(bin)/vegbien_dest; env schema=$(datasrc) table=$* \
183
errors_table_only=1 $(bin)/csv2db)
184
	echo 'SELECT * FROM "$(datasrc)"."$*" LIMIT 0;'|$(psqlAsBien) \
185
--no-align --field-separator=, --pset=footer=off >$*/header.csv
186

    
187
# The joined tables must be suffixed with ".src" to prevent the creation of a
188
# row_num column, which collides during joins.
189
isJoinedTable = $(filter %.src,$*)
190
hasRowNum = $(if $(isJoinedTable),,1)
191

    
192
%/install: _always
193
	$(if $(isCsv),$(import_install_))
194
import_install_ = (prefix=; . $(bin)/vegbien_dest; "time" nice -n +5\
195
env schema=$(datasrc) table=$* has_row_num=$(hasRowNum) $(bin)/csv2db\
196
$(catSrcs) $(if $(log),$(if $(quiet),&>,2>&1|tee )$*/logs/install.log.sql))
197

    
198
##### Maps building
199

    
200
# WARNING: You CANNOT make a subdir using `make inputs/<datasrc>/<subdir>/`.
201
# You must instead make the entire datasource dir: `make inputs/<datasrc>/`
202

    
203
# Maps to (try to) build are added to one of these
204
maps :=
205
createOnlyMaps :=
206

    
207
srcRoot = $(mappings)/root.sh
208
mkSrcMap = $(catSrcs)|(. $(srcRoot); env datasrc=$(datasrc) $(bin)/src_map >$@)
209

    
210
# Src maps cleanup
211
ifneq ($(filter %/.src.csv.last_cleanup,$(MAKECMDGOALS)),)
212
%/.src.csv.last_cleanup: %/src.csv
213
	$(bin)/in_place $< $(bin)/cols +
214
	touch $@
215
else
216
# Autogen src maps with known table names
217
%/src.csv: _always
218
	$(if $(wildcard $@),,$(if $(isCsv),$(mkSrcMap)))
219
	+$(if $(isCsv),$(selfMake) $(@:%/src.csv=%/.src.csv.last_cleanup))
220
# only build if doesn't exist
221
# only build if CSV srcs exist for that table name
222
endif
223

    
224
createOnlyMaps += $(tables:%=%/src.csv)
225

    
226
# Must come before $(root)/% to override it
227
$(coreSelfMap): _always
228
	-+$(subMake)
229
# ignore errors if $(coreSelfMap) does not exist
230

    
231
# Via maps cleanup
232
ifneq ($(filter %/.map.csv.last_cleanup,$(MAKECMDGOALS)),)
233
%/.map.csv.last_cleanup: %/map.csv $(coreMap)
234
	$(bin)/in_place $< env ignore=1 $(bin)/subtract $(word 2,$+) 0 1 2
235
	touch $@
236
# Include comment column so commented mappings are never removed
237
else
238
%/map.csv: %/src.csv _always
239
	$(if $(wildcard $@),,$(if $(wildcard $<),$(CP) $< $@))
240
	+$(selfMake) $(@:%/map.csv=%/.map.csv.last_cleanup)
241
endif
242

    
243
srcMap* = $*/src.csv
244
joinSrcMap = $(if $(hasSrc),$(bin)/in_place $@ $(bin)/intersect $(srcMap*) 0)
245
hasSrc = $(shell test -s $(word 3,$+) && echo t)
246

    
247
makeFullCsv = env ignore=1 $(bin)/union <$(wordlist 1,2,$+)|$(bin)/sort_map >$@
248

    
249
%/map.full.csv: %/map.csv $(coreSelfMap)
250
	$(makeFullCsv)
251
	$(joinSrcMap)
252
maps += $(patsubst %/map.csv,%/map.full.csv,$(viaMaps))
253

    
254
%/VegBIEN.csv: %/map.full.csv $(coreMap)
255
	$(bin)/join <$+|$(bin)/sort_map >$@
256
maps += $(autogenMaps)
257

    
258
maps: $(createOnlyMaps) $(maps) _always ;
259

    
260
all += $(maps)
261

    
262
##### Maps validation
263

    
264
missingMappingsCmd = +$(selfMake) remake 2>&1\
265
|$(SED) -n 's/^.*No $(*2Space) mapping for (.*)$$/\1/p'|$(SED) 's/\/_.*//'\
266
$(if $(filter non-empty_join,$*),|$(bin)/ucase_first 0)|sort|uniq
267

    
268
include $(root)/lib/mappings.Makefile
269

    
270
##### External dependencies
271

    
272
$(root)/%: _always
273
	+$(subMake)
274
.PRECIOUS: $(root)/% # let ext. dir's Makefile decide whether to delete on error
275

    
276
##### Mapping
277

    
278
+maps = $(filter %/map.csv %/VegBIEN.csv $(mappings)/%,$(+_))
279
map = $(if $(srcs),$(withCatSrcs) $(root)/map $(+maps),\
280
$(shell echo Warning: No input file $*/*.{$(exts)} >&2)false)
281
# need false to run some command, when prefixed by `env ...` below
282
map2db = $(if $(use_staged),env in_database=vegbien in_schema=$(datasrc)\
283
in_table=$*) env out_database=vegbien $(map)
284

    
285
##### Import to VegBIEN
286

    
287
profileTest = $(if $(profile),$(if $(test),1))
288
profileOnly = -env profile_to=/dev/fd/3 $(map2db) 3>&1 1>&2|\
289
$(bin)/profile_stats /dev/fd/0
290

    
291
log_ = $*/logs/$(if $(n),n=$(n).,)$(date).log.sql
292
trace = $(log_:.log.sql=.trace)
293
import = -$(if $(profileTest),$(profileOnly),(set -x; "time" env commit=1\
294
$(if $(profile),profile_to=$(trace)) $(map2db)) $(if $(log),\
295
$(if $(n),,&>$(log_)))$(if $(log),$(if $(n), 2>&1|tee -a $(log_))))
296
# don't abort on import errors, which often relate to invalid input data
297

    
298
import: $(tables:%=%/import) _always ;
299

    
300
%/import: %/VegBIEN.csv _always
301
	$(import)
302
# default:
303
%/import: _always ;
304

    
305
##### Log files from import
306

    
307
logs := $(wildcard */logs/*.log.sql */logs/*.trace)
308

    
309
rm_logs: _always
310
	$(RM) $(logs)
311

    
312
##### Verification of import
313

    
314
verify: $(tables:%=%/verify) _always ;
315

    
316
%/verify: verify/%.ref verify/%.out _always
317
	-$(diffVerbose) $(+_)
318
# don't abort on verification errors, which are expected during development
319
# default:
320
%/verify: verify/%.out _always
321
	$(if $(shell test -e $< && echo t),cat $<)
322
# don't run if verify/%.out's default do-nothing action was used
323
# can't use $(wildcard) because it won't recheck file after verify/%.out is run
324

    
325
psqlExport := "time" $(psqlAsBien) --no-align --field-separator=$$'\t'\
326
--pset=footer=off --pset=null=NULL
327
verify = $(if $(reverify),$(psqlExport) --set=datasource="'$(datasrc)'" <$< >$@)
328

    
329
verify/%.out: $(mappings)/verify.%.sql _always
330
	$(verify)
331
# default:
332
verify/%.out: _always ;
333

    
334
all += $(wildcard verify/*.out)
335

    
336
##### Editing import
337

    
338
rotate: _always
339
	echo "UPDATE party SET organizationname = organizationname||'.$(date)'\
340
WHERE organizationname = '$(datasrc)';"|$(psqlAsBien)
341

    
342
rm: _always
343
	echo "DELETE FROM party WHERE organizationname = '$(datasrc)';"|\
344
$(psqlAsBien)
345

    
346
##### Testing
347

    
348
testRefOutput = $(subst .by_col,,$(1))
349
testRef = $(testRefOutput).ref
350
hasOwnRef = $(filter $@,$(call testRefOutput,$@))
351
# filter returns non-empty if they are equal
352

    
353
# `rm $@`: Remove outputs of successful tests to reduce clutter
354
# `$(foreach use_staged...)`: Run with use_staged=1
355
define runTest
356
@echo "Testing $(abspath $@)..."
357
>$@ env test=1 n=$(test_n) $(1) $(foreach use_staged,1,$(map2db))
358
@(set -x; $(diffIgnoreSpace) $(call testRef,$@) $@) 2>&1 && rm $@ || { e=$$?;\
359
$(if $(wildcard $(call testRef,$@)),,cat $@;)\
360
$(if $(hasOwnRef),\
361
{\
362
read -p $(emph)'Accept new test output? (y/n)'$(endEmph) REPLY;\
363
if test "$$REPLY" = y; then\
364
(set -x; $(MAKE) $@-ok --directory=$(realpath .) --makefile=../input.Makefile);\
365
exit 0;\
366
fi;\
367
};,\
368
echo $(emph)"Note: The preceding failed test is compared to another test's\
369
output"$(endEmph);\
370
echo $(emph)"When it fails, this always indicates a bug"$(endEmph);\
371
)\
372
exit $$e;}
373
endef
374

    
375
tests :=
376

    
377
# Requires staging tables. To create them, run `make inputs/<datasrc>/install`.
378
# Non-flat-file inputs fall back to mimicking a successful test
379
%/test.xml: %/VegBIEN.csv _always
380
	$(if $(isCsv),$(call runTest,by_col=))
381
tests += %/test.xml
382

    
383
%/test.by_col.xml: %/VegBIEN.csv _always
384
	-$(if $(isCsv),$(call runTest,by_col=1))
385
# Don't abort tester if only by-column test fails
386

    
387
# Only run column-based tests if column-based mode enabled, because these tests
388
# are much slower than the row-based tests for small numbers of rows
389
ifneq ($(by_col),)
390
tests += %/test.by_col.xml
391
endif
392

    
393
testOutputs := $(foreach test,$(tests),$(tables:%=$(test)))
394

    
395
.PRECIOUS: $(testOutputs) # save outputs of failed tests so they can be accepted
396

    
397
test: _always $(testOutputs) ;
398

    
399
all += $(wildcard %/test*.xml)
400

    
401
# Accepts a test output: make <test_output_path>-ok
402
%-ok: _always
403
	mv $* $(call testRef,$*)
404

    
405
accept-all: _always
406
	+yes|$(selfMake) test
407

    
408
##### Documentation
409

    
410
steps = $(selfMake) -s import-$* test=1 by_col=1 verbosity=2 n=100\
411
2>&1|$(bin)/debug2redmine >$@
412

    
413
%/logs/steps.by_col.log.sql: _always
414
	+$(steps)
(4-4/4)