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 7245 aaronmk
full_import ?=
10 6592 aaronmk
import_source ?= 1
11 1190 aaronmk
log ?= $(if $(test),,1)
12 9833 aaronmk
noclobber ?=
13 1257 aaronmk
profile ?=
14 1983 aaronmk
quiet ?=
15 1620 aaronmk
reverify ?= 1
16 5810 aaronmk
schema_only ?=
17 1990 aaronmk
use_staged ?= $(by_col)
18 1921 aaronmk
19
# Makefile
20 7650 aaronmk
exts ?= csv tsv tab txt dat dmp xml
21 783 aaronmk
test_n ?= 2
22
23
##### Vars/functions
24
25 1821 aaronmk
# Paths
26 5008 aaronmk
datasrc := $(patsubst .%,%,$(notdir $(realpath .)))
27 1821 aaronmk
bin := $(root)/bin
28
mappings := $(root)/mappings
29
30 368 aaronmk
# Make
31 640 aaronmk
SHELL := /bin/bash
32 1509 aaronmk
selfMake = $(MAKE) --makefile=../input.Makefile
33 1821 aaronmk
subMake = $(MAKE) $(@:$(root)/%=%) --directory=$(root)
34 404 aaronmk
+_ = $(+:_%=)
35 368 aaronmk
addBeforeExt = $(basename $(2))$(1)$(suffix $(2))
36
37 640 aaronmk
# Terminal
38 1184 aaronmk
termCols := $(shell tput cols)
39 640 aaronmk
40 368 aaronmk
# Commands
41 1245 aaronmk
MKDIR = mkdir -p
42
mkdir = $(MKDIR) $(@D)
43 1184 aaronmk
diff = diff --unified=2
44 3721 aaronmk
diffIgnoreSpace = $(diff) --ignore-space-change
45 1184 aaronmk
diffVerbose = $(if $(verbose),diff --side-by-side --left-column\
46
--width=$(termCols),$(diff))
47 368 aaronmk
48 1918 aaronmk
# BIEN commands
49 4137 aaronmk
sortFilenames = $(shell $(bin)/sort_filenames $(1))
50 1524 aaronmk
selfMap = $(bin)/cols 0 0
51 8108 aaronmk
psqlAsBien := $(bin)/psql_verbose_vegbien
52 6186 aaronmk
psqlNoSearchPath := env no_search_path=1 $(psqlAsBien)
53 4408 aaronmk
# Usage: ($(inDatasrc); cat $(file))|$(psqlCmd)
54 6186 aaronmk
inDatasrc := echo 'SET search_path TO "$(datasrc)";'
55 353 aaronmk
56 1594 aaronmk
# SVN
57 1771 aaronmk
setSvnIgnore = svn propset svn:ignore $(2) $(1)
58 1594 aaronmk
define addDirWithIgnore
59
$(addDir)
60 1771 aaronmk
$(setSvnIgnore)
61 1594 aaronmk
endef
62
63 4414 aaronmk
##### Environment
64
65
export PATH := $(bin):$(PATH)
66
67 783 aaronmk
##### General targets
68 250 aaronmk
69 418 aaronmk
all: _always maps ;
70 247 aaronmk
71 383 aaronmk
clean: _always
72 256 aaronmk
	$(RM) $(all)
73
74 1742 aaronmk
remake: _always clean
75
	+$(selfMake)
76
# re-run make so that cache of existing files is reset
77
78 5876 aaronmk
# Only remake if doesn't exist. This prevents unintentional remaking when the
79
# make script is newly checked out from svn (which sets the mod time to now) but
80
# the output is synced externally.
81
# Can't remove prereq to do this, because it determines when the rule applies.
82
make_script = $(if $(wildcard $@),,"time" ./$< >$@)
83 368 aaronmk
84 6743 aaronmk
%/: %/map.csv _always ;
85
86 1604 aaronmk
%/: % _always ;
87
88 4379 aaronmk
%: %.make
89 5876 aaronmk
	$(make_script)
90 4182 aaronmk
.PRECIOUS: % # save partial outputs of aborted src make scripts
91 1627 aaronmk
92 6379 aaronmk
##### Tables discovery
93 250 aaronmk
94 6379 aaronmk
sortFile := import_order.txt
95 6156 aaronmk
noImportFile := _no_import
96
97 9880 aaronmk
dontImport = $(wildcard $(noImportFile))$(wildcard $(1)/$(noImportFile))$(if\
98 6592 aaronmk
$(import_source),,$(filter Source,$(1)))
99 6379 aaronmk
100
tables := $(if $(wildcard $(sortFile)),$(shell cat $(sortFile)))
101
    # $(shell) replaces "\n" with " "
102
allSubdirs := $(call wildcard/,*/)
103
allTables := $(call sortFilenames,$(filter-out _% verify logs,$(allSubdirs:%/=%)))
104
joinedTables := $(filter-out $(tables),$(allTables))
105
allTables := $(strip $(joinedTables) $(tables))# move joined tables to beginning
106 8383 aaronmk
has_visible_files = $(call wildcard/,$(1)/*)
107
allTables := $(foreach table,$(allTables),$(if\
108
$(call has_visible_files,$(table)),$(table)))
109 6379 aaronmk
ifeq ($(tables),)# none specified in sort file
110
tables := $(allTables)
111
endif
112
importTables := $(foreach table,$(tables),$(if\
113
$(call dontImport,$(table)),,$(table)))
114
115
##### SVN
116
117 6990 aaronmk
svnFilesGlob:= */{,$(noImportFile),{,.}{map,VegBIEN}.csv{,.*},*header.*,*.sql,test.xml.ref}
118
svnFilesGlob := {map.csv,*{schema,~}*.sql,{,*/}{*.make,*terms.csv},$(svnFilesGlob)}
119 9951 aaronmk
_svnFilesGlob := {_MySQL/{,*schema*.sql,*.make},{,*/}{*{run,Makefile},*.{md5,url,pdf},*README.TXT}}
120 9417 aaronmk
svnFiles = $(filter-out _% logs/% %.data.sql,$(call wildcard/,$(svnFilesGlob)))\
121 5923 aaronmk
$(call wildcard/,$(_svnFilesGlob))
122 5878 aaronmk
123 6806 aaronmk
# To update all inputs with these settings: make inputs/add
124 8254 aaronmk
add: _always Source/add $(allTables:%=%/add)
125 4182 aaronmk
	$(call setSvnIgnore,.,'*')
126 8801 aaronmk
	$(call addDirWithIgnore,logs,$$'*.gz\n*.log.sql\n*.trace')
127 6805 aaronmk
	$(call addDirWithIgnore,verify,$$'*.out\n*.csv\n*.xls\n*.xlsx')
128 4223 aaronmk
	$(call addFile,import_order.txt)
129 6014 aaronmk
	$(if $(wildcard _MySQL/),$(call addDirWithIgnore,_MySQL,'*'))
130 6064 aaronmk
	$(if $(wildcard _src/),$(call addDirWithIgnore,_src,'*'))
131 6733 aaronmk
	$(if $(wildcard _archive/),$(call addDirWithIgnore,_archive,'*'))
132 5921 aaronmk
	$(call add*,$(svnFiles))
133 766 aaronmk
134 4217 aaronmk
# Adds a new table subdir
135
%/add: _always
136
	$(call addDirWithIgnore,$*,'*')
137 8801 aaronmk
	$(call addDirWithIgnore,$*/logs,$$'*.gz\n*.log.sql\n*.trace')
138 4217 aaronmk
139 1969 aaronmk
##### Existing maps discovery
140 1955 aaronmk
141 5035 aaronmk
anyMap := %/map.csv %/VegBIEN.csv %/unmapped_terms.csv %/new_terms.csv
142 3574 aaronmk
143 7386 aaronmk
exts := $(call ci,$(exts))
144 4214 aaronmk
extsFilter := $(addprefix %.,$(exts))
145
dataOnly = $(filter $(extsFilter),$(1))
146
147 4182 aaronmk
anyTest = $*/test.%
148 4214 aaronmk
srcsOnly = $(filter-out $(anyMap) $(anyTest) %/logs,$(call dataOnly,$(1)))
149 4120 aaronmk
150 6071 aaronmk
srcDict := map.csv
151
152 7485 aaronmk
vocab := $(mappings)/VegCore.vocab.csv
153
thesaurus := $(mappings)/VegCore.thesaurus.csv
154 4658 aaronmk
coreMap := $(mappings)/VegCore-VegBIEN.csv
155 783 aaronmk
156 6157 aaronmk
viaMaps := $(importTables:%=%/map.csv)
157 783 aaronmk
158 4117 aaronmk
autogenMaps := $(subst map.,VegBIEN.,$(viaMaps))
159 728 aaronmk
directMaps := $(autogenMaps) $(filter-out $(autogenMaps),\
160 4182 aaronmk
$(wildcard */VegBIEN.csv))
161 728 aaronmk
162 1969 aaronmk
##### Sources
163
164 4182 aaronmk
srcs = $(call sortFilenames,$(call srcsOnly,$(wildcard $*/*)))
165 4450 aaronmk
nonHeaderSrcs = $(filter-out %/header.csv,$(srcs))
166
isRef = $(if $(nonHeaderSrcs),,1)
167
    # empty subdir, so references an already-installed staging table
168 4452 aaronmk
isXml = $(filter %.xml,$(nonHeaderSrcs))
169
nonXml = $(if $(isXml),,1)
170
isCsv = $(if $(nonHeaderSrcs),$(if $(isXml),,1))
171 1969 aaronmk
    # true if $(srcs) non-empty and contains no *.xml
172 8159 aaronmk
catSrcs = $(bin)/cat$(if $(nonXml),_csv) $(nonHeaderSrcs)
173 1969 aaronmk
withCatSrcs = $(catSrcs:$(bin)/%=$(bin)/with_%) --
174
175 7076 aaronmk
# Run with `make -s` to avoid echoing make commands
176 6157 aaronmk
cat: $(importTables:%=%/cat) _always ;
177 1969 aaronmk
178 4252 aaronmk
%/cat: _always
179 1969 aaronmk
	$(catSrcs)
180
181 6363 aaronmk
##### Input data retrieval
182
183
# Must come before `%.sql: _MySQL/%.sql` to override it
184
%.sql: %.sql.make
185
	$(make_script)
186
187
# The export must be created with:
188
# `--compatible=postgresql --add-locks=false --set-charset --no-create-info`
189
# Must come before `%.sql: _MySQL/%.sql` to override it
190
%.data.sql: _MySQL/%.data.sql
191
	$(if $(wildcard $@),,$(bin)/my2pg.data <$< >$@)
192
193
# The export must be created with:
194
# `--compatible=postgresql --add-locks=false --set-charset`
195
# Add `--no-data` to create a schema-only export.
196
%.sql: _MySQL/%.sql
197
	$(if $(wildcard $@),,$(bin)/my2pg <$< >$@)
198
199 4121 aaronmk
##### Staging tables installation
200 1921 aaronmk
201 5693 aaronmk
srcTable := %.src
202
203 6364 aaronmk
dbExportsWildcard = $(sort $(patsubst _MySQL/%.make,%,$(wildcard\
204
$(1) _MySQL/$(1).make)))
205
206
dbExports := $(call dbExportsWildcard,*schema*.sql)# schemas first
207
ifeq ($(schema_only),) # add rest of .sql files
208
dbExports += $(filter-out $(dbExports),$(call dbExportsWildcard,*.sql))
209 5810 aaronmk
endif
210 4438 aaronmk
dbExports := $(strip $(dbExports))# += adds extra whitespace
211 6565 aaronmk
allInstalls := $(if $(dbExports),sql) $(filter-out Source,$(allTables))
212 1921 aaronmk
213 4413 aaronmk
install: _always schema $(allInstalls:%=%/install) ;
214
215 5209 aaronmk
uninstall: _always confirm_rm_schema rm_schema ;
216 4121 aaronmk
# rm_schema will also drop all staging tables
217 1921 aaronmk
218
reinstall: _always uninstall install ;
219
220 5209 aaronmk
confirm_rm_schema: _always
221
	$(if $(filter TNRS,$(datasrc)),$(call confirm,WARNING: This will delete the\
222
TNRS cache!,To save it: make backups/TNRS.backup-remake))
223
224 4121 aaronmk
schema: _always
225 6186 aaronmk
	-echo 'CREATE SCHEMA "$(datasrc)";'|$(psqlNoSearchPath)
226 1921 aaronmk
# ignore errors if schema exists
227
228 4121 aaronmk
rm_schema: _always
229 6186 aaronmk
	echo 'DROP SCHEMA IF EXISTS "$(datasrc)" CASCADE;'|$(psqlNoSearchPath)
230 1921 aaronmk
231 5162 aaronmk
installLog := logs/install.log.sql
232 4453 aaronmk
233 8382 aaronmk
log_dir = $(1)logs
234 9833 aaronmk
has_log_dir = $(call and,$(wildcard $(log_dir)),$(call not,$(noclobber)))
235 8382 aaronmk
logInstall = $(if $(has_log_dir),$(if $(quiet),$(2)$(1)$(installLog)\
236
2>&1,2>&1|tee $(3) $(1)$(installLog)))
237 5692 aaronmk
logInstallRoot = $(call logInstall,,>)
238 5162 aaronmk
logInstall* = $(call logInstall,$*/,>)
239
logInstall*Add = $(call logInstall,$*/,>>,-a)# append to log
240
241 4413 aaronmk
# Must come before %/install to override it
242 4434 aaronmk
sql/install: $(dbExports)
243 8835 aaronmk
	($(inDatasrc); cat $+|pg_dump_limit)|"time" env no_search_path=1 \
244
$(bin)/psql_script_vegbien --set=schema='"$(datasrc)"' $(logInstallRoot)
245 4413 aaronmk
246 5795 aaronmk
cleanup = $(if $(wildcard $*/cleanup.sql),($(inDatasrc); cat $*/cleanup.sql)\
247 6186 aaronmk
|"time" $(psqlNoSearchPath) --echo-all --set=table='"$*"' $(logInstall*Add),\
248 7065 aaronmk
(export schema=$(datasrc) table=$*; . $(bin)/vegbien_dest; unset schemas; \
249 6186 aaronmk
$(bin)/csv2db) $(logInstall*Add))
250 4449 aaronmk
251 8160 aaronmk
%/header.csv:
252 8201 aaronmk
	echo 'COPY (SELECT * FROM "$(datasrc)"."$*" LIMIT 0) TO STDOUT CSV HEADER;'|\
253
env no_search_path=1 $(bin)/psql_script_vegbien >$*/header.csv
254 4449 aaronmk
255 8160 aaronmk
exportHeader = $(selfMake) $*/header.csv
256
257 7067 aaronmk
# Don't try to edit a view. Must come before %/install to override it.
258
%_view/install: _always ;
259
260 8238 aaronmk
%.sql/run: _always
261
	$(if $(wildcard $(@D)),($(inDatasrc); cat $(@D))|(cd '$(*D)';\
262
"time" env no_search_path=1 ../$(bin)/psql_verbose_vegbien \
263
--set=table='"$(*D)"' --set=table_str=\''"$(*D)"'\'))
264
265 9842 aaronmk
%/postprocess: _always
266
	$(if $(wildcard $*/run),,$(selfMake) $*/postprocess.sql/run)
267 8239 aaronmk
	$(if $(wildcard $*/import),$*/import)
268 8238 aaronmk
269 4260 aaronmk
# For staging tables which are derived by joining together other staging tables.
270 4263 aaronmk
%/install %/header.csv: %/create.sql _always
271 4527 aaronmk
	($(inDatasrc); echo 'CREATE TABLE "$*" AS'; cat $<; echo ';')|"time" \
272 6186 aaronmk
$(psqlNoSearchPath) --echo-all --set=schema='"$(datasrc)"' --set=table='"$*"' \
273 5912 aaronmk
$(logInstall*)
274 8238 aaronmk
	$(selfMake) $*/postprocess
275 4449 aaronmk
	$(exportHeader)
276 8157 aaronmk
	$(cleanup)
277 9968 aaronmk
ifneq ($(noclobber),)
278
.PRECIOUS: %/header.csv
279
endif
280 4260 aaronmk
281 4252 aaronmk
%/install: _always
282 8380 aaronmk
	$(if $(isCsv),$(import_install_))
283 8238 aaronmk
	$(selfMake) $*/postprocess
284 8381 aaronmk
	-$(exportHeader)
285 9843 aaronmk
	-$(if $(isCsv),,$(cleanup))
286 8381 aaronmk
# ignore errors if table does not exist (non-data dir)
287 5693 aaronmk
define import_install_
288 6950 aaronmk
(. $(bin)/vegbien_dest; unset schemas; "time" nice -n +5\
289 5028 aaronmk
env schema=$(datasrc) table=$* $(bin)/csv2db $(catSrcs) $(logInstall*))
290 5693 aaronmk
$(if $(filter $(srcTable),$*),($(inDatasrc);\
291
echo 'ALTER TABLE "$(datasrc)"."$*" RENAME row_num TO "$*.row_num";')|"time"\
292 6186 aaronmk
$(psqlNoSearchPath) --echo-all --set=table='"$*"' $(logInstall*Add))
293 5693 aaronmk
endef
294 5697 aaronmk
# table-scope src table's row_num col to allow joining it with other tables
295 1921 aaronmk
296 5819 aaronmk
%/uninstall: _always
297 6186 aaronmk
	echo 'DROP TABLE IF EXISTS "$(datasrc)"."$*" CASCADE;'|$(psqlNoSearchPath)
298 5819 aaronmk
299
%/reinstall: _always %/uninstall %/install ;
300
301 8241 aaronmk
postprocess: _always $(allTables:%=%/postprocess) ;
302 8113 aaronmk
303 5155 aaronmk
cleanup: _always $(tables:%=%/cleanup) ;
304
305 5184 aaronmk
# WARNING: This removes any index comments, due to a PostgreSQL bug.
306
# This occurs because ALTER TABLE recreates the index but not its comment.
307 5155 aaronmk
%/cleanup: _always
308
	$(cleanup)
309
310 1777 aaronmk
##### Maps building
311
312 4646 aaronmk
# Maps to (try to) build are added to this
313 1779 aaronmk
maps :=
314
315 4208 aaronmk
srcRoot = $(mappings)/root.sh
316 8156 aaronmk
mkSrcMap = (. $(srcRoot); env datasrc=$(datasrc) $(bin)/src_map <$*/header.csv\
317
>$@)
318 1779 aaronmk
319 7437 aaronmk
translate = $(if $(wildcard $(1)),$(bin)/in_place $< $(bin)/translate_ci 1 $(1))
320 6074 aaronmk
321 6076 aaronmk
$(srcDict):# empty target in case it doesn't exist
322
323 3572 aaronmk
# Via maps cleanup
324 4182 aaronmk
ifneq ($(filter %/.map.csv.last_cleanup,$(MAKECMDGOALS)),)
325 7484 aaronmk
%/.map.csv.last_cleanup: %/map.csv $(vocab) $(thesaurus) $(coreMap) $(srcDict)
326 7437 aaronmk
	$(call translate,$(srcDict))
327 4844 aaronmk
	$(bin)/in_place $< $(bin)/canon 1 $(vocab)
328 7484 aaronmk
	$(call translate,$(thesaurus))
329 8176 aaronmk
	$(bin)/in_place $< $(bin)/fix_line_endings
330 3572 aaronmk
	touch $@
331 4601 aaronmk
	+$(selfMake) $(<:%/map.csv=%/unmapped_terms.csv)
332
	+$(selfMake) $(<:%/map.csv=%/new_terms.csv)
333 5253 aaronmk
.PRECIOUS: %/.map.csv.last_cleanup
334 3572 aaronmk
else
335 4644 aaronmk
%/map.csv: _always
336 8252 aaronmk
	$(if $(wildcard $@),,$(mk_map_csv))
337 4182 aaronmk
	+$(selfMake) $(@:%/map.csv=%/.map.csv.last_cleanup)
338 8252 aaronmk
define mk_map_csv
339
+$(selfMake) $*/header.csv
340
$(if $(nonXml),$(mkSrcMap))
341
endef
342 3572 aaronmk
endif
343
344 4641 aaronmk
%/VegBIEN.csv: %/map.csv $(coreMap)
345 9875 aaronmk
	<$< $(bin)/cat_cols 1 2$(if $(wildcard\
346
$*/run),|$(bin)/cols 1 1 3|(head -1 $<; tail -n +2)\
347 9874 aaronmk
)|$(bin)/join $(coreMap)|$(bin)/sort_map >$@
348 1530 aaronmk
maps += $(autogenMaps)
349
350 4646 aaronmk
maps: $(maps) _always ;
351 1139 aaronmk
352
all += $(maps)
353
354 1742 aaronmk
##### Maps validation
355
356 4663 aaronmk
# `tail -n +2`: Remove header before running filter_out_ci because filter_out_ci
357
# only removes the header if it matches the vocabulary's header.
358 4601 aaronmk
359 4664 aaronmk
%/unmapped_terms.csv: %/map.csv $(coreMap)
360 4858 aaronmk
	tail -n +2 $<|$(bin)/cols 1|$(bin)/filter_out_ci 0 $(coreMap) >$@
361
	$(bin)/autoremove $@
362 4601 aaronmk
363 7484 aaronmk
%/new_terms.csv: %/map.csv $(vocab) $(thesaurus) %/unmapped_terms.csv
364 4857 aaronmk
	$(newTerms)
365
	$(bin)/autoremove $@
366
newTerms = tail -n +2 $<|$(bin)/filter_out_ci 0 $(vocab)|$(bin)/filter_out_ci 0\
367 7484 aaronmk
$(thesaurus) $(if $(wildcard $(word 4,$+)),|$(bin)/filter_out_ci 0 $(word 4,$+))\
368 7216 aaronmk
|grep -vE -e '^"?:' -e 'UNUSED' >$@; exit 0 # grep exits nonzero if no match
369 4601 aaronmk
370 6157 aaronmk
termsSubdirs := $(importTables)
371 4600 aaronmk
372 3764 aaronmk
include $(root)/lib/mappings.Makefile
373 1742 aaronmk
374 1509 aaronmk
##### External dependencies
375
376
$(root)/%: _always
377
	+$(subMake)
378 1628 aaronmk
.PRECIOUS: $(root)/% # let ext. dir's Makefile decide whether to delete on error
379 1509 aaronmk
380 783 aaronmk
##### Mapping
381 368 aaronmk
382 4182 aaronmk
+maps = $(filter %/map.csv %/VegBIEN.csv $(mappings)/%,$(+_))
383 4746 aaronmk
map2db = env in_database=vegbien in_schema=$(datasrc) in_table=$*\
384 4748 aaronmk
out_database=vegbien $(root)/map $(+maps)
385 368 aaronmk
386 783 aaronmk
##### Import to VegBIEN
387 772 aaronmk
388 7281 aaronmk
import: $(importTables:%=%/import) _always ;
389
390
rm: _always
391
	echo "DELETE FROM source WHERE shortname = '$(datasrc)';"\
392
|"time" $(psqlAsBien)
393
394 7284 aaronmk
reimport: _always rm import ;
395
396 1853 aaronmk
profileTest = $(if $(profile),$(if $(test),1))
397
profileOnly = -env profile_to=/dev/fd/3 $(map2db) 3>&1 1>&2|\
398
$(bin)/profile_stats /dev/fd/0
399
400 5457 aaronmk
log_ = $*/logs/$(if $(n),n=$(n).,)$(version).log.sql
401 7075 aaronmk
402
# Displays the import log file path
403 7076 aaronmk
# Run with `make -s` to avoid echoing make commands
404 7075 aaronmk
%/log_file: _always
405
	echo $$(pwd)/$(log_)
406
407 3317 aaronmk
trace = $(log_:.log.sql=.trace)
408 7080 aaronmk
import = -$(if $(profileTest),$(profileOnly),(set -x; date; "time" env commit=1\
409 7079 aaronmk
$(if $(profile),profile_to=$(trace)) $(map2db))$(if $(log), >>$(log_) 2>&1))
410 1088 aaronmk
# don't abort on import errors, which often relate to invalid input data
411 624 aaronmk
412 7245 aaronmk
import? = $(if $(call and,$(full_import),$(call dontImport,.)),,$(import))
413 7047 aaronmk
414 4252 aaronmk
%/import: %/VegBIEN.csv _always
415 7047 aaronmk
	$(import?)
416 772 aaronmk
# default:
417 4252 aaronmk
%/import: _always ;
418 718 aaronmk
419 7279 aaronmk
#### Taxonomic scrubbing
420
421 7282 aaronmk
scrub: _always
422 7275 aaronmk
	$(selfMake) $(root)/scrub & echo $$!
423 7265 aaronmk
424 7279 aaronmk
# Removes TNRS taxondeterminations
425
unscrub: _always
426
	echo "SELECT delete_scrubbed_taxondeterminations('$(datasrc)');"|\
427
"time" env no_query_results=1 $(psqlAsBien)
428
429 7283 aaronmk
rescrub: _always unscrub scrub ;
430
431 7282 aaronmk
import_scrub: _always import scrub ;
432
433 7285 aaronmk
reimport_scrub: _always reimport scrub ;
434
435 790 aaronmk
##### Log files from import
436
437 4182 aaronmk
logs := $(wildcard */logs/*.log.sql */logs/*.trace)
438 339 aaronmk
439 383 aaronmk
rm_logs: _always
440 339 aaronmk
	$(RM) $(logs)
441
442 783 aaronmk
##### Verification of import
443 264 aaronmk
444 5782 aaronmk
verifyTables := $(patsubst verify/%.ref,%,$(wildcard verify/*.ref))
445 1194 aaronmk
446 5782 aaronmk
verify: $(verifyTables:%=%/verify) _always ;
447
448 4252 aaronmk
%/verify: verify/%.ref verify/%.out _always
449 1184 aaronmk
	-$(diffVerbose) $(+_)
450 1088 aaronmk
# don't abort on verification errors, which are expected during development
451 1194 aaronmk
# default:
452 4252 aaronmk
%/verify: verify/%.out _always
453 1200 aaronmk
	$(if $(shell test -e $< && echo t),cat $<)
454 1199 aaronmk
# don't run if verify/%.out's default do-nothing action was used
455 1200 aaronmk
# can't use $(wildcard) because it won't recheck file after verify/%.out is run
456 368 aaronmk
457 6186 aaronmk
psqlExport := "time" $(psqlNoSearchPath) --no-align --field-separator=$$'\t'\
458 4401 aaronmk
--pset=footer=off --pset=null=NULL
459 5976 aaronmk
# Note that using $(inDatasrc) will not work with datasources whose tables are
460
# the same name as VegBIEN tables (likely only VegBank), because the datasource
461
# is first in the search_path.
462
verify = $(if $(reverify),($(inDatasrc); cat $<)|$(psqlExport)\
463
--set=datasource="'$(datasrc)'" >$@)
464 514 aaronmk
465 1199 aaronmk
verify/%.out: $(mappings)/verify.%.sql _always
466 1192 aaronmk
	$(verify)
467 5973 aaronmk
.PRECIOUS: verify/%.out # save partial output in case of error to help debugging
468 1194 aaronmk
# default:
469 1199 aaronmk
verify/%.out: _always ;
470 369 aaronmk
471 1199 aaronmk
all += $(wildcard verify/*.out)
472 1192 aaronmk
473 4402 aaronmk
%.ref: %.ref.sql
474 4409 aaronmk
	($(inDatasrc); cat $<)|$(psqlExport) >$@
475 5981 aaronmk
.PRECIOUS: %.ref # there must always be a .ref for the make rules to work
476 4402 aaronmk
477 1667 aaronmk
##### Editing import
478
479 6984 aaronmk
rename/%: _always
480
	echo "UPDATE source SET shortname = '$*' WHERE shortname = '$(datasrc)';"\
481
|$(psqlAsBien)
482 1667 aaronmk
483 783 aaronmk
##### Testing
484 368 aaronmk
485 4274 aaronmk
testRefOutput = $(subst .by_col,,$(1))
486 1986 aaronmk
testRef = $(testRefOutput).ref
487
hasOwnRef = $(filter $@,$(call testRefOutput,$@))
488
# filter returns non-empty if they are equal
489 630 aaronmk
490 4246 aaronmk
# `rm $@`: Remove outputs of successful tests to reduce clutter
491 4272 aaronmk
# `$(foreach use_staged...)`: Run with use_staged=1
492 897 aaronmk
define runTest
493 776 aaronmk
@echo "Testing $(abspath $@)..."
494 4272 aaronmk
>$@ env test=1 n=$(test_n) $(1) $(foreach use_staged,1,$(map2db))
495 6446 aaronmk
$(if $(hasOwnRef),,-)@(set -x; $(diffIgnoreSpace) $(call testRef,$@) $@) 2>&1\
496
&& rm $@ || { e=$$?; $(if $(wildcard $(call testRef,$@)),,cat $@;)\
497 777 aaronmk
$(if $(hasOwnRef),\
498 3179 aaronmk
{\
499
read -p $(emph)'Accept new test output? (y/n)'$(endEmph) REPLY;\
500
if test "$$REPLY" = y; then\
501
(set -x; $(MAKE) $@-ok --directory=$(realpath .) --makefile=../input.Makefile);\
502
exit 0;\
503
fi;\
504
};,\
505 777 aaronmk
echo $(emph)"Note: The preceding failed test is compared to another test's\
506
output"$(endEmph);\
507
echo $(emph)"When it fails, this always indicates a bug"$(endEmph);\
508
)\
509 652 aaronmk
exit $$e;}
510 628 aaronmk
endef
511
512 876 aaronmk
tests :=
513
514 6744 aaronmk
%/test: %/test.xml _always ;
515
516 4269 aaronmk
# Requires staging tables. To create them, run `make inputs/<datasrc>/install`.
517
# Non-flat-file inputs fall back to mimicking a successful test
518 4182 aaronmk
%/test.xml: %/VegBIEN.csv _always
519 4452 aaronmk
	$(if $(nonXml),$(call runTest,by_col=))
520 4182 aaronmk
tests += %/test.xml
521 627 aaronmk
522 4274 aaronmk
%/test.by_col.xml: %/VegBIEN.csv _always
523 4831 aaronmk
	$(if $(nonXml),$(call runTest,by_col=1))
524 4275 aaronmk
525
# Only run column-based tests if column-based mode enabled, because these tests
526
# are much slower than the row-based tests for small numbers of rows
527
ifneq ($(by_col),)
528 4274 aaronmk
tests += %/test.by_col.xml
529 4275 aaronmk
endif
530 4274 aaronmk
531 6157 aaronmk
testOutputs := $(foreach test,$(tests),$(importTables:%=$(test)))
532 724 aaronmk
533 1361 aaronmk
.PRECIOUS: $(testOutputs) # save outputs of failed tests so they can be accepted
534
535 876 aaronmk
test: _always $(testOutputs) ;
536
537 4182 aaronmk
all += $(wildcard %/test*.xml)
538 876 aaronmk
539 634 aaronmk
# Accepts a test output: make <test_output_path>-ok
540 626 aaronmk
%-ok: _always
541 4383 aaronmk
	mv $* $(call testRef,$*)
542 502 aaronmk
543 4120 aaronmk
accept-all: _always
544 3579 aaronmk
	+yes|$(selfMake) test
545
546 3133 aaronmk
##### Documentation
547
548 4494 aaronmk
steps = $(selfMake) -s $*/import test=1 by_col=1 verbosity=2 n=100\
549 3385 aaronmk
2>&1|$(bin)/debug2redmine >$@
550 3188 aaronmk
551 4182 aaronmk
%/logs/steps.by_col.log.sql: _always
552 3188 aaronmk
	+$(steps)