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 10582 aaronmk
continue ?=
10 10106 aaronmk
debug ?=
11 7245 aaronmk
full_import ?=
12 6592 aaronmk
import_source ?= 1
13 11251 aaronmk
is_view ?=
14 1190 aaronmk
log ?= $(if $(test),,1)
15 9833 aaronmk
noclobber ?=
16 1257 aaronmk
profile ?=
17 1983 aaronmk
quiet ?=
18 1620 aaronmk
reverify ?= 1
19 5810 aaronmk
schema_only ?=
20 1990 aaronmk
use_staged ?= $(by_col)
21 1921 aaronmk
22
# Makefile
23 7650 aaronmk
exts ?= csv tsv tab txt dat dmp xml
24 783 aaronmk
test_n ?= 2
25
26
##### Vars/functions
27
28 1821 aaronmk
# Paths
29 5008 aaronmk
datasrc := $(patsubst .%,%,$(notdir $(realpath .)))
30 1821 aaronmk
bin := $(root)/bin
31
mappings := $(root)/mappings
32
33 368 aaronmk
# Make
34 640 aaronmk
SHELL := /bin/bash
35 1509 aaronmk
selfMake = $(MAKE) --makefile=../input.Makefile
36 11253 aaronmk
subMake = $(MAKE) "$(@:$(root)/%=%)" --directory=$(root)
37 404 aaronmk
+_ = $(+:_%=)
38 10103 aaronmk
# used to filter the output of embedded $(shell make ...) invocations
39
filter_make := grep -vF -e lib -e ".Makefile'."
40 368 aaronmk
addBeforeExt = $(basename $(2))$(1)$(suffix $(2))
41
42 640 aaronmk
# Terminal
43 1184 aaronmk
termCols := $(shell tput cols)
44 640 aaronmk
45 368 aaronmk
# Commands
46 1245 aaronmk
MKDIR = mkdir -p
47
mkdir = $(MKDIR) $(@D)
48 1184 aaronmk
diff = diff --unified=2
49 3721 aaronmk
diffIgnoreSpace = $(diff) --ignore-space-change
50 1184 aaronmk
diffVerbose = $(if $(verbose),diff --side-by-side --left-column\
51
--width=$(termCols),$(diff))
52 368 aaronmk
53 1918 aaronmk
# BIEN commands
54 4137 aaronmk
sortFilenames = $(shell $(bin)/sort_filenames $(1))
55 1524 aaronmk
selfMap = $(bin)/cols 0 0
56 8108 aaronmk
psqlAsBien := $(bin)/psql_verbose_vegbien
57 6186 aaronmk
psqlNoSearchPath := env no_search_path=1 $(psqlAsBien)
58 4408 aaronmk
# Usage: ($(inDatasrc); cat $(file))|$(psqlCmd)
59 6186 aaronmk
inDatasrc := echo 'SET search_path TO "$(datasrc)";'
60 353 aaronmk
61 1594 aaronmk
# SVN
62 1771 aaronmk
setSvnIgnore = svn propset svn:ignore $(2) $(1)
63 1594 aaronmk
define addDirWithIgnore
64
$(addDir)
65 1771 aaronmk
$(setSvnIgnore)
66 1594 aaronmk
endef
67
68 4414 aaronmk
##### Environment
69
70
export PATH := $(bin):$(PATH)
71
72 783 aaronmk
##### General targets
73 250 aaronmk
74 418 aaronmk
all: _always maps ;
75 247 aaronmk
76 383 aaronmk
clean: _always
77 256 aaronmk
	$(RM) $(all)
78
79 1742 aaronmk
remake: _always clean
80
	+$(selfMake)
81
# re-run make so that cache of existing files is reset
82
83 5876 aaronmk
# Only remake if doesn't exist. This prevents unintentional remaking when the
84
# make script is newly checked out from svn (which sets the mod time to now) but
85
# the output is synced externally.
86
# Can't remove prereq to do this, because it determines when the rule applies.
87
make_script = $(if $(wildcard $@),,"time" ./$< >$@)
88 368 aaronmk
89 6743 aaronmk
%/: %/map.csv _always ;
90
91 1604 aaronmk
%/: % _always ;
92
93 4379 aaronmk
%: %.make
94 5876 aaronmk
	$(make_script)
95 4182 aaronmk
.PRECIOUS: % # save partial outputs of aborted src make scripts
96 1627 aaronmk
97 6379 aaronmk
##### Tables discovery
98 250 aaronmk
99 6379 aaronmk
sortFile := import_order.txt
100 6156 aaronmk
noImportFile := _no_import
101
102 9880 aaronmk
dontImport = $(wildcard $(noImportFile))$(wildcard $(1)/$(noImportFile))$(if\
103 6592 aaronmk
$(import_source),,$(filter Source,$(1)))
104 6379 aaronmk
105 10105 aaronmk
ifeq ($(sort_file_updated),)# keep $(sortFile) up-to-date
106
$(shell sort_file_updated=1 $(selfMake) $(sortFile)|$(filter_make) >&2)
107
export sort_file_updated=1
108 10104 aaronmk
endif
109
110 10101 aaronmk
sort_file_tables := $(if $(wildcard $(sortFile)),$(shell cat $(sortFile)))
111 6379 aaronmk
    # $(shell) replaces "\n" with " "
112 10101 aaronmk
tables := $(sort_file_tables)
113 6379 aaronmk
allSubdirs := $(call wildcard/,*/)
114
allTables := $(call sortFilenames,$(filter-out _% verify logs,$(allSubdirs:%/=%)))
115
joinedTables := $(filter-out $(tables),$(allTables))
116
allTables := $(strip $(joinedTables) $(tables))# move joined tables to beginning
117 8383 aaronmk
has_visible_files = $(call wildcard/,$(1)/*)
118
allTables := $(foreach table,$(allTables),$(if\
119
$(call has_visible_files,$(table)),$(table)))
120 6379 aaronmk
ifeq ($(tables),)# none specified in sort file
121
tables := $(allTables)
122
endif
123
importTables := $(foreach table,$(tables),$(if\
124
$(call dontImport,$(table)),,$(table)))
125
126 10100 aaronmk
list_tables: _always # use `make -s` to avoid echoing commands
127
	@for table in $(tables); do echo "$$table"; done
128
129 10101 aaronmk
$(sortFile): _always
130
	# add any missing tables to $(sortFile)
131 10102 aaronmk
	$(if $(filter-out $(sort_file_tables),$(tables)),$(selfMake) -s list_tables >$@)
132 10101 aaronmk
133 6379 aaronmk
##### SVN
134
135 10162 aaronmk
svnFilesGlob:= */{,$(noImportFile),{,.}{data,map,VegBIEN}.csv{,.*},*header.*,*.sql,test.xml.ref}
136 10107 aaronmk
svnFilesGlob := {map.csv,*{schema,grants,~}*.sql,{,*/}{*.make,*terms.csv},$(svnFilesGlob)}
137 9951 aaronmk
_svnFilesGlob := {_MySQL/{,*schema*.sql,*.make},{,*/}{*{run,Makefile},*.{md5,url,pdf},*README.TXT}}
138 9417 aaronmk
svnFiles = $(filter-out _% logs/% %.data.sql,$(call wildcard/,$(svnFilesGlob)))\
139 5923 aaronmk
$(call wildcard/,$(_svnFilesGlob))
140 5878 aaronmk
141 10167 aaronmk
add: _always $(if $(call dontImport,.),,add!) ;
142
143 6806 aaronmk
# To update all inputs with these settings: make inputs/add
144 10167 aaronmk
add!: _always Source/add $(allTables:%=%/add)
145 4182 aaronmk
	$(call setSvnIgnore,.,'*')
146 8801 aaronmk
	$(call addDirWithIgnore,logs,$$'*.gz\n*.log.sql\n*.trace')
147 6805 aaronmk
	$(call addDirWithIgnore,verify,$$'*.out\n*.csv\n*.xls\n*.xlsx')
148 4223 aaronmk
	$(call addFile,import_order.txt)
149 6014 aaronmk
	$(if $(wildcard _MySQL/),$(call addDirWithIgnore,_MySQL,'*'))
150 6064 aaronmk
	$(if $(wildcard _src/),$(call addDirWithIgnore,_src,'*'))
151 6733 aaronmk
	$(if $(wildcard _archive/),$(call addDirWithIgnore,_archive,'*'))
152 5921 aaronmk
	$(call add*,$(svnFiles))
153 766 aaronmk
154 4217 aaronmk
# Adds a new table subdir
155
%/add: _always
156
	$(call addDirWithIgnore,$*,'*')
157 8801 aaronmk
	$(call addDirWithIgnore,$*/logs,$$'*.gz\n*.log.sql\n*.trace')
158 4217 aaronmk
159 1969 aaronmk
##### Existing maps discovery
160 1955 aaronmk
161 5035 aaronmk
anyMap := %/map.csv %/VegBIEN.csv %/unmapped_terms.csv %/new_terms.csv
162 3574 aaronmk
163 7386 aaronmk
exts := $(call ci,$(exts))
164 4214 aaronmk
extsFilter := $(addprefix %.,$(exts))
165
dataOnly = $(filter $(extsFilter),$(1))
166
167 4182 aaronmk
anyTest = $*/test.%
168 4214 aaronmk
srcsOnly = $(filter-out $(anyMap) $(anyTest) %/logs,$(call dataOnly,$(1)))
169 4120 aaronmk
170 6071 aaronmk
srcDict := map.csv
171
172 7485 aaronmk
vocab := $(mappings)/VegCore.vocab.csv
173
thesaurus := $(mappings)/VegCore.thesaurus.csv
174 4658 aaronmk
coreMap := $(mappings)/VegCore-VegBIEN.csv
175 783 aaronmk
176 10205 aaronmk
viaMaps := $(tables:%=%/map.csv)
177 783 aaronmk
178 4117 aaronmk
autogenMaps := $(subst map.,VegBIEN.,$(viaMaps))
179 728 aaronmk
directMaps := $(autogenMaps) $(filter-out $(autogenMaps),\
180 4182 aaronmk
$(wildcard */VegBIEN.csv))
181 728 aaronmk
182 1969 aaronmk
##### Sources
183
184 4182 aaronmk
srcs = $(call sortFilenames,$(call srcsOnly,$(wildcard $*/*)))
185 4450 aaronmk
nonHeaderSrcs = $(filter-out %/header.csv,$(srcs))
186
isRef = $(if $(nonHeaderSrcs),,1)
187
    # empty subdir, so references an already-installed staging table
188 4452 aaronmk
isXml = $(filter %.xml,$(nonHeaderSrcs))
189
nonXml = $(if $(isXml),,1)
190
isCsv = $(if $(nonHeaderSrcs),$(if $(isXml),,1))
191 1969 aaronmk
    # true if $(srcs) non-empty and contains no *.xml
192 8159 aaronmk
catSrcs = $(bin)/cat$(if $(nonXml),_csv) $(nonHeaderSrcs)
193 1969 aaronmk
withCatSrcs = $(catSrcs:$(bin)/%=$(bin)/with_%) --
194
195 7076 aaronmk
# Run with `make -s` to avoid echoing make commands
196 6157 aaronmk
cat: $(importTables:%=%/cat) _always ;
197 1969 aaronmk
198 4252 aaronmk
%/cat: _always
199 1969 aaronmk
	$(catSrcs)
200
201 6363 aaronmk
##### Input data retrieval
202
203
# Must come before `%.sql: _MySQL/%.sql` to override it
204
%.sql: %.sql.make
205
	$(make_script)
206
207
# The export must be created with:
208
# `--compatible=postgresql --add-locks=false --set-charset --no-create-info`
209
# Must come before `%.sql: _MySQL/%.sql` to override it
210
%.data.sql: _MySQL/%.data.sql
211
	$(if $(wildcard $@),,$(bin)/my2pg.data <$< >$@)
212
213
# The export must be created with:
214
# `--compatible=postgresql --add-locks=false --set-charset`
215
# Add `--no-data` to create a schema-only export.
216
%.sql: _MySQL/%.sql
217
	$(if $(wildcard $@),,$(bin)/my2pg <$< >$@)
218
219 4121 aaronmk
##### Staging tables installation
220 1921 aaronmk
221 5693 aaronmk
srcTable := %.src
222
223 6364 aaronmk
dbExportsWildcard = $(sort $(patsubst _MySQL/%.make,%,$(wildcard\
224
$(1) _MySQL/$(1).make)))
225
226
dbExports := $(call dbExportsWildcard,*schema*.sql)# schemas first
227
ifeq ($(schema_only),) # add rest of .sql files
228
dbExports += $(filter-out $(dbExports),$(call dbExportsWildcard,*.sql))
229 5810 aaronmk
endif
230 10107 aaronmk
dbExports := $(filter-out grants.sql,$(dbExports))
231 4438 aaronmk
dbExports := $(strip $(dbExports))# += adds extra whitespace
232 10256 aaronmk
allInstalls := $(if $(dbExports),sql) $(allTables)
233 1921 aaronmk
234 4413 aaronmk
install: _always schema $(allInstalls:%=%/install) ;
235
236 5209 aaronmk
uninstall: _always confirm_rm_schema rm_schema ;
237 4121 aaronmk
# rm_schema will also drop all staging tables
238 1921 aaronmk
239
reinstall: _always uninstall install ;
240
241 5209 aaronmk
confirm_rm_schema: _always
242
	$(if $(filter TNRS,$(datasrc)),$(call confirm,WARNING: This will delete the\
243
TNRS cache!,To save it: make backups/TNRS.backup-remake))
244
245 4121 aaronmk
schema: _always
246 6186 aaronmk
	-echo 'CREATE SCHEMA "$(datasrc)";'|$(psqlNoSearchPath)
247 1921 aaronmk
# ignore errors if schema exists
248
249 4121 aaronmk
rm_schema: _always
250 6186 aaronmk
	echo 'DROP SCHEMA IF EXISTS "$(datasrc)" CASCADE;'|$(psqlNoSearchPath)
251 1921 aaronmk
252 5162 aaronmk
installLog := logs/install.log.sql
253 4453 aaronmk
254 8382 aaronmk
log_dir = $(1)logs
255 9833 aaronmk
has_log_dir = $(call and,$(wildcard $(log_dir)),$(call not,$(noclobber)))
256 8382 aaronmk
logInstall = $(if $(has_log_dir),$(if $(quiet),$(2)$(1)$(installLog)\
257
2>&1,2>&1|tee $(3) $(1)$(installLog)))
258 5692 aaronmk
logInstallRoot = $(call logInstall,,>)
259 5162 aaronmk
logInstall* = $(call logInstall,$*/,>)
260
logInstall*Add = $(call logInstall,$*/,>>,-a)# append to log
261
262 4413 aaronmk
# Must come before %/install to override it
263 4434 aaronmk
sql/install: $(dbExports)
264 10107 aaronmk
	($(inDatasrc); cat $+|pg_dump_limit; $(if $(wildcard\
265
grants.sql),cat grants.sql))|"time" env no_search_path=1 \
266 10106 aaronmk
$(bin)/psql_$(if $(debug),verbose,script)_vegbien --set=schema='"$(datasrc)"' \
267
$(logInstallRoot)
268 4413 aaronmk
269 10106 aaronmk
# $debug option runs the *.sql import verbosely, to display which statements are
270
# being run. this should only be used for SQL files that use COPY FROM to import
271
# data, to avoid echoing pages of insert statements.
272 5795 aaronmk
cleanup = $(if $(wildcard $*/cleanup.sql),($(inDatasrc); cat $*/cleanup.sql)\
273 6186 aaronmk
|"time" $(psqlNoSearchPath) --echo-all --set=table='"$*"' $(logInstall*Add),\
274 7065 aaronmk
(export schema=$(datasrc) table=$*; . $(bin)/vegbien_dest; unset schemas; \
275 6186 aaronmk
$(bin)/csv2db) $(logInstall*Add))
276 4449 aaronmk
277 8160 aaronmk
%/header.csv:
278 8201 aaronmk
	echo 'COPY (SELECT * FROM "$(datasrc)"."$*" LIMIT 0) TO STDOUT CSV HEADER;'|\
279
env no_search_path=1 $(bin)/psql_script_vegbien >$*/header.csv
280 4449 aaronmk
281 11253 aaronmk
exportHeader = $(selfMake) "$*/header.csv"
282 8160 aaronmk
283 7067 aaronmk
# Don't try to edit a view. Must come before %/install to override it.
284
%_view/install: _always ;
285
286 8238 aaronmk
%.sql/run: _always
287
	$(if $(wildcard $(@D)),($(inDatasrc); cat $(@D))|(cd '$(*D)';\
288
"time" env no_search_path=1 ../$(bin)/psql_verbose_vegbien \
289
--set=table='"$(*D)"' --set=table_str=\''"$(*D)"'\'))
290
291 10312 aaronmk
%/postprocess.sql: %/map.csv
292 10347 aaronmk
	$(if $(wildcard $*/run),$(bin)/in_place $@ env text=1 $(bin)/repl $<)
293 10312 aaronmk
294 9842 aaronmk
%/postprocess: _always
295 11253 aaronmk
	$(if $(wildcard $*/run),,$(selfMake) "$*/postprocess.sql/run")
296 8238 aaronmk
297 10240 aaronmk
%/map_table: _always
298
	$(if $(wildcard $*/run),$*/run map_table)
299
300 4260 aaronmk
# For staging tables which are derived by joining together other staging tables.
301 10994 aaronmk
%/install: %/create.sql _always
302 4527 aaronmk
	($(inDatasrc); echo 'CREATE TABLE "$*" AS'; cat $<; echo ';')|"time" \
303 6186 aaronmk
$(psqlNoSearchPath) --echo-all --set=schema='"$(datasrc)"' --set=table='"$*"' \
304 5912 aaronmk
$(logInstall*)
305 11253 aaronmk
	$(selfMake) "$*/postprocess"
306 4449 aaronmk
	$(exportHeader)
307 8157 aaronmk
	$(cleanup)
308 11253 aaronmk
	$(selfMake) "$*/map_table"
309 4260 aaronmk
310 4252 aaronmk
%/install: _always
311 8380 aaronmk
	$(if $(isCsv),$(import_install_))
312 11253 aaronmk
	$(selfMake) "$*/postprocess"
313 8381 aaronmk
	-$(exportHeader)
314 9843 aaronmk
	-$(if $(isCsv),,$(cleanup))
315 11253 aaronmk
	$(selfMake) "$*/map_table"
316 8381 aaronmk
# ignore errors if table does not exist (non-data dir)
317 5693 aaronmk
define import_install_
318 6950 aaronmk
(. $(bin)/vegbien_dest; unset schemas; "time" nice -n +5\
319 5028 aaronmk
env schema=$(datasrc) table=$* $(bin)/csv2db $(catSrcs) $(logInstall*))
320 5693 aaronmk
$(if $(filter $(srcTable),$*),($(inDatasrc);\
321
echo 'ALTER TABLE "$(datasrc)"."$*" RENAME row_num TO "$*.row_num";')|"time"\
322 6186 aaronmk
$(psqlNoSearchPath) --echo-all --set=table='"$*"' $(logInstall*Add))
323 5693 aaronmk
endef
324 5697 aaronmk
# table-scope src table's row_num col to allow joining it with other tables
325 1921 aaronmk
326 5819 aaronmk
%/uninstall: _always
327 11251 aaronmk
	echo 'DROP $(if\
328
$(is_view),VIEW,TABLE) IF EXISTS "$(datasrc)"."$*" CASCADE;'|$(psqlNoSearchPath)
329 5819 aaronmk
330
%/reinstall: _always %/uninstall %/install ;
331
332 8241 aaronmk
postprocess: _always $(allTables:%=%/postprocess) ;
333 8113 aaronmk
334 5155 aaronmk
cleanup: _always $(tables:%=%/cleanup) ;
335
336 5184 aaronmk
# WARNING: This removes any index comments, due to a PostgreSQL bug.
337
# This occurs because ALTER TABLE recreates the index but not its comment.
338 5155 aaronmk
%/cleanup: _always
339
	$(cleanup)
340
341 1777 aaronmk
##### Maps building
342
343 4646 aaronmk
# Maps to (try to) build are added to this
344 1779 aaronmk
maps :=
345
346 4208 aaronmk
srcRoot = $(mappings)/root.sh
347 8156 aaronmk
mkSrcMap = (. $(srcRoot); env datasrc=$(datasrc) $(bin)/src_map <$*/header.csv\
348
>$@)
349 1779 aaronmk
350 7437 aaronmk
translate = $(if $(wildcard $(1)),$(bin)/in_place $< $(bin)/translate_ci 1 $(1))
351 6074 aaronmk
352 6076 aaronmk
$(srcDict):# empty target in case it doesn't exist
353
354 3572 aaronmk
# Via maps cleanup
355 4182 aaronmk
ifneq ($(filter %/.map.csv.last_cleanup,$(MAKECMDGOALS)),)
356 7484 aaronmk
%/.map.csv.last_cleanup: %/map.csv $(vocab) $(thesaurus) $(coreMap) $(srcDict)
357 7437 aaronmk
	$(call translate,$(srcDict))
358 4844 aaronmk
	$(bin)/in_place $< $(bin)/canon 1 $(vocab)
359 7484 aaronmk
	$(call translate,$(thesaurus))
360 8176 aaronmk
	$(bin)/in_place $< $(bin)/fix_line_endings
361 3572 aaronmk
	touch $@
362 4601 aaronmk
	+$(selfMake) $(<:%/map.csv=%/unmapped_terms.csv)
363
	+$(selfMake) $(<:%/map.csv=%/new_terms.csv)
364 5253 aaronmk
.PRECIOUS: %/.map.csv.last_cleanup
365 3572 aaronmk
else
366 4644 aaronmk
%/map.csv: _always
367 8252 aaronmk
	$(if $(wildcard $@),,$(mk_map_csv))
368 4182 aaronmk
	+$(selfMake) $(@:%/map.csv=%/.map.csv.last_cleanup)
369 8252 aaronmk
define mk_map_csv
370 11253 aaronmk
+$(selfMake) "$*/header.csv"
371 8252 aaronmk
$(if $(nonXml),$(mkSrcMap))
372
endef
373 3572 aaronmk
endif
374
375 4641 aaronmk
%/VegBIEN.csv: %/map.csv $(coreMap)
376 11236 aaronmk
	$(if $(wildcard $*/run),-ln -s "../$(coreMap)" "$@"\
377 10174 aaronmk
,<$< $(bin)/cat_cols 1 2|$(bin)/join $(coreMap)|$(bin)/sort_map >$@)
378
# ignore errors if symlink exists
379 1530 aaronmk
maps += $(autogenMaps)
380
381 4646 aaronmk
maps: $(maps) _always ;
382 1139 aaronmk
383
all += $(maps)
384
385 1742 aaronmk
##### Maps validation
386
387 4663 aaronmk
# `tail -n +2`: Remove header before running filter_out_ci because filter_out_ci
388
# only removes the header if it matches the vocabulary's header.
389 4601 aaronmk
390 4664 aaronmk
%/unmapped_terms.csv: %/map.csv $(coreMap)
391 4858 aaronmk
	tail -n +2 $<|$(bin)/cols 1|$(bin)/filter_out_ci 0 $(coreMap) >$@
392
	$(bin)/autoremove $@
393 4601 aaronmk
394 7484 aaronmk
%/new_terms.csv: %/map.csv $(vocab) $(thesaurus) %/unmapped_terms.csv
395 4857 aaronmk
	$(newTerms)
396
	$(bin)/autoremove $@
397
newTerms = tail -n +2 $<|$(bin)/filter_out_ci 0 $(vocab)|$(bin)/filter_out_ci 0\
398 7484 aaronmk
$(thesaurus) $(if $(wildcard $(word 4,$+)),|$(bin)/filter_out_ci 0 $(word 4,$+))\
399 7216 aaronmk
|grep -vE -e '^"?:' -e 'UNUSED' >$@; exit 0 # grep exits nonzero if no match
400 4601 aaronmk
401 10205 aaronmk
termsSubdirs := $(tables)
402 4600 aaronmk
403 3764 aaronmk
include $(root)/lib/mappings.Makefile
404 1742 aaronmk
405 1509 aaronmk
##### External dependencies
406
407
$(root)/%: _always
408
	+$(subMake)
409 1628 aaronmk
.PRECIOUS: $(root)/% # let ext. dir's Makefile decide whether to delete on error
410 1509 aaronmk
411 783 aaronmk
##### Mapping
412 368 aaronmk
413 4182 aaronmk
+maps = $(filter %/map.csv %/VegBIEN.csv $(mappings)/%,$(+_))
414 4746 aaronmk
map2db = env in_database=vegbien in_schema=$(datasrc) in_table=$*\
415 10863 aaronmk
out_database=vegbien source=$(datasrc).new $(root)/map $(+maps)
416 368 aaronmk
417 783 aaronmk
##### Import to VegBIEN
418 772 aaronmk
419 11285 aaronmk
import_temp: $(importTables:%=%/import) _always ;
420 7281 aaronmk
421 11285 aaronmk
import: _always import_temp publish ; # removes temp suffix when done
422 10869 aaronmk
423 7281 aaronmk
rm: _always
424 10798 aaronmk
	echo "SELECT datasource_rm('$(datasrc)');"|"time" $(psqlAsBien)
425 7281 aaronmk
426 11285 aaronmk
reimport: _always import ; # import replaces the previous import
427 7284 aaronmk
428 1853 aaronmk
profileTest = $(if $(profile),$(if $(test),1))
429
profileOnly = -env profile_to=/dev/fd/3 $(map2db) 3>&1 1>&2|\
430
$(bin)/profile_stats /dev/fd/0
431
432 5457 aaronmk
log_ = $*/logs/$(if $(n),n=$(n).,)$(version).log.sql
433 7075 aaronmk
434
# Displays the import log file path
435 7076 aaronmk
# Run with `make -s` to avoid echoing make commands
436 7075 aaronmk
%/log_file: _always
437
	echo $$(pwd)/$(log_)
438
439 3317 aaronmk
trace = $(log_:.log.sql=.trace)
440 10582 aaronmk
restart_row = $(shell set -x; grep -F Partition: $(log_)|tail -1|$(sed)\
441
's/^.* rows ([[:digit:]]+)-.*$$/\1/')
442 7080 aaronmk
import = -$(if $(profileTest),$(profileOnly),(set -x; date; "time" env commit=1\
443 10582 aaronmk
$(if $(profile),profile_to=$(trace)) $(if $(continue),start=$(restart_row))\
444
$(map2db))$(if $(log), >>$(log_) 2>&1))
445 1088 aaronmk
# don't abort on import errors, which often relate to invalid input data
446 624 aaronmk
447 7245 aaronmk
import? = $(if $(call and,$(full_import),$(call dontImport,.)),,$(import))
448 7047 aaronmk
449 4252 aaronmk
%/import: %/VegBIEN.csv _always
450 7047 aaronmk
	$(import?)
451 772 aaronmk
# default:
452 4252 aaronmk
%/import: _always ;
453 718 aaronmk
454 7279 aaronmk
#### Taxonomic scrubbing
455
456 7282 aaronmk
scrub: _always
457 7275 aaronmk
	$(selfMake) $(root)/scrub & echo $$!
458 10748 aaronmk
# using & (background process) ignores TNRS errors, so that TNRS bugs do not
459
# prevent the remaining tables from being imported even if TNRS can't be run
460 7265 aaronmk
461 7279 aaronmk
# Removes TNRS taxondeterminations
462
unscrub: _always
463
	echo "SELECT delete_scrubbed_taxondeterminations('$(datasrc)');"|\
464
"time" env no_query_results=1 $(psqlAsBien)
465
466 7283 aaronmk
rescrub: _always unscrub scrub ;
467
468 7282 aaronmk
import_scrub: _always import scrub ;
469
470 7285 aaronmk
reimport_scrub: _always reimport scrub ;
471
472 10581 aaronmk
%/import_scrub: _always %/import scrub ;
473
474 790 aaronmk
##### Log files from import
475
476 4182 aaronmk
logs := $(wildcard */logs/*.log.sql */logs/*.trace)
477 339 aaronmk
478 383 aaronmk
rm_logs: _always
479 339 aaronmk
	$(RM) $(logs)
480
481 783 aaronmk
##### Verification of import
482 264 aaronmk
483 5782 aaronmk
verifyTables := $(patsubst verify/%.ref,%,$(wildcard verify/*.ref))
484 1194 aaronmk
485 5782 aaronmk
verify: $(verifyTables:%=%/verify) _always ;
486
487 4252 aaronmk
%/verify: verify/%.ref verify/%.out _always
488 1184 aaronmk
	-$(diffVerbose) $(+_)
489 1088 aaronmk
# don't abort on verification errors, which are expected during development
490 1194 aaronmk
# default:
491 4252 aaronmk
%/verify: verify/%.out _always
492 1200 aaronmk
	$(if $(shell test -e $< && echo t),cat $<)
493 1199 aaronmk
# don't run if verify/%.out's default do-nothing action was used
494 1200 aaronmk
# can't use $(wildcard) because it won't recheck file after verify/%.out is run
495 368 aaronmk
496 6186 aaronmk
psqlExport := "time" $(psqlNoSearchPath) --no-align --field-separator=$$'\t'\
497 4401 aaronmk
--pset=footer=off --pset=null=NULL
498 5976 aaronmk
# Note that using $(inDatasrc) will not work with datasources whose tables are
499
# the same name as VegBIEN tables (likely only VegBank), because the datasource
500
# is first in the search_path.
501
verify = $(if $(reverify),($(inDatasrc); cat $<)|$(psqlExport)\
502
--set=datasource="'$(datasrc)'" >$@)
503 514 aaronmk
504 1199 aaronmk
verify/%.out: $(mappings)/verify.%.sql _always
505 1192 aaronmk
	$(verify)
506 5973 aaronmk
.PRECIOUS: verify/%.out # save partial output in case of error to help debugging
507 1194 aaronmk
# default:
508 1199 aaronmk
verify/%.out: _always ;
509 369 aaronmk
510 1199 aaronmk
all += $(wildcard verify/*.out)
511 1192 aaronmk
512 4402 aaronmk
%.ref: %.ref.sql
513 4409 aaronmk
	($(inDatasrc); cat $<)|$(psqlExport) >$@
514 5981 aaronmk
.PRECIOUS: %.ref # there must always be a .ref for the make rules to work
515 4402 aaronmk
516 1667 aaronmk
##### Editing import
517
518 6984 aaronmk
rename/%: _always
519
	echo "UPDATE source SET shortname = '$*' WHERE shortname = '$(datasrc)';"\
520
|$(psqlAsBien)
521 1667 aaronmk
522 10860 aaronmk
%/publish: _always # usage: make inputs/src/src.version/publish
523
	echo "SELECT datasource_publish('$*');"|$(psqlAsBien)
524
525 10862 aaronmk
publish: _always $(datasrc).new/publish ; # usage: make inputs/src/publish
526
527 783 aaronmk
##### Testing
528 368 aaronmk
529 4274 aaronmk
testRefOutput = $(subst .by_col,,$(1))
530 1986 aaronmk
testRef = $(testRefOutput).ref
531
hasOwnRef = $(filter $@,$(call testRefOutput,$@))
532
# filter returns non-empty if they are equal
533 630 aaronmk
534 4246 aaronmk
# `rm $@`: Remove outputs of successful tests to reduce clutter
535 4272 aaronmk
# `$(foreach use_staged...)`: Run with use_staged=1
536 897 aaronmk
define runTest
537 776 aaronmk
@echo "Testing $(abspath $@)..."
538 4272 aaronmk
>$@ env test=1 n=$(test_n) $(1) $(foreach use_staged,1,$(map2db))
539 6446 aaronmk
$(if $(hasOwnRef),,-)@(set -x; $(diffIgnoreSpace) $(call testRef,$@) $@) 2>&1\
540
&& rm $@ || { e=$$?; $(if $(wildcard $(call testRef,$@)),,cat $@;)\
541 777 aaronmk
$(if $(hasOwnRef),\
542 3179 aaronmk
{\
543
read -p $(emph)'Accept new test output? (y/n)'$(endEmph) REPLY;\
544
if test "$$REPLY" = y; then\
545
(set -x; $(MAKE) $@-ok --directory=$(realpath .) --makefile=../input.Makefile);\
546
exit 0;\
547
fi;\
548
};,\
549 777 aaronmk
echo $(emph)"Note: The preceding failed test is compared to another test's\
550
output"$(endEmph);\
551
echo $(emph)"When it fails, this always indicates a bug"$(endEmph);\
552
)\
553 652 aaronmk
exit $$e;}
554 628 aaronmk
endef
555
556 876 aaronmk
tests :=
557
558 10839 aaronmk
%/test: %/test.xml $(if $(by_col),%/test.by_col.xml) _always ;
559 6744 aaronmk
560 4269 aaronmk
# Requires staging tables. To create them, run `make inputs/<datasrc>/install`.
561
# Non-flat-file inputs fall back to mimicking a successful test
562 4182 aaronmk
%/test.xml: %/VegBIEN.csv _always
563 4452 aaronmk
	$(if $(nonXml),$(call runTest,by_col=))
564 4182 aaronmk
tests += %/test.xml
565 627 aaronmk
566 4274 aaronmk
%/test.by_col.xml: %/VegBIEN.csv _always
567 4831 aaronmk
	$(if $(nonXml),$(call runTest,by_col=1))
568 4275 aaronmk
569
# Only run column-based tests if column-based mode enabled, because these tests
570
# are much slower than the row-based tests for small numbers of rows
571
ifneq ($(by_col),)
572 4274 aaronmk
tests += %/test.by_col.xml
573 4275 aaronmk
endif
574 4274 aaronmk
575 10205 aaronmk
testOutputs := $(foreach test,$(tests),$(tables:%=$(test)))
576 724 aaronmk
577 1361 aaronmk
.PRECIOUS: $(testOutputs) # save outputs of failed tests so they can be accepted
578
579 876 aaronmk
test: _always $(testOutputs) ;
580
581 4182 aaronmk
all += $(wildcard %/test*.xml)
582 876 aaronmk
583 634 aaronmk
# Accepts a test output: make <test_output_path>-ok
584 626 aaronmk
%-ok: _always
585 4383 aaronmk
	mv $* $(call testRef,$*)
586 502 aaronmk
587 4120 aaronmk
accept-all: _always
588 3579 aaronmk
	+yes|$(selfMake) test
589
590 3133 aaronmk
##### Documentation
591
592 4494 aaronmk
steps = $(selfMake) -s $*/import test=1 by_col=1 verbosity=2 n=100\
593 3385 aaronmk
2>&1|$(bin)/debug2redmine >$@
594 3188 aaronmk
595 4182 aaronmk
%/logs/steps.by_col.log.sql: _always
596 3188 aaronmk
	+$(steps)