##### Configuration

# Command line

import_source ?= 1

log ?= $(if $(test),,1)

profile ?=

quiet ?=

reverify ?= 1

schema_only ?=

use_staged ?= $(by_col)

exts ?= csv tsv txt dmp xml

test_n ?= 2

# Paths

datasrc := $(patsubst .%,%,$(notdir $(realpath .)))

bin := $(root)/bin

# Make

SHELL := /bin/bash

selfMake = $(MAKE) --makefile=../input.Makefile

subMake = $(MAKE) $(@:$(root)/%=%) --directory=$(root)

+_ = $(+:_%=)

addBeforeExt = $(basename $(2))$(1)$(suffix $(2))

# Terminal

termCols := $(shell tput cols)

esc := '['

# User interaction

# Commands

MKDIR = mkdir -p

CP = cp -p

diff = diff --unified=2

diffIgnoreSpace = $(diff) --ignore-space-change

diffVerbose = $(if $(verbose),diff --side-by-side --left-column\

# BIEN commands

sortFilenames = $(shell $(bin)/sort_filenames $(1))

selfMap = $(bin)/cols 0 0

psqlAsBien := $(bin)/psql_script_vegbien

psqlNoSearchPath := env no_search_path=1 $(psqlAsBien)

# Usage: ($(inDatasrc); cat $(file))|$(psqlCmd)

inDatasrc := echo 'SET search_path TO "$(datasrc)";'

# SVN

setSvnIgnore = svn propset svn:ignore $(2) $(1)

define addDirWithIgnore

$(setSvnIgnore)

endef

##### Environment

##### General targets

all: _always maps ;

clean: _always

	$(RM) $(all)

remake: _always clean

# Only remake if doesn't exist. This prevents unintentional remaking when the

%/: % _always ;

%: %.make

	$(make_script)

.PRECIOUS: % # save partial outputs of aborted src make scripts

##### Tables discovery

sortFile := import_order.txt

noImportFile := _no_import

dontImport = $(wildcard $(1)/$(noImportFile))$(if\

svnFilesGlob:= */{,$(noImportFile),{,.}{map,*terms,VegBIEN}.csv{,.*},*header.*,*.sql,test.xml.ref}

svnFilesGlob := {map.csv,*{schema,~}*.sql,{,*/}*.make,$(svnFilesGlob)}

_svnFilesGlob := {_MySQL/{,*schema*.sql,*.make},_src/*.{url,pdf}}

svnFiles = $(filter-out _% logs/%,$(call wildcard/,$(svnFilesGlob)))\

add: _always Source/add Source/map.csv $(allTables:%=%/add)

	$(call setSvnIgnore,.,'*')

	$(call addDirWithIgnore,logs,$$'*.log.sql\n*.trace')

	$(call addDirWithIgnore,verify,$$'*.out\n*.csv\n*.xls')

	$(call addFile,import_order.txt)

	$(if $(wildcard _MySQL/),$(call addDirWithIgnore,_MySQL,'*'))

	$(if $(wildcard _src/),$(call addDirWithIgnore,_src,'*'))

	$(if $(wildcard _archive/),$(call addDirWithIgnore,_archive,'*'))

	$(call add*,$(svnFiles))

# Adds a new table subdir

##### Existing maps discovery

anyMap := %/map.csv %/VegBIEN.csv %/unmapped_terms.csv %/new_terms.csv

extsFilter := $(addprefix %.,$(exts))

anyTest = $*/test.%

srcsOnly = $(filter-out $(anyMap) $(anyTest) %/logs,$(call dataOnly,$(1)))

srcDict := map.csv

vocab := $(mappings)/VegCore.csv

coreMap := $(mappings)/VegCore-VegBIEN.csv

dict := $(mappings)/Veg+-VegCore.csv

viaMaps := $(importTables:%=%/map.csv)

autogenMaps := $(subst map.,VegBIEN.,$(viaMaps))

directMaps := $(autogenMaps) $(filter-out $(autogenMaps),\

$(wildcard */VegBIEN.csv))

##### Sources

srcs = $(call sortFilenames,$(call srcsOnly,$(wildcard $*/*)))

nonHeaderSrcs = $(filter-out %/header.csv,$(srcs))

isXml = $(filter %.xml,$(nonHeaderSrcs))

    # true if $(srcs) non-empty and contains no *.xml

catSrcs = $(bin)/cat$(if $(nonXml),_csv) $(srcs)

withCatSrcs = $(catSrcs:$(bin)/%=$(bin)/with_%) --

# Usage: `make {--silent|-s} inputs/<datasrc>/cat` (don't echo make commands)

cat: $(importTables:%=%/cat) _always ;

%/cat: _always

	$(catSrcs)

##### Input data retrieval

##### Staging tables installation

srcTable := %.src

dbExportsWildcard = $(sort $(patsubst _MySQL/%.make,%,$(wildcard\

endif

dbExports := $(strip $(dbExports))# += adds extra whitespace

allInstalls := $(if $(dbExports),sql) $(filter-out Source,$(allTables))

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

uninstall: _always confirm_rm_schema rm_schema ;

# rm_schema will also drop all staging tables

confirm_rm_schema: _always

schema: _always

	-echo 'CREATE SCHEMA "$(datasrc)";'|$(psqlNoSearchPath)

# ignore errors if schema exists

rm_schema: _always

	echo 'DROP SCHEMA IF EXISTS "$(datasrc)" CASCADE;'|$(psqlNoSearchPath)

installLog := logs/install.log.sql

logInstall = $(if $(log),$(if $(quiet),$(2)$(1)$(installLog) 2>&1,2>&1|tee $(3)\

logInstallRoot = $(call logInstall,,>)

logInstall* = $(call logInstall,$*/,>)

# Must come before %/install to override it

sql/install: $(dbExports)

	($(inDatasrc); cat $+|pg_dump_limit)|"time" $(psqlNoSearchPath) \

--set=schema='"$(datasrc)"' $(logInstallRoot)

cleanup = $(if $(wildcard $*/cleanup.sql),($(inDatasrc); cat $*/cleanup.sql)\

|"time" $(psqlNoSearchPath) --echo-all --set=table='"$*"' $(logInstall*Add),\

echo 'SELECT * FROM "$(datasrc)"."$*" LIMIT 0;'|$(psqlNoSearchPath) \

--no-align --field-separator=, --pset=footer=off >$*/header.csv

# For staging tables which are derived by joining together other staging tables.

%/install %/header.csv: %/create.sql _always

	($(inDatasrc); echo 'CREATE TABLE "$*" AS'; cat $<; echo ';')|"time" \

$(psqlNoSearchPath) --echo-all --set=schema='"$(datasrc)"' --set=table='"$*"' \

$(logInstall*)

	$(exportHeader)

.PRECIOUS: %/header.csv

%/install: _always

	$(if $(isRef),$(exportHeader),$(if $(nonXml),$(import_install_)))

	$(if $(wildcard $*/postprocess.sql),($(inDatasrc); cat $*/postprocess.sql;)\

|"time" $(psqlNoSearchPath) --echo-all --set=table='"$*"' $(logInstall*Add))

define import_install_

(prefix=; . $(bin)/vegbien_dest; unset schemas; "time" nice -n +5\

env schema=$(datasrc) table=$* $(bin)/csv2db $(catSrcs) $(logInstall*))

$(if $(filter $(srcTable),$*),($(inDatasrc);\

$(psqlNoSearchPath) --echo-all --set=table='"$*"' $(logInstall*Add))

endef

# table-scope src table's row_num col to allow joining it with other tables

%/uninstall: _always

	echo 'DROP TABLE IF EXISTS "$(datasrc)"."$*" CASCADE;'|$(psqlNoSearchPath)

cleanup: _always $(tables:%=%/cleanup) ;

# WARNING: This removes any index comments, due to a PostgreSQL bug.

%/cleanup: _always

##### Maps building

# WARNING: You CANNOT make a subdir using `make inputs/<datasrc>/<subdir>/`.

# Maps to (try to) build are added to this

maps :=

srcRoot = $(mappings)/root.sh

mkSrcMap = $(catSrcs)|(. $(srcRoot); env datasrc=$(datasrc) $(bin)/src_map >$@)

define translate

$(srcDict):# empty target in case it doesn't exist

# Via maps cleanup

ifneq ($(filter %/.map.csv.last_cleanup,$(MAKECMDGOALS)),)

%/.map.csv.last_cleanup: %/map.csv $(vocab) $(dict) $(coreMap) $(srcDict)

	$(call translate?,$(srcDict))

	$(bin)/in_place $< $(bin)/canon 1 $(vocab)

	$(call translate?,$(dict))

	touch $@

	+$(selfMake) $(<:%/map.csv=%/unmapped_terms.csv)

.PRECIOUS: %/.map.csv.last_cleanup

else

%/map.csv: _always

	+$(selfMake) $(@:%/map.csv=%/.map.csv.last_cleanup)

.PRECIOUS: %/map.csv

endif

%/VegBIEN.csv: %/map.csv $(coreMap)

	<$< $(bin)/cat_cols 1 2|$(bin)/join $(coreMap)|$(bin)/sort_map >$@

maps += $(autogenMaps)

maps: $(maps) _always ;

##### Maps validation

# `tail -n +2`: Remove header before running filter_out_ci because filter_out_ci

%/unmapped_terms.csv: %/map.csv $(coreMap)

	tail -n +2 $<|$(bin)/cols 1|$(bin)/filter_out_ci 0 $(coreMap) >$@

%/new_terms.csv: %/map.csv $(vocab) $(dict) %/unmapped_terms.csv

	$(newTerms)

$(dict) $(if $(wildcard $(word 4,$+)),|$(bin)/filter_out_ci 0 $(word 4,$+))\

|grep -vE '^"?:' >$@; exit 0# because grep exits nonzero if no match

termsSubdirs := $(importTables)

include $(root)/lib/mappings.Makefile

##### External dependencies

.PRECIOUS: $(root)/% # let ext. dir's Makefile decide whether to delete on error

##### Mapping

+maps = $(filter %/map.csv %/VegBIEN.csv $(mappings)/%,$(+_))

map2db = env in_database=vegbien in_schema=$(datasrc) in_table=$*\

out_database=vegbien $(root)/map $(+maps)

##### Import to VegBIEN

profileTest = $(if $(profile),$(if $(test),1))

log_ = $*/logs/$(if $(n),n=$(n).,)$(version).log.sql

trace = $(log_:.log.sql=.trace)

import = -$(if $(profileTest),$(profileOnly),(set -x; "time" env commit=1\

$(if $(profile),profile_to=$(trace)) $(map2db)) $(if $(log),\

$(if $(n),,&>$(log_)))$(if $(log),$(if $(n), 2>&1|tee -a $(log_))))

# don't abort on import errors, which often relate to invalid input data

import: $(importTables:%=%/import) _always ;

%/import: %/VegBIEN.csv _always

	$(import)

# default:

%/import: _always ;

##### Log files from import

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

rm_logs: _always

	$(RM) $(logs)

##### Verification of import

verifyTables := $(patsubst verify/%.ref,%,$(wildcard verify/*.ref))

verify: $(verifyTables:%=%/verify) _always ;

%/verify: verify/%.ref verify/%.out _always

	-$(diffVerbose) $(+_)

# don't abort on verification errors, which are expected during development

# default:

%/verify: verify/%.out _always

	$(if $(shell test -e $< && echo t),cat $<)

# don't run if verify/%.out's default do-nothing action was used

# can't use $(wildcard) because it won't recheck file after verify/%.out is run

psqlExport := "time" $(psqlNoSearchPath) --no-align --field-separator=$$'\t'\

--pset=footer=off --pset=null=NULL

# Note that using $(inDatasrc) will not work with datasources whose tables are

verify/%.out: $(mappings)/verify.%.sql _always

	$(verify)

.PRECIOUS: verify/%.out # save partial output in case of error to help debugging

# default:

verify/%.out: _always ;

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

%.ref: %.ref.sql

	($(inDatasrc); cat $<)|$(psqlExport) >$@

.PRECIOUS: %.ref # there must always be a .ref for the make rules to work

##### Editing import

rotate: _always

	echo "UPDATE source SET shortname = shortname||'.$(version)'\

rm: _always

	echo "DELETE FROM source WHERE shortname = '$(datasrc)';"|$(psqlAsBien)

##### Testing

testRefOutput = $(subst .by_col,,$(1))

testRef = $(testRefOutput).ref

# `rm $@`: Remove outputs of successful tests to reduce clutter

# `$(foreach use_staged...)`: Run with use_staged=1

define runTest

@echo "Testing $(abspath $@)..."

>$@ env test=1 n=$(test_n) $(1) $(foreach use_staged,1,$(map2db))

$(if $(hasOwnRef),,-)@(set -x; $(diffIgnoreSpace) $(call testRef,$@) $@) 2>&1\

$(if $(hasOwnRef),\

{\

echo $(emph)"Note: The preceding failed test is compared to another test's\

exit $$e;}

endef

tests :=

# Requires staging tables. To create them, run `make inputs/<datasrc>/install`.

%/test.xml: %/VegBIEN.csv _always

	$(if $(nonXml),$(call runTest,by_col=))

tests += %/test.xml

%/test.by_col.xml: %/VegBIEN.csv _always

	$(if $(nonXml),$(call runTest,by_col=1))

tests += %/test.by_col.xml

endif

testOutputs := $(foreach test,$(tests),$(importTables:%=$(test)))

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

test: _always $(testOutputs) ;

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

# Accepts a test output: make <test_output_path>-ok

%-ok: _always

	mv $* $(call testRef,$*)

accept-all: _always

	+yes|$(selfMake) test

##### Documentation

steps = $(selfMake) -s $*/import test=1 by_col=1 verbosity=2 n=100\

2>&1|$(bin)/debug2redmine >$@

%/logs/steps.by_col.log.sql: _always

	+$(steps)