selfDir_uZPPqC := $(dir $(lastword $(MAKEFILE_LIST)))

root := $(selfDir_uZPPqC)..

include $(root)/lib/common.Makefile

##### Configuration

# Command line

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

profile ?=

quiet ?=

reverify ?= 1

use_staged ?= $(by_col)

# Makefile

exts ?= csv tsv txt xml

test_n ?= 2

##### Vars/functions

# Paths

datasrc := $(notdir $(realpath .))

bin := $(root)/bin

mappings := $(root)/mappings

# Make

SHELL := /bin/bash

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

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

+_ = $(+:_%=)

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

# OS

os := $(shell uname)

forOs = $(patsubst %,%-$(filter Linux Darwin,$(os)),$(1))

# Formatting

SED = sed -$(if $(filter Darwin,$(os)),E,r)

# System

date = $(shell date +"%Y-%m-%d-%H-%M-%S")

# Terminal

termCols := $(shell tput cols)

esc := '['

reset := $(esc)'0m'

emph := $(esc)'7m '

endEmph := ' '$(reset)

# Commands

MKDIR = mkdir -p

mkdir = $(MKDIR) $(@D)

CP = cp -p

diff = diff --unified=2

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

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

--width=$(termCols),$(diff))

# BIEN commands

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

selfMap = $(bin)/cols 0 0

psqlOpts := --set ON_ERROR_STOP=1 --quiet

psqlAsBien := $(bin)/psql_vegbien $(psqlOpts)

# SVN

addDir = $(if $(wildcard $(1)/),svn add --depth=empty $(1),svn mkdir $(1))

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

define addDirWithIgnore

$(addDir)

$(setSvnIgnore)

endef

##### General targets

all: _always maps ;

clean: _always

	$(RM) $(all)

remake: _always clean

	+$(selfMake)

# re-run make so that cache of existing files is reset

make_script = ./$< >$@

%/: % _always ;

# Must come before `%: %.make` to override it

%: %.make _always

	(set -x; $(make_script)) 2>>$<.log.sql

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

%: %.make _always

	$(make_script)

##### SVN

add: _always

	$(call setSvnIgnore,.,'*')

	$(call addDirWithIgnore,specimens,'*')

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

	$(call addDirWithIgnore,verify,'*.out')

##### Existing maps discovery

sortFile := import_order.txt

ifneq ($(wildcard $(sortFile)),)

tables := $(shell cat $(sortFile))# $(shell) replaces "\n" with " "

else

allSubdirs := $(call wildcard/,*/)

tables := $(call sortFilenames,$(filter-out _% src verify,$(allSubdirs:%/=%)))

endif

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

srcMap := %/src.csv

viaMap := %/map.csv

fullViaMap := %/map.full.csv

directMap := %/VegBIEN.csv

anyMap := $(srcMap) $(viaMap) $(fullViaMap) $(directMap)

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

dataOnly = $(filter $(extsFilter),$(1))

anyTest = $*/test.%

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

via := Veg+

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

coreSelfMap := $(mappings)/$(via).self.csv

viaMaps := $(wildcard $(tables:%=%/map.csv))

viaMaps += $(filter-out $(viaMaps),$(srcMaps:%/src.csv=%/map.csv))

viaMaps += $(filter-out $(viaMaps) $(fullViaMap),$(wildcard */map.csv))

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

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

$(wildcard */VegBIEN.csv))

##### Sources

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

isCsv = $(if $(srcs),$(if $(filter %.xml,$(srcs)),,1))

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

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

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

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

cat: $(addprefix cat-,$(tables)) _always ;

cat-%: _always

	$(catSrcs)

##### Staging tables installation

install: _always schema $(addprefix install-,$(tables)) ;

uninstall: _always rm_schema ;

# rm_schema will also drop all staging tables

reinstall: _always uninstall install ;

schema: _always

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

# ignore errors if schema exists

rm_schema: _always

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

install-%: _always

	$(if $(isCsv),$(import_install_))

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

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

$(if $(log),$(if $(quiet),&>,2>&1|tee )$*/logs/install.log.sql))

##### Maps building

# Maps to (try to) build are added to one of these

maps :=

createOnlyMaps :=

srcRoot = $(mappings)/root.sh

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

# Src maps cleanup

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

%/.src.csv.last_cleanup: %/src.csv

	$(bin)/in_place $< $(bin)/cols +

	touch $@

else

# Autogen src maps with known table names

%/src.csv: _always

	$(if $(wildcard $@),,$(if $(isCsv),$(mkSrcMap)))

	+$(if $(isCsv),$(selfMake) $(@:%/src.csv=%/.src.csv.last_cleanup))

# only build if doesn't exist

# only build if CSV srcs exist for that table name

endif

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

# Must come before $(root)/% to override it

$(coreSelfMap): _always

	-+$(subMake)

# ignore errors if $(coreSelfMap) does not exist

# Via maps cleanup

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

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

	$(bin)/in_place $< env ignore=1 $(bin)/subtract $(word 2,$+) 0 1 2

	touch $@

# Include comment column so commented mappings are never removed

else

%/map.csv: %/src.csv _always

	$(if $(wildcard $@),,$(if $(wildcard $<),$(CP) $< $@))

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

endif

srcMap* = $*/src.csv

joinSrcMap = $(if $(hasSrc),$(bin)/in_place $@ $(bin)/intersect $(srcMap*) 0)

hasSrc = $(shell test -s $(word 3,$+) && echo t)

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

%/map.full.csv: %/map.csv $(coreSelfMap)

	$(makeFullCsv)

	$(joinSrcMap)

maps += $(patsubst %/map.csv,%/map.full.csv,$(viaMaps))

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

	$(bin)/join <$+|$(bin)/sort_map >$@

maps += $(autogenMaps)

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

all += $(maps)

##### Maps validation

missingMappingsCmd = +$(selfMake) remake 2>&1\

|$(SED) -n 's/^.*No $(*2Space) mapping for (.*)$$/\1/p'|$(SED) 's/\/_.*//'\

$(if $(filter non-empty_join,$*),|$(bin)/ucase_first 0)|sort|uniq

include $(root)/lib/mappings.Makefile

##### External dependencies

$(root)/%: _always

	+$(subMake)

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

##### Mapping

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

map = $(if $(srcs),$(withCatSrcs) $(root)/map $(+maps),\

$(shell echo Warning: No input file $*/*.{$(exts)} >&2)false)

# need false to run some command, when prefixed by `env ...` below

map2db = $(if $(use_staged),env in_database=vegbien in_schema=$(datasrc)\

in_table=$*) env out_database=vegbien $(map)

##### Import to VegBIEN

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

profileOnly = -env profile_to=/dev/fd/3 $(map2db) 3>&1 1>&2|\

$(bin)/profile_stats /dev/fd/0

log_ = $*/logs/$(if $(n),n=$(n).,)$(date).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: $(addprefix import-,$(tables)) _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

verify: $(addprefix verify-,$(tables)) _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

verify = $(if $(reverify),"time" $(psqlAsBien) --set=datasource="'$(datasrc)'"\

--no-align --field-separator=$$'\t' --pset=footer=off --pset=null=NULL <$< >$@)

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

	$(verify)

# default:

verify/%.out: _always ;

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

##### Editing import

rotate: _always

	echo "UPDATE party SET organizationname = organizationname||'.$(date)'\

WHERE organizationname = '$(datasrc)';"|$(psqlAsBien)

rm: _always

	echo "DELETE FROM party WHERE organizationname = '$(datasrc)';"|\

$(psqlAsBien)

##### Testing

testRefOutput = $(subst .2-step,,$(subst .staging,,$(1)))

testRef = $(testRefOutput).ref

hasOwnRef = $(filter $@,$(call testRefOutput,$@))

# filter returns non-empty if they are equal

define runTest

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

>$@ env test=1 n=$(test_n) $(1)

@(set -x; $(diffIgnoreSpace) $(call testRef,$@) $@) 2>&1 || { e=$$?;\

$(if $(wildcard $(call testRef,$@)),,cat $@;)\

$(if $(hasOwnRef),\

{\

read -p $(emph)'Accept new test output? (y/n)'$(endEmph) REPLY;\

if test "$$REPLY" = y; then\

(set -x; $(MAKE) $@-ok --directory=$(realpath .) --makefile=../input.Makefile);\

exit 0;\

fi;\

};,\

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

output"$(endEmph);\

echo $(emph)"When it fails, this always indicates a bug"$(endEmph);\

)\

exit $$e;}

endef

test2Db = $(call runTest,$(map2db))

tests :=

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

	$(test2Db)

tests += %/test.xml

testStaged2Db = $(foreach use_staged,1,$(test2Db))

    # run with use_staged=1

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

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

	-$(if $(isCsv),$(testStaged2Db),cp -p $(@:.staging.xml=.xml) $@)

# Don't abort tester if only staging test fails, in case staging table missing

# Non-flat-file inputs fall back to mimicking a successful test

tests += %/test.staging.xml

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

	$(CP) $* $(call testRef,$*)

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)