setup Makefile for guess tests

Makefile

@@ -22,7 +22,8 @@ doc:
 readme:
 	cargo readme --no-title > README.md
 
-valve: valve_debug
+valve: src/*.rs src/*.lalrpop
+	@$(MAKE) valve_debug
 
 valve_release: src/*.rs src/*.lalrpop
 	rm -f valve
@@ -34,7 +35,7 @@ valve_debug: src/*.rs src/*.lalrpop
 	cargo build
 	ln -s target/debug/ontodev_valve valve
 
-build/valve.db: test/src/table.tsv valve clean | build
+build/valve.db: test/src/table.tsv clean valve | build
 	./valve $< $@
 
 test/output:
@@ -116,47 +117,70 @@ pg_random_test: valve clean random_test_data | build test/output
 	@echo "Test succeeded!"
 
 guess_test_dir = test/guess_test_data
+guess_test_db = build/valve_guess.db
+.PHONY: guess_test_data
+
+$(guess_test_dir)/table1.tsv: test/generate_random_test_data.py valve $(guess_test_dir)/*.tsv
+	./$< $$(date +"%s") 50000 5 $(guess_test_dir)/table.tsv $(guess_test_dir)
 
 $(guess_test_dir)/ontology:
 	mkdir -p $@
 
-guess_test_data: test/generate_random_test_data.py valve valve $(guess_test_dir)/table.tsv | $(guess_test_dir)/ontology
-	./$< $$(date +"%s") 10000 5 $(word 3,$^) $|
+guess_test_data: test/generate_random_test_data.py $(guess_test_dir)/table1.tsv valve confirm_overwrite.sh $(guess_test_dir)/*.tsv | $(guess_test_dir)/ontology
+	./confirm_overwrite.sh $(guess_test_dir)/ontology
+	rm -f $(guess_test_dir)/table1.tsv
+	./$< $$(date +"%s") 50000 5 $(guess_test_dir)/table.tsv $(guess_test_dir)
+	rm -f $(guess_test_dir)/ontology/*.tsv
+	./$< $$(date +"%s") 50000 5 $(guess_test_dir)/table_expected.tsv $|
+	rm -f $(guess_test_dir)/ontology/table1.tsv
+
+$(guess_test_db): valve guess_test_data $(guess_test_dir)/*.tsv | build $(guess_test_dir)/ontology
+	rm -f $@
+	./$< $(guess_test_dir)/table.tsv $@
+
+perf_test_dir = test/perf_test_data
+perf_test_db = build/valve_perf.db
+.PHONY: perf_test_data
+
+$(perf_test_dir)/ontology:
+	mkdir -p $@
 
-test/perf_test_data/ontology: test/generate_random_test_data.py valve test/perf_test_data/table.tsv
-	mkdir $@
-	./$< 1 10000 5 $(word 3,$^) $@
+perf_test_data: test/generate_random_test_data.py valve confirm_overwrite.sh $(perf_test_dir)/*.tsv | $(perf_test_dir)/ontology
+	./confirm_overwrite.sh $(perf_test_dir)/ontology
+	rm -f $(perf_test_dir)/ontology/*.tsv
+	./$< $$(date +"%s") 10000 5 $(perf_test_dir)/table.tsv $|
 
-build/valve_perf.db: valve | test/perf_test_data/ontology build
-	@if [ -f $@ ]; \
-	then \
-		echo "'$@' exists but is out of date. To rebuild '$@', run \`make cleanperfdb\`" \
-		"before running \`make $@\`" ; \
-		false; \
-	fi
-	time -p ./$< --verbose test/perf_test_data/table.tsv $@
+$(perf_test_db): valve perf_test_data $(perf_test_dir)/*.tsv | build $(perf_test_dir)/ontology
+	rm -f $@
+	time -p ./$< --verbose $(perf_test_dir)/table.tsv $@
 
 .PHONY: sqlite_perf_test
 sqlite_perf_test: build/valve_perf.db | test/output
 	time -p scripts/export.py messages $< $| $(tables_to_test)
 
 .PHONY: pg_perf_test
-pg_perf_test: valve test/perf_test_data/ontology | test/output
-	time -p ./$< --verbose test/perf_test_data/table.tsv postgresql:///valve_postgres
+pg_perf_test: valve $(perf_test_dir)/ontology | test/output
+	time -p ./$< --verbose $(perf_test_dir)/table.tsv postgresql:///valve_postgres
 	time -p scripts/export.py messages postgresql:///valve_postgres $| $(tables_to_test)
 
 .PHONY: perf_test
 perf_test: sqlite_perf_test pg_perf_test
 
 clean:
-	rm -Rf build/valve.db build/valve_random.db test/output $(random_test_dir)/ontology $(guess_test_dir)/ontology
+	rm -Rf build/valve.db* build/valve_random.db* test/output $(random_test_dir)/ontology valve
+
+clean_guess_db:
+	rm -Rf build/valve_guess.db
 
-cleanperfdb:
+clean_guess_data:
+	rm -Rf $(guess_test_dir)/table1.tsv $(guess_test_dir)/ontology
+
+clean_perf_db:
 	rm -Rf build/valve_perf.db
 
-cleanperfdata:
-	rm -Rf test/perf_test_data/ontology
+clean_perf_data:
+	rm -Rf $(perf_test_dir)/ontology
 
-cleanall: clean cleanperfdb cleanperfdata
+cleanall: clean clean_perf_db clean_perf_data clean_guess_db clean_guess_data
 	cargo clean
-	rm -Rf valve
+	rm -f valve

confirm_overwrite.sh

@@ -0,0 +1,14 @@
+#!/usr/bin/env sh
+
+if [ -d $1 -a ! -z "$(ls -A $1)" ]
+then
+	printf "$1 already exists and contains the following files: $(ls -A -m -w 0 $1)\nAre you sure (y/n)? "
+	read enter
+	if [ $enter = 'y' ]
+	then
+		exit 0
+	else
+    echo "Understood. Exiting with error code."
+		exit 1
+	fi
+fi

test/generate_random_test_data.py

@@ -11,7 +11,7 @@
 
 
 TOKEN_LENGTH = 9
-WINDOW_SIZE = 100
+WINDOW_SIZE = 50
 
 
 def get_special_tables(config):

test/guess_test_data/column.tsv

@@ -31,23 +31,3 @@ table1	prefix			prefix	primary
 table1	base			IRI	unique	
 table1	ontology IRI		empty	IRI		
 table1	version IRI		empty	IRI		
-table2	child			trimmed_line		
-table2	parent		empty	trimmed_line		
-table2	xyzzy		empty	trimmed_line		
-table2	foo		empty	integer		
-table2	bar		empty	text		
-table3	source			prefix		
-table3	id			CURIE	unique	
-table3	label			label	primary	
-table3	parent		empty	label		
-table3	related		empty	trimmed_line		
-table4	foreign_column			text	unique	
-table4	other_foreign_column			text	unique	
-table4	numeric_foreign_column			integer	primary	
-table5	foo			word	primary	
-table5	bar			integer		
-table6	child			integer		
-table6	parent		empty	integer		
-table6	xyzzy		empty	integer		
-table6	foo		empty	text		
-table6	bar		empty	integer		

test/guess_test_data/column_expected.tsv

@@ -0,0 +1,53 @@
+table	column	label	nulltype	datatype	structure	description
+table	table	table_name		table_name	primary	name of this table
+table	path			path		path to the TSV file for this table, relative to the table.tsv file
+table	type		empty	table_type		type of this table, used for tables with special meanings
+table	description		empty	description		a description of this table
+column	table			table_name	from(table.table)	the table that this column belongs to
+column	column			column_name		the name of this column
+column	label		empty	label		the human-readable label for this column
+column	nulltype		empty	datatype_name	from(datatype.datatype)	the datatype for NULL values in this column
+column	datatype			datatype_name	from(datatype.datatype)	the datatype for this column
+column	structure		empty	trimmed_line		schema information for this column
+column	description		empty	description		a description of this column
+datatype	datatype			datatype_name	primary	the name of this datatype
+datatype	parent		empty	datatype_name	tree(datatype)	the parent datatype
+datatype	transform		empty	word		
+datatype	condition		empty	datatype_condition		
+datatype	structure		empty	trimmed_line		
+datatype	description		empty	trimmed_text		
+datatype	SQLite type		empty	datatype_name		
+datatype	PostgreSQL type		empty	datatype_name		
+datatype	RDF type		empty	datatype_name		
+datatype	HTML type		empty	datatype_name		
+rule	table			table_name		
+rule	when column			column_name		
+rule	when condition			datatype_condition		
+rule	then column			column_name		
+rule	then condition			datatype_condition		
+rule	level			word		
+rule	description		empty	description		
+table1	prefix			prefix	primary	
+table1	base			IRI	unique	
+table1	ontology IRI		empty	IRI		
+table1	version IRI		empty	IRI		
+table2	child			trimmed_line	from(table4.other_foreign_column)	
+table2	parent		empty	trimmed_line	tree(child)	
+table2	xyzzy		empty	trimmed_line	under(table2.child, d)	
+table2	foo		empty	integer	from(table4.numeric_foreign_column)	
+table2	bar		empty	text		
+table3	source			prefix	from(table1.prefix)	
+table3	id			CURIE	unique	
+table3	label			label	primary	
+table3	parent		empty	label	tree(label)	
+table3	related		empty	trimmed_line		
+table4	foreign_column			text	unique	
+table4	other_foreign_column			text	unique	
+table4	numeric_foreign_column			integer	primary	
+table5	foo			word	primary	
+table5	bar			integer		
+table6	child			integer	from(table4.numeric_foreign_column)	
+table6	parent		empty	integer	tree(child)	
+table6	xyzzy		empty	integer	under(table6.child, 4)	
+table6	foo		empty	text		
+table6	bar		empty	integer		

test/guess_test_data/rule.tsv

@@ -1,9 +1,2 @@
 table	when column	when condition	then column	then condition	level	description
-table2	foo	null	bar	null	error	bar must be null whenever foo is null
-table2	foo	not null	bar	not null	error	bar cannot be null if foo is not null
-table2	foo	IRI	bar	label	error	bar must be a label if foo is an IRI
-table2	foo	equals(5)	bar	in('y', 'z')	error	bar must be 'y' or 'z' if foo = 5
-table6	foo	null	bar	null	error	bar must be null whenever foo is null
-table6	foo	not null	bar	not null	error	bar cannot be null if foo is not null
-table6	foo	IRI	bar	label	error	bar must be a label if foo is an IRI
-table6	foo	equals(e)	bar	in(25, 26)	error	bar must be 25 or 26 if foo = 'e'
+table1	ontology IRI	null	version IRI	null	error	'version IRI' must be null whenever 'ontology IRI' is null

test/guess_test_data/table.tsv

@@ -3,9 +3,4 @@ column	test/guess_test_data/column.tsv	Columns for all of the tables.	column
 datatype	test/guess_test_data/datatype.tsv	Datatypes for all of the columns	datatype
 rule	test/guess_test_data/rule.tsv	More complex "when" rules	rule
 table	test/guess_test_data/table.tsv	All of the user-editable tables in this project.	table
-table1	test/guess_test_data/ontology/table1.tsv	The first data table	
-table2	test/guess_test_data/ontology/table2.tsv	The second data table	
-table3	test/guess_test_data/ontology/table3.tsv	The third data table	
-table4	test/guess_test_data/ontology/table4.tsv	The fourth data table	
-table5	test/guess_test_data/ontology/table5.tsv	The fifth data table	
-table6	test/guess_test_data/ontology/table6.tsv	The sixth data table (like table2 but all numeric)	
+table1	test/guess_test_data/table1.tsv	The first data table	

test/guess_test_data/table_expected.tsv

@@ -0,0 +1,11 @@
+table	path	description	type
+column	test/guess_test_data/column_expected.tsv	Columns for all of the tables.	column
+datatype	test/guess_test_data/datatype.tsv	Datatypes for all of the columns	datatype
+rule	test/guess_test_data/rule.tsv	More complex "when" rules	rule
+table	test/guess_test_data/table_expected.tsv	All of the user-editable tables in this project.	table
+table1	test/guess_test_data/table1.tsv	The first data table	
+table2	test/guess_test_data/ontology/table2.tsv	The second data table	
+table3	test/guess_test_data/ontology/table3.tsv	The third data table	
+table4	test/guess_test_data/ontology/table4.tsv	The fourth data table	
+table5	test/guess_test_data/ontology/table5.tsv	The fifth data table	
+table6	test/guess_test_data/ontology/table6.tsv	The sixth data table (like table2 but all numeric)