:- module(test_eorscope_employer_costs, []). :- use_module(library(plunit)). :- use_module(library(apply)). :- use_module(library(lists)). :- use_module(library(pairs)). :- use_module(library(readutil)). :- use_module(library(sha)). :- use_module('../prolog/eorscope_employer_costs'). % Run with: swipl -g run_tests -t halt test/test_eorscope_employer_costs.pl :- prolog_load_context(directory, Dir), atomic_list_concat([Dir, '/../data'], Data), asserta(data_dir(Data)). data_file(Name, Path) :- data_dir(Dir), atomic_list_concat([Dir, '/', Name], Path). raw(Name, Text) :- data_file(Name, Path), read_file_to_string(Path, Text, [encoding(utf8)]). % Records in a CSV text counted independently of the parser: line breaks % outside quotes, minus the header. count_records(Text, N) :- string_codes(Text, Cs), foldl(count_lf, Cs, 0-false, N0-_), ( last(Cs, 0'\n) -> N1 = N0 ; N1 is N0 + 1 ), N is N1 - 1. count_lf(0'", N-Q0, N-Q) :- !, ( Q0 == true -> Q = false ; Q = true ). count_lf(0'\n, N0-false, N-false) :- !, N is N0 + 1. count_lf(_, S, S). iso_code(S, Len) :- string(S), string_length(S, Len), string_upper(S, S), string_codes(S, Cs), forall(member(C, Cs), code_type(C, upper)). :- begin_tests(eorscope_data_files). % The embedded files are the 2026-10-13 export, byte for byte. test(sha256, [forall(member(File-Hex, [ 'country_summary.csv'-'83e1cec0a902ff629be224e22e0c4c11b90adc2f77aaa1e2b2914a79e6075254', 'country_contributions.csv'-'5c55cd63e18a9b8e60813653b600cbe1f0e1bafb560482c949ea5c225900ec85', 'country_assumptions.csv'-'c204bf448d5aecc99968a5e37f71473233cc179b10238d18537e46eb573fd50a', 'LICENSE'-'105ff773cdd84d09900a475230ad58a7207718ac9d95b7ec09176a4cb044de01' ])), Got == Hex]) :- data_file(File, Path), read_file_to_codes(Path, Bytes, [type(binary)]), sha_hash(Bytes, Hash, [algorithm(sha256), encoding(octet)]), hash_atom(Hash, Got). :- end_tests(eorscope_data_files). :- begin_tests(eorscope_csv_parser). % Quotes, commas, doubled quotes, line breaks, CRLF, BOM, blank lines. test(vector) :- eorscope_employer_costs:parse_csv( "\uFEFFa,b,c\r\n\"x, y\",\"say \"\"hi\"\"\",\"l1\nl2\"\r\n\n,,\"\"\n1,2,3", t, R), assertion(R == [ ["a", "b", "c"], ["x, y", "say \"hi\"", "l1\nl2"], ["", "", ""], ["1", "2", "3"] ]). test(field_count, [error(syntax_error(csv(t, _)))]) :- eorscope_employer_costs:parse_csv("a,b\n1\n", t, _). test(after_quote, [error(syntax_error(csv(t, _)))]) :- eorscope_employer_costs:parse_csv("a\n\"x\"y\n", t, _). test(bare_quote, [error(syntax_error(csv(t, _)))]) :- eorscope_employer_costs:parse_csv("a\nx\"y\n", t, _). test(unterminated, [error(syntax_error(csv(t, _)))]) :- eorscope_employer_costs:parse_csv("a\n\"open\n", t, _). :- end_tests(eorscope_csv_parser). :- begin_tests(eorscope_countries). % 76 countries, valid codes, case-insensitive lookup. test(countries) :- countries(Cs), length(Cs, N), assertion(N == 76), raw('country_summary.csv', Text), count_records(Text, N2), assertion(N2 == N), maplist([C, I]>>get_dict(iso, C, I), Cs, Isos), assertion(is_set(Isos)), assertion(msort(Isos, Isos)), forall(member(C, Cs), check_country(C)), assertion(\+ country(xx, _)), assertion((country(de, DE), get_dict(iso, DE, "DE"))), assertion((country("de", DE2), get_dict(iso, DE2, "DE"))). check_country(C) :- _{iso:I, name:Name, region:Region, currency:Cur, example_salary_usd:Sal, fx_local_per_usd:FX, employer_cost_pct:Pct, employer_cost_usd_month:USD, last_reviewed:Rev, fx_date:FXD, basic_share_of_gross:Basic} :< C, assertion(iso_code(I, 2)), assertion((Name \== "", Region \== "", iso_code(Cur, 3))), assertion((integer(Sal), Sal > 0, FX > 0, Pct >= 0, USD >= 0)), assertion(Rev = date(_, _, _)), assertion((Cur == "USD" ; FXD = date(_, _, _))), ( Basic == null -> true ; assertion((Basic > 0, Basic =< 1)) ), assertion((country(I, C2), C2 == C)). % Mexico as printed on eorscope.com. test(mexico_printed) :- country("MX", MX), assertion(get_dict(employer_cost_pct_printed, MX, 30.9)). % Every country equals its row of country_summary.csv, split here without % the parser (the file has no quoted field). test(summary_rows) :- raw('country_summary.csv', Text), assertion(\+ sub_string(Text, _, _, _, "\"")), split_string(Text, "\n", "\r", Lines0), exclude(==(""), Lines0, [HeadLine|Lines]), split_string(HeadLine, ",", "", Head), forall(member(Line, Lines), check_row(Head, Line)). check_row(Head, Line) :- split_string(Line, ",", "", Fs), assertion(same_length(Fs, Head)), pairs_keys_values(V, Head, Fs), memberchk("iso"-Iso, V), country(Iso, C), forall(member(Col-Key, [ "country"-name, "region"-region, "currency"-currency, "example_role"-example_role, "example_place"-example_place, "total_excludes"-total_excludes ]), ( memberchk(Col-S, V), get_dict(Key, C, S2), assertion(S2 == S) )), memberchk("example_salary_usd"-SalS, V), assertion((number_string(Sal, SalS), get_dict(example_salary_usd, C, Sal))), forall(member(Col-Key, [ "basic_share_of_gross"-basic_share_of_gross, "employer_cost_pct"-employer_cost_pct, "employer_cost_pct_printed"-employer_cost_pct_printed, "employer_cost_usd_month"-employer_cost_usd_month, "fx_local_per_usd"-fx_local_per_usd ]), ( memberchk(Col-S, V), get_dict(Key, C, X), assertion(same_number(S, X)) )), forall(member(Col-Key, [ "total_declared_floor"-total_declared_floor, "total_declared_ceiling"-total_declared_ceiling ]), ( memberchk(Col-S, V), get_dict(Key, C, B), ( S == "true" -> B2 = true ; B2 = false ), assertion(B == B2) )), forall(member(Col-Key, [ "fx_date"-fx_date, "last_reviewed"-last_reviewed ]), ( memberchk(Col-S, V), get_dict(Key, C, D), assertion(same_day(S, D)) )). same_number("", null) :- !. same_number(S, X) :- number_string(N, S), X =:= N. same_day("", null) :- !. same_day(S, date(Y, M, D)) :- format(string(S2), "~d-~|~`0t~d~2+-~|~`0t~d~2+", [Y, M, D]), S2 == S. % Floor countries: an "at least" figure, flagged on the country and its lines. test(floor) :- country("AE", AE), assertion(get_dict(total_declared_floor, AE, true)), assertion(get_dict(total_declared_ceiling, AE, false)), contributions("AE", Ls), forall(member(L, Ls), assertion(get_dict(total_declared_floor, L, true))), country("DE", DE), assertion(get_dict(total_declared_floor, DE, false)), countries(Cs), include([C]>>get_dict(total_declared_floor, C, true), Cs, Floors), length(Floors, NF), assertion(NF == 13). % Regions cover every country once. test(regions) :- regions(Rs), foldl([R, N0, N]>>( countries_in_region(R, Cs), assertion(Cs \== []), forall(member(C, Cs), assertion(get_dict(region, C, R))), length(Cs, K), N is N0 + K ), Rs, 0, Total), countries(All), length(All, Total), countries_in_region("europe", E1), countries_in_region('Europe', E2), assertion(E1 == E2). :- end_tests(eorscope_countries). :- begin_tests(eorscope_contributions). % Contribution lines, base factors (Mexico's integrated base). test(lines) :- countries(Cs), foldl(check_lines, Cs, 0-[], Total-Factors), assertion(Total == 544), raw('country_contributions.csv', Text), count_records(Text, N2), assertion(N2 == Total), msort(Factors, Sorted), clumped(Sorted, Counts), assertion(Counts == [1.009-7, 1.049315-9, 1.049867-1, 1.08-4]), contributions(xx, None), assertion(None == []), contributions(mx, Mx1), contributions('MX', Mx2), length(Mx1, N3), length(Mx2, N4), assertion((N3 == N4, N3 > 0)). check_lines(C, T0-F0, T-F) :- get_dict(iso, C, Iso), contributions(Iso, Ls), assertion(Ls \== []), length(Ls, N), T is T0 + N, foldl(check_line(C), Ls, F0, F). check_line(C, L, F0, F) :- _{iso:Iso, name:Name, currency:Cur, total_declared_floor:Fl, total_declared_ceiling:Ce} :< C, _{iso:LIso, country:LName, currency:LCur, total_declared_floor:LFl, total_declared_ceiling:LCe, line_type:Type, line_id:ID, base:Base, base_factor:Factor, extra_kind:Kind, extra_value:Extra, rate:Rate, source_url:URL, source_checked_at:Checked} :< L, assertion((LIso == Iso, LCur == Cur, LName == Name)), assertion((LFl == Fl, LCe == Ce)), ( Type == "contribution" -> assertion((memberchk(Base, ["gross", "basic", "flat", "band"]), Kind == "")), assertion(( Base == "flat" -> Factor == null ; Factor \== null )), ( Factor \== null, Factor =\= 1 -> F = [Factor|F0] ; F = F0 ) ; Type == "statutory_extra" -> assertion(memberchk(Kind, ["percent", "months", "days"])), assertion((Extra \== null, Rate == null, Factor == null)), F = F0 ; assertion(fail) ), assertion((ID \== "", URL \== "", Checked = date(_, _, _))). :- end_tests(eorscope_contributions). :- begin_tests(eorscope_assumptions). test(assumptions) :- countries(Cs), foldl([C, N0, N]>>( get_dict(iso, C, Iso), assumptions(Iso, As), forall(member(A, As), assertion(( get_dict(iso, A, Iso), get_dict(id, A, ID), ID \== "", get_dict(label, A, Lb), Lb \== "" ))), length(As, K), N is N0 + K ), Cs, 0, Total), assertion(Total == 108), assumptions(xx, None), assertion(None == []). :- end_tests(eorscope_assumptions).