job-search/internal/skiprules/rules_test.go

package skiprules

import (
	"encoding/json"
	"os"
	"strings"
	"testing"

	"git.avsdev.uk/Go/job-search/internal/types"
)

// mustParse parses text as a rules file, failing the test on error.
func mustParse(t *testing.T, text string) *Rules {
	t.Helper()
	rules, err := Parse(strings.NewReader(text))
	if err != nil {
		t.Fatal(err)
	}
	return rules
}

// withFilter returns a rules file of one filter with fields (and a comment
// "c"); withRule one of an active filter holding rules.
func withFilter(fields string) string {
	return `[{"comment":"c",` + fields + `}]`
}

func withRule(rules ...string) string {
	return withFilter(`"active":true,"rules":[` + strings.Join(rules, ",") + `]`)
}

// ruleJSON returns a rule of ruleType with values as JSON, and extra fields.
func ruleJSON(ruleType, extra string, values ...string) string {
	encoded, _ := json.Marshal(values)
	return `{"rule_type":"` + ruleType + `","values":` + string(encoded) +
		extra + `}`
}

func buyer(values ...string) string { return ruleJSON("buyer", "", values...) }
func title(values ...string) string { return ruleJSON("title", "", values...) }
func cpv(values ...string) string   { return ruleJSON("cpv", "", values...) }

// cpvPrefixes is a CPV rule with exact false, cpvOnly one with exclusive
// true (the default), cpvAny one with exclusive false, and cpvAnyPrefixes
// one with exact and exclusive false.
func cpvPrefixes(values ...string) string {
	return ruleJSON("cpv", `,"exact":false`, values...)
}

func cpvOnly(values ...string) string {
	return ruleJSON("cpv", `,"exclusive":true`, values...)
}

func cpvAny(values ...string) string {
	return ruleJSON("cpv", `,"exclusive":false`, values...)
}

func cpvAnyPrefixes(values ...string) string {
	return ruleJSON("cpv", `,"exact":false,"exclusive":false`, values...)
}

// opportunity returns an opportunity with buyer, title and CPV codes.
func opportunity(buyer, title string, cpvs ...string) *types.Opportunity {
	o := &types.Opportunity{
		UUID:    "ocds-1",
		BuyerID: buyer,
		Title:   title,
	}
	for _, code := range cpvs {
		o.Classifications = append(o.Classifications, types.ClassificationCode{
			IsCPV: true,
			Code:  code,
		})
	}
	return o
}

// withCPVs and titled are opportunities with only CPV codes or a title.
func withCPVs(codes ...string) *types.Opportunity {
	return opportunity("", "x", codes...)
}

func titled(text string) *types.Opportunity { return opportunity("", text) }

func TestParseRefuses(t *testing.T) {
	active := `"active":true,`
	tests := []struct {
		name, text, want string
	}{
		{"not JSON", `{`, "reading JSON"},
		{"not a list", `{}`, "reading JSON"},
		{"null", `null`, "must be a list of filters"},
		{"trailing", `[] []`, "unexpected content"},
		{"unknown filter field", withFilter(`"filter_type":"x"`), `unknown field "filter_type"`},
		{"unknown rule field", withRule(`{"value":"a"}`), `unknown field "value"`},
		{"no comment", `[{"active":true,"rules":[` + buyer("a") + `]}]`, "filter 1: skiprules: comm"},
		{"no active", withFilter(`"rules":[` + buyer("a") + `]`), `filter 1 ("c"): skiprules: active`},
		{"no rules", withFilter(`"active":true`), "rules is required"},
		{"empty rules", withRule(), "rules is required"},
		{"bad logic", withFilter(active + `"logic_mode":"xor","rules":[` + buyer("a") + `]`), `"xor"`},
		{"empty logic", withFilter(active + `"logic_mode":"","rules":[` + buyer("a") + `]`), `""`},
		{"no rule type", withRule(`{"values":["a"]}`), "rule 1: skiprules: rule_type is required"},
		{"bad rule type", withRule(ruleJSON("status", "", "a")), `rule_type "status"`},
		{"no values", withRule(`{"rule_type":"buyer"}`), "values is required"},
		{"empty values", withRule(buyer()), "values is required"},
		{"exact on title", withRule(ruleJSON("title", `,"exact":true`, "a")), `exact is only for`},
		{"exclusive on buyer", withRule(ruleJSON("buyer", `,"exclusive":false`, "a")), "exclusive is"},
		{"no words", withRule(title("--")), "values has no words"},
		{"short CPV", withRule(cpv("7200000")), `"7200000" is not a CPV code`},
		{"letters CPV", withRule(cpv("7200000x")), "is not a CPV code"},
	}
	for _, tt := range tests {
		_, err := Parse(strings.NewReader(tt.text))
		if err == nil || !strings.Contains(err.Error(), tt.want) {
			t.Errorf("%s: err = %v; want one containing %q", tt.name, err, tt.want)
		}
	}
}

// An error names the filter and rule by their place in the file.
func TestParseNamesFilterAndRule(t *testing.T) {
	_, err := Parse(strings.NewReader(`[
		{"comment": "a", "active": true, "rules": [
			{"rule_type": "buyer", "values": ["a"]}
		]},
		{"comment": "b", "active": true, "rules": [
			{"rule_type": "buyer", "values": ["a"]},
			{"rule_type": "cpv", "values": ["x"]}
		]}
	]`))
	want := `filter 2 ("b"): rule 2: skiprules: "x" is not a CPV code`
	if err == nil || !strings.Contains(err.Error(), want) {
		t.Errorf("err = %v; want one containing %q", err, want)
	}
}

func TestParseAccepts(t *testing.T) {
	rules := mustParse(t, `[
		{"comment": "a", "active": true, "rules": [
			{"rule_type": "buyer", "values": ["b"]}
		]},
		{"comment": "", "active": false, "logic_mode": "and", "rules": [
			{"rule_type": "cpv", "values": ["72000000-5"], "exact": false},
			{"rule_type": "title", "values": ["road"]}
		]}
	]`)
	if filters, active := rules.Filters(); filters != 2 || active != 1 {
		t.Errorf("Filters() = %d, %d; want 2, 1", filters, active)
	}

	if filters, _ := mustParse(t, `[]`).Filters(); filters != 0 {
		t.Errorf("empty list: %d filters", filters)
	}
	var none *Rules
	if _, matched := none.Match(opportunity("b", "x")); matched {
		t.Error("nil rules matched")
	}
}

func TestMatch(t *testing.T) {
	tests := []struct {
		name  string
		rule  string
		o     *types.Opportunity
		match bool
	}{
		// buyer: any of the values
		{"buyer", buyer("GB-1", "GB-2"), opportunity("GB-2", "x"), true},
		{"other buyer", buyer("GB-1"), opportunity("GB-10", "x"), false},

		// title: every word, whole words, any case, punctuation splits
		{"title", title("road", "REPAIRS"), titled("Road-repairs, phase 2"), true},
		{"title phrase", title("road repairs"), titled("Repairs to the road"), true},
		{"title word missing", title("road", "bridge"), titled("Road repairs"), false},
		{"title part word", title("road"), titled("Railroad works"), false},
		{"title digits", title("phase 2"), titled("Road-repairs, phase 2"), true},

		// cpv: all the opportunity's CPVs in the values (a subset)
		{"cpv subset", cpv("03000000", "72000000", "48000000"), withCPVs("72000000", "48000000"), true},
		{"cpv equal", cpv("72000000"), withCPVs("72000000"), true},
		{"cpv one outside", cpv("72000000"), withCPVs("72000000", "48000000"), false},
		{"cpv none", cpv("72000000"), withCPVs(), false},
		{"cpv check digits", cpv("72000000-5"), withCPVs("72000000-1"), true},
		{"cpv exact by default", cpv("72000000"), withCPVs("72200000"), false},
		{"cpv prefix", cpvPrefixes("72000000"), withCPVs("72212000-4", "72000000"), true},
		{"cpv prefix group", cpvPrefixes("72200000"), withCPVs("72300000"), false},
		{"cpv prefix division", cpvPrefixes("70000000"), withCPVs("71000000"), false},
		{"cpv prefix in division", cpvPrefixes("70000000"), withCPVs("70110000"), true},
		{"cpv exclusive", cpvOnly("72000000"), withCPVs("72000000", "48000000"), false},

		// cpv, exclusive false: any of the opportunity's CPVs in the values
		{"cpv any: one of two", cpvAny("72000000"), withCPVs("48000000", "72000000"), true},
		{"cpv any: all", cpvAny("72000000", "48000000"), withCPVs("72000000", "48000000"), true},
		{"cpv any: none in values", cpvAny("72000000"), withCPVs("48000000", "03000000"), false},
		{"cpv any: no CPVs", cpvAny("72000000"), withCPVs(), false},
		{"cpv any: exact by default", cpvAny("72000000"), withCPVs("72200000", "48000000"), false},
		{"cpv any: prefix", cpvAnyPrefixes("72000000"), withCPVs("48000000", "72212000-4"), true},
		{"cpv any: check digits", cpvAny("72000000-5"), withCPVs("48000000", "72000000-1"), true},
	}
	for _, tt := range tests {
		rules := mustParse(t, withRule(tt.rule))
		comment, matched := rules.Match(tt.o)
		if matched != tt.match {
			t.Errorf("%s: matched %v; want %v", tt.name, matched, tt.match)
		}
		if matched && comment != "c" {
			t.Errorf("%s: comment %q; want c", tt.name, comment)
		}
	}
}

// Non-CPV classifications are ignored: only CPVs must be covered.
func TestMatchIgnoresOtherSchemes(t *testing.T) {
	rules := mustParse(t, withRule(cpv("72000000")))
	o := withCPVs("72000000")
	o.Classifications = append(o.Classifications, types.ClassificationCode{
		Code: "1234",
	})
	if _, matched := rules.Match(o); !matched {
		t.Error("a non-CPV code stopped the match")
	}
}

// Buyer X, or CPVs within 03000000 and 04000000; "and" needs both.
func TestMatchLogicModes(t *testing.T) {
	rules := buyer("X") + "," + cpv("03000000", "04000000")
	or, and := `"logic_mode":"or",`, `"logic_mode":"and",`
	tests := []struct {
		name, mode string
		o          *types.Opportunity
		match      bool
	}{
		{"default: buyer only", ``, opportunity("X", "t", "72000000"), true},
		{"or: cpv only", or, opportunity("Y", "t", "03000000"), true},
		{"default: neither", ``, opportunity("Y", "t", "72000000"), false},
		{"and: both", and, opportunity("X", "t", "03000000", "04000000"), true},
		{"and: buyer only", and, opportunity("X", "t", "72000000"), false},
		{"and: cpv only", and, opportunity("Y", "t", "03000000"), false},
	}
	for _, tt := range tests {
		text := withFilter(`"active":true,` + tt.mode + `"rules":[` + rules + `]`)
		if _, matched := mustParse(t, text).Match(tt.o); matched != tt.match {
			t.Errorf("%s: matched %v; want %v", tt.name, matched, tt.match)
		}
	}
}

// Inactive filters are ignored; any active filter that matches is enough,
// and the first one's comment is returned.
func TestMatchFilters(t *testing.T) {
	rules := mustParse(t, `[
		{"comment": "off", "active": false, "rules": [
			{"rule_type": "buyer", "values": ["X"]}
		]},
		{"comment": "title", "active": true, "rules": [
			{"rule_type": "title", "values": ["road"]}
		]},
		{"comment": "buyer", "active": true, "rules": [
			{"rule_type": "buyer", "values": ["X"]}
		]}
	]`)

	tests := []struct {
		o       *types.Opportunity
		comment string
		match   bool
	}{
		{opportunity("X", "Road works"), "title", true},
		{opportunity("X", "Bridge works"), "buyer", true},
		{opportunity("Y", "Road works"), "title", true},
		{opportunity("Y", "Bridge works"), "", false},
	}
	for _, tt := range tests {
		comment, matched := rules.Match(tt.o)
		if comment != tt.comment || matched != tt.match {
			t.Errorf(
				"%s/%s: %q, %v; want %q, %v",
				tt.o.BuyerID, tt.o.Title, comment, matched, tt.comment, tt.match,
			)
		}
	}
}

// The example in README.md is a valid rules file.
func TestREADMEExample(t *testing.T) {
	readme, err := os.ReadFile("README.md")
	if err != nil {
		t.Fatal(err)
	}
	_, example, found := strings.Cut(string(readme), "```json\n")
	example, _, _ = strings.Cut(example, "```")
	if !found {
		t.Fatal("README.md has no JSON example")
	}

	filters, active := mustParse(t, example).Filters()
	if filters != 4 || active != 3 {
		t.Errorf("example: %d filters, %d active; want 4, 3", filters, active)
	}
}