package missionclass

import "testing"

// TestClassifyLaborGolden pins ClassifyLabor to the exact labor category set the old inline laborKeywords
// logic in matching/intel.go produced for representative profile headlines. The matcher is a lowercased
// substring match over the same word list, so candidate intelligence output stays byte identical after
// the helper moved here. Results are returned in stable sorted order.
func TestClassifyLaborGolden(t *testing.T) {
	cases := []struct {
		text string
		want []string
	}{
		// An ISSO headline.
		{"Information System Security Officer with ISSM duties", []string{"ISSO"}},
		// A cyber headline that hits both the engineer and the SOC analyst words, plus intelligence via sigint.
		{"Senior Security Engineer and SOC Analyst, SIGINT background", []string{"Cyber Engineer", "Intelligence Analyst"}},
		// A program manager who is also a PMP.
		{"Program Manager, PMP certified capture manager", []string{"Program Manager"}},
		// A data scientist who does machine learning.
		{"Data Scientist focused on machine learning pipelines", []string{"Data Scientist"}},
		// A cloud and solutions architect.
		{"Cloud Architect and Solutions Architect for AWS", []string{"Cloud Architect"}},
		// An all source intelligence analyst.
		{"All Source Intelligence Analyst, GEOINT and intel analyst tradecraft", []string{"Intelligence Analyst"}},
		// A plain administrative headline matches nothing.
		{"General administrative support coordinator", nil},
		// Empty text matches nothing.
		{"", nil},
	}
	for _, tc := range cases {
		got := ClassifyLabor(tc.text)
		if len(got) != len(tc.want) {
			t.Fatalf("ClassifyLabor(%q) = %v, want %v", tc.text, got, tc.want)
		}
		for i := range got {
			if got[i] != tc.want[i] {
				t.Fatalf("ClassifyLabor(%q) = %v, want %v", tc.text, got, tc.want)
			}
		}
	}
}

// TestClassifyLaborCounts pins ClassifyLaborCounts to its deterministic occurrence count and first matched
// phrase evidence. Count is the literal sum of strings.Count over every matched keyword phrase, so a
// repeated mention counts again and a category named through two phrases counts both. Order follows the
// stable laborKeywords order. The labor mix is mined from genuine text matches only, never fabricated.
func TestClassifyLaborCounts(t *testing.T) {
	cases := []struct {
		name string
		text string
		want []LaborHit
	}{
		{
			// Two ISSO mentions: "isso" matches twice. The first matched phrase in the word list is "isso".
			name: "repeated isso counts twice",
			text: "We need an ISSO and a second ISSO for night shift",
			want: []LaborHit{{LCAT: "ISSO", Count: 2, Evidence: "isso"}},
		},
		{
			// A description naming three engineers and two cyber roles. "cyber engineer" hits once, and the
			// Intelligence Analyst category is hit by sigint once.
			name: "engineer and intel mix",
			text: "3 Cyber Engineer staff and 2 SIGINT analysts on the SIGINT desk",
			want: []LaborHit{
				{LCAT: "Cyber Engineer", Count: 1, Evidence: "cyber engineer"},
				{LCAT: "Intelligence Analyst", Count: 2, Evidence: "sigint"},
			},
		},
		{
			// A plain administrative headline matches nothing.
			name: "no match",
			text: "general administrative support services",
			want: nil,
		},
		{
			name: "empty",
			text: "",
			want: nil,
		},
	}
	for _, tc := range cases {
		t.Run(tc.name, func(t *testing.T) {
			got := ClassifyLaborCounts(tc.text)
			if len(got) != len(tc.want) {
				t.Fatalf("ClassifyLaborCounts(%q) = %v, want %v", tc.text, got, tc.want)
			}
			for i := range got {
				if got[i] != tc.want[i] {
					t.Fatalf("ClassifyLaborCounts(%q)[%d] = %+v, want %+v", tc.text, i, got[i], tc.want[i])
				}
			}
		})
	}
}