golib/io/transform/gsheet2csv/gsheet2csv_test.go

package gsheet2csv

import (
	"errors"
	"io"
	"net/http"
	"slices"
	"strings"
	"testing"
)

// mockHTTPClient allows controlling HTTP responses for testing.
type mockHTTPClient struct {
	resp *http.Response
	err  error
}

func (m *mockHTTPClient) Get(url string) (*http.Response, error) {
	return m.resp, m.err
}

// sampleCSV mimics the structure of ai-models.csv from the project README.
const sampleCSV = `# Generated by ollama list
"# Sample Quoted Comment, with ""quotes"" itself"
"NAME","ID","SIZE","MODIFIED"
"qwen3-coder:30b","06c1097efce0","18 GB","8 days ago"
"gpt-oss:20b","aa4295ac10c3","13 GB","8 days ago"

"gpt-oss:latest","aa4295ac10c3","13 GB","7 weeks ago"
`

// malformedCSV for testing error handling.
const malformedCSV = `# Comment
"NAME","ID","SIZE","MODIFIED
"qwen3-coder:30b","06c1097efce0","18 GB","8 days ago"
`

// TestParseIDs verifies the ParseIDs function for various URL formats.
func TestParseIDs(t *testing.T) {
	tests := []struct {
		name    string
		url     string
		wantDoc string
		wantGid string
	}{
		{
			name:    "Google Sheets Edit / Share URL with gid",
			url:     "https://docs.google.com/spreadsheets/d/1KdNsc63pk0QRerWDPcIL9cMnGQlG-9Ue9Jlf0PAAA34/edit?gid=559037238#gid=559037238",
			wantDoc: "1KdNsc63pk0QRerWDPcIL9cMnGQlG-9Ue9Jlf0PAAA34",
			wantGid: "559037238",
		},
		{
			name:    "Google Sheets CSV URL with gid",
			url:     "https://docs.google.com/spreadsheets/d/1KdNsc63pk0QRerWDPcIL9cMnGQlG-9Ue9Jlf0PAAA34/export?format=csv&usp=sharing&gid=559037238",
			wantDoc: "1KdNsc63pk0QRerWDPcIL9cMnGQlG-9Ue9Jlf0PAAA34",
			wantGid: "559037238",
		},
		{
			name:    "URL without gid",
			url:     "https://docs.google.com/spreadsheets/d/1KdNsc63pk0QRerWDPcIL9cMnGQlG-9Ue9Jlf0PAAA34/edit",
			wantDoc: "1KdNsc63pk0QRerWDPcIL9cMnGQlG-9Ue9Jlf0PAAA34",
			wantGid: "0",
		},
		{
			name:    "Invalid URL",
			url:     "https://example.com/invalid",
			wantDoc: "",
			wantGid: "",
		},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			gotDoc, gotGid := ParseIDs(tt.url)
			if gotDoc != tt.wantDoc {
				t.Errorf("ParseIDs() docid = %q, want %q", gotDoc, tt.wantDoc)
			}
			if gotGid != tt.wantGid {
				t.Errorf("ParseIDs() gid = %q, want %q", gotGid, tt.wantGid)
			}
		})
	}
}

// TestNewReaderFromURL tests initializing a Reader from a Google Sheets URL.
func TestNewReaderFromURL(t *testing.T) {
	originalGet := httpGet
	defer func() { httpGet = originalGet }()

	url := "https://docs.google.com/spreadsheets/d/1KdNsc63pk0QRerWDPcIL9cMnGQlG-9Ue9Jlf0PAAA34/edit?gid=559037238"

	// Test successful HTTP response
	mockResp := &http.Response{
		StatusCode: http.StatusOK,
		Body:       io.NopCloser(strings.NewReader(sampleCSV)),
	}
	client := &mockHTTPClient{resp: mockResp}
	httpGet = client.Get

	reader := NewReaderFromURL(url)
	if reader.err != nil {
		t.Errorf("NewReaderFromURL() unexpected error: %v", reader.err)
	}
	if reader.resp != mockResp {
		t.Error("NewReaderFromURL() did not set response correctly")
	}
	if !reader.close {
		t.Error("NewReaderFromURL() did not set close flag")
	}

	// Test HTTP failure
	client = &mockHTTPClient{resp: mockResp}
	client.err = errors.New("network error")
	httpGet = client.Get

	reader = NewReaderFromURL(url)
	if reader.err == nil {
		t.Error("NewReaderFromURL() expected error, got nil")
	}

	// Test non-200 status
	client = &mockHTTPClient{resp: &http.Response{
		StatusCode: http.StatusNotFound,
		Body:       io.NopCloser(strings.NewReader("these aren't the droids you're looking for")),
	}}
	httpGet = client.Get

	reader = NewReaderFromURL(url)
	if reader.err == nil {
		t.Error("NewReaderFromURL() expected error for non-200 status, got nil")
	}
}

// TestRead tests the Read method for comment handling.
func TestRead(t *testing.T) {
	tests := []struct {
		name             string
		preserveComments bool
		expected         [][]string
	}{
		{
			name: "Skip comments",
			expected: [][]string{
				{"NAME", "ID", "SIZE", "MODIFIED"},
				{"qwen3-coder:30b", "06c1097efce0", "18 GB", "8 days ago"},
				{"gpt-oss:20b", "aa4295ac10c3", "13 GB", "8 days ago"},
				{"gpt-oss:latest", "aa4295ac10c3", "13 GB", "7 weeks ago"},
			},
		},
		{
			name:             "Don't skip comments",
			preserveComments: true,
			expected: [][]string{
				{"# Generated by ollama list"},
				{"# Sample Quoted Comment, with \"quotes\" itself"},
				{"NAME", "ID", "SIZE", "MODIFIED"},
				{"qwen3-coder:30b", "06c1097efce0", "18 GB", "8 days ago"},
				{"gpt-oss:20b", "aa4295ac10c3", "13 GB", "8 days ago"},
				{"gpt-oss:latest", "aa4295ac10c3", "13 GB", "7 weeks ago"},
			},
		},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			reader := NewReader(strings.NewReader(sampleCSV))
			if tt.preserveComments {
				reader.Comment = 0
			}

			for i, want := range tt.expected {
				got, err := reader.Read()
				if err != nil {
					t.Errorf("Read() error at record %d: %v", i, err)
				}
				if !slices.Equal(got, want) {
					t.Errorf("Read() record %d = %v, want %v", i, got, want)
				}
			}

			// Verify EOF
			_, err := reader.Read()
			if !errors.Is(err, io.EOF) {
				t.Errorf("Read() expected EOF, got %v", err)
			}
		})
	}
}

// TestReadAll tests the ReadAll method for different configurations.
func TestReadAll(t *testing.T) {
	tests := []struct {
		name     string
		expected [][]string
	}{
		{
			name: "Skip comments",
			expected: [][]string{
				{"NAME", "ID", "SIZE", "MODIFIED"},
				{"qwen3-coder:30b", "06c1097efce0", "18 GB", "8 days ago"},
				{"gpt-oss:20b", "aa4295ac10c3", "13 GB", "8 days ago"},
				{"gpt-oss:latest", "aa4295ac10c3", "13 GB", "7 weeks ago"},
			},
		},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			reader := NewReader(strings.NewReader(sampleCSV))

			got, err := reader.ReadAll()
			if err != nil {
				t.Errorf("ReadAll() error: %v", err)
			}
			if len(got) != len(tt.expected) {
				t.Errorf("ReadAll() returned %d records, want %d", len(got), len(tt.expected))
			}
			for i, want := range tt.expected {
				if !slices.Equal(got[i], want) {
					t.Errorf("ReadAll() record %d = %v, want %v", i, got[i], want)
				}
			}
		})
	}
}

// TestNewReaderFromURLWithMalformedCSV tests NewReaderFromURL with malformed CSV.
func TestNewReaderFromURLWithMalformedCSV(t *testing.T) {
	mockResp := &http.Response{
		StatusCode: http.StatusOK,
		Body:       io.NopCloser(strings.NewReader(malformedCSV)),
	}
	client := &mockHTTPClient{resp: mockResp}
	originalGet := httpGet
	httpGet = client.Get
	defer func() { httpGet = originalGet }()

	url := "https://docs.google.com/spreadsheets/d/1KdNsc63pk0QRerWDPcIL9cMnGQlG-9Ue9Jlf0PAAA34/edit?gid=559037238"
	reader := NewReaderFromURL(url)
	if reader.err != nil {
		t.Errorf("NewReaderFromURL() unexpected error: %v", reader.err)
	}

	// Reading should fail due to malformed CSV
	_, err := reader.Read()
	if err == nil {
		t.Error("Read() expected error for malformed CSV, got nil")
	}
}