go-fa-api/listing_json_test.go

package fa

import (
	"bytes"
	"strings"
	"testing"

	"github.com/PuerkitoBio/goquery"
)

// TestReadListingJSON_RealFixture parses the embedded JSON blob out of the
// captured gallery_page1.html and asserts the shape we expect: keyed by
// SubmissionID, populated title/username/lower fields.
func TestReadListingJSON_RealFixture(t *testing.T) {
	raw := loadFixture(t, "gallery_page1.html")
	doc, err := goquery.NewDocumentFromReader(bytes.NewReader(raw))
	if err != nil {
		t.Fatalf("read doc: %v", err)
	}
	m := readListingJSON(doc)
	if m == nil {
		t.Fatal("readListingJSON returned nil script tag missing or malformed?")
	}
	if len(m) == 0 {
		t.Fatal("readListingJSON returned empty map")
	}
	t.Logf("parsed %d entries", len(m))

	missingTitle, missingLower, missingMtime := 0, 0, 0
	for id, entry := range m {
		if id == 0 {
			t.Errorf("entry has zero ID")
		}
		if entry.Title == "" {
			missingTitle++
		}
		if entry.Lower == "" {
			missingLower++
		}
		if entry.AvatarMtime == "" {
			missingMtime++
		}
	}
	// Allow some entries to lack avatar_mtime (FA sometimes omits it for
	// stale avatars), but title/lower should be on every entry.
	if missingTitle > 0 {
		t.Errorf("%d/%d entries had empty Title", missingTitle, len(m))
	}
	if missingLower > 0 {
		t.Errorf("%d/%d entries had empty Lower", missingLower, len(m))
	}
	t.Logf("missing avatar_mtime: %d/%d (acceptable)", missingMtime, len(m))
}

// TestParseGalleryPage_JSONMergePicksUpAvatars compares HTML-only and JSON
// merge modes on a favorites fixture (favorites lists submissions by many
// different artists, so the embedded JSON carries avatar_mtime for each;
// gallery pages don't, since every item is by the same artist).
//
// The JSON path should populate Author.AvatarURL on items where HTML
// scraping leaves it empty.
func TestParseGalleryPage_JSONMergePicksUpAvatars(t *testing.T) {
	raw := loadFixture(t, "favorites_page1.html")
	doc, err := goquery.NewDocumentFromReader(bytes.NewReader(raw))
	if err != nil {
		t.Fatalf("read doc: %v", err)
	}

	htmlOnly, _ := parseGalleryPage(doc, false)
	jsonMerged, _ := parseGalleryPage(doc, true)

	if len(htmlOnly) != len(jsonMerged) {
		t.Fatalf("item counts differ: html=%d json=%d", len(htmlOnly), len(jsonMerged))
	}
	if len(jsonMerged) == 0 {
		t.Fatal("no items parsed")
	}

	htmlAvatars, jsonAvatars := 0, 0
	for _, s := range htmlOnly {
		if s.Author.AvatarURL != "" {
			htmlAvatars++
		}
	}
	for _, s := range jsonMerged {
		if s.Author.AvatarURL != "" {
			jsonAvatars++
		}
	}
	t.Logf("html-only avatars: %d/%d; json-merged avatars: %d/%d",
		htmlAvatars, len(htmlOnly), jsonAvatars, len(jsonMerged))

	if jsonAvatars <= htmlAvatars {
		t.Errorf("expected JSON-merged parse to populate more AvatarURLs than HTML-only; got html=%d json=%d",
			htmlAvatars, jsonAvatars)
	}
	// Spot-check one avatar URL shape.
	for _, s := range jsonMerged {
		if s.Author.AvatarURL != "" {
			if !strings.HasPrefix(s.Author.AvatarURL, "https://a.furaffinity.net/") {
				t.Errorf("avatar URL has wrong prefix: %q", s.Author.AvatarURL)
			}
			if !strings.HasSuffix(s.Author.AvatarURL, "/"+s.Author.Name+".gif") {
				t.Errorf("avatar URL doesn't end with /%s.gif: %q", s.Author.Name, s.Author.AvatarURL)
			}
			break
		}
	}
}

// TestParseGalleryPage_JSONMergeGracefullyFallsBack confirms that
// useJSON=true on a doc with no js-submissionData blob produces the same
// result as useJSON=false i.e. the merge is non-destructive.
func TestParseGalleryPage_JSONMergeGracefullyFallsBack(t *testing.T) {
	// The synthetic gallery fixture has no js-submissionData tag.
	doc, err := goquery.NewDocumentFromReader(strings.NewReader(syntheticGalleryHTML))
	if err != nil {
		t.Fatalf("setup: %v", err)
	}
	htmlOnly, _ := parseGalleryPage(doc, false)
	jsonMerged, _ := parseGalleryPage(doc, true)

	if len(htmlOnly) != len(jsonMerged) {
		t.Fatalf("counts differ: %d vs %d", len(htmlOnly), len(jsonMerged))
	}
	for i := range htmlOnly {
		if htmlOnly[i].ID != jsonMerged[i].ID || htmlOnly[i].Title != jsonMerged[i].Title {
			t.Errorf("item %d: html=%+v json=%+v", i, htmlOnly[i], jsonMerged[i])
		}
	}
}

// TestReadListingJSON_MissingTagReturnsNil makes the fallback path
// explicit in a unit-style assertion.
func TestReadListingJSON_MissingTagReturnsNil(t *testing.T) {
	doc, _ := goquery.NewDocumentFromReader(strings.NewReader("<html><body></body></html>"))
	if m := readListingJSON(doc); m != nil {
		t.Errorf("expected nil, got %d entries", len(m))
	}
}

// TestReadListingJSON_MalformedJSONReturnsNil makes the fallback path
// explicit when FA's blob is present but somehow unparseable.
func TestReadListingJSON_MalformedJSONReturnsNil(t *testing.T) {
	doc, _ := goquery.NewDocumentFromReader(strings.NewReader(
		`<html><body><script id="js-submissionData" type="application/json">{not valid</script></body></html>`,
	))
	if m := readListingJSON(doc); m != nil {
		t.Errorf("expected nil, got %d entries", len(m))
	}
}