go-fa-api/user_parser.go

package fa

import (
	"fmt"
	"strings"

	"github.com/PuerkitoBio/goquery"

	"git.anthrove.art/public/go-fa-api/internal/urls"
)

// parseUser pulls a [User] out of /user/{name}/.
//
// FA's profile page has many optional sections; parser treats the headline
// (display name + avatar) as required, everything else as best-effort.
func parseUser(name string, doc *goquery.Document) (*User, error) {
	u := &User{UserRef: UserRef{Name: strings.ToLower(strings.TrimSpace(name))}}

	// Headline username + avatar.
	header := doc.Find("userpage-nav-user-details, div.userpage-nav-user-details, div.username").First()
	if header.Length() == 0 {
		header = doc.Find("h1.username, h2.username").First()
	}
	u.DisplayName = firstNonEmpty(
		// Scope the display name to the profile header first an unscoped
		// .c-usernameBlock__displayName also matches the logged-in viewer's
		// block elsewhere on the page.
		trimText(doc.Find("userpage-nav-user-details .js-displayName").First()),
		trimText(doc.Find("userpage-nav-user-details .c-usernameBlock__displayName").First()),
		trimText(doc.Find(".username h2 span").First()),
		trimText(doc.Find(".username h1").First()),
		trimText(doc.Find(".c-usernameBlock__displayName").First()),
		trimText(doc.Find(".c-usernameBlockSimple__displayName").First()),
		trimText(header),
		u.Name,
	)
	// The profile owner's avatar lives in the <userpage-nav-avatar> header
	// element. It must be scoped there: img.avatar / img.loggedin_user_avatar
	// in the site navigation belong to the logged-in viewer, and an unscoped
	// selector picks the viewer's avatar on every logged-in page load.
	u.AvatarURL = urls.AbsoluteCDN(firstNonEmpty(
		trimAttr(doc.Find("userpage-nav-avatar img").First(), "src"),
		trimAttr(doc.Find("div.userpage-nav-avatar img").First(), "src"),
		trimAttr(doc.Find("img.user-nav-avatar").First(), "src"),
	))
	if u.DisplayName == "" {
		return nil, fmt.Errorf("%w: user %q: missing display name", ErrParse, name)
	}

	// "Title" headline shown under the username.
	u.Title = firstNonEmpty(
		trimText(doc.Find(".userpage-flex-item.username .font-small").First()),
		trimText(doc.Find(".user-nav-user-details .c-usernameBlock__subtitle").First()),
	)

	// Registered-on date appears in profile-meta or in a span.popup_date.
	doc.Find("span.popup_date").EachWithBreak(func(_ int, sel *goquery.Selection) bool {
		raw := firstNonEmpty(trimAttr(sel, "title"), trimText(sel))
		if t, err := ParseFADate(raw); err == nil {
			u.Joined = t
			return false
		}
		return true
	})

	// Profile bio: large HTML block on the left column.
	bio := firstNonEmptySel(doc,
		"div.userpage-profile",
		"div.profile-page-body",
		"div.profile-description",
	)
	if bio != nil {
		u.BioHTML = htmlOf(bio)
		u.BioText = strings.TrimSpace(bio.Text())
	}

	// Stats: the "Stats" box in the right column is a flat run of
	//   <span class="highlight">Label:</span> value<br/>
	// pairs inside one or more <div class="cell">. The value is the bare
	// text node that immediately follows each highlight span.
	doc.Find("div.userpage-section-right div.cell").Each(func(_ int, cell *goquery.Selection) {
		nodes := cell.Contents()
		nodes.Each(func(i int, node *goquery.Selection) {
			if !node.Is("span.highlight") {
				return
			}
			label := strings.ToLower(strings.TrimRight(trimText(node), ":"))
			val := parseStatNumber(nodes.Eq(i + 1).Text())
			switch label {
			case "submissions":
				u.Stats.Submissions = val
			case "favs", "favorites":
				u.Stats.Favorites = val
			case "views", "page visits":
				u.Stats.Views = val
			case "comments earned", "comments":
				u.Stats.Comments = val
			case "journals":
				u.Stats.Journals = val
			}
		})
	})

	// Watcher / watching counts are NOT in the stats box FA renders them
	// in the "Recent Watchers" / "Recently Watched" section headers as
	// "View List (Watched by N)" / "View List (Watching N)".
	u.Stats.Watchers = parseStatNumber(trimText(doc.Find("section.watched-by-block .section-header a").First()))
	u.Stats.Watching = parseStatNumber(trimText(doc.Find("section.is-watching-block .section-header a").First()))

	// Contact information rows.
	doc.Find("div.user-contact-user-info, .userpage-contact-information li").Each(func(_ int, sel *goquery.Selection) {
		site := trimText(sel.Find("span.user-contact-item-name, .contact-site").First())
		linkSel := sel.Find("a").First()
		handle := trimText(linkSel)
		if handle == "" {
			handle = strings.TrimSpace(sel.Text())
		}
		href, _ := linkSel.Attr("href")
		if site != "" || handle != "" {
			u.Contacts = append(u.Contacts, UserContact{
				Site:   site,
				Handle: handle,
				URL:    urls.AbsoluteCDN(href),
			})
		}
	})

	// Featured submission: small preview thumbnail on the profile.
	if feat := doc.Find("div.userpage-featured-submission a, section.userpage-section-right figure a").First(); feat.Length() > 0 {
		href, _ := feat.Attr("href")
		if id := extractIntFromHref(href); id > 0 {
			u.FeaturedSub = &SubmissionRef{
				ID:       SubmissionID(id),
				Title:    trimAttr(feat, "title"),
				ThumbURL: urls.AbsoluteCDN(trimAttr(feat.Find("img").First(), "src")),
			}
		}
	}

	// Shouts: anchored by <a id="shout-NNN"> inside a
	// <div class="comment_container"> (underscore). Beta uses custom HTML5
	// elements <comment-container>/<comment-username>/<comment-date>/
	// <comment-user-text> within that wrapper goquery matches them by tag.
	doc.Find("a[id^='shout-']").Each(func(_ int, anchor *goquery.Selection) {
		container := anchor.ParentsFiltered("div.comment_container").First()
		if container.Length() == 0 {
			// Fallback for legacy markup where the anchor sits as a sibling
			// of a table or comment-container directly.
			container = anchor.Parent()
		}
		shout := Shout{}

		authorLink := container.Find("a.c-usernameBlock__displayName").First()
		if authorLink.Length() > 0 {
			href, _ := authorLink.Attr("href")
			shout.Author = UserRef{
				DisplayName: trimText(authorLink.Find("span.js-displayName").First()),
				AvatarURL:   urls.AbsoluteCDN(trimAttr(container.Find("img.comment_useravatar").First(), "src")),
			}
			if shout.Author.DisplayName == "" {
				shout.Author.DisplayName = trimText(authorLink)
			}
			if parts := strings.Split(strings.Trim(href, "/"), "/"); len(parts) >= 2 {
				shout.Author.Name = strings.ToLower(parts[1])
			}
		}

		shout.PostedAt = parsePopupDate(container.Find("comment-date span.popup_date").First())
		if shout.PostedAt.IsZero() {
			shout.PostedAt = parsePopupDate(container.Find("span.popup_date").First())
		}

		body := container.Find("comment-user-text").First()
		if body.Length() == 0 {
			body = container.Find(".comment_text, .comment-user-text").First()
		}
		shout.BodyHTML = htmlOf(body)

		if shout.Author.DisplayName != "" || shout.BodyHTML != "" {
			u.Shouts = append(u.Shouts, shout)
		}
	})

	// Watch state: the header carries a Watch/Unwatch button when the viewer
	// is logged in and looking at another user's page. An "/unwatch/" link
	// means the viewer currently watches this user.
	if _, unwatch := findWatchLinks(doc, u.Name); unwatch != "" {
		u.Watched = true
	}

	// Site banner: the <site-banner> element in the page header holds either
	// the artist's own banner (URL under /art/<name>/, uploaded via
	// /controls/profilebanner/) or when none is set FA's site-wide promo
	// banner (URL under /media/banners/).
	if banner := doc.Find("site-banner img").First(); banner.Length() > 0 {
		src := urls.AbsoluteCDN(trimAttr(banner, "src"))
		if src != "" {
			u.SiteBanner = &SiteBanner{
				ImageURL: src,
				IsCustom: strings.Contains(src, "/art/"+u.Name+"/"),
			}
		}
	}

	return u, nil
}

// firstNonEmptySel returns the first selection matching any of the selectors,
// or nil if none match. Useful for parser code that needs to tolerate
// alternate beta-theme markup.
func firstNonEmptySel(doc *goquery.Document, selectors ...string) *goquery.Selection {
	for _, sel := range selectors {
		s := doc.Find(sel).First()
		if s.Length() > 0 {
			return s
		}
	}
	return nil
}