inital commit

journal_parser.go

@@ -0,0 +1,96 @@
+package fa
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/PuerkitoBio/goquery"
+
+	"git.anthrove.art/public/go-fa-api/internal/urls"
+)
+
+// parseJournal lifts a [Journal] from a /journal/{id}/ document. FA renders
+// the journal view inside the author's profile shell, so the author is
+// derived from the userpage nav rather than from any byline in the journal
+// body itself.
+func parseJournal(id JournalID, doc *goquery.Document) (*Journal, error) {
+	j := &Journal{ID: id}
+
+	// Title.
+	j.Title = trimText(doc.Find("#c-journalTitleTop__subject h3").First())
+	if j.Title == "" {
+		j.Title = firstNonEmpty(
+			trimText(doc.Find("h2.journal-title").First()),
+			trimText(doc.Find("h3.journal-title").First()),
+		)
+	}
+	if j.Title == "" {
+		return nil, fmt.Errorf("%w: journal %d: missing title", ErrParse, id)
+	}
+
+	// Author from the userpage nav at the top of the rendered page.
+	authorLink := doc.Find("a.c-usernameBlock__displayName[href^='/user/']").First()
+	if authorLink.Length() > 0 {
+		href, _ := authorLink.Attr("href")
+		j.Author = UserRef{
+			DisplayName: trimText(authorLink.Find("span.js-displayName").First()),
+			AvatarURL:   urls.AbsoluteCDN(trimAttr(doc.Find("img.user-nav-avatar").First(), "src")),
+		}
+		if j.Author.DisplayName == "" {
+			j.Author.DisplayName = trimText(authorLink)
+		}
+		if parts := strings.Split(strings.Trim(href, "/"), "/"); len(parts) >= 2 {
+			j.Author.Name = strings.ToLower(parts[1])
+		}
+	}
+
+	// Date from the journal title block.
+	j.PostedAt = parsePopupDate(doc.Find("#c-journalTitleTop__date span.popup_date").First())
+	if j.PostedAt.IsZero() {
+		j.PostedAt = parsePopupDate(doc.Find("span.popup_date").First())
+	}
+
+	// Body.
+	body := firstNonEmptySel(doc,
+		"div.section-body.journal-body-theme div.journal-content",
+		"div.journal-content",
+		"div.journal-body",
+		"section.journal-body",
+	)
+	if body != nil {
+		j.BodyHTML = htmlOf(body)
+		j.BodyText = strings.TrimSpace(body.Text())
+	}
+
+	return j, nil
+}
+
+// parseUserJournalsPage parses a /journals/{user}/[page]/ listing page,
+// returning the journal entries it contains and whether a next page exists.
+//
+// FA renders each entry inside the listing differently from the standalone
+// journal page; selectors here target the listing's tile structure.
+func parseUserJournalsPage(doc *goquery.Document) (entries []*Journal, hasNext bool) {
+	doc.Find("section.journal, section[id^=jid], div.journal[id^=jid]").Each(func(_ int, sel *goquery.Selection) {
+		j := &Journal{}
+		idAttr, _ := sel.Attr("id")
+		idAttr = strings.TrimPrefix(idAttr, "jid:")
+		idAttr = strings.TrimPrefix(idAttr, "journal-")
+		if n, err := parseID[JournalID](strings.TrimSpace(idAttr)); err == nil {
+			j.ID = n
+		}
+		j.Title = firstNonEmpty(
+			trimText(sel.Find("h2 a, h3 a").First()),
+			trimText(sel.Find("h2, h3").First()),
+		)
+		j.PostedAt = parsePopupDate(sel.Find("span.popup_date").First())
+		body := sel.Find("div.journal-body, div.journal-content").First()
+		j.BodyHTML = htmlOf(body)
+		j.BodyText = strings.TrimSpace(body.Text())
+		if j.ID != 0 || j.Title != "" {
+			entries = append(entries, j)
+		}
+	})
+	hasNext = detectNextPage(doc)
+	return entries, hasNext
+}