inital commit

listing_json.go

@@ -0,0 +1,68 @@
+package fa
+
+import (
+	"encoding/json"
+	"strings"
+
+	"github.com/PuerkitoBio/goquery"
+)
+
+// listingJSONEntry is one row of FA's embedded js-submissionData blob —
+// a JSON dictionary every listing page emits with stable metadata for
+// each visible submission. The blob is keyed by submission ID (string)
+// and survives most HTML reshuffles because it's consumed by FA's own
+// front-end JS.
+//
+// Not every field appears on every page; description is sometimes truncated
+// to ~500 chars with a trailing "…". We use this purely as a hinting
+// source the HTML figure scrape remains authoritative for fields the
+// JSON doesn't carry (ID, rating, thumbnail).
+type listingJSONEntry struct {
+	Title       string `json:"title"`
+	Description string `json:"description"` // BBCode source (may be truncated)
+	Username    string `json:"username"`    // display name with original casing
+	Lower       string `json:"lower"`       // URL-safe lowercase login
+	AvatarMtime string `json:"avatar_mtime"`
+}
+
+// listingJSONMap is the parsed js-submissionData blob: SubmissionID → entry.
+type listingJSONMap map[SubmissionID]listingJSONEntry
+
+// readListingJSON pulls and parses the <script id="js-submissionData">
+// blob from doc. Returns nil if the tag is absent or the JSON is
+// malformed the caller then falls back to HTML-only scraping. The
+// JSON is keyed by string in the source; we deserialize directly into a
+// SubmissionID map by exploiting the typed int unmarshal path.
+func readListingJSON(doc *goquery.Document) listingJSONMap {
+	raw := strings.TrimSpace(doc.Find("script#js-submissionData").First().Text())
+	if raw == "" {
+		return nil
+	}
+	// Decode into string-keyed map first because Go's json package only
+	// accepts string-convertible keys from JSON object keys.
+	var stringKeyed map[string]listingJSONEntry
+	if err := json.Unmarshal([]byte(raw), &stringKeyed); err != nil {
+		return nil
+	}
+	out := make(listingJSONMap, len(stringKeyed))
+	for k, v := range stringKeyed {
+		id, err := parseID[SubmissionID](k)
+		if err != nil || id == 0 {
+			continue
+		}
+		out[id] = v
+	}
+	if len(out) == 0 {
+		return nil
+	}
+	return out
+}
+
+// avatarURLFromMtime builds FA's avatar CDN URL from the mtime + lowercase
+// login. Returns "" when either piece is missing.
+func avatarURLFromMtime(lower, mtime string) string {
+	if lower == "" || mtime == "" {
+		return ""
+	}
+	return "https://a.furaffinity.net/" + mtime + "/" + lower + ".gif"
+}