Phase 3: parse stored Compass pages into structured items

Turn the raw pages stored by Phase 2 into rows in the SQLite items table
(AGENTS.md §11 Phase 3, FR-3/FR-4/FR-5/FR-12).

- internal/parse: dependency-free HTML reader plus extractors for the ExtJS
  learning-task grid (subject, title, category, kind, due date, status), the
  home news feed (author, posted time, title, body) and the calendar week grid
  (day column + grid offset give the time; label gives title/room/teacher).
- internal/parse: fail loudly. A challenge page, login page, changed layout or
  unknown due-date format is an explicit error, never an empty day.
- FR-4: structure is parsed deterministically; importance is classified by an
  OmniRoute model (default openrouter/google/gemini-3.5-flash-lite) with a
  deterministic fallback. The classifier used is printed and recorded.
- internal/llm: small OpenAI-compatible OmniRoute client (handles SSE replies).
- internal/store: migration 002 adds source_page, importance and external_ref
  with a unique (child_id, source, type, external_ref) index; UpsertItems,
  CountItemsByType, SampleItems and ChildByName.
- cmd/school parse: reads var/raw/<child>/<run>, upserts items, records the scan
  row (ok=0 plus the error on failure) and prints counts and samples.
- scripts/parse-compass.sh and make parse run it in a one-off container against
  the service's own volume without restarting the service.
- docs/phase-3-parse-notes.md records the approach, the FR-4 decision and the
  known gaps.

Evidence on the live database: 56 rows for Harry (20 task, 1 message, 35 event).
Re-running the same raw pages inserts 0 and updates 56.

Co-Authored-By: Paperclip <noreply@paperclip.ing>

internal/parse/news.go

@@ -0,0 +1,102 @@
+package parse
+
+import (
+	"fmt"
+	"strings"
+	"time"
+)
+
+// newsTimeLayout is the aria-label format on the feed timestamp, for example
+// "14/10/2026 09:00 AM".
+const newsTimeLayout = "02/01/2006 03:04 PM"
+
+// parseNews parses the home-page news feed into message items. It fails loudly
+// when the page carries neither the feed toolbar nor a post, so a login or
+// challenge page is never mistaken for "no news" (FR-12).
+func parseNews(child string, page Page, loc *time.Location) ([]Item, error) {
+	items, found, err := extractNews(child, page, loc)
+	if err != nil {
+		return nil, err
+	}
+	if !found {
+		return nil, fmt.Errorf("parse: %s: no news feed or posts found in %s (layout changed or fetch failed)", child, page.Slug)
+	}
+	return items, nil
+}
+
+// extractNews returns news posts from the home page. found reports whether the
+// feed was present at all.
+func extractNews(child string, page Page, loc *time.Location) (items []Item, found bool, err error) {
+	root := parseHTML(page.HTML)
+
+	toolbar := find(root, func(n *node) bool {
+		return strings.EqualFold(n.Attrs["data-testid"], "feed-toolbar")
+	})
+	cards := findAll(root, func(n *node) bool { return n.hasClass("MuiCardContent-root") })
+	found = toolbar != nil || len(cards) > 0
+	if !found {
+		return nil, false, nil
+	}
+
+	for _, card := range cards {
+		titleNode := find(card, func(n *node) bool { return n.hasClassPart("postTitle") })
+		if titleNode == nil {
+			continue // not a news post
+		}
+		title := collapse(titleNode.Text())
+		if title == "" {
+			continue
+		}
+
+		var body string
+		if b := find(card, func(n *node) bool { return n.hasClassPart("froala-viewer") }); b != nil {
+			body = b.Text()
+		}
+
+		var author string
+		if a := find(card, func(n *node) bool { return n.hasClassPart("MuiTypography-dialogMdStrong") }); a != nil {
+			author = collapse(a.Text())
+		}
+
+		var posted *time.Time
+		if t := find(card, func(n *node) bool {
+			return n.hasClassPart("MuiTypography-dialogSmStrong") && n.hasAttr("aria-label")
+		}); t != nil {
+			raw := strings.TrimSpace(t.Attrs["aria-label"])
+			parsed, perr := time.ParseInLocation(newsTimeLayout, raw, loc)
+			if perr != nil {
+				return nil, false, fmt.Errorf("parse: %s: news %q: unrecognised posted time %q", page.Slug, title, raw)
+			}
+			posted = &parsed
+		}
+
+		pageURL := page.URL
+		if pageURL == "" {
+			pageURL = compassHost + "/"
+		}
+		ref := dedupeKey(author, title, relativeOrISO(posted))
+
+		items = append(items, Item{
+			Child:       child,
+			Source:      SourceCompass,
+			Type:        TypeMessage,
+			Teacher:     author, // messages carry the author, not a teacher
+			Title:       title,
+			Body:        body,
+			DueAt:       posted,
+			URL:         pageURL,
+			SourcePage:  page.Slug,
+			ExternalRef: ref,
+			Importance:  importanceNormal,
+		})
+	}
+	return items, true, nil
+}
+
+// relativeOrISO renders a posted time for the dedupe key.
+func relativeOrISO(t *time.Time) string {
+	if t == nil {
+		return ""
+	}
+	return t.Format(time.RFC3339)
+}