brain/internal/handlers/page_graph.go

package handlers

import (
	"context"
	"database/sql"
	"net/url"
	"regexp"
	"strings"

	"github.com/andrew/go-computer/internal/db"
)

// Connected pages (Notion-like upgrades phase 3).
//
// Page hierarchy, internal links and embedded collection views are all
// expressed in the page's Markdown, which stays the single source of truth:
//
//   - [Any label](computer:page/<id>) links to another page. /pages?id=<id>
//     and page/<id> are accepted as aliases so hand-written links work too.
//   - {{collection:<id>}} embeds a live collection view; an optional
//     {{collection:<id>|view=<viewId>}} pins a saved view's filters/sorts.
//
// On every content save the resolved PageLink and PageEmbed rows are rebuilt
// inside the same transaction. Only targets owned by the author are stored:
// backlinks therefore cannot enumerate or reveal titles from another account.
// Embedding displays data through the normal collection endpoints, so the
// underlying bookmark-visibility checks still apply and nothing is published.

var (
	markdownLinkRE = regexp.MustCompile(`\[[^\]]*\]\(\s*<?([^)\s>]+)(?:\s+"[^"]*")?\s*>?\)`)
	pageEmbedRE    = regexp.MustCompile(`\{\{\s*collection:([A-Za-z0-9_-]{1,128})(?:\s*\|\s*view=([A-Za-z0-9_-]{1,128}))?\s*\}\}`)
	pageRefRE      = regexp.MustCompile(`^[A-Za-z0-9_-]{1,128}$`)
)

// internalPageID normalizes a Markdown link target to a page id. It returns
// false for ordinary web links and malformed ids.
func internalPageID(href string) (string, bool) {
	href = strings.TrimSpace(strings.TrimSuffix(strings.TrimPrefix(href, "<"), ">"))
	raw := ""
	switch {
	case strings.HasPrefix(href, "computer://page/"):
		raw = strings.TrimPrefix(href, "computer://page/")
	case strings.HasPrefix(href, "computer:page/"):
		raw = strings.TrimPrefix(href, "computer:page/")
	case strings.HasPrefix(href, "page/"):
		raw = strings.TrimPrefix(href, "page/")
	case strings.HasPrefix(href, "/pages/"):
		raw = strings.TrimPrefix(href, "/pages/")
	case strings.HasPrefix(href, "/pages?"):
		if u, err := url.Parse(href); err == nil {
			raw = u.Query().Get("id")
		}
	case strings.HasPrefix(href, "computer:page?id="):
		if u, err := url.Parse(href); err == nil {
			raw = u.Query().Get("id")
		}
	case strings.HasPrefix(href, "computer://page?id="):
		if u, err := url.Parse(href); err == nil {
			raw = u.Query().Get("id")
		}
	}
	raw = strings.TrimSuffix(strings.TrimSpace(raw), "/")
	if !pageRefRE.MatchString(raw) {
		return "", false
	}
	return raw, true
}

type pageEmbedRef struct {
	collectionID string
	viewID       string
}

// pageGraph extracts the distinct page links and embeds from Markdown.
func pageGraph(content, selfID string) ([]string, []pageEmbedRef) {
	links := make([]string, 0)
	seenLinks := map[string]bool{}
	for _, m := range markdownLinkRE.FindAllStringSubmatchIndex(content, -1) {
		// Skip images: ![alt](target) is not an internal link.
		if m[0] > 0 && content[m[0]-1] == '!' {
			continue
		}
		id, ok := internalPageID(content[m[2]:m[3]])
		if !ok || id == selfID || seenLinks[id] {
			continue
		}
		seenLinks[id] = true
		links = append(links, id)
	}
	embeds := make([]pageEmbedRef, 0)
	seenEmbeds := map[string]bool{}
	for _, m := range pageEmbedRE.FindAllStringSubmatch(content, -1) {
		ref := pageEmbedRef{collectionID: m[1], viewID: m[2]}
		if seenEmbeds[ref.collectionID] {
			continue
		}
		seenEmbeds[ref.collectionID] = true
		embeds = append(embeds, ref)
	}
	return links, embeds
}

// syncPageGraph replaces one page's resolved links and embeds. It runs inside
// the content-save transaction, so the derived rows and the Markdown commit
// together (a failed save never leaves stale backlinks).
func syncPageGraph(ctx context.Context, tx *sql.Tx, pageID, uid, content string) error {
	links, embeds := pageGraph(content, pageID)

	if _, err := tx.ExecContext(ctx, `DELETE FROM "PageLink" WHERE "sourcePageId"=$1`, pageID); err != nil {
		return err
	}
	for _, target := range links {
		var visible bool
		if err := tx.QueryRowContext(ctx, `SELECT EXISTS(SELECT 1 FROM "Page" WHERE id=$1 AND "ownerId"=$2)`, target, uid).Scan(&visible); err != nil {
			return err
		}
		// A foreign/missing target is ignored rather than stored. The link
		// text remains in the author's Markdown, but it can never become a
		// cross-account backlink.
		if !visible {
			continue
		}
		if _, err := tx.ExecContext(ctx, `INSERT INTO "PageLink"("sourcePageId","targetPageId") VALUES($1,$2) ON CONFLICT DO NOTHING`, pageID, target); err != nil {
			return err
		}
	}

	if _, err := tx.ExecContext(ctx, `DELETE FROM "PageEmbed" WHERE "pageId"=$1`, pageID); err != nil {
		return err
	}
	for i, ref := range embeds {
		var visible bool
		if err := tx.QueryRowContext(ctx, `SELECT EXISTS(SELECT 1 FROM "Collection" WHERE id=$1 AND "ownerId"=$2)`, ref.collectionID, uid).Scan(&visible); err != nil {
			return err
		}
		if !visible {
			continue
		}
		viewID := any(nil)
		if ref.viewID != "" {
			var valid bool
			if err := tx.QueryRowContext(ctx, `SELECT EXISTS(SELECT 1 FROM "CollectionView" WHERE id=$1 AND "collectionId"=$2)`, ref.viewID, ref.collectionID).Scan(&valid); err != nil {
				return err
			}
			// An unknown view falls back to the collection's default view
			// instead of dropping the embed entirely.
			if valid {
				viewID = ref.viewID
			}
		}
		if _, err := tx.ExecContext(ctx, `INSERT INTO "PageEmbed"("pageId","collectionId","viewId",position) VALUES($1,$2,$3,$4)`, pageID, ref.collectionID, viewID, i); err != nil {
			return err
		}
	}
	return nil
}

// pageParent validates a requested parent and rejects cycles. The walk stays
// within the owner's pages and is bounded by the seen set, so a corrupt
// pre-existing cycle cannot loop forever.
func pageParent(ctx context.Context, q db.Queryer, pageID, uid string, raw any) (any, error) {
	if raw == nil {
		return nil, nil
	}
	parent, ok := raw.(string)
	if !ok {
		return nil, bad("parentId must be a page id or null")
	}
	parent = strings.TrimSpace(parent)
	if parent == "" {
		return nil, nil
	}
	if !pageRefRE.MatchString(parent) {
		return nil, bad("Invalid parentId")
	}
	if parent == pageID {
		return nil, bad("A page cannot be its own parent")
	}
	seen := map[string]bool{pageID: true}
	for current := parent; current != ""; {
		if seen[current] {
			return nil, bad("Page hierarchy cannot contain a cycle")
		}
		seen[current] = true
		var next sql.NullString
		err := q.QueryRowContext(ctx, `SELECT "parentId" FROM "Page" WHERE id=$1 AND "ownerId"=$2`, current, uid).Scan(&next)
		if err == sql.ErrNoRows {
			return nil, bad("Parent page not found")
		}
		if err != nil {
			return nil, err
		}
		current = next.String
	}
	return parent, nil
}