Merge pull request #185 from Southclaws/mentions

add mentions in content and notify members when mentioned in threads

api/openapi.yaml

@@ -3069,7 +3069,7 @@ components:
         The kind of event that triggered the notification.
         Identical to the `notification.Event` enumerated type.
       type: string
-      enum: [thread_reply, post_like, follow]
+      enum: [thread_reply, post_like, follow, profile_mention]
 
     NotificationStatus:
       type: string

app/resources/account/account.go

@@ -7,9 +7,8 @@ import (
 	"github.com/Southclaws/fault"
 	"github.com/Southclaws/fault/ftag"
 	"github.com/Southclaws/opt"
+	"github.com/Southclaws/storyden/app/resources/datagraph"
 	"github.com/rs/xid"
-
-	"github.com/Southclaws/storyden/app/resources/content"
 )
 
 var errSuspended = fault.Wrap(fault.New("suspended"), ftag.With(ftag.PermissionDenied))
@@ -22,7 +21,7 @@ type Account struct {
 	ID             AccountID
 	Handle         string
 	Name           string
-	Bio            content.Rich
+	Bio            datagraph.Content
 	Admin          bool
 	Followers      int
 	Following      int

app/resources/account/account_writer/account_writer.go

@@ -13,7 +13,7 @@ import (
 	"go.uber.org/fx"
 
 	"github.com/Southclaws/storyden/app/resources/account"
-	"github.com/Southclaws/storyden/app/resources/content"
+	"github.com/Southclaws/storyden/app/resources/datagraph"
 	"github.com/Southclaws/storyden/internal/ent"
 	"github.com/Southclaws/storyden/internal/ent/schema"
 )
@@ -47,7 +47,7 @@ func WithName(name string) Option {
 	}
 }
 
-func WithBio(v content.Rich) Option {
+func WithBio(v datagraph.Content) Option {
 	return func(a *account.Account) {
 		a.Bio = v
 	}

app/resources/account/mapping.go

@@ -8,7 +8,7 @@ import (
 	"github.com/Southclaws/fault"
 	"github.com/Southclaws/opt"
 
-	"github.com/Southclaws/storyden/app/resources/content"
+	"github.com/Southclaws/storyden/app/resources/datagraph"
 	"github.com/Southclaws/storyden/internal/ent"
 	"github.com/Southclaws/storyden/internal/ent/schema"
 )
@@ -18,7 +18,7 @@ func MapAccount(a *ent.Account) (*Account, error) {
 		return a.Service
 	})
 
-	bio, err := content.NewRichText(a.Bio)
+	bio, err := datagraph.NewRichText(a.Bio)
 	if err != nil {
 		return nil, err
 	}

app/resources/account/notification/event.go

@@ -5,7 +5,8 @@ package notification
 type eventEnum string
 
 const (
-	eventThreadReply eventEnum = "thread_reply"
-	eventPostLike    eventEnum = "post_like"
-	eventFollow      eventEnum = "follow"
+	eventThreadReply    eventEnum = "thread_reply"
+	eventPostLike       eventEnum = "post_like"
+	eventFollow         eventEnum = "follow"
+	eventProfileMention eventEnum = "profile_mention"
 )

app/resources/asset/asset.go

@@ -7,23 +7,22 @@ import (
 	"github.com/Southclaws/opt"
 	"github.com/rs/xid"
 
-	"github.com/Southclaws/storyden/app/resources/account"
 	"github.com/Southclaws/storyden/internal/ent"
 )
 
 var errInvalidFormat = fault.New("invalid format")
 
 type Repository interface {
 	Add(ctx context.Context,
-		owner account.AccountID,
+		owner xid.ID,
 		filename Filename,
 		size int,
 	) (*Asset, error)
 
 	Get(ctx context.Context, id Filename) (*Asset, error)
 	GetByID(ctx context.Context, id AssetID) (*Asset, error)
 
-	Remove(ctx context.Context, owner account.AccountID, id Filename) error
+	Remove(ctx context.Context, owner xid.ID, id Filename) error
 }
 
 type AssetID = xid.ID

app/resources/asset/db.go

@@ -8,7 +8,6 @@ import (
 	"github.com/Southclaws/fault/ftag"
 	"github.com/rs/xid"
 
-	"github.com/Southclaws/storyden/app/resources/account"
 	"github.com/Southclaws/storyden/internal/ent"
 	"github.com/Southclaws/storyden/internal/ent/asset"
 )
@@ -22,7 +21,7 @@ func New(db *ent.Client) Repository {
 }
 
 func (d *database) Add(ctx context.Context,
-	accountID account.AccountID,
+	accountID xid.ID,
 	filename Filename,
 	size int,
 ) (*Asset, error) {
@@ -68,7 +67,7 @@ func (d *database) GetByID(ctx context.Context, id AssetID) (*Asset, error) {
 	return FromModel(asset), nil
 }
 
-func (d *database) Remove(ctx context.Context, accountID account.AccountID, id Filename) error {
+func (d *database) Remove(ctx context.Context, accountID xid.ID, id Filename) error {
 	q := d.db.Asset.
 		Delete().Where(
 		asset.Filename(id.name),

app/resources/content/content.go → app/resources/datagraph/content.go

@@ -1,21 +1,44 @@
-package content
+package datagraph
 
 import (
 	"bytes"
 	"fmt"
+	"io"
 	"math"
 	"net/url"
+	"regexp"
 	"strings"
 	"unicode"
 
 	"github.com/Southclaws/fault"
+	"github.com/cixtor/readability"
 	"github.com/microcosm-cc/bluemonday"
 	"github.com/samber/lo"
+	"go.uber.org/zap"
 	"golang.org/x/net/html"
 	"golang.org/x/net/html/atom"
 )
 
-var policy = bluemonday.UGCPolicy()
+// RefScheme is used as a scheme for URIs to reference resources.
+// These can be used in content to refer to profiles, posts, nodes, etc.
+const RefScheme = "sdr"
+
+var policy = func() *bluemonday.Policy {
+	p := bluemonday.UGCPolicy()
+
+	p.AllowURLSchemes(
+		"mailto",
+		"http",
+		"https",
+		RefScheme,
+	)
+
+	p.AllowDataAttributes()
+
+	return p
+}()
+
+var spaces = regexp.MustCompile(`\s+`)
 
 // MaxSummaryLength is the maximum length of the short summary text
 const MaxSummaryLength = 128
@@ -24,14 +47,15 @@ const MaxSummaryLength = 128
 // string but on the read path, it'll get turned into this either way.
 const EmptyState = `<body></body>`
 
-type Rich struct {
+type Content struct {
 	html  *html.Node
 	short string
 	links []string
+	sdrs  RefList
 }
 
-func (r Rich) HTML() string {
-	if r.html == nil {
+func (r Content) HTML() string {
+	if r.html == nil || r.html.FirstChild == nil {
 		return EmptyState
 	}
 
@@ -45,34 +69,72 @@ func (r Rich) HTML() string {
 	return w.String()
 }
 
-func (r Rich) HTMLTree() *html.Node {
+func (r Content) HTMLTree() *html.Node {
 	return r.html
 }
 
-func (r Rich) Short() string {
+func (r Content) Short() string {
 	return r.short
 }
 
-func (r Rich) Links() []string {
+func (r Content) Links() []string {
 	return r.links
 }
 
+func (r Content) References() RefList {
+	return r.sdrs
+}
+
+type options struct {
+	baseURL string
+}
+type option func(*options)
+
 // NewRichText will pull out any meaningful structured information from markdown
 // document this includes a summary of the text and all link URLs for hydrating.
-func NewRichText(raw string) (Rich, error) {
-	sanitised := policy.Sanitize(raw)
-	htmlTree, err := html.Parse(strings.NewReader(sanitised))
+func NewRichText(raw string) (Content, error) {
+	return NewRichTextFromReader(strings.NewReader(raw))
+}
+
+func NewRichTextFromReader(r io.Reader, opts ...option) (Content, error) {
+	o := options{baseURL: "ignore:"}
+	for _, opt := range opts {
+		opt(&o)
+	}
+
+	buf, err := io.ReadAll(r)
+	if err != nil {
+		return Content{}, fault.Wrap(err)
+	}
+
+	sanitised := policy.SanitizeBytes(buf)
+
+	htmlTree, err := html.Parse(bytes.NewReader(sanitised))
+	if err != nil {
+		return Content{}, fault.Wrap(err)
+	}
+
+	result, err := readability.New().Parse(bytes.NewReader(sanitised), o.baseURL)
 	if err != nil {
-		return Rich{}, fault.Wrap(err)
+		return Content{}, fault.Wrap(err)
 	}
 
-	return NewRichTextFromHTML(htmlTree)
+	short := getSummary(result)
+
+	bodyTree, links, refs := extractReferences(htmlTree)
+
+	return Content{
+		html:  bodyTree,
+		short: short,
+		links: links,
+		sdrs:  refs,
+	}, nil
 }
 
-func NewRichTextFromHTML(htmlTree *html.Node) (Rich, error) {
+func extractReferences(htmlTree *html.Node) (*html.Node, []string, RefList) {
 	bodyTree := &html.Node{}
-	textonly := strings.Builder{}
 	links := []string{}
+	sdrs := []url.URL{}
 
 	if htmlTree.DataAtom == atom.Body {
 		bodyTree = htmlTree
@@ -89,18 +151,14 @@ func NewRichTextFromHTML(htmlTree *html.Node) (Rich, error) {
 
 				if hasHref {
 					if parsed, err := url.Parse(href.Val); err == nil {
-						links = append(links, parsed.String())
+						switch parsed.Scheme {
+						case "http", "https":
+							links = append(links, parsed.String())
+						case RefScheme:
+							sdrs = append(sdrs, *parsed)
+						}
 					}
 				}
-
-			case atom.P:
-				if n.Type == html.TextNode && len(n.Data) > 0 {
-
-					oneline := strings.ReplaceAll(n.Data, "\n", " ")
-					textonly.Write([]byte(oneline))
-					textonly.WriteByte(' ')
-					return
-				}
 			}
 		}
 
@@ -115,7 +173,24 @@ func NewRichTextFromHTML(htmlTree *html.Node) (Rich, error) {
 	}
 	walk(htmlTree)
 
-	paragraphs := []rune(strings.TrimSpace(textonly.String()))
+	var refs RefList
+	for _, v := range sdrs {
+		r, err := NewRefFromSDR(v)
+		if err != nil {
+			zap.L().Warn("invalid SDR in content", zap.Error(err))
+			continue
+		}
+		refs = append(refs, r)
+	}
+
+	return bodyTree, links, refs
+}
+
+func getSummary(article readability.Article) string {
+	trimmed := strings.TrimSpace(article.TextContent)
+	collapsed := spaces.ReplaceAllString(trimmed, " ")
+
+	paragraphs := []rune(collapsed)
 	end := int(math.Min(float64(len(paragraphs)-1), MaxSummaryLength))
 
 	var short string
@@ -147,9 +222,5 @@ func NewRichTextFromHTML(htmlTree *html.Node) (Rich, error) {
 		short = string(paragraphs)
 	}
 
-	return Rich{
-		html:  bodyTree,
-		short: short,
-		links: links,
-	}, nil
+	return short
 }