ferret/pkg/html/dynamic/helpers.go

package dynamic

import (
	"bytes"
	"context"
	"errors"
	"github.com/MontFerret/ferret/pkg/html/common"
	"github.com/MontFerret/ferret/pkg/html/dynamic/eval"
	"github.com/MontFerret/ferret/pkg/html/dynamic/events"
	"github.com/MontFerret/ferret/pkg/runtime/values"
	"github.com/PuerkitoBio/goquery"
	"github.com/mafredri/cdp"
	"github.com/mafredri/cdp/protocol/dom"
	"github.com/mafredri/cdp/protocol/page"
	"github.com/mafredri/cdp/protocol/runtime"
	"golang.org/x/sync/errgroup"
	"math"
	"strings"
)

type (
	batchFunc = func() error

	Quad struct {
		X float64
		Y float64
	}
)

func runBatch(funcs ...batchFunc) error {
	eg := errgroup.Group{}

	for _, f := range funcs {
		eg.Go(f)
	}

	return eg.Wait()
}

func getRootElement(ctx context.Context, client *cdp.Client) (*dom.GetDocumentReply, error) {
	d, err := client.DOM.GetDocument(ctx, dom.NewGetDocumentArgs().SetDepth(1))

	if err != nil {
		return nil, err
	}

	return d, nil
}

func fromProtocolQuad(quad dom.Quad) []Quad {
	return []Quad{
		{
			X: quad[0],
			Y: quad[1],
		},
		{
			X: quad[2],
			Y: quad[3],
		},
		{
			X: quad[4],
			Y: quad[5],
		},
		{
			X: quad[6],
			Y: quad[7],
		},
	}
}

func computeQuadArea(quads []Quad) float64 {
	var area float64

	for i := range quads {
		p1 := quads[i]
		p2 := quads[(i+1)%len(quads)]
		area += (p1.X*p2.Y - p2.X*p1.Y) / 2
	}

	return math.Abs(area)
}

func getClickablePoint(ctx context.Context, client *cdp.Client, id *HTMLElementIdentity) (Quad, error) {
	qargs := dom.NewGetContentQuadsArgs()

	if id.objectID != "" {
		qargs.SetObjectID(id.objectID)
	} else if id.backendID != 0 {
		qargs.SetBackendNodeID(id.backendID)
	} else {
		qargs.SetNodeID(id.nodeID)
	}

	res, err := client.DOM.GetContentQuads(ctx, qargs)

	if err != nil {
		return Quad{}, err
	}

	if res.Quads == nil || len(res.Quads) == 0 {
		return Quad{}, errors.New("node is either not visible or not an HTMLElement")
	}

	quads := make([][]Quad, 0, len(res.Quads))

	for _, q := range res.Quads {
		quad := fromProtocolQuad(q)

		if computeQuadArea(quad) > 1 {
			quads = append(quads, quad)
		}
	}

	if len(quads) == 0 {
		return Quad{}, errors.New("node is either not visible or not an HTMLElement")
	}

	// Return the middle point of the first quad.
	quad := quads[0]
	var x float64
	var y float64

	for _, q := range quad {
		x += q.X
		y += q.Y
	}

	return Quad{
		X: x / 4,
		Y: y / 4,
	}, nil
}

func parseAttrs(attrs []string) *values.Object {
	var attr values.String

	res := values.NewObject()

	for _, el := range attrs {
		el = strings.TrimSpace(el)
		str := values.NewString(el)

		if common.IsAttribute(el) {
			attr = str
			res.Set(str, values.EmptyString)
		} else {
			current, ok := res.Get(attr)

			if ok {
				if current.String() != "" {
					res.Set(attr, current.(values.String).Concat(values.SpaceString).Concat(str))
				} else {
					res.Set(attr, str)
				}
			}
		}
	}

	return res
}

func loadInnerHTML(ctx context.Context, client *cdp.Client, id *HTMLElementIdentity) (values.String, error) {
	var objID runtime.RemoteObjectID

	if id.objectID != "" {
		objID = id.objectID
	} else if id.backendID > 0 {
		repl, err := client.DOM.ResolveNode(ctx, dom.NewResolveNodeArgs().SetBackendNodeID(id.backendID))

		if err != nil {
			return "", err
		}

		if repl.Object.ObjectID == nil {
			return "", errors.New("unable to resolve node")
		}

		objID = *repl.Object.ObjectID
	} else {
		repl, err := client.DOM.ResolveNode(ctx, dom.NewResolveNodeArgs().SetNodeID(id.nodeID))

		if err != nil {
			return "", err
		}

		if repl.Object.ObjectID == nil {
			return "", errors.New("unable to resolve node")
		}

		objID = *repl.Object.ObjectID
	}

	// not a document
	if id.nodeID != 1 {
		res, err := eval.Property(ctx, client, objID, "innerHTML")

		if err != nil {
			return "", err
		}

		return values.NewString(res.String()), err
	}

	repl, err := client.DOM.GetOuterHTML(ctx, dom.NewGetOuterHTMLArgs().SetObjectID(objID))

	if err != nil {
		return "", err
	}

	return values.NewString(repl.OuterHTML), nil
}

func loadInnerText(ctx context.Context, client *cdp.Client, id *HTMLElementIdentity) (values.String, error) {
	var objID runtime.RemoteObjectID

	if id.objectID != "" {
		objID = id.objectID
	} else if id.backendID > 0 {
		repl, err := client.DOM.ResolveNode(ctx, dom.NewResolveNodeArgs().SetBackendNodeID(id.backendID))

		if err != nil {
			return "", err
		}

		if repl.Object.ObjectID == nil {
			return "", errors.New("unable to resolve node")
		}

		objID = *repl.Object.ObjectID
	} else {
		repl, err := client.DOM.ResolveNode(ctx, dom.NewResolveNodeArgs().SetNodeID(id.nodeID))

		if err != nil {
			return "", err
		}

		if repl.Object.ObjectID == nil {
			return "", errors.New("unable to resolve node")
		}

		objID = *repl.Object.ObjectID
	}

	// not a document
	if id.nodeID != 1 {
		res, err := eval.Property(ctx, client, objID, "innerText")

		if err != nil {
			return "", err
		}

		return values.NewString(res.String()), err
	}

	repl, err := client.DOM.GetOuterHTML(ctx, dom.NewGetOuterHTMLArgs().SetObjectID(objID))

	if err != nil {
		return "", err
	}

	return parseInnerText(repl.OuterHTML)
}

func parseInnerText(innerHTML string) (values.String, error) {
	buff := bytes.NewBuffer([]byte(innerHTML))

	parsed, err := goquery.NewDocumentFromReader(buff)

	if err != nil {
		return values.EmptyString, err
	}

	return values.NewString(parsed.Text()), nil
}

func createChildrenArray(nodes []dom.Node) []*HTMLElementIdentity {
	children := make([]*HTMLElementIdentity, len(nodes))

	for idx, child := range nodes {
		children[idx] = &HTMLElementIdentity{
			nodeID:    child.NodeID,
			backendID: child.BackendNodeID,
		}
	}

	return children
}

func contextWithTimeout() (context.Context, context.CancelFunc) {
	return context.WithTimeout(context.Background(), DefaultTimeout)
}

func waitForLoadEvent(ctx context.Context, client *cdp.Client) error {
	loadEventFired, err := client.Page.LoadEventFired(ctx)

	if err != nil {
		return err
	}

	_, err = loadEventFired.Recv()

	if err != nil {
		return err
	}

	return loadEventFired.Close()
}

func createEventBroker(client *cdp.Client) (*events.EventBroker, error) {
	var err error
	var onLoad page.LoadEventFiredClient
	var onReload dom.DocumentUpdatedClient
	var onAttrModified dom.AttributeModifiedClient
	var onAttrRemoved dom.AttributeRemovedClient
	var onChildCountUpdated dom.ChildNodeCountUpdatedClient
	var onChildNodeInserted dom.ChildNodeInsertedClient
	var onChildNodeRemoved dom.ChildNodeRemovedClient
	ctx := context.Background()

	onLoad, err = client.Page.LoadEventFired(ctx)

	if err != nil {
		return nil, err
	}

	onReload, err = client.DOM.DocumentUpdated(ctx)

	if err != nil {
		onLoad.Close()
		return nil, err
	}

	onAttrModified, err = client.DOM.AttributeModified(ctx)

	if err != nil {
		onLoad.Close()
		onReload.Close()
		return nil, err
	}

	onAttrRemoved, err = client.DOM.AttributeRemoved(ctx)

	if err != nil {
		onLoad.Close()
		onReload.Close()
		onAttrModified.Close()
		return nil, err
	}

	onChildCountUpdated, err = client.DOM.ChildNodeCountUpdated(ctx)

	if err != nil {
		onLoad.Close()
		onReload.Close()
		onAttrModified.Close()
		onAttrRemoved.Close()
		return nil, err
	}

	onChildNodeInserted, err = client.DOM.ChildNodeInserted(ctx)

	if err != nil {
		onLoad.Close()
		onReload.Close()
		onAttrModified.Close()
		onAttrRemoved.Close()
		onChildCountUpdated.Close()
		return nil, err
	}

	onChildNodeRemoved, err = client.DOM.ChildNodeRemoved(ctx)

	if err != nil {
		onLoad.Close()
		onReload.Close()
		onAttrModified.Close()
		onAttrRemoved.Close()
		onChildCountUpdated.Close()
		onChildNodeInserted.Close()
		return nil, err
	}

	broker := events.NewEventBroker(
		onLoad,
		onReload,
		onAttrModified,
		onAttrRemoved,
		onChildCountUpdated,
		onChildNodeInserted,
		onChildNodeRemoved,
	)

	err = broker.Start()

	if err != nil {
		onLoad.Close()
		onReload.Close()
		onAttrModified.Close()
		onAttrRemoved.Close()
		onChildCountUpdated.Close()
		onChildNodeInserted.Close()
		onChildNodeRemoved.Close()
		return nil, err
	}

	return broker, nil
}