mattermost-community-enterp.../vendor/github.com/JalfResi/justext/preprocess.go
Claude ec1f89217a Merge: Complete Mattermost Server with Community Enterprise
Full Mattermost server source with integrated Community Enterprise features.
Includes vendor directory for offline/air-gapped builds.

Structure:
- enterprise-impl/: Enterprise feature implementations
- enterprise-community/: Init files that register implementations
- enterprise/: Bridge imports (community_imports.go)
- vendor/: All dependencies for offline builds

Build (online):
  go build ./cmd/mattermost

Build (offline/air-gapped):
  go build -mod=vendor ./cmd/mattermost

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2025-12-17 23:59:07 +09:00

112 lines
2.4 KiB
Go

package justext
import (
"github.com/levigross/exp-html"
"regexp"
"strings"
)
func preprocess(htmlStr, encoding, defaultEncoding, encErrors string) (*html.Node, error) {
root, err := html.Parse(strings.NewReader(htmlStr))
if err != nil {
return nil, err
}
addKwTags(root)
removeElements(root, []string{"head", "script", "style"})
return root, nil
}
type nodeIterator func(n *html.Node)
func nodeIter(n *html.Node, f nodeIterator) {
f(n)
for _, c := range n.Child {
nodeIter(c, f)
}
}
func addKwTags(root *html.Node) *html.Node {
var blankText *regexp.Regexp = regexp.MustCompile("^[\n\r\t ]*$")
var nodesWithText []*html.Node
var markTextAndTail nodeIterator
markTextAndTail = func(node *html.Node) {
if node.Type != html.CommentNode || node.Type != html.DoctypeNode {
if node.Type == html.TextNode {
nodesWithText = append(nodesWithText, node)
}
}
}
nodeIter(root, markTextAndTail)
for _, node := range nodesWithText {
if blankText.MatchString(node.Data) {
node.Data = ""
} else {
kw := &html.Node{
Parent: nil,
Type: html.ElementNode,
Data: "kw",
}
node2 := CopyNode(node, true)
kw.Child = append(kw.Child, node2)
insertNode(node, kw)
node.Parent.Remove(node)
}
}
return root
}
func removeElements(root *html.Node, elementsToRemove []string) {
var toBeRemoved []*html.Node
var markRemovableNodes = func(node *html.Node) {
if node.Type == html.ElementNode {
for _, nodeName := range elementsToRemove {
if node.Data == nodeName {
toBeRemoved = append(toBeRemoved, node)
}
}
}
}
nodeIter(root, markRemovableNodes)
for _, node := range toBeRemoved {
node.Parent.Remove(node)
}
}
// insertsNode inserts a Node in a Node tree at the position of another node.
// Should be moved into html/utils package
func insertNode(originalNode *html.Node, newNode *html.Node) {
slice := originalNode.Parent.Child
for position, n := range slice {
if n == originalNode {
originalNode.Parent.Child = append(slice[:position], append([]*html.Node{newNode}, slice[position:]...)...)
return
}
}
}
func CopyNode(node *html.Node, deep bool) *html.Node {
newNode := &html.Node{
Type: node.Type,
Data: node.Data,
}
if deep && len(node.Child) > 0 {
for _, n := range node.Child {
newNode.Child = append(newNode.Child, CopyNode(n, true))
}
}
for _, i := range node.Attr {
newNode.Attr = append(newNode.Attr, html.Attribute{Key: i.Key, Val: i.Val})
}
return newNode
}