Merge remote-tracking branch 'backend/main'

1 files changed, 99 insertions, 0 deletions

diff --git a/backend/internal/ibd/html_helpers.go b/backend/internal/ibd/html_helpers.go
new file mode 100644
index 0000000..0176bc5
--- /dev/null
+++ b/backend/internal/ibd/html_helpers.go
@@ -0,0 +1,99 @@
+package ibd
+
+import (
+	"strings"
+
+	"golang.org/x/net/html"
+)
+
+func findChildren(node *html.Node, f func(node *html.Node) bool) (found []*html.Node) {
+	for c := node.FirstChild; c != nil; c = c.NextSibling {
+		if f(c) {
+			found = append(found, c)
+		}
+	}
+	return
+}
+
+func findChildrenRecursive(node *html.Node, f func(node *html.Node) bool) (found []*html.Node) {
+	if f(node) {
+		found = append(found, node)
+	}
+
+	for c := node.FirstChild; c != nil; c = c.NextSibling {
+		found = append(found, findChildrenRecursive(c, f)...)
+	}
+
+	return
+}
+
+func findClass(node *html.Node, className string) (found *html.Node) {
+	if isClass(node, className) {
+		return node
+	}
+
+	for c := node.FirstChild; c != nil; c = c.NextSibling {
+		if found = findClass(c, className); found != nil {
+			return
+		}
+	}
+
+	return
+}
+
+func isClass(node *html.Node, className string) bool {
+	if node.Type == html.ElementNode {
+		for _, attr := range node.Attr {
+			if attr.Key != "class" {
+				continue
+			}
+			classes := strings.Fields(attr.Val)
+			for _, class := range classes {
+				if class == className {
+					return true
+				}
+			}
+		}
+	}
+	return false
+}
+
+func extractText(node *html.Node) string {
+	var result strings.Builder
+	extractTextInner(node, &result)
+	return result.String()
+}
+
+func extractTextInner(node *html.Node, result *strings.Builder) {
+	if node.Type == html.TextNode {
+		result.WriteString(node.Data)
+	}
+	for c := node.FirstChild; c != nil; c = c.NextSibling {
+		extractTextInner(c, result)
+	}
+}
+
+func findId(node *html.Node, id string) (found *html.Node) {
+	if isId(node, id) {
+		return node
+	}
+
+	for c := node.FirstChild; c != nil; c = c.NextSibling {
+		if found = findId(c, id); found != nil {
+			return
+		}
+	}
+
+	return
+}
+
+func isId(node *html.Node, id string) bool {
+	if node.Type == html.ElementNode {
+		for _, attr := range node.Attr {
+			if attr.Key == "id" && attr.Val == id {
+				return true
+			}
+		}
+	}
+	return false
+}