From 472b2b67c342dd78fd4dd362b2cf66559e3cec8f Mon Sep 17 00:00:00 2001
From: "google-labs-jules[bot]"
 <161369871+google-labs-jules[bot]@users.noreply.github.com>
Date: Fri, 28 Nov 2025 11:56:42 +0000
Subject: [PATCH] Refactor HTML parser to treat p and h tags as blocks

- Updated `pkg/app/passage.go` to treat `p` and `h1`-`h4` tags as block elements, wrapping them in newlines.
- Ensured headers are still formatted as bold text using `platform.TelegramBold`.
- Cleaned up redundant logic and confirmed functionality with a reproduction test (which was deleted after verification).
- This ensures proper visual separation of paragraphs and headers in the Telegram output, preventing run-on text.
---
 pkg/app/passage.go | 30 +++++++++++++++++++++++++++---
 1 file changed, 27 insertions(+), 3 deletions(-)

diff --git a/pkg/app/passage.go b/pkg/app/passage.go
index c83ff51..7492bd1 100644
--- a/pkg/app/passage.go
+++ b/pkg/app/passage.go
@@ -82,7 +82,30 @@ func parseNode(node *html.Node) string {
 	if tag == "br" {
 		return "\n"
 	}
-	if !isFormattingTag(tag) && !isHeaderTag(tag) {
+
+	// Treat headers and paragraphs as block elements
+	if tag == "p" || isHeaderTag(tag) {
+		var content strings.Builder
+		content.WriteString("\n")
+
+		// Buffer to hold content of the block
+		var blockContent strings.Builder
+
+		for c := node.FirstChild; c != nil; c = c.NextSibling {
+			blockContent.WriteString(parseNode(c))
+		}
+
+		if isHeaderTag(tag) {
+			content.WriteString(platform.TelegramBold(blockContent.String()))
+		} else {
+			content.WriteString(blockContent.String())
+		}
+
+		content.WriteString("\n")
+		return content.String()
+	}
+
+	if !isFormattingTag(tag) {
 		var content strings.Builder
 		for c := node.FirstChild; c != nil; c = c.NextSibling {
 			content.WriteString(parseNode(c))
@@ -110,7 +133,8 @@ func parseNode(node *html.Node) string {
 	}
 
 	for c := node.FirstChild; c != nil; c = c.NextSibling {
-		if c.Type == html.ElementNode && (isFormattingTag(c.Data) || isHeaderTag(c.Data)) {
+		// Note: isHeaderTag is removed here because it's handled above as a block element
+		if c.Type == html.ElementNode && isFormattingTag(c.Data) {
 			flushTextBuffer()
 			content.WriteString(parseNode(c))
 		} else {
@@ -128,7 +152,7 @@ func ParsePassageFromHtml(rawHtml string) string {
 		log.Printf("Error parsing html: %v", err)
 		return rawHtml
 	}
-	return parseNode(doc)
+	return strings.TrimSpace(parseNode(doc))
 }
 
 // Deprecated: Using new API service