From c901c00166809395bb2bab65c3420f1bb593e43c Mon Sep 17 00:00:00 2001
From: Niklas Fasching <niklas.fasching@gmail.com>
Date: Tue, 20 Jul 2021 22:33:28 +0200
Subject: [PATCH] Add non-standard ealb (east asian line break) option

See pandoc: https://pandoc.org/MANUAL.html#extension-east_asian_line_breaks
---
 org/document.go                                |  3 ++-
 org/html_writer.go                             |  4 +++-
 org/inline.go                                  | 10 ++++++++--
 org/testdata/east_asian_line_breaks.html       |  8 ++++++++
 org/testdata/east_asian_line_breaks.org        | 11 +++++++++++
 org/testdata/east_asian_line_breaks.pretty_org | 11 +++++++++++
 6 files changed, 43 insertions(+), 4 deletions(-)
 create mode 100644 org/testdata/east_asian_line_breaks.html
 create mode 100644 org/testdata/east_asian_line_breaks.org
 create mode 100644 org/testdata/east_asian_line_breaks.pretty_org

diff --git a/org/document.go b/org/document.go
index 0f016c1..9bc7fc8 100644
--- a/org/document.go
+++ b/org/document.go
@@ -86,7 +86,7 @@ func New() *Configuration {
 		DefaultSettings: map[string]string{
 			"TODO":         "TODO | DONE",
 			"EXCLUDE_TAGS": "noexport",
-			"OPTIONS":      "toc:t <:t e:t f:t pri:t todo:t tags:t title:t",
+			"OPTIONS":      "toc:t <:t e:t f:t pri:t todo:t tags:t title:t ealb:nil",
 		},
 		Log:      log.New(os.Stderr, "go-org: ", 0),
 		ReadFile: ioutil.ReadFile,
@@ -179,6 +179,7 @@ func (d *Document) Get(key string) string {
 // - todo (export headline todo status)
 // - pri (export headline priority)
 // - tags (export headline tags)
+// - ealb (non-standard) (export with east asian line breaks / ignore line breaks between multi-byte characters)
 // see https://orgmode.org/manual/Export-settings.html for more information
 func (d *Document) GetOption(key string) string {
 	get := func(settings map[string]string) string {
diff --git a/org/html_writer.go b/org/html_writer.go
index 878560b..5ab9d8f 100644
--- a/org/html_writer.go
+++ b/org/html_writer.go
@@ -311,7 +311,9 @@ func (w *HTMLWriter) WriteStatisticToken(s StatisticToken) {
 }
 
 func (w *HTMLWriter) WriteLineBreak(l LineBreak) {
-	w.WriteString(strings.Repeat("\n", l.Count))
+	if w.document.GetOption("ealb") == "nil" || !l.BetweenMultibyteCharacters {
+		w.WriteString(strings.Repeat("\n", l.Count))
+	}
 }
 
 func (w *HTMLWriter) WriteExplicitLineBreak(l ExplicitLineBreak) {
diff --git a/org/inline.go b/org/inline.go
index e7a8f47..b1bb28c 100644
--- a/org/inline.go
+++ b/org/inline.go
@@ -7,6 +7,7 @@ import (
 	"strings"
 	"time"
 	"unicode"
+	"unicode/utf8"
 )
 
 type Text struct {
@@ -14,7 +15,10 @@ type Text struct {
 	IsRaw   bool
 }
 
-type LineBreak struct{ Count int }
+type LineBreak struct {
+	Count                      int
+	BetweenMultibyteCharacters bool
+}
 type ExplicitLineBreak struct{}
 
 type StatisticToken struct{ Content string }
@@ -159,7 +163,9 @@ func (d *Document) parseLineBreak(input string, start int) (int, Node) {
 	i := start
 	for ; i < len(input) && input[i] == '\n'; i++ {
 	}
-	return i - start, LineBreak{i - start}
+	_, beforeLen := utf8.DecodeLastRuneInString(input[:start])
+	_, afterLen := utf8.DecodeRuneInString(input[i:])
+	return i - start, LineBreak{i - start, beforeLen > 1 && afterLen > 1}
 }
 
 func (d *Document) parseInlineBlock(input string, start int) (int, int, Node) {
diff --git a/org/testdata/east_asian_line_breaks.html b/org/testdata/east_asian_line_breaks.html
new file mode 100644
index 0000000..3502d67
--- /dev/null
+++ b/org/testdata/east_asian_line_breaks.html
@@ -0,0 +1,8 @@
+<p>
+Line breaks between multi-byte characters are omitted when the <code class="verbatim">ealb</code> option is set:</p>
+<ul>
+<li>中午吃啥</li>
+<li>something else
+中午吃啥
+something else</li>
+</ul>
diff --git a/org/testdata/east_asian_line_breaks.org b/org/testdata/east_asian_line_breaks.org
new file mode 100644
index 0000000..9e4e124
--- /dev/null
+++ b/org/testdata/east_asian_line_breaks.org
@@ -0,0 +1,11 @@
+#+OPTIONS: ealb:t
+
+Line breaks between multi-byte characters are omitted when the =ealb= option is set:
+
+- 中午
+  吃啥
+
+- something else
+  中午
+  吃啥
+  something else
diff --git a/org/testdata/east_asian_line_breaks.pretty_org b/org/testdata/east_asian_line_breaks.pretty_org
new file mode 100644
index 0000000..9e4e124
--- /dev/null
+++ b/org/testdata/east_asian_line_breaks.pretty_org
@@ -0,0 +1,11 @@
+#+OPTIONS: ealb:t
+
+Line breaks between multi-byte characters are omitted when the =ealb= option is set:
+
+- 中午
+  吃啥
+
+- something else
+  中午
+  吃啥
+  something else