From 8a78c349175da90c7c79e3bec81698a6b910b821 Mon Sep 17 00:00:00 2001
From: eshrh <16175276+eshrh@users.noreply.github.com>
Date: Sat, 23 Jan 2021 18:29:43 -0500
Subject: [PATCH] Support title splitting for fullwidth CJK terminals (#1163)

* Split by fullwidth terminals without spaces.
* Add test
* Update write.feature
---
 features/write.feature | 14 ++++++++++++++
 jrnl/Entry.py          | 13 ++++++++-----
 2 files changed, 22 insertions(+), 5 deletions(-)

diff --git a/features/write.feature b/features/write.feature
index efa26201..eb22e480 100644
--- a/features/write.feature
+++ b/features/write.feature
@@ -28,6 +28,20 @@ Feature: Writing new entries.
         | basic_folder    |
         | basic_dayone    |
 
+    Scenario Outline: CJK entry should be split at fullwidth period without following space.
+        Given we use the config "<config_file>.yaml"
+        And we use the password "test" if prompted
+        When we run "jrnl 七転び。八起き"
+        And we run "jrnl -1"
+        Then the output should contain "| 八起き"
+
+        Examples: configs
+        | config_file     |
+        | basic_onefile   |
+        | basic_encrypted |
+        | basic_folder    |
+        | basic_dayone    |
+
     Scenario Outline: Writing an entry from command line should store the entry
         Given we use the config "<config_file>.yaml"
         And we use the password "bad doggie no biscuit" if prompted
diff --git a/jrnl/Entry.py b/jrnl/Entry.py
index 2a85e015..67ba84f3 100755
--- a/jrnl/Entry.py
+++ b/jrnl/Entry.py
@@ -204,14 +204,17 @@ class Entry:
 # https://github.com/fnl/segtok
 SENTENCE_SPLITTER = re.compile(
     r"""
-(                       # A sentence ends at one of two sequences:
-    [.!?\u2026\u203C\u203D\u2047\u2048\u2049\u22EF\u3002\uFE52\uFE57\uFF01\uFF0E\uFF1F\uFF61]                # Either, a sequence starting with a sentence terminal,
+    (
+    [.!?\u2026\u203C\u203D\u2047\u2048\u2049\u22EF\uFE52\uFE57] # Sequence starting with a sentence terminal,
     [\'\u2019\"\u201D]? # an optional right quote,
-    [\]\)]*             # optional closing brackets and
-    \s+                 # a sequence of required spaces.
-)""",
+    [\]\)]*             # optional closing bracket
+    \s+                 # AND a sequence of required spaces.
+    )
+    |[\uFF01\uFF0E\uFF1F\uFF61\u3002] # CJK full/half width terminals usually do not have following spaces.
+    """,
     re.VERBOSE,
 )
+
 SENTENCE_SPLITTER_ONLY_NEWLINE = re.compile("\n")