From a9f0d290af23f0fc8d16e89606c29980e413a545 Mon Sep 17 00:00:00 2001
From: tiye <jiyinyiyong@gmail.com>
Date: Thu, 27 Jun 2024 02:36:27 +0800
Subject: [PATCH] improve cjk in displaying; tag 0.6.11

---
 Cargo.toml             |  3 ++-
 src/edn.rs             | 21 +++++++++++++++++----
 src/lib.rs             |  4 +++-
 tests/display_tests.rs |  9 ++++++++-
 4 files changed, 30 insertions(+), 7 deletions(-)

diff --git a/Cargo.toml b/Cargo.toml
index d2beff4..425f585 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "cirru_edn"
-version = "0.6.10"
+version = "0.6.11"
 authors = ["jiyinyiyong <jiyinyiyong@gmail.com>"]
 edition = "2021"
 license = "MIT"
@@ -18,6 +18,7 @@ cirru_parser = "0.1.30"
 hex = "0.4.3"
 lazy_static = "1.4.0"
 bincode = "2.0.0-rc.3"
+cjk = "0.2.5"
 
 [dev-dependencies]
 criterion = "0.5.1"
diff --git a/src/edn.rs b/src/edn.rs
index 785be86..e5d625e 100644
--- a/src/edn.rs
+++ b/src/edn.rs
@@ -12,7 +12,7 @@ use std::{
   },
   collections::{HashMap, HashSet},
   convert::{TryFrom, TryInto},
-  fmt,
+  fmt::{self, Write},
   hash::{Hash, Hasher},
   iter::FromIterator,
   ptr,
@@ -63,7 +63,15 @@ impl fmt::Display for Edn {
         if is_simple_token(s) {
           f.write_fmt(format_args!("|{}", s))
         } else {
-          f.write_fmt(format_args!("\"|{}\"", s))
+          f.write_str("\"|")?;
+          for c in s.chars() {
+            if is_simple_char(c) {
+              f.write_char(c)?;
+            } else {
+              f.write_str(&c.escape_default().to_string())?;
+            }
+          }
+          f.write_char('"')
         }
       }
       Self::Quote(v) => f.write_fmt(format_args!("(quote {})", v)),
@@ -122,9 +130,14 @@ impl fmt::Display for Edn {
   }
 }
 
+/// check if a char is simple enough to be printed without quotes
+pub fn is_simple_char(c: char) -> bool {
+  matches!(c, '0'..='9' | 'A'..='Z' | 'a'..='z' | '-' | '?' | '.' | '$' | ',') || cjk::is_cjk_codepoint(c)
+}
+
 fn is_simple_token(tok: &str) -> bool {
-  for s in tok.bytes() {
-    if !matches!(s, b'0'..=b'9' | b'A'..=b'Z'| b'a'..=b'z'|  b'-' | b'?' | b'.'| b'$' | b',') {
+  for s in tok.chars() {
+    if !is_simple_char(s) {
       return false;
     }
   }
diff --git a/src/lib.rs b/src/lib.rs
index 2349dab..f49c821 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -9,7 +9,9 @@ use std::vec;
 
 use cirru_parser::{Cirru, CirruWriterOptions};
 
-pub use edn::{DynEq, Edn, EdnAnyRef, EdnListView, EdnMapView, EdnRecordView, EdnSetView, EdnTupleView};
+pub use edn::{
+  is_simple_char, DynEq, Edn, EdnAnyRef, EdnListView, EdnMapView, EdnRecordView, EdnSetView, EdnTupleView,
+};
 pub use tag::EdnTag;
 
 /// parse Cirru code into data
diff --git a/tests/display_tests.rs b/tests/display_tests.rs
index 269bc9e..ded630a 100644
--- a/tests/display_tests.rs
+++ b/tests/display_tests.rs
@@ -2,7 +2,7 @@ extern crate cirru_edn;
 
 use std::{sync::Arc, vec};
 
-use cirru_edn::{Edn, EdnRecordView, EdnTag};
+use cirru_edn::{Edn, EdnListView, EdnRecordView, EdnTag};
 
 #[test]
 fn display_data() {
@@ -22,3 +22,10 @@ fn display_data() {
   let t2 = Edn::from((Arc::new(Edn::tag("t")), vec![Edn::Number(1.0), Edn::Number(2.0)]));
   assert_eq!(format!("{t2}"), "(:: :t 1 2)");
 }
+
+#[test]
+fn display_with_cjk() {
+  let r = Edn::List(EdnListView(vec![Edn::str("你好"), Edn::str("世界"), Edn::str("海 洋")]));
+
+  assert_eq!(format!("{r}"), "([] |你好 |世界 \"|海 洋\")");
+}