Merge pull request #330 from jeffa5/experiment-graphemes

Remove grapheme splitting internally

automerge-cli/src/import.rs

@@ -28,7 +28,7 @@ fn import_map(
                 doc.put(obj, key, *b)?;
             }
             serde_json::Value::String(s) => {
-                doc.put(obj, key, s.as_ref())?;
+                doc.put(obj, key, s)?;
             }
             serde_json::Value::Array(vec) => {
                 let id = doc.put_object(obj, key, am::ObjType::List)?;
@@ -68,7 +68,7 @@ fn import_list(
                 doc.insert(obj, i, *b)?;
             }
             serde_json::Value::String(s) => {
-                doc.insert(obj, i, s.as_ref())?;
+                doc.insert(obj, i, s)?;
             }
             serde_json::Value::Array(vec) => {
                 let id = doc.insert_object(obj, i, am::ObjType::List)?;

automerge-wasm/Cargo.toml

@@ -30,7 +30,6 @@ getrandom = { version = "^0.2.2", features=["js"] }
 uuid = { version = "^0.8.2", features=["v4", "wasm-bindgen", "serde"] }
 serde-wasm-bindgen = "0.1.3"
 serde_bytes = "0.11.5"
-unicode-segmentation = "1.7.1"
 hex = "^0.4.3"
 regex = "^1.5"
 

automerge-wasm/src/interop.rs

@@ -4,7 +4,6 @@ use automerge::{Change, ChangeHash, Prop};
 use js_sys::{Array, Object, Reflect, Uint8Array};
 use std::collections::HashSet;
 use std::fmt::Display;
-use unicode_segmentation::UnicodeSegmentation;
 use wasm_bindgen::prelude::*;
 use wasm_bindgen::JsCast;
 
@@ -286,9 +285,9 @@ pub(crate) fn to_objtype(
         Some("text") => {
             let text = value.as_string()?;
             let text = text
-                .graphemes(true)
+                .chars()
                 .enumerate()
-                .map(|(i, ch)| (i.into(), ch.into()))
+                .map(|(i, ch)| (i.into(), ch.to_string().into()))
                 .collect();
             Some((am::ObjType::Text, text))
         }
@@ -311,9 +310,9 @@ pub(crate) fn to_objtype(
                 Some((am::ObjType::Map, map))
             } else if let Some(text) = value.as_string() {
                 let text = text
-                    .graphemes(true)
+                    .chars()
                     .enumerate()
-                    .map(|(i, ch)| (i.into(), ch.into()))
+                    .map(|(i, ch)| (i.into(), ch.to_string().into()))
                     .collect();
                 Some((am::ObjType::Text, text))
             } else {

automerge/Cargo.toml

@@ -24,7 +24,6 @@ smol_str = "^0.1.21"
 tracing = { version = "^0.1.29", features = ["log"] }
 fxhash = "^0.2.1"
 tinyvec = { version = "^1.5.1", features = ["alloc"] }
-unicode-segmentation = "1.7.1"
 serde = { version = "^1.0", features=["derive"] }
 dot = { version = "0.1.4", optional = true }
 js-sys = { version = "^0.3", optional = true }

automerge/src/automerge.rs

@@ -1735,4 +1735,47 @@ mod tests {
             ]
         );
     }
+
+    #[test]
+    fn can_insert_a_grapheme_into_text() {
+        let mut doc = Automerge::new();
+        let mut tx = doc.transaction();
+        let text = tx.put_object(ROOT, "text", ObjType::Text).unwrap();
+        let polar_bear = "🐻‍❄️";
+        tx.insert(&text, 0, polar_bear).unwrap();
+        tx.commit();
+        let s = doc.text(&text).unwrap();
+        assert_eq!(s, polar_bear);
+        let len = doc.length(&text);
+        assert_eq!(len, 1); // just one grapheme
+    }
+
+    #[test]
+    fn can_insert_long_string_into_text() {
+        let mut doc = Automerge::new();
+        let mut tx = doc.transaction();
+        let text = tx.put_object(ROOT, "text", ObjType::Text).unwrap();
+        let polar_bear = "🐻‍❄️";
+        let polar_bear_army = polar_bear.repeat(100);
+        tx.insert(&text, 0, &polar_bear_army).unwrap();
+        tx.commit();
+        let s = doc.text(&text).unwrap();
+        assert_eq!(s, polar_bear_army);
+        let len = doc.length(&text);
+        assert_eq!(len, 1); // many graphemes
+    }
+
+    #[test]
+    fn splice_text_uses_unicode_scalars() {
+        let mut doc = Automerge::new();
+        let mut tx = doc.transaction();
+        let text = tx.put_object(ROOT, "text", ObjType::Text).unwrap();
+        let polar_bear = "🐻‍❄️";
+        tx.splice_text(&text, 0, 0, polar_bear).unwrap();
+        tx.commit();
+        let s = doc.text(&text).unwrap();
+        assert_eq!(s, polar_bear);
+        let len = doc.length(&text);
+        assert_eq!(len, 4); // 4 chars
+    }
 }

automerge/src/transaction/transactable.rs

@@ -1,6 +1,5 @@
 use crate::exid::ExId;
 use crate::{AutomergeError, ChangeHash, Keys, KeysAt, ObjType, Prop, ScalarValue, Value};
-use unicode_segmentation::UnicodeSegmentation;
 
 /// A way of mutating a document within a single change.
 pub trait Transactable {
@@ -88,8 +87,7 @@ pub trait Transactable {
         del: usize,
         text: &str,
     ) -> Result<(), AutomergeError> {
-        let text = text.to_owned();
-        let vals = text.graphemes(true).map(|c| c.into());
+        let vals = text.chars().map(|c| c.into());
         self.splice(obj, pos, del, vals)
     }
 

automerge/src/value.rs

@@ -246,6 +246,12 @@ impl<'a> From<&str> for Value<'a> {
     }
 }
 
+impl<'a> From<&String> for Value<'a> {
+    fn from(s: &String) -> Self {
+        Value::Scalar(Cow::Owned(ScalarValue::Str(s.into())))
+    }
+}
+
 impl<'a> From<String> for Value<'a> {
     fn from(s: String) -> Self {
         Value::Scalar(Cow::Owned(ScalarValue::Str(s.into())))
@@ -626,6 +632,12 @@ impl From<&str> for ScalarValue {
     }
 }
 
+impl From<&String> for ScalarValue {
+    fn from(s: &String) -> Self {
+        ScalarValue::Str(s.into())
+    }
+}
+
 impl From<String> for ScalarValue {
     fn from(s: String) -> Self {
         ScalarValue::Str(s.into())