The IDE you know, rebuilt in Rust. IntelliJ IDEA–style keymap, tool windows and search-everywhere in a fast native app: editor, code insight, debugger, Git, terminal, database and HTTP clients for 60+ languages.
TypeScript code-editorcross-platformdapdatabase-clientdebuggerdesktop-appdeveloper-toolsgithttp-clientideintellijlsprusttauriterminal
fix(properties): keep existing Unicode escapes when saving with native-to-ascii (#650)
l0ng-ai376501a5fe57
diff --git a/crates/ide-host/tests/properties.rs b/crates/ide-host/tests/properties.rsindex 097d92a63..dce783dd4 100644--- a/crates/ide-host/tests/properties.rs+++ b/crates/ide-host/tests/properties.rs@@ -104,7 +104,7 @@ async fn properties_search_git_and_document_coordinates_agree() {"main",opened.id,opened.version,- ChangeSet { changes: vec![TextChange { from: 10, to: 16, insert: "世界".into() }] },+ ChangeSet { changes: vec![TextChange { from: 10, to: 16, insert: r"\u4E16\u754C".into() }] },).unwrap();let hunks = vcs::vcs_line_hunks(&host, "main", path.clone()).await.unwrap();@@ -129,7 +129,7 @@ async fn properties_search_git_and_document_coordinates_agree() {assert!(project.documents.snapshot(opened.id).unwrap().version > snap.version);assert!(vcs::vcs_line_hunks(&host, "main", path.clone()).await.unwrap().is_empty());// Partial commits identify hunks in displayed line coordinates as well.- let partial_text = displayed.replace("target", "世界").replace("logical", "changed");+ let partial_text = displayed.replace("target", r"\u4E16\u754C").replace("logical", "changed");project.documents.replace_all(opened.id, &partial_text).unwrap();doc::save(&host, "main", opened.id).await.unwrap();let hunks = ide_vcs::diff_hunks(displayed, &partial_text);@@ -169,7 +169,7 @@ async fn properties_search_git_and_document_coordinates_agree() {"main",RequestId(520),query,- "世界".into(),+ r"\u4E16\u754C".into(),false,vec![ide_search::ReplaceTarget {path: closed_path.clone().into(),@@ -181,5 +181,6 @@ async fn properties_search_git_and_document_coordinates_agree() {.unwrap();assert_eq!(std::fs::read_to_string(&closed_path).unwrap(), r"value=\u4F60\u597D\uD83D\uDE00\u4E16\u754C");assert!(project.documents.by_path(&closed_path).is_none());+ assert_eq!(project.text_of(&closed_path).unwrap(), "value=你好😀世界");host.close_project("main");}diff --git a/crates/ide-vcs/src/text_encoding.rs b/crates/ide-vcs/src/text_encoding.rsindex 038256a95..768504b6d 100644--- a/crates/ide-vcs/src/text_encoding.rs+++ b/crates/ide-vcs/src/text_encoding.rs@@ -3,7 +3,7 @@ use encoding_rs::{Encoding, UTF_8, UTF_16BE, UTF_16LE};use ide_core::{IdeError, IdeResult};use std::borrow::Cow;-/// File charset plus an optional reversible ASCII representation (for properties).+/// File charset plus an optional ASCII representation (for properties)./// The host supplies conversion functions so Git and documents share one codec.#[derive(Clone, Copy)]pub struct TextEncoding {diff --git a/crates/ide-vfs/src/encoding.rs b/crates/ide-vfs/src/encoding.rsindex 05504a1b2..92e0d999b 100644--- a/crates/ide-vfs/src/encoding.rs+++ b/crates/ide-vfs/src/encoding.rs@@ -62,20 +62,19 @@ pub fn decode_properties_escapes(text: &str) -> String {}/// Serializes displayed properties text as ASCII, using uppercase UTF-16 units.+/// Existing ASCII Unicode escapes retain their spelling and meaning on disk.pub fn encode_properties_ascii(text: &str) -> String {use std::fmt::Write;let mut out = String::with_capacity(text.len());let mut rest = text;while let Some(c) = rest.chars().next() {- let current = rest;rest = &rest[c.len_utf8()..];if c == '\\' {if rest.starts_with('\\') {out.push_str("\\\\");rest = &rest[1..];- } else if unicode_character(current).is_some() || rest.chars().next().is_some_and(|next| !next.is_ascii()) {- // Keep a literal backslash from swallowing a generated escape,- // or becoming a Unicode escape on the next load.+ } else if rest.chars().next().is_some_and(|next| !next.is_ascii()) {+ // Keep a literal backslash from swallowing a generated escape.out.push_str("\\u005C");} else {out.push(c);@@ -387,13 +386,32 @@ mod tests {for text in ["plain=ASCII\r\n", r"escaped=\\u0041", r"bad=\u12xz", r"bad=\uD800"] {assert_eq!(encode_properties_ascii(text), text);}- for text in ["你好😀", "\0", "\\中", "\\\\中", "\\\\\\😀", r"\u0041", r"\\u0041", r"\uD83D\uDE00"] {+ for text in ["你好😀", "\0", "\\中", "\\\\中", "\\\\\\😀", r"\\u0041"] {let ascii = encode_properties_ascii(text);assert!(ascii.is_ascii());assert_eq!(decode_properties_escapes(&ascii), text, "{ascii}");}}+ #[test]+ fn properties_ascii_escapes_are_not_escaped_again() {+ for text in [+ r"test.key2=\u4F60\u597D",+ r"test.key3=\u4f60\u597d",+ r"ascii=\u0041\u005C\u000A",+ r"emoji=\uD83D\uDE00",+ r"slashes=\\u4F60 \\\u4F60 \\\\u4F60",+ r"bad=\u123 \u12xz \uu0041 \uD800 \uDC00",+ ] {+ assert_eq!(encode_properties_ascii(text), text, "{text}");+ }+ let mixed = "raw=\\u4f60\\u597d; native=你好😀";+ let expected = r"raw=\u4f60\u597d; native=\u4F60\u597D\uD83D\uDE00";+ assert_eq!(encode_properties_ascii(mixed), expected);+ assert_eq!(encode_properties_ascii(expected), expected);+ assert_eq!(decode_properties_escapes(expected), "raw=你好; native=你好😀");+ }+#[test]fn properties_line_terminator_escapes_survive_an_edit() {let source = r"key=one\u000Atwo\u000Dthree";diff --git a/crates/ide-workspace/tests/documents.rs b/crates/ide-workspace/tests/documents.rsindex 028617eb8..02370eed5 100644--- a/crates/ide-workspace/tests/documents.rs+++ b/crates/ide-workspace/tests/documents.rs@@ -641,6 +641,57 @@ fn properties_conversion_uses_displayed_utf16_and_survives_reload() {assert!(!f.docs.is_dirty(doc.id).unwrap());}+#[test]+fn properties_pasted_escapes_survive_save_and_decode_on_reload() {+ let pasted = "test.key1=Hello\ntest.key2=\\u4F60\\u597D\ntest.key3=\\u4f60\\u597d\n";+ let displayed = "test.key1=Hello\ntest.key2=你好\ntest.key3=你好\n";+ for encoding in ["utf-8", "iso-8859-1", "utf-16le"] {+ for eol in ["\n", "\r\n"] {+ let f = fixture();+ let path = f.root.join("test_zh_CN.properties");+ let source = pasted.replace('\n', eol);+ fs::write(&path, ide_vfs::encode(&source, encoding, encoding == "utf-16le").unwrap()).unwrap();+ let mut defaults = f.docs.file_defaults();+ defaults.native_to_ascii = true;+ defaults.properties_fallback = encoding.into();+ f.docs.set_file_defaults(defaults.clone());+ let doc = f.docs.open(&path).unwrap();+ assert_eq!(doc.text, displayed);++ f.docs.replace_all(doc.id, pasted).unwrap();+ f.docs.save(doc.id).unwrap();+ assert_eq!(fs::read(&path).unwrap(), source.as_bytes(), "{encoding} {eol:?}");+ assert_eq!(f.docs.snapshot(doc.id).unwrap().text, pasted);+ assert!(!f.docs.is_dirty(doc.id).unwrap());+ f.docs.on_disk_changed(std::slice::from_ref(&path));+ assert_eq!(f.docs.snapshot(doc.id).unwrap().text, pasted, "own saves do not reload");++ f.docs.close(doc.id);+ let reopened = f.docs.open(&path).unwrap();+ assert_eq!(reopened.text, displayed);+ // An edit followed by its inverse preserves the original escape casing.+ f.docs.replace_all(reopened.id, pasted).unwrap();+ f.docs.replace_all(reopened.id, displayed).unwrap();+ f.docs.save(reopened.id).unwrap();+ assert_eq!(fs::read(&path).unwrap(), source.as_bytes());++ let external = format!("{source}test.key4=\\uD83D\\uDE00{eol}");+ fs::write(&path, &external).unwrap();+ f.docs.on_disk_changed(std::slice::from_ref(&path));+ assert_eq!(f.docs.snapshot(reopened.id).unwrap().text, format!("{displayed}test.key4=😀\n"));+ defaults.native_to_ascii = false;+ f.docs.set_file_defaults(defaults.clone());+ f.docs.reload_properties();+ assert_eq!(f.docs.snapshot(reopened.id).unwrap().text, external.replace("\r\n", "\n"));+ defaults.native_to_ascii = true;+ f.docs.set_file_defaults(defaults);+ f.docs.reload_properties();+ assert_eq!(f.docs.snapshot(reopened.id).unwrap().text, format!("{displayed}test.key4=😀\n"));+ assert_eq!(fs::read_to_string(&path).unwrap(), external);+ }+ }+}+#[test]fn properties_toggle_reloads_only_clean_properties_and_pins_dirty_codec() {let f = fixture();