Skip to content

multi-encoding.md

Latest commit

 

History

History
178 lines (138 loc) · 5.58 KB

multi-encoding.md

File metadata and controls

178 lines (138 loc) · 5.58 KB
Raw

Multi-Encoding File Support

Overview

Ferrite v0.2.6 adds comprehensive character encoding support, allowing users to open, edit, and save files in various encodings beyond UTF-8. The encoding is auto-detected on file open and can be manually changed via the status bar.

Key Files

  • src/state.rs - Tab struct encoding fields, detection logic, encode/decode methods
  • src/app.rs - Status bar encoding UI with dropdown picker
  • Cargo.toml - Added encoding_rs and chardetng dependencies

Implementation Details

Tab Struct Changes

Three new fields added to Tab:

pub struct Tab {
    // ... existing fields ...
    
    /// Detected encoding when file was opened (e.g., "UTF-8", "WINDOWS-1252")
    /// None for new/unsaved documents created in-app
    pub detected_encoding: Option<&'static str>,
    
    /// Original file bytes for re-decoding when user changes encoding
    /// Empty for new documents created in-app
    pub original_bytes: Vec<u8>,
    
    /// Currently selected encoding label (used for save operations)
    /// Defaults to "utf-8" for new documents
    pub current_encoding: &'static str,
}

Encoding Detection Flow

  1. Read file as bytes - std::fs::read(&path) instead of read_to_string()
  2. Check for BOM - encoding_rs::Encoding::for_bom(&bytes) detects UTF-8/16 BOMs
  3. Fallback to chardetng - Statistical detection for non-BOM files
  4. Decode to String - encoding.decode_without_bom_handling() for BOM-detected files
// BOM detection takes priority
if let Some((bom_encoding, bom_len)) = encoding_rs::Encoding::for_bom(&bytes) {
    let (decoded, _) = bom_encoding.decode_without_bom_handling(&bytes[bom_len..]);
    // Use bom_encoding
} else {
    // Use chardetng detection
    let mut detector = chardetng::EncodingDetector::new();
    detector.feed(&bytes, true);
    let detected = detector.guess(None, true);
    let (decoded, _, _) = detected.decode(&bytes);
}

Supported Encodings

pub const COMMON_ENCODINGS: &'static [&'static str] = &[
    "utf-8", "windows-1252", "iso-8859-1", "shift_jis", 
    "euc-jp", "gbk", "euc-kr", "iso-8859-15", 
    "utf-16le", "utf-16be",
];

Re-encoding on Save

When saving, content is encoded from UTF-8 String to bytes using the selected encoding:

pub fn encode_content(&self) -> Vec<u8> {
    if let Some(encoding) = encoding_rs::Encoding::for_label(self.current_encoding.as_bytes()) {
        let (bytes, _, _) = encoding.encode(&self.content);
        bytes.into_owned()
    } else {
        self.content.as_bytes().to_vec() // Fallback to UTF-8
    }
}

Manual Encoding Change

Users can re-decode file content with a different encoding:

pub fn set_encoding(&mut self, new_encoding: &'static str) -> Result<(), String> {
    if self.original_bytes.is_empty() {
        return Err("No original bytes available".to_string());
    }
    
    if let Some(encoding) = encoding_rs::Encoding::for_label(new_encoding.as_bytes()) {
        let (decoded, _) = encoding.decode_without_bom_handling(&self.original_bytes);
        self.content = decoded.into_owned();
        self.current_encoding = new_encoding;
        Ok(())
    } else {
        Err(format!("Unsupported encoding: {}", new_encoding))
    }
}

Status Bar UI

The encoding is displayed as a clickable button in the status bar. Clicking opens a dropdown with common encodings:

// In app.rs status bar
let encoding_btn = ui.button(tab.encoding_display_name());
if encoding_btn.clicked() {
    ui.memory_mut(|mem| mem.toggle_popup(popup_id));
}

egui::popup_below_widget(ui, popup_id, &encoding_btn, |ui| {
    for &enc in Tab::COMMON_ENCODINGS {
        if ui.selectable_label(enc == tab.current_encoding, enc.to_uppercase()).clicked() {
            pending_encoding_change = Some(enc);
            ui.memory_mut(|mem| mem.close_popup());
        }
    }
});

Dependencies Used

  • encoding_rs = "0.8" - Fast character encoding conversion (same library used by Firefox)
  • chardetng = "0.1" - Encoding detection (same algorithm as Firefox)

Session Restoration

When restoring sessions, files are re-read from disk with encoding detection:

enum ResolvedContent {
    /// Content recovered from crash backup (already UTF-8)
    Recovered(String),
    /// Content loaded from disk with encoding info
    FromDisk {
        content: String,
        original_bytes: Vec<u8>,
        encoding: &'static str,
    },
}

Testing

Test files are provided in test_md/encoding_tests/:

File Encoding Notes
utf8_no_bom.md UTF-8 Standard UTF-8
utf8_with_bom.md UTF-8 + BOM EF BB BF prefix
windows1252.txt Windows-1252 Curly quotes, Euro symbol
latin1.txt ISO-8859-1 Western European
shiftjis.txt Shift-JIS Japanese
eucjp.txt EUC-JP Japanese Unix
gbk.txt GBK Simplified Chinese
euckr.txt EUC-KR Korean
utf16le.txt UTF-16 LE With FF FE BOM
utf16be.txt UTF-16 BE With FE FF BOM

Regenerate with: python test_md/encoding_tests/create_test_files.py

Usage

  1. Open file - Encoding auto-detected, shown in status bar
  2. Check encoding - Look at status bar (e.g., "UTF-8", "SHIFT_JIS")
  3. Change encoding - Click encoding button, select from dropdown
  4. Save - File saved in currently selected encoding

Limitations

  • UTF-16 files require BOM for reliable detection
  • Some similar encodings may be mis-detected (e.g., ISO-8859-1 vs Windows-1252)
  • New documents always default to UTF-8
  • Original bytes are kept in memory for re-decoding (increases memory per tab)