text-document

A rich text document model for Rust, inspired by Qt's QTextDocument/QTextCursor API.

Built on Qleany-generated Clean Architecture with redb (embedded ACID database), full undo/redo, and multi-cursor support.

Features

• Rich text model: Frames, Blocks, InlineElements with InlineContent::Text | Image
• Multi-cursor editing: Qt-style cursors with automatic position adjustment
• Full undo/redo: Snapshot-based, with composite grouping (begin_edit_block / end_edit_block)
• Import/Export: Plain text, Markdown, HTML, LaTeX, DOCX
• Search: Find, find all, regex, replace (undoable)
• Formatting: Character format (bold, italic, underline, ...), block format (alignment, heading_level, ...), frame format
• Tables: Insert, remove, row/column operations, cell merge/split, table/cell formatting, cursor-position-based convenience methods
• Layout engine API: Read-only handles (TextBlock, TextFrame, TextTable, TextTableCell, TextList), flow traversal, fragment-based text shaping, atomic snapshots, incremental change events
• Event system: Callback-based (on_change) and polling-based (poll_events), with FormatChangeKind (Block vs Character) and flow-level insert/remove events
• Thread-safe: Send + Sync throughout, Arc<Mutex<...>> interior mutability
• Resources: Image and stylesheet storage with base64 encoding

Quick start

use text_document::{TextDocument, MoveMode, MoveOperation};

let doc = TextDocument::new();
doc.set_plain_text("Hello world").unwrap();

// Cursor-based editing
let cursor = doc.cursor();
cursor.move_position(MoveOperation::EndOfWord, MoveMode::KeepAnchor, 1);
cursor.insert_text("Goodbye").unwrap(); // replaces "Hello"

// Multiple cursors
let c1 = doc.cursor();
let c2 = doc.cursor_at(5);
c1.insert_text("A").unwrap();
// c2's position is automatically adjusted

// Undo
doc.undo().unwrap();

// Search
use text_document::FindOptions;
let matches = doc.find_all("world", &FindOptions::default()).unwrap();

// Export
let html = doc.to_html().unwrap();
let markdown = doc.to_markdown().unwrap();

Layout engine API

Read-only handles for building a layout/rendering engine on top of the document model:

use text_document::{TextDocument, FlowElement, FragmentContent};

let doc = TextDocument::new();
doc.set_plain_text("Hello\nWorld").unwrap();

// Walk the document's visual flow
for element in doc.flow() {
    match element {
        FlowElement::Block(block) => {
            println!("Block {}: {:?}", block.id(), block.text());
            // Get formatting runs for glyph shaping
            for frag in block.fragments() {
                match frag {
                    FragmentContent::Text { text, format, offset, length } => {
                        println!("  text at {offset}: {text} (len={length})");
                    }
                    FragmentContent::Image { name, width, height, offset, .. } => {
                        println!("  image at {offset}: {name} ({width}x{height})");
                    }
                }
            }
        }
        FlowElement::Table(table) => {
            println!("Table {}x{}", table.rows(), table.columns());
        }
        FlowElement::Frame(frame) => {
            // Nested frames have their own flow
            let nested = frame.flow();
        }
    }
}

// Atomic snapshot for full layout
let snap = doc.snapshot_flow();

// Direct block access
let block = doc.block_at_position(0).unwrap();
let next = block.next(); // O(n) traversal
let snap = block.snapshot(); // all data in one lock

Table operations

use text_document::TextDocument;

let doc = TextDocument::new();
doc.set_plain_text("Before table").unwrap();

let cursor = doc.cursor_at(12);

// Insert a 3x2 table, get a handle back
let table = cursor.insert_table(3, 2).unwrap();
assert_eq!(table.rows(), 3);

// Explicit-ID mutations
cursor.insert_table_row(table.id(), 1).unwrap();   // insert row at index 1
cursor.remove_table_column(table.id(), 0).unwrap(); // remove first column

// Position-based convenience (cursor must be inside a table cell)
// cursor.insert_row_above().unwrap();
// cursor.remove_current_column().unwrap();
// cursor.set_current_table_format(&format).unwrap();

CLI

A command-line tool for format conversion and text processing:

# Convert between formats (detected by file extension)
text-document convert README.md output.html
text-document convert article.html article.tex

# Show document statistics
text-document stats manuscript.md

# Find text (grep-like output)
text-document find paper.md "TODO" --case-sensitive

# Find and replace
text-document replace draft.md "colour" "color" --output fixed.md

# Print to stdout in a different format
text-document cat notes.html --format plain

Supported formats:

Document structure

Root
 +-- Document
     +-- Frame (root frame)
     |   +-- Block
     |   |   +-- InlineElement (Text "Hello ")
     |   |   +-- InlineElement (Text "world" with bold)
     |   |   +-- InlineElement (Image { name, width, height })
     |   +-- Block
     |       +-- InlineElement (Text "Second paragraph")
     +-- Table (rows: 2, columns: 3)
     |   +-- TableCell (row: 0, col: 0)
     |   |   +-- Frame (cell frame)
     |   |       +-- Block
     |   |           +-- InlineElement (Text "Cell content")
     |   +-- TableCell (row: 0, col: 1) ...
     +-- List (style: Decimal, indent: 1)
     +-- Resource (image data, stylesheets)

• Frame: contains Blocks and child Frames. child_order interleaves them (positive = block ID, negative = sub-frame ID).
• Block: a paragraph. Contains InlineElements. Has document_position for O(log n) lookup.
• InlineElement: either Text(String), Image { name, width, height, quality }, or Empty.
• List: styling for list items (Disc, Decimal, LowerAlpha, ...). Blocks reference lists via weak relationship.
• Table: grid of TableCells, each with an optional cell frame containing Blocks.
• Resource: binary data (images, stylesheets) stored as base64.

All format fields are Option<T> — None means "inherit from parent/default", Some(value) means "explicitly set".

Public API handles

All handles are Clone + Send + Sync (backed by Arc<Mutex<...>> + entity ID).

Architecture

Generated by Qleany v1.5.1, following Clean Architecture with Package by Feature (Vertical Slice):

crates/
+-- public_api/       # TextDocument, TextCursor, DocumentEvent (the public crate)
+-- cli/              # Command-line tool
+-- frontend/         # AppContext, commands, event hub client
+-- common/           # Entities, database (redb), events, undo/redo, repositories
+-- macros/           # #[uow_action] proc macro
+-- direct_access/    # Entity CRUD controllers + DTOs
+-- document_editing/ # 19 use cases (insert, delete, block, image, frame, list, fragment, table CRUD, merge/split cells, ...)
+-- document_formatting/ # 6 use cases (set/merge text format, block format, frame format, table format, cell format)
+-- document_io/      # 8 use cases (import/export plain text, markdown, HTML, LaTeX, DOCX)
+-- document_search/  # 3 use cases (find, find_all, replace)
+-- document_inspection/ # 4 use cases (stats, text at position, block at position, extract fragment)
+-- test_harness/       # Shared test setup utilities

Data flow: TextDocument / TextCursor -> frontend::commands -> controllers -> use cases -> UoW -> repositories -> redb

License

Licensed under either of

• Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
• MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)

at your option.

Contribution

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.