bayan-api / archive /phase_reports /EDITOR_REFACTOR_PLAN.md
youssefreda9's picture
Cleanup: Archive legacy scripts + phase reports, delete 35 orphaned files
78a516a
|
Raw
History Blame Contribute Delete
5.66 kB

Bayan (بيان) — Phase 1 Editor Stabilization Refactor Plan (Revised)

Goal

Refactor the current editor architecture to solve the core technical problems while avoiding unnecessary complexity.

This phase is strictly focused on editor stability and maintainability.

Success Criteria

  • Cursor position is preserved after analysis updates.
  • Text selection is preserved.
  • Multiple occurrences of the same word are highlighted correctly.
  • Suggestions use exact character offsets instead of string replacement.
  • Rendering is XSS-safe.
  • Editor code becomes modular and easier to extend.
  • Future features (DOCX import, export, database persistence) remain possible without major rewrites.

Scope

Included

  • Cursor preservation
  • Selection preservation
  • Offset-based highlighting
  • Duplicate occurrence handling
  • Secure rendering
  • Editor code modularization

Excluded

  • DOCX Import
  • TXT Import
  • DOCX Export
  • PDF Export
  • Supabase
  • Authentication
  • Autosave
  • Database integration
  • Deployment work

These belong to later phases.


Architectural Strategy

Keep Existing contenteditable

We will continue using a single contenteditable editor.

Current:

<div id="editor-container" contenteditable="true"></div>

No dual-layer editor.

No transparent text layer.

No click-forwarding system.

No overlay synchronization.


Why

The current project is a graduation project, not a full-scale IDE.

Keeping contenteditable provides:

  • Lower complexity
  • Faster implementation
  • Fewer bugs
  • Easier maintenance
  • Faster delivery

while still solving all current issues.


Milestone 1 — Modularize Editor Logic

Objective

Separate editor concerns from UI and networking logic.

New Structure

src/
│
├── js/
│   ├── api.js
│   ├── editor.js
│   ├── renderer.js
│   ├── selection.js
│   └── ui.js
│
├── index.html

Responsibilities

api.js

Handles:

  • /api/analyze
  • /api/spelling
  • /api/autocomplete
  • /api/summarize

No DOM manipulation.


editor.js

Handles:

  • Reading editor text
  • Writing editor text
  • Editor events
  • Debouncing

renderer.js

Handles:

  • Highlight rendering
  • Safe HTML generation
  • Offset mapping

selection.js

Handles:

  • Save selection
  • Restore selection
  • Caret positioning

ui.js

Handles:

  • Tooltips
  • Panels
  • Buttons
  • Notifications

Milestone 2 — Selection Preservation

Problem

Current rendering rewrites the editor DOM.

Result:

  • Cursor jumps
  • Selection disappears
  • Focus is lost

Solution

Before rendering:

const selection = saveSelection();

After rendering:

restoreSelection(selection);

Required Functions

saveSelection()
restoreSelection()
getCaretOffset()
setCaretOffset()

Expected Result

User can:

  • Type continuously
  • Select text
  • Accept suggestions

without cursor jumps.


Milestone 3 — Backend Offset Support

Current Problem

Suggestions identify words only.

Example:

{
  "original": "ذهبو",
  "correction": "ذهبوا"
}

This fails when the same word appears multiple times.


Required Change

Update backend responses.

Example:

{
  "suggestions": [
    {
      "start": 12,
      "end": 17,
      "original": "ذهبو",
      "correction": "ذهبوا",
      "type": "spelling"
    }
  ]
}

Benefits

  • Exact positioning
  • No ambiguity
  • Faster rendering
  • Supports duplicate words

Milestone 4 — Offset-Based Rendering

Current Problem

Rendering uses:

text.replace(...)

which only affects the first occurrence.


New Strategy

Use character offsets.

Example:

{
  "start": 45,
  "end": 50
}

Rendering process:

  1. Split text into segments.
  2. Create normal text nodes.
  3. Create highlighted spans.
  4. Assemble output safely.

Result

Every occurrence is highlighted correctly.


Milestone 5 — Secure Rendering

Current Problem

Direct innerHTML generation creates XSS risk.


Solution

Escape all user-generated content.

Example:

escapeHtml(text)

before inserting into DOM.


Requirements

Never inject:

userInput
suggestionText
apiResponse

directly into HTML.

Always sanitize first.


Milestone 6 — Highlight Engine Refactor

Objective

Create a reusable rendering pipeline.

Input

{
  "text": "...",
  "suggestions": [...]
}

Output

Safe highlighted HTML.


Responsibilities

The renderer must support:

  • Spelling highlights
  • Grammar highlights
  • Punctuation highlights

without changing editor logic.


Milestone 7 — Tooltip Mapping

Strategy

Each highlight receives:

<span
  class="error-highlight"
  data-suggestion-id="42">
</span>

Clicking a span retrieves:

currentSuggestions[id]

and opens the tooltip.


Benefits

  • Cleaner architecture
  • Easier debugging
  • Future extensibility

Deliverables

At the end of Phase 1:

Functional Deliverables

  • Stable editor
  • Cursor preservation
  • Selection preservation
  • Duplicate-word support
  • Offset-based highlighting
  • Secure rendering
  • Modular editor code

Technical Deliverables

api.js
editor.js
renderer.js
selection.js
ui.js

Explicitly Deferred

The following are NOT part of this phase:

  • Light/Dark Theme
  • UI Redesign
  • TXT Import/Export
  • DOCX Import/Export
  • PDF Export
  • Authentication
  • Supabase
  • Database Storage
  • Autosave
  • Deployment

These will be implemented in later phases.