Archive Reader
Cross-platform MHTML Reader and Manager
- Role
- Sole Developer (Architecture, Design, and Development)
- Client
- Personal Project / Public Distribution
- Duration
- 2026-07-15 — 2026-08-20
- Year
- 2026
- Team
- 1 persona
Problem
What needed to be solved?
Managing MHTML files is complex and unintuitive across mobile and desktop platforms, especially when handling large file sizes and datasets.
Solution
How was it solved?
Developed a cross-platform application that simplifies MHTML reading, management, and creation, with a strong focus on performance, reliability, and user experience.
Features
What it does
- Multi-library system with folder history and persistent metadata
- Custom multipart MIME parser compatible with real-world MHTML files across browsers
- Robust encoding detection (UTF-8, Windows-1252, ISO-8859-1) with charset sniffing from HTML
- Dual rendering strategy: local HTTP server on Android/iOS and file:// access on Windows
- Custom scrollbar with proportional positioning based on actual content length
- MHTML creation from image folders, ZIP files, or web URLs
- Full web page capture (HTML, images, styles) via headless WebView
- Dedicated Isolates for file generation with atomic writes to avoid UI blocking
- Stable handling of files larger than 130MB without memory leaks
Responsibilities
My role in the project
- Designed the complete application architecture
- Developed the app end-to-end in Flutter (UI, business logic, persistence)
- Designed and implemented a custom MIME/MHTML parser
- Resolved low-level issues (encoding handling, memory management, platform-specific WebView behavior)
- Performed manual testing on Android and Windows
Technical Challenges
What wasn’t trivial
MHTML compatibility
Achieved cross-compatibility between generated MHTML files and external readers (Chrome, Edge, other apps) by handling reference schemes (Content-Location vs cid:) and CRLF encoding
Memory optimization
Prevented UI blocking and memory leaks when processing files larger than 130MB by offloading heavy work to Isolates and streaming resources incrementally instead of loading them entirely into memory
Android Media Scanner
Diagnosed and mitigated interference from Android’s media scanner when writing files to public storage
WebView vs Flutter gestures
Resolved conflicts between native WebView gestures (e.g., text selection) and Flutter’s GestureDetector
CORS limitations
Worked around CORS restrictions when capturing external web resources without relying on a backend server
Results
What was achieved
- Delivered a functional application on Android and Windows with feature parity across platforms
- Generated MHTML files fully compatible with desktop browsers and third-party readers
- Enabled stable management of large file libraries without degrading UI performance
- file_selector.getSaveLocation is not supported on Android/iOS; getDirectoryPath is the practical cross-platform alternative
- Writing large files to a private temporary directory before moving them prevents corruption caused by external processes (media scanners, antivirus)
- Native WebView events lack consistent behavior across platforms; a JS↔Dart messaging bridge is more reliable than platform-specific callbacks
- Critically validate technical suggestions from external sources (including AI) before implementation, especially when internally inconsistent
- Prefer DOM-based HTML manipulation over string replacement for robustness against real-world markup variability