This is a maintenance release that removes unused legacy code while maintaining full backward compatibility.

Code Quality Improvements

• Removed 602 lines of dead code - Cleaned up old parsing functions that were replaced by the AST-based architecture in v1.2.0
• Eliminated all compiler warnings - Down from 47 warnings to 0
• No functional changes - All 298 tests still passing

What Was Removed

The following unused functions from the legacy (pre-AST) implementation were removed:

From erlmd_ast.erl:

• skip_blanks/2, skip_blanks_after_one/2 - Unused list item detection helpers
• parse_tag/2 - Unused HTML block conversion function

From erlmd.erl:

• parse/2, p1/4 - Old HTML rendering pipeline (replaced by erlmd_html:render/1)
• Old list/blockquote parsing helpers
• Legacy string processing and inline element functions
• Old character escaping and emphasis interpolation code

Note: The lexer and tokenization functions (is_double_indent/1, make_list_str/1, etc.) were preserved as they are still used by the new AST builder.

Upgrade Notes

No changes required - this release is fully backward compatible with v1.2.0.

This release represents a major refactoring of erlsci/erlmd, a fork of the erlmarkdown library

Breaking Changes

• Module renamed: markdown → erlmd
  • Old: markdown:conv/1
  • New: erlmd:conv/1
• Application renamed: markdown → erlmd (update your rebar.config dependencies)

New Features

AST-Based Architecture

• New intermediate representation: Markdown is now parsed into an Abstract Syntax Tree (AST) before HTML generation
• New module: erlmd_ast.erl - AST builder with full type definitions
• New module: erlmd_html.erl - Clean HTML renderer from AST
• Type definitions: include/types.hrl - Complete AST type specifications

New API Functions

• erlmd:conv_ast/1 - Convert Markdown string to AST (returns #document{} record)
• erlmd:conv_html/1 - Convert Markdown string or AST to HTML
• erlmd:conv/2 - High-level conversion with options: #{format => html | ast}
• erlmd:conv/1 - Convert to HTML using default options (backward compatible)

Improvements

Enhanced CommonMark Compliance (298/298 tests passing)

1. Phase 1: Fixed list and blockquote continuation line merging
2. Phase 2: Fixed codeblock merging with proper token handling
3. Phase 3: Implemented correct hard line breaks (two trailing spaces → <br />)
4. Phase 4: Proper list tight/loose detection and rendering
5. Phase 5: Improved blockquote hard break handling for better CommonMark compliance
6. Phase 6: Fixed tab expansion (tabs expand to spaces in text, preserved in code spans)
7. Phase 7: Fixed h2_or_hr disambiguation (setext headers vs horizontal rules)
8. Phase 8: Improved HTML tag pass-through handling

Test Infrastructure

• Migrated from JavaScript-generated tests to native Erlang EUnit tests
• 298 comprehensive test cases covering all Markdown features
• Removed old test generation infrastructure (tests/generate_tests.js, etc.)
• Added extensive test documentation in test/erlmd_tests.erl

Migration Guide (1.1.x → 1.2.0)

Step 1: Update your dependency in rebar.config:

% Old
{deps, [{markdown, "1.1.*"}]}.

% New
{deps, [{erlmd, "1.2.0"}]}.

Step 2: Update module calls in your code:

% Old
HTML = markdown:conv(MarkdownText).

% New
HTML = erlmd:conv(MarkdownText).

Step 3: If using UTF-8:

% Old
HTML = markdown:conv_utf8(Utf8Text).

% New
HTML = erlmd:conv_utf8(Utf8Text).

Step 4: Using the new AST API:

% Get AST
AST = erlmd:conv_ast("# Hello").

% Convert AST to HTML
HTML = erlmd:ast_to_html(AST).

% Or use conv/2 with options
AST = erlmd:conv("# Hello", #{format => ast}).
HTML = erlmd:conv("# Hello", #{format => html}).  % default

Documentation

• Design documents for AST architecture: docs/design/
• Phase-by-phase refactoring notes: workbench/PHASE*_FINDINGS.md