Add markdown-to-blocks parser for proper Logseq block hierarchy

[RobertoGongora]

Add markdown-to-blocks parser for proper Logseq block hierarchy

Fixes #7

This PR introduces intelligent markdown parsing that converts markdown content into Logseq's native block tree structure, fixing the issue where create_page and update_page would dump all content as a single block.

Key improvements:

• ✅ Markdown → proper Logseq block hierarchy (headings, lists, nesting)
• ✅ YAML frontmatter → page properties (no more MethodNotExist error)
• ✅ Code blocks preserved as single blocks (not split line-by-line)
• ✅ TODO/DONE checkboxes without duplicate markers
• ✅ Multi-line blockquotes join correctly
• ✅ All nested blocks returned by get_page_content

---

📖 Quick Examples

Example 1: Simple Page Creation - Proper Hierarchy

Before (v1.0.1) ❌

# All content dumped as a single flat block
create_page("Project Notes", """
# Project Notes
- Task 1
  - Subtask A
- Task 2
""")

# Result in Logseq:
# Single block containing:
# "# Project Notes\n- Task 1\n  - Subtask A\n- Task 2"

After (v1.1.0) ✅

# Properly parsed into hierarchical blocks
create_page("Project Notes", """
# Project Notes
- Task 1
  - Subtask A
- Task 2
""")

# Result in Logseq:
# ├─ # Project Notes
# │  ├─ Task 1
# │  │  └─ Subtask A
# │  └─ Task 2

Try it yourself:

from mcp_logseq.logseq import LogSeq
from mcp_logseq.parser import parse_content

content = """# Project Notes
- Task 1
  - Subtask A
- Task 2"""

parsed = parse_content(content)
print(f"Blocks created: {len(parsed.blocks)}")
print(f"First block: {parsed.blocks[0].content}")
print(f"First block children: {len(parsed.blocks[0].children)}")
# Output:
# Blocks created: 1
# First block: # Project Notes
# First block children: 2

Example 2: Properties + TODO Lists - No More Errors

Before (v1.0.1) ❌

create_page("Tasks", """
---
priority: high
due-date: 2026-01-15
---

# Tasks
- [ ] TODO: Design feature
- [x] DONE: Initial setup
""")

# Result in Logseq:
# - Properties: error: MethodNotExist: get_page_properties
# - TODO markers duplicated: "TODO TODO: Design feature"
# - Date serialization error: "Object of type date is not JSON serializable"

After (v1.1.0) ✅

create_page("Tasks", """
---
priority: high
due-date: 2026-01-15
---

# Tasks
- [ ] TODO: Design feature
- [x] DONE: Initial setup
""")

# Result in Logseq:
# - Properties: {priority: "high", due-date: "2026-01-15"} ✅
# - Clean TODO markers: "TODO Design feature" ✅
# - Date serialized correctly to ISO string ✅

Try it yourself:

from mcp_logseq.parser import parse_content

content = """---
priority: high
due-date: 2026-01-15
---

# Tasks
- [ ] TODO: Design feature
- [x] DONE: Initial setup"""

parsed = parse_content(content)
print(f"Properties: {parsed.properties}")
print(f"First task: {parsed.blocks[0].children[0].content}")
print(f"Second task: {parsed.blocks[0].children[1].content}")
# Output:
# Properties: {'priority': 'high', 'due-date': '2026-01-15'}
# First task: TODO Design feature
# Second task: DONE Initial setup

Summary

Changes: 11 files (+2543, -682)
Tests: 103 passing with comprehensive coverage
Type checking: 0 errors

See commit message for full details on new features, API extensions, and behavioral changes.

Recommended version: v1.1.0 (feature release)

[RobertoGongora]

Fixed an issue with capitalized markers not being properly parsed. I'm using this and improving as I find things but it's already in a pretty usable state.

Edit: Additional improvements made on the get_page_contents tool so its friendlier with LLMs and cross-tool e.g. for full page replacements.