docx-table-styler

name

docx-table-styler

description

Convert Obsidian/Markdown to styled DOCX for Korean government proposals and research reports. Styles all tables with navy header, horizontal row alternation, content-proportional column widths, 10pt font, tight cell margins, vertical center alignment. Preserves embedded images. Hancom HWP compatible. Triggers on requests like "DOCX 테이블 스타일", "한글 보고서 표 꾸미기", "정부 제안서 포맷 DOCX", "export Obsidian to styled Word", "all tables styled", "proposal document formatting".

DOCX Table Styler (Korean Government Report Preset)

When to use

Invoke when the user asks to:

• Convert a large Markdown document (especially Obsidian vault with ![[wikilinks]]) to a styled .docx
• Apply consistent branded table styling across many tables (dozens to hundreds)
• Produce a Korean government proposal, research report, or bilingual KR/EN document
• Prepare a DOCX that will later be opened in Hancom Office and saved as .hwp/.hwpx

What it does

Full pipeline from source .md to styled .docx:

1. Converts Obsidian ![[filename]] wikilinks to standard markdown links (sed preprocessing)
2. Runs Pandoc to produce unstyled .docx with TOC and embedded images
3. Applies table styling via python-docx to every table (including nested):
   • tblLook conditional formatting flags → zeroed (prevents style override)
   • Content-proportional column widths (Korean chars weighted 1.7×, min 900 dxa, budget 9000 dxa)
   • fixed table layout (widths respected in Hancom)
   • Row-based alternating shading — header navy, data rows blue/white alternating
   • Uniform column styling (no first/last column special case)
   • Cell vertical center alignment
   • Minimal cell margins (top/bottom 20 dxa, left/right 80 dxa)
   • Zero paragraph spacing, line spacing 1.0, empty paragraphs stripped
   • Font size 10pt throughout (run-level + pPr/rPr default)
4. (Optional) Prepends a full-page cover image with page break

Inputs

Required:

• Input .md file path
• Image folder path (for Pandoc to resolve ![[filename]])
• Output .docx path

Optional:

• Cover image (--cover path/to/cover.jpg)

Default palette (Korean gov proposal)

Element	Hex
Header background	#1F3864 (dark navy)
Header text	#FFFFFF (white, bold)
Odd data rows	#D9E2F3 (light blue)
Even data rows	#FFFFFF (white)
Body text	#000000 (black)

Customization: edit NAVY, LIGHT, WHITE constants at top of scripts/style_docx_tables.py.

Invocation

Full pipeline (most common)

bash ~/.claude/skills/docx-table-styler/scripts/md_to_docx.sh \
    "proposal.md" \
    "./도식" \
    "proposal.docx"

With cover page

bash ~/.claude/skills/docx-table-styler/scripts/md_to_docx.sh \
    "proposal.md" \
    "./도식" \
    "proposal.docx" \
    --cover "cover.jpg"

Style an existing DOCX (skip MD conversion)

python3 ~/.claude/skills/docx-table-styler/scripts/style_docx_tables.py \
    input.docx output.docx

Add cover page to existing DOCX

python3 ~/.claude/skills/docx-table-styler/scripts/add_cover_page.py \
    input.docx cover.jpg output.docx

Success criteria

• Script exits 0
• Output file size ≥ input (images preserved)
• unzip -p output.docx word/document.xml | tr '<' '\n' | grep -c '^w:tbl ' matches expected table count
• Every table has navy header row visible in Word/Hancom
• Short-text columns render narrow; long-text columns render wide
• No empty vertical whitespace within cells
• Embedded images count in word/media/ matches source
• Korean text renders without substitution (visual check)

Constraints & pitfalls

• Obsidian wikilinks: Pandoc does NOT understand ![[file]] — the sed preprocessing step is mandatory. The wrapper script handles this; if bypassing to call pandoc directly, run the sed step first.
• python-docx API quirks:
  • CT_PPr.get_or_add_rPr() does NOT exist — create w:rPr manually via OxmlElement.
  • table.style = "Table Grid" may raise KeyError if the style is not present in the source DOCX — wrap in try/except (already handled in script).
  • row.cells returns merged cells multiple times; the script guards with min(len(cells), num_cols).
• Nested tables: the script recursively descends through cell.tables.
• Hancom HWP compatibility: requires BOTH tblGrid/gridCol widths AND tcW per cell AND tblLayout w:type="fixed". The script sets all three.
• Text color on empty paragraphs: must set BOTH run-level color AND paragraph-level pPr/rPr/w:color so empty paragraphs under navy header render white, not black.
• Scale: 104-table document processes in ~5s on M-series Mac; larger docs scale linearly.

Known limitations / open branching decisions

• Row banding only (no column banding alternative)
• Single header row assumed (multi-row header detection not implemented)
• LTR languages only (no RTL handling)
• Wide tables (10+ columns) may look cramped at 9000 dxa budget — consider landscape page orientation
• HWP conversion is manual (open in Hancom → save as .hwp); no CLI automation included

After this skill

User opens the resulting .docx in Hancom 한글 → 파일 → 다른 이름으로 저장 → .hwp or .hwpx.

Hancom post-processing typically needed:

• Font substitution (제안서 표준: 함초롬바탕 / 함초롬돋움)
• Page margin adjustment (정부 표준: 상하좌우 25mm)
• TOC regeneration (입력 → 차례 → 차례 만들기)

Verification commands

# Table count preservation
unzip -p output.docx word/document.xml | tr '<' '\n' | grep -c '^w:tbl '

# Image count preservation
unzip -l output.docx | grep -c 'word/media/'

# Spot-check shading on first table
unzip -p output.docx word/document.xml | grep -o 'w:fill="[0-9A-Fa-f]*"' | head -5

File layout

~/.claude/skills/docx-table-styler/
├── SKILL.md                          # This file
└── scripts/
    ├── md_to_docx.sh                 # Full pipeline wrapper
    ├── style_docx_tables.py          # Table styling engine
    └── add_cover_page.py             # Optional cover page inserter