arxiv-translator

name	arxiv-translator
description	Automatically downloads LaTeX source from arXiv, translates it to a target language (default Chinese), and compiles it to PDF using xelatex. Uses modern xeCJK for multilingual support. Trigger when the user provides an arXiv URL, paper ID, or natural language query about an arXiv paper and wants it translated/compiled.

arXiv LaTeX Translator & Compiler (Modern Solution)

This skill handles the end-to-end workflow of converting an arXiv paper's LaTeX source into a translated PDF on Windows, macOS, and Linux using modern xeCJK and fontspec.

Workflow

0. Pre-flight Dependency Check

Before starting, verify that all required system tools are installed.

• Windows (PowerShell): Get-Command curl, tar, xelatex, bibtex -ErrorAction SilentlyContinue
• macOS / Linux (Bash): for cmd in curl tar xelatex bibtex; do command -v $cmd >/dev/null 2>&1 || echo "Missing: $cmd"; done

1. Identify & Download

• Extract the ID (e.g., 2508.11825v2) from the URL or find it via search.
• Create a target directory and download/extract the source and the paper pdf (for final check) using native tools (curl, tar).
  • Abstract: https://arxiv.org/abs/xxxx.xxxxx
  • Source: https://arxiv.org/src/xxxx.xxxxx
  • PDF: https://arxiv.org/pdf/xxxx.xxxxx.pdf

2. Analyze & Modularize (For Large Papers)

• Read the original .pdf file to build a mental map.
• Modularization Strategy (Crucial for Reliability):
  • DO NOT use the replace tool for content translation. It is prone to whitespace/encoding mismatches.
  • Instead, YOU MUST execute the standardized script located at scripts/split_tex.py (relative to this SKILL directory) to safely split the monolithic .tex file into sub-files by \section. Do not attempt to write a custom splitting script.
    • Command to run: Locate the skill directory and run python3 <path_to_skill_dir>/scripts/split_tex.py <main_file.tex>
  • For small files, you may still use replace or directly overwrite the file with write_file.
• Terminology Glossary: Create a local mental glossary of 5-10 core technical terms to ensure consistency.

3. Atomic Translation (Write-File Overwrite)

• Work on Chunks: Translate one sub-file (or one section) at a time.
• Overwrite Mode: Use the write_file tool to write the fully translated content into the sub-files. This avoids the "string not found" errors common with the replace tool.
• Content Preservation: Rigorously preserve LaTeX commands, math environments, citations, and references.
• Modern Chinese Support (xeCJK):
  • Inject \usepackage{xeCJK} and \setCJKmainfont{Microsoft YaHei} (or platform equivalent) into the main preamble.

4. Compile & Auto-Fix

• Use xelatex -interaction=nonstopmode.
• Standard compilation loop: xelatex -> bibtex -> xelatex x2.
• Error Auto-Fixing: If compilation fails, read the .log file, identify the error line, and fix the corresponding sub-file using write_file or replace.

5. Integrity Verification & Review

• Completeness Check: Verify all sub-files in sections/ are translated.
• Visual Check: Compare the translated PDF with the original.
• Clean up intermediate files and move the final _CN.pdf to the root.

Guidelines

• Prefer write_file over replace for Translation: Overwriting a small, dedicated section file is 100% reliable compared to searching for a literal string in a 20,000-line file.
• Font Selection:
  • Windows: Microsoft YaHei or SimSun.
  • macOS: PingFang SC.
  • Linux: Noto Sans CJK SC.
• Preserve Template: Use xeCJK to minimize style interference.
• Heredocs: Use single-quoted heredocs in shells to handle backslashes correctly.