PowerShell Chinese encoding + docx read + grep download + Word COM issues on Windows #27476

Description

On Windows 11, OpenCode encountered several toolchain issues when asked to process .docx files (read and write) and handle non-ASCII text through PowerShell:

1. Read tool: cannot handle .docx files

Read returns Cannot read binary file for .docx (ZIP+XML). This is a very common format for documents.

2. Grep tool: transport error downloading ripgrep

Transport error (GET https://github.com/BurntSushi/ripgrep/releases/download/...)

Prevents file content search when ripgrep is not yet installed.

3. PowerShell $var: syntax causes ParserError

When writing PowerShell scripts, patterns like "[$file:$lineNum] $line" trigger:

Variable reference is not valid. ':' was not followed by a valid variable name character.

Generated code snippets should use ${file} to be safe.

4. PowerShell console garbles non-ASCII characters

Chinese characters output via Write-Host appear as mojibake, requiring workarounds (write to file, then Read).

5. Word COM (New-Object -ComObject Word.Application) fails when Word is absent

Throws Unexpected HRESULT on systems without Microsoft Word installed. No fallback available.

6. Read tool truncates single-line files at 2000 chars

Files without line breaks (e.g., extracted docx text) appear as one line truncated to 2000 characters.

Steps to reproduce

1. Read a .docx file → Cannot read binary file
2. grep on Windows without preinstalled ripgrep → Transport error
3. Write-Host "[$file:$lineNum] $line" in PowerShell → ParserError
4. New-Object -ComObject Word.Application without Word → HRESULT

Operating System

Windows 11

Terminal

PowerShell 7 (pwsh.exe)

OpenCode version

1.14.50