Ticket #239: fixing_encoding_and_adding_encoding_directive.ps1

# Author: Patrik Svestka
# Licence: MIT
# Version 1.0.0

# Steps to ensure that all the *.st files, maintained in mercurial, are UTF-8 without BOM

# Powershell script:
# 0) find all files stored in mercurial
# 1) change the encoding
# changing encoding to UTF8 without BOM
# 2) Check if there is a directive:
# 3) Add it if missing 
# 4) Try to recompile the whole Smalltalk/X solution

# any encoding -> UTF8 without BOM
# inspired by https://stackoverflow.com/questions/18684793/powershell-batch-change-files-encoding-to-utf-8

$Utf8NoBomEncoding = New-Object System.Text.UTF8Encoding($False)

#$source_path = 'C:\prg_sdk\stx8-jv_swing\build\stx'
$source_path = (Resolve-Path .\).Path

$log_file = "$source_path\unicode_conversion.log"

$header = """{ Encoding: utf8 }""`n"

# get only smalltalk source files (*.st) that are stored only in Mercurial
ForEach ($file_with_path in (Get-ChildItem -Directory -Path $source_path -exclude CVS | where { !(Get-ChildItem $_ -Directory -Recurse -Filter CVS) } | Get-childItem -File -Recurse -Filter *.st | % { $_.FullName }) ) {

    # getting all required details from current file
    $destination_folder = (Get-Item $file_with_path).DirectoryName
    $content = Get-Content $file_with_path
    $file_only = Get-ChildItem $file_with_path | % {$_.Name}

    # change encoding only when file is not empty
    If ($content -ne $null) {
        If($content -match '"{ Encoding: utf8 }"') {
            Write-Output "$file_with_path --> correct file header (changing only encoding)" | Out-File -Append -Encoding UTF8 -FilePath $log_file
            [System.IO.File]::WriteAllLines($file_with_path, $content, $Utf8NoBomEncoding)

            # Keep the Unix style line end - https://stackoverflow.com/questions/19127741/replace-crlf-using-powershell/19128003#19128003
            # .NET solution for converting windows end (crlf) to (lf only)
            $text_with_unix_eol = [IO.File]::ReadAllText($file_with_path) -replace "`r`n", "`n"
            [IO.File]::WriteAllText($file_with_path, $text_with_unix_eol)

        } Else {
            Write-Output "$file_with_path -!-> Adding encoding header + changing the encoding" | Out-file -Append -Encoding UTF8 -FilePath $log_file

            # Changing encoding to UTF8 without BOM
            [System.IO.File]::WriteAllLines($file_with_path, $content, $Utf8NoBomEncoding)

            # storing the original path into a stack
            pushd -Path $destination_folder -StackName script_directory
            # adding header to a temp file (random name)
            $header | Set-Content asd3232d_tempfile.txt
            # Adding the original file at the end of the temp file
            Get-Content $file_with_path -ReadCount 8192 | Add-Content asd3232d_tempfile.txt
            # removing the original file
            Remove-item $file_with_path
            # renaming the temp file into the original file name
            Rename-Item asd3232d_tempfile.txt -NewName $file_only
            popd -StackName script_directory

            # Keep the Unix style line end
            # Convert CRLFs to LFs only.
            # .NET solution for converting windows end (crlf) to (lf only)
            $text_with_unix_eol = [IO.File]::ReadAllText($file_with_path) -replace "`r`n", "`n"
            [IO.File]::WriteAllText($file_with_path, $text_with_unix_eol)
        }
    } Else {
        Write-Host "Empty file at: $file_with_path"
    }
}