1 | # Author: Patrik Svestka
|
---|
2 | # Licence: MIT
|
---|
3 | # Version 1.0.0
|
---|
4 |
|
---|
5 | # Steps to ensure that all the *.st files, maintained in mercurial, are UTF-8 without BOM
|
---|
6 |
|
---|
7 | # Powershell script:
|
---|
8 | # 0) find all files stored in mercurial
|
---|
9 | # 1) change the encoding
|
---|
10 | # changing encoding to UTF8 without BOM
|
---|
11 | # 2) Check if there is a directive:
|
---|
12 | # 3) Add it if missing
|
---|
13 | # 4) Try to recompile the whole Smalltalk/X solution
|
---|
14 |
|
---|
15 | # any encoding -> UTF8 without BOM
|
---|
16 | # inspired by https://stackoverflow.com/questions/18684793/powershell-batch-change-files-encoding-to-utf-8
|
---|
17 |
|
---|
18 | $Utf8NoBomEncoding = New-Object System.Text.UTF8Encoding($False)
|
---|
19 |
|
---|
20 | #$source_path = 'C:\prg_sdk\stx8-jv_swing\build\stx'
|
---|
21 | $source_path = (Resolve-Path .\).Path
|
---|
22 |
|
---|
23 | $log_file = "$source_path\unicode_conversion.log"
|
---|
24 |
|
---|
25 | $header = """{ Encoding: utf8 }""`n"
|
---|
26 |
|
---|
27 | # get only smalltalk source files (*.st) that are stored only in Mercurial
|
---|
28 | ForEach ($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 }) ) {
|
---|
29 |
|
---|
30 | # getting all required details from current file
|
---|
31 | $destination_folder = (Get-Item $file_with_path).DirectoryName
|
---|
32 | $content = Get-Content $file_with_path
|
---|
33 | $file_only = Get-ChildItem $file_with_path | % {$_.Name}
|
---|
34 |
|
---|
35 | # change encoding only when file is not empty
|
---|
36 | If ($content -ne $null) {
|
---|
37 | If($content -match '"{ Encoding: utf8 }"') {
|
---|
38 | Write-Output "$file_with_path --> correct file header (changing only encoding)" | Out-File -Append -Encoding UTF8 -FilePath $log_file
|
---|
39 | [System.IO.File]::WriteAllLines($file_with_path, $content, $Utf8NoBomEncoding)
|
---|
40 |
|
---|
41 | # Keep the Unix style line end - https://stackoverflow.com/questions/19127741/replace-crlf-using-powershell/19128003#19128003
|
---|
42 | # .NET solution for converting windows end (crlf) to (lf only)
|
---|
43 | $text_with_unix_eol = [IO.File]::ReadAllText($file_with_path) -replace "`r`n", "`n"
|
---|
44 | [IO.File]::WriteAllText($file_with_path, $text_with_unix_eol)
|
---|
45 |
|
---|
46 | } Else {
|
---|
47 | Write-Output "$file_with_path -!-> Adding encoding header + changing the encoding" | Out-file -Append -Encoding UTF8 -FilePath $log_file
|
---|
48 |
|
---|
49 | # Changing encoding to UTF8 without BOM
|
---|
50 | [System.IO.File]::WriteAllLines($file_with_path, $content, $Utf8NoBomEncoding)
|
---|
51 |
|
---|
52 | # storing the original path into a stack
|
---|
53 | pushd -Path $destination_folder -StackName script_directory
|
---|
54 | # adding header to a temp file (random name)
|
---|
55 | $header | Set-Content asd3232d_tempfile.txt
|
---|
56 | # Adding the original file at the end of the temp file
|
---|
57 | Get-Content $file_with_path -ReadCount 8192 | Add-Content asd3232d_tempfile.txt
|
---|
58 | # removing the original file
|
---|
59 | Remove-item $file_with_path
|
---|
60 | # renaming the temp file into the original file name
|
---|
61 | Rename-Item asd3232d_tempfile.txt -NewName $file_only
|
---|
62 | popd -StackName script_directory
|
---|
63 |
|
---|
64 | # Keep the Unix style line end
|
---|
65 | # Convert CRLFs to LFs only.
|
---|
66 | # .NET solution for converting windows end (crlf) to (lf only)
|
---|
67 | $text_with_unix_eol = [IO.File]::ReadAllText($file_with_path) -replace "`r`n", "`n"
|
---|
68 | [IO.File]::WriteAllText($file_with_path, $text_with_unix_eol)
|
---|
69 | }
|
---|
70 | } Else {
|
---|
71 | Write-Host "Empty file at: $file_with_path"
|
---|
72 | }
|
---|
73 | }
|
---|
74 |
|
---|
75 |
|
---|
76 |
|
---|
77 |
|
---|