PowderCoatingLogix/tools/Fix-Encoding.ps1

# Fix-Encoding.ps1
# Replaces corrupted UTF-8-decoded-as-Windows-1252 characters with HTML entities
# in all .cshtml files under src/PowderCoating.Web/Views/.
#
# Each corruption pattern is built from exact Unicode code points at runtime so
# this script file itself never contains a literal non-ASCII character.
# Uses [System.IO.File]::ReadAllText / WriteAllText with explicit UTF-8 (no BOM)
# to avoid PowerShell 5.1's default ANSI encoding on Get-Content / Set-Content.
#
# Usage:
#   .\tools\Fix-Encoding.ps1            # fix all .cshtml files
#   .\tools\Fix-Encoding.ps1 -DryRun    # report what would change, write nothing

param(
    [switch]$DryRun
)

$utf8NoBom = New-Object System.Text.UTF8Encoding $false

# ---------------------------------------------------------------------------
# Corruption map — each entry is the 3-char corrupt sequence (as .NET strings)
# and the HTML entity to replace it with.
#
# How the corruption happens:
#   UTF-8 bytes for a Unicode char are misread as Windows-1252, producing 3
#   garbled characters.  Those garbled chars are then saved back as UTF-8,
#   so the file now contains their UTF-8 encoding instead of the original char.
#
# Pattern construction:
#   Original char  UTF-8 bytes   Win-1252 chars (Unicode)         Corrupt string
#   em dash        E2 80 94      U+00E2 U+20AC U+201D             a + euro + rdquo
#   en dash        E2 80 93      U+00E2 U+20AC U+201C             a + euro + ldquo
#   ellipsis       E2 80 A6      U+00E2 U+20AC U+00A6             a + euro + brvbar
#   lsaquo         E2 80 B9      U+00E2 U+20AC U+00B9             a + euro + sup1
#   rsaquo         E2 80 BA      U+00E2 U+20AC U+00BA             a + euro + ordm
#   box horiz      E2 94 80      U+00E2 U+201D U+20AC             a + rdquo + euro
#   lsquo          E2 80 98      U+00E2 U+20AC U+02DC             a + euro + tilde
#   rsquo          E2 80 99      U+00E2 U+20AC U+2122             a + euro + trade
#   ldquo          E2 80 9C      U+00E2 U+20AC U+0153             a + euro + oelig
#   bullet         E2 80 A2      U+00E2 U+20AC U+00A2             a + euro + cent
# ---------------------------------------------------------------------------

$a     = [char]0x00E2  # a with circumflex
$euro  = [char]0x20AC  # euro sign
$rdq   = [char]0x201D  # right double quotation mark
$ldq   = [char]0x201C  # left double quotation mark
$brvb  = [char]0x00A6  # broken bar
$sup1  = [char]0x00B9  # superscript one
$ordm  = [char]0x00BA  # masculine ordinal indicator
$tilde = [char]0x02DC  # small tilde
$trade = [char]0x2122  # trade mark sign
$oelig = [char]0x0153  # latin small letter oe
$cent  = [char]0x00A2  # cent sign

$replacements = @(
    # -- 3-char â€-corruption sequences (UTF-8 bytes misread as Windows-1252) ----
    @{ Find = "$a$euro$rdq";   Replace = '—'   }  # em dash   U+2014
    @{ Find = "$a$euro$ldq";   Replace = '–'   }  # en dash   U+2013
    @{ Find = "$a$euro$brvb";  Replace = '…'  }  # ellipsis  U+2026
    @{ Find = "$a$euro$sup1";  Replace = '‹'  }  # lsaquo    U+2039
    @{ Find = "$a$euro$ordm";  Replace = '›'  }  # rsaquo    U+203A
    @{ Find = "$a$rdq$euro";   Replace = '-'          }  # box horiz U+2500
    @{ Find = "$a$euro$tilde"; Replace = "'"          }  # lsquo     U+2018
    @{ Find = "$a$euro$trade"; Replace = "'"          }  # rsquo     U+2019
    @{ Find = "$a$euro$oelig"; Replace = '“'   }  # ldquo     U+201C
    @{ Find = "$a$euro$cent";  Replace = '•'    }  # bullet    U+2022

    # -- Partially-re-corrupted variants (3rd byte already normalised to ASCII) --
    # When an AI tool "fixed" the closing curly quote to ASCII before our script ran,
    # the corruption sequence ends with a plain ASCII double-quote (U+0022) or
    # single-quote (U+0027) instead of U+201C/201D. These won't match the 3-char
    # patterns above so they need their own entries.
    @{ Find = "$a$euro`"";  Replace = '—'  }  # ae + ASCII double-quote
    @{ Find = "$a$euro'";   Replace = '—'  }  # ae + ASCII single-quote (rare)

    # -- Right arrow (U+2192) corruption: UTF-8 E2 86 92 read as Win-1252 -----------
    # 0x86 = dagger (U+2020), 0x92 = right single quote (U+2019)
    # In JS comments we can safely render as ->
    @{ Find = [string][char]0x00E2 + [char]0x2020 + [char]0x2019; Replace = '->' }  # -> right arrow

    # -- Standalone smart/curly quotes introduced by AI tools or paste ----------
    # These have no place in .cshtml files. Curly double quotes break C# Razor
    # expressions; replace with ASCII. Em/en dashes in HTML → entities.
    @{ Find = [string][char]0x201C; Replace = '"'        }  # left double quote  → ASCII "
    @{ Find = [string][char]0x201D; Replace = '"'        }  # right double quote → ASCII "
    @{ Find = [string][char]0x2018; Replace = "'"        }  # left single quote  → ASCII '
    @{ Find = [string][char]0x2019; Replace = "'"        }  # right single quote → ASCII '
    @{ Find = [string][char]0x2014; Replace = '—'  }  # em dash (bare)
    @{ Find = [string][char]0x2013; Replace = '–'  }  # en dash (bare)
    @{ Find = [string][char]0x2026; Replace = '…' }  # ellipsis (bare)
)

$viewsPath = Join-Path $PSScriptRoot "..\src\PowderCoating.Web\Views"
$viewsPath = [System.IO.Path]::GetFullPath($viewsPath)

if (-not (Test-Path $viewsPath)) {
    Write-Error "Views directory not found: $viewsPath"
    exit 1
}

$files = Get-ChildItem -Path $viewsPath -Filter "*.cshtml" -Recurse
$changedCount = 0

foreach ($file in $files) {
    $content = [System.IO.File]::ReadAllText($file.FullName, [System.Text.Encoding]::UTF8)
    $updated = $content

    foreach ($r in $replacements) {
        $updated = $updated.Replace($r.Find, $r.Replace)
    }

    if ($updated -ne $content) {
        $rel = $file.FullName.Replace($viewsPath, 'Views')
        if ($DryRun) {
            Write-Host "[DRY RUN] Would fix: $rel" -ForegroundColor Yellow
        } else {
            [System.IO.File]::WriteAllText($file.FullName, $updated, $utf8NoBom)
            Write-Host "Fixed: $rel" -ForegroundColor Green
        }
        $changedCount++
    }
}

$verb = if ($DryRun) { "would be updated" } else { "updated" }
Write-Host "`nDone. $changedCount file(s) $verb." -ForegroundColor Cyan