# kb_sanitize_pass_v1.0.ps1
# Objet : Sanitation �tendue de la KB canonique (ASCII-safe, suppression mojibake et '@{'),
#         PREVIEW/EXECUTE avec SAFE-WRITE. PS 5.1 compatible, UTF-8 BOM.

[CmdletBinding()]
param(
  [switch]$Preview,
  [switch]$Execute,
  [string]$Root = "\\DS-918\chatgpt\ChatGPT-Gouvernance-Projets\_registry",
  [string]$StageRoot = "C:\Temp_Gouvernance"
)

function Ensure-Dir([string]$Path){ if(-not(Test-Path -LiteralPath $Path)){ New-Item -ItemType Directory -Force -Path $Path | Out-Null } }
function Ensure-Parent([string]$Target){ $p=Split-Path -Parent $Target; if($p){ Ensure-Dir $p } }
function Get-NowIso(){ (Get-Date).ToString("yyyy-MM-ddTHH:mm:ssK") }
function Write-SafeText([string]$Target,[string]$Content,[string]$StageRoot){
  Ensure-Dir $StageRoot; Ensure-Parent $Target
  $tmp=Join-Path $StageRoot ("write_" + [IO.Path]::GetRandomFileName())
  $utf8=New-Object Text.UTF8Encoding($true)
  [IO.File]::WriteAllText($tmp,$Content,$utf8)
  $bak=$null
  if(Test-Path -LiteralPath $Target){
    $bak=$Target+"."+(Get-Date -f yyyyMMdd_HHmmss)+".bak"
    Copy-Item -LiteralPath $Target -Destination $bak -Force
  }
  $tmpR="$Target.tmp"; Copy-Item $tmp $tmpR -Force; Move-Item $tmpR $Target -Force; Remove-Item $tmp -Force
  return $bak
}
function Read-JsonSansFooter([string]$Path){
  $sr=New-Object IO.StreamReader($Path)
  try{
    $L=New-Object 'System.Collections.Generic.List[string]'
    while(-not $sr.EndOfStream){
      $line=$sr.ReadLine()
      if($line -match '^\s*---\s*DOC-VERSION-FOOTER'){ break }
      $L.Add($line)|Out-Null
    }
    ($L -join "`n")
  } finally { $sr.Dispose() }
}
function Normalize-Ascii-Extended([string]$s){
  if($null -eq $s){ return "" }
  # 1) guillemets/quotes/dashes typographiques corrects -> ASCII
  $s = $s -replace "`u2018","'" -replace "`u2019","'" -replace "`u201A","'" -replace "`u201B","'"
  $s = $s -replace "`u201C",'"' -replace "`u201D",'"' -replace "`u201E",'"'
  $s = $s -replace "`u2013","-" -replace "`u2014","-"
  $s = $s -replace "`u00A0"," "

  # 2) mojibake fr�quents -> ASCII
  $s = $s -replace "’","'"
  $s = $s -replace "“",'"'
  $s = $s -replace "�?",'"'
  $s = $s -replace "–","-"
  $s = $s -replace "—","-"
  $s = $s -replace "•","-"
  $s = $s -replace "Â",""
  $s = $s -replace "�",""
  $s = $s -replace "?","?"

  # 3) r�sidus de hashtables stringifi�es
  $s = $s -replace "@\{","("    # neutraliser l'ouverture
  # (on laisse les '}' telles quelles pour ne pas manger des JSON valides si jamais)

  # 4) suppr. contr�les (hors CR/LF/TAB) + compaction des espaces
  $s = [regex]::Replace($s,"[^\P{C}\r\n\t]"," ")
  $s = [regex]::Replace($s,"\s{2,}"," ")
  $s.Trim()
}

function Sanitize-Entry($e){
  # Sanitize toutes les cha�nes simples et tableaux de cha�nes
  $id    = Normalize-Ascii-Extended ([string]$e.id)
  $title = Normalize-Ascii-Extended ([string]$e.title)
  $work  = Normalize-Ascii-Extended ([string]$e.workaround)
  $note  = Normalize-Ascii-Extended ([string]$e.note)
  $fix   = Normalize-Ascii-Extended ([string]$e.fix)

  $tags=@()
  if($e.tags){ foreach($t in $e.tags){ $tags += (Normalize-Ascii-Extended ([string]$t)) } }

  $seenIn=@()
  if($e.seen_in_threads){
    if($e.seen_in_threads -is [string]){ $seenIn=@(Normalize-Ascii-Extended ([string]$e.seen_in_threads)) }
    else { $seenIn=@($e.seen_in_threads | ForEach-Object { Normalize-Ascii-Extended ([string]$_) }) }
  }

  [ordered]@{
    id=$id; title=$title; blocking=[bool]$e.blocking; workaround=$work; note=$note; fix=$fix;
    tags=@($tags); seen_in_threads=@($seenIn); last_seen=[string]$e.last_seen
  }
}

# ---- Chargement KB canonique ----
$Kb = Join-Path $Root "bug_kb\BUG_KB.json.txt"
if(-not (Test-Path -LiteralPath $Kb)){ Write-Host "[ERR] KB absente: $Kb"; exit 2 }

$raw  = Get-Content -LiteralPath $Kb -Raw
$head = ($raw -split "(?m)^--- DOC-VERSION-FOOTER")[0]
$pat  = '�|�|’|“|�|?|@\{'
$badBefore = ([regex]::Matches($head,$pat)).Count

try { $obj = $head | ConvertFrom-Json -ErrorAction Stop } catch { Write-Host "[ERR] KB JSON invalide: $($_.Exception.Message)"; exit 3 }
if(-not $obj.entries){ $obj | Add-Member -Name entries -MemberType NoteProperty -Value @() }

# Sanitation
$clean = New-Object System.Collections.ArrayList
foreach($e in $obj.entries){ $null = $clean.Add( (Sanitize-Entry $e) ) }

$out = [ordered]@{ entries=@($clean); updated=(Get-NowIso) }
$json = ($out | ConvertTo-Json -Depth 6 -Compress)
$badAfter = ([regex]::Matches($json,$pat)).Count

$footer = "`r`n`r`n--- DOC-VERSION-FOOTER ---`r`nGenerated: $($out.updated)`r`nPolicy: TXT-ONLY v1.0; SAFE-WRITE v1.1; GOV_SCRIPT_GATE v1.3`r`nSource: KB_SANITIZE_PASS_v1.0`r`n"

if($Preview -or (-not $Execute)){
  Write-Host "== PREVIEW :: KB SANITIZE PASS =="
  Write-Host ("Entries={0}  badBefore={1}  badAfter_if_written={2}" -f $clean.Count, $badBefore, $badAfter)
  Write-Host ("Will write -> {0}  out-json-bytes�{1:n0}" -f $Kb, ([Text.Encoding]::UTF8.GetByteCount($json)))
  Write-Host "No write performed (Preview)."
  exit 0
}

$bak = Write-SafeText -Target $Kb -Content ($json + $footer) -StageRoot $StageRoot
Write-Host ("[OK] KB sanitiz�e et r��crite: {0}" -f $Kb)
Write-Host ("Backup: {0}" -f ($bak ? $bak : "<none>"))