JEI_Import_Documents-App/recon/fix_plan_dryrun.ps1

# JEI Line Fix Plan (dry-run)
# For each affected invoice, parse PDF + DB and output the proposed deletes + adds.
# Does NOT make any changes. Just produces a plan.

$ErrorActionPreference = 'Stop'

$pbUrl   = 'https://pb.agent.apphub.co.za'
$envPath = 'C:\Users\SJ\.minimax-agent\projects\JEI\.env.local'
$reconDir = 'C:\Users\SJ\.minimax-agent\projects\JEI\recon'
$txtsDir  = "$reconDir\txts"

$token = (Get-Content $envPath | Select-String -Pattern '^PB_TOKEN=' | ForEach-Object { $_ -replace '^PB_TOKEN=','' }).Trim()
$headers = @{ Authorization = "Bearer $token" }

# Invoices to dedupe
$dedupInvoices = @(
  '151778','151779','155655','155743','156092','156703','160107','160453',
  '161454','161611','162609','162923','163374','163841','163842','151346','161453'
)

$codeCache = @{}
function Get-StockCode($sid) {
  if ($codeCache.ContainsKey($sid)) { return $codeCache[$sid] }
  $r = Invoke-RestMethod -Uri "$pbUrl/api/collections/JEI_stock_codes/records/$sid`?fields=code" -Headers $headers -Method Get
  $codeCache[$sid] = $r.code
  $r.code
}

function Get-PdfLines($txtPath) {
  if (-not (Test-Path $txtPath)) { return @() }
  $txt = [System.IO.File]::ReadAllText($txtPath, [System.Text.Encoding]::UTF8)
  $pages = $txt -split "`f"
  $lines = @()
  for ($i=0; $i -lt $pages.Count; $i++) {
    $p = $pages[$i]
    $matches = [regex]::Matches($p, '(?m)^\s*(\d+\.\d+)\s+(\S+)')
    foreach ($m in $matches) {
      $code = $m.Groups[2].Value.Trim() -replace '\s+', ' '
      $qty  = [double]$m.Groups[1].Value
      $lines += [pscustomobject]@{ qty = $qty; code = $code; page = $i+1 }
    }
  }
  $lines
}

function Normalize-Code($code) {
  ($code -replace '\s+', ' ').Trim().ToUpper()
}

$totalDeletes = 0
$totalAdds = 0
$reportLines = @()
$plan = @()

$reportLines += "==================================================="
$reportLines += "FIX PLAN -- DRY RUN (no DB changes)"
$reportLines += "==================================================="

foreach ($invNo in $dedupInvoices) {
  $txtPath = "$txtsDir\$invNo.txt"
  if (-not (Test-Path $txtPath)) {
    $reportLines += ""
    $reportLines += "==== $invNo ===="
    $reportLines += "  SKIP -- no PDF text file"
    continue
  }

  $invRec = Invoke-RestMethod -Uri "$pbUrl/api/collections/JEI_invoices/records?filter=invoice_no=`"$invNo`"&fields=id" -Headers $headers -Method Get
  if ($invRec.items.Count -eq 0) { continue }
  $invId = $invRec.items[0].id

  $db = Invoke-RestMethod -Uri "$pbUrl/api/collections/JEI_invoice_lines/records?filter=invoice=`"$invId`"&perPage=500" -Headers $headers -Method Get
  $dbLines = foreach ($r in $db.items) {
    [pscustomobject]@{
      id       = $r.id
      line_no  = [int]$r.line_no
      code     = Get-StockCode $r.stock_code
      code_raw = $r.stock_code
      qty      = [double]$r.quantity
      net      = [double]$r.net_amount
      unit     = [double]$r.unit_price
      country  = $r.country_of_origin
      commodity = $r.commodity_code
    }
  }

  $pdfLines = Get-PdfLines $txtPath
  $pdfTuples = @{}
  foreach ($pl in $pdfLines) {
    $key = "$(Normalize-Code $pl.code)|$([string]$pl.qty)"
    if (-not $pdfTuples.ContainsKey($key)) {
      $pdfTuples[$key] = $pl
    }
  }

  $deletes = @()
  $adds    = @()
  $keep    = @()

  # Pass 1: exact (code, qty) match
  $matched = @{}
  foreach ($d in $dbLines) {
    $key = "$(Normalize-Code $d.code)|$([string]$d.qty)"
    if ($pdfTuples.ContainsKey($key)) {
      if (-not $matched.ContainsKey($key)) {
        $matched[$key] = @()
      }
      $matched[$key] += $d
    }
  }

  # Pass 2: fuzzy match — DB record's qty matches ANY PDF line's qty (same qty, same item but slightly different code name)
  # Already-matched DB records are excluded.
  $matchedDbIds = @{}
  foreach ($arr in $matched.Values) { foreach ($d in $arr) { $matchedDbIds[$d.id] = $true } }

  $qtyToPdfLines = @{}
  foreach ($pl in $pdfTuples.Values) {
    $qkey = "$([string]$pl.qty)"
    if (-not $qtyToPdfLines.ContainsKey($qkey)) { $qtyToPdfLines[$qkey] = @() }
    $qtyToPdfLines[$qkey] += $pl
  }

  foreach ($d in $dbLines) {
    if ($matchedDbIds.ContainsKey($d.id)) { continue }  # already exact-matched
    $qkey = "$([string]$d.qty)"
    if ($qtyToPdfLines.ContainsKey($qkey)) {
      # Match by qty alone — find any unused PDF line with this qty
      foreach ($pl in $qtyToPdfLines[$qkey]) {
        $plKey = "$(Normalize-Code $pl.code)|$([string]$pl.qty)"
        if (-not $matched.ContainsKey($plKey) -or ($matched[$plKey].Count -eq 1 -and $matched[$plKey][0].id -eq $d.id)) {
          # OK to match
          if (-not $matched.ContainsKey($plKey)) { $matched[$plKey] = @() }
          $matched[$plKey] += $d
          $matchedDbIds[$d.id] = $true
          break
        }
      }
    }
  }

  # Now unmatched DB records
  foreach ($d in $dbLines) {
    if (-not $matchedDbIds.ContainsKey($d.id)) {
      $deletes += [pscustomobject]@{ rec=$d; reason='NO_PDF_MATCH' }
    }
  }

  foreach ($k in $matched.Keys) {
    $arr = $matched[$k]
    if ($arr.Count -gt 1) {
      $sorted = $arr | Sort-Object id
      $keep += $sorted[0]
      foreach ($d in ($sorted | Select-Object -Skip 1)) {
        $deletes += [pscustomobject]@{ rec=$d; reason='DUPLICATE_OF_KEPT' }
      }
    } else {
      $keep += $arr[0]
    }
  }

  foreach ($k in $pdfTuples.Keys) {
    if (-not $matched.ContainsKey($k)) {
      $adds += $pdfTuples[$k]
    }
  }

  $reportLines += ""
  $reportLines += "==== $invNo ===="
  $reportLines += ("  PDF lines={0}  db_records={1}  will_keep={2}" -f $pdfTuples.Count, $dbLines.Count, $keep.Count)
  $reportLines += ("  Deletes: {0}" -f $deletes.Count)
  foreach ($d in $deletes) {
    $reportLines += ("    DELETE  rec={0,-22} line_no={1,3} code={2,-22} qty={3,5} net={4,10}  [{5}]" -f $d.rec.id, $d.rec.line_no, $d.rec.code, $d.rec.qty, $d.rec.net, $d.reason)
  }
  $reportLines += ("  Adds: {0}" -f $adds.Count)
  foreach ($a in $adds) {
    $reportLines += ("    ADD     code={0,-22} qty={1,5}  pdf-page={2}" -f $a.code, $a.qty, $a.page)
  }
  $totalDeletes += $deletes.Count
  $totalAdds += $adds.Count

  $plan += [pscustomobject]@{
    invoice   = $invNo
    pdf_lines = $pdfTuples.Count
    db_lines  = $dbLines.Count
    keep      = $keep.Count
    deletes   = @($deletes)
    adds      = @($adds)
  }
}

$reportLines += ""
$reportLines += "==================================================="
$reportLines += ("TOTALS -- would delete {0} records, add {1} records" -f $totalDeletes, $totalAdds)
$reportLines += "==================================================="

$reportLines | Out-File "$reconDir\fix_plan.txt" -Encoding UTF8

# Also save the structured plan as JSON for the apply step
$jsonOut = ($plan | ConvertTo-Json -Depth 8)
$jsonOut | Out-File "$reconDir\fix_plan.json" -Encoding UTF8

# Print to console
$reportLines | ForEach-Object { Write-Host $_ }