# =============================================================================
# check_rtl_text.ps1 —— detect stray non-Verilog text in the RTL sources.
#
# WHY: this project's .v files were twice corrupted by a text tool that inserted a
# line of README prose into them ("Tcl ..." / a Chinese sentence), which made
# Vivado fail with
#     ERROR: [Synth 8-2715] syntax error near "..."  'non-printable character 0xe4'
# and produced a cascade of bogus errors in the included ROM .vh. Running this
# script takes a second and catches it before spending a synthesis run.
#
# WHAT IT CHECKS
#   1) a line inside .v/.vh/.xdc that has non-ASCII characters *outside* a comment
#      (i.e. real code position) -> corruption;
#   2) a line that looks like pasted prose/markdown (leading "|", "**", or a line
#      that is bare text after `endmodule`).
#
# USAGE (from the repository root or from pipelined-cpu/):
#   powershell -ExecutionPolicy Bypass -File pipelined-cpu\tools\check_rtl_text.ps1
# Exit code 0 = clean, 1 = found something.
# =============================================================================
$ErrorActionPreference = 'Stop'

$srcs = Join-Path (Split-Path -Parent $PSScriptRoot) 'srcs'
if (-not (Test-Path $srcs)) { Write-Output "ERROR: cannot find $srcs"; exit 2 }

$bad = 0
$files = Get-ChildItem -Recurse -Path $srcs -Include *.v, *.vh, *.xdc -File

foreach ($f in $files) {
    $rel = $f.FullName.Substring($srcs.Length).TrimStart('\', '/')
    $lines = @(Get-Content -LiteralPath $f.FullName)
    $inBlockComment = $false

    for ($i = 0; $i -lt $lines.Count; $i++) {
        $line = $lines[$i]
        $lineNo = $i + 1

        # ---- strip comments so we only look at code positions ----
        $code = $line
        if ($f.Extension -eq '.xdc') {
            $h = $code.IndexOf('#'); if ($h -ge 0) { $code = $code.Substring(0, $h) }
        }
        else {
            $c = $code.IndexOf('//'); if ($c -ge 0) { $code = $code.Substring(0, $c) }
            if ($inBlockComment) { $code = '' }
            if ($line -match '/\*') { $inBlockComment = $true }
            if ($line -match '\*/') { $inBlockComment = $false }
        }

        $trim = $line.Trim()

        # ---- check 1: non-ASCII in code position ----
        if ($code -match '[^\x00-\x7F]') {
            Write-Output ("!! $rel`:$lineNo  non-ASCII text in code position:")
            Write-Output ("        " + $trim)
            $bad++
            continue
        }

        # ---- check 2: pasted prose / markdown leftovers ----
        if ($trim -match '^\||^\*\*|^#[^0-9#]' -and $f.Extension -ne '.xdc') {
            Write-Output ("!! $rel`:$lineNo  looks like pasted prose:")
            Write-Output ("        " + $trim)
            $bad++
        }
    }
}

Write-Output ""
Write-Output ("scanned " + $files.Count + " files under srcs/")
if ($bad -eq 0) {
    Write-Output "RTL text check: OK (no stray non-Verilog text found)."
    exit 0
}
Write-Output ("RTL text check: FAILED, " + $bad + " suspicious line(s) - fix them before synthesis.")
Write-Output "Hint: such lines are usually injected by an editor/tool; remove them, then re-run this script."
exit 1