# =============================================================================
# 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.
# =============================================================================
param(
    # -Fix: 自动删除"整行都是粘贴进来的正文"的坏行（保守：只删不含 ; ( ) = 等代码标点的整行）。
    #       被部分覆盖的行（例如某个端口行被改坏）只报告，需手工或 git 修。
    [switch]$Fix
)
$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('\', '/')
    # 用 .NET 按 UTF-8 读原始字节：PS 5.1 的 Get-Content 会按 GBK 解码，中文会乱。
    # 按行切开时保留行尾（\r\n / \n），写回时逐字节保持原样，不归一化换行/编码/BOM。
    $bytes = [System.IO.File]::ReadAllBytes($f.FullName)
    $text  = [System.Text.Encoding]::UTF8.GetString($bytes)
    $lines = @([regex]::Split($text, "(?<=\n)"))
    $drop  = @()

    for ($i = 0; $i -lt $lines.Count; $i++) {
        $line   = $lines[$i].TrimEnd("`r", "`n")
        $lineNo = $i + 1

        # ---- 只去掉 // (或 .xdc 的 #) 之后的部分，其余按代码位置检查 ----
        # 注意：这里不处理 /* */ 块注释（srcs 里没有块注释，之前因此漏报过坏行）
        $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) }
        }
        # 字符串字面量里的中文是合法的（如 $display("中文")），先去掉再判断
        $code = $code -replace '"[^"]*"', '""'

        $trim = $line.Trim()

        # ---- check 1: non-ASCII in code position ----
        if ($code -match '[^\x00-\x7F]') {
            # 整行只有正文（不含 ; ( ) = ` ' " 之类代码标点）⇒ 可以安全整行删掉
            $onlyProse = ($trim -notmatch '[;()=`''"]')
            $tag = if ($onlyProse) { '  [整行正文，可用 -Fix 删掉]' } else { '  [部分覆盖，需手工/git 修]' }
            Write-Output ("!! $rel`:$lineNo  non-ASCII text in code position${tag}:")
            Write-Output ("        " + $trim)
            if ($Fix -and $onlyProse) { $drop += $i }
            $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++
        }
    }

    if ($Fix -and $drop.Count -gt 0) {
        $kept = for ($i = 0; $i -lt $lines.Count; $i++) { if ($drop -notcontains $i) { $lines[$i] } }
        [System.IO.File]::WriteAllText($f.FullName, ($kept -join ''), (New-Object System.Text.UTF8Encoding($false)))
        Write-Output ("   -Fix: 已从 $rel 删除 " + $drop.Count + " 行（编码/换行保持不变），请重跑自检确认")
    }
}

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 "Hint1: 这类行通常是编辑器/工具粘贴进来的正文；整行正文可以用 -Fix 删掉："
Write-Output "       powershell -ExecutionPolicy Bypass -File tools\check_rtl_text.ps1 -Fix"
Write-Output "Hint2: 若报的是 [部分覆盖]，对照 git diff 手工修（被编辑器保存覆盖时，先关掉那个标签页再改）。"
exit 1