#!/bin/bash
set -u
BRAIN=/root/aria/brain
OUT=/root/aria-export-pack
rm -rf "$OUT"; mkdir -p "$OUT/notes"
SRC_DIRS=("02-Wissen" "03-Recherchen" "05-Referenzen" "00-Inbox/Videos")
HARD='IBAN|Steuer-ID|github_pat|sk-ant|\bGehalt\b|Vertragsstrafe|kaiis\.de|167\.233\.56|aseckzai'
CLIENT='BMW|Kadi|KADiCon|Durrani|Adesso'
inc=0; exc=0; scr=0
: > "$OUT/_excluded.txt"; : > "$OUT/_included.txt"; : > "$OUT/_scrubbed.txt"
for d in "${SRC_DIRS[@]}"; do
  [ -d "$BRAIN/$d" ] || continue
  while IFS= read -r f; do
    rel="${f#"$BRAIN"/}"
    if echo "$rel" | grep -qiE 'bmw|kadi|durrani|kadicon|adesso|kais'; then
      echo "$rel :: filename-client" >> "$OUT/_excluded.txt"; exc=$((exc+1)); continue
    fi
    if grep -qiE "$HARD" "$f"; then
      m=$(grep -ioE "$HARD" "$f" | sort -u | tr '\n' ',' | sed 's/,$//')
      echo "$rel :: hard[$m]" >> "$OUT/_excluded.txt"; exc=$((exc+1)); continue
    fi
    ch=$(grep -ioE "$CLIENT" "$f" | wc -l)
    if [ "$ch" -ge 6 ]; then
      echo "$rel :: client-core($ch)" >> "$OUT/_excluded.txt"; exc=$((exc+1)); continue
    fi
    mkdir -p "$OUT/notes/$(dirname "$rel")"
    sed -E \
      -e 's/\bBMW\b/[Automotive-Kunde]/gI' \
      -e 's/Kadi-v2/[SaaS-Projekt]/gI' -e 's/\bKadi\b/[SaaS-Projekt]/gI' \
      -e 's/\bKADiCon\b/[Firma]/gI' -e 's/\bKADi\b/[Firma]/gI' \
      -e 's/\bDurrani\b/[Projekt-D]/gI' \
      -e 's/\bAdesso\b/[Partner]/gI' \
      -e 's/Kais Aseckzai/[Owner]/gI' -e 's/\bKais\b/[Owner]/gI' -e 's/\bAseckzai\b/[Owner]/gI' \
      -e 's/#?037493/[brand-color]/gI' -e 's/#?1C69D4/[brand-color]/gI' \
      -e 's/[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Za-z]{2,}/[email]/g' \
      "$f" > "$OUT/notes/$rel"
    [ "$ch" -gt 0 ] && { echo "$rel :: $ch refs gescrubbt" >> "$OUT/_scrubbed.txt"; scr=$((scr+1)); }
    echo "$rel" >> "$OUT/_included.txt"; inc=$((inc+1))
  done < <(find "$BRAIN/$d" -name '*.md' -type f)
done
echo "INCLUDED=$inc EXCLUDED=$exc SCRUBBED=$scr"
