#!/bin/sh # EJ2 · 53 · texto de los PDF (sin red): pdftotext -layout; si < 200 caracteres, OCR (pdftoppm + tesseract spa) cd /w/interno/cotejo for f in docs/*.bin; do c=$(basename $f .bin); o=txt/$c.txt [ -s "$o" ] && continue pdftotext -layout "$f" "$o" 2>/dev/null n=$(wc -c < "$o" 2>/dev/null || echo 0) if [ "${n:-0}" -lt 200 ]; then mkdir -p /w/tmp/ocr/$c; pdftoppm -r 200 -l 6 -png "$f" /w/tmp/ocr/$c/p 2>/dev/null : > "$o"; for p in /w/tmp/ocr/$c/p*.png; do tesseract "$p" - -l spa 2>/dev/null >> "$o" || tesseract "$p" - 2>/dev/null >> "$o"; done echo "$c OCR $(wc -c < $o)" else echo "$c texto $n"; fi done