[WordPress → Astro] Phần 1 – Chuyển bài viết từ Database qua file Markdown

mkdir ~/blog-migration
mv ~/Downloads/*.xml ~/blog-migration/
cd ~/blog-migrationCode language: Nginx (nginx)

#!/usr/bin/env bash
# ==============================================================================
# wp-image-fixer.sh
# Replaces resized WordPress image URLs with originals in an XML export file.
# Usage:
#   ./wp-image-fixer.sh           # dry-run: scan and log results
#   ./wp-image-fixer.sh --apply   # apply replacements from cached log
# ==============================================================================

set -euo pipefail

LOG=".wp-image-check.log"
URLS=".wp-image-urls.tmp"
MAP=".wp-image-map.txt"
APPLY=false
PARALLEL=50
TIMEOUT=5

[[ "${1:-}" == "--apply" ]] && APPLY=true

info()    { echo "[info]  $*"; }
success() { echo "[ok]    $*"; }
warn()    { echo "[warn]  $*"; }
die()     { echo "[error] $*" >&2; exit 1; }

# ------------------------------------------------------------------------------
# Cleanup on exit or Ctrl+C
# ------------------------------------------------------------------------------
cleanup() {
    rm -f "$URLS"
    kill 0 2>/dev/null || true
}
trap cleanup INT TERM
trap 'rm -f "$URLS"' EXIT

# ------------------------------------------------------------------------------
# Detect existing log / prompt for input
# ------------------------------------------------------------------------------
if $APPLY; then
    [[ -f "$LOG" ]] || die "No scan found. Run the script without --apply first."
    INPUT=$(grep '^FILE|' "$LOG" | cut -d'|' -f2)
    info "Existing scan found for: $INPUT"
    echo
    read -rp "Apply fixes using cached results? (y/N): " confirm
    [[ "$confirm" != "y" ]] && exit 0
else
    if [[ -f "$LOG" ]]; then
        info "Removing previous scan results..."
        rm -f "$LOG"
    fi
    read -rp "Enter WordPress XML export file: " INPUT
    [[ -f "$INPUT" ]] || die "File not found: $INPUT"
    echo "FILE|$INPUT" > "$LOG"
fi

# ------------------------------------------------------------------------------
# DRY RUN — scan images and check originals
# ------------------------------------------------------------------------------
if ! $APPLY; then
    echo
    info "Extracting resized image URLs..."

    grep -oE 'https?://[^"[:space:]]+-[0-9]+x[0-9]+\.(jpg|jpeg|png|webp)' "$INPUT" \
        | sort -u > "$URLS"

    TOTAL=$(wc -l < "$URLS" | tr -d ' ')
    info "Found $TOTAL resized image URLs"
    echo

    if [[ "$TOTAL" -eq 0 ]]; then
        warn "No resized images found. Nothing to do."
        rm -f "$LOG"
        exit 0
    fi

    EST=$(( (TOTAL / PARALLEL) * TIMEOUT ))
    info "Checking originals (~${EST}s estimated)..."
    echo

    check_url() {
        local resized="$1"
        local log="$2"
        local timeout="$3"
        local original
        original=$(echo "$resized" | sed -E 's/-[0-9]+x[0-9]+(\.(jpg|jpeg|png|webp))/\1/')

        local code
        code=$(curl -s -o /dev/null -w "%{http_code}" \
            --max-time "$timeout" \
            --head \
            --retry 1 --retry-delay 1 \
            "$original" 2>/dev/null || echo "000")

        # Fall back to GET range if server blocks HEAD
        if [[ "$code" == "405" || "$code" == "403" || "$code" == "000" ]]; then
            code=$(curl -s -o /dev/null -w "%{http_code}" \
                --max-time "$timeout" \
                -H "Range: bytes=0-0" \
                "$original" 2>/dev/null || echo "000")
        fi

        if [[ "$code" == "200" || "$code" == "206" || "$code" == "301" || "$code" == "302" ]]; then
            echo "OK|$resized|$original" >> "$log"
        else
            echo "MISS|$resized|$original ($code)" >> "$log"
        fi
    }
    export -f check_url

    # Feed via cat to avoid xargs -a flag (not supported on all systems)
    cat "$URLS" | xargs -P "$PARALLEL" -I{} bash -c 'check_url "$@"' _ {} "$LOG" "$TIMEOUT" &
    WORKER_PID=$!

    while kill -0 "$WORKER_PID" 2>/dev/null; do
        DONE=$(grep -c '^\(OK\|MISS\)' "$LOG" 2>/dev/null || true)
        printf "\r    Progress: %d / %d" "$DONE" "$TOTAL" >&2
        sleep 0.3
    done
    wait "$WORKER_PID" 2>/dev/null || true

    DONE=$(grep -c '^\(OK\|MISS\)' "$LOG" 2>/dev/null || true)
    printf "\r    Progress: %d / %d\n" "$DONE" "$TOTAL" >&2

    OK_COUNT=$(grep -c '^OK'   "$LOG" || true)
    MISS_COUNT=$(grep -c '^MISS' "$LOG" || true)

    echo
    echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
    success "$OK_COUNT images ready to fix"
    [[ "$MISS_COUNT" -gt 0 ]] && warn "$MISS_COUNT originals not found — will be skipped"
    echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"

    if [[ "$MISS_COUNT" -gt 0 ]]; then
        echo
        info "Skipped URLs:"
        grep '^MISS' "$LOG" | cut -d'|' -f2
    fi

    echo
    info "To apply fixes run:  ./wp-image-fixer.sh --apply"
    echo
    exit 0
fi

# ------------------------------------------------------------------------------
# APPLY — fast single-pass Perl replacement
# ------------------------------------------------------------------------------
INPUT=$(grep '^FILE|' "$LOG" | cut -d'|' -f2)
OUTPUT="fixed-$(basename "$INPUT")"

echo
info "Applying fixes to $INPUT..."

grep '^OK' "$LOG" | awk -F'|' '{print $2 "\t" $3}' > "$MAP"
COUNT=$(wc -l < "$MAP" | tr -d ' ')

if [[ "$COUNT" -eq 0 ]]; then
    warn "No fixes to apply."
    rm -f "$MAP" "$LOG"
    exit 0
fi

REPLACED=$(perl - "$INPUT" "$MAP" "$OUTPUT" <<'PERL'
use strict;
use warnings;

my ($infile, $mapfile, $outfile) = @ARGV;

open(my $mfh, '<', $mapfile) or die "Cannot open map: $!";
my %map;
while (<$mfh>) {
    chomp;
    my ($from, $to) = split(/\t/, $_, 2);
    $map{$from} = $to if defined $from && defined $to;
}
close($mfh);

my $pattern = join('|', map { quotemeta($_) } sort { length($b) <=> length($a) } keys %map);
my $regex    = qr/$pattern/;

open(my $in,  '<', $infile)  or die "Cannot open input: $!";
open(my $out, '>', $outfile) or die "Cannot open output: $!";

my %seen;
while (my $line = <$in>) {
    $line =~ s/($regex)/do { $seen{$1} = 1; $map{$1} }/ge;
    print $out $line;
}
close($in);
close($out);

print scalar keys %seen, "\n";
PERL
)

rm -f "$MAP" "$LOG"

echo
echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
success "$REPLACED image URLs replaced — saved as $OUTPUT"
echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
echoCode language: Bash (bash)

chmod +x wp-image-fixer.shCode language: CSS (css)

./wp-image-fixer.sh

Enter WordPress XML export file: balampdplo.WordPress.2026-03-10.xml

[info]  Extracting resized image URLs...
[info]  Found 1782 resized image URLs

[info]  Checking originals (~175s estimated)...

    Progress: 1782 / 1782

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[ok]    1739 images ready to fix
[warn]  43 originals not found — will be skipped
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━


[info]  Skipped URLs:

[info]  To apply fixes run:  ./wp-image-fixer.sh --applyCode language: YAML (yaml)

./wp-image-fixer.sh --apply

[info]  Existing scan found for: balampdplo.WordPress.2026-03-08.xml

Apply fixes using cached results? (y/N): y

[info]  Applying fixes to balampdplo.WordPress.2026-03-08.xml...

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[ok]    1739 image URLs replaced — saved as fixed-balampdplo.WordPress.2026-03-08.xml
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━Code language: YAML (yaml)

npx wordpress-export-to-markdown \
  --prefix-date=true \
  --post-folders=false \
  --frontmatter-fields=title,author,date:pubDate,categories,tags,coverImage,draft,slug \
  --save-images=all \
  --date-folders=noneCode language: Nginx (nginx)

output
├── custom
│   ├── advanced_ads
│   │   └── images
│   ├── cp_popups
│   ├── foogallery
│   ├── google_maps
│   ├── surl
│   ├── tablepress_table
│   └── wpcf7_contact_form
├── pages
│   ├── _drafts
│   └── images
└── posts
    └── imagesCode language: Nginx (nginx)

[![](images/sansai-ryori-bldl01.jpg)](https://balodeplao.com/wp-content/uploads/2017/04/sansai-ryori-bldl01.jpg)Code language: Markdown (markdown)

grep -rn 'wp-content' output/posts/ --include='*.md'Code language: PHP (php)

find output/posts/ -name '*.md' -exec perl -i -pe '
s/\[!\[.*?\]\(([^)]+)\)\]\((https?:)?\/\/[^)]*wp-content\/uploads[^)]*\)/![]($1)/g;
s/\[\]\((https?:)?\/\/[^)]*wp-content\/uploads[^)]*\)//g;
' {} +Code language: Markdown (markdown)

# Find all images referenced in .md files, then check if the file actually exists
find output/posts/ -name "*.md" | while read f; do
  dir=$(dirname "$f")
  grep -oE '!\[[^]]*\]\(images/[^)]+\)' "$f" | grep -oE 'images/[^)]+' | while read img; do
    if [ ! -f "$dir/$img" ]; then
      echo "MISSING: $dir/$img (in $f)"
    fi
  done
doneCode language: Bash (bash)

MISSING: /home/mcj/test_folder/output/posts/_drafts/images/q (in /home/mcj/test_folder/output/posts/_drafts/id-4313.md)
MISSING: /home/mcj/test_folder/output/posts/_drafts/images/ir (in /home/mcj/test_folder/output/posts/_drafts/id-4313.md)
MISSING: /home/mcj/test_folder/output/posts/images/credit-card-statement-620x387.jpg (in /home/mcj/test_folder/output/posts/2015-07-10-lam-gi-khi-the-tin-dung-cua-ban-bi-hack.md)
MISSING: /home/mcj/test_folder/output/posts/images/ubersuv-voi-gia-uberBLACK.jpg (in /home/mcj/test_folder/output/posts/2015-12-07-ubersuv-nhieu-cho-hon-gia-khong-doi.md)Code language: JavaScript (javascript)

bash << 'EOF'
cd output/posts
fixed=0
for file in *.md; do
  original=$(cat "$file")
  content=$(perl -0777 -pe '
    s/\\\[caption\b[^\[]*?\\\]\s*(!\[[^\]]*\]\([^)]*\))\s*(.*?)\s*\\\[\/caption\\\]/
      my $img = $1; my $cap = $2;
      $cap =~ s|^\s+||; $cap =~ s|\s+$||;
      $img =~ m|!\[([^\]]*)\]\(([^)]*)\)|;
      my $alt = $1; my $src = $2;
      $cap ? "![$cap]($src)" : "![$alt]($src)"
    /gxse;
    s/<!--\s*wp:[^>]+-->//g;
    s/<!--\s*\/wp:[^>]+-->//g;
  ' "$file")
  content=$(printf '%s' "$content" | perl -0777 -pe 's/\n{3,}/\n\n/g')
  if [ "$content" != "$original" ]; then
    echo "Would fix: $file"
    ((fixed++))
  fi
done
echo ""
echo "Total: $fixed files would be fixed"
EOFCode language: Nginx (nginx)

bash << 'EOF'
cd output/posts
fixed=0
for file in *.md; do
  original=$(cat "$file")
  content=$(perl -0777 -pe '
    s/\\\[caption\b[^\[]*?\\\]\s*(!\[[^\]]*\]\([^)]*\))\s*(.*?)\s*\\\[\/caption\\\]/
      my $img = $1; my $cap = $2;
      $cap =~ s|^\s+||; $cap =~ s|\s+$||;
      $img =~ m|!\[([^\]]*)\]\(([^)]*)\)|;
      my $alt = $1; my $src = $2;
      $cap ? "![$cap]($src)" : "![$alt]($src)"
    /gxse;
    s/<!--\s*wp:[^>]+-->//g;
    s/<!--\s*\/wp:[^>]+-->//g;
  ' "$file")
  content=$(printf '%s' "$content" | perl -0777 -pe 's/\n{3,}/\n\n/g')
  if [ "$content" != "$original" ]; then
    printf '%s\n' "$content" > "$file"
    echo "✅ Fixed: $file"
    ((fixed++))
  fi
done
echo ""
echo "✨ Done! Fixed $fixed files."
EOFCode language: Nginx (nginx)

bash << 'EOF'
cd output/posts
{
  grep -ohE 'images/[^)]+' *.md
  grep -ohE '^coverImage:\s*"[^"]*"' *.md | grep -ohE '[^/"]+\.(jpg|jpeg|png|webp|gif)' | sed 's/^/images\//'
} | sort -u > .used_images.tmp
unused=0
while read img; do
  rel="images/$(basename "$img")"
  if ! grep -qF "$rel" .used_images.tmp; then
    echo "UNUSED: $img"
    ((unused++))
  fi
done < <(find images/ -type f)
rm .used_images.tmp
echo "---"
echo "Total unused: $unused"
EOFCode language: Bash (bash)

UNUSED: images/Nusa-Dua-BLDL-17.jpg
UNUSED: images/Flower-Dome-07.jpg
UNUSED: images/Cloud-Forest-09.jpg
UNUSED: images/Cloud-Forest-08.jpg
UNUSED: images/Flower-Dome-06.jpg
UNUSED: images/Wings-of-Time06.jpg
UNUSED: images/Nusa-Dua-BLDL-16.jpg
---
Total unused: 60Code language: YAML (yaml)

bash << 'EOF'
cd output/posts
{
  grep -ohE 'images/[^)]+' *.md
  grep -ohE '^coverImage:\s*"[^"]*"' *.md | grep -ohE '[^/"]+\.(jpg|jpeg|png|webp|gif)' | sed 's/^/images\//'
} | sort -u > .used_images.tmp
deleted=0
while read img; do
  rel="images/$(basename "$img")"
  if ! grep -qF "$rel" .used_images.tmp; then
    echo "DELETED: $img"
    rm "$img"
    ((deleted++))
  fi
done < <(find images/ -type f)
rm .used_images.tmp
echo "---"
echo "Total deleted: $deleted"
EOFCode language: Bash (bash)