#!/usr/bin/env bash
#
# migrate_content.sh
#
# Migrates content from the old jon.recoil.org-src site to the mono repo's
# site/ directory. Copies .mld files, static assets, and applies format
# transformations for the new odoc-based build system.
#
# Usage: ./scripts/migrate_content.sh
#   Run from the mono repo root.

set -euo pipefail

# Resolve paths
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
MONO_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
OLD_SITE="/home/jons-agent/workspace/jon.recoil.org-src"
NEW_SITE="$MONO_ROOT/site"

if [ ! -d "$OLD_SITE" ]; then
  echo "ERROR: Old site directory not found at $OLD_SITE"
  exit 1
fi

echo "=== Content Migration Script ==="
echo "  Source: $OLD_SITE"
echo "  Destination: $NEW_SITE"
echo ""

# Track files needing manual review
REVIEW_FILES_OCAMLTOP=()
REVIEW_FILES_DEFERRED=()
REVIEW_FILES_PROMPT=()

# ============================================================
# Step 1: Copy content directories
# ============================================================

echo "--- Step 1: Copying content ---"

# 1a. Copy blog .mld files, preserving directory structure
echo "  Copying blog/ ..."
if [ -d "$OLD_SITE/blog" ]; then
  find "$OLD_SITE/blog" -name "*.mld" -type f | while read -r src; do
    rel="${src#$OLD_SITE/}"
    dest="$NEW_SITE/$rel"
    mkdir -p "$(dirname "$dest")"
    cp "$src" "$dest"
  done
  blog_count=$(find "$NEW_SITE/blog" -name "*.mld" -type f | wc -l)
  echo "    Copied $blog_count .mld files"
else
  echo "    WARNING: $OLD_SITE/blog does not exist"
fi

# 1b. Copy notebooks .mld files, preserving directory structure
echo "  Copying notebooks/ ..."
if [ -d "$OLD_SITE/notebooks" ]; then
  find "$OLD_SITE/notebooks" -name "*.mld" -type f | while read -r src; do
    rel="${src#$OLD_SITE/}"
    dest="$NEW_SITE/$rel"
    mkdir -p "$(dirname "$dest")"
    cp "$src" "$dest"
  done
  nb_count=$(find "$NEW_SITE/notebooks" -name "*.mld" -type f | wc -l)
  echo "    Copied $nb_count .mld files"
else
  echo "    WARNING: $OLD_SITE/notebooks does not exist"
fi

# 1c. Copy static assets
echo "  Copying static/ ..."
if [ -d "$OLD_SITE/static" ]; then
  cp -r "$OLD_SITE/static/." "$NEW_SITE/static/"
  static_count=$(find "$NEW_SITE/static" -type f | wc -l)
  echo "    Copied $static_count static files"
else
  echo "    WARNING: $OLD_SITE/static does not exist"
fi

# 1d. Copy notes/ if it exists
echo "  Copying notes/ ..."
if [ -d "$OLD_SITE/notes" ] && [ -n "$(ls -A "$OLD_SITE/notes" 2>/dev/null)" ]; then
  mkdir -p "$NEW_SITE/notes"
  find "$OLD_SITE/notes" -name "*.mld" -type f | while read -r src; do
    rel="${src#$OLD_SITE/}"
    dest="$NEW_SITE/$rel"
    mkdir -p "$(dirname "$dest")"
    cp "$src" "$dest"
  done
  notes_count=$(find "$NEW_SITE/notes" -name "*.mld" -type f 2>/dev/null | wc -l)
  echo "    Copied $notes_count .mld files"
else
  echo "    Skipped (directory empty or does not exist)"
fi

# 1e. Copy drafts/ if it exists
echo "  Copying drafts/ ..."
if [ -d "$OLD_SITE/drafts" ] && [ -n "$(ls -A "$OLD_SITE/drafts" 2>/dev/null)" ]; then
  mkdir -p "$NEW_SITE/drafts"
  find "$OLD_SITE/drafts" -name "*.mld" -type f | while read -r src; do
    rel="${src#$OLD_SITE/}"
    dest="$NEW_SITE/$rel"
    mkdir -p "$(dirname "$dest")"
    cp "$src" "$dest"
  done
  drafts_count=$(find "$NEW_SITE/drafts" -name "*.mld" -type f 2>/dev/null | wc -l)
  echo "    Copied $drafts_count .mld files"
else
  echo "    Skipped (directory empty or does not exist)"
fi

# 1f. Copy reference/ .mld files only (not build output)
echo "  Copying reference/ ..."
if [ -d "$OLD_SITE/reference" ]; then
  mkdir -p "$NEW_SITE/reference"
  find "$OLD_SITE/reference" -name "*.mld" -type f | while read -r src; do
    rel="${src#$OLD_SITE/}"
    dest="$NEW_SITE/$rel"
    mkdir -p "$(dirname "$dest")"
    cp "$src" "$dest"
  done
  ref_count=$(find "$NEW_SITE/reference" -name "*.mld" -type f 2>/dev/null | wc -l)
  echo "    Copied $ref_count .mld files"
else
  echo "    Skipped (directory does not exist)"
fi

echo ""

# ============================================================
# Step 2: Apply transformations to all .mld files
# ============================================================

echo "--- Step 2: Applying transformations ---"

transform_count=0
libs_count=0
switch_count=0
notanotebook_count=0
autorun_count=0
ocamltop_plain_count=0

find "$NEW_SITE" -name "*.mld" -type f | sort | while read -r file; do
  modified=false

  # 2a. Replace {@ocamltop autorun[ with {@ocaml run-on=load[
  #     This handles variants like: {@ocamltop autorun hidden[
  if grep -q '{@ocamltop autorun' "$file" 2>/dev/null; then
    # Replace {@ocamltop autorun[ with {@ocaml run-on=load[
    # Also handle {@ocamltop autorun hidden[ -> {@ocaml run-on=load hidden[
    sed -i 's/{@ocamltop autorun\b/{@ocaml run-on=load/g' "$file"
    modified=true
    autorun_count=$((autorun_count + 1))
  fi

  # Also handle {x@ocamltop autorun[ -> {x@ocaml run-on=load[
  if grep -q '{x@ocamltop autorun' "$file" 2>/dev/null; then
    sed -i 's/{x@ocamltop autorun\b/{x@ocaml run-on=load/g' "$file"
    modified=true
  fi

  # Also handle {@ocaml autorun -> {@ocaml run-on=load (non-ocamltop variant)
  if grep -q '{@ocaml autorun' "$file" 2>/dev/null; then
    sed -i 's/{@ocaml autorun\b/{@ocaml run-on=load/g' "$file"
    modified=true
  fi

  # 2b. Replace {@ocamltop skip[ with {@ocaml skip[
  if grep -q '{@ocamltop skip' "$file" 2>/dev/null; then
    sed -i 's/{@ocamltop skip\b/{@ocaml skip/g' "$file"
    modified=true
  fi

  # 2c. Replace remaining {@ocamltop[ with {@ocaml[
  if grep -q '{@ocamltop\[' "$file" 2>/dev/null; then
    sed -i 's/{@ocamltop\[/{@ocaml[/g' "$file"
    modified=true
    ocamltop_plain_count=$((ocamltop_plain_count + 1))
  fi

  # Also handle {x@ocamltop[ -> {x@ocaml[
  if grep -q '{x@ocamltop\[' "$file" 2>/dev/null; then
    sed -i 's/{x@ocamltop\[/{x@ocaml[/g' "$file"
    modified=true
  fi

  # 2d. Transform @libs lines: space-separated to comma-separated @x-ocaml.requires
  #     e.g. @libs pkg1 pkg2 pkg3 -> @x-ocaml.requires pkg1,pkg2,pkg3
  if grep -q '^@libs ' "$file" 2>/dev/null; then
    sed -i '/^@libs /{s/^@libs //; s/ /,/g; s/^/@x-ocaml.requires /}' "$file"
    modified=true
    libs_count=$((libs_count + 1))
  fi

  # 2e. Remove @switch oxcaml lines entirely
  if grep -q '^@switch ' "$file" 2>/dev/null; then
    sed -i '/^@switch /d' "$file"
    modified=true
    switch_count=$((switch_count + 1))
  fi

  # 2f. Remove @notanotebook lines entirely
  if grep -q '^@notanotebook' "$file" 2>/dev/null; then
    sed -i '/^@notanotebook$/d' "$file"
    modified=true
    notanotebook_count=$((notanotebook_count + 1))
  fi

  if [ "$modified" = true ]; then
    transform_count=$((transform_count + 1))
  fi
done

echo "  Files transformed: see summary below"
echo ""

# ============================================================
# Step 3: Flag files for manual review
# ============================================================

echo "--- Step 3: Checking for files needing manual review ---"
echo ""

# 3a. Check for remaining ocamltop (incomplete transformation)
echo "  [CHECK] Files still containing 'ocamltop' (incomplete transformation):"
remaining_ocamltop=0
while IFS= read -r file; do
  rel="${file#$NEW_SITE/}"
  echo "    - $rel"
  remaining_ocamltop=$((remaining_ocamltop + 1))
done < <(grep -rl 'ocamltop' "$NEW_SITE" --include="*.mld" 2>/dev/null || true)
if [ "$remaining_ocamltop" -eq 0 ]; then
  echo "    (none -- all ocamltop references transformed)"
fi
echo ""

# 3b. Check for {x@ocaml[ (deferred blocks needing manual handling)
echo "  [CHECK] Files containing '{x@ocaml[' (deferred blocks, need manual review):"
deferred_count=0
while IFS= read -r file; do
  rel="${file#$NEW_SITE/}"
  echo "    - $rel"
  deferred_count=$((deferred_count + 1))
done < <(grep -rl '{x@ocaml\[' "$NEW_SITE" --include="*.mld" 2>/dev/null || true)
if [ "$deferred_count" -eq 0 ]; then
  echo "    (none)"
fi
echo ""

# 3c. Check for lines starting with # inside code blocks (old prompt prefixes)
echo "  [CHECK] Files with '# ' prompt prefixes inside code blocks (need manual review):"
prompt_count=0
while IFS= read -r file; do
  rel="${file#$NEW_SITE/}"
  echo "    - $rel"
  prompt_count=$((prompt_count + 1))
done < <(grep -rl '^# ' "$NEW_SITE" --include="*.mld" 2>/dev/null || true)
if [ "$prompt_count" -eq 0 ]; then
  echo "    (none)"
fi
echo ""

# ============================================================
# Summary
# ============================================================

echo "=== Migration Summary ==="
total_mld=$(find "$NEW_SITE" -name "*.mld" -type f | wc -l)
total_static=$(find "$NEW_SITE/static" -type f 2>/dev/null | wc -l)
echo "  Total .mld files migrated: $total_mld"
echo "  Total static files copied: $total_static"
echo ""
echo "  Transformations applied:"
echo "    @ocamltop autorun -> @ocaml run-on=load"
echo "    @ocamltop[ -> @ocaml["
echo "    @libs -> @x-ocaml.requires (comma-separated)"
echo "    @switch lines removed"
echo "    @notanotebook lines removed"
echo ""
echo "  Files needing manual review: $((remaining_ocamltop + deferred_count + prompt_count))"
echo "    - Remaining ocamltop: $remaining_ocamltop"
echo "    - Deferred blocks ({x@ocaml[): $deferred_count"
echo "    - Prompt prefixes (# ): $prompt_count"
echo ""
echo "=== Done ==="