#!/usr/bin/env python3
"""
EXT4 Filesystem Libraries
"""
import struct, os, sys, stat
from pathlib import Path

BLOCK=4096

def read_at(f, offset, size):
    f.seek(offset)
    return f.read(size)

def parse_superblock(data):
    sb = {}
    sb['inodes_count']     = struct.unpack_from('<I', data,  0)[0]
    sb['blocks_count']     = struct.unpack_from('<I', data,  4)[0]
    sb['blocks_per_group'] = struct.unpack_from('<I', data, 32)[0]
    sb['inodes_per_group'] = struct.unpack_from('<I', data, 40)[0]
    sb['inode_size']       = struct.unpack_from('<H', data, 88)[0]
    sb['magic']            = struct.unpack_from('<H', data, 56)[0]
    sb['desc_size']        = struct.unpack_from('<H', data, 254)[0] or 32
    return sb

def parse_gdt_entry(gdt_data, offset, desc_size):
    lo = struct.unpack_from('<I', gdt_data, offset +  8)[0]
    if desc_size >= 64:
        hi = struct.unpack_from('<I', gdt_data, offset + 40)[0]
        return lo | (hi << 32)
    return lo

def parse_extent_tree(data, inode_offset):
    base = inode_offset + 40
    magic, entries, _, depth = struct.unpack_from('<HHHH', data, base)
    if magic != 0xF30A:
        return []
    extents = []
    if depth == 0:
        for i in range(min(entries, 4)):
            o = base + 12 + i * 12
            if o + 12 > len(data): break
            l_block  = struct.unpack_from('<I', data, o    )[0]
            ee_len   = struct.unpack_from('<H', data, o + 4)[0]
            start_hi = struct.unpack_from('<H', data, o + 6)[0]
            start_lo = struct.unpack_from('<I', data, o + 8)[0]
            phys = (start_hi << 32) | start_lo
            if phys > 0:
                extents.append((l_block, phys, ee_len & 0x7FFF))
    else:
        # Depth > 0: extent index node - follow first child
        # (handles large dirs gracefully)
        o = base + 12
        ei_leaf_lo = struct.unpack_from('<I', data, o + 4)[0]
        ei_leaf_hi = struct.unpack_from('<H', data, o + 8)[0]
        extents.append((0, (ei_leaf_hi << 32) | ei_leaf_lo, 1))
    return extents

def read_extent_tree_blocks(f, data, inode_offset):
    """
    Fully recursive extent tree walker.
    Returns sorted list of (logical_block, phys_block) pairs.
    """
    base = inode_offset + 40
    magic, entries, _, depth = struct.unpack_from('<HHHH', data, base)
    if magic != 0xF30A:
        return []
    return _walk_extent_node(f, data, base, depth)

def _walk_extent_node(f, data, base, depth):
    magic, entries, _, _ = struct.unpack_from('<HHHH', data, base)
    if magic != 0xF30A:
        return []

    result = []
    if depth == 0:
        # Leaf node - actual extents
        for i in range(entries):
            o = base + 12 + i * 12
            l_block  = struct.unpack_from('<I', data, o    )[0]
            ee_len   = struct.unpack_from('<H', data, o + 4)[0]
            start_hi = struct.unpack_from('<H', data, o + 6)[0]
            start_lo = struct.unpack_from('<I', data, o + 8)[0]
            phys = (start_hi << 32) | start_lo
            if phys > 0:
                for b in range(ee_len & 0x7FFF):
                    result.append((l_block + b, phys + b))
    else:
        # Index node - recurse into each child
        for i in range(entries):
            o = base + 12 + i * 12
            leaf_lo = struct.unpack_from('<I', data, o + 4)[0]
            leaf_hi = struct.unpack_from('<H', data, o + 8)[0]
            leaf_block = (leaf_hi << 32) | leaf_lo
            try:
                child_data = read_at(f, leaf_block * BLOCK, BLOCK)
                result.extend(_walk_extent_node(f, child_data, 0, depth - 1))
            except OSError:
                pass
    return result

def read_inode(f, sb, gdt_data, inum):
    """Return raw inode block data and offset within it."""
    grp       = (inum - 1) // sb['inodes_per_group']
    local_idx = (inum - 1) %  sb['inodes_per_group']
    tbl_block = parse_gdt_entry(gdt_data, grp * sb['desc_size'], sb['desc_size'])
    byte_off  = local_idx * sb['inode_size']
    blk_off   = byte_off // BLOCK
    slot      = byte_off %  BLOCK
    data      = read_at(f, (tbl_block + blk_off) * BLOCK, BLOCK)
    return data, slot

def classify_inode(idata, slot):
    """
    Returns 'deleted', 'orphaned', or 'active' based on inode fields.
    """
    mode        = struct.unpack_from('<H', idata, slot +  0)[0]
    links_count = struct.unpack_from('<H', idata, slot + 26)[0]
    dtime       = struct.unpack_from('<I', idata, slot + 20)[0]
    flags       = struct.unpack_from('<I', idata, slot + 32)[0]

    if dtime != 0 and links_count == 0:
        return 'deleted'
    if dtime != 0 and links_count > 0:
        # Inconsistent - probably corruption
        return 'corrupt'
    if dtime == 0 and links_count == 0:
        # Unallocated inode - should not appear in dir entries
        return 'unallocated'
    return 'active'  # dtime=0, links_count>0 - normal live inode

def read_dir_entries(f, sb, gdt_data, inum):
    """Return dict of name -> (child_inum, ftype)."""
    idata, slot = read_inode(f, sb, gdt_data, inum)
    entries = {}
    for logical, phys in sorted(read_extent_tree_blocks(f, idata, slot)):
        try:
            bdata  = read_at(f, phys * BLOCK, BLOCK)
            offset = 0
            while offset < BLOCK - 8:
                e_ino, rec_len, name_len, ftype = \
                    struct.unpack_from('<IHBB', bdata, offset)
                if rec_len < 8 or offset + rec_len > BLOCK:
                    break
                if e_ino != 0 and name_len > 0:
                    name = bdata[offset+8:offset+8+name_len]\
                               .decode('utf-8', errors='replace')
                    entries[name] = (e_ino, ftype)
                offset += rec_len
        except OSError:
            pass
    return entries

def dump_file(f, sb, gdt_data, inum, dest_path):
    """Extract a regular file by inode to dest_path."""
    try:
        idata, slot = read_inode(f, sb, gdt_data, inum)
        size_lo = struct.unpack_from('<I', idata, slot +  4)[0]
        size_hi = struct.unpack_from('<I', idata, slot + 108)[0]
        size    = size_lo | (size_hi << 32)
        flags   = struct.unpack_from('<I', idata, slot + 32)[0]

        if flags & 0x10000000:
            # Inline data - stored in inode body after extent header
            inline = idata[slot+40:slot+40+size]
            with open(dest_path, 'wb') as out:
                out.write(inline)
            return True

        blocks = sorted(read_extent_tree_blocks(f, idata, slot))
        written = 0
        with open(dest_path, 'wb') as out:
            # Handle sparse files - fill holes with zeros
            for logical, phys in blocks:
                hole = logical * BLOCK
                if hole > written:
                    out.seek(hole)
                    written = hole
                remaining = size - written
                if remaining <= 0:
                    break
                chunk = read_at(f, phys * BLOCK, BLOCK)
                out.write(chunk[:min(BLOCK, remaining)])
                written += min(BLOCK, remaining)
            out.truncate(size)
        return True
    except OSError:
        return False

def dump_symlink(f, sb, gdt_data, inum, dest_path):
    try:
        idata, slot = read_inode(f, sb, gdt_data, inum)
        size = struct.unpack_from('<I', idata, slot + 4)[0]
        if size <= 60:
            target = idata[slot+40:slot+40+size].decode('utf-8', errors='replace')
        else:
            extents = read_extent_tree_blocks(f, idata, slot)
            if not extents:
                return False
            bdata  = read_at(f, extents[0][1] * BLOCK, BLOCK)
            target = bdata[:size].decode('utf-8', errors='replace')

        # Strip null terminator, control characters, and anything after first null
        target = target.split('\x00')[0].strip()

        if not target:
            print(f"  WARN empty symlink target for {dest_path}", file=sys.stderr)
            return False

        # Validate target looks like a path
        if any(ord(c) < 32 for c in target):
            print(f"  WARN control chars in symlink target {dest_path!r} -> {target!r}",
                  file=sys.stderr)
            return False

        if os.path.lexists(dest_path):
            return True  # already exists from a previous run

        os.symlink(target, dest_path)
        return True
    except (OSError, IndexError) as e:
        print(f"  WARN symlink {dest_path}: {e}", file=sys.stderr)
        return False

# ── recursive dumper ─────────────────────────────────────────────────────────

FTYPE_REG  = 1
FTYPE_DIR  = 2
FTYPE_SYM  = 7

def dump_tree(f, sb, gdt_data, inum, dest_dir, depth=0, visited=None):
    if visited is None:
        visited = set()
    if inum in visited:
        return
    visited.add(inum)

    try:
        entries = read_dir_entries(f, sb, gdt_data, inum)
    except Exception:
        return

    os.makedirs(dest_dir, exist_ok=True)

    for name, (child_inum, ftype) in entries.items():
        if name in ('.', '..'):
            continue
        safe_name = name.replace('/', '_').replace('\x00', '')
        dest = os.path.join(dest_dir, safe_name)

        try:
            # If ftype unknown, derive from inode mode
            if ftype == 0:
                idata, slot = read_inode(f, sb, gdt_data, child_inum)
                mode = struct.unpack_from('<H', idata, slot)[0]
                itype = mode & 0xF000
                if   itype == 0x4000: ftype = FTYPE_DIR
                elif itype == 0x8000: ftype = FTYPE_REG
                elif itype == 0xA000: ftype = FTYPE_SYM

            if ftype == FTYPE_DIR:
                dump_tree(f, sb, gdt_data, child_inum, dest,
                          depth+1, visited)
            elif ftype == FTYPE_REG:
                dump_file(f, sb, gdt_data, child_inum, dest)
            elif ftype == FTYPE_SYM:
                dump_symlink(f, sb, gdt_data, child_inum, dest)
            # ftype still 0 after mode check = special file, skip

        except Exception as e:
            print(f"  WARN: {dest}: {e}", file=sys.stderr)