Как посмотреть внешние связи и состав DOCX перед передачей ИИ в Windows

$DocxPython = 'python'
& $DocxPython -c "import sys; print(sys.version); sys.exit(0 if sys.version_info[:2] == (3,12) else 2)"
if ($LASTEXITCODE -ne 0) { throw 'Нужен уже установленный Python 3.12' }
$DocxWork = Join-Path (Get-Location) ('docx-inventory-' + [guid]::NewGuid().ToString('N'))
New-Item -ItemType Directory -Path $DocxWork -ErrorAction Stop | Out-Null
Set-Location -LiteralPath $DocxWork
Get-Location

"""Create only synthetic DOCX-shaped inventory fixtures; never opens Word."""
import base64
import sys
from pathlib import Path
from zipfile import ZipFile, ZIP_DEFLATED

P = 'http://schemas.openxmlformats.org/package/2006/relationships'
R = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/'
CT = '''<Types xmlns="http://schemas.openxmlformats.org/package/2006/content-types"><Default Extension="rels" ContentType="application/vnd.openxmlformats-package.relationships+xml"/><Default Extension="xml" ContentType="application/xml"/><Default Extension="txt" ContentType="text/plain"/><Default Extension="png" ContentType="image/png"/><Override PartName="/word/document.xml" ContentType="application/vnd.openxmlformats-officedocument.wordprocessingml.document.main+xml"/></Types>'''
DOC = '''<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main"><w:body><w:p><w:r><w:t>Synthetic inventory lesson only</w:t></w:r></w:p><w:sectPr/></w:body></w:document>'''
ROOT = f'<Relationships xmlns="{P}"><Relationship Id="rId1" Type="{R}officeDocument" Target="word/document.xml"/></Relationships>'
EMPTY = f'<Relationships xmlns="{P}"/>'
EXAMPLE = f'''<Relationships xmlns="{P}"><Relationship Id="rId1" Type="{R}hyperlink" Target="https://example.invalid/lesson" TargetMode="External"/><Relationship Id="rId2" Type="{R}image" Target="media/pixel.png"/><Relationship Id="rId3" Type="{R}package" Target="embeddings/note.txt"/><Relationship Id="rId4" Type="urn:thetimeai:lesson:unknown" Target="../customXml/item.xml"/></Relationships>'''
folder = Path('docx-fixtures')
try:
    folder.mkdir(exist_ok=False)
except FileExistsError:
    print('STOP: docx-fixtures already exists', file=sys.stderr)
    sys.exit(2)
png = base64.b64decode('iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mP8/x8AAwMCAO+jRZkAAAAASUVORK5CYII=')
for name, rels in (('clean.docx', EMPTY), ('example.docx', EXAMPLE), ('broken-xml.docx', '<Relationships>')):
    parts = {'[Content_Types].xml': CT, '_rels/.rels': ROOT, 'word/document.xml': DOC, 'word/_rels/document.xml.rels': rels}
    if name == 'example.docx':
        parts.update({'word/media/pixel.png': png, 'word/embeddings/note.txt': 'SYNTHETIC NOTE', 'customXml/item.xml': '<lesson/>'})
    with ZipFile(folder / name, 'x', compression=ZIP_DEFLATED) as archive:
        for part, data in parts.items():
            archive.writestr(part, data)
    print(folder / name)

(folder / 'not-zip.docx').write_bytes(b'SYNTHETIC NOT A ZIP')
print(folder / 'not-zip.docx')

& $DocxPython .\make_docx_fixtures.py
$DocxMakeExit = $LASTEXITCODE
if ($DocxMakeExit -ne 0) { throw 'Генератор не завершился' }
Get-ChildItem -LiteralPath .\docx-fixtures | Select-Object Name, Length

"""Bounded read-only DOCX ZIP/relationship inventory. No target fetching."""
import argparse
import hashlib
import io
import json
import posixpath
import re
import sys
import zipfile
import zlib
from pathlib import Path
from urllib.parse import urlsplit
from xml.etree import ElementTree as ET

MAX_FILE = 10 * 1024 * 1024
MAX_ENTRIES = 1000
MAX_PART = 10 * 1024 * 1024
MAX_TOTAL = 40 * 1024 * 1024
MAX_XML = 1024 * 1024
MAX_XML_TOTAL = 4 * 1024 * 1024
NS = 'http://schemas.openxmlformats.org/package/2006/relationships'
RT = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/'
KNOWN = {RT + x: x for x in ('hyperlink', 'image', 'oleObject', 'package')}

class Stop(Exception):
    pass

def require(ok, message):
    if not ok:
        raise Stop(message)

def member_ok(name):
    require(0 < len(name) <= 512 and not any(ord(c) < 32 for c in name), 'member name length/control')
    require(not name.startswith('/') and '\\' not in name and ':' not in name, 'member path form')
    require(all(p not in ('', '.', '..') for p in name.split('/')), 'member path components')

def xml_root(data):
    require(len(data) <= MAX_XML, 'XML size limit')
    require(b'\x00' not in data, 'only UTF-8 XML supported')
    text = data.decode('utf-8-sig')
    enc = re.match(r'''\s*<\?xml\b[^>]*\bencoding\s*=\s*['"]([^'"]+)''', text, re.I)
    require(not enc or enc[1].lower() in ('utf-8', 'utf8'), 'XML encoding')
    require(not re.search(r'<!\s*(DOCTYPE|ENTITY)\b', text, re.I), 'DTD/ENTITY unsupported')
    root = ET.fromstring(text)
    require(sum(1 for _ in root.iter()) <= 5000, 'XML node limit')
    return root

def source_part(name):
    if name == '_rels/.rels':
        return ''
    folder, filename = posixpath.split(name)
    require(posixpath.basename(folder) == '_rels' and filename.endswith('.rels'), 'rels path form')
    owner = posixpath.join(posixpath.dirname(folder), filename[:-5])
    member_ok(owner)
    return owner

def internal_target(owner, target):
    require(not any(c in target for c in ('\\', '%', '?', '#')), 'complex internal target unsupported')
    uri = urlsplit(target)
    require(not uri.scheme and not uri.netloc, 'internal target has scheme/host')
    components = [] if target.startswith('/') else posixpath.dirname(owner).split('/')
    components = [p for p in components if p]
    for p in target.split('/'):
        if p in ('', '.'):
            continue
        if p == '..':
            require(bool(components), 'target leaves package root')
            components.pop()
        else:
            components.append(p)
    result = '/'.join(components)
    member_ok(result)
    return result

def inspect(path):
    require(path.is_file() and not path.is_symlink(), 'regular file required')
    require(path.stat().st_size <= MAX_FILE, 'archive exceeds 10 MiB')
    with path.open('rb') as stream:
        data = stream.read(MAX_FILE + 1)
    require(len(data) <= MAX_FILE, 'archive exceeds 10 MiB')
    report = {'sha256': hashlib.sha256(data).hexdigest(), 'parts': [], 'relationships': []}
    with zipfile.ZipFile(io.BytesIO(data), 'r') as z:
        infos = z.infolist()
        require(len(infos) <= MAX_ENTRIES, 'entry count exceeds 1000')
        names = set()
        total = 0
        for info in infos:
            require(not info.is_dir(), 'directory entries unsupported')
            member_ok(info.orig_filename)
            member_ok(info.filename)
            require(info.filename not in names, 'duplicate ZIP name')
            names.add(info.filename)
            require(not info.flag_bits & 1, 'encrypted ZIP unsupported')
            require((info.external_attr >> 16) & 0o170000 != 0o120000, 'ZIP symlink unsupported')
            require(info.compress_type in (zipfile.ZIP_STORED, zipfile.ZIP_DEFLATED), 'compression unsupported')
            require(info.file_size <= MAX_PART, 'part exceeds 10 MiB')
            require(not info.file_size or (info.compress_size > 0 and info.file_size <= info.compress_size * 100), 'compression ratio exceeds 100:1')
            total += info.file_size
            require(total <= MAX_TOTAL, 'declared total exceeds 40 MiB')
            report['parts'].append({'name': info.filename, 'bytes': info.file_size})
        require({'[Content_Types].xml', '_rels/.rels', 'word/document.xml'} <= names, 'expected DOCX parts absent')
        xml_bytes = 0
        for name in sorted(n for n in names if n.endswith('.rels')):
            owner = source_part(name)
            require(not owner or owner in names, 'rels source part absent')
            info = z.getinfo(name)
            xml_bytes += info.file_size
            require(info.file_size <= MAX_XML and xml_bytes <= MAX_XML_TOTAL, 'relationship XML budget exceeded')
            with z.open(info) as stream:
                payload = stream.read(MAX_XML + 1)
            root = xml_root(payload)
            require(root.tag == '{' + NS + '}Relationships', 'unexpected relationship namespace/root')
            ids = set()
            for rel in root:
                require(rel.tag == '{' + NS + '}Relationship' and not list(rel), 'unexpected relationship element')
                require(set(rel.attrib) <= {'Id', 'Type', 'Target', 'TargetMode'}, 'unexpected relationship attributes')
                rid, typ, target = (rel.get(k, '') for k in ('Id', 'Type', 'Target'))
                require(all(0 < len(v) <= 2048 for v in (rid, typ, target)), 'relationship field length')
                require(rid not in ids, 'duplicate relationship Id')
                ids.add(rid)
                mode = rel.get('TargetMode', 'Internal')
                require(mode in ('Internal', 'External'), 'unexpected TargetMode')
                row = {'source': owner or '/', 'id': rid, 'kind': KNOWN.get(typ, 'OTHER'), 'type': typ, 'mode': mode, 'target': target}
                if mode == 'Internal':
                    resolved = internal_target(owner, target)
                    row.update(resolved=resolved, present=resolved in names)
                report['relationships'].append(row)
                require(len(report['relationships']) <= 2000, 'relationship count exceeds 2000')
    report['parts'].sort(key=lambda x: x['name'])
    report['note'] = 'Inventory only; no document/attachment content scan; no target requests.'
    return report

def main():
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument('docx', type=Path)
    args = parser.parse_args()
    try:
        result = inspect(args.docx)
    except (Stop, OSError, ValueError, ET.ParseError, zipfile.BadZipFile, RuntimeError, NotImplementedError, EOFError, zlib.error) as error:
        print('STOP: ' + ascii(str(error)), file=sys.stderr)
        return 2
    # ASCII escapes prevent terminal control characters from becoming commands/links.
    print(json.dumps(result, ensure_ascii=True, indent=2))
    return 0

if __name__ == '__main__':
    sys.exit(main())

$DocxInput = '.\docx-fixtures\example.docx'
$DocxBefore = (Get-FileHash -LiteralPath $DocxInput -Algorithm SHA256).Hash
& $DocxPython .\inspect_docx.py $DocxInput
$DocxInspectExit = $LASTEXITCODE
$DocxAfter = (Get-FileHash -LiteralPath $DocxInput -Algorithm SHA256).Hash
[pscustomobject]@{ ExitCode = $DocxInspectExit; InputUnchanged = ($DocxBefore -eq $DocxAfter) }
if ($DocxInspectExit -ne 0 -or $DocxBefore -ne $DocxAfter) { throw 'Инвентаризация не подтверждена' }

& $DocxPython .\inspect_docx.py .\docx-fixtures\clean.docx
$DocxCleanExit = $LASTEXITCODE
& $DocxPython .\inspect_docx.py .\docx-fixtures\broken-xml.docx
$DocxXmlExit = $LASTEXITCODE
& $DocxPython .\inspect_docx.py .\docx-fixtures\not-zip.docx
$DocxZipExit = $LASTEXITCODE
[pscustomobject]@{ Clean = $DocxCleanExit; BrokenXml = $DocxXmlExit; NotZip = $DocxZipExit }