Guide auto fixing links implemented

2026-08-05 13:01:30 -05:00
parent aed6a7c24d
commit c1659943fa
4 changed files with 248 additions and 2 deletions
@@ -1134,6 +1134,77 @@ class UGCCategoryTests(UGCMediaTestCase, UGCGatedTestCase):
        resp = self.client.get(reverse('library:guide_doc', args=['guide', version.pk, vf.file.uuid]))
        self.assertEqual(resp.status_code, 404)

+    def test_guide_rewrites_links_to_separately_uploaded_files(self):
+        self.project = Project.objects.create(
+            slug='guide', title='Guide', category='guide', owner=self.alice,
+        )
+        self._upload_version(
+            'README.md', b'# Title\n\n![logo](images/logo.png)\n', content_type='text/markdown',
+        )
+        self._upload_version('images/logo.png', b'PNGDATA', content_type='image/png')
+        self._upload_version_form()
+        version = self.project.versions.first()
+        md_vf = version.files.get(kind='markdown')
+        self.gate()
+        resp = self.client.get(reverse('library:guide_doc', args=['guide', version.pk, md_vf.file.uuid]))
+        self.assertEqual(resp.status_code, 200)
+        self.assertContains(resp, '<img src="/api/files/', count=1)
+        self.assertNotContains(resp, 'images/logo.png')
+
+    def test_guide_zip_docs_listed_and_links_rewritten(self):
+        self.project = Project.objects.create(
+            slug='guide', title='Guide', category='guide', owner=self.alice,
+        )
+        self._upload_version('guide.zip', _zip_bytes({
+            'README.md': '# Read\n\n![pic](images/a.png)\n\nNext: [part two](part2.md)\n',
+            'images/a.png': b'PNG2',
+            'part2.md': '# Two\n',
+        }))
+        self._upload_version_form()
+        version = self.project.versions.first()
+        self.gate()
+        # Both .md members appear in the guide switcher.
+        resp = self.client.get(reverse('library:project_detail', args=['guide']))
+        self.assertContains(resp, 'README.md')
+        self.assertContains(resp, 'part2.md')
+        # Zip doc renders with image + cross-doc link rewritten.
+        resp = self.client.get(reverse('library:guide_zip_doc', args=['guide', version.pk, 'README.md']))
+        self.assertEqual(resp.status_code, 200)
+        self.assertContains(resp, '<img src="/packs/guide/versions/%d/asset/images/a.png"' % version.pk, count=1)
+        self.assertContains(resp, 'href="/packs/guide/guide/zip/%d/part2.md"' % version.pk, count=1)
+
+    def test_guide_zip_doc_rejects_non_guide_or_missing(self):
+        self.project = Project.objects.create(
+            slug='guide', title='Guide', category='guide', owner=self.alice,
+        )
+        self._upload_version('guide.zip', _zip_bytes({'README.md': '# Read\n'}))
+        self._upload_version_form()
+        version = self.project.versions.first()
+        self.gate()
+        self.assertEqual(
+            self.client.get(reverse('library:guide_zip_doc', args=['guide', version.pk, 'missing.md'])).status_code,
+            404,
+        )
+        self.assertEqual(
+            self.client.get(reverse('library:guide_zip_doc', args=['guide', version.pk, 'data.json'])).status_code,
+            404,
+        )
+
+    def test_guide_links_not_rewritten_outside_guide_category(self):
+        self.project = Project.objects.create(
+            slug='pack', title='Pack', category='non_pack', owner=self.alice,
+        )
+        self._upload_version(
+            'notes.md', b'# Notes\n\n![x](images/x.png)\n', content_type='text/markdown',
+        )
+        self._upload_version_form()
+        version = self.project.versions.first()
+        md_vf = version.files.get(kind='markdown')
+        self.gate()
+        resp = self.client.get(reverse('library:guide_doc', args=['pack', version.pk, md_vf.file.uuid]))
+        self.assertEqual(resp.status_code, 200)
+        self.assertContains(resp, 'src="images/x.png"')
+
    def test_regular_category_has_no_extra_metadata(self):
        self.project = Project.objects.create(
            slug='mod', title='Mod', category='mod', owner=self.alice,
@@ -16,6 +16,7 @@ urlpatterns = [
    path('packs/<slug:slug>/versions/<int:version_id>/files/<uuid:file_uuid>/download/', views.version_file_download, name='version_file_download'),
    path('packs/<slug:slug>/versions/<int:version_id>/asset/<path:asset_path>', views.pack_asset, name='pack_asset'),
    path('packs/<slug:slug>/guide/<int:version_id>/<uuid:file_uuid>/', views.guide_doc, name='guide_doc'),
+    path('packs/<slug:slug>/guide/zip/<int:version_id>/<path:asset_path>', views.guide_zip_doc, name='guide_zip_doc'),
    path('packs/<slug:slug>/gallery/upload/', views.asset_upload, name='asset_upload'),
    path('packs/<slug:slug>/gallery/<int:asset_id>/thumb/', views.asset_thumbnail, name='asset_thumbnail'),
    path('packs/<slug:slug>/gallery/<int:asset_id>/delete/', views.asset_delete, name='asset_delete'),
@@ -347,7 +347,38 @@ def project_detail(request, slug):
    anim_stats = None
    latest = versions.first()
    if project.category == 'guide' and latest is not None:
-        guide_docs = [vf for vf in latest.files.all() if vf.is_markdown]
+        guide_docs = []
+        seen = set()
+        for vf in latest.files.all():
+            if vf.is_markdown:
+                guide_docs.append({
+                    'filename': vf.filename,
+                    'url': reverse('library:guide_doc', args=[project.slug, latest.pk, vf.file.uuid]),
+                })
+                seen.add((vf.filename or '').lower())
+        release = latest.files.filter(kind='release').first()
+        if release is not None and (release.filename or '').lower().endswith('.zip'):
+            try:
+                with zipfile.ZipFile(_stored_path(release.file)) as zf:
+                    members = [
+                        m for m in zf.namelist()
+                        if not m.endswith('/') and m.lower().endswith('.md')
+                    ]
+            except (zipfile.BadZipFile, OSError, FileNotFoundError):
+                members = []
+            members.sort(key=lambda m: (
+                not _guide_norm_path(m).rsplit('/', 1)[-1].startswith(('readme', 'index')),
+                m,
+            ))
+            for m in members:
+                bname = _guide_norm_path(m).rsplit('/', 1)[-1]
+                if bname in seen:
+                    continue
+                seen.add(bname)
+                guide_docs.append({
+                    'filename': bname,
+                    'url': reverse('library:guide_zip_doc', args=[project.slug, latest.pk, m]),
+                })
    if latest is not None:
        animation_manifest = latest.animation_manifest or None
        models_manifest = latest.models_manifest or None
@@ -487,6 +518,117 @@ def api_packs_latest(request, namespace, pack_id):
    })


+_GUIDE_LINK_RE = re.compile(r'(!?\[[^\]]*\]\()([^\s)]+)([^)]*)(\))')
+_GUIDE_REF_RE = re.compile(r'^(\[[^\]]+\]:\s*)(\S+)\s*$', re.MULTILINE)
+_GUIDE_LINK_SKIP = ('http://', 'https://', '//', 'data:', 'mailto:', '#', 'tel:', 'javascript:')
+
+
+def _guide_norm_path(path):
+    """Normalize a link target / stored path for matching (lowercase, no ./ or /)."""
+    p = (path or '').strip().replace('\\', '/')
+    while p.startswith('./'):
+        p = p[2:]
+    p = p.lstrip('/')
+    return p.lower()
+
+
+def _guide_asset_maps(version, base_dir=None, source='zip'):
+    """Build normalized path/basename -> served URL maps for a guide version.
+
+    ``full_map`` keys are normalized full paths, ``base_map`` keys are
+    normalized basenames. First match wins; priority is: members inside the
+    current doc's own zip directory -> standalone uploaded files -> other zip
+    members. Markdown members map to the rendered-doc endpoints, everything
+    else to the raw file/asset endpoints."""
+    slug = version.project.slug
+    full = {}
+    base = {}
+
+    def _add(path, url):
+        norm = _guide_norm_path(path)
+        full.setdefault(norm, url)
+        bname = norm.rsplit('/', 1)[-1]
+        if bname and bname not in base:
+            base[bname] = url
+
+    release = version.files.filter(kind='release').first()
+    zip_members = []
+    if release is not None and (release.filename or '').lower().endswith('.zip'):
+        try:
+            with zipfile.ZipFile(_stored_path(release.file)) as zf:
+                zip_members = [m for m in zf.namelist() if not m.endswith('/')]
+        except (zipfile.BadZipFile, OSError, FileNotFoundError):
+            zip_members = []
+
+    def _zip_url(member):
+        if member.lower().endswith('.md'):
+            return reverse('library:guide_zip_doc', args=[slug, version.pk, member])
+        return reverse('library:pack_asset', args=[slug, version.pk, member])
+
+    def _under_base(member):
+        if not base_dir:
+            return False
+        norm = _guide_norm_path(member)
+        return norm == base_dir or norm.startswith(base_dir + '/')
+
+    if source == 'zip':
+        for m in zip_members:
+            if _under_base(m):
+                _add(m, _zip_url(m))
+    for vf in version.files.all():
+        name = (vf.filename or '').replace('\\', '/')
+        if not name or (vf.kind == 'release' and name.lower().endswith('.zip')):
+            continue
+        if vf.is_markdown:
+            url = reverse('library:guide_doc', args=[slug, version.pk, vf.file.uuid])
+        else:
+            url = reverse('library:file_request', args=[vf.file.uuid])
+        _add(name, url)
+    for m in zip_members:
+        if not (source == 'zip' and _under_base(m)):
+            _add(m, _zip_url(m))
+    return full, base
+
+
+def _resolve_guide_src(src, full_map, base_map, base_dir=None):
+    """Rewrite one markdown target to its served URL, or None when unknown."""
+    if not src or src.startswith(_GUIDE_LINK_SKIP):
+        return None
+    path = src.split('#')[0].split('?')[0]
+    if not path:
+        return None
+    norm = _guide_norm_path(path)
+    if base_dir:
+        cand = _guide_norm_path(f'{base_dir}/{path}')
+        if cand in full_map:
+            return full_map[cand]
+    if norm in full_map:
+        return full_map[norm]
+    bname = norm.rsplit('/', 1)[-1]
+    if bname in base_map:
+        return base_map[bname]
+    return None
+
+
+def _rewrite_guide_links(md, version, base_dir=None, source='zip'):
+    """Rewrite markdown link/image/reference targets to served URLs whenever a
+    file with the same name is present in the version's uploads (individual
+    files or the release zip). Guide category only."""
+    full_map, base_map = _guide_asset_maps(version, base_dir=base_dir, source=source)
+
+    def _sub_inline(m):
+        new = _resolve_guide_src(m.group(2), full_map, base_map, base_dir)
+        return f'{m.group(1)}{new or m.group(2)}{m.group(3)}{m.group(4)}'
+
+    def _sub_ref(m):
+        new = _resolve_guide_src(m.group(2), full_map, base_map, base_dir)
+        return f'{m.group(1)}{new or m.group(2)}'
+
+    md = _GUIDE_LINK_RE.sub(_sub_inline, md)
+    md = _GUIDE_REF_RE.sub(_sub_ref, md)
+    return md
+
+
 def guide_doc(request, slug, version_id, file_uuid):
    """Render one markdown guide document server-side (used by the switcher)."""
    version = get_object_or_404(
@@ -503,6 +645,38 @@ def guide_doc(request, slug, version_id, file_uuid):
        text = fh.read().decode('utf-8', 'replace')
    finally:
        fh.close()
+    if version.project.category == 'guide':
+        text = _rewrite_guide_links(text, version, source='file')
+    return HttpResponse(render_markdown(text), content_type='text/html; charset=utf-8')
+
+
+def guide_zip_doc(request, slug, version_id, asset_path):
+    """Render one markdown document stored inside the version's release zip."""
+    version = get_object_or_404(
+        Version.objects.select_related('project'),
+        pk=version_id, project__slug=slug,
+    )
+    if version.project.category != 'guide':
+        raise Http404
+    if not asset_path or asset_path.startswith('/') or '..' in asset_path.split('/'):
+        raise Http404
+    if not asset_path.lower().endswith('.md'):
+        raise Http404
+    vf = (
+        version.files.select_related('file').filter(kind='release').first()
+        or version.files.select_related('file').first()
+    )
+    if vf is None or not (vf.filename or '').lower().endswith('.zip'):
+        raise Http404
+    with zipfile.ZipFile(_stored_path(vf.file)) as zf:
+        if asset_path not in zf.namelist():
+            raise Http404
+        data = zf.read(asset_path)
+    text = data.decode('utf-8', 'replace')
+    text = _rewrite_guide_links(
+        text, version,
+        base_dir=posixpath.dirname(asset_path), source='zip',
+    )
    return HttpResponse(render_markdown(text), content_type='text/html; charset=utf-8')