lino 25.2.2__py3-none-any.whl → 25.3.0__py3-none-any.whl

lino/modlib/uploads/actions.py

@@ -2,10 +2,8 @@
 # Copyright 2023 Rumma & Ko Ltd
 # License: GNU Affero General Public License v3 (see file COPYING for details)
 
-import base64
-from django.utils import timezone
 from lino.api import rt, dd, _
-from .mixins import make_captured_image
+from .mixins import base64_to_image
 
 
 class CameraStream(dd.Action):  # TODO: rename this to CaptureImage
@@ -29,12 +27,8 @@ class CameraStream(dd.Action):  # TODO: rename this to CaptureImage
     description
     """
 
-    def base64_to_image(self, imgstring):
-        imgdata = base64.b64decode(imgstring.split("base64,")[1])
-        return make_captured_image(imgdata, timezone.now())
-
     def handle_uploaded_file(self, ar, **kwargs):
-        file = self.base64_to_image(ar.request.POST["image"])
+        file = base64_to_image(ar.request.POST["image"])
         upload = rt.models.uploads.Upload(file=file, user=ar.get_user(), **kwargs)
         upload.save_new_instance(ar)
         return upload

lino/modlib/uploads/choicelists.py

@@ -6,6 +6,16 @@ from lino.api import dd, rt, _
 from lino.modlib.office.roles import OfficeStaff
 
 
+class UploadAreas(dd.ChoiceList):
+    required_roles = dd.login_required(OfficeStaff)
+    verbose_name = _("Upload area")
+    verbose_name_plural = _("Upload areas")
+
+
+add = UploadAreas.add_item
+add("90", _("Uploads"), "general")
+
+
 class Shortcut(dd.Choice):
     """Represents a shortcut field."""
 
@@ -31,15 +41,5 @@ class Shortcuts(dd.ChoiceList):
     max_length = 50  # fields get created before the values are known
 
 
-class UploadAreas(dd.ChoiceList):
-    required_roles = dd.login_required(OfficeStaff)
-    verbose_name = _("Upload area")
-    verbose_name_plural = _("Upload areas")
-
-
-add = UploadAreas.add_item
-add("90", _("Uploads"), "general")
-
-
 def add_shortcut(*args, **kw):
     return Shortcuts.add_item(*args, **kw)

lino/modlib/uploads/fixtures/std.py

@@ -0,0 +1,17 @@
+# -*- coding: UTF-8 -*-
+# Copyright 2015-2025 Rumma & Ko Ltd
+# License: GNU Affero General Public License v3 (see file COPYING for details)
+
+from lino.api import dd, rt
+from lino.modlib.system.choicelists import Recurrences
+from lino.modlib.uploads.choicelists import Shortcuts
+
+
+def objects():
+    UploadType = rt.models.uploads.UploadType
+
+    kw = dict(max_number=1, wanted=True)
+
+    for us in Shortcuts.get_list_items():
+        kw.update(dd.str2kw('name', us.text))
+        yield UploadType(shortcut=us, **kw)

lino/modlib/uploads/mixins.py

@@ -2,6 +2,7 @@
 # Copyright 2008-2024 Rumma & Ko Ltd
 # License: GNU Affero General Public License v3 (see file COPYING for details)
 
+import base64
 import os
 import shutil
 import uuid
@@ -17,6 +18,7 @@ from django.core.files.storage import default_storage
 from django.core.exceptions import ValidationError, FieldError
 from django.template.defaultfilters import filesizeformat
 from django.utils.html import format_html
+from django.utils import timezone
 
 from lino.core import constants
 from lino.utils import DATE_TO_DIR_TPL
@@ -57,7 +59,8 @@ def make_uploaded_file(filename, src=None, upload_date=None):
         raise Exception(f"Source {src} does not exist")
     filename = default_storage.generate_filename(safe_filename(filename))
     upload_to = Path(upload_date.strftime(upload_to_tpl))
-    upload_filename = default_storage.generate_filename(str(upload_to / filename))
+    upload_filename = default_storage.generate_filename(
+        str(upload_to / filename))
     dest = settings.SITE.media_root / upload_filename
     if needs_update(src, dest):
         print("cp {} {}".format(src, dest))
@@ -66,17 +69,26 @@ def make_uploaded_file(filename, src=None, upload_date=None):
     return upload_filename
 
 
-def make_captured_image(imgdata, upload_date=None):
+def base64_to_image(imgstring):
+    type, file = imgstring.split(";base64,")
+    imgdata = base64.b64decode(file)
+    return make_captured_image(imgdata, timezone.now(), ext=f".{type.split('/')[1]}")
+
+
+def make_captured_image(imgdata, upload_date=None, filename=None, ext='.jpg'):
     if upload_date is None:
-        upload_date = dd.demo_date()
-    filename = default_storage.generate_filename(
-        safe_filename(str(uuid.uuid4()) + ".jpg"))
+        upload_date = dd.today()
+    if not filename:
+        filename = str(uuid.uuid4()) + ext
+    filename = default_storage.generate_filename(safe_filename(filename))
     upload_to = Path(upload_date.strftime(upload_to_tpl))
-    upload_filename = default_storage.generate_filename(str(upload_to / filename))
+    upload_filename = default_storage.generate_filename(
+        str(upload_to / filename))
     dest = Path(settings.MEDIA_ROOT) / upload_filename
     dest.parent.mkdir(parents=True, exist_ok=True)
-    with dest.open("wb") as f:
-        f.write(imgdata)
+    dest.write_bytes(imgdata)
+    # with dest.open("wb") as f:
+    #     f.write(imgdata)
     return upload_filename
 
 

lino/modlib/uploads/models.py

@@ -2,6 +2,7 @@
 # Copyright 2008-2024 Rumma & Ko Ltd
 # License: GNU Affero General Public License v3 (see file COPYING for details)
 
+from .ui import *
 import os
 from os.path import join, exists
 from pathlib import Path
@@ -26,13 +27,14 @@ from lino.modlib.users.mixins import UserAuthored, My
 # from lino.modlib.office.roles import OfficeUser, OfficeStaff, OfficeOperator
 from lino.modlib.office.roles import OfficeStaff
 from lino.mixins import Referrable
+from lino.utils.soup import register_sanitizer
 from lino.utils.mldbc.mixins import BabelNamed
 from lino.modlib.checkdata.choicelists import Checker
 from lino.modlib.publisher.mixins import Publishable
 
 from .actions import CameraStream
-from .choicelists import Shortcuts, UploadAreas
-from .mixins import UploadBase
+from .choicelists import Shortcuts, UploadAreas, add_shortcut
+from .mixins import UploadBase, base64_to_image
 from .utils import previewer, UploadMediaFile
 
 from . import VOLUMES_ROOT
@@ -57,7 +59,7 @@ class Volume(Referrable):
 
     def full_clean(self, *args, **kw):
         super().full_clean(*args, **kw)
-        pth = dd.plugins.uploads.get_volumes_root() / self.ref
+        pth = dd.plugins.uploads.volumes_root / self.ref
         if pth.exists():
             if pth.resolve().absolute() != Path(self.root_dir).resolve().absolute():
                 raise ValidationError(
@@ -121,8 +123,10 @@ class Upload(UploadBase, UserAuthored, Controllable, Publishable):
     upload_area = UploadAreas.field(default="general")
     type = dd.ForeignKey("uploads.UploadType", blank=True, null=True)
     volume = dd.ForeignKey("uploads.Volume", blank=True, null=True)
-    library_file = models.CharField(_("Library file"), max_length=255, blank=True)
-    description = models.CharField(_("Description"), max_length=200, blank=True)
+    library_file = models.CharField(
+        _("Library file"), max_length=255, blank=True)
+    description = models.CharField(
+        _("Description"), max_length=200, blank=True)
     source = dd.ForeignKey("sources.Source", blank=True, null=True)
 
     camera_stream = CameraStream()
@@ -176,7 +180,7 @@ class Upload(UploadBase, UserAuthored, Controllable, Publishable):
         if self.file:
             return self.file.size
         if self.volume_id and self.library_file:
-            pth = dd.plugins.uploads.get_volumes_root() / self.volume.ref / self.library_file
+            pth = dd.plugins.uploads.volumes_root / self.volume.ref / self.library_file
             return pth.stat().st_size
             # return os.path.getsize(pth)
 
@@ -286,7 +290,8 @@ class UploadChecker(Checker):
             if not exists(join(settings.MEDIA_ROOT, obj.file.name)):
                 yield (
                     False,
-                    format_lazy(_("Upload entry {} has no file"), obj.file.name),
+                    format_lazy(_("Upload entry {} has no file"),
+                                obj.file.name),
                 )
                 return
 
@@ -299,6 +304,7 @@ class UploadChecker(Checker):
         for i in obj.check_previews(fix):
             yield i
 
+
 UploadChecker.activate()
 
 
@@ -308,7 +314,7 @@ class UploadsFolderChecker(Checker):
     def get_checkdata_problems(self, obj, fix=False):
         assert obj is None  # this is an unbound checker
         Upload = rt.models.uploads.Upload
-        pth = dd.plugins.uploads.get_uploads_root()
+        pth = dd.plugins.uploads.uploads_root
         assert str(pth).startswith(settings.MEDIA_ROOT)
         start = len(settings.MEDIA_ROOT) + 1
         for filename in Path(pth).rglob("*"):
@@ -319,7 +325,8 @@ class UploadsFolderChecker(Checker):
             qs = Upload.objects.filter(file=rel_filename)
             n = qs.count()
             if n == 0:
-                msg = format_lazy(_("File {} has no upload entry."), rel_filename)
+                msg = format_lazy(
+                    _("File {} has no upload entry."), rel_filename)
                 # print(msg)
                 yield (dd.plugins.uploads.remove_orphaned_files, msg)
                 if fix and dd.plugins.uploads.remove_orphaned_files:
@@ -357,31 +364,33 @@ def before_analyze(sender, **kwargs):
             items = []
             target = sender.modules.resolve(i.target)
             sar = ar.spawn_request(
-                actor=target, master_instance=obj, known_values=dict(type=utype)
-            )
+                actor=target, master_instance=obj, known_values=dict(
+                    type=utype))
             # param_values=dict(pupload_type=et))
             n = sar.get_total_count()
             if n == 0:
-                iar = target.insert_action.request_from(sar, master_instance=obj)
+                iar = target.insert_action.request_from(
+                    sar, master_instance=obj)
                 btn = iar.ar2button(
                     None,
-                    _("Upload"),
-                    icon_name="page_add",
-                    title=_("Upload a file from your PC to the server."),
-                )
+                    "⊕",  # _("Upload"),
+                    # icon_name="page_add",
+                    title=_("Upload a file from your PC to the server."))
                 items.append(btn)
-            elif n == 1:
+            # elif n == 1:
+            else:
                 after_show = ar.get_status()
                 obj = sar.data_iterator[0]
                 if (mf := obj.get_media_file()) is not None:
                     items.append(
                         sar.renderer.href_button(
                             mf.get_download_url(),
-                            _("show"),
+                            "⎙",  # Unicode symbol Print Screen
                             target="_blank",
-                            icon_name="page_go",
-                            style="vertical-align:-30%;",
-                            title=_("Open the uploaded file in a new browser window"),
+                            # icon_name="page_go",
+                            # style="vertical-align:-30%;",
+                            title=_(
+                                "Open the uploaded file in a new browser window"),
                         )
                     )
                 after_show.update(record_id=obj.pk)
@@ -389,23 +398,25 @@ def before_analyze(sender, **kwargs):
                     sar.window_action_button(
                         sar.ah.actor.detail_action,
                         after_show,
-                        _("Edit"),
-                        icon_name="application_form",
-                        title=_("Edit metadata of the uploaded file."),
+                        "⎆",  # Unicode symbol Enter
+                        # icon_name="application_form",
+                        title=_("Edit the information about the uploaded file."),
                     )
                 )
-            else:
-                obj = sar.sliced_data_iterator[0]
-                items.append(ar.obj2html(obj, pgettext("uploaded file", "Last")))
-
-                btn = sar.renderer.action_button(
-                    obj,
-                    sar,
-                    sar.bound_action,
-                    _("All {0} files").format(n),
-                    icon_name=None,
-                )
-                items.append(btn)
+            # else:
+            #     obj = sar.sliced_data_iterator[0]
+            #     items.append(ar.obj2html(
+            #         obj, pgettext("uploaded file", "Last")))
+
+            btn = sar.renderer.action_button(
+                obj,
+                sar,
+                sar.bound_action,
+                "⏏",  # _("All {0} files").format(n),
+                icon_name=None,
+                title=_("Manage the list of uploaded files.")
+            )
+            items.append(btn)
 
             return E.div(*join_elems(items, ", "))
 
@@ -415,8 +426,6 @@ def before_analyze(sender, **kwargs):
         #             i.model_spec, i.name)
 
 
-from .ui import *
-
 # raise Exception("20241112")
 
 
@@ -466,3 +475,18 @@ def setup_memo_commands(sender=None, **kwargs):
 
     mp = sender.plugins.memo.parser
     mp.register_django_model('file', rt.models.uploads.Upload, rnd=file2html)
+
+
+def on_sanitize(soup, save=False, ar=None):
+    # raise Exception(f"20250301")
+    for tag in soup.find_all():
+        tag_name = tag.name.lower()
+        if tag_name == "img" and tag['src'].startswith("data:image") and ar is not None and save:
+            file = base64_to_image(tag['src'])
+            upload = rt.models.uploads.Upload(file=file, user=ar.get_user())
+            sar = upload.get_default_table().request(parent=ar)
+            upload.save_new_instance(sar)
+            tag.replace_with(f'[file {upload.pk}]')
+
+
+register_sanitizer(on_sanitize)

lino/modlib/uploads/ui.py

@@ -21,7 +21,7 @@ from lino.core import constants
 def filename_leaf(name):
     i = name.rfind("/")
     if i != -1:
-        return name[i + 1 :]
+        return name[i + 1:]
     return name
 
 
@@ -153,6 +153,7 @@ class AreaUploads(Uploads):
     required_roles = dd.login_required(UploadsReader)
     stay_in_grid = True
     default_display_modes = {None: constants.DISPLAY_MODE_SUMMARY}
+    detailed_summary = False
 
     # 20180119
     # @classmethod
@@ -170,7 +171,8 @@ class AreaUploads(Uploads):
         return obj.description or filename_leaf(obj.file.name) or str(obj.id)
 
     @classmethod
-    def get_table_summary(self, obj, ar):
+    def get_table_summary(self, ar):
+        obj = ar.master_instance
         if obj is None:
             return
         UploadType = rt.models.uploads.UploadType
@@ -185,12 +187,14 @@ class AreaUploads(Uploads):
                 qs = qs.filter(upload_area=area)
         else:
             return E.div(
-                "{} is not an UploadController!".format(model_class_path(obj.__class__))
+                "{} is not an UploadController!".format(
+                    model_class_path(obj.__class__))
             )
         volume = obj.get_uploads_volume()
         # print(20190208, volume)
         for ut in qs:
-            sar = ar.spawn(self, master_instance=obj, known_values=dict(type_id=ut.id))
+            sar = ar.spawn(self, master_instance=obj,
+                           known_values=dict(type_id=ut.id))
             # logger.info("20140430 %s", sar.data_iterator.query)
             files = []
             for m in sar:
@@ -209,7 +213,8 @@ class AreaUploads(Uploads):
                         mf.get_download_url(),
                         # u"\u21A7",  # DOWNWARDS ARROW FROM BAR (↧)
                         # u"\u21E8",
-                        "\u21f2",  # SOUTH EAST ARROW TO CORNER (⇲)
+                        # "\u21f2",  # SOUTH EAST ARROW TO CORNER (⇲)
+                        "⎙",  # Unicode symbol Print Screen
                         style="text-decoration:none;",
                         # _(" [show]"),  # fmt(m),
                         target="_blank",
@@ -237,7 +242,8 @@ class AreaUploads(Uploads):
                 types.append(chunks)
         # logger.info("20140430 %s", [tostring(e) for e in types])
         # elems += [str(ar.bound_action.action.__class__), " "]
-        if ar.bound_action.action.window_type == "d":
+        # if ar.bound_action.action.window_type == "d":
+        if self.detailed_summary:
             if len(types) == 0:
                 elems.append(E.ul(E.li(str(ar.no_data_text))))
             else:
@@ -245,12 +251,12 @@ class AreaUploads(Uploads):
         else:
             if len(types) == 0:
                 elems.append(str(ar.no_data_text))
-                elems.append(" / ")
+                # elems.append(" / ")
             else:
                 for chunks in types:
                     elems.extend(chunks)
-                    elems.append(" / ")
-            elems.append(obj.show_uploads.as_button_elem(ar))
+                    # elems.append(" / ")
+            # elems.append(obj.show_uploads.as_button_elem(ar))
         # ba = self.find_action_by_name("show_uploads")
         return E.div(*elems)
 

lino/utils/__init__.py

@@ -40,7 +40,6 @@ function for general use. It has also many subpackages and submodules.
     pyuca
     quantities
     ranges
-    requests
     restify
     screenshots
     sqllog

lino/utils/jscompressor.py

@@ -39,18 +39,18 @@ class JSCompressor(object):
 
     literalMarker = "@_@%d@_@"  # temporary replacement
     # put the string literals back in
-    backSubst = re.compile("@_@(\d+)@_@")
+    backSubst = re.compile(r"@_@(\d+)@_@")
 
     # /* ... */ comments on single line
     mlc1 = re.compile(r"(\/\*.*?\*\/)")
     mlc = re.compile(r"(\/\*.*?\*\/)", re.DOTALL)  # real multiline comments
-    slc = re.compile("\/\/.*")  # remove single line comments
+    slc = re.compile(r"\/\/.*")  # remove single line comments
 
     # collapse successive non-leading white space characters into one
-    collapseWs = re.compile("(?<=\S)[ \t]+")
+    collapseWs = re.compile(r"(?<=\S)[ \t]+")
 
     squeeze = re.compile(
-        """
+        r"""
         \s+(?=[\}\]\)\:\&\|\=\;\,\.\+])   |     # remove whitespace preceding control characters
         (?<=[\{\[\(\:\&\|\=\;\,\.\+])\s+  |     # ... or following such
         [ \t]+(?=\W)                      |     # remove spaces or tabs preceding non-word characters

lino/utils/media.py

@@ -5,6 +5,7 @@
 """
 
 from os.path import join
+from pathlib import Path
 
 from django.conf import settings
 from lino.core.utils import is_devserver
@@ -17,10 +18,10 @@ has_davlink = False
 class MediaFile(object):
     """
     Represents a file on the server below :setting:`MEDIA_ROOT` with
-    two properties :attr:`name` and :attr:`url`.
+    two properties :attr:`path` and :attr:`url`.
 
-    It also takes into consideration the settings
-    :attr:`webdav_root <lino.core.site.Site.webdav_root>`
+    It takes into consideration the settings
+    :attr:`webdav_root <lino.core.site.Site.webdav_root>`,
     :attr:`webdav_protocol <lino.core.site.Site.webdav_protocol>`
     and
     :attr:`webdav_url <lino.core.site.Site.webdav_url>`
@@ -28,28 +29,49 @@ class MediaFile(object):
 
     def __init__(self, editable, *parts):
         self.editable = editable
-        self.parts = parts
-
-    @property
-    def name(self):
-        "return the filename on the server"
-        if self.editable and (has_davlink or settings.SITE.webdav_protocol):
-            return join(settings.SITE.webdav_root, *self.parts)
-        return join(settings.MEDIA_ROOT, *self.parts)
-
-    def get_url(self, request):
-        "return the url that points to file on the server"
-        if self.editable and request is not None:
-            if is_devserver():
-                url = "file://" + join(settings.SITE.webdav_root, *self.parts)
+        # self.parts = parts
+        if editable and settings.SITE.webdav_protocol:
+            path = Path(settings.SITE.webdav_root, *parts)
+            # 20250302 Removed the file:// trick on a devserver because anyway
+            # it doesn't work anymore. For editable media files we need a a
+            # webdav server and a protocol handler.
+            # if is_devserver():
+            if False:
+                url = "file://" + join(settings.SITE.webdav_root, *parts)
             else:
-                url = settings.SITE.webdav_url + "/".join(self.parts)
-                url = request.build_absolute_uri(url)
-            if settings.SITE.webdav_protocol:
-                url = settings.SITE.webdav_protocol + "://" + url
-            return url
+                url = settings.SITE.webdav_url + "/".join(parts)
+                if settings.SITE.webdav_protocol:
+                    url = settings.SITE.webdav_protocol + "://" + url
+        else:
+            path = Path(settings.MEDIA_ROOT, *parts)
+            url = settings.SITE.build_media_url(*parts)
+        self.url = url
+        self.path = path
 
-        return settings.SITE.build_media_url(*self.parts)
+    # @property
+    # def path(self):
+    #     "Return the full filename on the server as a Path object."
+
+    # @property
+    # def name(self):
+    #     "return the filename on the server"
+    #     if self.editable and (has_davlink or settings.SITE.webdav_protocol):
+    #         return join(settings.SITE.webdav_root, *self.parts)
+    #     return join(settings.MEDIA_ROOT, *self.parts)
+
+    # def get_url(self, request):
+    #     "return the url that points to file on the server"
+    #     if self.editable and request is not None:
+    #         if is_devserver():
+    #             url = "file://" + join(settings.SITE.webdav_root, *self.parts)
+    #         else:
+    #             url = settings.SITE.webdav_url + "/".join(self.parts)
+    #             url = request.build_absolute_uri(url)
+    #         if settings.SITE.webdav_protocol:
+    #             url = settings.SITE.webdav_protocol + "://" + url
+    #         return url
+    #
+    #     return settings.SITE.build_media_url(*self.parts)
 
 
 class TmpMediaFile(MediaFile):

lino/utils/report.py

@@ -37,10 +37,10 @@ class EmptyTableRow(VirtualRow, Printable):
     def __str__(self):
         return str(self._table.label)
 
-    def before_printable_build(self, bm):
-        pass
+    def must_build_printable(self, bm):
+        return True
 
-    def filename_root(self):
+    def get_printable_target_stem(self):
         return self._table.app_label + "." + self._table.__name__
 
     # def get_print_language(self):
@@ -239,7 +239,8 @@ class Report(EmptyTable):
 
     @classmethod
     def as_appy_pod_xml(cls, self, apr):
-        chunks = tuple(apr.story2odt(self.get_story(apr.ar), master_instance=self))
+        chunks = tuple(apr.story2odt(
+            self.get_story(apr.ar), master_instance=self))
         return str("").join(chunks)  # must be utf8 encoded
 
     @classmethod

lino/utils/restify.py

@@ -17,7 +17,7 @@ import re
 
 # This regular expression finds the indentation of every non-blank
 # line in a string.
-_INDENT_RE = re.compile("^([ ]*)(?=\S)", re.MULTILINE)
+_INDENT_RE = re.compile(r"^([ ]*)(?=\S)", re.MULTILINE)
 
 
 def min_indent(s):
@@ -451,7 +451,7 @@ def rst2latex(
 
 
 if __name__ == "__main__":
-    test = """
+    test = r"""
 Test example
 ============
 

lino/utils/soup.py

@@ -7,7 +7,8 @@
 import re
 from bs4 import BeautifulSoup, NavigableString, Comment, Doctype
 from bs4.element import Tag
-import logging; logger = logging.getLogger(__file__)
+import logging
+logger = logging.getLogger(__file__)
 # from lino.api import dd
 
 
@@ -39,6 +40,7 @@ WHITESPACE_TAGS = PARAGRAPH_TAGS | {
 SHORT_PREVIEW_IMAGE_HEIGHT = "8em"
 REMOVED_IMAGE_PLACEHOLDER = "⌧"
 
+
 class Style:
     # TODO: Extend rstgen.sphinxconf.sigal_image.Format to incoroporate this.
     def __init__(self, s):
@@ -289,6 +291,7 @@ ALLOWED_ATTRIBUTES["p"] = GENERALLY_ALLOWED_ATTRS | {"align"}
 #         return re.sub("(width|height):[^;]+;", "", css)
 #     return css
 
+
 def sanitized_soup(old):
 
     # Inspired by https://chase-seibert.github.io/blog/2011/01/28/sanitize-html-with-beautiful-soup.html
@@ -322,23 +325,38 @@ def sanitized_soup(old):
             tag.attrs = dict()
 
     # remove all comments because they might contain scripts
-    comments = soup.find_all(text=lambda text:isinstance(text, (Comment, Doctype)))
+    comments = soup.find_all(
+        text=lambda text: isinstance(text, (Comment, Doctype)))
     for comment in comments:
         comment.extract()
 
     # remove the wrapper tag if it is useless
-    if len(soup.contents) == 1:
-        main_tag = soup.contents[0]
-        if main_tag.name in useless_main_tags and not main_tag.attrs:
-            main_tag.unwrap()
+    # if len(soup.contents) == 1:
+    #     main_tag = soup.contents[0]
+    #     if main_tag.name in useless_main_tags and not main_tag.attrs:
+    #         main_tag.unwrap()
 
     return soup
 
-def sanitize(s):
+
+def sanitize(s, **kwargs):
     s = s.strip()
     if not s:
         return s
+
+    soup = sanitized_soup(s)
+
+    for func in SANITIZERS:
+        func(soup, **kwargs)
+
     # do we want to remove whitespace between tags?
     # s = re.sub(">\s+<", "><", s)
     # return sanitized_soup(s).decode(formatter="html").strip()
-    return str(sanitized_soup(s)).strip()
+    return str(soup).strip()
+
+
+SANITIZERS = []
+
+
+def register_sanitizer(func):
+    SANITIZERS.append(func)