Plone 5 Upgrade Without UUID Links

If you're upgrading a Plone site to 5 and the site was not use UUID links, you'll notice a lot of your links are probably broken. The reason this happens is because Plone 5 no longer provides a <base> tag in the HTML output which non-uuid linked sites required to work properly.

To fix, create a python file named migrate-to-uuid.py with the following contents:

from AccessControl.SecurityManagement import newSecurityManager
from Acquisition import aq_parent
from Products.Archetypes.Field import TextField
from Products.CMFPlone.interfaces.siteroot import IPloneSiteRoot
from lxml.html import fromstring
from lxml.html import tostring
from plone.app.textfield.value import RichTextValue
from plone.dexterity.interfaces import IDexterityContent
from plone.uuid.interfaces import IUUID
import transaction


def getUID(ob):
    try:
        return IUUID(ob)
    except TypeError:
        return ob.UID()


def findObject(from_obj, path):
    while not IPloneSiteRoot.providedBy(from_obj):
        found = from_obj.restrictedTraverse(path, None)
        if found:
            return found
        from_obj = aq_parent(from_obj)


def transform_links_to_uuid(ob, html):
    if not html:
        return html
    changes = False
    try:
        dom = fromstring(html)
    except:
        print('Error parsing DOM from content: %s' % '/'.join(ob.getPhysicalPath()))
        return html
    for el in dom.cssselect('a'):
        href = el.attrib.get('href', '')
        if (not href or href.startswith('http://') or
                href.startswith('https://') or href.startswith('/') or
                'resolveuid' in href):
            continue
        link_obj = findObject(ob, href)
        if link_obj:
            uuid = getUID(link_obj)
            el.attrib.update({
                'href': 'resolveuid/%s' % uuid,
                'data-linktype': 'internal',
                'data-val': uuid
            })
            changes = True

    for el in dom.cssselect('img'):
        src = el.attrib.get('src', '')
        if (not src or src.startswith('http://') or
                src.startswith('https://') or src.startswith('/') or
                'resolveuid' in src):
            continue

        src, _, scale = src.partition('/@@images/image/')
        if not scale:
            src, _, scale = src.partition('/image_')
        img_obj = findObject(ob, src)
        if img_obj:
            uuid = getUID(img_obj)
            new_href = 'resolveuid/%s' % uuid
            if scale:
                new_href += '/@@images/image/' + scale

            attribs = {
                'src': new_href,
                'data-linktype': 'image',
                'data-val': uuid
            }
            if scale:
                attribs['data-scale'] = scale
            el.attrib.update(attribs)
            changes = True

    if changes:
        return tostring(dom)
    else:
        return html


def migrate_dexterity_to_uuid_links(ob):
    try:
        orig = ob.text.raw
        new = transform_links_to_uuid(ob, orig)
        if new != orig:
            ob.text = RichTextValue(
                new, mimeType=ob.text.mimeType, outputMimeType=ob.text.outputMimeType)
            print('Fixed content not using UUID links: %s' % '/'.join(ob.getPhysicalPath()))
            return 1
    except AttributeError:
        pass
    return 0


def migrate_archetypes_to_uuid_links(ob):
    for field in ob.Schema().fields():
        if type(field) == TextField and field.getContentType(ob) == 'text/html':
            orig = field.getRaw(ob)
            new = transform_links_to_uuid(ob, orig)
            if orig != new:
                field.set(ob, new)
                print('Fixed content not using UUID links: %s' % '/'.join(ob.getPhysicalPath()))
                return 1

    return 0


def migrate_to_uuid_links(site):
    catalog = site.portal_catalog
    count = 0
    for brain in catalog():
        ob = brain.getObject()
        if IDexterityContent.providedBy(ob):
            count += migrate_dexterity_to_uuid_links(ob)
        else:
            count += migrate_archetypes_to_uuid_links(ob)
        if count % 100 == 0:
            transaction.commit()
            print('finished processing %i' % count)


user = app.acl_users.getUser('admin')  # noqa
newSecurityManager(None, user.__of__(app.acl_users))  # noqa

for oid in app.objectIds():  # noqa
    _obj = app[oid]  # noqa
    if IPloneSiteRoot.providedBy(_obj):
        migrate_to_uuid_links(_obj)
        transaction.commit()

This script will run through every Plone site on your instance. If you want to target, just one Plone site, you'll need to modify the bottom of the script.

Then, just run the script against your instance:

./bin/client1 run migrate-to-uuid.py