Merge ae72c5a into c21822e

CHANGES.rst

@@ -6,6 +6,9 @@ Changelog
 
 - Lowercased email address after validation.
   [sgeulette]
+- Added `xhtml.imagesToData` that turns the src of images used in a xhtml
+  content from an `http` or equivalent URL to a data base64 value.
+  [gbastien]
 
 0.42 (2021-04-30)
 -----------------

src/imio/helpers/cache.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 from datetime import datetime
+from persistent.mapping import PersistentMapping
 from plone import api
 from plone.i18n.normalizer import IIDNormalizer
 from plone.memoize import ram
@@ -9,7 +10,6 @@
 from zope.component import getUtility
 from zope.component import queryUtility
 from zope.schema.interfaces import IVocabularyFactory
-from persistent.mapping import PersistentMapping
 
 import logging
 

src/imio/helpers/emailer.py

@@ -1,24 +1,25 @@
 # -*- coding: utf-8 -*-
 
+from email import encoders
 from email.mime.base import MIMEBase
 from email.mime.multipart import MIMEMultipart
 from email.mime.text import MIMEText
 from email.utils import parseaddr
-from email import encoders
 from imio.helpers import _
 from imio.helpers.content import safe_encode
 from plone import api
 from Products.CMFDefault.exceptions import EmailAddressInvalid
 from Products.CMFDefault.utils import checkEmailAddress
 from Products.CMFPlone.utils import safe_unicode
 from smtplib import SMTPException
-from zope.component import getMultiAdapter
 from zope import schema
+from zope.component import getMultiAdapter
 
 import csv
 import logging
 import socket
 
+
 logger = logging.getLogger("imio.helpers")
 EMAIL_CHARSET = 'utf-8'
 

src/imio/helpers/patches.py

@@ -2,6 +2,7 @@
 
 import pkg_resources
 
+
 try:
     pkg_resources.get_distribution("collective.solr")
 except pkg_resources.DistributionNotFound:

src/imio/helpers/security.py

@@ -10,12 +10,12 @@
 from random import sample
 from random import seed
 from time import time
+from zope.component import getMultiAdapter
 
 import logging
 import os
 import string
 
-from zope.component import getMultiAdapter
 
 logger = logging.getLogger("imio.helpers")
 

src/imio/helpers/tests/test_content.py

@@ -14,8 +14,8 @@
 from imio.helpers.content import get_state_infos
 from imio.helpers.content import get_vocab
 from imio.helpers.content import normalize_name
-from imio.helpers.content import object_values
 from imio.helpers.content import object_ids
+from imio.helpers.content import object_values
 from imio.helpers.content import restore_link_integrity_checks
 from imio.helpers.content import richtextval
 from imio.helpers.content import safe_delattr

src/imio/helpers/tests/test_security.py

@@ -5,6 +5,7 @@
 from imio.helpers.security import get_environment
 from imio.helpers.security import get_user_from_criteria
 from imio.helpers.security import is_develop_environment
+from imio.helpers.security import setup_logger
 from imio.helpers.testing import IntegrationTestCase
 from plone import api
 
@@ -58,3 +59,7 @@ def test_get_user_from_criteria(self):
         self.assertEqual(len(get_user_from_criteria(self.portal, fullname='Stéph')), 1)
         self.assertEqual(len(get_user_from_criteria(self.portal, email='.be')), 2)
         self.assertEqual(len(get_user_from_criteria(self.portal, fullname='Smith')), 2)
+
+    def test_setup_logger(self):
+        # just call it to check that it is not broken
+        self.assertIsNone(setup_logger())

src/imio/helpers/tests/test_xhtml.py

@@ -2,6 +2,7 @@
 from imio.helpers.testing import IntegrationTestCase
 from imio.helpers.xhtml import addClassToContent
 from imio.helpers.xhtml import addClassToLastChildren
+from imio.helpers.xhtml import imagesToData
 from imio.helpers.xhtml import imagesToPath
 from imio.helpers.xhtml import markEmptyTags
 from imio.helpers.xhtml import object_link
@@ -16,6 +17,7 @@
 
 import urllib
 
+
 picsum_image1_url = 'https://i.picsum.photos/id/10/200/300.jpg?hmac=94QiqvBcKJMHpneU69KYg2pky8aZ6iBzKrAuhSUBB9s'
 picsum_image2_url = 'https://i.picsum.photos/id/1082/200/200.jpg?hmac=3usO1ziO7kCseIG52ruhRigxyk39W_L9eECWe1Hs6fY'
 
@@ -32,6 +34,20 @@
     "OXORpldeTR7KdNjU1QVosfhpsfhpt/cwHP6JojjAmhovhRofL5/KTY7NdGBPhToBdjz+L9gTfb28/QDsHx4Ge5rVu2+zHWNDQ3InVk+YbvWe" \
     "+HEAkp6v/6cnemATSvoCmtFvb0Kvy47BR72AfwDXsx4tZedcTQAAAABJRU5ErkJggg=="
 
+base64_gif_img_data = "data:image/gif;base64,R0lGODlhCgAKAPcAAP////79/f36+/3z8/zy8/rq7Prm6Pnq7Pje4vTx8v" \
+    "Pg5O6gqe2gqOq4v+igqt9tetxSYNs7Tdo5TNc5TNUbMdUbMNQRKNIKIdIJINIGHdEGHtEDGtACFdAAFtAAFNAAEs8AFAAAAAAAA" \
+    "AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA" \
+    "AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA" \
+    "AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA" \
+    "AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA" \
+    "AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA" \
+    "AAAAAAAAAAAAAAAAAAAAAAAAAAAAACIAIP3sAAoACBL2OAAAAPjHWRQAABUKiAAAABL2FAAAAhL+dPkBhPm4MP///xL2YPZNegA" \
+    "AAAAQAAAABgAAAAAAABL2wAAAAAAARRL25PEPfxQAAPEQAAAAAAAAA/3r+AAAAAAAGAAAABL2uAAAQgAAABL2pAAAAAAAAAAAAAA" \
+    "ADAAAAgABAfDTAAAAmAAAAAAAAAAAABL27Mku0AAQAAAAAAAAAEc1MUaDsAAAGBL3FPELyQAAAAABEgAAAwAAAQAAAxL22AAAmB" \
+    "L+dPO4dPPKQP///0c70EUoOQAAmMku0AAQABL3TAAAAEaDsEc1MUaDsAAABkT98AABEhL3wAAALAAAQAAAQBL3kQAACSH5BAEAA" \
+    "AAALAAAAAAKAAoAQAhGAAEIHEhw4IIIFCAsKCBwQAQQFyKCeKDAIEKFDAE4hBiRA0WCAwwUBLCAA4cMICg0YIjAQoaIFzJMSCCw" \
+    "5EkOKjM2FDkwIAA7"
+
 
 class TestXHTMLModule(IntegrationTestCase):
     """
@@ -456,6 +472,27 @@ def test_imagesToPath(self):
         text = '<img src="resolveuid/unknown_uid" alt="Image" title="Image">'
         self.assertEqual(imagesToPath(doc2, text), text)
 
+    def test_imagesToData(self):
+        """
+          Test that images src contained in a XHTML content are correctly changed to
+          the a data base64 value.
+          Method is based on same as imagesToPath so we do not redo
+          tests that are already done in test_imagesToPath.
+        """
+        # create a document and an image
+        docId = self.portal.invokeFactory('Document', id='doc', title='Document')
+        doc = getattr(self.portal, docId)
+        file_path = path.join(path.dirname(__file__), 'dot.gif')
+        data = open(file_path, 'r')
+        img = self.portal.invokeFactory('Image', id='img', title='Image', file=data.read())
+        img = getattr(self.portal, img)
+        # has a blob
+        self.assertEqual(img.get_size(), 873)
+        text = '<p>Image <img src="{0}/img"> end of text.</p>'.format(self.portal_url)
+        self.assertEqual(
+            imagesToData(doc, text),
+            '<p>Image <img src="{0}"> end of text.</p>'.format(base64_gif_img_data))
+
     def test_storeExternalImagesLocally(self):
         """
           Test that images src contained in a XHTML that reference external images is
@@ -694,6 +731,10 @@ def test_object_link(self):
                          u'<a href="http://nohost/plone/folder/edit" target="_blank"></a>')
 
     def test_separate_images(self):
+        # no image, content is returned as is
+        text = '<p>My text.</p><p>My text.</p><p>My text.</p>'
+        result = separate_images(text)
+        self.assertEqual(text, result)
         # one image, nothing changed
         text = '<p><img src="http://plone/nohost/image1.png"></p>'
         result = separate_images(text)

src/imio/helpers/xhtml.py

@@ -9,6 +9,7 @@
 from zExceptions import NotFound
 from zope.container.interfaces import INameChooser
 
+import base64
 import cgi
 import logging
 import lxml.html
@@ -254,7 +255,48 @@ def removeCssClasses(xhtmlContent,
                     for x in tree.iterchildren()])
 
 
-def imagesToPath(context, xhtmlContent, pretty_print=False):
+def _img_from_src(context, img, portal, portal_url):
+    """ """
+    # check if it is a local or an external image
+    img_src = img.attrib.get('src', None)
+    # wrong <img> without src or external image
+    if not img_src or (img_src.startswith('http') and not img_src.startswith(portal_url)):
+        return
+    # here, we have an image contained in the portal
+    # either absolute path (http://...) or relative (../images/myimage.png)
+    imageObj = None
+    # absolute path
+    if img_src.startswith(portal_url):
+        img_src = img_src.replace(portal_url, '')
+        try:
+            # get the image but remove leading '/'
+            imageObj = portal.unrestrictedTraverse(img_src[1:])
+        except (KeyError, AttributeError, NotFound):
+            return
+    # relative path
+    else:
+        try:
+            imageObj = context.unrestrictedTraverse(img_src)
+        # in case we have a wrong resolveuid/unknown_uid, it raises NotFound
+        except (KeyError, AttributeError, NotFound):
+            return
+
+    # maybe we have a ImageScale instead of the real Image object?
+    if isinstance(imageObj, ImageScale):
+        imageObj = imageObj.aq_inner.aq_parent
+    return imageObj
+
+
+def _get_image_blob(imageObj):
+    """Be defensinve in case this is a wrong <img> with a src
+       to someting else than an image... """
+    blob = None
+    if hasattr(aq_base(imageObj), 'getBlobWrapper') and imageObj.get_size():
+        blob = imageObj.getBlobWrapper()
+    return blob
+
+
+def _transform_images(context, xhtmlContent, pretty_print=False, transform_type="path"):
     '''Turn <img> source contained in given p_xhtmlContent to a FileSystem absolute path
        to the .blob binary stored on the server.  This is usefull when generating documents
        with XHTML containing images that are private, LibreOffice is not able to access these
@@ -284,39 +326,23 @@ def imagesToPath(context, xhtmlContent, pretty_print=False):
     portal = api.portal.get()
     portal_url = portal.absolute_url()
     for img in imgs:
-        # check if it is a local or an external image
-        img_src = img.attrib.get('src', None)
-        # wrong <img> without src or external image
-        if not img_src or (img_src.startswith('http') and not img_src.startswith(portal_url)):
+        imageObj = _img_from_src(context, img, portal, portal_url)
+        if imageObj is None:
             continue
-        # here, we have an image contained in the portal
-        # either absolute path (http://...) or relative (../images/myimage.png)
-        imageObj = None
-        # absolute path
-        if img_src.startswith(portal_url):
-            img_src = img_src.replace(portal_url, '')
-            try:
-                # get the image but remove leading '/'
-                imageObj = portal.unrestrictedTraverse(img_src[1:])
-            except (KeyError, AttributeError, NotFound):
-                continue
-        # relative path
-        else:
-            try:
-                imageObj = context.unrestrictedTraverse(img_src)
-            # in case we have a wrong resolveuid/unknown_uid, it raises NotFound
-            except (KeyError, AttributeError, NotFound):
-                continue
-        # maybe we have a ImageScale instead of the real Image object?
-        if isinstance(imageObj, ImageScale):
-            imageObj = imageObj.aq_inner.aq_parent
+
+        blob = _get_image_blob(imageObj)
+        # change img src only if a blob was found
         blob_path = None
-        # be defensinve in case this is a wrong <img> with a src to someting else than an image...
-        if hasattr(aq_base(imageObj), 'getBlobWrapper') and imageObj.get_size():
-            blob_path = imageObj.getBlobWrapper().blob._p_blob_committed
-        # change img src only if a blob_path was found
-        if blob_path:
-            img.attrib['src'] = blob_path
+        if blob:
+            if transform_type == "path":
+                blob_path = blob.blob._p_blob_committed
+                if blob_path:
+                    img.attrib['src'] = blob_path
+            elif transform_type == "data":
+                blob_path = blob.blob._p_blob_committed
+                if blob_path and blob.content_type.startswith('image/'):
+                    img.attrib['src'] = "data:{0};base64,{1}".format(
+                        blob.content_type, base64.b64encode(blob.data))
 
     # use encoding to 'ascii' so HTML entities are translated to something readable
     return ''.join([lxml.html.tostring(x,
@@ -325,6 +351,29 @@ def imagesToPath(context, xhtmlContent, pretty_print=False):
                                        method='html') for x in tree.iterchildren()])
 
 
+def imagesToPath(context, xhtmlContent, pretty_print=False):
+    '''Turn <img> source contained in given p_xhtmlContent to a FileSystem absolute path
+       to the .blob binary stored on the server.  This is usefull when generating documents
+       with XHTML containing images that are private, LibreOffice is not able to access these
+       images using the HTTP request.
+       <img src='http://mysite/myfolder/myimage.png' /> becomes
+       <img src='/absolute/path/to/blobstorage/myfile.blob'/>,
+       external images are left unchanged.
+       The image_scale is not kept, so :
+       <img src='http://mysite/myfolder/myimage.png/image_preview' /> becomes
+       <img src='/absolute/path/to/blobstorage/myfile.blob'/>.'''
+
+    return _transform_images(context, xhtmlContent, pretty_print, transform_type="path")
+
+
+def imagesToData(context, xhtmlContent, pretty_print=False):
+    '''Turn <img> source contained in given p_xhtmlContent to a data:image/png;base64... value.
+       External images are left unchanged.
+       The image_scale is not kept, so we get the full image.'''
+
+    return _transform_images(context, xhtmlContent, pretty_print, transform_type="data")
+
+
 def storeImagesLocally(context,
                        xhtmlContent,
                        imagePortalType='Image',