Merge pull request #980 from maartenbreddels/fix_non_unique_attachments

Improved attachments +fix supporting non-unique names

nbconvert/exporters/html.py

@@ -8,6 +8,7 @@
 
 from traitlets import default, Unicode
 from traitlets.config import Config
+from jinja2 import contextfilter
 
 from nbconvert.filters.highlight import Highlight2HTML
 from nbconvert.filters.markdown_mistune import IPythonRenderer, MarkdownWithMath
@@ -66,9 +67,12 @@ def default_config(self):
         c.merge(super(HTMLExporter,self).default_config)
         return c
 
-    def markdown2html(self, source):
+    @contextfilter
+    def markdown2html(self, context, source):
         """Markdown to HTML filter respecting the anchor_link_text setting"""
-        renderer = IPythonRenderer(escape=False,
+        cell = context['cell']
+        attachments = cell.get('attachments', {})
+        renderer = IPythonRenderer(escape=False, attachments=attachments,
                                    anchor_link_text=self.anchor_link_text)
         return MarkdownWithMath(renderer=renderer).render(source)
 
@@ -77,22 +81,9 @@ def default_filters(self):
             yield pair
         yield ('markdown2html', self.markdown2html)
 
-    def process_attachments(self, nb, output):
-        for cell in nb.cells:
-            if 'attachments' in cell:
-                for key, attachment in cell['attachments'].items():
-                    for att_type, att in attachment.items():
-                        output = output.replace(
-                            'attachment:{}'.format(key),
-                            'data:' + att_type + ';base64, ' + attachment[att_type])
-        return output
-
     def from_notebook_node(self, nb, resources=None, **kw):
         langinfo = nb.metadata.get('language_info', {})
         lexer = langinfo.get('pygments_lexer', langinfo.get('name', None))
         highlight_code = self.filters.get('highlight_code', Highlight2HTML(pygments_lexer=lexer, parent=self))
         self.register_filter('highlight_code', highlight_code)
-
-        output, resources = super(HTMLExporter, self).from_notebook_node(nb, resources, **kw)
-        att_output = self.process_attachments(nb, output)
-        return att_output, resources
+        return super(HTMLExporter, self).from_notebook_node(nb, resources, **kw)

nbconvert/exporters/tests/test_html.py

@@ -117,8 +117,14 @@ def test_attachments(self):
         )
         check_for_png = re.compile(r'<img src="[^"]*?"([^>]*?)>')
         result = check_for_png.search(output)
-        self.assertTrue(result.group(0).strip().startswith('<img src="data:image/png;base64, iVBOR'))
-        self.assertTrue(result.group(1).strip().startswith('alt="python.png"'))
+        self.assertTrue(result.group(0).strip().startswith('<img src="data:image/png;base64,iVBOR'))
+        self.assertTrue(result.group(1).strip().startswith('alt="image.png"'))
+
+        check_for_data = re.compile(r'<img src="(?P<url>[^"]*?)"')
+        results = check_for_data.findall(output)
+        assert results[0] != results[1], 'attachments only need to be unique within a cell'
+        assert 'image/svg' in results[1], 'second image should use svg'
+
 
     def test_custom_filter_highlight_code(self):
         # Overwriting filters takes place at: Exporter.from_notebook_node

nbconvert/filters/markdown_mistune.py

@@ -147,6 +147,31 @@ def latex_environment(self, name, text):
     def inline_math(self, text):
         return '$%s$' % self.escape_html(text)
 
+    def image(self, src, title, text):
+        """Rendering a image with title and text.
+
+        :param src: source link of the image.
+        :param title: title text of the image.
+        :param text: alt text of the image.
+        """
+        attachments = self.options.get('attachments', {})
+        attachment_prefix = 'attachment:'
+        if src.startswith(attachment_prefix):
+            name = src[len(attachment_prefix):]
+            assert name in attachments, "missing attachment: {}".format(name)
+            attachment = attachments[name]
+            # we choose vector over raster, and lossless over lossy
+            preferred_mime_types = ['image/svg+xml', 'image/png', 'image/jpeg']
+            for preferred_mime_type in preferred_mime_types:
+                if preferred_mime_type in attachment:
+                    break
+            else:  # otherwise we choose the first mimetype we can find
+                preferred_mime_types = attachment.keys()[0]
+            mime_type = preferred_mime_type
+            data = attachment[mime_type]
+            src = 'data:' + mime_type + ';base64,' + data
+        return super(IPythonRenderer, self).image(src, title, text)
+
 
 def markdown2html_mistune(source):
     """Convert a markdown string to HTML using mistune"""

setup.py

@@ -198,7 +198,7 @@ def run(self):
 
 setup_args['install_requires'] = [
     'mistune>=0.8.1',
-    'jinja2',
+    'jinja2>=2.4',
     'pygments',
     'traitlets>=4.2',
     'jupyter_core',