Custom Export Configurations¶
The simplest application of this, would be to copy
latex_ipypublish_all.json
(or the html/slides variants) and make changes to the
cell_defaults
and nb_defaults
dictionaries, to suit your output
needs, then run:
nbpublish -f path/to/latex_ipypublish_all.json notebook.ipynb
The Conversion Process¶
iPyPublish uses export configuration files to control how the Notebook(s) will be exported. As shown in the figure below, they define two key components:
- The export class, and its associated pre-processors and filter functions.
- The jinja template outline and segments to be inserted into it.
This process extends nbconvert
in a number of ways:
- Merging of notebooks is handled automatically
- Numerous additional
ipypublish.preprocessors
andipypublish.filters
are supplied. - jinja templates are constructed via segment insertions, into a skeleton (outline) template, rather than by inheritance only. This allows for greater control and modularity in their construction.
- The use of
latexmk
with XeLaTeX to convert TeX to PDF, and correct resolution of file references and citations.
The Configuration File Format¶
The configuration file is a JSON file, with a validation schema given in Export Configuration Schema. Below is a minimal example:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 | {
"description": [
"A description of the configuration"
],
"exporter": {
"class": "nbconvert.exporters.LatexExporter",
"preprocessors": [
{
"class": "ipypublish.preprocessors.latex_doc_links.LatexDocLinks",
"args":
{
"metapath": "${meta_path}",
"filesfolder": "${files_path}"
}
}
],
"filters": {
"remove_dollars": "ipypublish.filters.filters.remove_dollars",
},
"other_args": {}
},
"template": {
"outline": {
"module": "ipypublish.templates.outline_schemas",
"file": "latex_tplx_schema.json"
},
"segments": [
{
"module": "ipypublish.templates.segments",
"file": "std-standard_packages.latex-tpl.json"
},
{
"directory": "path/to/folder",
"file": "a_user_defined_segment.json"
}
]
}
}
|
Exporter Class¶
In line 6, we define the exporter class, which can be any class in the python
environment namespace that inherits from
nbconvert.exporters.Exporter
.
Exporters can be parsed any number of preprocessors
(inheriting from nbconvert.preprocessors.Preprocessor
),
which act on the notebook in the order supplied.
The args
field is used to set any configurable traitlets
the class exposes.
Two special placeholders are available:
${meta_path}
will be set dynamically as the path to the (primary) ipynb file, containing the document level meta-data.${files_path}
will be set dynamically as the path to the folder where, additional files (such as internal images) will be output to.
Filters provide functions or classes to transform particular content of the notebook, and are parsed to the jinja templating engine.
See also
- The classes available natively in nbconvert:
nbconvert.exporters
,nbconvert.preprocessors
,nbconvert.filters
. - How Filters are used in jinja.
Template Construction¶
In line 22, we define how to construct the jinja template.
The outline
key defines the path to an outline schema,
such as in Template Outline Configuration Example.
This file achieves two things; to define an outline of the jinja template
structural blocks,
with placeholders to be replaced by str.format()
, and to
provide a schema for segment files which are used to replace
one or more of the placeholders.
This approach allows independent aspects of the document to be stored separately then pieced together in the desired manner. For example, the segment defined in Example of a Template Segment defines only parts of the document which define how the bibliography is constructed. This could be removed or replaced by a custom export configuration. Similarly, input and output prompts can be added/removed in html documents.
Segments are applied in the order they are defined, and the outline schema defines whether they are appended above or below existing content. For example, these segments:
[
{
"notebook_input_markdown_pre": "<div class='inner'>",
"notebook_input_markdown": " test",
"notebook_input_markdown_post": "</div>",
},
{
"notebook_input_markdown_pre": "<div class='outer'>",
"notebook_input_markdown_post": "</div>",
}
]
will result in a template containing:
<div class='outer'>
<div class='inner'>
test
</div>
</div>
Segment configuration files also have an optional overwrite
key, which
define segments that overwrite any previously defined content in that section.
See also
- The jinja documentation on Template Designer Documentation
- The nbconvert documentation on Customizing nbconvert
Loading Custom Configurations¶
Custom configurations can be parsed directly to nbpublish
:
nbpublish -f path/to/configs/export_config.json notebook.ipynb
Or used as a key, by providing nbpublish
with additional folders to scan
(in addition to the ipypublish.export_plugins
module folder):
nbpublish -ep path/to/configs -f export_config notebook.ipynb
Conversion of Plugins From Old API¶
The old style export plugins (defined as python scripts)
can be converted to the new JSON style, using the
ipypublish.port_api.plugin_to_json.convert_to_json()
function.
The old style template segment dictionaries (defined as python scripts)
can be converted to the new JSON style, using the
ipypublish.port_api.tpl_dct_to_json.py_to_json()
function.