NBInclude
NBInclude is a package for the Julia language which allows you to include and execute IJulia (Julia-language Jupyter) notebook files just as you would include an ordinary Julia file.
The goal of this package is to make notebook files just as easy to incorporate into Julia programs as ordinary Julia (.jl
) files, giving you the advantages of a notebook (integrated code, formatted text, equations, graphics, and other results) while retaining the modularity and re-usability of .jl
files.
Basic usage
Analogous to include("myfile.jl")
in Julia to execute myfile.jl
, you can do
using NBInclude
@nbinclude("myfile.ipynb")
to execute all of the code cells in the IJulia notebook myfile.ipynb
. Similar to include
, the value of the last evaluated expression in the last evaluated code cell is returned.
We also export an in_nbinclude()
function, which returns true
only when it is
executed in code run via @nbinclude
. Using this, you can selectively run code
in a notebook only interactively or only via @nbinclude
.
There is also a function
nbexport("myfile.jl", "myfile.ipynb")
that can be used to convert an IJulia notebook file to an ordinary Julia file, with Markdown text in the notebook converted to formatted comments in the Julia file.
Detailed features
Key features of @nbinclude
are:
- The path of the notebook is relative to the path of the current file (if any),
and nested inclusions can use paths relative to the notebook, just as for
include
. - In a module, included notebooks work fine with precompilation in Julia (and re-compilation is automatically triggered if the notebook changes).
- Code is associated with accurate line numbers (e.g. for backtraces when exceptions are thrown), in the form of
myfile.ipynb:In[N]:M
for lineM
in input cellN
of themyfile.ipynb
notebook. Un-numbered cells (e.g. unevaluated cells) are given a number+N
for theN
-th nonempty cell in the notebook. You can use@nbinclude("myfile.ipynb", renumber=true)
to automatically renumber the cells in sequence (as if you had selected Run All from the Jupyter Cell menu), without altering the file. - The Julia
@__FILE__
macro returns/path/to/myfile.ipynb:In[N]
for input cellN
. - In IJulia, cells beginning with
;
or?
are interpreted as shell commands or help requests, respectively. Such cells are ignored by@nbinclude
. counters
andregex
keywords can be used to include a subset of notebook cells to those for whichcounter β counters
and the cell text matchesregex
. For example,@nbinclude("notebook.ipynb"; counters=1:10, regex=r"#\s*EXECUTE")
would include cells 1 to 10 fromnotebook.ipynb
that contain comments like# EXECUTE
.- A keyword
anshook
can be used to run a passed function on the return value of all the cells. - No Python or Jupyter dependency.
- The
softscope
flag mentioned below.
Note: Scoping rules differ between interactive (IJulia, REPL) and non-interactive Julia code. Running a notebook as @nbinclude("foo.ipynb"; softscope=true)
will load notebooks using "soft" global scoping similar to interactive (REPL) code in Julia 1.5+ or for IJulia with any Julia version. That flag's default value, false
, will load notebooks with the "hard" scoping rule that Julia uses for non-interactive code (e.g. in include
); see also the SoftGlobalScope package for more details.
Key features of nbexport
are:
- You can either call
nbexport(filename, notebookfile)
to export to a file, ornbexport(io, notebookfile)
to write to anIO
stream (e.g.stdout
or a buffer). - To export to a string, use
sprint(nbexport, notebookfile)
. - Like
@nbinclude
, you can pass aregex
keyword to specify a subset of the notebook code cells to export. - Markdown cells in the notebook are parsed and formatted as pretty-printed text comments with the help of Julia's Markdown standard library.
- Markdown cells can be ignored by passing
markdown=false
tonbexport
.
Contact
NBInclude was written by Steven G. Johnson and is free/open-source software under the MIT/Expat license. Please file bug reports and feature requests at the NBInclude github page.