Re: Executing a Notebook and saving the modified Notebook - all without the GUI

*To*: mathgroup at smc.vnet.net*Subject*: [mg125141] Re: Executing a Notebook and saving the modified Notebook - all without the GUI*From*: A Retey <awnl at gmx-topmail.de>*Date*: Fri, 24 Feb 2012 00:57:14 -0500 (EST)*Delivered-to*: l-mathgroup@mail-archive0.wolfram.com

Am 23.02.2012 11:47, schrieb C. Mundi: > Surely there is a "simple" way to tell Mathematica from the command > line to run a notebook and save the state of the notebook! We would > have hoped for something like "mathematica.exe --nogui --notebook > Test.nb" would work, especially if we embedded something like > "NotebookSave[]; End[];" at the end of the file. > > We've been searching for an answer to this question for days, and none > of the answers we've found actually solves our problem. So before you > answer, please read my question, which I will try to make very clear > with a very simple example at the risk of seeming pedantic. > > Given a Notebook file, Test.nb with a single cell > > In[1]:= Plot[Cos[x], {x, 0, 2 Pi}] > > we can open the Notebook in the front end, Evaluate the (whole) > Notebook and get the expected Out which is a plot of a single cycle. > We now save the notebook, using the GUI menus. We then open Test.nb > in the front end and we see both the input expression and the output > plot saved previously. Good. > > Now we want to do everything *without* the GUI. Let our INPUT be the > original Test.nb file, i.e. inputs but no outputs yet. Let our RESULT > be a modified Test.nb including the plot. We seek a sequence of batch- > like commands which will transform our INPUT into our RESULT, viz. > with no GUI interaction and in fact under the control of an external > unix shell script. It is permissible for the script to cause > mathematica to access files besides Test.nb. > > We need the output to be in NB form, because we need the embedded > graphics. But we also need to evaluate the notebook in an automated, > unattended way. > > At this time, solutions based on MathLink are not attractive to us. > > We have looked at FrontEndToken but have not yet found a way to effect > the result we seek. We also note that some of the interesting > FrontEndTokens -- particularly those related to Evaluate -- come with > warnings that they are not yet fully integrated into the long-term > plan for Mathematica. That sounds ominous. No need to refer to FrontEndTokens for this. You should be able to write a script that would be run by the kernel with using UsingFrontEnd, NotebookOpen, NotebookEvaluate, NotebookSave. I have not time to actually try that but I don't think it would be very difficult to achieve. See the documentation for details, for NotebookEvaluate you will need to set InsertResults -> True. hth, albert