🚀 go-pugleaf

At each iteration the script asks InDesign to create an array of topics, so maybe you could speed things up a bit by going about it like this:

topics = app.activeDocument.indexes[0].allTopics; 
topics_length = topics.length; 
for (var i=0; i < topics_length; i++) {  
  $.writeln(i+' '+topics[i].name);  
}



As there are so many topics, this could make a difference.

Peter

Hi Rick,

Two things:

1. Large arrays of text references are very slow in ExtendScript.

2. Try running the script via app.doScript and turn off Undo. Often, the performance problem is related to InDesign's Undo feature (which writes to disk with each change to the document database).

Thanks,

Ole

Hi Peter,

I am a bit embarrassed, but you are exactly right. When I get the topics ahead of time, it cruises right through the list! Thank you very much.

Ole, thanks for the suggestion. I will try this later in the process.

Rick

Olav,

is it really true, that reading something - without changing anything in document - will affect Undo ?

robin
-- 
www.adobescripts.com

Thanks, Ole.

Ole,

Are you sure that collapsing the undo stack really increases performance?

My understanding is that the undos are still recorded the same way, but 
they are just collapsed into one general item in the undo stack. If 
anything, there should be more overhead (collapsing them) -- not less.

I guess I should really do performance tests to find out...

-- 
Harbs
http://www.in-tools.com

Ole was right: doScript() with UndoModes.fastEntireScript gives the best results. The test was a document without any formatting, no undo history, containing 36,000 characters. The script creates a character style and sets underline to true. Then it finds all "A" and "a" (2,680 instances), changes the contents of each found instance with "x" and applies the character style.

Results:

UndoModes.fastEntireScript	22.6, 22.8, 23.1
UndoModes.autoUndo	25.8, 25.8, 25.9
UndoModes.entireScript	26.3, 26.4, 26.5
UndoModes.scriptRequest	29.9, 29,3, 29.2

Conclusion: fastEntireScript lives up to its name and wins hands-down.

My next step, I thought, was to add from now on this line at the beginning of labour-intensive scripts: 

      app.activeScriptUndoMode = UndoModes.fastEntireScript;




But unfortunately, activeScriptUndoMode is read only. Time for a feature request.

Peter

The timings in post #10 are from a script run from the ESTK. It has been mentioned that running scripts from the palette is quicker, which is confirmed:

UndoModes.fastEntireScript 21, 21
UndoModes.autoUndo 50, 43 / 30, 24
UndoModes.entireScript 52, 33 / 31, 26
UndoModes.scriptRequest 24, 25

The timings for autoUndo and entireScript look bizarre here, I've no idea what was going on. Timings after the slash were obtained after restarting InDesign. Didn't improve. The downward trend of these two modes in seuccessive runs is strange.

Anyway, all this confirms that the quickest way to run a script is to have it fired up by doScript() using UndoModes.fastEntireScript.

Peter

The longer times for entireScript and autoUndo seems to confirm my 
suspicion that there's extra overhead with collapsing the undo stack. I 
wonder what's done differently with fastEntireScript, and why do we need 
the regular entireScript if we are presented with a faster option?

The fact that the UndoModes are read-only is not really that big of a 
deal. All you need to do is add to the beginning of the script:
var undoName = "Script Name";
app.doScript(main, undefined , undefined,UndoModes.fastEntireScript , 
undoName );

instead of:
main().

-- 
Harbs
http://www.in-tools.com

Back to an index question. I need to collect all of the index entries in an InDesign book and create an index. Can anyone suggest a good data structure that I can use to collect the index entries? I will need to collect the entries and the corresponding page numbers. After they are collected, I will need to do a primary sort on the entries and a secondary sort on the page numbers. Any suggestions would be appreciated.

Rick Quatro
585-659-8267

Harbs,

That looks promising, but when I try that, I get an error "UndoModes is undefined". Any ideas?

Thanks,

Peter

Rick,

Are you trying to mimick "Generate index" from the Index panel?

Peter

I'm telling you: somethings fishy with the enums in CS4. I've had this 
kind of problem on many occasions... :(

try:

app.doScript(main,undefined,undefined,1699964501,undoName);


-- 
Harbs
http://www.in-tools.com

Peter,

Yes, I am trying to mimic "Generate Index". My client has a book with a large number of index entries. The index is taking about 4 hours to generate, and then it dies with an error message about an invalid character. I don't have the exact error message right now, but apparently this is a known issue with large indexes.

So my goal is to roll my own index generator. Right now I am using an array of objects for the index entries. Each object has two members: entry, which is the index entry itself; and pages, which is an array of page numbers. Once I collect all of the entries in the book, sort them, and sort and remove duplicates from the page numbers, I should be able to write the index.

Rick

Harbs,

If only I had remembered that UndoModes were introduced in CS4... Tried it in there and it works fine. Thanks very much for the pointer.

Peter

Rick,

Here's (a variant of) a script that I sometimes use to deal with indexes. Its output is an array of strings of topics followed by their page numbers. Topics and numbers are separated by // so you can find them easily. Subtopics are separated by #. So you get something like this:

Basque//1, 2, 3, 4
Romance#Italian//1, 2, 3, 4

Peter



#target indesign;

rough_index = manual_index (app.activeDocument);

function manual_index (doc) 
   { 
   var indexx = []; 
   t = app.activeDocument.indexes[0].allTopics; 
   for (i = 0; i < t.length; i++) 
      { 
      prefs = t[i].pageReferences.everyItem().getElements(); 
      if (prefs.length > 0) 
         { 
         topic = topic_path (t[i], t[i].name); 
         pages = find_pages (prefs); 
         indexx.push (topic + '//' + pages) 
         } 
      } 
   return indexx 
   }

function topic_path (top, str) 
   { 
   if (top.parent.constructor.name == 'Index') 
      return str; 
   else 
      return topic_path (top.parent, top.parent.name + '#' + str) 
   }

function find_pages (page_refs) 
   { 
   var temp = []; 
   for (var i = 0; i < page_refs.length; i++) 
      { 
      try {temp.push (page_refs[i].sourceText.parentTextFrames[0].parent.name)} 
      catch (_){} 
      } 
   return temp.join (', ') 
   } 

Hi Peter,

Thanks. I will have to go through your code and digest it. I have mine working at the book level, except for sorting the entries and writing the index to InDesign. Right now, it writes the index to a text file just so I can test the results, performance, etc.

Thank you very much for your help.

Rick