!Edit search/replace

I should like !Edit’s search/replace facility to be able to store the ten most recently used search/replace strings rather than just one. I should also like to be able to load a script file that would populate these ‘last ten’ so that I could apply them one by one.

Ideally I would like !Edit to be able to process a script file containing a list of successive search/replace strings. I frequently have the need to do the following set of search/replace operations, with the end-of-file replace option:

[09][0D]$ to [09]\ [0D]$ - replace blank R.H. cell with hard space
<br>$ to <br> - delete [CTRL]-ENTER after <br> (repeat until none found)
[0D]$ to \\ - remember end of full line as '\'
$ to <space> - replace [CTRL]-ENTER with space other than end of line
\\ to $ - replace end of full line

Ideally I would like !Edit to be able to process a script file containing a list of successive search/replace strings.

I think we may have been here before, but I think you are basically asking for StrongED.
A small quote from the StrongED helpfile:

InsertFile (filename)
#Parent !ref_function
This function inserts the given file into the text. Is not currently used by any StrongED modes, but could be used by modes to insert common phrases into the text. Modes should use it in conjunction with the () function.

Example:

#Wrap off; Indent 2; fcode
SetTmp() InsertFile(“\<StrongED$Tmp_ModePath\>.SomeFile”)

I cannot get my head around StrongEd’s search/replace syntax though to get an exactly equivalent search/replace to the one I describe for !Edit.

I cannot get my head around StrongEd’s search/replace syntax though to get an exactly equivalent search/replace to the one I describe for !Edit.

Sorry, I wasn’t aware that Edit had that complex a search and replace facility.
Just looked and still can’t find it.

A brief comparison (by me) of Edit vs StrongED searching can be found here

If you need any further help, just ask.

That’s very helpful. I’ll try it when I have a moment.

I would like !Edit to be able to process a script file containing a list of successive search/replace strings.

This script

#!lua
local replace = { {from_1, to_1}, ... , {from_n, to_n} }
for line in io.lines (arg[1]) do
  for _, x in ipairs (replace) do
    line = line:gsub (x[1], x[2])
  end -- for
  print (line)
end -- for

Gavin’s Lua script is one way of automating this with StrongED. Another is to use named search/replace expressions and then use them in the Replace function.

For this you need to add the required definitions to a ModeFile. If we call your list of replacements S1,R1 … S5,R5 then you’d have to make the following additions to an appropriate mode (either an existing mode or a newly created one). Search, Replace and Functions are sections in a ModeFile.

Search
   S1   \t \r \n
   S2   "<br>" \n
   S3   \r \n
   S4   \n
   S5   "\\"
End

Replace
   R1   \t "&nbsp;" \r \n
   R2   "<br>"
   R3   "\\"
   R4   " "
   R5   \n
End

Functions
   Key      <keydef>
   Icon     <sprite>
   Menu     <item text>
   Help     <help text>
   Select   Replace(S1,R1,,,) ... Replace(S5,R5,,,)
End

You can then execute the list of replacements by using the associated key/icon/menu. It might seem a bit cryptic right now but once you’ve done this a few times it becomes very easy to add this kind of functionality through ModeFiles.

Chris:

I should like !Edit’s search/replace facility to be able to store the ten most recently used search/replace strings rather than just one. I should also like to be able to load a script file that would populate these ‘last ten’ so that I could apply them one by one.

Or you could just use a different application that can process search and replace [and various other] operations from a script. A fugly script language, but that’s another matter. ;)

I don’t know the Edit syntax (having not touched edit since around the time the dinosaurs died out), so I can’t be 100% sure – but I’d hope it can do the specific search and replace tasks you need. If not, you know the developer.

Looking at your list and trying to remember, though:

[09][0D]$ to [09]\ [0D]$ - replace blank R.H. cell with hard space

Would become:

Replace.ChangeFrom = "|i|m|j"
Replace.ChangeTo = "|i |m|j"
Content.Replace

I’m not sure why you’re calling that a “blank R.H. cell” though – it appears to be a tab followed by an alien end of line – and you aren’t replacing it with a non-breaking space, you’re inserting a non-breaking space between the tab and the line ending.

<br>$ to <br> - delete [CTRL]-ENTER after <br> (repeat until none found)

That’s a more difficult one. The basic search and replace is easy enough:

Replace.ChangeFrom = "<br>|j"
Replace.ChangeTo = "<br>"
Content.Replace

The problem is making it repeat until it doesn’t find any – there’s currently no mechanism in WebChange to do that. Should be possible to add, though.

[0D]$ to \\ - remember end of full line as '\'

That would be:

Replace.ChangeFrom = "|m|j"
Replace.ChangeTo = "\"
Content.Replace

I’m curious as to how a line with a carriage return as well as a line feed is a “full line” though, and one without (next task) is presumably not. Something to do with how the file(s) are generated?

$ to <space> - replace [CTRL]-ENTER with space other than end of line

That’s:

Replace.ChangeFrom = "|j"
Replace.ChangeTo = " "
Content.Replace

And finally, [sort of] reversing the one before the last:

\\ to $ - replace end of full line

Is:

Replace.ChangeFrom = "\"
Replace.ChangeTo = "|j"
Content.Replace

The full script, then, with some added comments and conveniently ignoring the need for conditional repetition on the 2nd:

WebChange.Script

; None of what follows uses wildcards so let's switch 'em off for efficiency
Replace.Wildcards = "No"

; Case insensitivity is also unnecessary - so a bit more efficiency
Replace.CaseSensitive = "Yes"

; Insert a non-breaking space between a tab and CRLF line ending
Replace.ChangeFrom = "|i|m|j"
Replace.ChangeTo = "|i |m|j"
Content.Replace

; Get rid of linefeeds after <br> tags - the bit that needs to be conditionally repeated
Replace.ChangeFrom = "<br>|j"
Replace.ChangeTo = "<br>"
Content.Replace

; 'hide' CRLF line endings from the next search and replace
Replace.ChangeFrom = "|m|j"
Replace.ChangeTo = "\"
Content.Replace

; Replace linefeeds with spaces
Replace.ChangeFrom = "|j"
Replace.ChangeTo = " "
Content.Replace

; now put those original line endings back - but as LFs only
Replace.ChangeFrom = "\"
Replace.ChangeTo = "|j"
Content.Replace

Since part of the process effectively changes those CRLF line endings with LF only, it’s a pity you need to change LF-only line endings with spaces first – WebChange currently provides no mechanism to replace something conditionally in the way that’s needed. Otherwise we could reduce the number of steps, and change one search/replace with a different function.

Edit: while I was correctly turning your [0D] into |m in each case, I was forgetting in the case of $ (i.e. [0A]) and using |a instead of |j – do’h!

Also, I’m not 100% sure OTTOMH it’s “Yes”/“No” for the settings. It might be “On”/“Off”.

Perhaps I do need a manual after all ;)

Steve:

Sorry, I wasn’t aware that Edit had that complex a search and replace facility.
Just looked and still can’t find it.

Go to its “Find” menu item, and in the dialogue select the “Magic characters” option.

Since part of the process effectively changes those CRLF line endings with LF only, it’s a pity you need to change LF-only line endings with spaces first – WebChange currently provides no mechanism to replace something conditionally in the way that’s needed.

That’s he sort of search & replace where you drop in a temporary alternate string for one string containing a specific substring and then do the replace on the remaining substring instances then replace the temporary string with the intended target string.

I’m not sure why you’re calling that a “blank R.H. cell” though

The source file is a TAB-separated, no-quotes csv file saved by OpenOfficeCalc (Excel 97 can’t cope with this). If the rightmost cell is blank rather than space, this makes sure it is a (non-breaking) space (for later processing). CR’s entered in a cell (to make it readable) appear as LF, end of line as CRLF. Deliberate ‘new lines’ are entered in a cell as “<br>”.

The full, gory detail is here and I have written it up for Archive magazine. I used to produce a book relying on IE6 to format a rather complex HTML table, printed by Acrobat Distiller. I decided to justify the text and render the pages myself as IE kept getting updated and (in Windows-speak) improved.

That’s he sort of search & replace where you drop in a temporary alternate string for one string containing a specific substring and then do the replace on the remaining substring instances then replace the temporary string with the intended target string.

Quite so – and that’s what Chris’ original sequence does in Edit, and my script therefore does in WebChange.

It would be feasible for me to add a couple of variables to the search and replace function (one to specify a string that needs to be protected and another to define the protection string) which would therefore allow the final three steps to be rolled into one in the script but in reality WebChange would still be performing three search and replace operations, so it wouldn’t really add much – other than to make the script itself shorter.

The source file is a TAB-separated, no-quotes csv file saved by OpenOfficeCalc

Ah, okay – that explains not only the inserted non-breaking space, but also the reason for the mixed line endings.