Alan’s python script is more powerful !

Indeed, especially when one notices in the data that some of the references come before the definitions!

No problem. Let’s give it a new try !

From the initial lines :

===<span id='X382'>X382</span>===
1841 England - Census transcript - John COOMBE - Household

The regex S/R, described in my previous post, changed it as below :

===<span id='1841Census'>X382</span>===
1841 England - Census transcript - John COOMBE - Household

But, may be, you would expect the following result, where the string X382 is changed, both, inside the single quotes and outside :

===<span id='1841Census'>1841Census</span>===
1841 England - Census transcript - John COOMBE - Household

If so, use this new regex S/R, below :

SEARCH (?-si)(===<span id=).+?(?=</span>.+\R(\d+))

REPLACE \1'\2Census'>\2Census

Which can be used, indifferently, with the Replace or the Replace All buttons !

Cheers,

guy038

P.S. :

Of course, the Alan’s python script is more powerful !

Maybe something like this:

search_repl_pairs_list = []
editor.research(r"(?-s)===<span id='(\D\d\d\d).+\R(\d{4})", lambda m: search_repl_pairs_list.append((m.group(1), m.group(2))))
for tup in search_repl_pairs_list: editor.replace(tup[0], tup[1] + 'Census')

You have more “logic” in your problem statement than a regular expression can handle, I’m afraid.

In such cases you should probably turn to a scripting plugin, e.g. Pythonscript, that can work with regular expression data, but can also incorporate more logic into it.

Ah… I’m sorry ! I should have examined your text more carefully. No trouble, there’s a solution, anyway !

If you’ll use the Replace All button, exclusively, here is the right regex S/R, which may be used, either, in a macro :

SEARCH (?-si)===<span id='\K.+?(?='.+\R(\d+))

REPLACE \1Census

Notes :

• The (?-si) in-line modifiers forces the regex engine :

  • To consider any dot . symbol as matching a single standard char, and not any EOL character

  • To process the S/R in a non-insensitive way

• Then, the ===<span id=' matches the identical string ===<span id=’

• The special \K syntax resets the match process and the regex engine position

• Therefore, the part .+? matches the shortest range of standard characters… ( our string X### )

• With the condition, due to the look-ahead structure (?=.........), that it must be followed with :

  • A single quote and some standard characters '.+

  • Followed with a line-break \R of current line

  • Followed with some digits characters, stored as group 1, due to the embedded parentheses (\d+)

• In the replacement \1Census, the match ( string X### ) is replaced with the number, located at beginning of next line, followed with the string Census

Now, if you want to see, at once, the result of each step by step replacement, use this alternate syntax :

SEARCH (?-si)(===<span id=').+?(?='.+\R(\d+))

REPLACE \1\2Census

Notes :

• The \K syntax is not present. So, the literal string ===<span id=’ is embedded, itself, in parentheses as the group 1 (===<span id=') and will be re-used in replacement. And, the (\d+) represents the group 2

• In replacement, it first rewrites the beginning of current line \1, followed with the number \2, at beginning of the 2nd line

Cheers,

guy038

You said :

e.g. In the following text I want to Replace all occurrences of X382 by 1841Census and all occurrences of X391 by 1861Census

If so, the correct regex syntax which will process all values in one go, is :

SEARCH (?-i)\bX3((82)|91)\b

REPLACE 18(?{2}4:6)1Census

Notes :

• The part (?-i) forces a non-insensitive search

• The part X3 looks for the string X3, with that exact case

• The part ((82)|91) means that the string X3 must be followed with, either, the number 82 or the number 91

• The inner parentheses represents the group 2. So, if it matches the string X391, then group 2 is not defined

• The two assertions \b forces the string X… to be surrounded with non-word chars for matching

• In replacement :

  • It first writes the string 18

  • According to group 2, the part (?{2}4:6) rewrites digit 4 or digit 6

  • it finally writes ths string 1Census

Of course, if you need to run this regex S/R very often, it would be sensible to record this S/R in a macro, and use it with a shortcut ;-))

Best Regards,

guy038