Question

Project:

Take Wikipedia’s list of Roman consuls, put the data in a CSV so I can make a graph of the rise and fall of various gens in terms of consulage

Example data source:

509,L. Iunius Brutus,L. Tarquinius Collatinus
suff.,Sp. Lucretius Tricipitinus,P. Valerius Poplicola
suff.,M. Horatius Pulvillus,
508,P. Valerius Poplicola II,T. Lucretius Tricipitinus
507,P. Valerius Poplicola III,M. Horatius Pulvillus II

Vim search:

/\v(\d+|suff\.),((\w+\.=) (\w+)(\s\w+)=(\s\w+)=(\s[iv]+)=(\s\(.{-}\))=,=){,2}

So essentially:

1. Find the year at the beginning (or indication of suffect consul): (\d+|suff\.)
2. The next grouping (let’s call it the outer group) needs to be found up to two times: (outer group){,2}
3. For each of these two outer groups, find:
   1. Praenomen, with optional period (sometimes this isn’t present): (\w+.=)
   2. Nomen: (\w+)
   3. Optional cognomen (includes space, as do all below): (\s\w+)=
   4. Optional agnomen: (\s\w+)=
   5. Optional iteration (indicates the nth time he’s been consul). Data source does not have more than 8 iterations (so I and V will suffice): (\s[iv]+)=
   6. Optional explanatory note like “Sicinius (Sabinus?)”: (\s\(.{-}\))=

(Last comma is optional since it’s the end of the row.)

So the back references turn out to be:

\1: year or suffect
\2: the entire second outer group
\3: Praenomen of second outer group (same with all below)
\4: Nomen
\5: Cognomen
\6: Agnomen
\7: Iteration
\8: Explanatory note

The problem is I can’t figure out how to capture that first outer group. It’s like the \2 and \3-\8 references get overwritten when it sees that second outer group.

Using this replace:

:%s//1:{\1}^I2:{\2}^I3:{\3}^I4:{\4}^I5:{\5}^I6:{\6}^I7:{\7}^I8:{\8}^I9:{\9} 

I get this output:

1:{509} 2:{L. Tarquinius Collatinus}    3:{L.}  4:{Tarquinius}  5:{ Collatinus} 6:{}    7:{}    8:{}    9:{}
1:{suff.}   2:{P. Valerius Poplicola}   3:{P.}  4:{Valerius}    5:{ Poplicola}  6:{}    7:{}    8:{}    9:{}
1:{suff.}   2:{M. Horatius Pulvillus,}  3:{M.}  4:{Horatius}    5:{ Pulvillus}  6:{}    7:{}    8:{}    9:{}
1:{508} 2:{T. Lucretius Tricipitinus}   3:{T.}  4:{Lucretius}   5:{ Tricipitinus}   6:{ II} 7:{}    8:{}    9:{}
1:{507} 2:{M. Horatius Pulvillus II}    3:{M.}  4:{Horatius}    5:{ Pulvillus}  6:{ II} 7:{}    8:{}    9:{}

I can’t access those groups within the first outer group. I think they’re being overwritten: are they being overwritten? If so, is there a way around this?

Edit:
Original title
Vim regex (or any compatible regex): how to reference a group (within a group) if the outer group is iterated?

Answer 1

I’d break it down in substeps, employing vim functions instead of doing it all the normal (pun intended) way:

/\v(.{-}),(.{-}),(.*)

See what I did? made that much simpler and clearer

Edit Getting slightly less lazy, let’s define a helper function to split into a minimum of 3 substrings and tabseparate them:

function! Consul(s)        
    return join((split(a:s) + ["","",""])[0:2], "\t")
endf

Now reduce the substitution to (linebreaks for SO only)

%s/\v(.{-}),(.{-}),(.*)/\=join(
  [submatch(1), Consul(submatch(2)), Consul(submatch(3))], "\t")/g

Running that beauty on your input yields

509 L.  Iunius  Brutus  L.  Tarquinius  Collatinus
suff.   Sp. Lucretius   Tricipitinus    P.  Valerius    Poplicola
suff.   M.  Horatius    Pulvillus           
508 P.  Valerius    Poplicola   T.  Lucretius   Tricipitinus
507 P.  Valerius    Poplicola   M.  Horatius    Pulvillus

I’m pretty sure it will be a very easy step to further decorate the now neatly tab-separated columns to your liking. I might add it, but for now, here’s simplest thing I can think of:

:%s/\v(.{-})\t(.{-})\t(.{-})\t(.{-})\t(.{-})\t(.{-})\t(.{-})$/1:{\1}\t2:{\2}\t3:{\3}\t4:{\4}\t5:{\5}\t6:{\6}\t7:{\7}/g

Result:

1:{509} 2:{L.}  3:{Iunius}  4:{Brutus}  5:{L.}  6:{Tarquinius}  7:{Collatinus}
1:{suff.}   2:{Sp.} 3:{Lucretius}   4:{Tricipitinus}    5:{P.}  6:{Valerius}    7:{Poplicola}
1:{suff.}   2:{M.}  3:{Horatius}    4:{Pulvillus}   5:{}    6:{}    7:{}
1:{508} 2:{P.}  3:{Valerius}    4:{Poplicola}   5:{T.}  6:{Lucretius}   7:{Tricipitinus}
1:{507} 2:{P.}  3:{Valerius}    4:{Poplicola}   5:{M.}  6:{Horatius}    7:{Pulvillus}