paste lines of string characters in raw text - regex

Question

I work on raw textual data from a scanned catalog.
Here is an example:
ABADIE-LANDEL (Pierre) — 1920 — né à Paris. — 17, rue Campagne-Première
ABOU (Albert) — 1930 — né à Marseille.
— 41, rue de Seine, 6e.
ANGER (Jacques) — 1925 — né à Paris. — 33, rue Vineuse, 16e.
ANTHONE (Armand) — 1908 — né à Paris. — 4, avenue Victor-Hugo
Rue des Tournelles
ANTRAL (Jean) — 1920
This is a list of names with occasional lines including address mentions.
The data is imported into R with:
readlines ("clipboard", encoding = " latin1 ")
I am able to identify lines including artist names in capital letters with different regex
[A-ZÁÀÂÄÃÅÇÉÈÊËÍÌÎÏÑÓÒÔÖÕÚÙÛÜÝYÆO][A-ZÁÀÂÄÃÅÇÉÈÊËÍÌÎÏÑÓÒÔÖÕÚÙÛÜÝYÆO |']
or (ICU)
[\p{Uppercase Letter}][\p{Uppercase Letter}|']
I am able to identify lines including artworks
^[0-9]+[\s][^bis]`
I am able to extract artists names
".+(?=- [0-9]{4})"
or
(.+)[0-9]{4}.+ # with backreference \1
For more data, here is a sample of data from a 1930 catalog:
https://docs.google.com/document/d/1nF3CQmZbDsCGKMp_OgZymxWIfoOx5xrNdTmDXZANwuc/edit?usp=sharing
I wish I could paste the pieces of adress substrings but My final goal is to create a data.frame object structured as follows:
1st column: NAME artist and surname;
2nd column: supplements (address, nationality ...)
3rd columns: works or better ...
Column 3: 1 work
4th column 2 work, etc.
Thank you in advance for your help.

Answer 1

If I understand your question correctly, you want to extract names and addresses from your records some of which may span across different lines.
One solution may be to exploit the fact that the character — works as a field separator. So assuming that the structure of your records is regular you could do:
(data is a variable holding your example string)
## Replace newlines with the separator character
data <- gsub("\\n(\\s*—)?", " — ", data)
## Normalize space
data <- gsub("\\s+", " ", data)
## Now split by the separator character
tokens <- strsplit(data, "\\s—\\s")[[1]]
tokens now contains:
[1] "ABADIE-LANDEL (Pierre)" "1920" "né à Paris." "17, rue Campagne-Première" "ABOU (Albert)"
[6] "1930" "né à Marseille." "41, rue de Seine, 6e." "ANGER (Jacques)" "1925"
[11] "né à Paris." "33, rue Vineuse, 16e." "ANTHONE (Armand)" "1908" "né à Paris."
[16] "4, avenue Victor-Hugo" "Rue des Tournelles" "ANTRAL (Jean)" "1920"
Each complete record should have 4 sequential indices in this vector but since there may be incomplete records we must work a little more.
We exploit the fact that people's names are all-capitals and follow a strict pattern. We get the indices of the names in tokens and then split tokens on those indices. Each subvector produced is now a complete record:
## Get the indices of names
idx <- which(grepl("^[A-Z-]+\\s\\(", tokens))
## Use the indices to partition tokens to subvectors
records <- list()
for (i in 1:length(idx)) {
start <- idx[i]
if (i == length(idx)) {
stop <- length(tokens)
}
else {
stop <- idx[i+1] - 1
}
records[[i]] <- tokens[start:stop]
}
Here is the final list of results:
[[1]]
[1] "ABADIE-LANDEL (Pierre)" "1920" "né à Paris." "17, rue Campagne-Première"
[[2]]
[1] "ABOU (Albert)" "1930" "né à Marseille." "41, rue de Seine, 6e."
[[3]]
[1] "ANGER (Jacques)" "1925" "né à Paris." "33, rue Vineuse, 16e."
[[4]]
[1] "ANTHONE (Armand)" "1908" "né à Paris." "4, avenue Victor-Hugo" "Rue des Tournelles"
[[5]]
[1] "ANTRAL (Jean)" "1920"
Hope this helps or leads to better ideas.