Creating an url mapping when migrating from Wordpress to Hugo

url-mapping-disqus.R

library(xml2)
library(rvest)

doc <- read_html(x = "path_to_xml_file")

threads <- xml_find_all(doc, ".//thread")

old_links <- gsub(".* (http.*)", "\\1", xml_text(xml_find_all(threads, xpath = ".//id")))
old_titles <- xml_text(xml_find_all(threads, xpath = ".//title"))

old_titles <- iconv(enc2utf8(old_titles), "utf-8", "ascii//translit")
old_titles <- tolower(old_titles)
old_titles <- gsub("'", "", old_titles)
old_titles <- gsub("[[:punct:]]", "", old_titles)
old_titles <- gsub("  ", " ", old_titles)
old_titles <- gsub(" ", "-", old_titles)

# here change for you domain
new_links <- paste0("http://szychtawdanych.pl/post/", old_titles, "/")

mapping <- unique(data.frame(old_links, new_links))
write.csv(file = "szychta_url_mapping.csv", mapping, row.names = F, col.names = FALSE)

# more on migrating to Hugo at https://szychtawdanych.pl/post/przenoszenie-bloga-do-hugo/