{"id":12216,"date":"2019-05-21T07:49:07","date_gmt":"2019-05-21T12:49:07","guid":{"rendered":"https:\/\/rud.is\/b\/?p=12216"},"modified":"2019-05-22T07:56:31","modified_gmt":"2019-05-22T12:56:31","slug":"add-dressbarn-to-the-continued-retailpocalypse","status":"publish","type":"post","link":"https:\/\/rud.is\/b\/2019\/05\/21\/add-dressbarn-to-the-continued-retailpocalypse\/","title":{"rendered":"Add Dressbarn to the Continued Retailpocalypse"},"content":{"rendered":"

I’ve talked about the retailpocalypse before<\/a> and this morning I was greeted with the news about Dressbarn closing all 650 stores<\/a> as I fired up a browser.<\/p>\n

I tweeted some pix and data but not everyone is on Twitter so I’m just posting a blog-blurb here with the code and data links.<\/p>\n

Code is below and at https:\/\/paste.sr.ht\/~hrbrmstr\/af6da1af0314426255c65bc2fc254e0abb2190c3<\/a>.<\/p>\n

Data is at https:\/\/rud.is\/dl\/dressbarn-locations.json.gz<\/a>.<\/p>\n

Images are in a gallery below the code.<\/p>\n

library(rvest)\nlibrary(stringi)\nlibrary(urltools)\nlibrary(worldtilegrid) # install from sh\/gl\/gh or just remove the theme_enhange_wtg() calls\nlibrary(statebins)\nlibrary(tidyverse)\n\n# this is the dressbarn locations directory page\npg <- read_html(\"https:\/\/locations.dressbarn.com\/\")\n\n# this is the selector to get the main links\nhtml_nodes(pg, \"a.Directory-listLink\") %>% \n  html_attr(\"href\") -> locs\n\n# PRE-NOTE\n# No sleep() code (I looked at the web site, saw how many self-requests it makes for all DB\n# resources and concluded that link scrapes + full page captures would not be burdensome\n# plus they're going out of business)\n\n# basic idea here is to get all the main state location pages\n# some states only have one store so the link goes right to that so handle that condition\n# for ones with multiple stores get all the links on the state index page\n# for links on state index page that have multiple stores in one area,\n# grab all those; then, concatenate all the final target store links into one \n# character vector.\n\nkeep(locs, ~nchar(.x) == 2) %>% \n  sprintf(\"https:\/\/locations.dressbarn.com\/%s\", .) %>% # state has multiple listings\n  map(\n    ~read_html(.x) %>% \n      html_nodes(\"a.Directory-listLink\") %>% \n      html_attr(\"href\") %>% \n      sprintf(\"https:\/\/locations.dressbarn.com\/%s\", .)\n  ) %>% \n  append(\n    keep(locs, ~nchar(.x) > 2) %>% sprintf(\"https:\/\/locations.dressbarn.com\/%s\", .) # state has one store\n  ) %>% \n  flatten_chr() %>% \n  map_if(\n    ~stri_count_fixed(.x, \"\/\") == 4, # 4 URL parts == there's another listing page layer\n    ~read_html(.x) %>% \n      html_nodes(\"a.Teaser-titleLink\") %>% \n      html_attr(\"href\") %>% \n      stri_replace_first_fixed(\"..\/\", \"\") %>% \n      sprintf(\"https:\/\/locations.dressbarn.com\/%s\", .)\n  )  %>% \n  flatten_chr() -> listings\n\n# make a tibble with the HTML source for the final store location pages\n# so we don't end up doing multiple retrievals\n\ntibble(\n  listing = listings,\n  html_src = map_chr(listings, ~httr::GET(.x) %>% httr::content(as = \"text\"))\n) -> dress_barn\n\n# save off our work in the event we have a (non-R-crashing) issue\ntf <- tempfile(fileext = \".rds\")\nprint(tf)\nsaveRDS(dress_barn, tf) \n\n# now, get data from the pages\n#\n# first, turn all the character vectors into something we can get HTML nodes from\n#\n# dressbarn web folks handliy put an \"uber\" link on each page so we get lon\/lat for free in that URL\n# they also handily used an <address> semantic tag in the proper PostalAddress schema format\n# so we can get locality and actual address, too\nmutate(\n  dress_barn,\n  parsed = map(html_src, read_html),\n  uber_link = \n    map_chr(\n      parsed, ~html_nodes(.x, xpath=\".\/\/a[contains(@href, 'uber')]\") %>% \n        html_attr(\"href\") \n    ), \n  locality = map_chr(\n    parsed, ~html_node(.x, xpath=\".\/\/address\/meta[@itemprop = 'addressLocality']\") %>% \n      html_attr(\"content\")\n  ),\n  address = map_chr(\n    parsed, ~html_node(.x, xpath=\".\/\/address\/meta[@itemprop = 'streetAddress']\") %>% \n      html_attr(\"content\")\n  ),\n  state = stri_match_first_regex(\n    dress_barn$listing, \n    \"https:\/\/locations.dressbarn.com\/([[:alpha:]]+)\/.*$\"\n  )[,2]\n) %>% \n  bind_cols(\n    param_get(.$uber_link, c(\"dropoff%5Blatitude%5D\", \"dropoff%5Blongitude%5D\")) %>% \n      as_tibble() %>% \n      set_names(c(\"lat\", \"lon\")) %>%\n      mutate_all(as.double)\n  ) -> dress_barn\n\n# save off our hard work with the HTML source so we can do more later if need be\nselect(dress_barn, -parsed) %>% \n  saveRDS(\"~\/Data\/dressbarn-with-src.rds\")\n\n# save off something others will want\nselect(dress_barn, -parsed, -html_src, -listing) %>% \n  jsonlite::toJSON() %>% \n  write_lines(\"~\/Data\/dressbarn-locations.json.gz\")\n\n# simple map\nggplot(dress_barn, aes(lon, lat)) + \n  geom_jitter(size = 0.25, color = ft_cols$yellow, alpha = 1\/2) +\n  coord_map(\"polyconic\") +\n  labs(\n    title = \"Locations of U.S. Dressbarn Stores\",\n    subtitle = \"All 650 locations closing\",\n    caption = \"Source: Dressbarn HTML store listings;\\nData: <https:\/\/rud.is\/dl\/dressbarn-locations.json.gz> via @hrbrmstr\"\n  ) +\n  theme_ft_rc(grid=\"\") +\n  theme_enhance_wtg()\n\nunlink(tf) # cleanup \n\ncount(dress_barn, state) %>% \n  left_join(tibble(name = state.name, state = tolower(state.abb))) %>% \n  left_join(usmap::statepop, by = c(\"name\"=\"full\")) %>% \n  mutate(per_capita = (n\/pop_2015) * 1000000) %>% \n  arrange(desc(per_capita)) %>% \n  select(name, n, per_capita) %>% \n  arrange(desc(per_capita)) %>% \n  complete(name = state.name) %>% \n  statebins(state_col = \"name\", value_col = \"per_capita\", ) +\n  labs(title = \"Dressbarn State per-capita closings\") +\n  theme_ipsum_rc(grid=\"\") +\n  theme_enhance_wtg()\n<\/code><\/pre>\n\n\t\t