R/populate_layers.r

In OHI-Science/ohirepos: Create OHI tailored repositories

## populate_layers.r

#' Populate new OHI repos with layers from global assessments
#'
#' @param repo_registry data frame with information about the repo
#' @param gh_org GitHub organization, defaults to 'OHI-Science'
#' @param multi_nation T/F whether to pull information from multiple nations (i.e. Baltic, Arctic)
#'
#' @return repo
#' @export
#'
#' @examples
#'
#'
populate_layers <- function(repo_registry,
                            multi_nation = FALSE){

  ## create variables
  key           <- repo_registry$study_key
  study_area    <- repo_registry$study_area
  dir_repo      <- repo_registry$dir_repo
  dir_origin    <- repo_registry$dir_origin
  lyrs_origin   <- readr::read_csv(file.path(dir_origin, 'layers.csv'))
  dir_scenario  <- file.path(dir_repo, repo_registry$scenario_name)
  dir_shp_out   <- repo_registry$dir_shp_out
  gh_org        <- repo_registry$gh_org


  ## clone repo master branch
  unlink(dir_repo, recursive=TRUE, force=TRUE)
  repo <- ohirepos::clone_repo(dir_repo,
                               sprintf('https://github.com/%s/%s.git',
                                       gh_org, key))

  ## modify layers dataframe
  lyrs_key = lyrs_origin %>%
    select(
      targets, layer, filename, fld_value, units,
      name, description) %>%
    mutate(
      layer_gl = layer,
      path_in  = file.path(dir_origin, 'layers', filename),
      rgns_in  = 'global',              ## TODO long term: don't think we need this rgns_in field
      filename = sprintf('%s_%s.csv', layer, repo_registry$suffix_origin)) %>%
    arrange(targets, layer)

  ## swap out rgn_area in directory; save a copy of the file
  lyr_area <- 'rgn_area'
  csv      <- 'rgn_offshore_data.csv'
  ix       <- which(lyrs_key$layer == lyr_area)
  lyrs_key$rgns_in[ix]  <-  'subcountry'
  # lyrs_key$path_in[ix]  <-  file.path(dir_shp_out, csv)
  lyrs_key$filename[ix] <-  sprintf('%s.csv', lyr_area)

  ## drop cntry_* layers ## TODO July 2017 delete?
  lyrs_key = filter(lyrs_key, !grepl('^cntry_', layer))

  ## drop all layers no longer being used (especially LE) ## TODO July 2017 JSL clean this up. 117 -> 102
  lyrs_le_rm = c(
    'le_gdp_pc_ppp',
    'le_jobs_cur_adj_value',
    'le_jobs_cur_base_value',
    'le_jobs_ref_adj_value',
    'le_jobs_ref_base_value',
    'le_rev_cur_adj_value',
    'le_rev_cur_base_value',
    'le_rev_cur_base_value',
    'le_rev_ref_adj_value',
    'le_rev_ref_base_value',
    'le_rev_sector_year',
    'le_revenue_adj',
    'le_wage_cur_adj_value',
    'le_wage_cur_base_value',
    'le_wage_ref_adj_value',
    'le_wage_ref_base_value',
    'liveco_status',
    'liveco_trend',
    'cntry_rgn',
    'cntry_georegions',

    "t_average_visitors",
    "t_visitor_gdp")
  lyrs_key <- filter(lyrs_key, !layer %in% lyrs_le_rm)


  ## match OHI+ regions to global regions ----
  rgns_key <- read_csv(file.path(dir_scenario, 'spatial/regions_list.csv')) %>%
    select(rgn_id, rgn_name) %>%
    mutate(rgn_id_origin   = repo_registry$rgn_id_global,
           rgn_name_origin = repo_registry$rgn_name_global)


  ## setup for copying layers over
  dir.create(sprintf('%s/layers', dir_scenario), showWarnings=FALSE)

  # ## copy layers one by one, saving differently if multi_nation
  if (!multi_nation) {

    #   ## old global to new custom countries
    #   if (dim(rlist)[1] != dim(rgns_key)[1]) { # make sure Guayaquil doesn't match to both ECU and Galapagos
    #     sc_cntries = subset(sc_studies, sc_key == key, gl_rgn_key, drop=T)
    #     rgns_key <- rgns_key %>%
    #       filter(cntry_key %in% sc_cntries)
    #   }

    ## element layers
    elements <- c('element_wts_cp_km2_x_protection',
                  'element_wts_cs_km2_x_storage',
                  'element_wts_hab_pres_abs')

    ## for each layer (not multi_nation)...
    for (lyr in lyrs_key$layer){ # lyr = "ao_access_mhi2017"   lyr = 'hd_subtidal_hb'  lyr = 'rgn_global' lyr = 'rgn_labels'

      if ( lyr %in% elements ) {

        ## copy elements files directly
        readr::read_csv(lyrs_key$path_in[lyrs_key$layer == lyr]) %>%
          readr::write_csv(sprintf('%s/layers/%s', dir_scenario,
                                   lyrs_key$filename[lyrs_key$layer == lyr]), na="")

      } else {

        ## call copy_layer and write to layer to csv
        d <- ohirepos::copy_layer(lyr,
                                  rgns_key,
                                  dir_origin,
                                  dir_scenario,
                                  lyrs_key,
                                  write_to_csv = TRUE)

        ## update filename (will change if placeholder)
        lyrs_key$filename[lyrs_key$layer == lyr] <- basename(d)

      }

    }

  } else { # multi_nation == TRUE

    ## overwrite rgns_key if multi_nation
    sc_rgns_lookup <- read.csv(sprintf('~/github/ohi-webapps/custom/%s/sc_rgns_lookup.csv', key))
    rgns_key = sc_rgns_lookup %>%
      merge(
        gl_rgns,
        by='gl_rgn_name', all.x=T) %>%
      select(rgn_id, sc_rgn_name, rgn_id_global, gl_rgn_name) %>%
      arrange(sc_rgn_name)

    ## old global to multi_nation
    rgns_key = rgns_key %>%
      group_by(cntry_key) %>%
      filter(row_number() == 1) %>%
      ungroup()
    if (dim(rlist)[1] != dim(rgns_key)[1]) { # so GYE doesn't match both ECU+Galapagos
      rgns_key = rgns_key %>%
        filter(cntry_key %in% unique(sc_rgns_lookup$gl_rgn_key))
    }

    ## for each layer...(multi_nation)
    for (lyr in lyrs_key$layer){ # lyr = "ao_access"

      ## call copy_layer and then write to layer to csv as separate step
      d <- ohirepos::copy_layer(lyr, rgns_key,
                                dir_origin, repo_registry$suffix_origin,
                                lyrs_key, write_to_csv = FALSE)

      if ('rgn_id' %in% names(d))    d <- d %>% arrange(rgn_id)
      if ('region_id' %in% names(d)) d <- d %>% arrange(region_id)

      ## update filename (will change if placeholder)
      lyrs_key$filename[lyrs_key$layer == lyr] <- basename(d)

      ## write to csv as separate step
      csv_out = d
      #TODO update this so data and filename are returned
      write_csv(d, csv_out)

    }
  } ## end if (!multi_nation)


  ## create layers.csv registry ----
  lyrs_reg = lyrs_key %>%
    select(
      targets,
      layer,
      filename,
      fld_value,
      units,
      name,
      description,
      layer_gl,
      path_in)

  ## save layers.csv
  layers_csv <- sprintf('%s/layers.csv', dir_scenario)
  readr::write_csv(lyrs_reg, layers_csv, na='')

  ## check for empty layers
  ohicore::CheckLayers(layers_csv, file.path(dir_scenario, 'layers'),
                       flds_id=c('rgn_id','country_id','saup_id','fao_id','fao_saup_id'))

  return(repo)

}