Bundling Habib et al. 2020 to a DwC Archive

This is an R Markdown Notebook for converting the species checklist found in the following reference to DarwinCore format for upload into OBIS as part of UNESCO’s eDNA Expeditions project:

Habib, Kazi & Neogi, Amit & Oh, Jina & Nahar, Najmun & Kim, Choong-Gon & Lee, Youn-Ho. (2020). An overview of fishes of the Sundarbans, Bangladesh and their present conservation status. Journal of Threatened Taxa. 12. 15154–15172.

Setup

Call the necessary libraries and variables. Suppresses loading messages.

library(magrittr)                       # To use %<>% pipes
suppressMessages(library(janitor))      # To clean input data
suppressMessages(library(dplyr))        # To clean input data
library(stringr)                        # To clean input data
suppressMessages(library(rgnparser))    # To clean species names
suppressMessages(library(taxize))       # To get WoRMS IDs
library(worrms)                         # To get WoRMS IDs
library(digest)                         # To generate hashes
suppressMessages(library(obistools))    # To generate centroid lat/long and uncertainty
suppressMessages(library(sf))           # To generate wkt polygon
suppressMessages(library(EML))          # To create eml.xml file
library(xml2)                           # To create the meta.xml file
suppressMessages(library(zip))          # To zip DwC file

Input Parameters and Paths

path_to_project_root <- "../../.."
site_dir_name <- "the_sundarbans"
dataset_dir_name <- "Habib_et_al_2020"
original_pdf <- "ravichandra.pdf"
short_name <- "bangla-sundarbans-habib-2020"

Parsing PDF table to CSV

The data for this reference is formatted as a text-based table inside a PDF across multiple sheets. First, we use pdf_to_table to parse out the table to a CSV.

#conda environment
condaenv <- "mwhs-data-mobilization"

# Path to the Python script
script <- paste(path_to_project_root, "scripts_data/pdf_to_tables/pdf_to_table.py", sep="/")

# Input PDF file path
input_pdf <- paste(path_to_project_root, "datasets", site_dir_name, dataset_dir_name, "raw", original_pdf, sep="/")

# Output directory for OCR/table files
output_dir <- paste(path_to_project_root, "datasets", site_dir_name, dataset_dir_name, "processed", sep="/")

# Define page numbers and table areas (see documentation)
page_args <- c(
  "-a 123.148,268.99,737.77,378.372 -p 7",
  "-a 88.919,269.734,741.491,378.372 -p 8",
  "-a 88.919,270.479,728.097,379.116 -p 9",
  "-a 88.175,270.479,742.235,378.372 -p 10",
  "-a 88.175,269.734,740.002,378.372 -p 11",
  "-a 88.919,271.223,752.652,377.628 -p 12",
  "-a 88.175,271.223,745.211,377.628 -p 13",
  "-a 88.175,270.479,730.329,379.116 -p 14",
  "-a 87.431,269.734,588.951,376.884 -p 15"
)

# Define run parameters (see documentation)
run_parameters <- "-s -c"

# Combine page arguments and execute
page_args_combined <- paste(page_args, collapse = " ")
command <- paste("conda run -n", condaenv, "python", script, "-i", input_pdf, run_parameters, page_args_combined, "-o", output_dir)
system(command, intern=TRUE)

##  [1] ""                                                                                                                
##  [2] "Script Execution Summary"                                                                                        
##  [3] "Date and Time: 2023-09-12 13:24:17"                                                                              
##  [4] "------------------------------"                                                                                  
##  [5] ""                                                                                                                
##  [6] "PDF input: ../../../datasets/the_sundarbans/Habib_et_al_2020/raw/ravichandra.pdf"                                
##  [7] "Perform Table Parsing: TRUE"                                                                                     
##  [8] "Selected Areas:"                                                                                                 
##  [9] "  Area 1: [123.148, 268.99, 737.77, 378.372]"                                                                    
## [10] "  Area 2: [88.919, 269.734, 741.491, 378.372]"                                                                   
## [11] "  Area 3: [88.919, 270.479, 728.097, 379.116]"                                                                   
## [12] "  Area 4: [88.175, 270.479, 742.235, 378.372]"                                                                   
## [13] "  Area 5: [88.175, 269.734, 740.002, 378.372]"                                                                   
## [14] "  Area 6: [88.919, 271.223, 752.652, 377.628]"                                                                   
## [15] "  Area 7: [88.175, 271.223, 745.211, 377.628]"                                                                   
## [16] "  Area 8: [88.175, 270.479, 730.329, 379.116]"                                                                   
## [17] "  Area 9: [87.431, 269.734, 588.951, 376.884]"                                                                   
## [18] "Pages: 7, 8, 9, 10, 11, 12, 13, 14, 15"                                                                          
## [19] "Concatenate: True"                                                                                               
## [20] "Concatenate across headers: False"                                                                               
## [21] "Stream Extraction: True"                                                                                         
## [22] "Lattice Extraction: False"                                                                                       
## [23] ""                                                                                                                
## [24] "Parsing Tables"                                                                                                  
## [25] "------------------------------"                                                                                  
## [26] ""                                                                                                                
## [27] ""                                                                                                                
## [28] "Saving to CSV"                                                                                                   
## [29] "CSV file: ../../../datasets/the_sundarbans/Habib_et_al_2020/processed/ravichandra_tables_parsed_concatenated.csv"
## [30] "------------------------------"                                                                                  
## [31] ""                                                                                                                
## [32] ""                                                                                                                
## [33] "Run Details: ../../../datasets/the_sundarbans/Habib_et_al_2020/processed/ravichandra_parameters.txt"             
## [34] "Finished"                                                                                                        
## [35] ""

Read source data

Now we’ll read in the csv table outputted from the previous step

processed_csv <- "ravichandra_tables_parsed_concatenated.csv"

input_data <- read.csv(paste(path_to_project_root, "datasets", site_dir_name, dataset_dir_name, "processed", processed_csv, sep="/"))

#to preview pretty table
knitr::kable(head(input_data))

Scientific.name
Chiloscyllium griseum
Scoliodon laticaudus
Glyphis glyphis
Rhizoprionodon acutus
Carcharhinus melanopterus
Sphyrna lewini

Preprocessing

Here we tidy the data up, since OCR and table parsing errors are common and only take the list of species, since this is a checklist.

Data parsed from PDF is already pretty clean, so we don’t have to do much here.

Tidy Data

input_data %<>%
  remove_empty(c("rows", "cols")) %>%       # Remove empty rows and columns
  clean_names()                             # Have sensible (lowercase) column names

cleaned_data <- input_data

#to preview pretty table
knitr::kable(head(cleaned_data))

scientific_name
Chiloscyllium griseum
Scoliodon laticaudus
Glyphis glyphis
Rhizoprionodon acutus
Carcharhinus melanopterus
Sphyrna lewini

Get WoRMS IDs

Auto matching

First we will try to do this automatically by first cleaning the species names using gnparser and then using the taxise library to call the WoRMS database.

#Parse names
parsed_names <- rgnparser::gn_parse(cleaned_data[,])

#Function to get WoRMS IDs. Search for accepted names first and if not found, search for unaccepted. If still not found, use the worrms package to search.
get_worms_id_from_element <- function(element) {
  worms_id <- get_wormsid(element$canonical$full, searchtype="scientific", fuzzy=TRUE, messages = FALSE, accepted = TRUE)
  if (attr(worms_id, "match") == "not found") {
    worms_id <- get_wormsid(element$canonical$full, searchtype="scientific", messages = FALSE, fuzzy=TRUE)
    if (attr(worms_id, "match") == "not found") {
      worms_id <- NA
    }
  }
  return(worms_id)
}

#Call the function
worms_ids <- lapply(parsed_names, function(element) {
  if (element$parsed) {
    return(get_worms_id_from_element(element))
  } else {
    return(NA)
  }
})

#combine original names, parsed data and WoRMS ID into one data frame
combined_dataframe <- data.frame()

for (i in 1:nrow(cleaned_data)) {
  cleaned_value <- cleaned_data[i,]
  canonical_value <- parsed_names[[i]]$canonical$full
  worms_id_value <- worms_ids[[i]][1]
  if (is.null(canonical_value)){
    canonical_value <- NA
  }
  temp_row <- data.frame(CleanedData = cleaned_value, CanonicalFull = canonical_value, WormsIDs = worms_id_value)
  combined_dataframe <- rbind(combined_dataframe, temp_row)
}

knitr::kable(head(combined_dataframe))

CleanedData	CanonicalFull	WormsIDs
Chiloscyllium griseum	Chiloscyllium griseum	277829
Scoliodon laticaudus	Scoliodon laticaudus	217364
Glyphis glyphis	Glyphis glyphis	277184
Rhizoprionodon acutus	Rhizoprionodon acutus	105802
Carcharhinus melanopterus	Carcharhinus melanopterus	105795
Sphyrna lewini	Sphyrna lewini	105816

Human Verification

Sometimes there are misspellings in the original text or incorrect OCR that can be searched for and fixed by hand. To do this, view the combined dataframe, search for unmatched species in WoRMS and add the ID, and remove rows that were not autoremoved in the earlier cleaning steps

#Run this command to view the dataframe and inspect for missing IDs
#View(combined_dataframe)

combined_dataframe[35,2:3] = c("Anguilla bengalensis", 217456)
combined_dataframe[68,2:3] = c("Puntius chola", 1026489)
combined_dataframe[69,2:3] = c("Puntius terio", 1026600)
combined_dataframe[77,2:3] = c("Mystus tengara", 1022781)
combined_dataframe[78,2:3] = c("Mystus bleekeri", 1020266)
combined_dataframe[177,2:3] = c("Uranoscopus inermis", 305463)
combined_dataframe[201,2:3] = c("Otolithoides pama", 281950)
combined_dataframe[231,2:3] = c("Sicamugil cascasia", 1022773)
combined_dataframe[248,2:3] = c("Pogonogobius planifrons", 309788)
combined_dataframe[260,2:3] = c("Parambassis baculis", 1022774)
combined_dataframe[313,2:3] = c("Chelonodontops bengalensis", NA)
combined_dataframe[316,2:3] = c("Carinotetraodon travancoricus", 1014801)

Darwin Core mapping

Required Terms

OBIS currently has eight required DwC terms: scientificName, scientificNameID, occurrenceID, eventDate, decimalLongitude, decimalLatitude, occurrenceStatus, basisOfRecord.

scientificName/scientificNameID

Create a dataframe with unique taxa only (though this should already be unique). This will be our primary DarwinCore data frame.

#rename and restructure WoRMSIDs to OBIS requirements
occurrence <- combined_dataframe %>%
  distinct(CanonicalFull, WormsIDs) %>%
  rename(scientificName = CanonicalFull) %>%
  rename(scientificNameID = WormsIDs) %>%
  mutate(scientificNameID = ifelse(!is.na(scientificNameID), paste("urn:lsid:marinespecies.org:taxname:", scientificNameID, sep = ""), NA))

occurrenceID

OccurrenceID is an identifier for the occurrence record and should be persistent and globally unique. It is a combination of dataset-shortname:occurrence: and a hash based on the scientific name.

# Vectorize the digest function (The digest() function isn't vectorized. So if you pass in a vector, you get one value for the whole vector rather than a digest for each element of the vector):
vdigest <- Vectorize(digest)

# Generate taxonID:
occurrence %<>% mutate(occurrenceID = paste(short_name, "occurrence", vdigest (scientificName, algo="md5"), sep=":"))

eventDate

This is NULL since this is technically a checklist and we do not know the collection date.

eventDate <- ""
occurrence %<>% mutate(eventDate)

decimalLongitude/decimalLatitude

Use obistools::calculate_centroid to calculate a centroid and radius for WKT strings. This is useful for populating decimalLongitude, decimalLatitude and coordinateUncertaintyInMeters. The WKT strings are from https://github.com/iobis/mwhs-shapes.

if (!file.exists(paste(path_to_project_root, "scripts_data/marine_world_heritage.gpkg", sep="/"))) {
  download.file("https://github.com/iobis/mwhs-shapes/blob/master/output/marine_world_heritage.gpkg?raw=true", paste(path_to_project_root, "scripts_data/marine_world_heritage.gpkg", sep="/"))
}

shapes <- st_read(paste(path_to_project_root, "scripts_data/marine_world_heritage.gpkg", sep="/"))

## Reading layer `marine_world_heritage' from data source 
##   `/mnt/c/Users/Chandra Earl/Desktop/Labs/UNESCO/mwhs-data-mobilization/scripts_data/marine_world_heritage.gpkg' 
##   using driver `GPKG'
## Simple feature collection with 60 features and 4 fields
## Geometry type: MULTIPOLYGON
## Dimension:     XY
## Bounding box:  xmin: -180 ymin: -55.32282 xmax: 180 ymax: 71.81381
## Geodetic CRS:  4326

#For some sites, the GeoPackage has core as well as buffer areas. Merge the geometries by site.
shapes_processed <- shapes %>%
  group_by(name) %>%
  summarize()

#Sundarbans National Park
ind_shape <- shapes_processed$geom[which(shapes_processed$name == "Sundarbans National Park")]


#convert shape to WKT
wkt <- st_as_text(ind_shape, digits = 6)

localities <- calculate_centroid(wkt)

occurrence %<>% mutate(decimalLatitude = localities$decimalLatitude)
occurrence %<>% mutate(decimalLongitude = localities$decimalLongitude)

occurrenceStatus

occurrenceStatus <- "present"
occurrence %<>% mutate(occurrenceStatus)

basisOfRecord

basisOfRecord <- "HumanObservation"
occurrence %<>% mutate(basisOfRecord)

Extra Terms

footprintWKT

occurrence %<>% mutate(footprintWKT = wkt)

coordinateUncertaintyInMeters

occurrence %<>% mutate(coordinateUncertaintyInMeters = localities$coordinateUncertaintyInMeters)

geodeticDatum

geodeticDatum <- "WGS84"
occurrence %<>% mutate(geodeticDatum)

country

country <- "Bangladesh"
occurrence %<>% mutate(country)

locality

locality <- "Bangla Sundarbans"
occurrence %<>% mutate(locality)

Post-processing

Check data

Use the check_fields command from obistools to check if all OBIS required fields are present in an occurrence table and if any values are missing.

#Reorganize columns
occurrence = occurrence %>% select(occurrenceID, scientificName, scientificNameID, eventDate, country, locality, decimalLatitude, decimalLongitude, coordinateUncertaintyInMeters, footprintWKT, geodeticDatum, occurrenceStatus, basisOfRecord)

#Check fields
check_fields(occurrence)

## Warning: `data_frame()` was deprecated in tibble 1.1.0.
## ℹ Please use `tibble()` instead.
## ℹ The deprecated feature was likely used in the obistools package.
##   Please report the issue to the authors.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.

## # A tibble: 317 × 4
##    level field       row message                                 
##    <chr> <chr>     <int> <chr>                                   
##  1 error eventDate     1 Empty value for required field eventDate
##  2 error eventDate     2 Empty value for required field eventDate
##  3 error eventDate     3 Empty value for required field eventDate
##  4 error eventDate     4 Empty value for required field eventDate
##  5 error eventDate     5 Empty value for required field eventDate
##  6 error eventDate     6 Empty value for required field eventDate
##  7 error eventDate     7 Empty value for required field eventDate
##  8 error eventDate     8 Empty value for required field eventDate
##  9 error eventDate     9 Empty value for required field eventDate
## 10 error eventDate    10 Empty value for required field eventDate
## # ℹ 307 more rows

Create the EML file

This is a file which contains the dataset’s metadata and is required in a DarwinCore-Archive.

emld::eml_version("eml-2.1.1")

## [1] "eml-2.1.1"

#Title
title <- "An overview of fishes of the Sundarbans, Bangladesh and their present conservation status: Fishes Checklist"

#AlternateIdentifier
alternateIdentifier <- paste("https://ipt.obis.org/secretariat/resource?r=", short_name, sep="")

#Abstract
abstract <- eml$abstract(
  para = "Sundarbans, the largest mangrove forest of the world is located in Bangladesh and India. Studies done on the diversity of fish fauna in the Sundarbans mangrove forest of Bangladesh are sparse and patchy. Here we take the opportunity to provide an updated checklist of the fishes of the Sundarbans, Bangladesh based on primary and secondary data. Field surveys were undertaken in the aquatic habitat of Sundarbans core area along with its adjacent marine habitat from June 2015 to July 2017. Based on published information and primary observations the updated list of fishes covers a total of 322 species belonging to 217 genera, 96 families, and 22 orders. Additionally, four species of fishes, are newly reported in Bangladesh waters, viz., Mustelus mosis Hemprich & Ehrenberg, 1899; Lagocephalus guentheri Miranda Ribeiro, 1915; Carangoides hedlandensis Whitley, 1934; Uranoscopus cognatus Cantor, 1849. The global IUCN Red List status of each species has been enlisted. The updated checklist will constitute the reference inventory of fish biodiversity for the Sundarbans, a natural world heritage site."
)

People

Here we add the people involved in the project:

The creator is the person or organization responsible for creating the resource itself.

The contact is the person or institution to contact with questions about the use, interpretation of a data set.

The metadataProvider is the person responsible for providing the metadata documentation for the resource.

The associatedParty (in this case the Data Curator) is the person who mobilized the data from the original resource.

creator <- list(eml$creator(
    individualName = eml$individualName(
      givenName = "Kazi Ahsan", 
      surName = "Habib"),
    electronicMailAddress = "ahsan.sau@gmail.com",
    organizationName = "Sher-e-Bangla Agricultural University"
  ), eml$creator(
    individualName = eml$individualName(
      givenName = "Amit Kumer", 
      surName = "Neogi"),
    electronicMailAddress = "neogi3710@gmail.com",
    organizationName = "Sher-e-Bangla Agricultural University"
  ), eml$creator(
    individualName = eml$individualName(
      givenName = "Najmun", 
      surName = "Nahar"),
    electronicMailAddress = "naharnajmun887@gmail.com",
    organizationName = "Sher-e-Bangla Agricultural University"
  ), eml$creator(
    individualName = eml$individualName(
      givenName = "Jina", 
      surName = "Oh"),
    electronicMailAddress = "jnoh@kiost.ac.kr",
    organizationName = "Korea Institute of Ocean Science and Technology"
  ), eml$creator(
    individualName = eml$individualName(
      givenName = "Youn-Ho", 
      surName = "Lee"),
    electronicMailAddress = "ylee@kiost.ac",
    organizationName = "Korea Institute of Ocean Science and Technology"
  ), eml$creator(
    individualName = eml$individualName(
      givenName = "Choong-Gon", 
      surName = "Kim"),
    electronicMailAddress = "kimcg@kiost.ac.kr",
    organizationName = "Korea Institute of Ocean Science and Technology"
  )
)

contact <- eml$creator(
  individualName = eml$individualName(
    givenName = "OBIS", 
    surName = "Secretariat"),
  electronicMailAddress = "helpdesk@obis.org",
  organizationName = "OBIS",
  positionName = "Secretariat"
)

metadataProvider <- eml$metadataProvider(
  individualName = eml$individualName(
    givenName = "Chandra", 
    surName = "Earl"),
  electronicMailAddress = "c.earl@unesco.org",
  organizationName = "UNESCO",
  positionName = "eDNA Scientific Officer"
)

associatedParty <- eml$associatedParty(
  role = "processor",
  individualName = eml$individualName(
    givenName = "Chandra", 
    surName = "Earl"),
  electronicMailAddress = "c.earl@unesco.org",
  organizationName = "UNESCO",
  positionName = "eDNA Scientific Officer"
)

Additional Metadata

Here we add the additionalMetadata element, which is required for a GBIF-type EML file and contains information such as the citation of the dataset, the citation of the original resource and the creation timestamp of the EML.

#{dataset.authors} ({dataset.pubDate}) {dataset.title}. [Version {dataset.version}]. {organization.title}. {dataset.type} Dataset {dataset.doi}, {dataset.url}

additionalMetadata <- eml$additionalMetadata(
  metadata = list(
    gbif = list(
      dateStamp = paste0(format(Sys.time(), "%Y-%m-%dT%H:%M:%OS3"), paste0(substr(format(Sys.time(), "%z"), 1, 3), ":", paste0(substr(format(Sys.time(), "%z"), 4, 5)))),
      hierarchyLevel = "dataset",
      citation = "IPT will autogenerate this",
      bibliography = list(
        citation = "Habib KA, Neogi AN, Nahar N, Oh J and Lee Y. (2020). An overview of fishes of the Sundarbans, Bangladesh and their present conservation status. Journal of Threatened Taxa. 12. 15154–15172.")
    )
  )
)

citationdoi <- "https://doi.org/10.11609/jott.4893.12.1.15154-15172"

Coverage

Here we describe the dataset’s geographic, taxonomic and temporal coverage.

#Coverage
coverage <- eml$coverage(
  geographicCoverage = eml$geographicCoverage(
    geographicDescription = "Indian Sundarbans",
    boundingCoordinates = eml$boundingCoordinates(
      westBoundingCoordinate = st_bbox(ind_shape)$xmax,
      eastBoundingCoordinate = st_bbox(ind_shape)$xmin,
      northBoundingCoordinate = st_bbox(ind_shape)$ymax,
      southBoundingCoordinate = st_bbox(ind_shape)$ymin)
    ),
  taxonomicCoverage = eml$taxonomicCoverage(
    generalTaxonomicCoverage = "Fishes",
    taxonomicClassification = list(
      eml$taxonomicClassification(
        taxonRankName = "Superclass",
        taxonRankValue = "Agnatha"),
      eml$taxonomicClassification(
        taxonRankName = "unranked",
        taxonRankValue = "Chondrichthyes"),
      eml$taxonomicClassification(
        taxonRankName = "unranked",
        taxonRankValue = "Osteichthyes")
      )
  ),
  temporalCoverage = eml$temporalCoverage(
    rangeOfDates = eml$rangeOfDates(
      beginDate = eml$beginDate(
        calendarDate = "2015-06-01"
      ),
      endDate = eml$endDate(
        calendarDate = "2017-07-31"
      )
    )
  )
)

Extra MetaData

These fields are not required, though they make the metadata more complete.

methods <- eml$methods(
  methodStep = eml$methodStep(
    description = eml$description(
      para = paste("See Github <a href=\"https://github.com/iobis/mwhs-data-mobilization\">Project</a> and <a href=\"https://iobis.github.io/mwhs-data-mobilization/notebooks/", site_dir_name, "/", dataset_dir_name, "\"> R Notebook</a> for dataset construction methods", sep="")
    )
  )
)


#publication date of dataset
pubDate <- "2023-10-15"

#language of original document
language <- "eng"

keywordSet <- eml$keywordSet(
  keyword = "Occurrence",
  keywordThesaurus = "GBIF Dataset Type Vocabulary: http://rs.gbif.org/vocabulary/gbif/dataset_type_2015-07-10.xml"
)

maintenance <- eml$maintenance(
  description = eml$description(
    para = ""),
  maintenanceUpdateFrequency = "notPlanned"
)

#Universal CC
intellectualRights <- eml$intellectualRights(
  para = "To the extent possible under law, the publisher has waived all rights to these data and has dedicated them to the <ulink url=\"http://creativecommons.org/publicdomain/zero/1.0/legalcode\"><citetitle>Public Domain (CC0 1.0)</citetitle></ulink>. Users may copy, modify, distribute and use the work, including for commercial purposes, without restriction."
)

purpose <- eml$purpose(
  para = "These data were made accessible through UNESCO's eDNA Expeditions project to mobilize available marine species and occurrence datasets from World Heritage Sites."
)

additionalInfo <- eml$additionalInfo(
  para = "marine, harvested by iOBIS"
)

Create and Validate EML

#Put it all together
my_eml <- eml$eml(
           packageId = paste("https://ipt.obis.org/secretariat/resource?id=", short_name, "/v1.0", sep = ""),  
           system = "http://gbif.org",
           scope = "system",
           dataset = eml$dataset(
               alternateIdentifier = alternateIdentifier,
               title = title,
               creator = creator,
               metadataProvider = metadataProvider,
               associatedParty = associatedParty,
               pubDate = pubDate,
               coverage = coverage,
               language = language,
               abstract = abstract,
               keywordSet = keywordSet,
               contact = contact,
               methods = methods,
               intellectualRights = intellectualRights,
               purpose = purpose,
               maintenance = maintenance,
               additionalInfo = additionalInfo),
           additionalMetadata = additionalMetadata
)

eml_validate(my_eml)

## [1] TRUE
## attr(,"errors")
## character(0)

Create meta.xml file

This is a file which describes the archive and data file structure and is required in a DarwinCore-Archive. It is based on the template file “meta_occurrence_checklist_template.xml”

meta_template <- paste(path_to_project_root, "scripts_data/meta_occurrence_checklist_template.xml", sep="/")
meta <- read_xml(meta_template)

fields <- xml_find_all(meta, "//d1:field")

for (field in fields) {
  term <- xml_attr(field, "term")
  if (term == "http://rs.tdwg.org/dwc/terms/eventDate") {
    xml_set_attr(field, "default", eventDate)
  } else if (term == "http://rs.tdwg.org/dwc/terms/country") {
    xml_set_attr(field, "default", country)
  } else if (term == "http://rs.tdwg.org/dwc/terms/locality") {
    xml_set_attr(field, "default", locality)
  } else if (term == "http://rs.tdwg.org/dwc/terms/decimalLatitude") {
    xml_set_attr(field, "default", localities$decimalLatitude)
  } else if (term == "http://rs.tdwg.org/dwc/terms/decimalLongitude") {
    xml_set_attr(field, "default", localities$decimalLongitude)
  } else if (term == "http://rs.tdwg.org/dwc/terms/coordinateUncertaintyInMeters") {
    xml_set_attr(field, "default", localities$coordinateUncertaintyInMeters)
  } else if (term == "http://rs.tdwg.org/dwc/terms/footprintWKT") {
    xml_set_attr(field, "default", wkt)
  } else if (term == "http://rs.tdwg.org/dwc/terms/geodeticDatum") {
    xml_set_attr(field, "default", geodeticDatum)
  } else if (term == "http://rs.tdwg.org/dwc/terms/occurrenceStatus") {
    xml_set_attr(field, "default", occurrenceStatus)
  } else if (term == "http://rs.tdwg.org/dwc/terms/basisOfRecord") {
    xml_set_attr(field, "default", basisOfRecord)
  }
}

Save outputs

dwc_output_dir <- paste(path_to_project_root, "output", site_dir_name, dataset_dir_name, sep="/")

write.csv(occurrence, paste(dwc_output_dir, "/occurrence.csv", sep = ""), na = "", row.names = FALSE)
write_xml(meta, file = paste(dwc_output_dir, "/meta.xml", sep = ""))
write_eml(my_eml, paste(dwc_output_dir, "/eml.xml", sep = ""))

Edit EML

We have to further edit the eml file to conform to GBIF-specific requirements that cannot be included in the original EML construction. This includes changing the schemaLocation and rearranging the GBIF element, since the construction automatically arranges the children nodes to alphabetical order.

#edit the schemaLocation and rearrange gbif node for gbif specific eml file
eml_content <- read_xml(paste(dwc_output_dir, "/eml.xml", sep = ""))

#change schemaLocation attributes for GBIF
root_node <- xml_root(eml_content)
xml_set_attr(root_node, "xsi:schemaLocation", "https://eml.ecoinformatics.org/eml-2.1.1 http://rs.gbif.org/schema/eml-gbif-profile/1.2/eml.xsd")
xml_set_attr(root_node, "xmlns:dc", "http://purl.org/dc/terms/")
xml_set_attr(root_node, "xmlns:stmml", NULL)
xml_set_attr(root_node, "xml:lang", "eng")


#rearrange children nodes under the GBIF element
hierarchyLevel <- eml_content %>% xml_find_all(".//hierarchyLevel")
dateStamp <- eml_content %>% xml_find_all(".//dateStamp")
citation <- eml_content %>% xml_find_all("./additionalMetadata/metadata/gbif/citation")
bibcitation <- eml_content %>% xml_find_all("./additionalMetadata/metadata/gbif/bibliography/citation")
xml_set_attr(bibcitation, "identifier", citationdoi)

eml_content %>% xml_find_all(".//hierarchyLevel") %>% xml_remove()
eml_content %>% xml_find_all(".//dateStamp") %>% xml_remove()
eml_content %>% xml_find_all("./additionalMetadata/metadata/gbif/citation") %>% xml_remove()
eml_content %>% xml_find_all(".//gbif") %>% xml_add_child(citation, .where=0)
eml_content %>% xml_find_all(".//gbif") %>% xml_add_child(hierarchyLevel, .where=0)
eml_content %>% xml_find_all(".//gbif") %>% xml_add_child(dateStamp, .where=0)

write_xml(eml_content, paste(dwc_output_dir, "/eml.xml", sep = ""))

Zip files to DwC-A

output_zip <- paste(dwc_output_dir, "DwC-A.zip", sep="/")

if (file.exists(output_zip)) {
  unlink(output_zip)
}

file_paths <- list.files(dwc_output_dir, full.names = TRUE)
zip(zipfile = output_zip, files = file_paths, mode = "cherry-pick")

if (file.exists(output_zip)) {
  unlink(file_paths)
}