From 1220ea7e85229530341f66e572726c8cd34ced81 Mon Sep 17 00:00:00 2001
From: Carl Boettiger <cboettig@gmail.com>
Date: Thu, 28 Jun 2018 20:48:09 +0000
Subject: [PATCH] WIP

---
 data-raw/fishbase.R | 43 ++++++++++++++++++++++++++++++++++---------
 data-raw/gbif.R     |  9 ++++-----
 data-raw/itis.R     |  5 +++++
 data-raw/tpl.R      |  3 ++-
 data-raw/wd.R       |  6 ++++++
 5 files changed, 51 insertions(+), 15 deletions(-)
 create mode 100644 data-raw/wd.R

diff --git a/data-raw/fishbase.R b/data-raw/fishbase.R
index 46e982c..64ee127 100644
--- a/data-raw/fishbase.R
+++ b/data-raw/fishbase.R
@@ -1,4 +1,4 @@
-library(rfishbase)
+library(rfishbase) # 3.0
 library(tidyverse)
 
 fb <- as_tibble(rfishbase::load_taxa())
@@ -6,21 +6,46 @@ fb_wide <- fb %>%
   select( id = SpecCode, 
           genus = Genus, 
           species = Species, 
-          subfamily = SubFamily, 
+          subfamily = Subfamily, 
           family = Family,
           order = Order, 
           class = Class,
-          common_name = FBname) %>%
+          superclass = SuperClass) %>%
   mutate(phylum = "Chorodata", 
          kingdom = "Animalia",
-         id = paste0("FB:", id)) %>%
-  select(id, species, genus, subfamily, 
-          family, order, class, phylum,
-          kingdom, common_name)
+         id = paste0("FB:", id))
+
+write_tsv(fb_wide, "data/fb_hierarchy.tsv.bz2")
 
-write_tsv(fb_wide, "data/fb_wide.tsv.bz2")
+species <- rfishbase:::fb_species()
+synonyms <- rfishbase::synonyms(NULL) %>%
+  left_join(species) %>% 
+  rename(id = SpecCode)  %>% 
+  select(id,
+         species = Species,
+         synonym,
+         type = Status,
+         syn_id = SynCode,
+         rank = TaxonLevel, 
+         tsn = TSN, 
+         col = CoL_ID, 
+         worms = WoRMS_ID, 
+         zoobank = ZooBank_ID)
 
+## Consider preserving stock code?
+common <- rfishbase:::fb_tbl("comnames")  %>%
+  left_join(species) %>% 
+  rename(id = SpecCode)  %>% 
+  select(id, 
+         species = Species,
+         synonym = ComName,
+         language = Language) %>% 
+  mutate(type = "common") 
 
+fb_synonyms <- 
+common %>% 
+  bind_rows(synonyms)
+         
 
 slb <- as_tibble(rfishbase::load_taxa(server = "https://fishbase.ropensci.org/sealifebase"))
 slb_wide <- slb %>% 
@@ -36,7 +61,7 @@ slb_wide <- slb %>%
   select(id, species, genus, subfamily, 
          family, order, class, common_name)
 
-write_tsv(slb_wide, "data/slb_wide.tsv.bz2")
+write_tsv(slb_wide, "data/slb_hierarchy.tsv.bz2")
 
 
 
diff --git a/data-raw/gbif.R b/data-raw/gbif.R
index 9d99f12..2e8debe 100644
--- a/data-raw/gbif.R
+++ b/data-raw/gbif.R
@@ -25,16 +25,15 @@ gbif_wide <- gbif_taxa %>%
          kingdom, phylum, class, order, family, genus, species,
          specific_epithet = specificEpithet, infraspecific_epithet = infraspecificEpithet,
          genericName, taxonomicStatus)
-write_tsv(gbif_wide, "data/gbif_wide.tsv.bz2")
+write_tsv(gbif_wide, "data/gbif_hierarchy.tsv.bz2")
 rm(gbif_wide)
 
 gbif_long <- 
   gbif_taxa %>% 
-  select(taxon_id, name = scientificName, rank = rank, genericName, taxonomicStatus)
+  select(id = taxon_id, name = scientificName, rank = rank, genericName, taxonomicStatus)
 write_tsv(gbif_long, "data/gbif_long.tsv.bz2")
 
-
-write_tsv(gbif_taxa, "data/gbif.tsv.bz2")
-rm(gbif_taxa)
+gbif_long %>% select(id, name, rank) %>% distinct() %>%
+write_tsv("data/gbif_taxonid.tsv.bz2")
 
 
diff --git a/data-raw/itis.R b/data-raw/itis.R
index 3aa1c58..f241a61 100644
--- a/data-raw/itis.R
+++ b/data-raw/itis.R
@@ -117,12 +117,15 @@ system.time({
   write_tsv(itis_wide, bzfile("data/itis_wide.tsv.bz2", compression=9))
 })
 
+#########################################################
 
 ## Database prep for ITIS
 library(tidyverse)
 itis_long <- read_tsv("data/itis_long.tsv.bz2")
 itis_wide <- read_tsv("data/itis_wide.tsv.bz2")
 
+fs::file_move("data/itis_wide.tsv.bz2", "data/itis_hierarchy.tsv.bz2")
+
 ## accepted == valid
 ### https://www.itis.gov/submit_guidlines.html#usage
 
@@ -136,6 +139,8 @@ itis_taxonid <- taxonid %>% filter(name_usage %in% c("accepted", "valid")) %>% s
 ## assert ids are unique
 itis_taxonid %>% pull(id) %>% duplicated() %>% any() %>% testthat::expect_false()
 
+write_tsv(itis_taxonid, "data/itis_taxonid.tsv.bz2")
+
 
 itis_paths  <- itis_long %>% select(id, path_id, path, path_rank, path_rank_id) %>%
   distinct() %>% arrange(id)
diff --git a/data-raw/tpl.R b/data-raw/tpl.R
index eb20a21..2d2c766 100644
--- a/data-raw/tpl.R
+++ b/data-raw/tpl.R
@@ -29,7 +29,8 @@ write_tsv(tpl_wide, "data/tpl_wide.tsv.bz2")
 
 tpl_long <- tpl_wide %>% 
   gather(path_rank, path, -id, -kewid, -ipni_id, -confidence_level) %>%
-  left_join(select(tpl_wide, id, name=species)) %>% mutate(rank = "species") %>%
+  left_join(select(tpl_wide, id, name=species)) %>% 
+  mutate(rank = "species") %>%
   select(id, name, rank, path, path_rank, confidence_level, kew_id=kewid, ipni_id)
 
 write_tsv(tpl_long, "data/tpl_long.tsv.bz2")
diff --git a/data-raw/wd.R b/data-raw/wd.R
new file mode 100644
index 0000000..5a0aaa1
--- /dev/null
+++ b/data-raw/wd.R
@@ -0,0 +1,6 @@
+
+# WikiData
+# https://doi.org/10.5281/zenodo.1213476
+# 
+download.file("https://zenodo.org/record/1213477/files/wikidata-taxon-info20171227.tsv.gz", "wd-taxon.tsv.gz")
+download.file("https://zenodo.org/record/1213477/files/links-globi-wd-ott.tsv.gz", "wd-links.tsv.gz")
\ No newline at end of file