mfdb_import_* to accept tibbles

[bthe]

I've stumble upon an error when trying to import tibbles (tidyverse's spin on data.frames) to mfdb.

Error: Must use a vector in `[`, not an object of class matrix.

At the moment the only work around seems to cast the tibble to a data.frame prior to import. I assume this has something to do with the R postgres package but I would find it cleaner that the mfdb_import_* functions would allow for tibbles in the import.

[lentinj]

Not necessarily, tibbles wrap data.frames then break backwards-compatibility, so there's likely to be sharp edges when using them with code not expecting them.

I had a quick go and did the commit above, in branch tibble-support, which fixes an obvious problem when you feed tibbles into mfdb_import_survey(). I'm not convinced it's related to the error above, but most of what mfdb_import_data.R does is on individual columns, not the table object as a whole.

If it isn't, what function are you feeding a tibble into? Could be I'm looking in the wrong place.

[bthe]

Here is an example of what I was doing (sorry for the lengthy example), the last bit doesn't work unless the data is stored as a data.frame

library(icesDatras)
library(mfdb)
library(tidyices)
library(tidyverse)


mdb <- mfdb('datras')


yrs <- 2017  # years
qs <- c(1)   # quarters

hh <-
  icesDatras::getDATRAS(record = "HH",
                        survey = "NS-IBTS",
                        years = yrs,
                        quarters = qs) %>%
  dtrs_tidy_hh()


hl <-
  icesDatras::getDATRAS(record = "HL",
                        survey = "NS-IBTS",
                        years = yrs,
                        quarters = qs) %>%
  dtrs_tidy_hl(hh)


ca <-
  icesDatras::getDATRAS(record = "CA",
                        survey = "NS-IBTS",
                        years = yrs,
                        quarters = qs) %>%
  dtrs_tidy_ca()


## Setup the spatial taxonomy
hh <- hh %>% 
  mutate(areacell = geo::d2sr(shootlat,shootlong))

hh %>% 
  select(areacell) %>% 
  distinct() %>% 
  mutate(id = 1:n(),
         lat = geo::sr2d(areacell)$lat,
         lon = geo::sr2d(areacell)$lon,
         area = geo::srA(areacell)) %>% 
  select(id, name=areacell,size = area) %>% 
  mfdb_import_area(mdb,.)
  
## Setup sampling types 

mfdb_import_sampling_type(mdb, data.frame(
  id = 1:3,
  name = c('NS-IBTS','SEA','LND'),
  description = c('North Sea IBTS','Catch samples','Landings')))

## upload tow taxonomy (optional)

# select(synis_id,ar,man,lat=kastad_n_breidd,lon=kastad_v_lengd,lat1=hift_n_breidd,lon1=hift_v_lengd,
# gear,sampling_type,depth=dypi_kastad,vessel,reitur,smareitur)

hh %>%
  select(sampling_type = survey,
         name = id, 
         year, 
         month = quarter,
         latitude = shootlat,
         longitude = shootlong,
         towlength = hauldur, 
         depth,
         vessel = ship) %>% 
  mfdb_import_tow_taxonomy(mdb,.)

## setup vessel taxonomy 

hh %>% 
  select(name = ship) %>% 
  distinct() %>% 
  mutate(length=NA,tonnage=NA,power=NA, full_name = name) %>% 
mfdb_import_vessel_taxonomy(mdb,.)


## import length data

hh %>%
  left_join(hl) %>% 
  select(sampling_type = survey,
         areacell,
         tow = id, 
         year, 
         month = quarter,
         latitude = shootlat,
         longitude = shootlong,
         towlength = hauldur, 
         depth,
         vessel = ship,
         latin, 
         length,
         sex, 
         count = n)%>% 
  mutate(latin = tolower(latin)) %>% 
  left_join(mfdb::species %>% 
              mutate(latin = gsub('.+\\((.+)\\)','\\1',description),
                     latin = tolower(latin)) %>% 
              rename(species = name)) %>%
  mutate(species = as.character(species)) %>% 
  select(-c(id,latin,description)) %>% 
  as.data.frame() %>% ## here I need to cast the tibble to data.frame
  mfdb_import_survey(mdb,.,data_source = 'DATRAS-lengths')

[lentinj]

No worries. I'm not quite sure where the matrix comes into it, but the above commit does fix this example, I'll merge into 6.x