spanish: R package for functions on spanish data

There are some special data in spain that is not addressed by any R package or function. Mainly on data processing and NLP. This package is my humble (and objectionable) attempt of helping programmers working with this kind of data creating some functions for it.

Documentation here: https://ropenspain.github.io/spanish/

Installation:

spanish is available on CRAN:

install.package("spanish")

Development version:

devtools::install_github("verajosemanuel/spanish")

to_number()

to_number() is a quick & dirty function to translate spanish spelled monetary quantities into their numerical counterparts. Given a numerical quantity spelled in spanish to_number translates it to integer.

to_number("dosmil ciento cuarenta y ocho")
[1] 2148

This function can be used on dataframes with lapply. Try the provided example dataframe (cantidades).

cantidades$var3 <- lapply(cantidades$var2, to_number)

head(cantidades[ , c("var2","var3")])
                                                                var2    var3
1                                                                DOS       2
2                                       CINCO MIL NOVECIENTOS VEINTE    5920
3 DOS MILLONES QUINIENTOS VEINTISIETEMIL DOSCIENTOS CUARENTA Y CINCO 2527245
4                   cientoveintisietemil cuatrocientos ochenta y dos  127482
5                               Dos mil cuatrocientos noventa y seis    2496
6                                                    dosmil cuarenta    2040

to_words()

to_words() is a quick & dirty (optimization pending) function to translate integers to spanish spelled monetary quantities. Given a numerical quantity to_words translates it to spanish.

to_words(3245235)
[1] "tres millones doscientos cuarenta y cincomil doscientos treinta y cinco"

This function can be used on dataframes with apply. Try random numbers.

a <- rnorm(n = 12,mean = 1000000,sd = 6000000)
a <- abs(a)  # non negative

> sapply(a, to_words)
 [1] "un millón setecientos ochenta y cuatromil quinientos diez"                 "cuatro millones setecientos sesenta y seismil novecientos noventa y nueve"
 [3] "cinco millones setecientos mil setecientos once"                           "cinco millones trescientos noventa y ochomil setecientos treinta y cuatro"
 [5] "setecientos sesenta y mil cuatrocientos  y cinco"                          "doscientos ochenta y seismil setecientos "                                
 [7] "un millón novecientos sesenta y sietemil cincuenta y tres"                 "ciento noventa y seismil novecientos treinta y seis"                      
 [9] "siete millones doscientos ochenta y ochomil cuatrocientos ochenta y cinco" "dos millones ciento  y sietemil quinientos cincuenta y siete"             
[11] "seis millones ochocientos ochenta y ochomil doscientos  y uno"             "seis millones sesentamil novecientos cuarenta y siete"                    

geocode_cadastral()

geocode_cadastral obtains longitude/latitude from valid cadastral references or kml files from catastro.

geocode_cadastral("0636105UF3403N", parse_files = FALSE)

[1] "36.5209422288168,-4.89298751473745"

Use lapply to geocode cadastral references from dataframe columns.

cadastral_references$new <- lapply(cadastral_references$cadref1, geocode_cadastral)

cadastral_references
         cadref1        cadref2                                new
1 2614501VK6621S 1449804NH1014N -3.44129716598736,40.3007548071216
2 2715807VK6621S 3880109CD6038S -3.44034755388475,40.3008975152619
3 2744702DF3824D 9501722DD6890B  2.19479121426678,41.4065056978248
4 0944328DF3804D 1617647TF8611F  2.17206636096108,41.4071068008502
5 8742806DF2884B 7893306CS7479S  2.14651391794721,41.4046077850113
6 9314903NH3591C 9701703XG8890B -8.51948430067999,42.9131469314961

separate generated geocode "new" into columns (lon/lat) usign tidyr

library(tidyr)

separate(cadastral_references, new, into = c('longitude','latitude'), sep = "," )

         cadref1        cadref2         longitude         latitude
1 2614501VK6621S 1449804NH1014N -3.44129716598736 40.3007548071216
2 2715807VK6621S 3880109CD6038S -3.44034755388475 40.3008975152619
3 2744702DF3824D 9501722DD6890B  2.19479121426678 41.4065056978248
4 0944328DF3804D 1617647TF8611F  2.17206636096108 41.4071068008502
5 8742806DF2884B 7893306CS7479S  2.14651391794721 41.4046077850113
6 9314903NH3591C 9701703XG8890B -8.51948430067999 42.9131469314961

When folder is source. We may process each kml the same way as remote URL.

 files <- list.files("folder", full.names = T)

 for (f in files) {
  coords <- geocode_cadastral(f, parse_files = TRUE)
  df <- as.data.frame(rbind(df , as.data.frame(coords, stringsAsFactors = F )))
 }

separate lat/lon into columns the same as before

 df <- tidyr::separate(coords, into = c("longitude","latitude"), sep = "," )
 

Requirements:

• magrittr must be installed.
• to_number() needs clean text. So it must be previously cleaned & removed extraneous words, symbols or cents.
• to_number() quantities MUST be written in a correct Spanish (this is not a grammar tool).
• to_words() does not keep sign. If negative numbers are provided, their positive is used for conversion.
• The upper limit is up to the millions range in both cases.
• geocode_cadastral() requests to catastro website. So notice if website is alive and working.

BE WARNED: You'll be banned if many requests to catastro are made in short time.

• Because of that geocode_cadastral() waits two seconds between requests.

TO DO

• Try to clean text prior to be used by to_number()
• Include decimals (cents) in to_number() function
• Force to_words() keep sign