clean up royal 92

Author: smasongarrison

data-raw/df_royal92.R

@@ -5,7 +5,12 @@ library(readr)
 library(usethis)
 library(BGmisc)
 
-
+date_qualifier_regex <- "\\b(?:A[BF]T|BE[TF])\\b\\s*"
+text_cleanup_regex <- c(
+  "/|\\(twin\\)" = "",
+  "_" = " ",
+  "\\s*-\\s*" = "-"
+)
 
 ## Create dataframe
 royal92 <- df_raw <- readGedcom("data-raw/royal92.ged")
@@ -16,40 +21,131 @@ royal92 <- ped2fam(royal92, personID = "personID") %>%
     -name_surn,
     -FAMC,
     -FAMS
-  ) %>%
+  )  %>%
+  addPersonToPed(
+    personID = 1147,
+    name = "Henry de Montfort",
+    sex = "M",
+    momID = 1370,
+    dadID = 873,
+    overwrite = TRUE
+  )  %>%  addPersonToPed(
+    personID = 3011,
+    name = "Simon de Montfort the Younger",
+    sex = "M",
+    momID = 1370,
+    dadID = 873
+  )
+
+
+royal92_cleaned <- royal92 %>%
   mutate(
     momID = as.numeric(momID),
     dadID = as.numeric(dadID),
-    name = str_remove(name, "/"),
-    aproximated_dob = case_when(
-      str_detect(birth_date, "ABT|BET|AFT|BEF") == TRUE ~ TRUE,
+    approximated_dob = case_when(
+      str_detect(birth_date, date_qualifier_regex) ~ TRUE,
       str_length(birth_date) == 4 ~ TRUE,
       TRUE ~ FALSE
     ),
-    aproximated_dod = case_when(
-      str_detect(death_date, "ABT|BET|AFT|BEF") == TRUE ~ TRUE,
+    approximated_dod = case_when(
+      str_detect(death_date, date_qualifier_regex) ~ TRUE,
       str_length(death_date) == 4 ~ TRUE,
       TRUE ~ FALSE
     ),
-    birth_date = str_replace_all(birth_date, "ABT|BET|AFT|BEF", ""),
-    death_date = str_replace_all(death_date, "ABT|BET|AFT|BEF", ""),
+    birth_date = str_replace_all(
+      birth_date, date_qualifier_regex, "") %>%
+      str_squish(),
+    death_date = str_replace_all(
+      death_date, date_qualifier_regex, "") %>%
+      str_squish(),
+    # if only year is given, assign 15th June as the date
     birth_date = case_when(
+      personID == 1147 ~ "15 NOV 1238",
+      personID == 1149 ~ "7 OCT 1816",
+      personID == 2456 ~ "15 JUN 1070",
+      personID == 2985 ~ "26 APR 1924",
+      personID == 3011 ~ "15 APR 1240",
       str_length(birth_date) == 0 ~ NA_character_,
       str_length(birth_date) == 4 ~ paste0("15 JUN ", birth_date),
       TRUE ~ birth_date
-    ),
-    birth_date = as.Date(str_trim(birth_date), format = "%d %b %Y"),
+    )  %>%
+      str_trim() %>%
+      as.Date(format = "%d %b %Y"),
+    # if only year is given, assign 15th June as the date
     death_date = case_when(
+      personID == 1147 ~ "4 AUG 1265",
+      personID == 1149 ~ "12 APR 1817",
+      personID == 2456 ~ "14 FEB 1117",
+      personID == 2985 ~ "14 DEC 1997",
+      personID == 3011 ~ "15 JUN 1271",
       str_length(death_date) == 0 ~ NA_character_,
       str_length(death_date) == 4 ~ paste0("15 JUN ", death_date),
       TRUE ~ death_date
-    ),
-    death_date = as.Date(str_trim(death_date), format = "%d %b %Y"),
+    ) %>%
+      str_trim() %>%
+      as.Date(format = "%d %b %Y"),
     name = case_when(
+      personID == 12 ~ "Alexandra of Denmark (Alix)",
+      personID == 27 ~  "Victoria Eugenie (Ena)",
+      personID == 39  ~ "Alexandra Fedorovna (Alix)",
+      personID == 41 ~ "Dagmar (Marie) of Denmark",
+      personID == 84 ~ "Elizabeth (Ella)",
+      personID == 85 ~ "Mary (May)",
+      personID == 136 ~  "Mary Adelaide (Fat Mary)",
+      personID == 155 ~ "Michael (Mischa) Alexandrovich Romanov",
+      personID == 785 ~ "Richard Curzon-Howe",
+      personID == 788 ~ "James Hamilton",
+      personID == 1197 ~ "Karl Theodor (Gackl)",
+      personID == 1200 ~ "Sophie Charlotte Auguste",
+      personID == 1442 ~ "Ferdinand Philippe Marie d'Orléans",
+      personID == 1594 ~ "John IV (the Conqueror) of Montfort",
+      personID == 1709 ~ "Henry Somerset",
+      personID == 2851 ~ "Ernest Frederick III of Saxe-Hildburghausen",
       personID == 2944 ~ "William Scott of Buccleuch Montagu-Douglas",
-      TRUE ~ str_replace_all(name, "_", " ")
-    )
+      personID == 2990 ~ "William Legge",
+      personID == 2991 ~ "Rupert Legge",
+      personID == 2992 ~ "Charlotte Legge",
+      personID == 2993 ~ "Henry Legge",
+
+      TRUE ~ name %>%
+        str_replace_all(text_cleanup_regex) %>%
+        str_squish()
+    ),
+    twinID = case_when(
+     personID == 223 ~ 222,
+     personID == 222 ~ 223,
+     personID == 1116 ~ 1117,
+     personID == 1117 ~ 1116,
+     personID == 1155 ~ 1156,
+     personID == 1156 ~ 1155,
+     TRUE ~ NA_real_
+  ),
+  attribute_title = attribute_title %>%
+      str_replace_all(text_cleanup_regex) %>%
+      str_squish(),
+  sex = case_when(
+    personID %in% c(1098,
+                    1753,
+                    1755,
+                    1756,
+                    1803,
+                    2033,
+                    2509,
+                    2990,
+                    2991,
+                    2993
+                    ) ~ "M",
+    personID %in% c(1149,
+                    2992
+                    ) ~ "F",
+    TRUE ~ sex
   )
+  )
+
+
+
+royal92 <- royal92_cleaned %>%
+  select(-approximated_dob, -approximated_dod)
 # "22 AUG 1485"
 # if missing momID or dadID, assign the next available ID
 
@@ -68,7 +164,6 @@ checkis_acyclic <- checkPedigreeNetwork(royal92,
 checkis_acyclic
 if (checkis_acyclic$is_acyclic) {
   message("The pedigree is acyclic.")
-
   write_csv(royal92, here("data-raw", "royal92.csv"))
   usethis::use_data(royal92, overwrite = TRUE, compress = "xz")
 } else {