Add 2023 tax bills

data-raw/sample_tax_bills/sample_tax_bills_detail.R

@@ -26,9 +26,11 @@
 
 # Different tax bills can have different table sizes depending on the number of
 # taxing district.
-extract_tax_bill <- function(file) {
+
+extract_tax_bill <- function(file){
   base_file <- basename(file)
-  tbl <- pdf_text(file)[[1]] %>%
+  tbl <- pdf_text(file) %>%
+    paste(., collapse='\n') %>%
     str_extract(., regex("MISCELLANEOUS TAXES.*", dotall = TRUE)) %>%
     str_split(., "\n") %>%
     unlist() %>%
@@ -46,12 +48,12 @@
       agency_name = str_squish(agency_name),
       flag = is.na(prev_tax),
       prev_tax = if_else(flag,
-        pension,
-        prev_tax
+                         pension,
+                         prev_tax
       ),
       pension = if_else(flag,
-        NA,
-        pension
+                        NA,
+                        pension
       )
     ) %>%
     select(-flag) %>%
@@ -61,7 +63,8 @@
         agency_name,
         paste0(
           "TAXES|Assess|Property|EAV|Local Tax|",
-          "Total Tax|Do not|Equalizer|cookcountyclerk.com"
+          "Total Tax|Do not|Equalizer|cookcountyclerk.com|",
+          "Pursuant|meaning of|If paying later|\\d{15}+|By \\d{2}/"
         )
       )
     )
@@ -76,6 +79,7 @@
   return(out)
 }
 
+
 # Collect all scanned tables + meta data in a data frame
 bills <- map(list_pdf_inputs, extract_tax_bill)
 bills_df <- bind_rows(bills)