bool_detect.R

require(stringr)

bool_detect = function(x, qry, ignore_case = TRUE, print_call = FALSE){
  ops = c(AND = '&', `&` = '&', OR = '|', `|` = '|', `(` = '(', `)` = ')')  # boolean operators
  # add spaces around parentheses;
  qry = qry %>% str_trim %>% str_replace_all(fixed('('), '( ') %>% str_replace_all(fixed(')'), ' )') %>%
    str_replace_all("'", '"') %>% # enforce double quotes
    str_replace_all('(?<="[^ ]{0,100}) (?=[^ ]+")', '\u00A0') # sub spaces in quoted text for a unicode space
  # split elements by ascii space character
  comps = str_split(qry, '[ ]+')[[1]] %>% str_replace_all('\u00A0', ' ') %>% str_remove_all('"')
  comp_ops = comps %in% names(ops)  # identify operators from query terms
  # construct string for a composite logical call to evaluate
  to_negate = str_detect(comps[!comp_ops], '^-') # args to negate (NOT)
  qry_neg = comps[!comp_ops] %>% str_replace('^-', '')
  # build a str_detect command for each argument, negated as required
  comps[!comp_ops] = paste0(c('','!')[to_negate+1], 'str_detect(x, regex("',
                            qry_neg, '", ignore_case=', ignore_case,'))')
  comps[comp_ops] = ops[comps[comp_ops]]  # convert AND/OR to R logical operators
  qry_final = paste(comps, collapse = ' ')  # reduce to single string
  if(print_call) message(qry_final)
  eval(parse(text = qry_final))
}

# Example usage

#' # simple AND
#' sentences[ bool_detect(sentences, 'cat AND dog', print_call=T) ]
#' 
#' # simple OR
#' sentences[ bool_detect(sentences, 'cat OR dog', print_call=T) ]
#' 
#' # nested OR
#' sentences[ bool_detect(sentences, '(corner OR tree) AND (cat OR dog)') ]
#'
#' # quoted phrases
#' sentences[ bool_detect(sentences, '"the big"') ]
#'
#' # Report hoe query is interpreted 
#' x = bool_detect(sentences, 'cat OR dog', print_call=TRUE)
#'