Handling a Boolean query

makeboolean.R

library(stringr)
library(tidyverse)

#---------------------------------------------------------------
# https://twitter.com/nealhaddaway/status/1362512900716433409
#
# His example query
#--------------------------------------------------------------
q <- "((arable OR agricult* OR farm* OR crop* OR cultivat* OR field*) AND 
 (plough* OR plow* OR till* OR 'direct drill*' OR fertili* OR biosolid* OR 
 'bio solid' OR organic OR manur* or sewage OR compost* OR amendment* OR biochar* 
 OR digestate* OR 'crop residue*' OR 'crop straw*' OR mulch* OR 'crop rotat*' OR 'break crop*' 
 OR 'grass ley' OR 'clover ley' OR legume* OR 'bioenergy crop*' OR 'cover crop*' OR 'grass clover'
 OR 'cropping system*' OR 'crop system' OR 'winter crop*' OR 'spring crop*' OR 'summer fallow*' 
 OR 'catch crop*' OR intercrop* OR conservation) AND (CH4 OR methane OR CO2 OR 'carbon dioxide' 
 OR N2O OR 'nitrous oxide' OR GHG* OR 'greenhouse gas*' OR 'green-house gas*') AND (flux* OR dynamic* 
 OR emission* OR exchang* OR balanc*))"

#----------------------------------
# First, replace hyphens with a Å,
# ie make them word chars
#----------------------------------
q <- gsub("-", "Å", q)

#------------------------------------------------------------------
# Detect two-word terms aka phrases (within apostrophes), 
# and replace white space in them with a Ä, ie make them word chars
#
# https://stackoverflow.com/a/50205051
#------------------------------------------------------------------

repl <- regex("
  (?<=')  # positive lookbehind for an ' char
  \\w+    # one or more word chars
  (\\s?)  # optional white space
  \\w*    # optional word(s)
  \\*?    # optional ending * char          
              ", comments = TRUE)

replace_whitespace <- function(str) {
  str_replace_all(str, "\\s", "Ä")
}

q_whitespace_replaced <- q %>% 
  str_replace_all(repl, replace_whitespace)

word_chars <- regex("
  [\\w\\*]+    # one or more word characters or a * char
              ", comments = TRUE)

#------------------------
# Extract words to a list
#------------------------
words <- str_extract_all(q_whitespace_replaced, word_chars)

#---------------------------------------
# Detect OR clauses before the next AND
#---------------------------------------

do_q <- function(w, start, end) {
  
  l <- vector("list", end)
  
  stopped <- 1
  
  for ( i in start:end ) {
    if ( grepl("^AND", w[[1]][i]) ) {
      stopped <- i
      break 
    } else {
      l[i] <- w[[1]][i]
      names(l)[i] <- paste0("col", i)
    }
  }
  return_obj <- list(l, stopped)
  return(return_obj)
}

#-------------------------------
# Store all OR clauses in a list
#-------------------------------
res_list <- vector("list")

# First round
list_length <- length(words[[1]])
res <- do_q(words, 1, list_length) 
res_list[1] <- res[1]
next_list_item <- unlist(res[2])+1

# and the rest
i <- 1
while ( next_list_item != 2 ) {
  i <- i+1
  res <- do_q(words, next_list_item, list_length) 
  res_list[i] <- res[1]
  next_list_item <- unlist(res[2])+1
}  

#--------------------------------------------------------------------------------------------------
# Save only list items with a name "col*", i.e. those with and OR 
#
# https://community.rstudio.com/t/use-of-grep-to-sub-list-the-list-based-on-its-items-name-in-r/53798/2
#--------------------------------------------------------------------------------------------------
res_list_ors <- vector("list")

for ( i in 1:length(res_list) ){
  index_list <- grep("col.*", names(res_list[[i]]))
  res_list_ors[i] <- list(res_list[[i]][index_list])
}

#-------------------------------
# Construct the Boolean query
#-------------------------------
bstring <- ""

for ( i in 1:length(res_list_ors) ) {
  # Remove the help Ä char, and add apostrophes, ie restore phrases
  b_raw <- gsub("^([^X]+)Ä(.*)", "'\\1 \\2'", res_list_ors[[i]])
  # Replace the help Å char with the original hyphen
  b_raw <- gsub("Å", "-", b_raw)
  # Concat OR terms
  b <- paste(unlist(b_raw), collapse = " ")
  # Concat all ORs with AND
  bstring <- paste0(bstring, ") AND (", b)
}

#-------------------------------------------
# Clean: remove the leading AND,
# and wrap the whole string with parentheses
#-------------------------------------------
bstring <- gsub('^) AND ', '', bstring)
bstring <- paste0('(', bstring, '))')

(bstring)
# [1] "((arable OR agricult* OR farm* OR crop* OR cultivat* OR field*) AND 
# (plough* OR plow* OR till* OR 'direct drill*' OR fertili* OR biosolid* OR 'bio solid' 
# OR organic OR manur* or sewage OR compost* OR amendment* OR biochar* OR digestate* OR 'crop residue*' 
# OR 'crop straw*' OR mulch* OR 'crop rotat*' OR 'break crop*' OR 'grass ley' OR 'clover ley' OR legume* 
# OR 'bioenergy crop*' OR 'cover crop*' OR 'grass clover' OR 'cropping system*' OR 'crop system' OR 'winter crop*' 
# OR 'spring crop*' OR 'summer fallow*' OR 'catch crop*' OR intercrop* OR conservation) AND (CH4 OR methane OR CO2 
# OR 'carbon dioxide' OR N2O OR 'nitrous oxide' OR GHG* OR 'greenhouse gas*' OR 'green-house gas*') AND (flux* 
# OR dynamic* OR emission* OR exchang* OR balanc*))"