functions.R

# Loading libraries -------------------------------------------------------
if (!require(XLConnect)) {
  install.packages("XLConnect")
  library(XLConnect)
}

if (!require(ggrepel)){
  install.packages("ggrepel")
  library(ggrepel)
}

if (!require(gridExtra)){
  install.packages("gridExtra")
  library(gridExtra)
}

if (!require(readxl)){
  install.packages("readxl")
  library(readxl)
}

if (!require(tidyverse)) {
  install.packages("tidyverse")
  library(tidyverse)
}

if (!require(stringr)) {
  install.packages("stringr")
  library(stringr)
}
if (!require(stringi)){
  install.packages("stringi")
  library(stringi)
}

if (!require(ggmap)){
  install.packages("ggmap")
  library(ggmap)
}

if (!require(maps)){
  install.packages("maps")
  library(maps)
}

if (!require(scales)){
  install.packages("scales")
  library(scales)
}

if (!require(forcats)){
  install.packages("forcats")
  library(forcats)
}

if (!require(googlesheets)){
  install.packages("googlesheets")
  library(googlesheets)
}

# Defining functions ------------------------------------------------------

# Extract a theme's questions from a dataframe --------------------------------------------
## Questions around a central themem usually share the theme's name, and then the question 
## itself is placed between square parenthesis. This function takes the theme's name, 
## and removes it from the column names of the input dataframe
theme_extract <- function(dataframe, theme){
  x <- dataframe %>%
    select(contains(theme))
  names(x) <- names(x) %>%
    str_replace(".* \\[", "") %>%
    str_replace("\\]", "")
  x
}


# Plot a theme ------------------------------------------------------------
## This function plots the columns corresponding to a certain theme, typically extracted 
## using  the theme_extract function, and shows them as bar graph
theme_plot <- function(dataframe, fill_by = NULL){
  if (is.null(fill_by)) {
    g <- vector("list", length(dataframe))
    for (i in seq_along(dataframe)){
      g[[i]] <- dataframe %>% plot_bars( col = names(.)[i+1])
      g[[i]]
    }
  } else {
    g <- vector("list", length(dataframe)-1)
    for (i in seq(1, length(dataframe)-1)){ 
      g[[i]] <- dataframe %>% plot_bars( col = names(.)[i+1], fill = fill_by)
      print(i)
      # g[[i]] <- dataframe %>% plot_dot_plots( col = names(.)[i], fill = fill_by)
    }
  }
    
  g
}

# Plot a bar plot of a dataframe's column --------------------------------------------------
## The bar plot can also be filled according to another column as thought convenient
plot_bars <- function(df, col, fill = NULL) {
  
  col2 <- str_replace_all(col, " ", "_")
  names(df)[names(df)==col] <- col2
  
  if (is.null(fill)) {
    plot <- ggplot(df, aes(x= get(col2), y = (..count..)/sum(..count..)) ) +
      geom_bar(position = position_dodge(width = 0.5)) +
      scale_y_continuous(labels=percent, breaks = pretty_breaks(n=5), limits = c(0,1)) + 
      ylab("Percent") +
      scale_x_discrete(drop=FALSE) +
      xlab(col) + 
      geom_label_repel(stat = "count", aes(label = scales::percent((..count..)/sum(..count..))),
                       size = 3, alpha = 0.9) # +
    # theme(axis.title.x = element_text(face = "bold.italic"))
  } else {
  fill2 <- str_replace_all(fill, " ", "_")
  names(df)[names(df)==fill] <- fill2
  plot <- ggplot(df, aes(x= get(col2), y = (..count..)/sum(..count..), fill = get(fill2)) ) +
    geom_bar(position = position_dodge(width = 0.5)) +
    scale_y_continuous(labels=percent, breaks = pretty_breaks(n=5), limits = c(0,.7)) + 
    ylab("Percent") +
    scale_fill_discrete(drop=FALSE) + scale_x_discrete(drop=FALSE) +
    xlab(col) + guides(fill=guide_legend(title=str_replace_all(fill, " ", "\n"))) # +
    # geom_label_repel(stat = "count", aes(label = scales::percent((..count..)/sum(..count..))),
    #                 size = 3, alpha = 0.7) # +
    # theme(axis.title.x = element_text(face = "bold.italic"))
  }
  plot
}


# Plot a dot plot of a dataframe's column --------------------------------------------------
## The dot plot will have shapes according to how many variables are present. Caution not to plot 
## more than 6 at once
plot_dot_plots <- function(df, col, fill = NULL) {
  # df = logistics
  # col = "The timing of the webinar was convenient"
  col2 <- str_replace_all(col, " ", "_")
  names(df)[names(df)==col] <- col2
  df <- df %>%
    group_by(get(col2)) %>%
    summarise( n = n())
  names(df)[1] <- col2
  plot <-
    ggplot(df, aes(x = get(col2), y = n/sum(n))) +
    geom_point( ) +
    coord_flip() +
    geom_segment(aes(xend = get(col2), yend = 0)) +
    scale_y_continuous(labels=percent) + ylab("Percent") + xlab(col)
  # else {
  #   fill2 <- str_replace_all(fill, " ", "_")
  #   names(df)[names(df)==fill] <- fill2
  #   plot <- ggplot( df, aes(x= (get(col2)), fill = get(fill2))    ) +
  #     geom_bar(aes(y = (..count..)/sum(..count..)), 
  #              stat="count", 
  #              # position = position_dodge(preserve = "single")
  #              position = position_dodge(width = 0.5)
  #              ) + scale_fill_discrete(drop=FALSE) + scale_x_discrete(drop=FALSE) + 
  #     scale_y_continuous(labels=percent) + ylab("Percent") +
  #     xlab(col) + guides(fill=guide_legend(title=str_replace_all(fill, " ", "\n")))
  # }
  plot
}



country_from_instit <-  function (instit.name) {
  country = ifelse(instit.name == "Institut Pasteur de Tunis", 'Tunisia',
                   ifelse(instit.name == "University of Cape Town", 'South Africa',
                          ifelse(instit.name == "University of Mauritius", 'Mauritius',
                                 ifelse(instit.name== "University of Khartoum", 'Sudan',
                                        ifelse(instit.name  == "Covenant University", 'Nigeria',
                                               ifelse(instit.name  == "University of the Western Cape", 'South Africa',
                                                      ifelse( instit.name  == "Uganda Virus Research Institute", 'Uganda',
                                                              ifelse(instit.name  == "Noguchi Memorial Institute for Medical Research, University of Ghana",
                                                                     'Ghana',
                                                                     ifelse(instit.name  == "University of the Free State", 'South Africa',
                                                                            ifelse(instit.name== "The Egyptian Center of Bioinformatics and Genomics ", 'Egypt',
                                                                                   'NA'))))))))))
  # country = data.frame(Country = country)
  # return(country)
}

# To change plot order of bars, change levels in underlying factor
reorder_size <- function(x) {
  factor(x, levels = names(sort(table(x))))
}

# plot_bars <- function(df, col) {
#   col2 <- str_replace_all(col, " ", "_")
#   names(df)[names(df)==col] <- col2
#   plot <- ggplot(df, aes(x= reorder_size(get(col2)))) +
#     geom_bar(aes(y = (..count..)/sum(..count..))) +         
#     scale_y_continuous(labels=percent) + ylab("Percent") +
#     xlab(col)
#   plot
# }


poll <- function(data, string){
  poll <- data %>%
    select(contains(string))
  names(poll) <- names(poll) %>%
    str_replace(".* \\[", "") %>%
    str_replace("\\]", "")
  poll
}


#  ------------------------------------------------------------------------

# Re-level satisfaction levels to a human format --------------------------
releveling <- function (col){
  col %>%
    parse_factor(levels = NULL) %>%
    fct_recode("Very\nconvenient" = "Strongly agree", "Convenient" = "Agree",
               "Neutral" = "Neutral", "Inconvenient" = "Disagree",
               "Very\ninconvenient" = "Strongly disagree" ) %>%
    fct_expand("Very\ninconvenient", "Inconvenient", "Neutral", "Convenient", "Very\nconvenient") %>%
    fct_relevel("Very\ninconvenient", "Inconvenient", "Neutral", "Convenient", "Very\nconvenient")
  
  # col %>%
  #   parse_factor(levels = NULL) %>%
  #   fct_expand("Very agree", "Agree", "Neutral", "Disagree", "Strongly disagree") %>%
  #   fct_recode("Strongly\nagree" = "Strongly agree", "Agree" = "Agree",
  #              "Neutral" = "Neutral", "Disagree" = "Disagree",
  #              "Strongly\ndisagree" = "Strongly disagree" ) %>%
  #   fct_relevel("Strongly\ndisagree", "Disagree", "Neutral", "Agree", "Strongly\nagree")
    
}


# Arrange a shared legend, from https://stackoverflow.com/question --------

g_legend<-function(a.gplot){
  tmp <- ggplot_gtable(ggplot_build(a.gplot))
  leg <- which(sapply(tmp$grobs, function(x) x$name) == "guide-box")
  legend <- tmp$grobs[[leg]]
  return(legend)}