### TCI TRANSACTION VERIFICATION TOOL
# This script extracts text and specific information from mobile money transaction receipt screenshots and saves it to a CSV file.

## 1. INSTALL AND LOAD LIBRARIES
# Uncomment the install.packages() lines to install necessary packages if not already installed.

# install.packages('tesseract')  # For OCR (Optical Character Recognition)
# install.packages('magick')    # For image processing
# install.packages('stringr')   # For advanced string operations

library(tidyverse)  # Data manipulation and tidying
library(stringr)    # String operations
library(tesseract)  # OCR for extracting text from images
library(magick)     # Image reading and manipulation

## 2. SET FILE PATHS
# Specify file paths for input images and output data.

# Path to the folder with the image files (assumed to be JPEG files)
image_files <- list.files(path = "C:/Users/imagefolder/", pattern = "\\.jpg$", full.names = TRUE)

# Path to the output CSV file
output_file_path <- "C:/Users/outputfolder/outputfile.csv"

## 3. INITIALIZE OUTPUT DATA FRAME
# Create an empty data frame to store extracted information
combined_data <- tibble(
  FileName = character(),
  Contents = character(),
  AgentID = character(),
  AgentName = character(),
  Balance = character(),
  Charge = character(),
  Fee = character(),
  Tax = character()
)

# Function to read each image and clean the text
read_file_to_character <- function(file_path) {
  file_contents <- readLines(file_path, warn = FALSE, encoding = "UTF-8")
  # Replace non-ASCII characters with an empty string, but preserve original spacing
  file_contents <- stringi::stri_replace_all_regex(file_contents, "[^\\x20-\\x7E]", "")
  # Collapse the lines into a single string, if necessary
  paste(file_contents, collapse = " ")
}

## 4. REGULAR EXPRESSION PATTERNS
# Define patterns for extracting various pieces of information from the receipt text.

# Agent ID (e.g., Airtel receipts)
agent_id_pattern <- "Agent\\s*[^:]*:\\s*(\\d+)"

# Agent Name (e.g., MTN receipts)
agent_name_pattern <- "from\\s+([A-Z]+\\s[A-Z]+)"

# Transaction balance
balance_pattern <- "[Bb][a-z]+\\s?:?\\s*UG[Xx]\\s*([0-9]+,?[0-9]+)"

# Transaction charges
charge_pattern <- "[Cc]hargeUGX([0-9]+\\s?,?[0-9]+)"

# Transaction fee
fee_pattern <- "[Ff]ee\\s?i?s?\\s?:?\\s?UGX\\s+([0-9]+,?[0-9]+)"

# Transaction tax
tax_pattern <- "Tax\\s?:?\\s?UGX\\s?([0-9]+,?[0-9]+)"

# Additional patterns (examples)
phone_pattern <- "^[+]?(?:[0-9\\-.]\\s?){6,15}[0-9]$"  # Phone numbers
amount_pattern <- "UGX\\s+\\b\\d{1,3}(?:,\\d{3})*(?:\\.\\d+)?\\b"  # Amounts
email_pattern <- "^[a-zA-Z0-9]+(?:\\.[a-zA-Z0-9]+)*@[a-zA-Z0-9-]+\\.[a-zA-Z]{2,}$"  # Emails
date_pattern <- "\\d{2}-\\d{2}-\\d{4}"  # Dates (DD-MM-YYYY)

## 5. EXTRACT INFORMATION
# Loop through each image file, perform OCR, and extract relevant details.

for (file_path in image_files) {
  
  #Clean up text
  text <- read_file_to_character(file_path)
  # Read image and extract text using OCR
  image <- image_read(file_path)
  text <- ocr(image)
  
  # Extract file name
  file_name <- basename(file_path)
  
  # Extract balance
  balance <- str_extract(text, balance_pattern)
  balance <- if(!is.na(balance)) str_match(balance, balance_pattern)[,2] else NA
  
  # Extract fee
  fee <- str_extract(text, fee_pattern)
  fee <- if(!is.na(fee)) str_match(fee, fee_pattern)[,2] else NA
  
  # Extract charge
  charge <- str_extract(text, charge_pattern)
  charge <- if(!is.na(charge)) str_match(charge, charge_pattern)[,2] else NA
  
  # Extract tax
  tax <- str_extract(text, tax_pattern)
  tax <- if(!is.na(tax)) str_match(tax, tax_pattern)[,2] else NA
  
  # Extract agent ID
  agent_id <- str_extract_all(text, agent_id_pattern) %>% unlist()
  # Extract just the first occurrence of the Agent IDs if multiple are found
  agent_id <- if(length(agent_id) > 0) str_match(agent_id[1], agent_id_pattern)[,2] else NA
    
  # Extract agent name
  agentname <- str_extract(text, agent_name_pattern)
  agentname <- if(!is.na(agentname)) str_match(agentname, agent_name_pattern)[,2] else NA
  
  # Combine extracted data
  combined_data <- bind_rows(
    combined_data,
    tibble(
      FileName = file_name,
      Contents = text,
      AgentID = agent_id,
      AgentName = agent_name,
      Balance = balance,
      Fee = fee,
      Charge = charge,
      Tax = tax
    )
  )
}

## 6. SAVE RESULTS
# Write the extracted data to a CSV file
write_csv(combined_data, output_file_path)