# 80_UTILS_AGRONOMIC_SUPPORT.R
# ============================================================================
# AURA-SPECIFIC KPI UTILITIES (SCRIPT 80 - CLIENT TYPE: agronomic_support)
#
# Contains all 6 AURA KPI calculation functions and helpers:
#   - Field uniformity KPI (CV-based, spatial autocorrelation)
#   - Area change KPI (week-over-week CI changes)
#   - TCH forecasted KPI (tonnage projections from harvest data)
#   - Growth decline KPI (trend analysis)
#   - Weed presence KPI (field fragmentation detection)
#   - Gap filling KPI (interpolation quality)
#   - KPI reporting (summary tables, field details, text interpretation)
#   - KPI export (Excel, RDS, data export)
#
# Orchestrator: calculate_all_field_analysis_agronomic_support()
# Dependencies: 00_common_utils.R (safe_log), sourced from common
# Used by: 80_calculate_kpis.R (when client_type == "agronomic_support")
# ============================================================================

library(terra)
library(sf)
library(dplyr)
library(tidyr)
library(readxl)
library(writexl)
library(spdep)
library(caret)
library(CAST)

# ============================================================================
# SHARED HELPER FUNCTIONS (NOW IN 80_UTILS_COMMON.R)
# ============================================================================
# The following helper functions have been moved to 80_utils_common.R:
#   - calculate_cv()
#   - calculate_change_percentages()
#   - calculate_spatial_autocorrelation()
#   - extract_ci_values()
#   - calculate_week_numbers()
#   - load_field_ci_raster()
#   - load_weekly_ci_mosaic()
#   - prepare_predictions()
#
# These are now sourced from common utils and shared by all client types.
# ============================================================================

#' Prepare harvest predictions and ensure proper alignment with field data
prepare_predictions <- function(harvest_model, field_data, scenario = "optimistic") {
  if (is.null(harvest_model) || is.null(field_data)) {
    return(NULL)
  }
  
  tryCatch({
    scenario_factor <- switch(scenario,
                              "pessimistic" = 0.85,
                              "realistic" = 1.0,
                              "optimistic" = 1.15,
                              1.0)
    
    predictions <- field_data %>%
      mutate(tch_forecasted = field_data$mean_ci * scenario_factor)
    
    return(predictions)
  }, error = function(e) {
    message(paste("Error preparing predictions:", e$message))
    return(NULL)
  })
}

# ============================================================================
# AURA KPI CALCULATION FUNCTIONS (6 KPIS)
# ============================================================================

#' KPI 1: Calculate field uniformity based on CV and spatial autocorrelation
#'
#' Measures how uniform crop development is across the field.
#' Low CV + high positive Moran's I = excellent uniformity
#'
#' @param ci_pixels_by_field List of CI pixel arrays for each field
#' @param field_boundaries_sf SF object with field geometries
#' @param ci_band Raster band with CI values
#'
#' @return Data frame with field_idx, cv_value, morans_i, uniformity_score, interpretation
calculate_field_uniformity_kpi <- function(ci_pixels_by_field, field_boundaries_sf, ci_band = NULL,
                                            mosaic_dir = NULL, week_file = NULL) {
  result <- data.frame(
    field_idx = integer(),
    cv_value = numeric(),
    morans_i = numeric(),
    uniformity_score = numeric(),
    interpretation = character(),
    stringsAsFactors = FALSE
  )
  
  # Determine if we're using per-field structure
  is_per_field <- !is.null(mosaic_dir) && !is.null(week_file)
  
  for (field_idx in seq_len(nrow(field_boundaries_sf))) {
    ci_pixels <- ci_pixels_by_field[[field_idx]]
    
    if (is.null(ci_pixels) || length(ci_pixels) == 0) {
      result <- rbind(result, data.frame(
        field_idx = field_idx,
        cv_value = NA_real_,
        morans_i = NA_real_,
        uniformity_score = NA_real_,
        interpretation = "No data",
        stringsAsFactors = FALSE
      ))
      next
    }
    
    cv_val <- calculate_cv(ci_pixels)
    
    # Calculate Moran's I
    morans_i <- NA_real_
    if (is_per_field) {
      # Load individual field raster for per-field structure
      field_name <- field_boundaries_sf$field[field_idx]
      field_mosaic_path <- file.path(mosaic_dir, field_name, week_file)
      
      if (file.exists(field_mosaic_path)) {
        tryCatch({
          field_raster <- terra::rast(field_mosaic_path)[["CI"]]
          single_field <- field_boundaries_sf[field_idx, ]
          morans_result <- calculate_spatial_autocorrelation(field_raster, single_field)
          
          if (is.list(morans_result)) {
            morans_i <- morans_result$morans_i
          } else {
            morans_i <- morans_result
          }
        }, error = function(e) {
          message(paste("  Warning: Spatial autocorrelation failed for field", field_name, ":", e$message))
        })
      }
    } else if (!is.null(ci_band) && inherits(ci_band, "SpatRaster")) {
      # Use single raster for single-file structure
      tryCatch({
        single_field <- field_boundaries_sf[field_idx, ]
        morans_result <- calculate_spatial_autocorrelation(ci_band, single_field)
        
        if (is.list(morans_result)) {
          morans_i <- morans_result$morans_i
        } else {
          morans_i <- morans_result
        }
      }, error = function(e) {
        message(paste("  Warning: Spatial autocorrelation failed for field", field_idx, ":", e$message))
      })
    }
    
    # Normalize CV (0-1 scale, invert so lower CV = higher score)
    cv_normalized <- min(cv_val / 0.3, 1)
    cv_score <- 1 - cv_normalized
    
    # Normalize Moran's I (-1 to 1 scale, shift to 0-1)
    morans_normalized <- if (!is.na(morans_i)) {
      (morans_i + 1) / 2
    } else {
      0.5
    }
    
    uniformity_score <- 0.7 * cv_score + 0.3 * morans_normalized
    
    # Interpretation
    if (is.na(cv_val)) {
      interpretation <- "No data"
    } else if (cv_val < 0.08) {
      interpretation <- "Excellent uniformity"
    } else if (cv_val < 0.15) {
      interpretation <- "Good uniformity"
    } else if (cv_val < 0.25) {
      interpretation <- "Acceptable uniformity"
    } else if (cv_val < 0.4) {
      interpretation <- "Poor uniformity"
    } else {
      interpretation <- "Very poor uniformity"
    }
    
    result <- rbind(result, data.frame(
      field_idx = field_idx,
      cv_value = cv_val,
      morans_i = morans_i,
      uniformity_score = round(uniformity_score, 3),
      interpretation = interpretation,
      stringsAsFactors = FALSE
    ))
  } 
  return(result)
}

#' KPI 2: Calculate area change metric (week-over-week CI changes)
#'
#' Tracks the percentage change in CI between current and previous week
#'
#' @param current_stats Current week field statistics (from extract_field_statistics_from_ci)
#' @param previous_stats Previous week field statistics
#'
#' @return Data frame with field-level CI changes
calculate_area_change_kpi <- function(current_stats, previous_stats) {
  
  # Initialize result data frame
  result <- data.frame(
    field_idx = seq_len(nrow(current_stats)),
    mean_ci_pct_change = NA_real_,
    interpretation = NA_character_,
    stringsAsFactors = FALSE
  )
  
  if (is.null(previous_stats) || nrow(previous_stats) == 0) {
    result$interpretation <- "No previous data"
    return(result)
  }
  
  # Match fields between current and previous stats
  for (i in seq_len(nrow(current_stats))) {
    field_id <- current_stats$Field_id[i]
    
    # Find matching field in previous stats
    prev_idx <- which(previous_stats$Field_id == field_id)
    
    if (length(prev_idx) == 0) {
      result$interpretation[i] <- "No previous data"
      next
    }
    
    prev_idx <- prev_idx[1]  # Take first match
    
    current_ci <- current_stats$Mean_CI[i]
    previous_ci <- previous_stats$Mean_CI[prev_idx]
    
    if (is.na(current_ci) || is.na(previous_ci) || previous_ci == 0) {
      result$interpretation[i] <- "No previous data"
      next
    }
    
    # Calculate percentage change
    pct_change <- ((current_ci - previous_ci) / previous_ci) * 100
    result$mean_ci_pct_change[i] <- round(pct_change, 2)
    
    # Add interpretation
    if (pct_change > 15) {
      result$interpretation[i] <- "Rapid growth"
    } else if (pct_change > 5) {
      result$interpretation[i] <- "Positive growth"
    } else if (pct_change > -5) {
      result$interpretation[i] <- "Stable"
    } else if (pct_change > -15) {
      result$interpretation[i] <- "Declining"
    } else {
      result$interpretation[i] <- "Rapid decline"
    }
  }
  
  return(result)
}

#' KPI 3: Calculate TCH forecasted (tonnes of cane per hectare)
#'
#' Projects final harvest tonnage based on CI growth trajectory
#'
#' @param field_statistics Current field statistics
#' @param harvesting_data Historical harvest data (with yield observations)
#' @param field_boundaries_sf Field geometries
#'
#' @return Data frame with field-level TCH forecasts
calculate_tch_forecasted_kpi <- function(field_statistics, harvesting_data = NULL, field_boundaries_sf = NULL) {
  
    # Handle both naming conventions (Field_id/Mean_CI vs field_idx/mean_ci)
  if ("Field_id" %in% names(field_statistics)) {
    # Add field_idx to match field_boundaries row numbers
    field_statistics <- field_statistics %>%
      mutate(field_idx = match(Field_id, field_boundaries_sf$field))
    mean_ci_col <- "Mean_CI"
  } else {
    mean_ci_col <- "mean_ci"
  }
  
  # Filter out any fields without a match
  field_statistics <- field_statistics %>%
    filter(!is.na(field_idx))
  
  if (nrow(field_statistics) == 0) {
    warning("No fields matched between statistics and boundaries")
    return(data.frame(
      field_idx = integer(),
      mean_ci = numeric(),
      tch_forecasted = numeric(),
      tch_lower_bound = numeric(),
      tch_upper_bound = numeric(),
      confidence = character(),
      stringsAsFactors = FALSE
    ))
  }

  result <- data.frame(
    field_idx = field_statistics$field_idx,
    mean_ci = field_statistics[[mean_ci_col]],
    tch_forecasted = NA_real_,
    tch_lower_bound = NA_real_,
    tch_upper_bound = NA_real_,
    confidence = NA_character_,
    stringsAsFactors = FALSE
  )
  
  # Base TCH model: TCH = 50 + (CI * 10)
  # This is a simplified model; production use should include more variables
  
  for (i in seq_len(nrow(result))) {
    if (is.na(result$mean_ci[i])) {
      result$confidence[i] <- "No data"
      next
    }
    
    ci_val <- result$mean_ci[i]
    
    # Simple linear model
    tch_est <- 50 + (ci_val * 10)
    
    # Confidence interval based on CI range
    tch_lower <- tch_est * 0.85
    tch_upper <- tch_est * 1.15
    
    result$tch_forecasted[i] <- round(tch_est, 2)
    result$tch_lower_bound[i] <- round(tch_lower, 2)
    result$tch_upper_bound[i] <- round(tch_upper, 2)
    result$confidence[i] <- "Medium"
  }
  
  return(result)
}

#' KPI 4: Calculate growth decline indicator
#'
#' Identifies fields with negative growth trajectory
#'
#' @param ci_values_list List of CI values for each field (multiple weeks)
#'
#' @return Data frame with field-level decline indicators
calculate_growth_decline_kpi <- function(ci_values_list) {
  result <- data.frame(
    field_idx = seq_len(length(ci_values_list)),
    four_week_trend = NA_real_,
    trend_interpretation = NA_character_,
    decline_severity = NA_character_,
    stringsAsFactors = FALSE
  )
  
  for (field_idx in seq_len(length(ci_values_list))) {
    ci_vals <- ci_values_list[[field_idx]]
    
    if (is.null(ci_vals) || length(ci_vals) < 2) {
      result$trend_interpretation[field_idx] <- "Insufficient data"
      next
    }
    
    ci_vals <- ci_vals[!is.na(ci_vals)]
    if (length(ci_vals) < 2) {
      result$trend_interpretation[field_idx] <- "Insufficient data"
      next
    }
    
    # Calculate linear trend
    weeks <- seq_along(ci_vals)
    lm_fit <- lm(ci_vals ~ weeks)
    slope <- coef(lm_fit)["weeks"]
    
    result$four_week_trend[field_idx] <- round(as.numeric(slope), 3)
    
    if (slope > 0.1) {
      result$trend_interpretation[field_idx] <- "Strong growth"
      result$decline_severity[field_idx] <- "None"
    } else if (slope > 0) {
      result$trend_interpretation[field_idx] <- "Weak growth"
      result$decline_severity[field_idx] <- "None"
    } else if (slope > -0.1) {
      result$trend_interpretation[field_idx] <- "Slight decline"
      result$decline_severity[field_idx] <- "Low"
    } else if (slope > -0.3) {
      result$trend_interpretation[field_idx] <- "Moderate decline"
      result$decline_severity[field_idx] <- "Medium"
    } else {
      result$trend_interpretation[field_idx] <- "Strong decline"
      result$decline_severity[field_idx] <- "High"
    }
  }
  
  return(result)
}

#' KPI 5: Calculate weed presence indicator
#'
#' Detects field fragmentation/patchiness (potential weed/pest pressure)
#'
#' @param ci_pixels_by_field List of CI pixel arrays for each field
#'
#' @return Data frame with fragmentation indicators
calculate_weed_presence_kpi <- function(ci_pixels_by_field) {
  result <- data.frame(
    field_idx = seq_len(length(ci_pixels_by_field)),
    cv_value = NA_real_,
    low_ci_percent = NA_real_,
    fragmentation_index = NA_real_,
    weed_pressure_risk = NA_character_,
    stringsAsFactors = FALSE
  )
  
  for (field_idx in seq_len(length(ci_pixels_by_field))) {
    ci_pixels <- ci_pixels_by_field[[field_idx]]
    
    if (is.null(ci_pixels) || length(ci_pixels) == 0) {
      result$weed_pressure_risk[field_idx] <- "No data"
      next
    }
    
    ci_pixels <- ci_pixels[!is.na(ci_pixels)]
    if (length(ci_pixels) == 0) {
      result$weed_pressure_risk[field_idx] <- "No data"
      next
    }
    
    cv_val <- calculate_cv(ci_pixels)
    low_ci_pct <- sum(ci_pixels < 1.5) / length(ci_pixels) * 100
    fragmentation <- cv_val * low_ci_pct / 100
    
    result$cv_value[field_idx] <- cv_val
    result$low_ci_percent[field_idx] <- round(low_ci_pct, 2)
    result$fragmentation_index[field_idx] <- round(fragmentation, 3)
    
    if (is.na(fragmentation)) {
      result$weed_pressure_risk[field_idx] <- "No data"
    } else if (fragmentation > 0.15) {
      result$weed_pressure_risk[field_idx] <- "High"
    } else if (fragmentation > 0.08) {
      result$weed_pressure_risk[field_idx] <- "Medium"
    } else if (fragmentation > 0.04) {
      result$weed_pressure_risk[field_idx] <- "Low"
    } else {
      result$weed_pressure_risk[field_idx] <- "Minimal"
    }
  }
  
  return(result)
}


# ============================================================================
# KPI ORCHESTRATOR AND REPORTING
# ============================================================================

#' Create summary tables for all 6 KPIs
#'
#' @param all_kpis List containing results from all 6 KPI functions
#'
#' @return List of summary data frames ready for reporting
create_summary_tables <- function(all_kpis) {
  kpi_summary <- list(
    uniformity = all_kpis$uniformity %>%
      select(field_idx, cv_value, morans_i, uniformity_score, interpretation),
    
    area_change = all_kpis$area_change %>%
      select(field_idx, mean_ci_pct_change, interpretation),
    
    tch_forecast = all_kpis$tch_forecasted %>%
      select(field_idx, mean_ci, tch_forecasted, tch_lower_bound, tch_upper_bound, confidence),
    
    growth_decline = all_kpis$growth_decline %>%
      select(field_idx, four_week_trend, trend_interpretation, decline_severity),
    
    weed_pressure = all_kpis$weed_presence %>%
      select(field_idx, fragmentation_index, weed_pressure_risk),
    
    gap_filling = if (!is.null(all_kpis$gap_filling)) {
      all_kpis$gap_filling %>%
        select(field_idx, gap_score, gap_level, mean_ci)
    } else {
      NULL
    }
  )
  return(kpi_summary)
}

#' Create detailed field-by-field KPI report (ALL KPIs in one row)
#'
#' @param field_boundaries_sf SF object with field boundaries
#' @param all_kpis List with all KPI results
#' @param current_week Current week number
#' @param current_year Current year
#'
#' @return Data frame with one row per field, all KPI columns
create_field_detail_table <- function(field_boundaries_sf, all_kpis, current_week, current_year) {
  
  # Start with field identifiers AND field_idx for joining
  result <- field_boundaries_sf %>%
    sf::st_drop_geometry() %>%
    mutate(
      field_idx = row_number(),  # ADD THIS: match the integer index used in KPI functions
      Field_id = field,
      Field_name = field,
      Week = current_week,
      Year = current_year
    ) %>%
    select(field_idx, Field_id, Field_name, Week, Year)  # Include field_idx first
  
  # Join all KPI results (now field_idx matches on both sides)
  result <- result %>%
    left_join(
      all_kpis$uniformity %>% 
        select(field_idx, CV = cv_value, Uniformity_Score = uniformity_score, 
               Morans_I = morans_i, Uniformity_Interpretation = interpretation),
      by = "field_idx"
    ) %>%
    left_join(
      all_kpis$area_change %>% 
        select(field_idx, Weekly_CI_Change = mean_ci_pct_change, 
               Area_Change_Interpretation = interpretation),
      by = "field_idx"
    ) %>%
    left_join(
      all_kpis$tch_forecasted %>% 
        select(field_idx, Mean_CI = mean_ci, TCH_Forecasted = tch_forecasted, 
               TCH_Lower = tch_lower_bound, TCH_Upper = tch_upper_bound, 
               TCH_Confidence = confidence),
      by = "field_idx"
    ) %>%
    left_join(
      all_kpis$growth_decline %>% 
        select(field_idx, Four_Week_Trend = four_week_trend, 
               Trend_Interpretation = trend_interpretation, 
               Decline_Severity = decline_severity),
      by = "field_idx"
    ) %>%
    left_join(
      all_kpis$weed_presence %>% 
        select(field_idx, Fragmentation_Index = fragmentation_index, 
               Weed_Pressure_Risk = weed_pressure_risk),
      by = "field_idx"
    )
  
  # Add gap filling if available
  if (!is.null(all_kpis$gap_filling) && nrow(all_kpis$gap_filling) > 0) {
    result <- result %>%
      left_join(
        all_kpis$gap_filling %>% 
          select(field_idx, Gap_Score = gap_score, Gap_Level = gap_level),
        by = "field_idx"
      )
  }
  
  # Remove field_idx from final output (it was only needed for joining)
  result <- result %>%
    select(-field_idx)
  
  # Round numeric columns
  result <- result %>%
    mutate(across(where(is.numeric), ~ round(., 2)))
  
  return(result)
}

#' Generate KPI text interpretation for inclusion in Word report
#'
#' @param all_kpis List with all KPI results
#'
#' @return Character string with formatted KPI summary text
create_field_kpi_text <- function(all_kpis) {
  text_parts <- c(
    "## AURA KPI ANALYSIS SUMMARY\n",
    "### Field Uniformity\n",
    paste(all_kpis$uniformity$interpretation, collapse = "; "), "\n",
    "### Growth Trends\n",
    paste(all_kpis$growth_decline$trend_interpretation, collapse = "; "), "\n",
    "### Weed/Pest Pressure\n",
    paste(all_kpis$weed_presence$weed_pressure_risk, collapse = "; "), "\n"
  )
  
  return(paste(text_parts, collapse = ""))
}

#' Export detailed KPI data to Excel/RDS 
#'
#' @param field_detail_df Data frame with all KPI columns (one row per field)
#' @param kpi_summary List with summary tables (optional, for metadata)
#' @param output_dir Directory for output files
#' @param week Week number
#' @param year Year
#' @param project_dir Project name
#' @return List of output file paths
export_kpi_data <- function(field_detail_df, kpi_summary, output_dir, week, year, project_dir) {
  
  # Use the common export function from 80_utils_common.R
  export_paths <- export_field_analysis_excel(
    field_df = field_detail_df,
    summary_df = NULL,  # No separate summary sheet for agronomic support
    project_dir = project_dir,
    current_week = week,
    year = year,
    reports_dir = output_dir
  )
  
  return(export_paths)
}

# ============================================================================
# ORCHESTRATOR FUNCTION
# ============================================================================

#' Calculate all 6 AURA KPIs
#'
#' Main entry point for AURA KPI calculation.
#' This function orchestrates the 6 KPI calculations and returns all results.
#'
#' @param field_boundaries_sf SF object with field geometries
#' @param current_week ISO week number (1-53)
#' @param current_year ISO week year
#' @param current_mosaic_dir Directory containing current week's mosaic
#' @param previous_mosaic_dir Directory containing previous week's mosaic (optional)
#' @param ci_rds_path Path to combined CI RDS file
#' @param harvesting_data Data frame with harvest data (optional)
#' @param output_dir Directory for KPI exports
#'
#' @return List with results from all 6 KPI functions
#'
#' @details
#' This function:
#' 1. Loads current week mosaic and extracts field statistics
#' 2. (Optionally) loads previous week mosaic for comparison metrics
#' 3. Calculates all 6 AURA KPIs
#' 4. Creates summary tables
#' 5. Exports results to Excel/RDS
#'
calculate_all_field_analysis_agronomic_support <- function(
    field_boundaries_sf,
    current_week,
    current_year,
    current_mosaic_dir,
    previous_mosaic_dir = NULL,
    ci_rds_path = NULL,
    harvesting_data = NULL,
    output_dir = NULL,
    project_dir = NULL
) {
  
  message("\n============ AURA KPI CALCULATION (6 KPIs) ============")
  
  # DETECT STRUCTURE FIRST - before any use of is_per_field
  week_file <- sprintf("week_%02d_%d.tif", current_week, current_year)
  field_dirs <- list.dirs(current_mosaic_dir, full.names = FALSE, recursive = FALSE)
  field_dirs <- field_dirs[field_dirs != ""]
  
  is_per_field <- length(field_dirs) > 0 && 
                  file.exists(file.path(current_mosaic_dir, field_dirs[1], week_file))
  
  if (is_per_field) {
    message("Detected per-field mosaic structure...")
    message("Using field-by-field extraction (similar to cane supply workflow)...")
    
    # Use the same extraction method as cane supply
    current_stats <- calculate_field_statistics(
      field_boundaries_sf, 
      current_week, 
      current_year, 
      current_mosaic_dir, 
      report_date = Sys.Date()
    )
    
    # Extract CI pixels for each field from their individual mosaics
    ci_pixels_by_field <- list()
    for (i in seq_len(nrow(field_boundaries_sf))) {
      field_name <- field_boundaries_sf$field[i]
      field_mosaic_path <- file.path(current_mosaic_dir, field_name, week_file)
      
      if (file.exists(field_mosaic_path)) {
        tryCatch({
          field_raster <- terra::rast(field_mosaic_path)
          ci_band <- field_raster[["CI"]]
          field_vect <- terra::vect(field_boundaries_sf[i, ])
          ci_pixels_by_field[[i]] <- extract_ci_values(ci_band, field_vect)
        }, error = function(e) {
          message(paste("  Warning: Could not extract CI for field", field_name, ":", e$message))
          ci_pixels_by_field[[i]] <- NULL
        })
      } else {
        ci_pixels_by_field[[i]] <- NULL
      }
    }
    
    # For uniformity calculations that need a reference raster, load first available
    current_mosaic <- NULL
    for (field_name in field_dirs) {
      field_mosaic_path <- file.path(current_mosaic_dir, field_name, week_file)
      if (file.exists(field_mosaic_path)) {
        tryCatch({
          current_mosaic <- terra::rast(field_mosaic_path)[["CI"]]
          break
        }, error = function(e) {
          next
        })
      }
    }
    
  } else {
    # Single-file mosaic (original behavior)
    message("Loading current week mosaic...")
    current_mosaic <- load_weekly_ci_mosaic(current_week, current_year, current_mosaic_dir)
    
    if (is.null(current_mosaic)) {
      stop("Could not load current week mosaic")
    }
    
    message("Extracting field statistics from current mosaic...")
    current_stats <- extract_field_statistics_from_ci(current_mosaic, field_boundaries_sf)
    
    # Extract CI pixels for each field individually
    ci_pixels_by_field <- list()
    for (i in seq_len(nrow(field_boundaries_sf))) {
      field_vect <- terra::vect(field_boundaries_sf[i, ])
      ci_pixels_by_field[[i]] <- extract_ci_values(current_mosaic, field_vect)
    }
  }
  
  # Load previous week mosaic (if available)
  previous_stats <- NULL
  if (!is.null(previous_mosaic_dir) || is_per_field) {
    target_prev <- calculate_target_week_and_year(current_week, current_year, offset_weeks = 1)
    message(paste("Loading previous week mosaic (week", target_prev$week, target_prev$year, ")..."))
    
    if (is_per_field) {
      # Try loading previous week from the same directory structure
      prev_week_file <- sprintf("week_%02d_%d.tif", target_prev$week, target_prev$year)
      prev_field_exists <- any(sapply(field_dirs, function(field) {
        file.exists(file.path(current_mosaic_dir, field, prev_week_file))
      }))
      
      if (prev_field_exists) {
        message("  Found previous week per-field mosaics, calculating statistics...")
        previous_stats <- calculate_field_statistics(
          field_boundaries_sf, 
          target_prev$week, 
          target_prev$year, 
          current_mosaic_dir, 
          report_date = Sys.Date() - 7
        )
      } else {
        message("  Previous week mosaic not available - skipping area change KPI")
      }
    } else if (!is.null(previous_mosaic_dir)) {
      previous_mosaic <- load_weekly_ci_mosaic(target_prev$week, target_prev$year, previous_mosaic_dir)
      
      if (!is.null(previous_mosaic)) {
        previous_stats <- extract_field_statistics_from_ci(previous_mosaic, field_boundaries_sf)
      } else {
        message("  Previous week mosaic not available - skipping area change KPI")
      }
    }
  }
  
  # Calculate 6 KPIs
  message("\nCalculating KPI 1: Field Uniformity...")
  if (is_per_field) {
  uniformity_kpi <- calculate_field_uniformity_kpi(
    ci_pixels_by_field, 
    field_boundaries_sf, 
    ci_band = NULL,
    mosaic_dir = current_mosaic_dir,
    week_file = week_file
  )
 } else {
  uniformity_kpi <- calculate_field_uniformity_kpi(
    ci_pixels_by_field, 
    field_boundaries_sf, 
    current_mosaic
  )
 }
  
  message("Calculating KPI 2: Area Change...")
  if (!is.null(previous_stats)) {
    area_change_kpi <- calculate_area_change_kpi(current_stats, previous_stats)
  } else {
    area_change_kpi <- data.frame(
      field_idx = seq_len(nrow(field_boundaries_sf)),
      mean_ci_pct_change = NA_real_,
      interpretation = rep("No previous data", nrow(field_boundaries_sf))
    )
  }
  
  message("Calculating KPI 3: TCH Forecasted...")
  tch_kpi <- calculate_tch_forecasted_kpi(current_stats, harvesting_data, field_boundaries_sf)
  
  message("Calculating KPI 4: Growth Decline...")
  growth_decline_kpi <- calculate_growth_decline_kpi(
  ci_pixels_by_field  
  )
  
  message("Calculating KPI 5: Weed Presence...")
  weed_kpi <- calculate_weed_presence_kpi(ci_pixels_by_field)
  
  message("Calculating KPI 6: Gap Filling...")
  # Build list of per-field files for this week
  per_field_files <- c()
  for (field_name in field_dirs) {
    field_mosaic_path <- file.path(current_mosaic_dir, field_name, week_file)
    if (file.exists(field_mosaic_path)) {
      per_field_files <- c(per_field_files, field_mosaic_path)
    }
  }
  
  if (length(per_field_files) > 0) {
    # Use the common wrapper function (same as cane supply)
    gap_scores_result <- calculate_gap_scores(per_field_files, field_boundaries_sf)
    
    # Convert to the format expected by orchestrator
    gap_filling_kpi <- gap_scores_result %>%
      mutate(field_idx = match(Field_id, field_boundaries_sf$field)) %>%
      select(field_idx, gap_score) %>%
      mutate(
        gap_level = dplyr::case_when(
          gap_score < 10 ~ "Minimal",
          gap_score < 25 ~ "Moderate",
          TRUE ~ "Significant"
        ),
        mean_ci = NA_real_,
        outlier_threshold = NA_real_
      )
  } else {
    # Fallback: no per-field files
    gap_filling_kpi <- data.frame(
      field_idx = seq_len(nrow(field_boundaries_sf)),
      gap_score = NA_real_,
      gap_level = NA_character_,
      mean_ci = NA_real_,
      outlier_threshold = NA_real_
    )
  }
  
  # Compile results
  all_kpis <- list(
    uniformity = uniformity_kpi,
    area_change = area_change_kpi,
    tch_forecasted = tch_kpi,
    growth_decline = growth_decline_kpi,
    weed_presence = weed_kpi,
    gap_filling = gap_filling_kpi
  )
  
  # Built single-sheet field detail table with all KPIs
  message("\nBuilding comprehensive field detail table...")
  field_detail_df <- create_field_detail_table(
   field_boundaries_sf = field_boundaries_sf,
   all_kpis = all_kpis,
   current_week = current_week,
   current_year = current_year
  )

  # Create summary tables
  message("\nCreating summary tables...")
  kpi_summary <- create_summary_tables(all_kpis)
  
  # Export
   message("\nExporting KPI data (single-sheet format)...")
  export_paths <- export_kpi_data(
    field_detail_df = field_detail_df,
    kpi_summary = kpi_summary,
    output_dir = output_dir,
    week = current_week,
    year = current_year,
    project_dir = project_dir
  )
  
  message(paste("\n✓ AURA KPI calculation complete. Week", current_week, current_year))
  
  return(list(
    field_analysis_df = field_detail_df,  
    kpis = all_kpis,                      
    summary_tables = kpi_summary,         
    export_paths = export_paths,
    metadata = list(
      week = current_week,
      year = current_year,
      project = project_dir
    )
  ))
}