"""
    Subhourly

Module for converting UC instances from 40-minute time periods (36 periods/day)
to 20-minute time periods (72 periods/day).
"""
module Subhourly

using Dates
import ..UnitCommitment

export convert_to_subhourly, interpolate_values, repeat_values

"""
    convert_to_subhourly(instance_path::AbstractString, next_day_path::AbstractString)

Convert a UC instance from 36 time periods (40 minutes each) to 72 time periods (20 minutes each).

# Arguments
- `instance_path::AbstractString`: Path to the current day's instance
- `next_day_path::AbstractString`: Path to the next day's instance (needed for interpolation at boundaries)

# Returns
- Modified `UnitCommitmentInstance` with 72 time periods

# Details
The function performs the following transformations:
1. **Interpolated quantities** (demands, profiled unit outputs): Linear interpolation using next day's first period
2. **Repeated quantities** (max_power, min_power, flow limits): Each value repeated twice
3. **Ramping capacities**: Halved (since time periods are half the duration)
4. **Time period counts** (min_uptime, min_downtime, startup delay): Doubled
5. **Power production costs (not the fixed startup costs)**: Halved

# Example
```julia
current_instance = convert_to_subhourly(
    "matpower/case14/2017-01-01",
    "matpower/case14/2017-01-02"
)
println("Total time periods: ", current_instance.time)  # Should be 72
```
"""
function convert_to_subhourly(instance_path::AbstractString, next_day_path::AbstractString)
    # Read instances
    instance = read_instance(instance_path)
    next_instance = read_instance(next_day_path)
    
    return convert_to_subhourly(instance, next_instance)
end

"""
    convert_to_subhourly(instance::UnitCommitment.UnitCommitmentInstance, 
                         next_instance::UnitCommitment.UnitCommitmentInstance)

Convert a UC instance from 36 time periods to 72 time periods using instance objects.
"""
function convert_to_subhourly(instance, next_instance)
    sc_current = instance.scenarios[1]
    sc_next = next_instance.scenarios[1]
    
    # Process buses - interpolate loads
    for i in 1:length(sc_current.buses)
        load_current = sc_current.buses[i].load
        load_next_first = sc_next.buses[i].load[1]
        sc_current.buses[i].load = interpolate_values(load_current, load_next_first)
    end
    
    # Process thermal units
    for i in 1:length(sc_current.thermal_units)
        unit = sc_current.thermal_units[i]
        unit_next = sc_next.thermal_units[i]
        
        # Repeat time-dependent vector quantities
        unit.max_power = repeat_values(unit.max_power)
        unit.min_power = repeat_values(unit.min_power)
        unit.must_run = repeat_values(unit.must_run)
        unit.min_power_cost = repeat_values(unit.min_power_cost)
        
        # Process cost segments
        for j in 1:length(unit.cost_segments)
            # cost should be repeated and then halved
            unit.cost_segments[j].cost = interpolate_values(unit.cost_segments[j].cost, unit.cost_segments[j].cost[1]) ./ 2.0
            unit.cost_segments[j].mw = interpolate_values(unit.cost_segments[j].mw, unit.cost_segments[j].mw[1])
        end
        
        # Repeat commitment status
        unit.commitment_status = repeat_values(unit.commitment_status)
        
        # Halve ramping capacities per time period (since time periods are half the duration)
        unit.ramp_up_limit = unit.ramp_up_limit / 2.0
        unit.ramp_down_limit = unit.ramp_down_limit / 2.0
        # Note: startup_limit and shutdown_limit are NOT modified (they are power limits, not per-period rates)
        
        # Double time period counts
        unit.min_uptime = unit.min_uptime * 2
        unit.min_downtime = unit.min_downtime * 2
        
        # Double startup delays in startup categories
        for startup_cat in unit.startup_categories
            startup_cat.delay = startup_cat.delay * 2
        end
    end
    
    # Process transmission lines
    for i in 1:length(sc_current.lines)
        line = sc_current.lines[i]
        
        # Repeat flow limits
        line.normal_flow_limit = repeat_values(line.normal_flow_limit)
        line.emergency_flow_limit = repeat_values(line.emergency_flow_limit)
        line.flow_limit_penalty = repeat_values(line.flow_limit_penalty)
    end
    
    # Process reserves - interpolate
    for i in 1:length(sc_current.reserves)
        reserve = sc_current.reserves[i]
        reserve_next = sc_next.reserves[i]
        
        # Interpolate reserve requirements
        reserve.amount = interpolate_values(reserve.amount, reserve_next.amount[1])
    end
    
    # Process price-sensitive loads - interpolate
    for i in 1:length(sc_current.price_sensitive_loads)
        psl = sc_current.price_sensitive_loads[i]
        psl_next = sc_next.price_sensitive_loads[i]
        
        # Interpolate demand and revenue
        psl.demand = interpolate_values(psl.demand, psl_next.demand[1])
        psl.revenue = interpolate_values(psl.revenue, psl_next.revenue[1])
    end
    
    # Process profiled units (renewables)
    for i in 1:length(sc_current.profiled_units)
        pu = sc_current.profiled_units[i]
        pu_next = sc_next.profiled_units[i]
        
        # Interpolate renewable profiles
        pu.min_power = interpolate_values(pu.min_power, pu_next.min_power[1])
        pu.max_power = interpolate_values(pu.max_power, pu_next.max_power[1])
        pu.cost = interpolate_values(pu.cost, pu_next.cost[1])
    end
    
    # Process storage units
    for i in 1:length(sc_current.storage_units)
        su = sc_current.storage_units[i]
        su_next = sc_next.storage_units[i]
        
        # Interpolate storage levels
        su.min_level = interpolate_values(su.min_level, su_next.min_level[1])
        su.max_level = interpolate_values(su.max_level, su_next.max_level[1])
        
        # Repeat other storage parameters
        su.simultaneous_charge_and_discharge = repeat_values(su.simultaneous_charge_and_discharge)
        su.charge_cost = interpolate_values(su.charge_cost, su_next.charge_cost[1])
        su.discharge_cost = interpolate_values(su.discharge_cost, su_next.discharge_cost[1])
        su.charge_efficiency = repeat_values(su.charge_efficiency)
        su.discharge_efficiency = repeat_values(su.discharge_efficiency)
        su.loss_factor = repeat_values(su.loss_factor)
        
        # Repeat rate limits
        su.min_charge_rate = repeat_values(su.min_charge_rate)
        su.max_charge_rate = repeat_values(su.max_charge_rate)
        su.min_discharge_rate = repeat_values(su.min_discharge_rate)
        su.max_discharge_rate = repeat_values(su.max_discharge_rate)
    end
    
    # Process scenario-level fields
    sc_current.power_balance_penalty = repeat_values(sc_current.power_balance_penalty)
    
    # Update time count
    sc_current.time = 72
    instance.time = 72
    
    return instance
end

"""
    interpolate_values(values::Vector{T}, next_first::T) where T

Interpolate a vector of 36 values to 72 values using linear interpolation.

# Arguments
- `values::Vector{T}`: Original 36-element vector
- `next_first::T`: First value from the next day (for boundary interpolation)

# Returns
- 72-element vector with interpolated values

# Details
For each pair of consecutive values, inserts an interpolated midpoint.
Uses `next_first` to interpolate the value after the last period.

# Example
```julia
interpolate_values([10.0, 20.0, 30.0], 40.0)
# Returns: [10.0, 15.0, 20.0, 25.0, 30.0, 35.0]
```
"""
function interpolate_values(values::Vector{T}, next_first::T) where T
    n = length(values)
    result = Vector{T}(undef, 2 * n)
    
    for i in 1:n-1
        result[2*i-1] = values[i]
        result[2*i] = (values[i] + values[i+1]) / 2
    end
    
    # Handle the last period using next day's first value
    result[2*n-1] = values[n]
    result[2*n] = (values[n] + next_first) / 2
    
    return result
end

"""
    repeat_values(values::Vector{T}) where T

Repeat each element of a 36-element vector to create a 72-element vector.

# Arguments
- `values::Vector{T}`: Original 36-element vector

# Returns
- 72-element vector with each value repeated twice

# Example
```julia
repeat_values([1, 2, 3])
# Returns: [1, 1, 2, 2, 3, 3]
```
"""
function repeat_values(values::Vector{T}) where T
    n = length(values)
    result = Vector{T}(undef, 2 * n)
    
    for i in 1:n
        result[2*i-1] = values[i]
        result[2*i] = values[i]
    end
    
    return result
end

"""
    read_instance(path::AbstractString)

Read a UC instance from a local file or benchmark.

# Arguments
- `path::AbstractString`: Path to instance file or benchmark name

# Returns
- `UnitCommitmentInstance` object
"""
function read_instance(path::AbstractString)
    # Check if path exists locally with common extensions
    if isfile(path)
        return UnitCommitment.read(path)
    elseif isfile(path * ".json.gz")
        return UnitCommitment.read(path * ".json.gz")
    elseif isfile(path * ".json")
        return UnitCommitment.read(path * ".json")
    else
        # Try reading as benchmark
        return UnitCommitment.read_benchmark(path)
    end
end

end # module