Crunch voter turnout information from registration data

crunch.go

package main

import (
	"cmp"
	"encoding/csv"
	"flag"
	"fmt"
	"maps"
	"os"
	"path/filepath"
	"slices"
	"strconv"
	"strings"

	csvutil "github.com/earthboundkid/csv/v2"
)

func main() {
	app := parse()
	if err := app.run(); err != nil {
		fmt.Fprintf(os.Stderr, "Error: %v\n", err)
		os.Exit(1)
	}
}

type appEnv struct {
	in, out string
}

func parse() *appEnv {
	var app appEnv
	flag.StringVar(&app.in, "input", "voters.csv", "path` for input")
	flag.StringVar(&app.out, "out", ".", "`directory` for output")
	flag.CommandLine.Init("crunch", flag.ExitOnError)
	flag.Parse()
	return &app
}

func (app *appEnv) run() error {
	return app.counted()
}

type Counts struct {
	Total             int
	Turnout           int
	Male              int
	MaleTurnout       int
	Female            int
	FemaleTurnout     int
	Republican        int
	RepublicanTurnout int
	Democratic        int
	DemocraticTurnout int
	Other             int
	OtherTurnout      int
}

type InputRow struct {
	Gender                string `csv:"Gender"`
	DOB                   string `csv:"DOB"`
	FirstRegistered       string `csv:"First registered"`
	LastVoted             string `csv:"Last voted"`
	Party                 string `csv:"Party"`
	ZIP                   string `csv:"ZIP"`
	HouseDistrict         string `csv:"State House District"`
	SenateDistrict        string `csv:"State Senate District"`
	CongressionalDistrict string `csv:"Congressional District"`
	County                string `csv:"County"`
}

func (app *appEnv) counted() error {
	f, err := os.Open(app.in)
	if err != nil {
		return err
	}
	defer f.Close()
	counts := make(map[string]*Counts)
	var pennCounts Counts

	var row InputRow
	for range csvutil.Scan(csvutil.Options{
		Reader: f,
	}, &row) {
		pennCounts.increment(row)
		county := "County " + row.County
		setDefault(counts, county).increment(row)
		setDefault(counts, row.CongressionalDistrict).increment(row)
		setDefault(counts, row.HouseDistrict).increment(row)
		setDefault(counts, row.SenateDistrict).increment(row)
	}

	out, err := os.Create(filepath.Join(app.out, "counts.csv"))
	if err != nil {
		return err
	}
	defer func() {
		err = cmp.Or(err, out.Close())
	}()

	w := csv.NewWriter(out)
	if err := w.Write([]string{
		0: "Area",

		1: "Total registered",
		2: "Recent voter turnout",
		3: "Turnout percentage",

		4: "Males registered",
		5: "Male turnout",
		6: "Male turnout percentage",
		7: "Male percentage of registration",
		8: "Male percentage of total turnout",
		9: "Male turnout performance",

		10: "Females registered",
		11: "Female turnout",
		12: "Female turnout percentage",
		13: "Female percentage of registration",
		14: "Female percentage of total turnout",
		15: "Female turnout performance",

		16: "Republicans registered",
		17: "Republican turnout",
		18: "Republican turnout percentage",
		19: "Republican percentage of registration",
		20: "Republican percentage of total turnout",
		21: "Republican turnout performance",

		22: "Democrats registered",
		23: "Democratic turnout",
		24: "Democratic turnout percentage",
		25: "Democratic percentage of registration",
		26: "Democratic percentage of total turnout",
		27: "Democratic turnout performance",

		28: "Other party registered",
		29: "Other party turnout",
		30: "Other party turnout percentage",
		31: "Other party percentage of registration",
		32: "Other party percentage of total turnout",
		33: "Other party turnout performance",
	}); err != nil {
		return err
	}

	if err := w.Write(pennCounts.toOutputRow("Pennsylvania")); err != nil {
		return err
	}

	for _, k := range slices.Sorted(maps.Keys(counts)) {
		if err := w.Write(counts[k].toOutputRow(k)); err != nil {
			return err
		}
	}

	w.Flush()
	return w.Error()
}

func (c *Counts) increment(row InputRow) {
	c.Total++
	turnout := row.LastVoted == "05/20/2025"
	if turnout {
		c.Turnout++
	}
	if row.Gender == "M" {
		c.Male++
	}
	if row.Gender == "M" && turnout {
		c.MaleTurnout++
	}
	if row.Gender == "F" {
		c.Female++
	}
	if row.Gender == "F" && turnout {
		c.FemaleTurnout++
	}
	if row.Party == "R" {
		c.Republican++
		if turnout {
			c.RepublicanTurnout++
		}
	}
	if row.Party == "D" {
		c.Democratic++
		if turnout {
			c.DemocraticTurnout++
		}
	}
	if row.Party != "R" && row.Party != "D" {
		c.Other++
		if turnout {
			c.OtherTurnout++
		}
	}
}

func (c *Counts) toOutputRow(area string) []string {
	return []string{
		// 0: "Area",
		0: strings.TrimPrefix(area, "County "),
		// 1: "Total registered",
		1: strconv.Itoa(c.Total),
		// 2: "Recent voter turnout",
		2: strconv.Itoa(c.Turnout),
		// 3: "Turnout percentage",
		3: pct(c.Turnout, c.Total),

		// 4: "Males registered",
		// 5: "Male turnout",
		// 6: "Male turnout percentage",
		// 7: "Male percentage of registration",
		// 8: "Male percentage of total turnout",
		// 9: "Male turnout performance",
		4: strconv.Itoa(c.Male),
		5: strconv.Itoa(c.MaleTurnout),
		6: pct(c.MaleTurnout, c.Male),
		7: pct(c.Male, c.Total),
		8: pct(c.MaleTurnout, c.Turnout),
		9: performance(c.Male, c.MaleTurnout, c.Total, c.Turnout),

		// 10: "Females registered",
		// 11: "Female turnout",
		// 12: "Female turnout percentage",
		// 13: "Female percentage of registration",
		// 14: "Female percentage of total turnout",
		// 15: "Female turnout performance",
		10: strconv.Itoa(c.Female),
		11: strconv.Itoa(c.FemaleTurnout),
		12: pct(c.FemaleTurnout, c.Female),
		13: pct(c.Female, c.Total),
		14: pct(c.FemaleTurnout, c.Turnout),
		15: performance(c.Female, c.FemaleTurnout, c.Total, c.Turnout),

		// 16: "Republicans registered",
		// 17: "Republican turnout",
		// 18: "Republican turnout percentage",
		// 19: "Republican percentage of registration",
		// 20: "Republican percentage of total turnout",
		// 21: "Republican turnout performance",
		16: strconv.Itoa(c.Republican),
		17: strconv.Itoa(c.RepublicanTurnout),
		18: pct(c.RepublicanTurnout, c.Republican),
		19: pct(c.Republican, c.Total),
		20: pct(c.RepublicanTurnout, c.Turnout),
		21: performance(c.Republican, c.RepublicanTurnout, c.Total, c.Turnout),

		// 22: "Democrats registered",
		// 23: "Democratic turnout",
		// 24: "Democratic turnout percentage",
		// 25: "Democratic percentage of registration",
		// 26: "Democratic percentage of total turnout",
		// 27: "Democratic turnout performance",
		22: strconv.Itoa(c.Democratic),
		23: strconv.Itoa(c.DemocraticTurnout),
		24: pct(c.DemocraticTurnout, c.Democratic),
		25: pct(c.Democratic, c.Total),
		26: pct(c.DemocraticTurnout, c.Turnout),
		27: performance(c.Democratic, c.DemocraticTurnout, c.Total, c.Turnout),

		// 28: "Other party registered",
		// 29: "Other party turnout",
		// 30: "Other party turnout percentage",
		// 31: "Other party percentage of registration",
		// 32: "Other party percentage of total turnout",
		// 33: "Other party turnout performance",
		28: strconv.Itoa(c.Other),
		29: strconv.Itoa(c.OtherTurnout),
		30: pct(c.OtherTurnout, c.Other),
		31: pct(c.Other, c.Total),
		32: pct(c.OtherTurnout, c.Turnout),
		33: performance(c.Other, c.OtherTurnout, c.Total, c.Turnout),
	}
}

// pct returns a string representing num divided by denom as a percentage
func pct(num, denom int) string {
	return fmt.Sprintf("%.1f%%", 100*float64(num)/float64(denom))
}

// performance returns a percentage represening how much more or less a group turned out versus their percentage of registrants
func performance(groupReg, groupTurnout, total, turnout int) string {
	g, gt, r, t := float64(groupReg), float64(groupTurnout), float64(total), float64(turnout)
	perf := (gt * r / g / t) - 1
	return fmt.Sprintf("%.1f%%", 100*perf)
}

// setDefault is like Python's dict.setdefault.
// It adds a map entry for key if one doesn't exist
// and returns the existing entry if it does.
func setDefault[M ~map[K]*V, K comparable, V any](m M, k K) *V {
	if v := m[k]; v != nil {
		return v
	}
	v := new(V)
	m[k] = v
	return v
}