package main
import (
"html/template"
"log"
"math/big"
"net/http"
"strings"
"crypto/rand"
"database/sql"
"github.com/gin-gonic/gin"
_ "github.com/mattn/go-sqlite3"
)
// TODO multiple backends
// I've been waffling on the question of whether to support multiple backends.
// there are two narratives here.
// NARRATIVE ONE: TRUNKLESS IS AN ART MACHINE BASED ON GUTENBERG
// Trunkless is specifically about exploring the project gutenberg corpus. It's a massive corpus that depicts a particular era of the English language. This focus means anyone working with it is exploring the same phrase space.
//
// Poets wishing for a similar interface that can be used generally should consult $TOOLTHATDOESNOTEXIST for doing digital cutup poetry locally.
// NARRATIVE TWO: TRUNKLESS IS A GENERAL PURPOSE CUTUP POETRY WORKBENCH
// Embodied in trunkless is a fundamental user interface for making digital cut-up poetry. This interface can support any kind of phrase backend. Because preparing the phrase backends is time and labor consuming, the user-poets cannot add their own; however, they can submit them for addition.
// If I wanted to support multiple backends what would it take?
// - a backend selector on the web tool -- drop down
// - extension of schema to include more clear metadata -- specifically a an optional sourceURL
// - changes to this file to support pairs of (dsn, maxID)s
//
// it's not that hard but given that i'm a stopgap for adding new phrase backends it feels like wasted effort.
// what about this narrative? I launch just with gutenberg and see if I hear "wow i want this for other stuff"
// the thing is, I want it for my stuff. but I consistently don't want a web app. I want a CLI or a controller-based interface. I can sit myself and put time into working on poems from gutenberg and then see if I want to build something totally new or just add new phrase backends.
// no matter what there are improvements to phrasing I want to see before I let anyone see this (even to beta test) so the plan now is:
// ship phraser improvements
// get a deployment pipeline
// seek beta feedback
const (
dsn = "phrase.db?cache=shared&mode=r"
maxID = 467014991
)
func connectDB() (*sql.DB, error) {
db, err := sql.Open("sqlite3", dsn)
if err != nil {
return nil, err
}
return db, nil
}
type source struct {
ID int64
Name string
}
type phrase struct {
ID int64
Text string
Source source
}
func main() {
r := gin.Default()
r.SetFuncMap(template.FuncMap{
"upper": strings.ToUpper,
})
r.LoadHTMLFiles("templates/index.tmpl")
r.StaticFile("/cutive.ttf", "./assets/cutive.ttf")
r.StaticFile("/favicon.ico", "./assets/favicon.ico")
r.StaticFile("/bg_light.gif", "./assets/bg_light.gif")
r.StaticFile("/bg_dark.gif", "./assets/bg_dark.gif")
r.StaticFile("/main.js", "./assets/main.js")
r.StaticFile("/html2canvas.min.js", "./assets/html2canvas.min.js")
randMax := big.NewInt(maxID)
r.HEAD("/", func(c *gin.Context) {
c.String(http.StatusOK, "")
})
r.GET("/", func(c *gin.Context) {
c.HTML(http.StatusOK, "index.tmpl", struct {
MaxID int
// TODO anything else?
}{maxID})
})
r.GET("/line", func(c *gin.Context) {
db, err := connectDB()
if err != nil {
log.Println(err.Error())
c.String(http.StatusInternalServerError, "oh no.")
return
}
defer db.Close()
id, err := rand.Int(rand.Reader, randMax)
if err != nil {
log.Println(err.Error())
c.String(http.StatusInternalServerError, "oh no.")
return
}
stmt, err := db.Prepare("select p.phrase, p.id, s.name from phrases p join sources s on p.sourceid = s.id where p.id = ?")
if err != nil {
log.Println(err.Error())
c.String(http.StatusInternalServerError, "oh no.")
return
}
row := stmt.QueryRow(id.Int64())
var p phrase
var s source
err = row.Scan(&p.Text, &s.ID, &s.Name)
if err != nil {
log.Println(err.Error())
c.String(http.StatusInternalServerError, "oh no.")
}
p.Source = s
p.ID = id.Int64()
c.JSON(http.StatusOK, p)
})
r.Run() // 8080
}