go_tooling/tsv2htaccess/main.go

package main

import (
	"encoding/csv"
	"fmt"
	"html/template"
	"io"
	"os"
)

const apacheTemplate = `
{{/* Do not edit: generated by tsv2htaccess*/}}
{{ range . }}
{{- /*gotype: code.osinet.fr/fgm/go_tooling/tsv2htaccess.redirect*/ -}}
RewriteRule ^{{ .Clean }}$	{{ .Target }}	[L,NC,NE,R=301]
{{ end }}
`

type redirect struct {
	Clean, Target string
}

func main() {
	var err error

	if len(os.Args) != 4 {
		fmt.Fprintf(os.Stderr, `
Syntax:
	%s source destination prefix

Where:
- source		URL pairs in TSV format
- destination	Apache .htaccess fragment 
- prefix		Prefix to strip from source URLs in pairs
`, os.Args[0])
		return
	}

	source := os.Args[1]
	dest := os.Args[2]
	prefix := os.Args[3]

	fmt.Fprintf(os.Stderr, "Converting %s to %s, stripping %s\n", source, dest, prefix)
	var tsv io.ReadCloser
	tsv, err = os.Open(source)
	if err != nil {
		panic(err)
	}
	defer tsv.Close()

	redirects, err := read(tsv, prefix)
	if err != nil {
		panic(err)
	}

	hta, err := os.Create(dest)
	if err != nil {
		panic(err)
	}
	defer hta.Close()

	write(hta, redirects)
}

func read(tsv io.Reader, expectedPrefix string) ([]redirect, error) {
	var err error
	l := len(expectedPrefix)

	r := csv.NewReader(tsv)
	r.Comma = '\t'
	r.FieldsPerRecord = 2
	rows, err := r.ReadAll()
	if err != nil {
		return nil, err
	}

	redirects := make([]redirect, len(rows))
	for i, rawRow := range rows {
		absClean := rawRow[0]
		actualPrefix := absClean[0:l]
		if actualPrefix != expectedPrefix {
			panic(fmt.Errorf("incorrect source prefix: [%s], should be [%s]", actualPrefix, expectedPrefix))
		}
		clean := absClean[l:]
		redirects[i] = redirect{Clean: clean, Target: rawRow[1]}
	}
	return redirects, nil
}

func write(w io.Writer, redirects []redirect) {
	tpl := template.Must(template.New("htaccess").Parse(apacheTemplate))
	err := tpl.Execute(w, redirects)
	if err != nil {
		panic(err)
	}
	tpl.Execute(os.Stderr, redirects)
}