convert_evernote_xml.go

package main

import (
	"bufio"
	"flag"
	"fmt"
	"io/ioutil"
	"os"
	"regexp"
	"strings"
	"time"
)

const OUTDIR = "output"

type Note struct {
	Title    string
	Contents string
	Updated  time.Time
}

func main() {
	flag.Parse()
	args := flag.Args()

	file, err := os.Open(args[0])
	if err != nil {
		fmt.Println("file error")
		os.Exit(1)
	}
	defer file.Close()

	sc := bufio.NewScanner(file)

	if _, err := os.Stat(OUTDIR); err != nil {
		if err = os.Mkdir(OUTDIR, 0777); err != nil {
			fmt.Printf("make dir ./%s failed", OUTDIR)
			os.Exit(1)
		}
	}

	var inContents = false
	var note Note
	for sc.Scan() {
		text := sc.Text()
		if strings.Contains(text, "<title>") {
			re := regexp.MustCompile(".+<title>(.+)</title>.+")
			note.Title = re.ReplaceAllString(text, "$1") + ".txt"
		} else if strings.Contains(text, "<en-note") {
			if strings.Contains(text, "</en-note>") {
				re := regexp.MustCompile(".*<en-note>(.+)</en-note>.*")
				contents := re.ReplaceAllString(text, "$1")
				note.Contents = note.Contents + "\n" + contents
			} else {
				inContents = true
			}
		} else if strings.Contains(text, "</en-note>") {
			inContents = false
		} else if inContents {
			note.Contents = note.Contents + "\n" + text
		} else if strings.Contains(text, "<updated>") {
			re := regexp.MustCompile(".+<updated>(.+)</updated>.+")
			timeString := re.ReplaceAllString(text, "$1")
			note.Updated, _ = time.Parse("20060102T150405Z", timeString)

			re = regexp.MustCompile("</?[a-zA-Z]+/?>")
			note.Contents = re.ReplaceAllString(note.Contents, "")
			note.Title = strings.Replace(note.Title, "/", "-", -1)
			path := OUTDIR + "/" + note.Title
			ioutil.WriteFile(path, []byte(note.Contents+"\n"), 0666)
			if err := os.Chtimes(path, note.Updated, note.Updated); err != nil {
				fmt.Printf("write error %s\n", path)
			}
			note.Contents = ""
		}
	}
}