package main

import (
	// "github.com/paulrosania/go-charset/charset"
	"encoding/xml"
	"fmt"
	"github.com/rug-compling/alpinocorpus-go/alpinocorpus"
	"os"
	"runtime"
	"strings"
)

func main() {
	for _, filename := range os.Args[1:] {
		doFile(filename)
	}
}

func doFile(filename string) {
	reader, err := alpinocorpus.NewReader(filename)
	defer reader.Close()
	checkErr(err)

	entries, err := reader.GetAll()
	checkErr(err)

	name := reader.Name()
	for pair := range entries.KeysValues() {
		doXML(name, pair.Key, &pair.Value)
	}
}

func doXML(filename, dataname string, data *string) {

	type Node struct {
		Nodes []Node `xml:"node"`
		Root  string `xml:"root,attr"`
		Lemma string `xml:"lemma,attr"`
		Word  string `xml:"word,attr"`
	}

	type Nodes struct {
		Nodes    []Node `xml:"node"`
		Sentence string `xml:"sentence"`
	}

	nodes := &Nodes{}
	p := xml.NewDecoder(strings.NewReader(*data))
	//p.CharsetReader = charset.NewReader // needed if xml is not in UTF-8
	checkErr(p.Decode(nodes))

	roots := make([]string, 0, 100)
	lemmas := make([]string, 0, 100)
	var f func(node Node)
	f = func(node Node) {
		if node.Root != "" && node.Root != node.Word {
			roots = append(roots, node.Root)
		}
		if node.Lemma != "" && node.Lemma != node.Word && node.Lemma != node.Root {
			lemmas = append(lemmas, node.Lemma)
		}
		for _, n := range node.Nodes {
			f(n)
		}
	}
	for _, n := range nodes.Nodes {
		f(n)
	}

	fmt.Printf("FILE: %v/%v\n%v", filename, dataname, nodes.Sentence)
	for _, root := range roots {
		fmt.Print(" ", root)
	}
	for _, lemma := range lemmas {
		fmt.Print(" ", lemma)
	}
	fmt.Print("\n\n")

}

func checkErr(err error) {
	if err != nil {
		_, filename, lineno, ok := runtime.Caller(1)
		if ok {
			fmt.Fprintf(os.Stderr, "\n\n%v:%v: Error: %v\n\n", filename, lineno, err)
		}
		panic(err)
	}
}