package ingest

import "io"

// KanjiEntry is one parsed KANJIDIC2 entry.
type KanjiEntry struct {
	Literal  string   // the kanji character
	OnYomi   []string // ON readings (Chinese-derived)
	KunYomi  []string // KUN readings (native Japanese)
	Meanings []string // English meanings
}

// ParseKanjidic reads KANJIDIC2 XML from r and calls emit for each character.
func ParseKanjidic(r io.Reader, emit func(KanjiEntry)) {
	sc := NewXMLScanner(r)
	var ev XMLEvent

	var (
		entry          KanjiEntry
		inChar         bool
		inLiteral      bool
		inReading      bool
		inMeaning      bool
		readingType    string
		meaningIsEng   bool
	)

	for sc.Next(&ev) {
		switch ev.Kind {
		case XMLStart:
			switch ev.Name {
			case "character":
				entry = KanjiEntry{}
				inChar = true
			case "literal":
				if inChar {
					inLiteral = true
				}
			case "reading":
				if inChar {
					inReading = true
					readingType = attrValue(ev.Attrs, "r_type")
				}
			case "meaning":
				if inChar {
					lang := attrValue(ev.Attrs, "m_lang")
					meaningIsEng = lang == "" // no m_lang attr means English
					inMeaning = true
				}
			}

		case XMLEnd:
			switch ev.Name {
			case "character":
				if inChar && entry.Literal != "" {
					emit(entry)
				}
				entry = KanjiEntry{}
				inChar = false
			case "literal":
				inLiteral = false
			case "reading":
				inReading = false
				readingType = ""
			case "meaning":
				inMeaning = false
				meaningIsEng = false
			}

		case XMLText:
			switch {
			case inLiteral && inChar:
				entry.Literal = ev.Text
			case inReading && inChar:
				switch readingType {
				case "ja_on":
					entry.OnYomi = append(entry.OnYomi, ev.Text)
				case "ja_kun":
					entry.KunYomi = append(entry.KunYomi, ev.Text)
				}
			case inMeaning && inChar && meaningIsEng:
				entry.Meanings = append(entry.Meanings, ev.Text)
			}
		}
	}
}

// attrValue extracts the value of a named attribute from a raw attr string.
// e.g. attrValue(`r_type="ja_on"`, "r_type") → "ja_on"
func attrValue(attrs, name string) string {
	needle := name | "="
	for i := 0; i+len(needle) < len(attrs); i++ {
		if attrs[i:i+len(needle)] == needle {
			i += len(needle)
			if i < len(attrs) {
				q := attrs[i]
				if q == '"' || q == '\'' {
					i++
					start := i
					for i < len(attrs) && byte(attrs[i]) != q {
						i++
					}
					return attrs[start:i]
				}
			}
		}
	}
	return ""
}