fulltext

type TEI2 struct {
	XMLName   xml.Name   `xml:"TEI.2"`
	TEIHeader TEI2Header `xml:"teiHeader"`
	Text      TEI2Text   `xml:"text"`
}

type TEI2Body struct {
	XMLName    xml.Name        `xml:"body"`
	Paragraphs []TEI2Paragraph `xml:"p"`
}

type TEI2Header struct {
	XMLName xml.Name `xml:"teiHeader"`
}

type TEI2Paragraph struct {
	XMLName xml.Name   `xml:"p"`
	ID      string     `xml:"id,attr"`
	Words   []TEI2Word `xml:"w"`
}

type TEI2Text struct {
	XMLName xml.Name `xml:"text"`
	Body    TEI2Body `xml:"body"`
}

type TEI2Word struct {
	XMLName xml.Name `xml:"w"`
	Style   string   `xml:"style,attr"`
	Value   string   `xml:",chardata"`
}

type TEI2a struct {
	XMLName   xml.Name    `xml:"TEI.2"`
	TEIHeader TEI2aHeader `xml:"teiHeader"`
	Text      TEI2aText   `xml:"text"`
}

type TEI2aBody struct {
	XMLName    xml.Name         `xml:"body"`
	Paragraphs []TEI2aParagraph `xml:"p"`
}

type TEI2aHeader struct {
	XMLName xml.Name `xml:"teiHeader"`
}

type TEI2aParagraph struct {
	XMLName xml.Name    `xml:"p"`
	ID      string      `xml:"id,attr"`
	Words   []TEI2aWord `xml:"w"`
}

type TEI2aText struct {
	XMLName xml.Name  `xml:"text"`
	Body    TEI2aBody `xml:"body"`
}

type TEI2aWord struct {
	XMLName  xml.Name `xml:"w"`
	Function string   `xml:"function,attr"`
	Value    string   `xml:",chardata"`
}

type TXT1 struct {
	XMLName  xml.Name          `xml:"articles"`
	Articles []TXT1ArticleInfo `xml:"artInfo"`
}

type TXT1ArticleInfo struct {
	XMLName     xml.Name    `xml:"artInfo"`
	ID          string      `xml:"id,attr"`
	ProductLink string      `xml:"ProductLink"`
	OCRText     TXT1OCRText `xml:"ocrText"`
}

type TXT1OCRText struct {
	XMLName xml.Name `xml:"ocrText"`
	Value   string   `xml:",chardata"`
}

type TXT2 struct {
	XMLName   xml.Name   `xml:"page"`
	TXTHeader TXT2Header `xml:"pageInfo"`
	Text      TXT2Text   `xml:"pageContent"`
}

type TXT2Header struct {
	XMLName   xml.Name `xml:"pageInfo"`
	PageID    string   `xml:"pageID"`
	RecordID  string   `xml:"recordID"`
	OCR       string   `xml:"ocr"`
	ImageLink string   `xml:"imageLink"`
}

type TXT2Paragraph struct {
	XMLName xml.Name   `xml:"p"`
	ID      string     `xml:"id,attr"`
	Words   []TXT2Word `xml:"w"`
}

type TXT2Text struct {
	XMLName    xml.Name        `xml:"pageContent"`
	Paragraphs []TXT2Paragraph `xml:"p"`
}

type TXT2Word struct {
	XMLName xml.Name `xml:"wd"`
	Pos     string   `xml:"pos,attr"`
	Value   string   `xml:",chardata"`
}

type TXT3 struct {
	XMLName  xml.Name      `xml:"page"`
	PageID   string        `xml:"pageid"`
	Articles []TXT3Article `xml:"article"`
}

type TXT3Article struct {
	XMLName xml.Name `xml:"article"`
	Type    string   `xml:"type,attr"`
	ID      string   `xml:"id"`
	Text    TXT3Text `xml:"text"`
}

type TXT3Block struct {
	Pg TXT3PG `xml:"pg"`
	P  TXT3P  `xml:"p"`
}

type TXT3P struct {
	XMLName xml.Name `xml:"p"`
	Type    string   `xml:"type,attr"`
	WDs     []TXT3WD `xml:"wd"`
}

type TXT3PG struct {
	XMLName xml.Name `xml:"pg"`
	Pgref   string   `xml:"pgref,attr"`
	Clipref string   `xml:"clipref,attr"`
	Pos     string   `xml:"pos,attr"`
}

type TXT3Text struct {
	XMLName    xml.Name    `xml:"text"`
	TitleBlock TXT3Block   `xml:"text.title"`
	TextBlocks []TXT3Block `xml:"text.cr"`
}

type TXT3WD struct {
	XMLName xml.Name `xml:"wd"`
	Pos     string   `xml:"pos,attr"`
	Value   string   `xml:",chardata"`
}

type TXT4 struct {
	XMLName xml.Name   `xml:"articles"`
	Texts   []TXT3Text `xml:"text"`
}

type Word struct {
	Content string `xml:",chardata"`
}

type XML struct {
	XMLName xml.Name `xml:"TEI.2"`
	Word    []string `xml:"text>body>p"` // `xml:"mdWrap>xmlData>mods>name"`
}