fulltext

type PAGEXML1 struct {
	XMLName  xml.Name         `xml:"PcGts"`
	Metadata PAGEXML1Metadata `xml:"Metadata"`
	Page     PAGEXML1Page     `xml:"Page"` // array?
}

type PAGEXML1Baseline struct {
	XMLName xml.Name `xml:"Baseline"`
	Points  string   `xml:"points,attr"`
}

type PAGEXML1Border struct {
	XMLName xml.Name       `xml:"Border"`
	Coords  PAGEXML1Coords `xml:"Coords"` // array?
}

type PAGEXML1Coords struct {
	XMLName xml.Name `xml:"Coords"`
	Points  string   `xml:"points,attr"`
}

type PAGEXML1Metadata struct {
	XMLName    xml.Name `xml:"Metadata"`
	Creator    string   `xml:"Creator"`
	Created    string   `xml:"Created"`
	LastChange string   `xml:"LastChange"`
}

type PAGEXML1OrderedGroup struct {
	XMLName xml.Name `xml:"OrderedGroup"`
	ID      string   `xml:"id,attr"`
	Caption string   `xml:"caption,attr"`

	RegionRefIndexed []PAGEXML1RegionRefIndexed `xml:"RegionRefIndexed"` // array?
}

type PAGEXML1Page struct {
	XMLName          xml.Name `xml:"Page"`
	ImageFilename    string   `xml:"imageFilename,attr"`
	ImageWidth       string   `xml:"imageWidth,attr"`
	ImageHeight      string   `xml:"imageHeight,attr"`
	ImageXResolution string   `xml:"imageXResolution,attr"`
	ImageYResolution string   `xml:"imageYResolution,attr"`
	Type             string   `xml:"type,attr"`

	Border       PAGEXML1Border       `xml:"Border"`
	ReadingOrder PAGEXML1ReadingOrder `xml:"ReadingOrder"`
	Relations    PAGEXML1Relations    `xml:"Relations"`
	TextRegion   PAGEXML1TextRegion   `xml:"TextRegion"`
}

type PAGEXML1Paragraph struct {
	XMLName xml.Name       `xml:"p"`
	ID      string         `xml:"id,attr"`
	Words   []PAGEXML1Word `xml:"w"`
}

type PAGEXML1ReadingOrder struct {
	XMLName xml.Name `xml:"ReadingOrder"`

	OrderedGroup PAGEXML1OrderedGroup `xml:"OrderedGroup"` // array?
}

type PAGEXML1RegionRefIndexed struct {
	XMLName   xml.Name `xml:"RegionRefIndexed"`
	Index     string   `xml:"index,attr"`
	RegionRef string   `xml:"regionRef,attr"`
}

type PAGEXML1Relation struct {
	XMLName xml.Name `xml:"Relation"`
	ID      string   `xml:"id,attr"`
	Type    string   `xml:"type,attr"`

	SourceRegionRef []PAGEXML1SourceRegionRef `xml:"SourceRegionRef"`
	TargetRegionRef []PAGEXML1TargetRegionRef `xml:"TargetRegionRef"`
}

type PAGEXML1Relations struct {
	XMLName xml.Name `xml:"Relations"`

	Relation PAGEXML1Relation `xml:"Relation"` // array?
}

type PAGEXML1SourceRegionRef struct {
	XMLName   xml.Name `xml:"SourceRegionRef"`
	RegionRef string   `xml:"regionRef,attr"`
}

type PAGEXML1TargetRegionRef struct {
	XMLName   xml.Name `xml:"TargetRegionRef"`
	RegionRef string   `xml:"regionRef,attr"`
}

type PAGEXML1TextEquiv struct {
	XMLName xml.Name `xml:"TextEquiv"`

	Unicode PAGEXML1Unicode `xml:"Unicode"` // array?
}

type PAGEXML1TextLine struct {
	XMLName         xml.Name `xml:"TextLine"`
	ID              string   `xml:"id,attr"`
	PrimaryLanguage string   `xml:"primaryLanguage,attr"`

	Coords    PAGEXML1Coords    `xml:"Coords"`    // array?
	TextEquiv PAGEXML1TextEquiv `xml:"TextEquiv"` // array?
	Baseline  PAGEXML1Baseline  `xml:"Baseline"`  // array?

	Word PAGEXML1Word `xml:"Word"` // array?
}

type PAGEXML1TextRegion struct {
	XMLName           xml.Name `xml:"TextRegion"`
	ID                string   `xml:"id,attr"`
	Type              string   `xml:"type,attr"`
	PrimaryLanguage   string   `xml:"primaryLanguage,attr"`
	SecondaryLanguage string   `xml:"secondaryLanguage,attr"`

	Coords    PAGEXML1Coords      `xml:"Coords"`    // array?
	TextLine  []PAGEXML1TextLine  `xml:"TextLine"`  // array?
	TextEquiv []PAGEXML1TextEquiv `xml:"TextEquiv"` // array?
}

type PAGEXML1Unicode struct {
	XMLName xml.Name `xml:"Unicode"`

	Value string `xml:",chardata"` // array?
}

type PAGEXML1Word struct {
	XMLName  xml.Name `xml:"Word"`
	ID       string   `xml:"id,attr"`
	Language string   `xml:"language,attr"`

	Coords    PAGEXML1Coords    `xml:"Coords"`    // array?
	TextEquiv PAGEXML1TextEquiv `xml:"TextEquiv"` // array?
}

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"`
	SourcePage string   `xml:"sourcePage"`
	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 TXT3a struct {
	XMLName xml.Name    `xml:"page"`
	PageID  string      `xml:"page-id"`
	IssueID string      `xml:"issue-id"`
	Texts   []TXT3aText `xml:"text"`
}

type TXT3aBlock struct {
	Pg TXT3aPG  `xml:"pg"`
	P  []TXT3aP `xml:"p"`
}

type TXT3aP struct {
	XMLName xml.Name `xml:"p"`
	Type    string   `xml:"type,attr"`
	WDs     []TXT3WD `xml:"wd"`
}

type TXT3aPG struct {
	XMLName xml.Name `xml:"pg"`
	Pgref   string   `xml:"pgref,attr"`
	Clipref string   `xml:"clipref,attr"`
	Pos     string   `xml:"pos,attr"`
}

type TXT3aText struct {
	XMLName    xml.Name     `xml:"text"`
	TitleBlock TXT3aBlock   `xml:"text.title"`
	TextBlocks []TXT3aBlock `xml:"text.cr"`
}

type TXT3aWD struct {
	XMLName xml.Name `xml:"wd"`
	Pos     string   `xml:"pos,attr"`
	Value   string   `xml:",chardata"`
}

type TXT3b struct {
	XMLName    xml.Name     `xml:"page"`
	PageID     string       `xml:"page-id"`
	IssueID    string       `xml:"issue-id"`
	TextBlocks []TXT3aBlock `xml:"text.cr"`
}

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

type TXT5 struct {
	XMLName xml.Name `xml:"IMAGE"`
	Name    string   `xml:"NAME,attr"`
	WDs     []TXT5WD `xml:"WD"`
}

type TXT5WD struct {
	XMLName xml.Name `xml:"WD"`
	Pos     string   `xml:"POS,attr"`
	Value   string   `xml:",chardata"`
}

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"`
}