func getParagraphsToTagEnd(reader *xmltextreader.XmlTextReaderPtr, tag string, currentPointer string, c chan *Word) {
res := reader.Read()
eventType := reader.NodeType()
currentElementNum := 0
for !(eventType == xmltextreader.XML_END_ELEMENT && tag == reader.Name()) && res != -1 {
if xmltextreader.XML_START_ELEMENT == eventType {
currentElementNum += 1
name := reader.Name()
if name == paragraph_tag {
getParaToTagEnd(reader, paragraph_tag, currentPointer+"/"+strconv.Itoa(currentElementNum), 1.5, c)
}
}
res = reader.Read()
eventType = reader.NodeType()
}
}
func processPersonInfo(reader *xmltextreader.XmlTextReaderPtr, tag string, currentPointer string, c chan *Word) {
res := reader.Read()
eventType := reader.NodeType()
currentElementNum := 0
for !(eventType == xmltextreader.XML_END_ELEMENT && tag == reader.Name()) && res != -1 {
if eventType == xmltextreader.XML_START_ELEMENT {
currentElementNum += 1
name := reader.Name()
if first_name_tag == name {
sendWords(getText(reader), currentPointer+"/"+strconv.Itoa(currentElementNum), 1.3, c)
} else if last_name_tag == name {
sendWords(getText(reader), currentPointer+"/"+strconv.Itoa(currentElementNum), 1.3, c)
} else if middle_name_tag == name {
sendWords(getText(reader), currentPointer+"/"+strconv.Itoa(currentElementNum), 1.3, c)
} else if nick_name_tag == name {
sendWords(getText(reader), currentPointer+"/"+strconv.Itoa(currentElementNum), 1.1, c)
}
}
res = reader.Read()
eventType = reader.NodeType()
}
}
func processTitleInfo(reader *xmltextreader.XmlTextReaderPtr, tag string, currentPointer string, c chan *Word) {
res := reader.Read()
eventType := reader.NodeType()
currentElementNum := 0
for !(eventType == xmltextreader.XML_END_ELEMENT && tag == reader.Name()) && res != -1 {
if eventType == xmltextreader.XML_START_ELEMENT {
currentElementNum += 1
name := reader.Name()
if author_tag == name {
processPersonInfo(reader, name, currentPointer+"/"+strconv.Itoa(currentElementNum), c)
} else if book_title_tag == name {
sendWords(getText(reader), currentPointer+"/"+strconv.Itoa(currentElementNum), 2.0, c)
} else if annotation_tag == name {
getParagraphsToTagEnd(reader, annotation_tag, currentPointer+"/"+strconv.Itoa(currentElementNum), c)
} else if translator_tag == name {
sendWords(getText(reader), currentPointer+"/"+strconv.Itoa(currentElementNum), 1.1, c)
}
}
res = reader.Read()
eventType = reader.NodeType()
}
}
func processBody(reader *xmltextreader.XmlTextReaderPtr, tag string, currentPointer string, c chan *Word) {
currentElementNum := 0
res := reader.Read()
eventType := reader.NodeType()
if currentPointer == "/1" {
currentElementNum = 1
}
for !(xmltextreader.XML_END_ELEMENT == eventType && tag == reader.Name()) && -1 != res {
name := reader.Name()
if xmltextreader.XML_START_ELEMENT == eventType {
currentElementNum += 1
if paragraph_tag == name || text_author_tag == name || subtitle_tag == name || verse_tag == name {
getParaToTagEnd(reader, name, currentPointer+"/"+strconv.Itoa(currentElementNum), 1, c)
} else if body_tag == name {
hasAttr, err := reader.HasAttributes()
if nil == err && hasAttr == false {
processBody(reader, name, currentPointer+"/"+strconv.Itoa(currentElementNum), c)
}
} else if section_tag == name {
hasAttr, err := reader.HasAttributes()
if nil == err && hasAttr {
// Skip section with id attr, notes definitions
getParaToTagEnd(reader, name, "", 1, c)
} else {
processBody(reader, name, currentPointer+"/"+strconv.Itoa(currentElementNum), c)
}
} else if title_tag == name || epigraph_tag == name {
processBody(reader, name, currentPointer+"/"+strconv.Itoa(currentElementNum), c)
}
}
res = reader.Read()
eventType = reader.NodeType()
}
}
func getParaToTagEnd(reader *xmltextreader.XmlTextReaderPtr, tag string, fb2Pointer string, weightMult float32, c chan *Word) {
res := reader.Read()
eventType := reader.NodeType()
var (
strong byte = 1 << 0
emphasis byte = 1 << 1
link byte = 1 << 2
)
var name string
var styles byte = 0
for !(eventType == xmltextreader.XML_END_ELEMENT && tag == reader.Name()) && res != -1 {
if xmltextreader.XML_TEXT_NODE == eventType {
var weight float32 = 1.0
if styles&strong != 0 {
weight += 0.2
} else if styles&emphasis != 0 {
weight += 0.1
} else if styles&link != 0 {
weight += 0.1
}
if len(fb2Pointer) > 0 {
sendWords(reader.Value(), fb2Pointer, weight*weightMult, c)
}
} else if xmltextreader.XML_START_ELEMENT == eventType {
name = reader.Name()
if emphasis_tag == name {
styles |= emphasis
} else if strong_tag == name {
styles |= strong
} else if link_tag == name {
styles |= link
}
} else if xmltextreader.XML_END_ELEMENT == eventType {
name = reader.Name()
if emphasis_tag == name {
styles &^= emphasis
} else if strong_tag == name {
styles &^= strong
} else if link_tag == name {
styles &^= link
}
}
res = reader.Read()
eventType = reader.NodeType()
}
}
func processDescription(reader *xmltextreader.XmlTextReaderPtr, currentPointer string, c chan *Word) int {
nextTag(reader)
if reader.Name() != description_tag {
return -1
}
res := reader.Read()
eventType := reader.NodeType()
currentElementNum := 0
for !(eventType == xmltextreader.XML_END_ELEMENT && reader.Name() == description_tag) && res != -1 {
if eventType == xmltextreader.XML_START_ELEMENT {
currentElementNum += 1
name := reader.Name()
if name == title_info_tag || name == src_title_info_tag {
processTitleInfo(reader, name, currentPointer+"/"+strconv.Itoa(currentElementNum), c)
}
}
res = reader.Read()
eventType = reader.NodeType()
}
return 1
}