func trNodeToSchedule(scheduleNode xml.Node) (item ScheduleItem, err error) {
results, err := scheduleNode.Search("./td/text()")
if err != nil {
return ScheduleItem{}, err
}
item = ScheduleItem{
TrainNumber: strings.TrimSpace(results[1].String()),
Misc: strings.TrimSpace(results[2].String()),
Class: strings.TrimSpace(results[3].String()),
Relation: strings.TrimSpace(results[4].String()),
StartingStation: strings.TrimSpace(results[5].String()),
CurrentStation: strings.TrimSpace(results[6].String()),
ArrivingTime: strings.TrimSpace(results[7].String()),
DepartingTime: strings.TrimSpace(results[8].String()),
Ls: strings.TrimSpace(results[9].String()),
}
if len(results) > 10 {
item.Status = strings.TrimSpace(results[10].String())
}
stationParts := strings.FieldsFunc(item.Relation, func(r rune) bool {
return r == '-'
})
item.EndStation = stationParts[1] // [ANGKE BOGOR] BOGOR is end station
return
}
func Link(performance xml.Node) string {
anchor, err := performance.Search(".//a")
if err != nil {
fmt.Println(err)
}
return "http://www.bso.org" + anchor[0].Attr("href")
}
func getContent(node gxtml.Node, cssQuery string) string {
result, err := node.Search(toXpath(cssQuery))
if err != nil {
panic(fmt.Errorf("Failed to find %v node", cssQuery))
}
return result[0].Content()
}
func parseSource(m xml.Node) string {
res, _ := m.Search("source")
if len(res) > 0 {
return res[0].Content()
}
return ""
}
func parseRights(m xml.Node) string {
res, _ := m.Search("rights")
if len(res) > 0 {
return res[0].Content()
}
return ""
}
func parsePublisher(m xml.Node) string {
publisher := ""
res, _ := m.Search("publisher")
if len(res) > 0 {
publisher = res[0].Content()
}
return publisher
}
func parseLanguages(m xml.Node) []string {
languages := []string{}
res, _ := m.Search("language")
for _, n := range res {
languages = append(languages, n.Content())
}
return languages
}
func parseTitles(m xml.Node) []string {
titles := []string{}
res, _ := m.Search("title")
for _, n := range res {
titles = append(titles, n.Content())
}
return titles
}
func parseSubjects(m xml.Node) []string {
subjects := []string{}
res, _ := m.Search("subject")
for _, n := range res {
subjects = append(subjects, n.Content())
}
return subjects
}
func parseDescription(m xml.Node) string {
description := ""
res, _ := m.Search("description")
if len(res) > 0 {
description = res[0].Content()
}
return description
}
func parseIdentifiers(m xml.Node) []*Identifier {
identifiers := []*Identifier{}
res, _ := m.Search("identifier")
for _, n := range res {
identifier := Identifier{Identifier: n.Content(), Scheme: n.Attr("scheme")}
identifiers = append(identifiers, &identifier)
}
return identifiers
}
func Pieces(details xml.Node) string {
pieces, _ := details.Search(".//div[@class='program-media-collapse']/h3")
var piecesString string
piecesString = "<ul class=\"works\">"
for _, piece := range pieces {
piecesString += "<li>"
piecesString += piece.Content()
piecesString += "</li>"
}
piecesString += "</ul>"
return piecesString
}
func linkDensity(node xml.Node) float64 {
links, err := node.Search("a")
if err != nil {
return 0.0
}
llength := 0.0
for _, link := range links {
llength += float64(len(link.Content()))
}
tlength := float64(len(node.Content()))
return llength / tlength
}
func processNode(node xml.Node, row string) {
row = row + node.Attr("TEXT") + "|"
kids, err := node.Search("node")
if err != nil {
log.Println("Error searching for node:", err)
return
}
if len(kids) > 0 { // has children, not a leaf node
for i := range kids {
processNode(kids[i], row)
}
} else {
fmt.Println(row) // print leaf node
}
}
/*
Not working because:
http://www.sc2ratings.com/players.php?realname=Yang,%20Hee-Soo
is parsed as:
http://www.sc2ratings.com/players.php?realname=Yang, Hee-Soo
*/
func parseLeagues(player xml.Node) []string {
out := []string{}
partialUrl, err := player.Search(".//a/@href")
errorHandler(err)
if len(partialUrl) == 1 {
playerPageUrl := "http://www.sc2ratings.com/" + partialUrl[0].String()
playerPageSource := retrievePageSource(playerPageUrl)
playerPage, err := gokogiri.ParseHtml(playerPageSource)
errorHandler(err)
defer playerPage.Free()
fmt.Println(playerPage)
}
return out
}
func parseDates(m xml.Node) []*Date {
dates := []*Date{}
res, _ := m.Search("date")
for _, n := range res {
date := Date{Date: n.Content(), Event: n.Attr("event")}
dates = append(dates, &date)
}
res, _ = m.Search("meta[@property='dcterms:modified']")
if len(res) > 0 {
date := Date{Date: res[0].Content(), Event: "modified"}
dates = append(dates, &date)
}
return dates
}
func loadInputs(doc xml.Node) ([]*Input, error) {
nodes, e := doc.Search(".//input")
if e != nil {
return nil, e
}
out := []*Input{}
for _, n := range nodes {
i := &Input{
Type: n.Attr("type"),
Name: n.Attr("name"),
Value: n.Attr("value"),
Checked: n.Attr("checked") == "checked",
}
out = append(out, i)
}
return out, nil
}
func Title(performance xml.Node) string {
title, _ := performance.Search(".//div[@class='performance-title']")
return title[0].Content()
}
func Time(performance xml.Node) string {
time, _ := performance.Search(".//span[@class='performance-time']")
return time[0].Content()
}
func Day(performance xml.Node) string {
day, _ := performance.Search(".//span[contains(@class, 'performance-day')]")
return day[0].Content()
}
// Parse a match row into a managed object
func GetParsedMatch(n xml.Node) (pm *ParsedMatch, err error) {
pm = &ParsedMatch{}
match_url, _ := n.Search("td/a/@href")
red, _ := n.Search("td/a/span[@class='redtext']/text()")
redvalue, _ := n.Search("td/a/span[@class='redtext']/following-sibling::text()")
blue, _ := n.Search("td/a/span[@class='bluetext']/text()")
bluevalue, _ := n.Search("td/a/span[@class='bluetext']/following-sibling::text()")
winner, _ := n.Search("td[position() = 2]/span/text()")
bettors, _ := n.Search("td[last()]/text()")
if len(match_url) > 0 {
pm.MatchId, _ = strconv.Atoi(numRx.FindString(match_url[0].String()))
}
if len(redvalue) > 0 {
pm.RedBets, _ = strconv.Atoi(numRx.FindString(redvalue[0].String()))
}
if len(bluevalue) > 0 {
pm.BlueBets, _ = strconv.Atoi(numRx.FindString(bluevalue[0].String()))
}
pm.Red = nameSub(html.UnescapeString(red[0].String()))
pm.Blue = nameSub(html.UnescapeString(blue[0].String()))
pm.Bettors, _ = strconv.Atoi(bettors[0].String())
if len(winner) > 0 {
pm.Winner = nameSub(html.UnescapeString(winner[0].String()))
if pm.Winner == pm.Red {
pm.FightWinner = spicerack.WINNER_RED
} else if pm.Winner == pm.Blue {
pm.FightWinner = spicerack.WINNER_BLUE
}
}
if pm.MatchId == 0 {
err = errors.New("Unable to parse match id.")
} else if len(pm.Red) == 0 || len(pm.Blue) == 0 {
err = errors.New("Red or Blue fighter is an empty string.")
} else if int(pm.FightWinner) == 0 {
err = errors.New("No winner found.")
} else if pm.MatchId < 51966 {
err = errors.New("Pre-matchmaking fight. Ignored.")
}
return pm, err
}
