// ClearDocs removes old docs.
func ClearDocs() {
// rm docs and create a new one.
path := "./" + docs
err := os.RemoveAll(path)
if err != nil {
cognilog.Log("red", err)
}
err = os.Mkdir(path, 0755)
if err != nil {
cognilog.Log("red", err)
}
}
func jsnLinks(s string) []string {
config, err := ioutil.ReadFile(s)
if err != nil {
cognilog.Log("red", err)
}
jsn := make(map[string][]string)
err = json.Unmarshal(config, &jsn)
if err != nil {
cognilog.Log("red", err)
}
// slice of stopwords
links := jsn["links"]
return links
}
// NewCmd returns an initialized Cmd
func NewCmd(s string) *Cmd {
url, err := url.Parse(s)
if err != nil {
cognilog.Log("red", err)
}
return &Cmd{
U: url,
M: "GET",
}
}
// BotCmd returns an initialized robot Cmd
func BotCmd(s string) *Cmd {
lnk, err := url.Parse(s)
if err != nil {
cognilog.Log("red", err)
}
rob := lnk.ResolveReference(robotPath)
return &Cmd{
U: rob,
M: "GET",
}
}
// parseCmd returns cmd by resolving a url
func parseCmd(s string, lnk *url.URL) (*Cmd, error) {
href, err := url.Parse(s)
if err != nil {
cognilog.Log("red", err)
return nil, err
}
url := lnk.ResolveReference(href)
return &Cmd{
U: url,
M: "GET",
}, nil
}
// Seed creates a robot type, appends it to hostinfo and appends RootURL to
// the Queue.
func (f *Fetch) Seed(args ...string) {
for _, str := range args {
res, err := f.DoRequest(BotCmd(str))
if err != nil {
cognilog.Log("red", err)
}
robot := MakeBot(res)
if !robot.FullDisallow { // if not FullDisallow add
f.HostInfo = append(f.HostInfo, robot)
f.addSeed(NewCmd(str)) // add RootURL to Queue or Frontier.
cognilog.LogINFO("green", "added to Queue", str)
}
}
}
// MakeBot takes an http.Response and returns a Robot
func MakeBot(res *http.Response) *Robot {
var robot = new(Robot)
// treat all 4xx errors in the same way. Assume that there are no restrictions.
if res.StatusCode >= 400 && res.StatusCode > 500 {
robot.RootURL = NewCmd(strings.TrimSuffix(res.Request.URL.String(), "robots.txt")).URL()
robot.FullAllow = true
robot.CrawDelay = DefaultCrawlDelay
return robot
} else if res.StatusCode == 200 {
byt, err := ioutil.ReadAll(res.Body)
if err != nil {
cognilog.LogINFO("red", "Body read error", err)
}
redr := strings.NewReader(string(byt))
scanner := bufio.NewScanner(redr)
var groups = make(map[string][]string)
var key string
var status bool
for scanner.Scan() {
txt := strings.ToLower(scanner.Text())
if txt == "" {
continue
}
// new group
if strings.HasPrefix(txt, "user-agent:") {
key = trimSpaces(txt)
status = true
continue
}
if status && key != "" {
groups[key] = append(groups[key], trimSpaces(txt))
}
}
if err := scanner.Err(); err != nil {
cognilog.Log("red", err)
}
if isAllowAll(groups) {
robot.FullAllow = true
} else if isDisallowAll(groups) {
robot.FullDisallow = true
}
robot.RootURL = NewCmd(strings.TrimSuffix(res.Request.URL.String(), "robots.txt")).URL()
robot.Groups = groups
robot.CrawDelay = DefaultCrawlDelay
return robot
}
robot.RootURL = NewCmd(strings.TrimSuffix(res.Request.URL.String(), "robots.txt")).URL()
return robot
}
// It creates a robot type, appends it to hostinfo and RootURL to the Queue.
func (f *Fetch) SeedSlice(jsn string) {
args := jsnLinks(jsn)
cln := resolv(args)
for _, str := range cln {
res, err := f.DoRequest(BotCmd(str))
if err != nil {
cognilog.Log("red", err)
continue
}
robot := MakeBot(res)
if !robot.FullDisallow { // if not FullDisallow add
f.HostInfo = append(f.HostInfo, robot)
f.addSeed(NewCmd(str)) // add RootURL to Queue or Frontier.
cognilog.LogINFO("green", "Root added", str)
} else {
cognilog.LogINFO("red", "Root Not added", str)
}
}
}
