apiclassic/parsedescription.go

   1 package apiclassic
   2
   3 import (
   4         "errors"
   5         "golang.org/x/net/html"
   6         "bytes"
   7         "strings"
   8         "github.com/terorie/yt-mango/util"
   9 )
  10
  11 const descriptionSelector = "#eow-description"
  12
  13 func (p *parseVideoInfo) parseDescription() error {
  14         // Find description root
  15         descNode := p.doc.Find(descriptionSelector).First()
  16         if len(descNode.Nodes) == 0 { return errors.New("could not find description") }
  17
  18         // Markdown text
  19         var buffer bytes.Buffer
  20
  21         // Enumerate nodes
  22         for c := descNode.Nodes[0].FirstChild; c != nil; c = c.NextSibling {
  23                 switch c.Type {
  24                 case html.TextNode:
  25                         // FIXME: "&amp;lt;" gets parsed to => "<"
  26                         // Write text to buffer, escaping markdown
  27                         err := util.MarkdownTextEscape.ToBuffer(c.Data, &buffer)
  28                         if err != nil { return err }
  29                 case html.ElementNode:
  30                         switch c.Data {
  31                         // Newline
  32                         case "br":
  33                                 err := buffer.WriteByte(0x0a)
  34                                 if err != nil { return err }
  35                         // Link
  36                         case "a":
  37                                 err := parseLink(c, &buffer)
  38                                 if err != nil { return err }
  39                         }
  40                 }
  41         }
  42
  43         // Save description
  44         p.v.Description = buffer.String()
  45
  46         return nil
  47 }
  48
  49 func parseLink(c *html.Node, dest *bytes.Buffer) error {
  50         // Find text
  51         if c.FirstChild == nil { return nil } // Empty link
  52         if c.FirstChild.Type != html.TextNode {
  53                 return errors.New("unexpected non-text node")
  54         }
  55         text := c.FirstChild.Data
  56
  57         // Find href
  58         for _, attr := range c.Attr {
  59                 if attr.Key == "href" {
  60                         switch {
  61                         // hashtag
  62                         case strings.HasPrefix(attr.Val, "/results"):
  63                                 dest.WriteString(text)
  64
  65                         // real link
  66                         case strings.HasPrefix(attr.Val, "/redirect"):
  67                                 /*
  68                                 Not needed:
  69                                 // Decode link from href
  70                                 link, err := decodeLink(attr.Val)
  71                                 if err != nil { return err }
  72                                 // Escape to markdown
  73                                 link, err = net.MarkdownLinkEscape.ToString(link)
  74                                 if err != nil { return err }
  75                                 // Write to buffer
  76                                 dest.WriteString(fmt.Sprintf("[%s](%s)\n", text, link))
  77                                 */
  78                                 dest.WriteString(text)
  79
  80                         default:
  81                                 return errors.New("unknown link")
  82                         }
  83                         break
  84                 }
  85         }
  86         return nil
  87 }
  88
  89 /* Not needed
  90
  91 func decodeLink(href string) (string, error) {
  92         url, err := url2.Parse(href)
  93         if err != nil { return "", err }
  94
  95         query := url.Query()
  96         link := query.Get("q")
  97         if link == "" { return "", errors.New("empty link") }
  98
  99         return link, nil
 100 }
 101
 102  */