parser.nim 3.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103
  1. import xmltree, sequtils, strtabs, strutils, strformat, json, times
  2. import nimquery, regex
  3. import ./types, ./formatters
  4. proc getAttr(node: XmlNode; attr: string; default=""): string =
  5. if node.isNIl or node.attrs.isNil: return default
  6. return node.attrs.getOrDefault(attr)
  7. proc selectAttr(node: XmlNode; selector: string; attr: string; default=""): string =
  8. let res = node.querySelector(selector)
  9. if res == nil: "" else: res.getAttr(attr, default)
  10. proc selectText(node: XmlNode; selector: string): string =
  11. let res = node.querySelector(selector)
  12. result = if res == nil: "" else: res.innerText()
  13. proc parseProfile*(node: XmlNode): Profile =
  14. let profile = node.querySelector(".profile-card")
  15. result.fullname = profile.selectText(".fullname")
  16. result.username = profile.selectText(".username").strip(chars={'@', ' '})
  17. result.description = profile.selectText(".bio")
  18. result.verified = profile.selectText("li.verified").len > 0
  19. result.protected = profile.selectText(".Icon.Icon--protected").len > 0
  20. result.userpic = profile.selectAttr(".ProfileCard-avatarImage", "src").getUserpic()
  21. result.banner = profile.selectAttr("svg > image", "xlink:href").replace("600x200", "1500x500")
  22. if result.banner == "":
  23. result.banner = profile.selectAttr(".ProfileCard-bg", "style")
  24. let stats = profile.querySelectorAll(".ProfileCardStats-statLink")
  25. for s in stats:
  26. let text = s.getAttr("title").split(" ")[0]
  27. case s.getAttr("href").split("/")[^1]
  28. of "followers": result.followers = text
  29. of "following": result.following = text
  30. else: result.tweets = text
  31. proc parseTweetProfile*(tweet: XmlNode): Profile =
  32. result = Profile(
  33. fullname: tweet.getAttr("data-name"),
  34. username: tweet.getAttr("data-screen-name"),
  35. userpic: tweet.selectAttr(".avatar", "src").getUserpic(),
  36. verified: tweet.selectText(".Icon.Icon--verified").len > 0
  37. )
  38. proc parseTweet*(tweet: XmlNode): Tweet =
  39. result.id = tweet.getAttr("data-item-id")
  40. result.link = tweet.getAttr("data-permalink-path")
  41. result.text = tweet.selectText(".tweet-text").stripTwitterUrls()
  42. result.retweetBy = tweet.selectText(".js-retweet-text > a > b")
  43. result.pinned = "pinned" in tweet.getAttr("class")
  44. result.profile = parseTweetProfile(tweet)
  45. let time = tweet.querySelector(".js-short-timestamp")
  46. result.time = fromUnix(parseInt(time.getAttr("data-time", "0")))
  47. result.shortTime = time.innerText()
  48. result.replies = "0"
  49. result.likes = "0"
  50. result.retweets = "0"
  51. for action in tweet.querySelectorAll(".ProfileTweet-actionCountForAria"):
  52. let
  53. text = action.innerText.split()
  54. num = text[0]
  55. case text[1]
  56. of "replies": result.replies = num
  57. of "likes": result.likes = num
  58. of "retweets": result.retweets = num
  59. else: discard
  60. for photo in tweet.querySelectorAll(".AdaptiveMedia-photoContainer"):
  61. result.photos.add photo.attrs["data-image-url"]
  62. let player = tweet.selectAttr(".PlayableMedia-player", "style")
  63. if player.len > 0:
  64. let thumb = player.replace(re".+:url\('([^']+)'\)", "$1")
  65. if "tweet_video" in thumb:
  66. result.gif = thumb.replace(re".+thumb/([^\.']+)\.jpg.+", "$1")
  67. else:
  68. result.videoThumb = thumb
  69. proc parseTweets*(node: XmlNode): Tweets =
  70. if node.isNil: return
  71. node.querySelectorAll(".tweet").map(parseTweet)
  72. template selectTweets*(node: XmlNode; class: string): untyped =
  73. parseTweets(node.querySelector(class))
  74. proc parseConversation*(node: XmlNode): Conversation =
  75. result.tweet = parseTweet(node.querySelector(".permalink-tweet-container > .tweet"))
  76. result.before = node.selectTweets(".in-reply-to")
  77. let replies = node.querySelector(".replies-to")
  78. if replies.isNil: return
  79. result.after = replies.selectTweets(".ThreadedConversation--selfThread")
  80. for reply in replies.querySelectorAll("li > .stream-items"):
  81. let thread = parseTweets(reply)
  82. if not thread.anyIt(it in result.after):
  83. result.replies.add thread