parserutils.nim 5.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153
  1. import xmltree, htmlparser, strtabs, strformat, times
  2. import nimquery, regex
  3. import ./types, ./formatters, ./api
  4. const
  5. thumbRegex = re".+:url\('([^']+)'\)"
  6. gifRegex = re".+thumb/([^\.']+)\.jpg.*"
  7. proc getAttr*(node: XmlNode; attr: string; default=""): string =
  8. if node.isNil or node.attrs.isNil: return default
  9. return node.attrs.getOrDefault(attr)
  10. proc selectAttr*(node: XmlNode; selector: string; attr: string; default=""): string =
  11. let res = node.querySelector(selector)
  12. if res == nil: "" else: res.getAttr(attr, default)
  13. proc selectText*(node: XmlNode; selector: string): string =
  14. let res = node.querySelector(selector)
  15. result = if res == nil: "" else: res.innerText()
  16. proc getHeader(profile: XmlNode): XmlNode =
  17. result = profile.querySelector(".permalink-header")
  18. if result.isNil:
  19. result = profile.querySelector(".stream-item-header")
  20. if result.isNil:
  21. result = profile.querySelector(".ProfileCard-userFields")
  22. proc isVerified*(profile: XmlNode): bool =
  23. getHeader(profile).selectText(".Icon.Icon--verified").len > 0
  24. proc isProtected*(profile: XmlNode): bool =
  25. getHeader(profile).selectText(".Icon.Icon--protected").len > 0
  26. proc getName*(profile: XmlNode; selector: string): string =
  27. profile.selectText(selector).stripText()
  28. proc getUsername*(profile: XmlNode; selector: string): string =
  29. profile.selectText(selector).strip(chars={'@', ' '})
  30. proc emojify*(node: XmlNode) =
  31. for i in node.querySelectorAll(".Emoji"):
  32. i.add newText(i.getAttr("alt"))
  33. proc getTweetText*(tweet: XmlNode): string =
  34. let
  35. selector = ".tweet-text > a.twitter-timeline-link.u-hidden"
  36. link = tweet.selectAttr(selector, "data-expanded-url")
  37. quote = tweet.querySelector(".QuoteTweet")
  38. text = tweet.querySelector(".tweet-text")
  39. hasEmojis = not text.querySelector(".Emoji").isNil
  40. if hasEmojis:
  41. emojify(text)
  42. result = stripText(selectText(text, ".tweet-text"))
  43. if not quote.isNil and link.len > 0:
  44. result = result.replace(link, "")
  45. result = stripTwitterUrls(result)
  46. proc getTime(tweet: XmlNode): XmlNode =
  47. tweet.querySelector(".js-short-timestamp")
  48. proc getTimestamp*(tweet: XmlNode): Time =
  49. let time = getTime(tweet).getAttr("data-time", "0")
  50. fromUnix(parseInt(time))
  51. proc getShortTime*(tweet: XmlNode): string =
  52. getTime(tweet).innerText()
  53. proc getBio*(profile: XmlNode; selector: string): string =
  54. profile.selectText(selector).stripText()
  55. proc getAvatar*(profile: XmlNode; selector: string): string =
  56. profile.selectAttr(selector, "src").getUserpic()
  57. proc getBanner*(tweet: XmlNode): string =
  58. let url = tweet.selectAttr("svg > image", "xlink:href")
  59. if url.len > 0:
  60. result = url.replace("600x200", "1500x500")
  61. else:
  62. result = tweet.selectAttr(".ProfileCard-bg", "style")
  63. if result.len == 0:
  64. result = "background-color: #161616"
  65. proc getPopupStats*(profile: var Profile; node: XmlNode) =
  66. for s in node.querySelectorAll( ".ProfileCardStats-statLink"):
  67. let text = s.getAttr("title").split(" ")[0]
  68. case s.getAttr("href").split("/")[^1]
  69. of "followers": profile.followers = text
  70. of "following": profile.following = text
  71. else: profile.tweets = text
  72. proc getIntentStats*(profile: var Profile; node: XmlNode) =
  73. profile.tweets = "?"
  74. for s in node.querySelectorAll( "dd.count > a"):
  75. let text = s.innerText()
  76. case s.getAttr("href").split("/")[^1]
  77. of "followers": profile.followers = text
  78. of "following": profile.following = text
  79. proc getTweetStats*(tweet: Tweet; node: XmlNode) =
  80. tweet.replies = "0"
  81. tweet.retweets = "0"
  82. tweet.likes = "0"
  83. for action in node.querySelectorAll(".ProfileTweet-actionCountForAria"):
  84. let text = action.innerText.split()
  85. case text[1]
  86. of "replies": tweet.replies = text[0]
  87. of "likes": tweet.likes = text[0]
  88. of "retweets": tweet.retweets = text[0]
  89. proc getGif(player: XmlNode): Gif =
  90. let
  91. thumb = player.getAttr("style").replace(thumbRegex, "$1")
  92. id = thumb.replace(gifRegex, "$1")
  93. url = fmt"https://video.twimg.com/tweet_video/{id}.mp4"
  94. Gif(url: url, thumb: thumb)
  95. proc getTweetMedia*(tweet: Tweet; node: XmlNode) =
  96. for photo in node.querySelectorAll(".AdaptiveMedia-photoContainer"):
  97. tweet.photos.add photo.attrs["data-image-url"]
  98. let player = node.querySelector(".PlayableMedia")
  99. if player.isNil:
  100. return
  101. if "gif" in player.getAttr("class"):
  102. tweet.gif = some(getGif(player.querySelector(".PlayableMedia-player")))
  103. elif "video" in player.getAttr("class"):
  104. tweet.video = some(Video())
  105. proc getQuoteMedia*(quote: var Quote; node: XmlNode) =
  106. let sensitive = node.querySelector(".QuoteTweet--sensitive")
  107. if not sensitive.isNil:
  108. quote.sensitive = true
  109. return
  110. let media = node.querySelector(".QuoteMedia")
  111. if not media.isNil:
  112. quote.thumb = some(media.selectAttr("img", "src"))
  113. let badge = node.querySelector(".AdaptiveMedia-badgeText")
  114. let gifBadge = node.querySelector(".Icon--gifBadge")
  115. if not badge.isNil:
  116. quote.badge = some(badge.innerText())
  117. elif not gifBadge.isNil:
  118. quote.badge = some("GIF")