extension/packages/find-link.coffee

   1 utils = require 'utils'
   2
   3 # All the following elements qualify as a link
   4 LINK_ELEMENTS = [
   5   "a"
   6   "area[@href]"
   7   "button"
   8 ]
   9
  10 # All elements that have one or more of the following properties
  11 # qualify as a link
  12 LINK_ELEMENT_PROPERTIES = [
  13   "@onclick"
  14   "@onmousedown"
  15   "@onmouseup"
  16   "@oncommand"
  17   "@role='link'"
  18   "@role='button'"
  19   "contains(@class, 'button')"
  20 ]
  21
  22 # Find a link that match with the patterns
  23 findLinkMatchPattern = (document, patterns) ->
  24   links = getLinkElements(document)
  25   candidateLinks = []
  26
  27   # filter visible links that contain patterns and put in candidateLinks
  28   for i in [0...links.snapshotLength] by 1
  29     link = links.snapshotItem(i)
  30
  31     if isVisibleElement(link) and isElementMatchPattern(link, patterns)
  32       candidateLinks.push(link)
  33
  34   return if candidateLinks.length == 0
  35
  36   for link in candidateLinks
  37     link.wordCount = link.textContent.trim().split(/\s+/).length
  38
  39   # favor shorter links, links near the end of a page
  40   candidateLinks = candidateLinks.sort (a, b) ->
  41     if a.wordCount == b.wordCount then 1 else a.wordCount - b.wordCount
  42
  43   results = []
  44
  45   # match patterns. Sort them to match shorter patterns first.
  46   # The latter is to prevent matching first links that contain
  47   # longer words like `next`, `more`, etc.
  48   for pattern in patterns.sort((a, b) -> a.length > b.length)
  49     console.log(pattern)
  50     # if the pattern is a word, wrapped it in word boundaries.
  51     # thus we won't match words like 'previously' to 'previous'
  52     exactWordRegex =
  53       if /^\b|\b$/.test(pattern)
  54         new RegExp('\\b' + pattern + '\\b', 'i')
  55       else
  56         new RegExp(pattern, 'i')
  57
  58     for candidateLink in candidateLinks
  59       if exactWordRegex.test(candidateLink.textContent)
  60         results.push(candidateLink)
  61
  62   return results
  63
  64
  65 # Returns elements that qualify as links
  66 getLinkElements = do ->
  67   elements = [
  68     LINK_ELEMENTS...
  69     "*[#{ LINK_ELEMENT_PROPERTIES.join(' or ') }]"
  70   ]
  71
  72   return utils.getDomElements(elements)
  73
  74
  75 # Determine if the link is visible
  76 isVisibleElement = (element) ->
  77   document = element.ownerDocument
  78   window   = document.defaultView
  79
  80   # element that isn't visible on the page
  81   computedStyle = window.getComputedStyle(element, null)
  82   if computedStyle.getPropertyValue('visibility') != 'visible' or
  83       computedStyle.getPropertyValue('display') == 'none' or
  84       computedStyle.getPropertyValue('opacity') == '0'
  85     return false
  86
  87   # element that has zero dimension
  88   clientRect = element.getBoundingClientRect()
  89   if clientRect.width == 0 or clientRect.height == 0
  90     return false
  91
  92   return true
  93
  94
  95 # Determine if the link has a pattern matched
  96 isElementMatchPattern = (element, patterns) ->
  97   for pattern in patterns
  98     if element.textContent.toLowerCase().contains(pattern)
  99       return true
 100
 101   return false
 102
 103 exports.find = findLinkMatchPattern