PDF TextPosition and TextQuote starting to work

TextPositionAnchor can be generated and re-anchored within a PDF so
long as the selection doesn't span a page boundary.

TextQuoteAnchor is generated, but not re-anchored.

h/static/scripts/annotator/anchoring/html.coffee

@@ -16,7 +16,11 @@
 # :return: A Promise that resolves to a Range on success.
 # :rtype: Promise
 ####
-exports.anchor = (selectors, options) ->
+exports.anchor = (selectors) ->
+  options =
+    root: document.body
+    ignoreSelector: '[class^="annotator-"]'
+
   # Selectors
   fragment = null
   position = null
@@ -48,34 +52,38 @@ exports.anchor = (selectors, options) ->
   if fragment?
     promise = promise.catch =>
       Promise.resolve(FragmentAnchor.fromSelector(fragment, options))
-      .then((a) -> Promise.resolve(a.toRange(options)))
+      .then((a) -> a.toRange(options))
       .then(assertQuote)
 
   if range?
     promise = promise.catch =>
       Promise.resolve(RangeAnchor.fromSelector(range, options))
-      .then((a) -> Promise.resolve(a.toRange(options)))
+      .then((a) -> a.toRange(options))
       .then(assertQuote)
 
   if position?
     promise = promise.catch =>
       Promise.resolve(TextPositionAnchor.fromSelector(position, options))
-      .then((a) -> Promise.resolve(a.toRange(options)))
+      .then((a) -> a.toRange(options))
       .then(assertQuote)
 
   if quote?
     promise = promise.catch =>
       Promise.resolve(TextQuoteAnchor.fromSelector(quote, options))
-      .then((a) -> Promise.resolve(a.toRange(options)))
+      .then((a) -> a.toRange(options))
 
   return promise
 
 
-exports.describe = (range, options) ->
+exports.describe = (range) ->
+  options =
+    root: document.body
+    ignoreSelector: '[class^="annotator-"]'
+
   maybeDescribeWith = (type) ->
     return Promise.resolve(type)
-    .then((t) -> Promise.resolve(t.fromRange(range, options)))
-    .then((a) -> Promise.resolve(a.toSelector(options)))
+    .then((t) -> t.fromRange(range, options))
+    .then((a) -> a.toSelector(options))
     .catch(-> null)
 
   selectors = (maybeDescribeWith(type) for type in [

h/static/scripts/annotator/anchoring/main.coffee

+if (window.PDFViewerApplication)
+  module.exports = require('./pdf')
+else
+  module.exports = require('./html')

h/static/scripts/annotator/anchoring/pdf.coffee

+seek = require('dom-seek')
+
+Annotator = require('annotator')
+xpathRange = Annotator.Range
+
+{
+  TextPositionAnchor
+  TextQuoteAnchor
+} = require('./types')
+
+
+getSiblingIndex = (node) ->
+  siblings = Array.prototype.slice.call(node.parentNode.childNodes)
+  return siblings.indexOf(node)
+
+
+getNodeTextLayer = (node) ->
+  until node.classList?.contains('page')
+    node = node.parentNode
+  return node.getElementsByClassName('textLayer')[0]
+
+
+getPageTextLayer = (pageIndex) ->
+  return PDFViewerApplication.pdfViewer.pages[pageIndex].textLayer
+
+
+getPageTextContent = (pageIndex) ->
+  return PDFViewerApplication.pdfViewer.getPageTextContent(pageIndex)
+
+
+getPageOffset = (pageIndex) ->
+  index = -1
+
+  next = (offset) ->
+    if ++index is pageIndex
+      return Promise.resolve(offset)
+
+    return getPageTextContent(index)
+    .then(getTextContentLength)
+    .then((length) -> next(offset + length))
+
+  return next(0)
+
+
+getTextContentLength = (textContent) ->
+  sum = 0
+  for item in textContent.items
+    sum += item.str.length
+  return sum
+
+
+findPage = (offset) ->
+  index = 0
+  total = 0
+
+  next = (length) ->
+    if total + length >= offset
+      offset = total
+      return Promise.resolve({index, offset})
+    else
+      index++
+      total += length
+      return count()
+
+  count = ->
+    return getPageTextContent(index)
+    .then(getTextContentLength)
+    .then(next)
+
+  return count()
+
+
+###*
+# Anchor a set of selectors.
+#
+# This function converts a set of selectors into a document range using.
+# It encapsulates the core anchoring algorithm, using the selectors alone or
+# in combination to establish the best anchor within the document.
+#
+# :param Array selectors: The selectors to try.
+# :return: A Promise that resolves to a Range on success.
+# :rtype: Promise
+####
+exports.anchor = (selectors) ->
+  options =
+    root: document.getElementById('viewer')
+    ignoreSelector: '[class^="annotator-"]'
+
+  # Selectors
+  position = null
+  quote = null
+
+  # Collect all the selectors
+  for selector in selectors ? []
+    switch selector.type
+      when 'TextPositionSelector'
+        position = selector
+      when 'TextQuoteSelector'
+        quote = selector
+
+  # Until we successfully anchor, we fail.
+  promise = Promise.reject('unable to anchor')
+
+  # Assert the quote matches the stored quote, if applicable
+  assertQuote = (range) ->
+    if quote?.exact? and range.toString() != quote.exact
+      throw new Error('quote mismatch')
+    else
+      return range
+
+  if position?
+    promise = promise.catch =>
+      findPage(position.start)
+      .then(({index, offset}) ->
+        textLayer = getPageTextLayer(index)
+        start = position.start - offset
+        end = position.end - offset
+        root = textLayer.textLayerDiv
+        Promise.resolve(TextPositionAnchor.fromSelector({start, end}, {root}))
+        .then((a) -> Promise.resolve(a.toRange({root})))
+      )
+      .then(assertQuote)
+
+  return promise
+
+
+exports.describe = (range) ->
+  range = new xpathRange.BrowserRange(range).normalize()
+
+  startTextLayer = getNodeTextLayer(range.start)
+  endTextLayer = getNodeTextLayer(range.end)
+
+  # XXX: range covers only one page
+  if startTextLayer isnt endTextLayer
+    throw new Error('selecting across page breaks is not supported')
+
+  startRange = range.limit(startTextLayer)
+  endRange = range.limit(endTextLayer)
+
+  startPageIndex = getSiblingIndex(startTextLayer.parentNode)
+  endPageIndex = getSiblingIndex(endTextLayer.parentNode)
+
+  iter = seek.createTextIterator(startTextLayer)
+
+  return iter.seek(range.start).then (start) ->
+    iter.seek(range.end).then (length) ->
+      end = start + length + range.end.textContent.length
+      getPageOffset(startPageIndex).then (offset) ->
+        # XXX: range covers only one page
+        start += offset
+        end += offset
+        position = new TextPositionAnchor(start, end).toSelector()
+
+        options = {root: startTextLayer}
+
+        r = document.createRange()
+        r.setStartBefore(startRange.start)
+        r.setEndAfter(endRange.end)
+
+        quote = Promise.resolve(TextQuoteAnchor.fromRange(r, options))
+        .then((a) -> a.toSelector())
+
+        return Promise.all([position, quote])

h/static/scripts/annotator/guest.coffee

@@ -3,7 +3,7 @@ Promise = global.Promise or require('es6-promise').Promise
 Annotator = require('annotator')
 $ = Annotator.$
 
-anchoring = require('./anchoring/html')
+anchoring = require('./anchoring/main')
 highlighter = require('./highlighter')
 
 
@@ -145,19 +145,17 @@ module.exports = class Guest extends Annotator
     this.removeEvents()
 
   setupAnnotation: (annotation) ->
-    {anchored, unanchored, plugins} = this
-    root = @element[0]
-    ignoreSelector = '[class^="annotator-"]'
+    {anchored, unanchored, element, plugins} = this
 
     maybeAnchor = (target) ->
-      anchoring.anchor(target.selector, {root, ignoreSelector})
+      anchoring.anchor(target.selector)
       .then(highlightRange)
       .then((highlights) -> {annotation, target, highlights})
       .catch((reason) -> {annotation, target, reason})
 
     highlightRange = (range) ->
       return new Promise(raf).then ->
-        normedRange = Annotator.Range.sniff(range).normalize(root)
+        normedRange = Annotator.Range.sniff(range).normalize(element[0])
         return highlighter.highlightRange(normedRange)
 
     storeAndSync = (results) ->
@@ -179,16 +177,12 @@ module.exports = class Guest extends Annotator
     annotation
 
   createAnnotation: (annotation = {}) ->
-    root = @element[0]
-    ignoreSelector = '[class^="annotator-"]'
-
-    options = {root, ignoreSelector}
     ranges = @selectedRanges
     @selectedRanges = null
 
     info = this.getDocumentInfo()
 
-    Promise.all(ranges.map((r) -> anchoring.describe(r, options)))
+    Promise.all(ranges.map(anchoring.describe))
     .then((targets) ->
       info.then((info) ->
         if info.metadata?

h/static/scripts/annotator/plugin/pdf.coffee

+raf = require('raf')
 Promise = require('es6-promise').Promise
 Annotator = require('annotator')
 
@@ -44,10 +45,21 @@ class PDF extends Annotator.Plugin
 
       return {title, link}
 
-  onpagerendered: =>
-    unanchored = @annotator.unanchored.splice(0, @annotator.unanchored.length)
+  onpagerendered: (event) =>
+    annotator = @annotator
+    unanchored = @annotator.unanchored
+    page = PDFViewerApplication.pdfViewer.pages[event.detail.pageNumber - 1]
+
+    waitForTextLayer = ->
+      unless (page.textLayer.renderingDone)
+        return new Promise(raf).then(waitForTextLayer)
+
+    reanchor = ->
+      unanchored = unanchored.splice(0, unanchored.length)
       for obj in unanchored
-      @annotator.setupAnnotation(obj.annotation)
+        annotator.setupAnnotation(obj.annotation)
+
+    waitForTextLayer().then(reanchor)
 
 Annotator.Plugin.PDF = PDF