Merge pull request #790 from wordpress-mobile/feature/performance-improvement

Performance improvement on parsing from html

Author: hypest

aztec/src/main/kotlin/org/wordpress/aztec/AztecParser.kt

@@ -22,6 +22,7 @@ import android.content.Context
 import android.support.v4.text.TextDirectionHeuristicsCompat
 import android.text.Editable
 import android.text.Spannable
+import android.text.SpannableString
 import android.text.SpannableStringBuilder
 import android.text.Spanned
 import android.text.TextUtils
@@ -57,6 +58,21 @@ import java.util.Comparator
 
 class AztecParser @JvmOverloads constructor(val plugins: List<IAztecPlugin> = listOf(),
                                             private val ignoredTags: List<String> = listOf("body", "html")) {
+    /**
+     * A faster version of fromHtml(), intended for inspecting the span structure only. It doesn't prepare the text for
+     * visual editing.
+     */
+    @Suppress("unused") // this method is used in wpandroid so, suppress the inspection
+    fun parseHtmlForInspection(source: String, context: Context): Spanned {
+        val tidySource = tidy(source)
+
+        val spanned = SpannableString(Html.fromHtml(tidySource,
+                AztecTagHandler(context, plugins), context, plugins, ignoredTags))
+
+        postprocessSpans(spanned)
+
+        return spanned
+    }
 
     fun fromHtml(source: String, context: Context): Spanned {
         val tidySource = tidy(source)
@@ -117,7 +133,7 @@ class AztecParser @JvmOverloads constructor(val plugins: List<IAztecPlugin> = li
         return html
     }
 
-    private fun postprocessSpans(spannable: SpannableStringBuilder) {
+    private fun postprocessSpans(spannable: Spannable) {
         plugins.filter { it is ISpanPostprocessor }
             .map { it as ISpanPostprocessor }
             .forEach {
@@ -236,7 +252,7 @@ class AztecParser @JvmOverloads constructor(val plugins: List<IAztecPlugin> = li
     }
 
     // Always try to put a visual newline before block elements and only put one after if needed
-    fun syncVisualNewlinesOfBlockElements(spanned: Editable) {
+    fun syncVisualNewlinesOfBlockElements(spanned: Spannable) {
         // clear any visual newline marking. We'll mark them with a fresh set of passes
         spanned.getSpans(0, spanned.length, AztecVisualLinebreak::class.java).forEach {
             spanned.removeSpan(it)

aztec/src/main/kotlin/org/wordpress/aztec/AztecTagHandler.kt

@@ -53,6 +53,9 @@ import java.util.ArrayList
 class AztecTagHandler(val context: Context, val plugins: List<IAztecPlugin> = ArrayList()) : Html.TagHandler {
     private val loadingDrawable: Drawable
 
+    // Simple LIFO stack to track the html tag nesting for easy reference when we need to handle the ending of a tag
+    private val tagStack = mutableListOf<Any>()
+
     init {
         val styles = context.obtainStyledAttributes(R.styleable.AztecText)
         loadingDrawable = ContextCompat.getDrawable(context, styles.getResourceId(R.styleable.AztecText_drawableLoading, R.drawable.ic_image_loading))!!
@@ -163,8 +166,8 @@ class AztecTagHandler(val context: Context, val plugins: List<IAztecPlugin> = Ar
             start(output, AztecMediaClickableSpan(mediaSpan))
             output.append(Constants.IMG_CHAR)
         } else {
-            end(output, mediaSpan.javaClass)
             end(output, AztecMediaClickableSpan::class.java)
+            end(output, mediaSpan.javaClass)
         }
     }
 
@@ -177,11 +180,24 @@ class AztecTagHandler(val context: Context, val plugins: List<IAztecPlugin> = Ar
     }
 
     private fun start(output: Editable, mark: Any) {
+        tagStack.add(mark)
+
         output.setSpan(mark, output.length, output.length, Spanned.SPAN_MARK_MARK)
     }
 
     private fun end(output: Editable, kind: Class<*>) {
-        val last = output.getLast(kind)
+        // Get most recent tag type from the stack instead of `getLast()`. This is a speed optimization as `getLast()`
+        //  doesn't know that the tags are in fact nested and in pairs (since it's html), including empty html elements
+        //  (they are treated as pairs by tagsoup anyway).
+        val last = if (tagStack.size > 0 && kind.equals(tagStack[tagStack.size - 1].javaClass)) {
+            tagStack.removeAt(tagStack.size - 1) // remove and return the top mark on the stack
+        } else {
+            // Warning: the tags stack is apparently inconsistent at this point
+
+            // fall back to getting the last tag type from the Spannable
+            output.getLast(kind)
+        }
+
         val start = output.getSpanStart(last)
         val end = output.length
 

aztec/src/main/kotlin/org/wordpress/aztec/plugins/CssUnderlinePlugin.kt

@@ -1,5 +1,6 @@
 package org.wordpress.aztec.plugins
 
+import android.text.Spannable
 import android.text.SpannableStringBuilder
 import android.text.Spanned
 import org.wordpress.aztec.plugins.html2visual.ISpanPostprocessor
@@ -50,7 +51,7 @@ class CssUnderlinePlugin : ISpanPostprocessor, ISpanPreprocessor {
         }
     }
 
-    override fun afterSpansProcessed(spannable: SpannableStringBuilder) {
+    override fun afterSpansProcessed(spannable: Spannable) {
         spannable.getSpans(0, spannable.length, HiddenHtmlSpan::class.java).forEach {
             if (it.TAG == SPAN_TAG && CssStyleFormatter.containsStyleAttribute(it.attributes, CssStyleFormatter.CSS_TEXT_DECORATION_ATTRIBUTE)) {
                 CssStyleFormatter.removeStyleAttribute(it.attributes, CssStyleFormatter.CSS_TEXT_DECORATION_ATTRIBUTE)
@@ -62,4 +63,4 @@ class CssUnderlinePlugin : ISpanPostprocessor, ISpanPreprocessor {
             }
         }
     }
-}
+}

aztec/src/main/kotlin/org/wordpress/aztec/plugins/html2visual/ISpanPostprocessor.kt

@@ -1,8 +1,8 @@
 package org.wordpress.aztec.plugins.html2visual
 
-import android.text.SpannableStringBuilder
+import android.text.Spannable
 import org.wordpress.aztec.plugins.IAztecPlugin
 
 interface ISpanPostprocessor : IAztecPlugin {
-    fun afterSpansProcessed(spannable: SpannableStringBuilder)
-}
+    fun afterSpansProcessed(spannable: Spannable)
+}

aztec/src/main/kotlin/org/wordpress/aztec/source/Format.kt

@@ -298,7 +298,7 @@ object Format {
     }
 
     @JvmStatic
-    fun preProcessSpannedText(text: SpannableStringBuilder, isCalypsoFormat: Boolean) {
+    fun preProcessSpannedText(text: Spannable, isCalypsoFormat: Boolean) {
         if (isCalypsoFormat) {
             text.getSpans(0, text.length, AztecVisualLinebreak::class.java).forEach {
                 val spanStart = text.getSpanStart(it)