Improve performance of generating synthetic changesets

Summary: Ref T7776. This could get better, but I think I got most of the big stuff. It's ~4x faster now.

Test Plan:
Before:

{F393338}

After:

{F393339}

Reviewers: btrahan

Reviewed By: btrahan

Subscribers: epriestley

Maniphest Tasks: T7776

Differential Revision: https://secure.phabricator.com/D12730

src/applications/differential/parser/DifferentialHunkParser.php

@@ -418,9 +418,7 @@ final class DifferentialHunkParser {
     $old_lines = array();
     $new_lines = array();
     foreach ($hunks as $hunk) {
-
+      $lines = $hunk->getSplitLines();
-      $lines = $hunk->getChanges();
-      $lines = phutil_split_lines($lines);
 
       $line_type_map = array();
       $line_text = array();
@@ -514,22 +512,32 @@ final class DifferentialHunkParser {
     // Put changes side by side.
     $olds = array();
     $news = array();
+    $olds_cursor = -1;
+    $news_cursor = -1;
     foreach ($changeset_hunks as $hunk) {
       $n_old = $hunk->getOldOffset();
       $n_new = $hunk->getNewOffset();
-      $changes = phutil_split_lines($hunk->getChanges());
+      $changes = $hunk->getSplitLines();
       foreach ($changes as $line) {
         $diff_type = $line[0]; // Change type in diff of diffs.
         $orig_type = $line[1]; // Change type in the original diff.
         if ($diff_type == ' ') {
           // Use the same key for lines that are next to each other.
-          $key = max(last_key($olds), last_key($news)) + 1;
+          if ($olds_cursor > $news_cursor) {
+            $key = $olds_cursor + 1;
+          } else {
+            $key = $news_cursor + 1;
+          }
           $olds[$key] = null;
           $news[$key] = null;
+          $olds_cursor = $key;
+          $news_cursor = $key;
         } else if ($diff_type == '-') {
           $olds[] = array($n_old, $orig_type);
+          $olds_cursor++;
         } else if ($diff_type == '+') {
           $news[] = array($n_new, $orig_type);
+          $news_cursor++;
         }
         if (($diff_type == '-' || $diff_type == ' ') && $orig_type != '-') {
           $n_old++;
@@ -660,11 +668,15 @@ final class DifferentialHunkParser {
     $offsets = array();
     $n = 1;
     foreach ($hunks as $hunk) {
-      for ($i = 0; $i < $hunk->getNewLen(); $i++) {
+      $new_length = $hunk->getNewLen();
-        $offsets[$n] = $hunk->getNewOffset() + $i;
+      $new_offset = $hunk->getNewOffset();
+
+      for ($i = 0; $i < $new_length; $i++) {
+        $offsets[$n] = $new_offset + $i;
         $n++;
       }
     }
+
     return $offsets;
   }
 }

src/applications/differential/storage/DifferentialHunk.php

@@ -10,6 +10,7 @@ abstract class DifferentialHunk extends DifferentialDAO
   protected $newLen;
 
   private $changeset;
+  private $splitLines;
   private $structuredLines;
   private $structuredFiles = array();
 
@@ -109,9 +110,16 @@ abstract class DifferentialHunk extends DifferentialDAO
     return $this->structuredFiles[$kind];
   }
 
+  public function getSplitLines() {
+    if ($this->splitLines === null) {
+      $this->splitLines = phutil_split_lines($this->getChanges());
+    }
+    return $this->splitLines;
+  }
+
   private function getStructuredLines() {
     if ($this->structuredLines === null) {
-      $lines = phutil_split_lines($this->getChanges());
+      $lines = $this->getSplitLines();
 
       $structured = array();
       foreach ($lines as $line) {
@@ -154,13 +162,20 @@ abstract class DifferentialHunk extends DifferentialDAO
   }
 
   final private function makeContent($include) {
+    $lines = $this->getSplitLines();
     $results = array();
-    $lines = explode("\n", $this->getChanges());
 
+    $include_map = array();
+    for ($ii = 0; $ii < strlen($include); $ii++) {
+      $include_map[$include[$ii]] = true;
+    }
+
+    if (isset($include_map['+'])) {
+      $n = $this->newOffset;
+    } else {
+      $n = $this->oldOffset;
+    }
 
-    $n = (strpos($include, '+') !== false ?
-      $this->newOffset :
-      $this->oldOffset);
     $use_next_newline = false;
     foreach ($lines as $line) {
       if (!isset($line[0])) {
@@ -171,14 +186,14 @@ abstract class DifferentialHunk extends DifferentialDAO
         if ($use_next_newline) {
           $results[last_key($results)] = rtrim(end($results), "\n");
         }
-      } else if (strpos($include, $line[0]) === false) {
+      } else if (empty($include_map[$line[0]])) {
         $use_next_newline = false;
       } else {
         $use_next_newline = true;
-        $results[$n] = substr($line, 1)."\n";
+        $results[$n] = substr($line, 1);
       }
 
-      if ($line[0] == ' ' || strpos($include, $line[0]) !== false) {
+      if ($line[0] == ' ' || isset($include_map[$line[0]])) {
         $n++;
       }
     }