Fix explosive runtime of detectCopiedCode()

Summary:
Fixes T5041. Pretty sure this is the issue: if a diff contains a large number of identical lines longer than 30 characters, we end up paying O(N^2) for each set.

Instead, when N > 16, opt to pay 0.

Test Plan: Added a test which dropped from ~100s to ~0 after changes (this diff includes a reduced-strenght version of the test, since parsing a 4,000 line diff is a little bit pricey).

Reviewers: btrahan

Reviewed By: btrahan

Subscribers: epriestley

Maniphest Tasks: T5041

Differential Revision: https://secure.phabricator.com/D9178

src/applications/differential/parser/DifferentialChangesetParser.php

@@ -1176,6 +1176,16 @@ final class DifferentialChangesetParser {
         $added = array_map('trim', $hunk->getAddedLines());
         for (reset($added); list($line, $code) = each($added); ) {
           if (isset($map[$code])) { // We found a long matching line.
+
+            if (count($map[$code]) > 16) {
+              // If there are a large number of identical lines in this diff,
+              // don't try to figure out where this block came from: the
+              // analysis is O(N^2), since we need to compare every line
+              // against every other line. Even if we arrive at a result, it
+              // is unlikely to be meaningful. See T5041.
+              continue 2;
+            }
+
             $best_length = 0;
             foreach ($map[$code] as $val) { // Explore all candidates.
               list($file, $orig_line) = $val;

src/applications/differential/storage/__tests__/DifferentialDiffTestCase.php

@@ -15,4 +15,41 @@ final class DifferentialDiffTestCase extends ArcanistPhutilTestCase {
       ipull($copies, 1));
   }
 
+  public function testDetectSlowCopiedCode() {
+    // This tests that the detector has a reasonable runtime when a diff
+    // contains a very large number of identical lines. See T5041.
+
+    $parser = new ArcanistDiffParser();
+
+    $line = str_repeat('x', 60);
+    $oline = '-'.$line."\n";
+    $nline = '+'.$line."\n";
+
+    $n = 1000;
+    $oblock = str_repeat($oline, $n);
+    $nblock = str_repeat($nline, $n);
+
+    $raw_diff = <<<EODIFF
+diff --git a/dst b/dst
+new file mode 100644
+index 0000000..1234567
+--- /dev/null
++++ b/dst
+@@ -0,0 +1,{$n} @@
+{$nblock}
+diff --git a/src b/src
+deleted file mode 100644
+index 123457..0000000
+--- a/src
++++ /dev/null
+@@ -1,{$n} +0,0 @@
+{$oblock}
+EODIFF;
+
+    $diff = DifferentialDiff::newFromRawChanges($parser->parseDiff($raw_diff));
+
+    $this->assertTrue(true);
+  }
+
+
 }