public function testGenerateUTF8IntralineDiff()
{
// Both Strings Empty.
$left = '';
$right = '';
$result = array(array(array(0, 0)), array(array(0, 0)));
$this->assertEqual($result, ArcanistDiffUtils::generateIntralineDiff($left, $right));
// Left String Empty.
$left = '';
$right = "Grumpy☃at";
$result = array(array(array(0, 0)), array(array(0, 11)));
$this->assertEqual($result, ArcanistDiffUtils::generateIntralineDiff($left, $right));
// Right String Empty.
$left = "Grumpy☃at";
$right = '';
$result = array(array(array(0, 11)), array(array(0, 0)));
$this->assertEqual($result, ArcanistDiffUtils::generateIntralineDiff($left, $right));
// Both Strings Same
$left = "Grumpy☃at";
$right = "Grumpy☃at";
$result = array(array(array(0, 11)), array(array(0, 11)));
$this->assertEqual($result, ArcanistDiffUtils::generateIntralineDiff($left, $right));
// Both Strings are different.
$left = "Grumpy☃at";
$right = 'Smiling Dog';
$result = array(array(array(1, 11)), array(array(1, 11)));
$this->assertEqual($result, ArcanistDiffUtils::generateIntralineDiff($left, $right));
// String with one difference in the middle.
$left = 'GrumpyCat';
$right = "Grumpy☃at";
$result = array(array(array(0, 6), array(1, 1), array(0, 2)), array(array(0, 6), array(1, 3), array(0, 2)));
$this->assertEqual($result, ArcanistDiffUtils::generateIntralineDiff($left, $right));
// Differences in middle, not connected to each other.
$left = 'GrumpyCat';
$right = "Grumpy☃a☃t";
$result = array(array(array(0, 6), array(1, 2), array(0, 1)), array(array(0, 6), array(1, 7), array(0, 1)));
$this->assertEqual($result, ArcanistDiffUtils::generateIntralineDiff($left, $right));
// String with difference at the beginning.
$left = "GrumpyC☃t";
$right = "DrumpyC☃t";
$result = array(array(array(1, 1), array(0, 10)), array(array(1, 1), array(0, 10)));
$this->assertEqual($result, ArcanistDiffUtils::generateIntralineDiff($left, $right));
// String with difference at the end.
$left = "GrumpyC☃t";
$right = "GrumpyC☃P";
$result = array(array(array(0, 10), array(1, 1)), array(array(0, 10), array(1, 1)));
$this->assertEqual($result, ArcanistDiffUtils::generateIntralineDiff($left, $right));
// String with differences at the beginning and end.
$left = "GrumpyC☃t";
$right = "DrumpyC☃P";
$result = array(array(array(1, 1), array(0, 9), array(1, 1)), array(array(1, 1), array(0, 9), array(1, 1)));
$this->assertEqual($result, ArcanistDiffUtils::generateIntralineDiff($left, $right));
// This is a unicode combining character, "COMBINING DOUBLE TILDE".
$cc = "͠";
$left = 'Senor';
$right = "Sen{$cc}or";
$result = array(array(array(0, 2), array(1, 1), array(0, 2)), array(array(0, 2), array(1, 3), array(0, 2)));
$this->assertEqual($result, ArcanistDiffUtils::generateIntralineDiff($left, $right));
}
public function process()
{
$old = array();
$new = array();
$n = 0;
$this->old = array_reverse($this->old);
$this->new = array_reverse($this->new);
$whitelines = false;
$changed = false;
$skip_intra = array();
while (count($this->old) || count($this->new)) {
$o_desc = array_pop($this->old);
$n_desc = array_pop($this->new);
$oend = end($this->old);
if ($oend) {
$o_next = $oend['type'];
} else {
$o_next = null;
}
$nend = end($this->new);
if ($nend) {
$n_next = $nend['type'];
} else {
$n_next = null;
}
if ($o_desc) {
$o_type = $o_desc['type'];
} else {
$o_type = null;
}
if ($n_desc) {
$n_type = $n_desc['type'];
} else {
$n_type = null;
}
if ($o_type != null && $n_type == null) {
$old[] = $o_desc;
$new[] = null;
if ($n_desc) {
array_push($this->new, $n_desc);
}
$changed = true;
continue;
}
if ($n_type != null && $o_type == null) {
$old[] = null;
$new[] = $n_desc;
if ($o_desc) {
array_push($this->old, $o_desc);
}
$changed = true;
continue;
}
if ($this->whitespaceMode != self::WHITESPACE_SHOW_ALL) {
$similar = false;
switch ($this->whitespaceMode) {
case self::WHITESPACE_IGNORE_TRAILING:
if (rtrim($o_desc['text']) == rtrim($n_desc['text'])) {
if ($o_desc['type']) {
// If we're converting this into an unchanged line because of
// a trailing whitespace difference, mark it as a whitespace
// change so we can show "This file was modified only by
// adding or removing trailing whitespace." instead of
// "This file was not modified.".
$whitelines = true;
}
$similar = true;
}
break;
default:
// In this case, the lines are similar if there is no change type
// (that is, just trust the diff algorithm).
if (!$o_desc['type']) {
$similar = true;
}
break;
}
if ($similar) {
$o_desc['type'] = null;
$n_desc['type'] = null;
$skip_intra[count($old)] = true;
} else {
$changed = true;
}
} else {
$changed = true;
}
$old[] = $o_desc;
$new[] = $n_desc;
}
$this->old = $old;
$this->new = $new;
$unchanged = false;
if ($this->subparser) {
$unchanged = $this->subparser->isUnchanged();
$whitelines = $this->subparser->isWhitespaceOnly();
} else {
if (!$changed) {
$filetype = $this->changeset->getFileType();
if ($filetype == DifferentialChangeType::FILE_TEXT || $filetype == DifferentialChangeType::FILE_SYMLINK) {
$unchanged = true;
}
}
}
$this->specialAttributes = array(self::ATTR_UNCHANGED => $unchanged, self::ATTR_DELETED => array_filter($this->old) && !array_filter($this->new), self::ATTR_WHITELINES => $whitelines);
if ($this->isSubparser) {
// The rest of this function deals with formatting the diff for display;
// we can exit early if we're a subparser and avoid doing extra work.
return;
}
if ($this->subparser) {
// Use this parser's side-by-side line information -- notably, the
// change types -- but replace all the line text with the subparser's.
// This lets us render whitespace-only changes without marking them as
// different.
$old = $this->old;
$new = $this->new;
$old_text = ipull($this->subparser->old, 'text', 'line');
$new_text = ipull($this->subparser->new, 'text', 'line');
foreach ($old as $k => $desc) {
if (empty($desc)) {
continue;
}
$old[$k]['text'] = idx($old_text, $desc['line']);
}
foreach ($new as $k => $desc) {
if (empty($desc)) {
continue;
}
$new[$k]['text'] = idx($new_text, $desc['line']);
// If there's a corresponding "old" text and the line is marked as
// unchanged, test if there are internal whitespace changes between
// non-whitespace characters, e.g. spaces added to a string or spaces
// added around operators. If we find internal spaces, mark the line
// as changed.
//
// We only need to do this for "new" lines because any line that is
// missing either "old" or "new" text certainly can not have internal
// whitespace changes without also having non-whitespace changes,
// because characters had to be either added or removed to create the
// possibility of internal whitespace.
if (isset($old[$k]['text']) && empty($new[$k]['type'])) {
if (trim($old[$k]['text']) != trim($new[$k]['text'])) {
// The strings aren't the same when trimmed, so there are internal
// whitespace changes. Mark this line changed.
$old[$k]['type'] = '-';
$new[$k]['type'] = '+';
}
}
}
$this->old = $old;
$this->new = $new;
}
$min_length = min(count($this->old), count($this->new));
for ($ii = 0; $ii < $min_length; $ii++) {
if ($this->old[$ii] || $this->new[$ii]) {
if (isset($this->old[$ii]['text'])) {
$otext = $this->old[$ii]['text'];
} else {
$otext = '';
}
if (isset($this->new[$ii]['text'])) {
$ntext = $this->new[$ii]['text'];
} else {
$ntext = '';
}
if ($otext != $ntext && empty($skip_intra[$ii])) {
$this->intra[$ii] = ArcanistDiffUtils::generateIntralineDiff($otext, $ntext);
}
}
}
$lines_context = self::LINES_CONTEXT;
$max_length = max(count($this->old), count($this->new));
$old = $this->old;
$new = $this->new;
$visible = false;
$last = 0;
for ($cursor = -$lines_context; $cursor < $max_length; $cursor++) {
$offset = $cursor + $lines_context;
if (isset($old[$offset]) && $old[$offset]['type'] || isset($new[$offset]) && $new[$offset]['type']) {
$visible = true;
$last = $offset;
} else {
if ($cursor > $last + $lines_context) {
$visible = false;
}
}
if ($visible && $cursor > 0) {
$this->visible[$cursor] = 1;
}
}
// NOTE: Micro-optimize a couple of ipull()s here since it gives us a
// 10% performance improvement for certain types of large diffs like
// Phriction changes.
$old_corpus = array();
foreach ($this->old as $o) {
$old_corpus[] = $o['text'];
}
$old_corpus_block = implode("\n", $old_corpus);
$new_corpus = array();
foreach ($this->new as $n) {
$new_corpus[] = $n['text'];
}
$new_corpus_block = implode("\n", $new_corpus);
$old_future = $this->getHighlightFuture($old_corpus_block);
$new_future = $this->getHighlightFuture($new_corpus_block);
$futures = array('old' => $old_future, 'new' => $new_future);
foreach (Futures($futures) as $key => $future) {
try {
switch ($key) {
case 'old':
$this->oldRender = $this->processHighlightedSource($this->old, $future->resolve());
break;
case 'new':
$this->newRender = $this->processHighlightedSource($this->new, $future->resolve());
break;
}
} catch (Exception $ex) {
phlog($ex);
throw $ex;
}
}
$this->applyIntraline($this->oldRender, ipull($this->intra, 0), $old_corpus);
$this->applyIntraline($this->newRender, ipull($this->intra, 1), $new_corpus);
$generated = strpos($new_corpus_block, '@' . 'generated') !== false;
$this->specialAttributes[self::ATTR_GENERATED] = $generated;
}
public function generateIntraLineDiffs()
{
$old = $this->getOldLines();
$new = $this->getNewLines();
$diffs = array();
foreach ($old as $key => $o) {
$n = $new[$key];
if (!$o || !$n) {
continue;
}
if ($o['type'] != $n['type']) {
$diffs[$key] = ArcanistDiffUtils::generateIntralineDiff($o['text'], $n['text']);
}
}
$this->setIntraLineDiffs($diffs);
return $this;
}