public function testUTF8v()
{
$strings = array('' => array(), 'x' => array('x'), 'quack' => array('q', 'u', 'a', 'c', 'k'), "x東y" => array('x', "東", 'y'), "x͠y" => array("x", "͠", 'y'));
foreach ($strings as $str => $expect) {
$this->assertEqual($expect, phutil_utf8v($str), 'Vector of ' . $str);
}
}
public function render()
{
$viewer = $this->getViewer();
$rows = array();
foreach ($this->events as $event) {
// Limit display log size. If a daemon gets stuck in an output loop this
// page can be like >100MB if we don't truncate stuff. Try to do cheap
// line-based truncation first, and fall back to expensive UTF-8 character
// truncation if that doesn't get things short enough.
$message = $event->getMessage();
$more = null;
if (!$this->showFullMessage) {
$more_lines = null;
$more_chars = null;
$line_limit = 12;
if (substr_count($message, "\n") > $line_limit) {
$message = explode("\n", $message);
$more_lines = count($message) - $line_limit;
$message = array_slice($message, 0, $line_limit);
$message = implode("\n", $message);
}
$char_limit = 8192;
if (strlen($message) > $char_limit) {
$message = phutil_utf8v($message);
$more_chars = count($message) - $char_limit;
$message = array_slice($message, 0, $char_limit);
$message = implode('', $message);
}
if ($more_chars) {
$more = new PhutilNumber($more_chars);
$more = pht('Show %d more character(s)...', $more);
} else {
if ($more_lines) {
$more = new PhutilNumber($more_lines);
$more = pht('Show %d more line(s)...', $more);
}
}
if ($more) {
$id = $event->getID();
$more = array("\n...\n", phutil_tag('a', array('href' => "/daemon/event/{$id}/"), $more));
}
}
$row = array($event->getLogType(), phabricator_date($event->getEpoch(), $viewer), phabricator_time($event->getEpoch(), $viewer), array($message, $more));
if ($this->combinedLog) {
array_unshift($row, phutil_tag('a', array('href' => '/daemon/log/' . $event->getLogID() . '/'), pht('Daemon %s', $event->getLogID())));
}
$rows[] = $row;
}
$classes = array('', '', 'right', 'wide prewrap');
$headers = array('Type', 'Date', 'Time', 'Message');
if ($this->combinedLog) {
array_unshift($classes, 'pri');
array_unshift($headers, 'Daemon');
}
$log_table = new AphrontTableView($rows);
$log_table->setHeaders($headers);
$log_table->setColumnClasses($classes);
return $log_table->render();
}
public function render()
{
$rows = array();
if (!$this->user) {
throw new Exception("Call setUser() before rendering!");
}
foreach ($this->events as $event) {
// Limit display log size. If a daemon gets stuck in an output loop this
// page can be like >100MB if we don't truncate stuff. Try to do cheap
// line-based truncation first, and fall back to expensive UTF-8 character
// truncation if that doesn't get things short enough.
$message = $event->getMessage();
$more_lines = null;
$more_chars = null;
$line_limit = 12;
if (substr_count($message, "\n") > $line_limit) {
$message = explode("\n", $message);
$more_lines = count($message) - $line_limit;
$message = array_slice($message, 0, $line_limit);
$message = implode("\n", $message);
}
$char_limit = 8192;
if (strlen($message) > $char_limit) {
$message = phutil_utf8v($message);
$more_chars = count($message) - $char_limit;
$message = array_slice($message, 0, $char_limit);
$message = implode('', $message);
}
$more = null;
if ($more_chars) {
$more = number_format($more_chars);
$more = "\n<... {$more} more characters ...>";
} else {
if ($more_lines) {
$more = number_format($more_lines);
$more = "\n<... {$more} more lines ...>";
}
}
$row = array(phutil_escape_html($event->getLogType()), phabricator_date($event->getEpoch(), $this->user), phabricator_time($event->getEpoch(), $this->user), str_replace("\n", '<br />', phutil_escape_html($message . $more)));
if ($this->combinedLog) {
array_unshift($row, phutil_render_tag('a', array('href' => '/daemon/log/' . $event->getLogID() . '/'), phutil_escape_html('Daemon ' . $event->getLogID())));
}
$rows[] = $row;
}
$classes = array('', '', 'right', 'wide wrap');
$headers = array('Type', 'Date', 'Time', 'Message');
if ($this->combinedLog) {
array_unshift($classes, 'pri');
array_unshift($headers, 'Daemon');
}
$log_table = new AphrontTableView($rows);
$log_table->setHeaders($headers);
$log_table->setColumnClasses($classes);
return $log_table->render();
}
public function getHighlightFuture($source)
{
$color = 0;
$colors = array('rbw_r', 'rbw_o', 'rbw_y', 'rbw_g', 'rbw_b', 'rbw_i', 'rbw_v');
$result = array();
foreach (phutil_utf8v($source) as $character) {
if ($character == ' ' || $character == "\n") {
$result[] = $character;
continue;
}
$result[] = '<span class="' . $colors[$color] . '">' . phutil_escape_html($character) . '</span>';
$color = ($color + 1) % count($colors);
}
$result = implode('', $result);
return new ImmediateFuture($result);
}
public function getHighlightFuture($source)
{
$color = 0;
$colors = array('rbw_r', 'rbw_o', 'rbw_y', 'rbw_g', 'rbw_b', 'rbw_i', 'rbw_v');
$result = array();
foreach (phutil_utf8v($source) as $character) {
if ($character == ' ' || $character == "\n") {
$result[] = $character;
continue;
}
$result[] = phutil_tag('span', array('class' => $colors[$color]), $character);
$color = ($color + 1) % count($colors);
}
$result = phutil_implode_html('', $result);
return new ImmediateFuture($result);
}
public function getStringParts()
{
$input_text = $this->inputText;
$text_array = phutil_utf8v($input_text);
for ($ii = 0; $ii < count($text_array); $ii++) {
$char = $text_array[$ii];
$char_hex = bin2hex($char);
if (array_key_exists($char, self::$invisibleChars)) {
$text_array[$ii] = array('special' => true, 'value' => '<' . self::$invisibleChars[$char] . '>');
} else {
if (ord($char) < 32) {
$text_array[$ii] = array('special' => true, 'value' => '<0x' . $char_hex . '>');
} else {
$text_array[$ii] = array('special' => false, 'value' => $char);
}
}
}
return $text_array;
}
public function correctSpelling($input, array $options)
{
$matrix = $this->getEditDistanceMatrix();
if (!$matrix) {
throw new PhutilInvalidStateException('setEditDistanceMatrix');
}
$max_distance = $this->getMaximumDistance();
if (!$max_distance) {
throw new PhutilInvalidStateException('setMaximumDistance');
}
$input = $this->normalizeString($input);
foreach ($options as $key => $option) {
$options[$key] = $this->normalizeString($option);
}
$distances = array();
$inputv = phutil_utf8v($input);
foreach ($options as $option) {
$optionv = phutil_utf8v($option);
$matrix->setSequences($optionv, $inputv);
$distances[$option] = $matrix->getEditDistance();
}
asort($distances);
$best = min($max_distance, head($distances));
foreach ($distances as $option => $distance) {
if ($distance > $best) {
unset($distances[$option]);
}
}
// Before filtering, check if we have multiple equidistant matches and
// return them if we do. This prevents us from, e.g., matching "alnd" with
// both "land" and "amend", then dropping "land" for being too short, and
// incorrectly completing to "amend".
if (count($distances) > 1) {
return array_keys($distances);
}
foreach ($distances as $option => $distance) {
if (phutil_utf8_strlen($option) < $distance) {
unset($distances[$option]);
}
}
return array_keys($distances);
}
private function wrapICSLine($line)
{
$out = array();
$buf = '';
// NOTE: The line may contain sequences of combining characters which are
// more than 80 bytes in length. If it does, we'll split them in the
// middle of the sequence. This is okay and generally anticipated by
// RFC5545, which even allows implementations to split multibyte
// characters. The sequence will be stitched back together properly by
// whatever is parsing things.
foreach (phutil_utf8v($line) as $character) {
// If adding this character would bring the line over 75 bytes, start
// a new line.
if (strlen($buf) + strlen($character) > 75) {
$out[] = $buf . "\r\n";
$buf = ' ';
}
$buf .= $character;
}
$out[] = $buf . "\r\n";
return implode('', $out);
}
/**
* Split a UTF-8 string into an array of characters. Combining characters
* are not split.
*
* @param string A valid utf-8 string.
* @return list A list of characters in the string.
*/
function phutil_utf8v_combined($string)
{
$components = phutil_utf8v($string);
$array_length = count($components);
// If the first character in the string is a combining character,
// prepend a space to the string.
if ($array_length > 0 && phutil_utf8_is_combining_character($components[0])) {
$string = ' ' . $string;
$components = phutil_utf8v($string);
$array_length++;
}
for ($index = 1; $index < $array_length; $index++) {
if (phutil_utf8_is_combining_character($components[$index])) {
$components[$index - 1] = $components[$index - 1] . $components[$index];
unset($components[$index]);
$components = array_values($components);
$index--;
$array_length = count($components);
}
}
return $components;
}
/**
* Find the words which are part of the query string, and bold them in a
* result string. This makes it easier for users to see why a result
* matched their query.
*/
private function emboldenQuery($str)
{
$query = $this->query->getParameter('query');
if (!strlen($query) || !strlen($str)) {
return $str;
}
// This algorithm is safe but not especially fast, so don't bother if
// we're dealing with a lot of data. This mostly prevents silly/malicious
// queries from doing anything bad.
if (strlen($query) + strlen($str) > 2048) {
return $str;
}
// Keep track of which characters we're going to make bold. This is
// byte oriented, but we'll make sure we don't put a bold in the middle
// of a character later.
$bold = array_fill(0, strlen($str), false);
// Split the query into words.
$parts = preg_split('/ +/', $query);
// Find all occurrences of each word, and mark them to be emboldened.
foreach ($parts as $part) {
$part = trim($part);
$part = trim($part, '"+');
if (!strlen($part)) {
continue;
}
$matches = null;
$has_matches = preg_match_all('/(?:^|\\b)(' . preg_quote($part, '/') . ')/i', $str, $matches, PREG_OFFSET_CAPTURE);
if (!$has_matches) {
continue;
}
// Flag the matching part of the range for boldening.
foreach ($matches[1] as $match) {
$offset = $match[1];
for ($ii = 0; $ii < strlen($match[0]); $ii++) {
$bold[$offset + $ii] = true;
}
}
}
// Split the string into ranges, applying bold styling as required.
$out = array();
$buf = '';
$pos = 0;
$is_bold = false;
foreach (phutil_utf8v($str) as $chr) {
if ($bold[$pos] != $is_bold) {
if (strlen($buf)) {
if ($is_bold) {
$out[] = phutil_tag('strong', array(), $buf);
} else {
$out[] = $buf;
}
$buf = '';
}
$is_bold = !$is_bold;
}
$buf .= $chr;
$pos += strlen($chr);
}
if (strlen($buf)) {
if ($is_bold) {
$out[] = phutil_tag('strong', array(), $buf);
} else {
$out[] = $buf;
}
}
return $out;
}
private function unescapeTextValue($data)
{
$result = array();
$buf = '';
$esc = false;
foreach (phutil_utf8v($data) as $c) {
if (!$esc) {
if ($c == '\\') {
$esc = true;
} else {
if ($c == ',') {
$result[] = $buf;
$buf = '';
} else {
$buf .= $c;
}
}
} else {
switch ($c) {
case 'n':
case 'N':
$buf .= "\n";
break;
default:
$buf .= $c;
break;
}
}
}
if ($esc) {
$this->raiseParseFailure(self::PARSE_UNESCAPED_BACKSLASH, pht('ICS document contains TEXT value ending with unescaped ' . 'backslash.'));
}
$result[] = $buf;
return $result;
}
protected function withNgramsConstraint(PhabricatorSearchNgrams $index, $value)
{
if (strlen($value)) {
$this->ngrams[] = array('index' => $index, 'value' => $value, 'length' => count(phutil_utf8v($value)));
}
return $this;
}
/**
* Format the log string, replacing "%x" variables with values.
*
* @return string Finalized, log string for writing to disk.
* @task internals
*/
private function format()
{
// Always convert '%%' to literal '%'.
$map = array('%' => '%') + $this->data;
$result = '';
$saw_percent = false;
foreach (phutil_utf8v($this->format) as $c) {
if ($saw_percent) {
$saw_percent = false;
if (array_key_exists($c, $map)) {
$result .= addcslashes($map[$c], ".. \\..ÿ");
} else {
$result .= '-';
}
} else {
if ($c == '%') {
$saw_percent = true;
} else {
$result .= $c;
}
}
}
return rtrim($result) . "\n";
}
/**
* Hard-wrap a piece of UTF-8 text with embedded HTML tags and entities.
*
* @param string An HTML string with tags and entities.
* @return string Hard-wrapped string.
*/
protected function lineWrap($line)
{
$c = 0;
$break_here = array();
// Convert the UTF-8 string into a list of UTF-8 characters.
$vector = phutil_utf8v($line);
$len = count($vector);
$byte_pos = 0;
for ($ii = 0; $ii < $len; ++$ii) {
// An ampersand indicates an HTML entity; consume the whole thing (until
// ";") but treat it all as one character.
if ($vector[$ii] == '&') {
do {
++$ii;
} while ($vector[$ii] != ';');
++$c;
// An "<" indicates an HTML tag, consume the whole thing but don't treat
// it as a character.
} else {
if ($vector[$ii] == '<') {
do {
++$ii;
} while ($vector[$ii] != '>');
} else {
++$c;
}
}
// Keep track of where we need to break the string later.
if ($c == $this->lineWidth) {
$break_here[$ii] = true;
$c = 0;
}
}
$result = array();
foreach ($vector as $ii => $char) {
$result[] = $char;
if (isset($break_here[$ii])) {
$result[] = "<span class=\"over-the-line\">⬅</span><br />";
}
}
return implode('', $result);
}
/**
* Shorten a string to provide a summary, respecting UTF-8 characters. This
* function attempts to truncate strings at word boundaries.
*
* NOTE: This function makes a best effort to apply some reasonable rules but
* will not work well for the full range of unicode languages. For instance,
* no effort is made to deal with combining characters.
*
* @param string UTF-8 string to shorten.
* @param int Maximum length of the result.
* @param string If the string is shortened, add this at the end. Defaults to
* horizontal ellipsis.
* @return string A string with no more than the specified character length.
*/
function phutil_utf8_shorten($string, $length, $terminal = "…")
{
$terminal_len = count(phutil_utf8v($terminal));
if ($terminal_len >= $length) {
// If you provide a terminal we still enforce that the result (including
// the terminal) is no longer than $length, but we can't do that if the
// terminal is too long.
throw new Exception("String terminal length must be less than string length!");
}
$string_v = phutil_utf8v($string);
$string_len = count($string_v);
if ($string_len <= $length) {
// If the string is already shorter than the requested length, simply return
// it unmodified.
return $string;
}
// NOTE: This is not complete, and there are many other word boundary
// characters and reasonable places to break words in the UTF-8 character
// space. For now, this gives us reasonable behavior for latin langauges. We
// don't necessarily have access to PCRE+Unicode so there isn't a great way
// for us to look up character attributes.
// If we encounter these, prefer to break on them instead of cutting the
// string off in the middle of a word.
static $break_characters = array(' ' => true, "\n" => true, ';' => true, ':' => true, '[' => true, '(' => true, ',' => true, '-' => true);
// If we encounter these, shorten to this character exactly without appending
// the terminal.
static $stop_characters = array('.' => true, '!' => true, '?' => true);
// Search backward in the string, looking for reasonable places to break it.
$word_boundary = null;
$stop_boundary = null;
// If we do a word break with a terminal, we have to look beyond at least the
// number of characters in the terminal.
$terminal_area = $length - $terminal_len;
for ($ii = $length; $ii >= 0; $ii--) {
$c = $string_v[$ii];
if (isset($break_characters[$c]) && $ii <= $terminal_area) {
$word_boundary = $ii;
} else {
if (isset($stop_characters[$c]) && $ii < $length) {
$stop_boundary = $ii + 1;
break;
} else {
if ($word_boundary !== null) {
break;
}
}
}
}
if ($stop_boundary !== null) {
// We found a character like ".". Cut the string there, without appending
// the terminal.
$string_part = array_slice($string_v, 0, $stop_boundary);
return implode('', $string_part);
}
// If we didn't find any boundary characters or we found ONLY boundary
// characters, just break at the maximum character length.
if ($word_boundary === null || $word_boundary === 0) {
$word_boundary = $length - $terminal_len;
}
$string_part = array_slice($string_v, 0, $word_boundary);
$string_part = implode('', $string_part);
return $string_part . $terminal;
}
public function truncateString($string)
{
// First, check if the string has fewer bytes than the most restrictive
// limit. Codepoints and glyphs always take up at least one byte, so we can
// just return the string unmodified if we're under all of the limits.
$byte_len = strlen($string);
if ($byte_len <= $this->minimumLimit) {
return $string;
}
// If we need the vector of codepoints, build it.
$string_pv = null;
if ($this->maximumCodepoints) {
$string_pv = phutil_utf8v($string);
$point_len = count($string_pv);
}
// We always need the combined vector, even if we're only doing byte or
// codepoint truncation, because we don't want to truncate to half of a
// combining character.
$string_gv = phutil_utf8v_combined($string);
$glyph_len = count($string_gv);
// Now, check if we're still over the limits. For example, a string may
// be over the raw byte limit but under the glyph limit if it contains
// several multibyte characters.
$too_long = false;
if ($this->maximumBytes && $byte_len > $this->maximumBytes) {
$too_long = true;
}
if ($this->maximumCodepoints && $point_len > $this->maximumCodepoints) {
$too_long = true;
}
if ($this->maximumGlyphs && $glyph_len > $this->maximumGlyphs) {
$too_long = true;
}
if (!$too_long) {
return $string;
}
// This string is legitimately longer than at least one of the limits, so
// we need to truncate it. Find the minimum cutoff point: this is the last
// glyph we can possibly return while satisfying the limits and having space
// for the terminator.
$cutoff = $glyph_len;
if ($this->maximumBytes) {
if ($byte_len <= $this->maximumBytes) {
$cutoff = $glyph_len;
} else {
$bytes = $this->terminatorBytes;
for ($ii = 0; $ii < $glyph_len; $ii++) {
$bytes += strlen($string_gv[$ii]);
if ($bytes > $this->maximumBytes) {
$cutoff = $ii;
break;
}
}
}
}
if ($this->maximumCodepoints) {
if ($point_len <= $this->maximumCodepoints) {
$cutoff = min($cutoff, $glyph_len);
} else {
$points = 0;
for ($ii = 0; $ii < $glyph_len; $ii++) {
$glyph_bytes = strlen($string_gv[$ii]);
while ($points < $point_len) {
$glyph_bytes -= strlen($string_pv[$points]);
$points++;
if ($glyph_bytes <= 0) {
break;
}
}
$points_total = $points + $this->terminatorCodepoints;
if ($points_total > $this->maximumCodepoints) {
$cutoff = min($cutoff, $ii);
break;
}
}
}
}
if ($this->maximumGlyphs) {
if ($glyph_len <= $this->maximumGlyphs) {
$cutoff = min($cutoff, $glyph_len);
} else {
$cutoff = min($cutoff, $this->maximumGlyphs - $this->terminatorGlyphs);
}
}
// If we don't have enough characters for anything, just return the
// terminator.
if ($cutoff <= 0) {
return $this->terminator;
}
// Otherwise, we're going to try to cut the string off somewhere reasonable
// rather than somewhere arbitrary.
// NOTE: This is not complete, and there are many other word boundary
// characters and reasonable places to break words in the UTF-8 character
// space. For now, this gives us reasonable behavior for latin languages. We
// don't necessarily have access to PCRE+Unicode so there isn't a great way
// for us to look up character attributes.
// If we encounter these, prefer to break on them instead of cutting the
// string off in the middle of a word.
static $break_characters = array(' ' => true, "\n" => true, ';' => true, ':' => true, '[' => true, '(' => true, ',' => true, '-' => true);
// If we encounter these, shorten to this character exactly without
// appending the terminal.
static $stop_characters = array('.' => true, '!' => true, '?' => true);
// Search backward in the string, looking for reasonable places to break it.
$word_boundary = null;
$stop_boundary = null;
// If we do a word break with a terminal, we have to look beyond at least
// the number of characters in the terminal. If the terminal is longer than
// the required length, we'll skip this whole block and return it on its
// own.
// Only search backward for a while. At some point we don't get a better
// result by looking through the whole string, and if this is "MMM..." or
// a non-latin language without word break characters we're just wasting
// time.
$search = max(0, $cutoff - 256);
for ($ii = min($cutoff, $glyph_len - 1); $ii >= $search; $ii--) {
$c = $string_gv[$ii];
if (isset($break_characters[$c])) {
$word_boundary = $ii;
} else {
if (isset($stop_characters[$c])) {
$stop_boundary = $ii + 1;
break;
} else {
if ($word_boundary !== null) {
break;
}
}
}
}
if ($stop_boundary !== null) {
// We found a character like ".". Cut the string there, without appending
// the terminal.
$string_part = array_slice($string_gv, 0, $stop_boundary);
return implode('', $string_part);
}
// If we didn't find any boundary characters or we found ONLY boundary
// characters, just break at the maximum character length.
if ($word_boundary === null || $word_boundary === 0) {
$word_boundary = $cutoff;
}
$string_part = array_slice($string_gv, 0, $word_boundary);
$string_part = implode('', $string_part);
return $string_part . $this->terminator;
}
/**
* Hard-wrap a block of UTF-8 text with embedded HTML tags and entities.
*
* @param string An HTML string with tags and entities.
* @return list List of hard-wrapped lines.
* @group utf8
*/
function phutil_utf8_hard_wrap_html($string, $width)
{
$break_here = array();
// Convert the UTF-8 string into a list of UTF-8 characters.
$vector = phutil_utf8v($string);
$len = count($vector);
$char_pos = 0;
for ($ii = 0; $ii < $len; ++$ii) {
// An ampersand indicates an HTML entity; consume the whole thing (until
// ";") but treat it all as one character.
if ($vector[$ii] == '&') {
do {
++$ii;
} while ($vector[$ii] != ';');
++$char_pos;
// An "<" indicates an HTML tag, consume the whole thing but don't treat
// it as a character.
} else {
if ($vector[$ii] == '<') {
do {
++$ii;
} while ($vector[$ii] != '>');
} else {
++$char_pos;
}
}
// Keep track of where we need to break the string later.
if ($char_pos == $width) {
$break_here[$ii] = true;
$char_pos = 0;
}
}
$result = array();
$string = '';
foreach ($vector as $ii => $char) {
$string .= $char;
if (isset($break_here[$ii])) {
$result[] = $string;
$string = '';
}
}
if (strlen($string)) {
$result[] = $string;
}
return $result;
}
/**
* Soft wrap text for display on a console, respecting UTF8 character boundaries
* and ANSI color escape sequences.
*
* @param string Text to wrap.
* @param int Optional indent level.
* @return string Wrapped text.
*
* @group console
*/
function phutil_console_wrap($text, $indent = 0)
{
$lines = array();
$width = 78 - $indent;
$esc = chr(27);
$break_pos = null;
$len_after_break = 0;
$line_len = 0;
$line = array();
$lines = array();
$vector = phutil_utf8v($text);
$vector_len = count($vector);
for ($ii = 0; $ii < $vector_len; $ii++) {
$chr = $vector[$ii];
// If this is an ANSI escape sequence for a color code, just consume it
// without counting it toward the character limit. This prevents lines
// with bold/color on them from wrapping too early.
if ($chr == $esc) {
for ($ii; $ii < $vector_len; $ii++) {
$line[] = $vector[$ii];
if ($vector[$ii] == 'm') {
break;
}
}
continue;
}
$line[] = $chr;
++$line_len;
++$len_after_break;
if ($line_len > $width) {
if ($break_pos !== null) {
$slice = array_slice($line, 0, $break_pos);
while (count($slice) && end($slice) == ' ') {
array_pop($slice);
}
$slice[] = "\n";
$lines[] = $slice;
$line = array_slice($line, $break_pos);
$line_len = $len_after_break;
$len_after_break = 0;
$break_pos = null;
}
}
if ($chr == " ") {
$break_pos = count($line);
$len_after_break = 0;
}
if ($chr == "\n") {
$lines[] = $line;
$line = array();
$len_after_break = 0;
$line_len = 0;
$break_pos = null;
}
}
if ($line) {
if ($line) {
$lines[] = $line;
}
}
$pre = null;
if ($indent) {
$pre = str_repeat(' ', $indent);
}
foreach ($lines as $idx => $line) {
$lines[$idx] = $pre . implode('', $line);
}
return implode('', $lines);
}
