foreach ($out as $idx => $str) {
$str = trim($str);
if (strlen($str) == 0) {
continue;
}
$this->ChangedVerbs[$this->get_ed($str)] = $str;
$this->ChangedVerbs[$this->get_ing($str)] = $str;
}
return $this;
}
public function show()
{
foreach ($this->ChangedVerbs as $key => $past) {
printf("%s => %s <br>", $key, $past);
}
}
public function getBaseForm($word, &$bChanged)
{
if (array_key_exists($word, $this->ChangedVerbs)) {
$bChanged = true;
return $this->ChangedVerbs[$word];
} else {
$bChanged = false;
return $word;
}
}
}
if (0) {
$rv = new EnglishRegularVerbs();
$rv->show();
}
public function load2freq($filename)
{
$this->init2();
$IrregVerb = new EnglishIrregularVerbs();
$RegVerb = new EnglishRegularVerbs();
$data = file_get_contents($filename);
if (false == $data) {
print "failed6 to read file:" . $filename;
die;
}
//$data = str_replace("'s ", " ", $data);
$data = strtolower($data);
$data = preg_replace("/([a-z]+[0-9]+[:][0-9]+[\\s])/i", " ", $data);
//del verse name.
$data = preg_replace("/[a-z]['][s][\\s]/i", " ", $data);
//del ---'s .
//$data = preg_replace("/[:'\"\?\.,=+()_[]{}|\\\/~`!@#\$%^&\*<>]/i", " ", $data); //del ---'s .
$data = preg_replace("/[^a-z]/i", " ", $data);
//del symbols . ^:not letters.
//$pattern = "/src=[\"']?([^\"']?.*(png|jpTotalWordsg|gif))[\"']?/i";
//$pattern = "/[0-9]+[\s]+[A-Za-z']+[\s]*/i"; //^:start of line; \s:space; $:end line; +:>=1;
//$tot = preg_match_all("/[\s]+/", $data, $out);
//print_r($out[0]);
$out = preg_split("/[\\s]+/", $data);
//space
foreach ($out as $idx => $str) {
$str = trim($str);
if (strlen($str) == 0) {
continue;
}
//
$word = $str;
$bChanged = false;
$s = substr($str, -1);
//get last char;
if ("s" == $s) {
$this->wordends($str);
$word = $this->del_s($str, $bChanged);
}
if (!$bChanged) {
$word = $IrregVerb->getBaseForm($str, $bChanged);
}
if (!$bChanged) {
$word = $RegVerb->getBaseForm($str, $bChanged);
}
if (!$bChanged) {
//iiregular noun plural, verb tense.
if (array_key_exists($word, $this->Ignores)) {
$word = $this->Ignores[$word];
}
}
//ignored words
if (strlen($word) == 0) {
echo count($out) . "," . $idx . $str . " == bad word process<br>";
continue;
}
if (!isset($this->WordFreqArr[$word])) {
$this->WordFreqArr[$word] = array("sort" => 0, "afreq" => 0, "arank" => 0, "rfreq" => 0, "rrank" => 0);
}
$arr = $this->WordFreqArr[$word];
$arr["afreq"] += 1;
$arr["arank"] = 0;
$arr["sort"] = $arr["afreq"];
$this->WordFreqArr[$word] = $arr;
$this->TotalWords += 1;
}
//$IrregVerb->show_FoundUsed();
$this->Ignores = array_merge($this->Ignores, $IrregVerb->ChangedVerbs);
$this->Ignores = array_merge($this->Ignores, $RegVerb->ChangedVerbs);
}
