有没有可以清理内容的 PHP 类？

Question

我一直在尝试使用一系列正则表达式和 PHP 函数 preg_replace 在 PHP 中编写。

我的主要目标是整理内容，例如确保句子的开头是大写字母；逗号后面有一个空格；等等。

我试图实现的整理的一些例子：

// Remove any spaces around slashes
$content_replacements_from[] = "/\s*\/\s*/";
$content_replacements_to[] = "/";

// Remove any new lines or tabs
$content_replacements_from[] = "/[\r\n\t]/";
$content_replacements_to[] = " ";

// Remove any extra spaces
$content_replacements_from[] = "/\s{2,}/";
$content_replacements_to[] = " ";

// Tidy up joined full stops
$content_replacements_from[] = "/([a-zA-Z]{1})\s*[\.]{1}\s*([^(jpeg|jpg|png|pdf|gif|doc|xls|docx|xlsx|ppt|pptx|html|php|htm)]{1})/";
$content_replacements_to[] = "$1. $2";

// Tidy up joined commas
$content_replacements_from[] = "/([a-zA-Z0-9]{1})\s*[\,]{1}\s*([a-zA-Z0-9]{1})/";
$content_replacements_to[] = "$1, $2";

// Tidy up joined exclamation marks
$content_replacements_from[] = "/([a-zA-Z0-9]{1})\s*[\!]{1}\s*([a-zA-Z0-9]{1})/";
$content_replacements_to[] = "$1! $2";

// Tidy up joined question marks
$content_replacements_from[] = "/([a-zA-Z0-9]{1})\s*[\?]{1}\s*([a-zA-Z0-9]{1})/";
$content_replacements_to[] = "$1? $2";

// Tidy up joined semi colons
$content_replacements_from[] = "/([a-zA-Z0-9]{1})\s*[\;]{1}\s*([a-zA-Z0-9]{1})/";
$content_replacements_to[] = "$1; $2";

// Tidy up joined colons
$content_replacements_from[] = "/([a-zA-Z0-9]{1})\s*[\:]{1}\s*([a-zA-Z0-9]{1})/";
$content_replacements_to[] = "$1: $2";

// Tidy up fluid ounces
$content_replacements_from[] = "/[Ff]{1}[Ll]{1}.?\s?[Oo]{1}[Zz]{1}/";
$content_replacements_to[] = "fl oz";

// Tidy up rpm
$content_replacements_from[] = "/[Rr]{1}[Pp]{1}[Mm]{1}/";
$content_replacements_to[] = "rpm";

// Tidy up UK
$content_replacements_from[] = "/[Uu]{1}[Kk]{1}/";
$content_replacements_to[] = "UK";

// Tidy up Maxi-sense
$content_replacements_from[] = "/[Mm]{1}axi[\s\-]?[Ss]{1}ense/";
$content_replacements_to[] = "maxi-sense";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Mm]{1}axi[\s\-]?[Ss]{1}ense/";
$content_replacements_to[] = ". Maxi-sense";
$content_replacements_from[] = "/^[Mm]{1}axi[\s\-]?[Ss]{1}ense/";
$content_replacements_to[] = "Maxi-sense";

// Tidy up Side-by-side
$content_replacements_from[] = "/[Ss]{1}ide[\s\-]?[Bb]{1}y[\s\-]?[Ss]{1}ide/";
$content_replacements_to[] = "side-by-side";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Ss]{1}ide[\s\-]?[Bb]{1}y[\s\-]?[Ss]{1}ide/";
$content_replacements_to[] = ". Side-by-side";
$content_replacements_from[] = "/^[Ss]{1}ide[\s\-]?[Bb]{1}y[\s\-]?[Ss]{1}ide/";
$content_replacements_to[] = "Side-by-side";

// Tidy up extra large
$content_replacements_from[] = "/[Xx]{1}[Ll]{l}/";
$content_replacements_to[] = "extra large";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Xx]{1}[Ll]{l}/";
$content_replacements_to[] = "Extra large";
$content_replacements_from[] = "/^[Xx]{1}[Ll]{l}/";
$content_replacements_to[] = "Extra large";

// Tidy up D-radius
$content_replacements_from[] = "/[Dd]{1}[\s\-]?[Rr]{1}adius/";
$content_replacements_to[] = "D-radius";

// Tidy up A-rate
$content_replacements_from[] = "/[Aa]{1}[\s\-]?[Rr]{1}ate/";
$content_replacements_to[] = "A-rate";

// Tidy up In-column
$content_replacements_from[] = "/[Ii]{1}n[\s\-]?[Cc]{1}olum[n]?/";
$content_replacements_to[] = "in-column";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Ii]{1}n[\s\-]?[Cc]{1}olum[n]?/";
$content_replacements_to[] = "In-column";
$content_replacements_from[] = "/^[Ii]{1}n[\s\-]?[Cc]{1}olum[n]?/";
$content_replacements_to[] = "In-column";

// Tidy up kW
$content_replacements_from[] = "/[Kk]{1}[Ww]{1}/";
$content_replacements_to[] = "kW";

// Tidy up Built-in
$content_replacements_from[] = "/[Bb]{1}uilt[\s\-]?[Ii]{1}n/";
$content_replacements_to[] = "built-in";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Bb]{1}uilt[\s\-]?[Ii]{1}n/";
$content_replacements_to[] = "Built-in";
$content_replacements_from[] = "/^[Bb]{1}uilt[\s\-]?[Ii]{1}n/";
$content_replacements_to[] = "Built-in";

// Tidy up Built-under
$content_replacements_from[] = "/[Bb]{1}uilt[\s\-]?[Uu]{1}nder/";
$content_replacements_to[] = "built-under";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Bb]{1}uilt[\s\-]?[Uu]{1}nder/";
$content_replacements_to[] = "Built-under";
$content_replacements_from[] = "/^[Bb]{1}uilt[\s\-]?[Uu]{1}nder/";
$content_replacements_to[] = "Built-under";

// Tidy up Under-counter
$content_replacements_from[] = "/[Uu]{1}nder[\s\-]?[Cc]{1}ounter/";
$content_replacements_to[] = "under-counter";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Uu]{1}nder[\s\-]?[Cc]{1}ounter/";
$content_replacements_to[] = "Under-counter";
$content_replacements_from[] = "/^[Uu]{1}nder[\s\-]?[Cc]{1}ounter/";
$content_replacements_to[] = "Under-counter";

// Tidy up Under-cabinet
$content_replacements_from[] = "/[Uu]{1}nder[\s\-]?[Cc]{1}abinet/";
$content_replacements_to[] = "under-cabinet";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Uu]{1}nder[\s\-]?[Cc]{1}abinet/";
$content_replacements_to[] = "Under-cabinet";
$content_replacements_from[] = "/^[Uu]{1}nder[\s\-]?[Cc]{1}abinet/";
$content_replacements_to[] = "Under-cabinet";

// Tidy up integrated
$content_replacements_from[] = "/([a-zA-Z0-9]{1})[\s]{1}[\-]{1}[Ii]{1}ntegrated/";
$content_replacements_to[] = "$1-integrated";

// Tidy up Semi-integrated
$content_replacements_from[] = "/[Ss]{1}emi[\s\-]?[Ii]{1}ntegrated/";
$content_replacements_to[] = "semi-integrated";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Ss]{1}emi[\s\-]?[Ii]{1}ntegrated/";
$content_replacements_to[] = "Semi-integrated";
$content_replacements_from[] = "/^[Ss]{1}emi[\s\-]?[Ii]{1}ntegrated/";
$content_replacements_to[] = "Semi-integrated";

// Tidy up Fully-integrated
$content_replacements_from[] = "/[Ff]{1}ully[\s\-]?[Ii]{1}ntegrated/";
$content_replacements_to[] = "fully-integrated";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Ff]{1}ully[\s\-]?[Ii]{1}ntegrated/";
$content_replacements_to[] = "Fully-integrated";
$content_replacements_from[] = "/^[Ff]{1}ully[\s\-]?[Ii]{1}ntegrated/";
$content_replacements_to[] = "Fully-integrated";

// Tidy up Semi-automatic
$content_replacements_from[] = "/[Ss]{1}emi[\s\-]?[Aa]{1}utomatic/";
$content_replacements_to[] = "semi-automatic";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Ss]{1}emi[\s\-]?[Aa]{1}utomatic/";
$content_replacements_to[] = "Semi-automatic";
$content_replacements_from[] = "/^[Ss]{1}emi[\s\-]?[Aa]{1}utomatic/";
$content_replacements_to[] = "Semi-automatic";

// Tidy up Fully-automatic
$content_replacements_from[] = "/[Ff]{1}ully[\s\-]?[Aa]{1}utomatic/";
$content_replacements_to[] = "fully-automatic";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Ff]{1}ully[\s\-]?[Aa]{1}utomatic/";
$content_replacements_to[] = "Fully-automatic";
$content_replacements_from[] = "/^[Ff]{1}ully[\s\-]?[Aa]{1}utomatic/";
$content_replacements_to[] = "Fully-automatic";

// Tidy up Pull-out
$content_replacements_from[] = "/[Pp]{1}ull[\s\-]?[Oo]{1}ut/";
$content_replacements_to[] = "pull-out";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Pp]{1}ull[\s\-]?[Oo]{1}ut/";
$content_replacements_to[] = "Pull-out";
$content_replacements_from[] = "/^[Pp]{1}ull[\s\-]?[Oo]{1}ut/";
$content_replacements_to[] = "Pull-out";

// Tidy up including
$content_replacements_from[] = "/\s[Ii]{1}nc[l]?[\.]?\s/";
$content_replacements_to[] = " including ";

// Tidy up use
$content_replacements_from[] = "/\s[Uu]{1}se\s/";
$content_replacements_to[] = " use ";

// Tidy up ?-piece
$content_replacements_from[] = "/([2345TtYy]{1})[\s\-]?[Pp]{1}iece/";
$content_replacements_to[] = "$1-piece";

// Tidy up ?-spout
$content_replacements_from[] = "/([Cc]{1})[\s\-]?[Ss]{1}pout/";
$content_replacements_to[] = "$1-spout";

// Tidy up ?-end
$content_replacements_from[] = "/([Cc]{1})[\s\-]?[Ee]{1}nd/";
$content_replacements_to[] = "$1-end";

// Tidy up Brushed Steel
$content_replacements_from[] = "/[Bb]{1}[\-\/]{1}[Ss]{1}teel/";
$content_replacements_to[] = "brushed steel";

// Tidy up Stainless Steel
$content_replacements_from[] = "/[Ss]{1}[\-\/]{1}[Ss]{1}teel/";
$content_replacements_to[] = "stainless steel";

// Tidy up Silk Steel
$content_replacements_from[] = "/[Ss]{1}ilk[\s]?[Ss]{1}teel/";
$content_replacements_to[] = "silk steel";

// Remove trade marks
$content_replacements_from[] = "/™/";
$content_replacements_to[] = "";

// Replace long dashes
$content_replacements_from[] = "/–/";
$content_replacements_to[] = "-";

// Replace single quotes
$content_replacements_from[] = "/’/";
$content_replacements_to[] = "'";
$content_replacements_from[] = "/`/";
$content_replacements_to[] = "'";

// Tidy up m
$content_replacements_from[] = "/[\s]?[Mm]{1}etre/";
$content_replacements_to[] = "m";

// Tidy up m3
$content_replacements_from[] = "/([0-9]{1})[\s]?[Mm]{1}3/";
$content_replacements_to[] = "$1m³";
$content_replacements_from[] = "/\&sup3\;/";
$content_replacements_to[] = html_entity_decode("³");

// Tidy up to in between numbers
$content_replacements_from[] = "/([0-9]{1})[\s]?to[\s]?([0-9]{1})/";
$content_replacements_to[] = "$1 - $2";

// Tidy up per hour
$content_replacements_from[] = "/\s[Aa]{1}nd\s[Hh]{1}[Rr]?$/";
$content_replacements_to[] = "ph";

// Tidy up l
$content_replacements_from[] = "/[\s]?[Ll]{1}itre/";
$content_replacements_to[] = "l";

// Tidy up -in
$content_replacements_from[] = "/\-[Ii]{1}n/";
$content_replacements_to[] = "-in";

// Tidy up plus
$content_replacements_from[] = "/\s[Pp]{1}lus\s/";
$content_replacements_to[] = " plus ";

// Tidy up including
$content_replacements_from[] = "/\s[Ii]{1}ncluding\s/";
$content_replacements_to[] = " including ";

// Tidy up including
$content_replacements_from[] = "/[Ii]{1}nc\s/";
$content_replacements_to[] = "Including "; 

// Tidy up Push/pull
$content_replacements_from[] = "/[Pp]{1}ush\/[Pp]{1}ull/";
$content_replacements_to[] = "push/pull";
$content_replacements_from[] = "/[\.|\!|\?]{1}\s{1}[Pp]{1}ush\/[Pp]{1}ull/";
$content_replacements_to[] = "Push/pull";
$content_replacements_from[] = "/^[Pp]{1}ush\/[Pp]{1}ull/";
$content_replacements_to[] = "Push/pull";

// Tidy up +
$content_replacements_from[] = "/\s\+\s/";
$content_replacements_to[] = " and ";

// Tidy up *
$content_replacements_from[] = "/\*/";
$content_replacements_to[] = "";

// Tidy up with
$content_replacements_from[] = "/\s[Ww]{1}ith\s/";
$content_replacements_to[] = " with ";

// Tidy up without
$content_replacements_from[] = "/\s[Ww]{1}ithout\s/";
$content_replacements_to[] = " without ";

// Tidy up in
$content_replacements_from[] = "/\s[Ii]{1}n\s/";
$content_replacements_to[] = " in ";

// Tidy up of
$content_replacements_from[] = "/\s[Oo]{1}f\s/";
$content_replacements_to[] = " of ";

// Tidy up for
$content_replacements_from[] = "/\s[Ff]{1}or\s/";
$content_replacements_to[] = " for ";

// Tidy up or
$content_replacements_from[] = "/\s[Oo]{1}r\s/";
$content_replacements_to[] = " or ";

// Tidy up and
$content_replacements_from[] = "/\s[Aa]{1}nd\s/";
$content_replacements_to[] = " and ";

// Tidy up to
$content_replacements_from[] = "/\s[Tt]{1}o\s/";
$content_replacements_to[] = " to ";

// Tidy up too
$content_replacements_from[] = "/\s[Tt]{1}oo\s/";
$content_replacements_to[] = " too ";

// Tidy up &
$content_replacements_from[] = "/\s&\s/";
$content_replacements_to[] = " and ";

// Tidy up &
$content_replacements_from[] = "/\s&\s/";
$content_replacements_to[] = " and ";

// Tidy up mm
$content_replacements_from[] = "/M[Mm]{1}/";
$content_replacements_to[] = "mm";

// Tidy up ize to ise
$content_replacements_from[] = "/([a-zA-Z]{2})ize{1}/";
$content_replacements_to[] = "$1ise";

// Tidy up izer to iser
$content_replacements_from[] = "/([a-zA-Z]{2})izer{1}/";
$content_replacements_to[] = "$1iser";

// Tidy up yze to yse
$content_replacements_from[] = "/([a-zA-Z]{2})yze{1}/";
$content_replacements_to[] = "$1yse";

// Tidy up ization to isation
$content_replacements_from[] = "/([a-zA-Z]{2})ization{1}/";
$content_replacements_to[] = "$1isation";

// Tidy up times symbol
$content_replacements_from[] = "/([0-9]{1})\s*[Xx]\s*([0-9A-Za-z]{1})/";
$content_replacements_to[] = "$1 × $2";

// Tidy up times symbol
$content_replacements_from[] = "/\&times\;/";
$content_replacements_to[] = html_entity_decode("×");

// Tidy up inches
$content_replacements_from[] = "/([0-9]{1})\s*[Ii]{1}nches/";
$content_replacements_to[] = "$1\"";

// Tidy up inch
$content_replacements_from[] = "/([0-9]{1})\s*[Ii]{1}nch/";
$content_replacements_to[] = "$1\"";

// Make the replacements
$content = preg_replace($content_replacements_from, $content_replacements_to, $content);

这显然是复杂和冗长的。

有谁知道更好的方法或知道可以做到这一点的课程吗？

如果可能的话，我也想将其应用于 HTML 中的内容。

Answer 1

正则表达式非常适合文本搜索和替换。您收到的报告表明还有改进的空间。但我的答案不是要优化这些，而是​​建议开始构建您自己的 StringCleaner 集，它可以执行不同的操作，但都具有相同的界面：

interface StringCleaner
{
    public function clean($string);
}

接下来，对于 HTML，一个想法我要做的就是创建一个 FilterIterator 来提供对所有文本节点的访问，这样它们就可以更轻松地使用任何标准清理器进行更改。

为了一次应用多个StringCleaner（并创建这些集合），我使用了复合模式 （通过从 SplObjectStore 扩展）它本身也是一个 StringCleaner。

没有类定义的示例：

$cleanerTrim = new TrimCleaner();

$cleanerBasics = new RegexCleaner();

// Remove any spaces around slashes
$cleanerBasics->addRule('\s*\/\s*', '/');

// Remove any new lines or tabs
$cleanerBasics->addRule('[\r\n\t]', ' ');

// Tidy up joined full stops
$cleanerBasics->addRule('(\w+)\.(?!jpeg|jpg|png|pdf|gif|doc|xls|docx|xlsx|ppt|pptx|html|php|htm)(\w+)', '$1. $2');

// Remove any extra spaces
$cleanerBasics->addRule('\s{2,}', ' ');

// Remove single spaces
$cleanerBasics->addRule('^\s
如示例所示，当您将复杂性隐藏到具体的 StringCleaner 对象中时，您可以开始创建更多动态规则。这可以通过添加更多对不同于正则表达式的操作的 StringCleaner 类型进行扩展，TrimCleaner 中给出了一个非常简单的 trim 示例。
但可以肯定的是，正则表达式也非常强大。正如您在 RegexCleaner 中看到的，我已将每个正则表达式分隔符移至类本身中，因此当您定义规则时，无需一遍又一遍地键入它们。这只是另一个简单的示例，当您使用已定义的操作接口将替换封装到它自己的类中时，您可以简化事情。
 完整示例。
, '');

$cleanerInches = new RegexCleaner();

// Tidy up inches
$cleanerInches->addRule('([0-9])\s*[Ii]nches', '$1"');


$cleaner = new CleanerComposite();
$cleaner->attach($cleanerBasics);
$cleaner->attach($cleanerInches);
$cleaner->attach($cleanerTrim);


$htmlString = <<<HTML
<html>
  <head>
    <title>
        hello world.hello earth.
    </title>
  </head>
  <body>
<table><tr><td>test. 
</td></tr></table>
     <h1>Get it 1 more time.</h1>
     <p>When 12 inches were not enough;      hickup.</p>

  </body>
</html>
HTML;


// load HTML
$dom = new DOMDocument();
$dom->preserveWhiteSpace = FALSE;
$dom->loadHTML($htmlString);

// create XPath
$xpath = new DomXPath($dom);

$it = new DOMTextWhiteSpaceFilter($xpath->query('//text()'));
foreach($it as $node)
{
    $node->data = $cleaner->clean($node->data);
}

// remove whitespace only nodes
$it = new DOMTextWhiteSpaceFilter($xpath->query('//text()'), DOMTextWhiteSpaceFilter::WHITESPACE);
foreach($it as $node)
{
    $node->parentNode->removeChild($node);
}

$dom->formatOutput = true;
echo $dom->saveHTML();

如示例所示，当您将复杂性隐藏到具体的 StringCleaner 对象中时，您可以开始创建更多动态规则。这可以通过添加更多对不同于正则表达式的操作的 StringCleaner 类型进行扩展，TrimCleaner 中给出了一个非常简单的 trim 示例。

但可以肯定的是，正则表达式也非常强大。正如您在 RegexCleaner 中看到的，我已将每个正则表达式分隔符移至类本身中，因此当您定义规则时，无需一遍又一遍地键入它们。这只是另一个简单的示例，当您使用已定义的操作接口将替换封装到它自己的类中时，您可以简化事情。

完整示例。

Answer 2

可能有一种比许多正则表达式更好的方法来做到这一点，但如果没有其他人能想出更好的工具，那么我将如何使用 PHP 正则表达式来做到这一点。

可读性和易于维护几乎总是比速度更重要。 preg_replace 确实需要两个单独的字符串或数组来匹配和替换，但我们可以通过在使用时重新排列数据来解决这个问题。因此，我建议使用以下更具可读性的格式：

$content_replacements = array(array('From' => "/pattern 1/", 'To' => "$1 $2"),
                              array('From' => "/pattern 2/", 'To' => "$1,$2."));

它有一个很大的优点，如果您忘记“从”或“到”，您的模式和替换不会失去同步。

然后要运行所有这些，您可以使用循环：

   foreach ($content_replacements as $replacement)
   {
      $content = preg_replace($replacement['From'], $replacement['To'], $content);
   }

Answer 3

我认为您想要的第一件事是找到关于正则表达式的良好参考，并温习诸如哪些字符需要在字符类内部和外部转义（方括号：[]）之类的事情。然后通过删除 {1} 并在有意义时使用不区分大小写的匹配（'//i'）来清理您的正则表达式。然后我提供了一些代码，您可以使用它们来合并一些大小写规则，希望对您有所帮助。

// capitalize first word of each sentence
function tidy_sentences($str){
    $tokens = array();
    foreach(preg_split('/([.?!]\s*)/', $str, 0, PREG_SPLIT_DELIM_CAPTURE) as $token){
        if(!preg_match('/([.?!]\s*)/', $token, $m)) $token[0] = strtoupper($token[0]);
        $tokens[] = $token;
    }
    return implode($tokens);
}

// apply capitalization rules to individual words
function tidy_words($str){
    $tokens = array();
    foreach(preg_split('/([^\w-])/', $str, 0, PREG_SPLIT_DELIM_CAPTURE) as $token){
        switch(true){
            // tokens you want to uppercase go here
            case preg_match('/^(uk|kw)$/i', $token, $m): $tokens[] = strtoupper($m[0]); break;
            // tokens you want to lowercase go here
            case preg_match('/^(rpm|fl|oz)$/i', $token, $m): $tokens[] = strtolower($m[0]); break;
            // tokens you want to capitalize first letter go here
            case preg_match('/^(maxi-sense|side-by-side|d-radius)$/i', $token, $m): $tokens[] = ucfirst($m[0]); break;
            default: $tokens[] = $token;
        }
    }
    return implode($tokens);
}

function tidy($str){
    return tidy_sentences(tidy_words($str));
}

echo tidy('foo bar rpm maxi-sense uk Fl OZ baz! pull out');
// => Foo bar rpm Maxi-sense UK fl oz baz! Pull out

Answer 4

我认为最好的方法不一定是使用纯正则表达式。正则表达式旨在用作适合遵循特定模式的字符串的工具。您这里的用例似乎并不那么具体，所以我认为是时候探索不同的途径了。

我不确定您要清理的字符串有多复杂，或者内容通常是什么（也许它通常少于 100 个单词并且属于某个特定主题，因此常见词汇量很小）。但是，我认为更灵活（从长远来看可能更容易）的解决方案将涉及更简单的正则表达式（或正则表达式集）来识别字符串中的标记。识别出特定标记后，您可以采取非常具体的操作，并根据识别标记对字符串执行更高级的算法。

我相信 hakre 的 StringCleaner 类走在正确的道路上，其想法是从核心逻辑中抽象出不同类型的字符串，以帮助清理代码并使其更易于管理和灵活。

我知道这是非常通用和理论性的，但是你正在谈论的库的大小会很快膨胀 - 因为它涉及（基本上）自然语言处理，这是人工智能的一种形式。