Handles common string operations in PHP applications:
- Filter a string (remove newlines, remove excess spaces, wrap long words, etc.)
- Test a string (whether a string is valid UTF8 or whether it is in a valid datetime format)
- Generate a random string with a set of characters
- Condense a number into a string, and convert/expand a string into a number
- Process an URL and modify it in a safe way (convert to relative URL, change parts of it, etc.)
- Regex wrapper to better handle type hints and errors with the most common regex functions
composer require squirrelphp/strings
Alternatively squirrelphp/strings-bundle can be installed for easy integration in Symfony projects, as it provides many servides for dependency injection by default and lets you easily register additional functionality based on this library.
Filters a string and returns back a string, using the Squirrel\Strings\StringFilterInterface interface:
public function filter(string $string): string;Each filter does exactly one thing (ideally) so they can be combined depending on how an input needs to be changed / processed.
Additional filters can easily be defined by implementing Squirrel\Strings\StringFilterInterface. Possible ideas for custom filters in applications:
- Process HTML tags, usually highly application dependent (which tags are allowed in which context)
- Streamline user input and combine multiple filters into one filter
- Convert HTML to Markdown, or the other way around
This library has some basic filters which cover a lot of ground and can be used individually or combined. Each filter class ends with a Filter suffix, which is not mentioned in the following list in order to keep the titles and texts more readable.
Normalizing newlines and removing excess spaces is helpful to not waste space before storing data and to always end up with consistent data, especially if you want to transform spaces and newlines to HTML or some other format later.
Replaces all types of newlines (in unicode there are currently 8 different newline characters) with unix newlines (\n) so you will not have a mixture of different newlines with unexpected results.
Replaces all unicode whitespace characters (currently 16 different ones) with a regular space if you do not care about the minute differences between the spaces (like width, or non-breaking, or mathematical). For most input this makes sense in order to be able to trim and limit unnecessary spaces.
Removes any unnecessary spaces, which are:
- Any spaces at the beginning or end of the string
- Any spaces around unix newlines
- Reduce consecutive spaces to just one space
Just operates on regular spaces (unicode 0020, decimal 32) and ignores other unicode whitespaces.
Limits the number of unix newlines which can appear right after each other. This only handles unix newlines and does not factor in spaces between newlines, so running the above three filters first makes sense.
The first argument of this filter is the number of consecutive newlines allowed, and it defaults to two, but can be set to any number above zero.
Zero width spaces are usually not something you want to save in a database, so this filter removes the three main zero width spaces defined in unicode.
If you do not need any newlines in a text or they are not allowed, this filter replaces any type of newline (in unicode there are currently 8 different newline characters) with a space (to avoid the risk of combining content which is only separated by a newline). Internally this uses the NormalizeNewlinesToUnixStyle filter first and then replaces the unix style newlines with spaces.
Trims characters from the beginning and end of the string, using the PHP trim function if the characters given to the constructor are only ASCII, or using regex if unicode characters are trimmed.
By default (if no constructor argument is used) the same characters are trimmed as the PHP trim function trims by default, meaning: " \t\n\r\0\x0B" (ordinary space, horizontal tab, new line, carriage return, NUL byte and vertical tab)
Replaces all horizontal tabs with spaces. This is the only filter that deals with horizontal tabs, as tabs might have a different/specific meaning compared to the other unicode spaces.
Long sequences of characters without a breaking character (like a space or newline) can break layouts and be difficult to display, and it can easily occur in user input or even by accident in regular content you write yourself.
This filter adds a zero-width space after a certain amount of characters (20 by default) in which no unix newlines or regular spaces occur. So if there is ample space even long words are not broken up, but if the space is tight the long word is split up into multiple lines.
This variant of the filter assumes no HTML is allowed, so it may break up any long sequence of characters.
Long sequences of characters without a breaking character (like a space or newline) can break layouts and be difficult to display, and it can easily occur in user input or even by accident in regular content you write yourself.
This filter adds a zero-width space after a certain amount of characters (20 by default) in which no unix newlines or regular spaces occur. So if there is ample space even long words are not broken up, but if the space is tight the long word is split up into multiple lines.
This variant of the filter looks for HTML tags in the string. If there are none, it behaves like WrapLongWordsNoHTML, if HTML tags do occur, each one is temporarily replaced by a substitute character and only counts as one character, and will not be broken up by the filter. So words might be split up "too early" when many HTML tags occur, as HTML tags count as one character for wrapping.
Converts all unicode characters to their lowercase equivalent.
Converts all unicode characters to their uppercase equivalent.
Convert the first character in the string to uppercase, correctly handles a unicode character as first character.
Convert the first character of every word in the string to uppercase, correctly handles unicode characters.
Convert from CamelCase to snake_case. Only supports alphanumeric characters (A-Z, a-z, 0-9), ignores all others!
Convert from snake_case to CamelCase. Only supports alphanumeric characters (A-Z, a-z, 0-9), ignores all others!
Removes all HTML tags. If HTML tags are malformed this might remove more than expected, as it does not try to validate the HTML, it just removes anything that looks like a HTML tag.
Removes the three main characters used in HTML tags: < , > and "
For HTML you often want to process newlines in a predictable way, this filter is one possibility:
- Convert double newlines
\n\nto</p><p> - Convert single newlines
\nto<br/> - Add
<p>to the beginning of the string and</p>to the end
For simple content without block level HTML tags this is often ideal to structure text and show it on a HTML page.
Encode &"'<> into their HTML entities (&, ", ', <, >), which is mainly helpful for correctly and securely displaying text in a HTML context.
Does the reverse of EncodeBasicHTMLEntities, sensible if you know input might contain HTML entities and you want to streamline the text and avoid something like &amp;.
This decodes all HTML entities according to the HTML5 standard (using html_entity_decode internally). This is usually not necessary but might make sense if you receive text and know it contains a lot of HTML entities and you do not know exactly which or how many.
Removes any characters which are not valid according to the UTF8 specification. This filter is recommended for anything coming from outside of your application (user input, web services, data import) so you can continue to operate on a valid UTF8 string afterwards. Most string functions or databases will otherwise reject a string if it contains invalid characters.
If invalid UTF8 characters must never appear in your application it might make sense to instead check the encoding in your application and throw an exception in this way:
// Checks if $string contains only valid UTF8 characters
if (!\mb_check_encoding($string, 'UTF-8')) {
// Invalid characters found, log this or throw an exception
}Yet this might be overkill for generic user input, where you just want to try to work things out even if the input is partly malformed (might be better than to fail completely).
Remove any characters which are not letters or numbers, so only A-Z, a-z and 0-9 are allowed. Can be handy for tokens, parts of an URL, an entered code, or other things where you know no other characters are allowed and you just want to ignore anything non-alphanumeric.
Remove any characters which are not letters, so only A-Z and a-z are allowed. Can be handy for tokens, country or language codes, or other things where you know no other characters are allowed and you just want to ignore them.
Remove any characters which are not numbers, so only 0-9 are allowed. Can be handy for tokens, parts of an URL, an entered code, or other things where you know no other characters are allowed and you just want to ignore them.
Remove any characters which are not letters, numbers and basic ASCII characters, so only A-Z, a-z, 0-9, space and !"#$'()*+,-./:;<=>?@[\]^_`{|}~ are allowed (no newlines, control characters, or unicode - all that is removed).
Remove anything that looks like an email, meaning any string part with non-space characters before and after an @ symbol, so this is quite "greedy".
Originally added to be able to analyze texts and detect the language, where email addresses would only confuse a language detection algorithm, so removing anything that looks like an email from a string should lead to "just text" or at least more analyzable text.
Remove anything that looks like an URL, meaning any string part that starts with a valid looking scheme, followed by "://", followed by zero or more non-space characters.
Originally added to be able to analyze texts and detect the language, where URLs would only confuse a language detection algorithm, so removing anything that looks like an URL from a string should lead to "just text" or at least more analyzable text.
Sometimes unicode with its plethora of characters can be a hindrance - for example in these cases:
- In a database of blocked customers you would want an entered first name like
émilto also matchemil, so a customer cannot slightly change his name to circumvent your security measures - When a user enters his address which you check against a database of known addresses you want it to be user-friendly, so if a user enters
Leon Breitling-StrasseorLeon Breitlingstrasseyou want both of these to matchLéon Breitling-Strasseeven though the characters do not match 1:1 - For an URL you want to map to ASCII letters as much as possible, so a blog post with a title like
L'école d'Humanitébecomesl-ecole-d-humanite(for an URL likehttps://my-blog.com/2019-07-12/l-ecole-d-humanite) which is both readable for users and search engines
Beware: This works well for countries and languages with latin characters (like most of Europe, North America, South America, most of Africa, Australia), yet not so well with other scripts, like Cyrillic, Arabic, Hanzi or Greek, to name just a few.
Reduces most letters to their base latin ASCII character (A-Z, a-z), if it is possible, so é becomes e, Â becomes A, etc. It is very thorough and uses both the Normalizer from the Intl extension and a long list of custom conversions. Some characters are converted to two ASCII characters (like Æ => AE, or ß to ss), so your string might get longer.
Runs NormalizeLettersToAscii from above and then removes any non-alphanumeric characters, so:
Léon Breitling-Strasse 13becomesLeonBreitlingStrasse13- 'Pré Raguel Strasse de l'école' becomes 'PreRaguelStrassedelecole'
Runs NormalizeToAlphanumeric from above and then converts all characters to lowercase, so:
Léon Breitling-Strasse 13becomesleonbreitlingstrasse13- 'Pré Raguel Strasse de l'école' becomes 'preraguelstrassedelecole'
If you process both the user inputs and the known values in your database in this manner, you can match them and get more matches/results, as spaces, dashes, diacritics etc. are not taken into account.
Sometimes you do not want to remove non-alphanumeric characters but instead replace them with a character, for example for URLs you want to convert L'école d'Humanité to l-ecole-d-humanite.
This is what ReplaceNonAlphanumeric does by default - replace all non-alphanumeric characters with a dash, and if multiple non-alphanumeric characters occur in sequence they are replaced by just one dash.
In the contructor of ReplaceNonAlphanumeric you can set another replacement character instead of a dash - for example a dot, or a slash, depending on your use case.
These filters combine other filters into a sensible package to run on user input, and is more of an example than something you might want to use directly in your application.
You can do your own combination of filters by using the Squirrel\Strings\StringFilterRunner class.
Runs the following filters:
- RemoveNonUTF8Characters
- ReplaceUnicodeWhitespaces
- ReplaceTabsWithSpaces
- NormalizeNewlinesToUnixStyle
- RemoveExcessSpaces
- LimitConsecutiveUnixNewlines
This makes sure the string is valid UTF8 and normalizes all whitespace characters and removes unnecessary whitespace characters, while leaving the content itself alone (works with or without HTML, does not convert HTML entities).
Runs the following filters:
- RemoveNonUTF8Characters
- ReplaceUnicodeWhitespaces
- ReplaceTabsWithSpaces
- ReplaceNewlinesWithSpaces
- RemoveExcessSpaces
Basically the same as StreamlineInputWithNewlines but newlines are converted to spaces. This is good for common user input like names, emails addresses and any other fields where newlines make no sense.
Tests a string and returns back true or false (whether the test was successful), using the Squirrel\Strings\StringTesterInterface interface:
public function test(string $string): bool;Additional testers can easily be defined by implementing Squirrel\Strings\StringTesterInterface. Possible ideas for custom testers in applications:
- Check the structure of external data (which can be highly application dependent)
- Check the string for allowed values or allowed characters
This library has two default testers. Each tester class ends with a Tester suffix, which is not mentioned in the following list in order to keep the titles and texts more readable.
Checks that only valid UTF8 characters are contained within a string. If your application wants to be strict about external data it can make sense to reject any data with non-UTF8 characters (a less strict way of dealing with non-UTF8 characters would be the RemoveNonUTF8Characters filter).
Checks the string according to a datetime format accepted by the date function given in the constructor of this class (default is Y-m-d for the ISO date format with dashes between year, month and day) and makes sure the given date exists (2021-02-29 would return false, yet 2020-02-29 would return true). When validating input this makes it easy to ensure a date is in the format you expect and can be used for further processing.
Generates random strings according to a list of possible characters allowed in the string.
With the two included classes (one with unicode support, one for ASCII-only) it is easy to define a random generator with your own set of characters which should be allowed to appear in a random string. These are sensible values:
ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789for 62 possible values per character, each can be A-Z, a-z or 0-9 and these values are very safe to use in applications (no special characters, only alphanumeric)abcdefghijklmnopqrstuvwxyz0123456789for 36 possible values per character, same as above except this is the case insensitive version, for when there should be no difference between "A" and "a" (for example)234579ACDEFGHKMNPQRSTUVWXYZor234579acdefghkmnpqrstuvwxyzfor 27 read-friendly uppercase or lowercase characters: if a person has to enter a code it is good to avoid characters which are very similar and easily confusable, like 0 (number zero) and O (letter), or 8 (number eight) and B (letter)
Defining your own range of possible characters is easy, and even unicode characters can be used.
Convert an integer to a string with a given "character set" - this way we can encode an integer to condense it (so an integer with 8 numbers is now only a 4-character-string) and later convert it back when needed.
The main use case are tokens in URLs, so less space is needed, as even large numbers become short strings if you use 36 or 62 values per character: with 62 possible characters (ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789) a string which is three characters long can cover numbers up to 238'328, with five characters you can cover numbers up to 916'132'832.
A side benefit of condensing is that it becomes less obvious an integer is used - tokens just look random and do not divulge their intent.
Defining your own range of possible characters is easy, and even unicode characters can be used.
The URL class accepts an URL in the constructor and then lets you get or change certain parts of the URL to do the following:
- Get scheme, host, path, query string and specific query string variables
- Change an absolute URL to a relative URL
- Change scheme, host, path and query string
- Replace query string variables, or add/remove them
This can be used to easily build or change your URLs, or to sanitize certain parts of a given URL, for example when redirecting: use the relative URL instead of the absolute URL to avoid malicious redirecting to somewhere outside of your control.
Using the built-in preg_match, preg_match_all, preg_replace and preg_replace_callback PHP functions often makes code less readable and harder to understand for static analyzers because of its uses of references ($matches) and the many possible return values. Squirrel\Strings\Regex wraps the basic functionality of these preg functions, creates easier to understand return values and throws a Squirrel\Strings\Exception\RegexException if anything goes wrong. These are the available static methods for the Regex class:
Wraps preg_match to check if $pattern exists in $subject.
Wraps preg_match_all to retrieve all occurences of $pattern in $subject with PREG_UNMATCHED_AS_NULL flag always set and the possibility to add additional flags. Returns null if no matches are found, otherwise the array of results as set by preg_match_all for $matches.
Regex::replace(string|array $pattern, string|array $replacement, string $subject, int $limit): string
Wraps preg_replace to replace occurences of $pattern with $replacement and only accepts a string as $subject.
Regex::replaceArray(string|array $pattern, string|array $replacement, array $subject, int $limit): array
Wraps preg_replace to replace occurences of $pattern with $replacement and only accepts an array as $subject.
Regex::replaceWithCallback(string|array $pattern, callback $callback, string $subject, int $limit, int $flags): string
Wraps preg_replace_callback to call a callback with the signature function(array $matches): string for each occurence of $pattern in $subject and only accepts a string as $subject.
Regex::replaceArrayWithCallback(string|array $pattern, callback $callback, array $subject, int $limit, int $flags): array
Wraps preg_replace_callback to call a callback with the signature function(array $matches): string for each occurence of $pattern in $subject and only accepts an array as $subject.