Regex

Added in version 2.14.

class Regex(**kwargs)

A GRegex is the “compiled” form of a regular expression pattern.

GRegex implements regular expression pattern matching using syntax and semantics similar to Perl regular expression. See the PCRE documentation) for the syntax definition.

Some functions accept a start_position argument, setting it differs from just passing over a shortened string and setting NOTBOL in the case of a pattern that begins with any kind of lookbehind assertion. For example, consider the pattern “BissB” which finds occurrences of “iss” in the middle of words. (”B” matches only if the current position in the subject is not a word boundary.) When applied to the string “Mississipi” from the fourth byte, namely “issipi”, it does not match, because “B” is always false at the start of the subject, which is deemed to be a word boundary. However, if the entire string is passed , but with start_position set to 4, it finds the second occurrence of “iss” because it is able to look behind the starting point to discover that it is preceded by a letter.

Note that, unless you set the RAW flag, all the strings passed to these functions must be encoded in UTF-8. The lengths and the positions inside the strings are in bytes and not in characters, so, for instance, “xc3xa0” (i.e. “à”) is two bytes long but it is treated as a single character. If you set RAW the strings can be non-valid UTF-8 strings and a byte is treated as a character, so “xc3xa0” is two bytes and two characters long.

When matching a pattern, “n” matches only against a “n” character in the string, and “r” matches only a “r” character. To match any newline sequence use “R”. This particular group matches either the two-character sequence CR + LF (”rn”), or one of the single characters LF (linefeed, U+000A, “n”), VT vertical tab, U+000B, “v”), FF (formfeed, U+000C, “f”), CR (carriage return, U+000D, “r”), NEL (next line, U+0085), LS (line separator, U+2028), or PS (paragraph separator, U+2029).

The behaviour of the dot, circumflex, and dollar metacharacters are affected by newline characters, the default is to recognize any newline character (the same characters recognized by “R”). This can be changed with G_REGEX_NEWLINE_CR, G_REGEX_NEWLINE_LF and G_REGEX_NEWLINE_CRLF compile options, and with G_REGEX_MATCH_NEWLINE_ANY, G_REGEX_MATCH_NEWLINE_CR, G_REGEX_MATCH_NEWLINE_LF and G_REGEX_MATCH_NEWLINE_CRLF match options. These settings are also relevant when compiling a pattern if G_REGEX_EXTENDED is set, and an unescaped “#” outside a character class is encountered. This indicates a comment that lasts until after the next newline.

Creating and manipulating the same GRegex structure from different threads is not a problem as GRegex does not modify its internal state between creation and destruction, on the other hand GMatchInfo is not threadsafe.

The regular expressions low-level functionalities are obtained through the excellent PCRE library written by Philip Hazel.

Constructors

class Regex

classmethod new(pattern: str, compile_options: RegexCompileFlags, match_options: RegexMatchFlags) → Regex | None

Compiles the regular expression to an internal form, and does the initial setup of the Regex structure.

Added in version 2.14.

Parameters:

pattern – the regular expression
compile_options – compile options for the regular expression, or 0
match_options – match options for the regular expression, or 0

Methods

class Regex

check_replacement(replacement: str) → tuple[bool, bool]

Checks whether replacement is a valid replacement string (see replace()), i.e. that all escape sequences in it are valid.

If has_references is not None then replacement is checked for pattern references. For instance, replacement text ‘foon’ does not contain references and may be evaluated without information about actual match, but ‘01’ (whole match followed by first subpattern) requires valid MatchInfo object.

Added in version 2.14.

Parameters:: replacement – the replacement string

error_quark() → int

escape_nul(string: str, length: int) → str

Escapes the nul characters in string to “x00”. It can be used to compile a regex with embedded nul characters.

For completeness, length can be -1 for a nul-terminated string. In this case the output string will be of course equal to string.

Added in version 2.30.

Parameters:

string – the string to escape
length – the length of string

escape_string(string: str, length: int) → str

Escapes the special characters used for regular expressions in string, for instance “a.b*c” becomes “a.b*c”. This function is useful to dynamically generate regular expressions.

string can contain nul characters that are replaced with “0”, in this case remember to specify the correct length of string in length.

Added in version 2.14.

Parameters:

string – the string to escape
length – the length of string, in bytes, or -1 if string is nul-terminated

get_capture_count() → int: Returns the number of capturing subpatterns in the pattern.

Added in version 2.14.

get_compile_flags() → RegexCompileFlags

Returns the compile options that regex was created with.

Depending on the version of PCRE that is used, this may or may not include flags set by option expressions such as (?i) found at the top-level within the compiled pattern.

Added in version 2.26.

get_has_cr_or_lf() → bool: Checks whether the pattern contains explicit CR or LF references.

Added in version 2.34.

get_match_flags() → RegexMatchFlags: Returns the match options that regex was created with.

Added in version 2.26.

get_max_backref() → int: Returns the number of the highest back reference in the pattern, or 0 if the pattern does not contain back references.

Added in version 2.14.

get_max_lookbehind() → int: Gets the number of characters in the longest lookbehind assertion in the pattern. This information is useful when doing multi-segment matching using the partial matching facilities.

Added in version 2.38.

get_pattern() → str: Gets the pattern string associated with regex, i.e. a copy of the string passed to new().

Added in version 2.14.

get_string_number(name: str) → int

Retrieves the number of the subexpression named name.

Added in version 2.14.

Parameters:: name – name of the subexpression

match(string: str, match_options: RegexMatchFlags) → tuple[bool, MatchInfo]

Scans for a match in string for the pattern in regex. The match_options are combined with the match options specified when the regex structure was created, letting you have more flexibility in reusing Regex structures.

Unless RAW is specified in the options, string must be valid UTF-8.

A MatchInfo structure, used to get information on the match, is stored in match_info if not None. Note that if match_info is not None then it is created even if the function returns False, i.e. you must free it regardless if regular expression actually matched.

To retrieve all the non-overlapping matches of the pattern in string you can use next().

static void
print_uppercase_words (const gchar *string)
{
  // Print all uppercase-only words.
  GRegex *regex;
  GMatchInfo *match_info;

  regex = g_regex_new ("[A-Z]+", G_REGEX_DEFAULT, G_REGEX_MATCH_DEFAULT, NULL);
  g_regex_match (regex, string, 0, &match_info);
  while (g_match_info_matches (match_info))
    {
      gchar *word = g_match_info_fetch (match_info, 0);
      g_print ("Found: ``%s``\n", word);
      g_free (word);
      g_match_info_next (match_info, NULL);
    }
  g_match_info_free (match_info);
  g_regex_unref (regex);
}

string is not copied and is used in MatchInfo internally. If you use any MatchInfo method (except free()) after freeing or modifying string then the behaviour is undefined.

Added in version 2.14.

Parameters:

string – the string to scan for matches
match_options – match options

match_all(string: str, match_options: RegexMatchFlags) → tuple[bool, MatchInfo]

Using the standard algorithm for regular expression matching only the longest match in the string is retrieved. This function uses a different algorithm so it can retrieve all the possible matches. For more documentation see match_all_full().

string is not copied and is used in MatchInfo internally. If you use any MatchInfo method (except free()) after freeing or modifying string then the behaviour is undefined.

Added in version 2.14.

Parameters:

string – the string to scan for matches
match_options – match options

match_all_full(string: Sequence[str], start_position: int, match_options: RegexMatchFlags) → tuple[bool, MatchInfo]

Using the standard algorithm for regular expression matching only the longest match in the string is retrieved, it is not possible to obtain all the available matches. For instance matching “<a> <c>” against the pattern “<.*>” you get “<a> <c>”.

This function uses a different algorithm (called DFA, i.e. deterministic finite automaton), so it can retrieve all the possible matches, all starting at the same point in the string. For instance matching “<a> <c>” against the pattern “<.*>;” you would obtain three matches: “<a> <c>”, “<a> ” and “<a>”.

The number of matched strings is retrieved using get_match_count(). To obtain the matched strings and their position you can use, respectively, fetch() and fetch_pos(). Note that the strings are returned in reverse order of length; that is, the longest matching string is given first.

Note that the DFA algorithm is slower than the standard one and it is not able to capture substrings, so backreferences do not work.

Setting start_position differs from just passing over a shortened string and setting NOTBOL in the case of a pattern that begins with any kind of lookbehind assertion, such as “b”.

Unless RAW is specified in the options, string must be valid UTF-8.

string is not copied and is used in MatchInfo internally. If you use any MatchInfo method (except free()) after freeing or modifying string then the behaviour is undefined.

Added in version 2.14.

Parameters:

string – the string to scan for matches
start_position – starting index of the string to match, in bytes
match_options – match options

match_full(string: Sequence[str], start_position: int, match_options: RegexMatchFlags) → tuple[bool, MatchInfo]

Setting start_position differs from just passing over a shortened string and setting NOTBOL in the case of a pattern that begins with any kind of lookbehind assertion, such as “b”.

Unless RAW is specified in the options, string must be valid UTF-8.

string is not copied and is used in MatchInfo internally. If you use any MatchInfo method (except free()) after freeing or modifying string then the behaviour is undefined.

To retrieve all the non-overlapping matches of the pattern in string you can use next().

static void
print_uppercase_words (const gchar *string)
{
  // Print all uppercase-only words.
  GRegex *regex;
  GMatchInfo *match_info;
  GError *error = NULL;

  regex = g_regex_new ("[A-Z]+", G_REGEX_DEFAULT, G_REGEX_MATCH_DEFAULT, NULL);
  g_regex_match_full (regex, string, -1, 0, 0, &match_info, &error);
  while (g_match_info_matches (match_info))
    {
      gchar *word = g_match_info_fetch (match_info, 0);
      g_print ("Found: ``%s``\n", word);
      g_free (word);
      g_match_info_next (match_info, &error);
    }
  g_match_info_free (match_info);
  g_regex_unref (regex);
  if (error != NULL)
    {
      g_printerr ("Error while matching: ``%s``\n", error->message);
      g_error_free (error);
    }
}

Added in version 2.14.

Parameters:

string – the string to scan for matches
start_position – starting index of the string to match, in bytes
match_options – match options

match_simple(pattern: str, string: str, compile_options: RegexCompileFlags, match_options: RegexMatchFlags) → bool

Scans for a match in string for pattern.

This function is equivalent to match() but it does not require to compile the pattern with new(), avoiding some lines of code when you need just to do a match without extracting substrings, capture counts, and so on.

If this function is to be called on the same pattern more than once, it’s more efficient to compile the pattern once with new() and then use match().

Added in version 2.14.

Parameters:

pattern – the regular expression
string – the string to scan for matches
compile_options – compile options for the regular expression, or 0
match_options – match options, or 0

replace(string: Sequence[str], start_position: int, replacement: str, match_options: RegexMatchFlags) → str

Replaces all occurrences of the pattern in regex with the replacement text. Backreferences of the form ‘number’ or ‘g<number>’ in the replacement text are interpolated by the number-th captured subexpression of the match, ‘g<name>’ refers to the captured subexpression with the given name. ‘0’ refers to the complete match, but ‘0’ followed by a number is the octal representation of a character. To include a literal ‘' in the replacement, write ‘\\’.

There are also escapes that changes the case of the following text:

l: Convert to lower case the next character
u: Convert to upper case the next character
L: Convert to lower case till E
U: Convert to upper case till E
E: End case modification

If you do not need to use backreferences use replace_literal().

The replacement string must be UTF-8 encoded even if RAW was passed to new(). If you want to use not UTF-8 encoded strings you can use replace_literal().

Setting start_position differs from just passing over a shortened string and setting NOTBOL in the case of a pattern that begins with any kind of lookbehind assertion, such as “b”.

Added in version 2.14.

Parameters:

string – the string to perform matches against
start_position – starting index of the string to match, in bytes
replacement – text to replace each match with
match_options – options for the match

replace_eval(string: Sequence[str], start_position: int, match_options: RegexMatchFlags, eval: Callable[[...], bool], *user_data: Any) → str

Replaces occurrences of the pattern in regex with the output of eval for that occurrence.

Setting start_position differs from just passing over a shortened string and setting NOTBOL in the case of a pattern that begins with any kind of lookbehind assertion, such as “b”.

The following example uses replace_eval() to replace multiple strings at once:

static gboolean
eval_cb (const GMatchInfo *info,
         GString          *res,
         gpointer          data)
{
  gchar *match;
  gchar *r;

   match = g_match_info_fetch (info, 0);
   r = g_hash_table_lookup ((GHashTable *)data, match);
   g_string_append (res, r);
   g_free (match);

   return FALSE;
}

...

GRegex *reg;
GHashTable *h;
gchar *res;

h = g_hash_table_new (g_str_hash, g_str_equal);

g_hash_table_insert (h, "1", "ONE");
g_hash_table_insert (h, "2", "TWO");
g_hash_table_insert (h, "3", "THREE");
g_hash_table_insert (h, "4", "FOUR");

reg = g_regex_new ("1|2|3|4", G_REGEX_DEFAULT, G_REGEX_MATCH_DEFAULT, NULL);
res = g_regex_replace_eval (reg, text, -1, 0, 0, eval_cb, h, NULL);
g_hash_table_destroy (h);

...

Added in version 2.14.

Parameters:

string – string to perform matches against
start_position – starting index of the string to match, in bytes
match_options – options for the match
eval – a function to call for each match
user_data – user data to pass to the function

replace_literal(string: Sequence[str], start_position: int, replacement: str, match_options: RegexMatchFlags) → str

Replaces all occurrences of the pattern in regex with the replacement text. replacement is replaced literally, to include backreferences use replace().

Setting start_position differs from just passing over a shortened string and setting NOTBOL in the case of a pattern that begins with any kind of lookbehind assertion, such as “b”.

Added in version 2.14.

Parameters:

string – the string to perform matches against
start_position – starting index of the string to match, in bytes
replacement – text to replace each match with
match_options – options for the match

split(string: str, match_options: RegexMatchFlags) → list[str]

Breaks the string on the pattern, and returns an array of the tokens. If the pattern contains capturing parentheses, then the text for each of the substrings will also be returned. If the pattern does not match anywhere in the string, then the whole string is returned as the first token.

As a special case, the result of splitting the empty string “” is an empty vector, not a vector containing a single string. The reason for this special case is that being able to represent an empty vector is typically more useful than consistent handling of empty elements. If you do need to represent empty elements, you’ll need to check for the empty string before calling this function.

A pattern that can match empty strings splits string into separate characters wherever it matches the empty string between characters. For example splitting “ab c” using as a separator “s*”, you will get “a”, “b” and “c”.

Added in version 2.14.

Parameters:

string – the string to split with the pattern
match_options – match time option flags

split_full(string: Sequence[str], start_position: int, match_options: RegexMatchFlags, max_tokens: int) → list[str]

Setting start_position differs from just passing over a shortened string and setting NOTBOL in the case of a pattern that begins with any kind of lookbehind assertion, such as “b”.

Added in version 2.14.

Parameters:

string – the string to split with the pattern
start_position – starting index of the string to match, in bytes
match_options – match time option flags
max_tokens – the maximum number of tokens to split string into. If this is less than 1, the string is split completely

split_simple(pattern: str, string: str, compile_options: RegexCompileFlags, match_options: RegexMatchFlags) → list[str]

This function is equivalent to split() but it does not require to compile the pattern with new(), avoiding some lines of code when you need just to do a split without extracting substrings, capture counts, and so on.

If this function is to be called on the same pattern more than once, it’s more efficient to compile the pattern once with new() and then use split().

Added in version 2.14.

Parameters:

pattern – the regular expression
string – the string to scan for matches
compile_options – compile options for the regular expression, or 0
match_options – match options, or 0