ruby--ruby/include/ruby/re.h

#ifndef RUBY_RE_H                                    /*-*-C++-*-vi:se ft=cpp:*/
#define RUBY_RE_H 1
/**
 * @file
 * @author     $Author$
 * @date       Thu Sep 30 14:18:32 JST 1993
 * @copyright  Copyright (C) 1993-2007 Yukihiro Matsumoto
 * @copyright  This  file  is   a  part  of  the   programming  language  Ruby.
 *             Permission  is hereby  granted,  to  either redistribute  and/or
 *             modify this file, provided that  the conditions mentioned in the
 *             file COPYING are met.  Consult the file for details.
 */
#include "ruby/internal/config.h"

#ifdef HAVE_SYS_TYPES_H
# include <sys/types.h>
#endif

#include <stdio.h>

#include "ruby/regex.h"
#include "ruby/internal/core/rmatch.h"
#include "ruby/internal/dllexport.h"

struct re_registers;            /* Defined in onigmo.h */

RBIMPL_SYMBOL_EXPORT_BEGIN()

/**
 * Creates a  new instance of  ::rb_cRegexp.  It can  be seen as  a specialised
 * version of rb_reg_new_str() where it does not take options.
 *
 * @param[in]  str  Source code in String.
 * @return     Allocated new instance of ::rb_cRegexp.
 */
VALUE rb_reg_regcomp(VALUE str);

/**
 * Runs  the  passed  regular  expression   over  the  passed  string.   Unlike
 * rb_reg_search()  this function  also  takes position  and  direction of  the
 * search, which make  it possible for this  function to run from  in middle of
 * the string.
 *
 * @param[in]  re               Regular expression to execute.
 * @param[in]  str              Target string to search.
 * @param[in]  pos              Offset in `str` to start searching, in bytes.
 * @param[in]  dir              `pos`' direction; 0  means left-to-right, 1 for
 *                              the opposite.
 * @exception  rb_eArgError     `re` is broken.
 * @exception  rb_eRegexpError  `re` is malformed.
 * @retval     -1               Match failed.
 * @retval     otherwise        Offset of first such byte where match happened.
 * @post       `Regexp.last_match` is updated.
 * @post       `$&`, `$~`, etc., are updated.
 *
 * @internal
 *
 * Distinction  between raising  ::rb_eArgError  and  ::rb_eRegexpError is  not
 * obvious, at least to @shyouhei.
 */
long rb_reg_search(VALUE re, VALUE str, long pos, int dir);

/**
 * Substitution.  This  is basically the implementation  of `String#sub`.  Also
 * `String#gsub` repeatedly calls this function.
 *
 * @param[in]  repl  Replacement string, e.g. `"\\1\\2"`
 * @param[in]  src   Source string, to be replaced.
 * @param[in]  regs  Matched data generated by applying `rexp` to `src`.
 * @param[in]  rexp  Regular expression.
 * @return     A substituted string.
 *
 * @internal
 *
 * This  function does  not  check for  encoding compatibility.   `String#sub!`
 * etc. employ their own checker.
 *
 * `regs` should  have been `const  struct re_registers  *` because it  is read
 * only.  Kept as-is for compatibility.
 */
VALUE rb_reg_regsub(VALUE repl, VALUE src, struct re_registers *regs, VALUE rexp);

/**
 * Tell us if this is a wrong idea,  but it seems this function has no usage at
 * all.  Just remains here for theoretical backwards compatibility.
 *
 * @param[in]  re               Regular expression to execute.
 * @param[in]  str              Target string to search.
 * @param[in]  pos              Offset in `str` to start searching, in bytes.
 * @param[in]  dir              `pos`' direction; 0  means left-to-right, 1 for
 *                              the opposite.
 * @return     Adjusted nearest  offset to  `pos` inside of  `str`, where  is a
 *             character boundary.
 *
 */
long rb_reg_adjust_startpos(VALUE re, VALUE str, long pos, int dir);

/**
 * Escapes  any  characters  that  would  have special  meaning  in  a  regular
 * expression.
 *
 * @param[in]  str  Target string to escape.
 * @return     A copy of `str` whose contents are escaped.
 */
VALUE rb_reg_quote(VALUE str);

/**
 * Exercises  various  checks  and  preprocesses  so  that  the  given  regular
 * expression can be applied to the given string.  The preprocess here includes
 * (but not limited to) for instance encoding conversion.
 *
 * @param[in]  re                  Target regular expression.
 * @param[in]  str                 What `re` is about to run on.
 * @exception  rb_eArgError        `re` does not fit for `str`.
 * @exception  rb_eEncCompatError  `re` and `str` are incompatible.
 * @exception  rb_eRegexpError     `re` is malformed.
 * @return     A preprocessesed pattern buffer ready to be applied to `str`.
 * @note       The return value is manages by our GC.  Don't free.
 *
 * @internal
 *
 * The  return  type,  `regex_t  *`, is  defined  in  `<ruby/onigmo.h>`,  _and_
 * _conflicts_ with POSIX's  `<regex.h>`.  We can no longer  save the situation
 * at this point.  Just don't mix the two.
 */
regex_t *rb_reg_prepare_re(VALUE re, VALUE str);

/**
 * Duplicates a match data.  This  is roughly the same as `onig_region_copy()`,
 * except it tries to GC when there is not enough memory.
 *
 * @param[out]  dst             Target registers to fill.
 * @param[in]   src             Source registers to duplicate.
 * @exception   rb_eNoMemError  Not enough memory.
 * @retval      0               Successful
 * @retval      ONIGERR_MEMORY  Not enough memory, even after GC (unlikely).
 * @post        `dst` has identical contents to `src`.
 *
 * @internal
 *
 * It seems this function is here for `ext/strscan` and nothing else.
 */
int rb_reg_region_copy(struct re_registers *dst, const struct re_registers *src);

RBIMPL_SYMBOL_EXPORT_END()

#endif /* RUBY_RE_H */