source: libcfa/src/collections/string.hfa@ d8a2f98

//
// Cforall Version 1.0.0 Copyright (C) 2016 University of Waterloo
//
// The contents of this file are covered under the licence agreement in the
// file "LICENCE" distributed with Cforall.
//
// string -- variable-length, mutable run of text, with value semantics
//
// Author           : Michael L. Brooks
// Created On       : Fri Sep 03 11:00:00 2021
// Last Modified By : Peter A. Buhr
// Last Modified On : Mon May  4 21:54:23 2026
// Update Count     : 332
//

#pragma once

#include <iostream.hfa>
#include <string_res.hfa>

static struct __cfa_string_preference_boost_t {} __cfa_string_preference_boost;
#define PBOOST forall ( | { __cfa_string_preference_boost_t __cfa_string_preference_boost; } )

struct string {
        string_res * inner;
};

// RAII, assignment
void ^?{}( string & s );

void ?{}( string & s );                                                                 // empty string
void ?{}( string & s, string s2, size_t maxlen );
PBOOST void ?{}( string & s, string s2 );
void ?{}( string & s, char );
void ?{}( string & s, const char * cs );                                // copy from string literal (NULL-terminated)
void ?{}( string & s, const char * cs, size_t size );   // copy specific length from buffer

void ?{}( string & s, signed long int rhs );
void ?{}( string & s, size_t rhs );
void ?{}( string & s, double rhs );
void ?{}( string & s, long double rhs );
void ?{}( string & s, double _Complex rhs );
void ?{}( string & s, long double _Complex rhs );
static inline void ?{}( string & s, int rhs ) { (s){(signed long int) rhs}; }

PBOOST string & ?=?( string & s, string c );
string & ?=?( string & s, const char * cs );                    // copy from "literal"
string & ?=?( string & s, char c );                                             // copy from 'l'
string & assign( string & s, const string & c, size_t n );
string & assign( string & s, const char * cs, size_t n );
string & ?=?( string & s, signed long int rhs );
string & ?=?( string & s, size_t rhs );
string & ?=?( string & s, double rhs );
string & ?=?( string & s, long double rhs );
string & ?=?( string & s, double _Complex rhs );
string & ?=?( string & s, long double _Complex rhs );
static inline string & ?=?( string & s, int rhs ) { return s = ((signed long int) rhs); } // to match cost of (char * int): int

static inline string tostr( char c ) { string s = c; return s; }
static inline string tostr( const char * cs ) { string s = cs; return s; }
static inline string tostr( signed long int v ) { string s = v; return s; }
static inline string tostr( size_t v ) { string s = v; return s; }
static inline string tostr( double v ) { string s = v; return s; }
static inline string tostr( long double v ) { string s = v; return s; }
static inline string tostr( long double _Complex v ) { string s = v; return s; }

static inline string & strcpy( string & s, const char * cs ) { s = cs; return s; }
static inline string & strncpy( string & s, const char * cs, size_t n ) { assign( s, cs, n ); return s; }
static inline string & strcpy( string & s1, const string & s2 ) { s1 = s2; return s1; }
static inline string & strncpy( string & s, const string & cs, size_t n ) { assign( s, cs, n ); return s; }
char * strncpy( char * dst, string & src, size_t n );
char * ?=?( char *& dst, string & src );
void ?{}( char *& dst, string & src );

// Alternate construction: request shared edits
struct string_Share {
        string * s;
};
string_Share ?`share( string & s );
void ?{}( string & s, string_Share src );

// Getters
static inline size_t len( const string & s ) { return len( *s.inner ); }
static inline size_t len( const char * cs ) { return strlen( cs ); };
static inline size_t strlen( const string & s ) { return len( s ); }
size_t strnlen( const string & s, size_t maxlen );

// IO Operator
forall( ostype & | basic_ostream( ostype ) ) {
        ostype & ?|?( ostype & out, string s );
        void ?|?( ostype & out, string s );
}
forall( istype & | basic_istream( istype ) )
istype & ?|?( istype & in, string & s );

static inline {
        _Ostream_Manip(string) bin( string s ) { return (_Ostream_Manip(string))@{ .val = s, .wd = 1, .pc = 0, .base = 'b', { .all = 0 } }; }
        _Ostream_Manip(string) oct( string s ) { return (_Ostream_Manip(string))@{ .val = s, .wd = 1, .pc = 0, .base = 'o', { .all = 0 } }; }
        _Ostream_Manip(string) hex( string s ) { return (_Ostream_Manip(string))@{ .val = s, .wd = 1, .pc = 0, .base = 'x', { .all = 0 } }; }
        _Ostream_Manip(string) quote( string s, const char qleft = '"', const char qright = '\0' ) {
                return (_Ostream_Manip(string))@{ .val = s, .wd = 1, .pc = 0, .base = 's', { .flags.quote = true }, .qleft = qleft, .qright = qright }; }
        _Ostream_Manip(string) wd( unsigned int wd, string s ) { return (_Ostream_Manip(string))@{ .val = s, .wd = wd, .pc = 0, .base = 's', { .all = 0 } }; }
        _Ostream_Manip(string) wd( unsigned int wd, unsigned int pc, string s ) { return (_Ostream_Manip(string))@{ .val = s, .wd = wd, .pc = pc, .base = 's', { .flags.pc = true } }; }
        _Ostream_Manip(string) & wd( unsigned int wd, _Ostream_Manip(string) & fmt ) { fmt.wd = wd; return fmt; }
        _Ostream_Manip(string) & wd( unsigned int wd, unsigned int pc, _Ostream_Manip(string) & fmt ) { fmt.wd = wd; fmt.pc = pc; fmt.flags.pc = true; return fmt; }
        _Ostream_Manip(string) & left( _Ostream_Manip(string) & fmt ) { fmt.flags.left = true; return fmt; }
        _Ostream_Manip(string) & nobase( _Ostream_Manip(string) & fmt ) { fmt.flags.nobsdp = true; return fmt; }
        _Ostream_Manip(string) & upcase( _Ostream_Manip(string) & fmt ) { if ( fmt.base == 'x' || fmt.base == 'b' ) fmt.base -= 32; /* upper case */ return fmt; }
        _Ostream_Manip(string) & quote( _Ostream_Manip(string) & fmt, const char qleft = '"', const char qright = '\0' ) {
                fmt.flags.quote = true; fmt.qleft = qleft, fmt.qright = qright; return fmt; }
} // distribution

forall( ostype & | basic_ostream( ostype ) ) {
        ostype & ?|?( ostype & os, _Ostream_Manip(string) f );
        void ?|?( ostype & os, _Ostream_Manip(string) );
}

struct _Istream_Swidth {
        string & s;
        inline _Istream_str_base;
}; // _Istream_Swidth

struct _Istream_Squote {
        _Istream_Swidth sstr;
}; // _Istream_Squote

struct _Istream_Sstr {
        string & s;
        inline _Istream_str_base;
//      _Istream_Swidth sstr;
}; // _Istream_Sstr

static inline {
        // read width does not include null terminator
        _Istream_Swidth wdi( unsigned int rwd, string & s ) { return (_Istream_Swidth)@{ .s = s, { {.scanset = 0p}, .wd = rwd, {.flags.rwd = true} } }; }
        _Istream_Sstr getline( string & s, const char delimiter = '\n' ) {
                return (_Istream_Sstr)@{ .s = s, { {.delimiters = { delimiter, '\0' } }, .wd = -1, {.flags.delimiter = true} } };
        }
        _Istream_Sstr & getline( _Istream_Swidth & f, const char delimiter = '\n' ) {
                f.delimiters[0] = delimiter; f.delimiters[1] = '\0'; f.flags.delimiter = true; return (_Istream_Sstr &)f;
        }
        _Istream_Squote quote( string & s, const char Ldelimiter = '\"', const char Rdelimiter = '\0' ) {
                return (_Istream_Squote)@{ { .s = s, { {.delimiters = { Ldelimiter, Rdelimiter, '\0' }}, .wd = -1, {.flags.rwd = true} } } };
        }
        _Istream_Squote & quote( _Istream_Swidth & f, const char Ldelimiter = '"', const char Rdelimiter = '\0' ) {
                f.delimiters[0] = Ldelimiter;  f.delimiters[1] = Rdelimiter;  f.delimiters[2] = '\0';
                return (_Istream_Squote &)f;
        }
        _Istream_Sstr incl( const char scanset[], string & s ) { return (_Istream_Sstr)@{ .s = s, { {.scanset = scanset}, .wd = -1, {.flags.inex = false} } }; }
        _Istream_Sstr & incl( const char scanset[], _Istream_Swidth & f ) { f.scanset = scanset; f.flags.inex = false; return (_Istream_Sstr &)f; }
        _Istream_Sstr excl( const char scanset[], string & s ) { return (_Istream_Sstr)@{ .s = s, { {.scanset = scanset}, .wd = -1, {.flags.inex = true} } }; }
        _Istream_Sstr & excl( const char scanset[], _Istream_Swidth & f ) { f.scanset = scanset; f.flags.inex = true; return (_Istream_Sstr &)f; }
        _Istream_Sstr ignore( string & s ) { return (_Istream_Sstr)@{ .s = s, { {.scanset = 0p}, .wd = -1, {.flags.ignore = true} } }; }
        _Istream_Sstr & ignore( _Istream_Swidth & f ) { f.flags.ignore = true; return (_Istream_Sstr &)f; }
        _Istream_Squote & ignore( _Istream_Squote & f ) { f.sstr.flags.ignore = true; return (_Istream_Squote &)f; }
        _Istream_Sstr & ignore( _Istream_Sstr & f ) { f.flags.ignore = true; return (_Istream_Sstr &)f; }
} // distribution

forall( istype & | basic_istream( istype ) ) {
        istype & ?|?( istype & is, _Istream_Squote f );
        istype & ?|?( istype & is, _Istream_Sstr f );
        static inline istype & ?|?( istype & is, _Istream_Swidth f ) { return is | *(_Istream_Sstr *)&f; }
}

// Concatenation
void ?+=?( string & s, char c );
PBOOST void ?+=?( string & s, string );
void append( string & s, const string & s2, size_t maxlen );
void ?+=?( string & s, const char * cs );
void append( string & s, const char * buffer, size_t bsize );

string ?+?( string s, char c );
string ?+?( char c, string s );
PBOOST string ?+?( string s, string s2 );
string ?+?( const char * cs, char c );                                  // not backwards compatible
string ?+?( char c, const char * cs );
string ?+?( const char * cs1 , const char * cs2 );
string ?+?( const char * cs , string s );
string ?+?( string s, const char * cs  );
string ?+?( char, char );                                                               // not being called 8-(

static inline string & strcat( string & s1, const string & s2 ) { s1 += s2; return s1; }
static inline string & strcat( string & s, const char * cs  ) { s += cs; return s; }
static inline string & strncat( string & s1, const string & s2, size_t maxlen ) { append( s1, s2, maxlen ); return s1; }
static inline string & strncat( string & s, const char * buffer, size_t bsize ) { append( s, buffer, bsize ); return s; }

// Repetition

// Type `signed long long int` chosen for `factor` argument to achieve cost detente.
// This way, the call `'a' * 3` gets the same safe conversion cost calling here as for
// the built-in definition `int * int`.
typedef signed long long int strmul_factor_t;

void ?*=?( string & s, strmul_factor_t factor );
string ?*?( char c, strmul_factor_t factor );                                   // not backwards compatible
PBOOST string ?*?( string s, strmul_factor_t factor );
string ?*?( const char * cs, strmul_factor_t factor );
static inline string ?*?( strmul_factor_t factor, char c ) { return c * factor; }
PBOOST static inline string ?*?( strmul_factor_t factor, string s ) { return s * factor; }
static inline string ?*?( strmul_factor_t factor, const char * cs ) { return cs * factor; }

// Character access
char ?[?]( const string & s, size_t index );
string ?[?]( string & s, size_t index );                                // mutable length-1 slice of original
//char codePointAt(const string & s, size_t index );            // to revisit under Unicode

// Comparisons
static inline int strcmp( const string & s1, const string & s2 ) { return strcmp( *s1.inner, *s2.inner ); }
int strncmp( const string & s1, const string & s2, size_t maxlen );
static inline bool ?==?( const string & s1, const string & s2 ) { return *s1.inner == *s2.inner; }
static inline bool ?!=?( const string & s1, const string & s2 ) { return *s1.inner != *s2.inner; }
static inline bool ?>? ( const string & s1, const string & s2 ) { return *s1.inner >  *s2.inner; }
static inline bool ?>=?( const string & s1, const string & s2 ) { return *s1.inner >= *s2.inner; }
static inline bool ?<=?( const string & s1, const string & s2 ) { return *s1.inner <= *s2.inner; }
static inline bool ?<? ( const string & s1, const string & s2 ) { return *s1.inner <  *s2.inner; }

static inline int strcmp( const string & s1, const char * s2 ) { return strcmp( *s1.inner, s2 ); }
int strncmp( const string & s1, const char * s2, size_t maxlen );
static inline bool ?==?( const string & s1, const char * s2 ) { return *s1.inner == s2; }
static inline bool ?!=?( const string & s1, const char * s2 ) { return *s1.inner != s2; }
static inline bool ?>? ( const string & s1, const char * s2 ) { return *s1.inner >  s2; }
static inline bool ?>=?( const string & s1, const char * s2 ) { return *s1.inner >= s2; }
static inline bool ?<=?( const string & s1, const char * s2 ) { return *s1.inner <= s2; }
static inline bool ?<? ( const string & s1, const char * s2 ) { return *s1.inner <  s2; }

static inline int strcmp( const char * s1, const string & s2 ) { return strcmp( s1, *s2.inner ); }
int strncmp( const char * s1, const string & s2, size_t maxlen );
static inline bool ?==?( const char * s1, const string & s2 ) { return s1 == *s2.inner; }
static inline bool ?!=?( const char * s1, const string & s2 ) { return s1 != *s2.inner; }
static inline bool ?>? ( const char * s1, const string & s2 ) { return s1 >  *s2.inner; }
static inline bool ?>=?( const char * s1, const string & s2 ) { return s1 >= *s2.inner; }
static inline bool ?<=?( const char * s1, const string & s2 ) { return s1 <= *s2.inner; }
static inline bool ?<? ( const char * s1, const string & s2 ) { return s1 <  *s2.inner; }

// String search
bool contains( const string & s, char ch );                             // single character

//int find( const string & s, size_t start, size_t len, const string & key, size_t kstart, size_t klen );
size_t find$( const string_res & s, size_t start, size_t len, const string & key_res, size_t kstart, size_t klen );

size_t find( const string & s, char key );
size_t find( const string & s, const char * key );
size_t find( const string & s, const string & key );
size_t find( const string & s, const char * key, size_t keysize );

size_t find( const string & s, size_t start, char key );
size_t find( const string & s, size_t start, const string & key );
size_t find( const string & s, size_t start, const char * key );
size_t find( const string & s, size_t start, const char * key, size_t keysize );

bool includes( const string & s, const string & mask );
bool includes( const string & s, const char * mask );
bool includes( const string & s, const char * mask, size_t masksize );

bool startsWith( const string & s, const string & prefix );
bool startsWith( const string & s, const char * prefix );
bool startsWith( const string & s, const char * prefix, size_t prefixsize );

bool endsWith( const string & s, const string & suffix );
bool endsWith( const string & s, const char * suffix );
bool endsWith( const string & s, const char * suffix, size_t suffixsize );

// Slicing
string ?()( string & s, ssize_t start, ssize_t len );
static inline string ?()( const string & s, ssize_t start, ssize_t len ) { string & w = (string &)s; return w( start, len ); } // FIX ME
string ?()( string & s, ssize_t start );
static inline string ?()( const string & s, ssize_t start ) { string & w = (string &)s; return w( start ); } // FIX ME
static inline string ?()( string & s, char m ) { return s( find( s, m ), 1 )`share; }
static inline string ?()( const string & s, char m ) { string & w = (string &)s; return w( find( s, m ), 1 )`share; } // FIX ME
static inline string ?()( string & s, const char * m ) { return s( find( s, m ), len( m ) )`share; }
static inline string ?()( const string & s, const char * m ) { string & w = (string &)s; return w( find( s, m ), len( m ) )`share; } // FIX ME
static inline string ?()( string & s, const string & m ) { return s( find( s, m ), len( m ) )`share; }
static inline string ?()( const string & s, const string & m ) { string & w = (string &)s; return w( find( s, m ), len( m ) )`share; } // FIX ME

struct charclass {
        charclass_res * inner;
};

void ?{}( charclass & ) = void;
void ?{}( charclass &, charclass ) = void;
charclass ?=?( charclass &, charclass ) = void;

void ?{}( charclass &, const string & chars );
void ?{}( charclass &, const char * chars );
void ?{}( charclass &, const char * chars, size_t charssize );
void ^?{}( charclass & );

size_t include( const string & s, const charclass & mask );
static inline size_t include( const string & s, const char * mask ) { return include( s, (charclass){ mask } ); }
static inline size_t include( const string & s, const string & mask ) { return include( s, (charclass){ mask } ); }
static inline size_t include( const char * cs, const charclass & mask ) { return include( (string){ cs }, mask ); }
static inline size_t include( const char * cs, const char * mask ) { return include( (string){ cs }, (charclass){ mask } ); }
static inline size_t include( const char * cs, const string & mask ) { return include( (string){ cs }, (charclass){ mask } ); }

static inline string include( const string & s, const charclass & mask ) { return s( 0, include( s, mask ) ); }
static inline string include( const string & s, const char * mask ) { return s( 0, include( s, (charclass){ mask } ) ); }
static inline string include( const string & s, const string & mask ) { return s( 0, include( s, (charclass){ mask } ) ); }
static inline string include( const char * cs, const charclass & mask ) { const string s = cs; return s( 0, include( s, mask ) ); }
static inline string include( const char * cs, const char * mask ) { const string s = cs; return s( 0, include( s, (charclass){ mask } ) ); }
static inline string include( const char * cs, const string & mask ) { const string s = cs; return s( 0, include( s, (charclass){ mask } ) ); }

size_t exclude( const string & s, const charclass & mask );
static inline size_t exclude( const string & s, const char * mask ) { return exclude( s, (charclass){ mask } ); }
static inline size_t exclude( const string & s, const string & mask ) { return exclude( s, (charclass){ mask } ); }
static inline size_t exclude( const char * cs, const charclass & mask ) { return exclude( (string){ cs }, mask ); }
static inline size_t exclude( const char * cs, const string & mask ) { return exclude( (string){ cs }, (charclass){ mask } ); }
static inline size_t exclude( const char * cs, const char * mask ) { return exclude( (string){ cs }, (charclass){ mask } ); }

static inline string exclude( const string & s, const charclass & mask ) { return s( 0, exclude( s, mask ) ); }
static inline string exclude( const string & s, const char * mask ) { return s( 0, exclude( s, (charclass){ mask } ) ); }
static inline string exclude( const string & s, const string & mask ) { return s( 0, exclude( s, (charclass){ mask } ) ); }
static inline string exclude( const char * cs, const charclass & mask ) { const string s = cs; return s( 0, exclude( s, mask ) ); }
static inline string exclude( const char * cs, const string & mask ) { const string s = cs; return s( 0, exclude( s, (charclass){ mask } ) ); }
static inline string exclude( const char * cs, const char * mask ) { const string s = cs; return s( 0, exclude( s, (charclass){ mask } ) ); }

size_t include( const string & s, int (* f)( int ) );   // for C character-class functions, e.g., isdigit
static inline size_t include( const char * cs, int (* f)( int ) ) { return include( (string){ cs }, f ); }
static inline string include( const string & s, int (* f)( int ) ) { return s( 0, include( s, f ) ); }
static inline string include( const char * cs, int (* f)( int ) ) { const string s = cs; return s( 0, include( s, f ) ); }

static inline size_t include( const string & s, bool (* f)( char ) ) { return include( s, (int (*)( int ))f ); }
static inline size_t include( const char * cs, bool (* f)( char ) ) { return include( (string){ cs }, f ); }
static inline string include( const string & s, bool (* f)( char ) ) { return s( 0, include( s, f ) ); }
static inline string include( const char * cs, bool (* f)( char ) ) { const string s = cs; return s( 0, include( s, f ) ); }

size_t exclude( const string & s, int (* f)( int ) );   // for C character-class functions, e.g., isdigit
static inline size_t exclude( const char * cs, int (* f)( int ) ) { return exclude( (string){ cs }, f ); }
static inline string exclude( const string & s, int (* f)( int ) ) { return s( 0, exclude( s, f ) ); }
static inline string exclude( const char * cs, int (* f)( int ) ) { const string s = cs; return s( 0, exclude( s, f ) ); }

static inline size_t exclude( const string & s, bool (* f)( char ) ) { return exclude( s, (int (*)( int ))f ); }
static inline size_t exclude( const char * cs, bool (* f)( char ) ) { return exclude( (string){ cs }, f ); }
static inline string exclude( const string & s, bool (* f)( char ) ) { return s( 0, exclude( s, f ) ); }
static inline string exclude( const char * cs, bool (* f)( char ) ) { const string s = cs; return s( 0, exclude( s, f ) ); }

string replace( const string & s, const string & from, const string & to );
static inline string replace( const char * cs, const char * from, const char * to ) { return replace( (string){ cs }, (string){ from }, (string){ to } ); }
static inline string replace( const string & s, const char * from, const char * to ) { return replace( s, (string){ from }, (string){ to } ); }
static inline string replace( const string & s, const char * from, const string & to ) { return replace( s, (string){ from }, to ); }
static inline string replace( const string & s, string & from, const char * to ) { return replace( s, from, (string){ to } ); }

string translate( const string & s, int (* f)( int ) ); // for C character-class functions, e.g., isdigit
static inline string translate( const char * cs, int (* f)( int ) ) { return translate( (string){ cs }, f ); }

static inline string translate( const string & s, bool (* f)( char ) ) { return translate( s, (int (*)( int ))f ); }
static inline string translate( const char * cs, bool (* f)( char ) ) { return translate( (string){ cs }, f ); }

#ifndef _COMPILING_STRING_CFA_
#undef PBOOST
#endif