source: libcfa/src/collections/string.cfa@ b14d0d97

//
// 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 Sep 15 10:26:35 2025
// Update Count     : 394
//

#define _COMPILING_STRING_CFA_

#include "string.hfa"
#include "string_res.hfa"
#include <stdlib.hfa>

#pragma GCC visibility push(default)

/*
Implementation Principle: typical operation translates to the equivalent
operation on `inner`.  Exceptions are implementing new RAII pattern for value
semantics and some const-hell handling.
*/

////////////////////////////////////////////////////////
// string RAII

// private (not in header)
static void ?{}( string & s, string_res & src, size_t start, size_t len ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, src, SHARE_EDITS, start, len );
}

void ?{}( string & s ) {
        (s.inner) { malloc() };
        ?{}( *s.inner );
}

PBOOST void ?{}( string & s, string c ) {  // c is a memcpy of the real src string
        (s.inner) { malloc() };
        ?{}( *s.inner, *c.inner, COPY_VALUE );
}

void ?{}( string & s, string s2, size_t maxlen ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, *s2.inner, COPY_VALUE, maxlen );
}

void ?{}( string & s, char c ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, c );
}

void ?{}( string & s, const char * c ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, c );
}

void ?{}( string & s, const char * c, size_t size ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, c, size );
}

void ?{}( string & s, signed long int rhs ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, rhs );
}

void ?{}( string & s, size_t rhs ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, rhs );
}

void ?{}( string & s, double rhs ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, rhs );
}

void ?{}( string & s, long double rhs ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, rhs );
}

void ?{}( string & s, double _Complex rhs ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, rhs );
}

void ?{}( string & s, long double _Complex rhs ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, rhs );
}

void ^?{}( string & s ) {
        ^(*s.inner){};
        free( s.inner );
        s.inner = 0p;
}

////////////////////////////////////////////////////////
// Alternate construction: request shared edits

string_Share ?`share( string & s ) {
        string_Share ret = { &s };
        return ret;
}

void ?{}( string & s, string_Share src ) {
        ?{}( s, *src.s->inner, 0, src.s->inner->Handle.lnth );
}

////////////////////////////////////////////////////////
// Assignment

PBOOST string & ?=?( string & s, string c ) {
        (*s.inner) = (*c.inner);
        return s;
}

string & ?=?( string & s, const char * val ) {
        (*s.inner) = val;
        return s;
}

string & ?=?( string & s, char val ) {
        (*s.inner) = val;
        return s;
}

string & assign( string & s, const string & c, size_t n ) {
        assign( *s.inner, *c.inner, n );
        return s;
}

string & assign( string & s, const char * c, size_t n ) {
        assign( *s.inner, c, n );
        return s;
}

string & ?=?( string & s, signed long int rhs ) {
        (*s.inner) = rhs;
        return s;
}

string & ?=?( string & s, size_t rhs ) {
        (*s.inner) = rhs;
        return s;
}

string & ?=?( string & s, double rhs ) {
        (*s.inner) = rhs;
        return s;
}

string & ?=?( string & s, long double rhs ) {
        (*s.inner) = rhs;
        return s;
}

string & ?=?( string & s, double _Complex rhs ) {
        (*s.inner) = rhs;
        return s;
}

string & ?=?( string & s, long double _Complex rhs ) {
        (*s.inner) = rhs;
        return s;
}

////////////////////////////////////////////////////////
// C-style

// safe conversion from string to char *
char * strncpy( char * dst, string & src, size_t n ) {
        size_t l = min( n - 1, len( src ) );                            // ensure null terminated
        for ( i; l ) dst[i] = src[i];
        dst[l] = '\0';
        return dst;
}
char * ?=?( char *& dst, string & src ) {
        dst = aalloc( len( src ) + 1 );                                 // ensure null terminated
        for ( i; len( src ) ) dst[i] = src[i];
        dst[len(src)] = '\0';
        return dst;
}
void ?{}( char *& dst, string & src ) {
        dst = aalloc( len( src ) + 1 );                                 // ensure null terminated
        for ( i; len( src ) ) dst[i] = src[i];
        dst[len(src)] = '\0';
}

size_t strnlen( const string & s, size_t maxlen ) { return min( len( s ), maxlen ); }

////////////////////////////////////////////////////////
// Input-Output

forall( ostype & | basic_ostream( ostype ) ) {
        ostype & ?|?( ostype & out, string s ) {
                return out | (*s.inner); // print internal string_res
        }

        void ?|?( ostype & out, string s ) with ( basic_ostream_table ) {
                (ostype &)(out | (*s.inner)); ends( out );
        }

        ostype & ?|?( ostype & os, _Ostream_Manip(string) f ) {
                size_t l = len( f.val );
                char cstr[l + 1];                                                                       // room for null terminator
                for ( i; l ) cstr[i] = f.val[i];                                        // copy string
                cstr[l] = '\0';                                                                         // terminate
                _Ostream_Manip(const char *) cf @= { cstr, f.wd, f.pc, f.base, {f.all} };
                return os | cf | nonl;
        } // ?|?

        void ?|?( ostype & os, _Ostream_Manip(string) f ) with ( basic_ostream_table ) {
                (ostype &)(os | f); ends( os );
        }
}

forall( istype & | basic_istream( istype ) ) {
        istype & ?|?( istype & in, string & s ) {
                return in | (*s.inner); // read to internal string_res
        }

        istype & ?|?( istype & is, _Istream_Squote f ) {
                _Istream_Rquote f2 = { { f.sstr.s.inner, (_Istream_str_base)f.sstr } };
                return is | f2;
        } // ?|?

        istype & ?|?( istype & is, _Istream_Sstr f ) {
        //      _Istream_Rstr f2 = {f.sstr.s.inner, (_Istream_str_base)f.sstr};
                _Istream_Rstr f2 = {f.s.inner, (_Istream_str_base)f};
                return is | f2;
        } // ?|?
}

////////////////////////////////////////////////////////
// Slicing

string ?()( string & s, ssize_t start, ssize_t len ) {
        if ( start < 0 ) start += len( s );
        if ( len < 0 ) { len = -len; start -= len - 1; }
        if ( start < 0 || start >= len( s ) ) return (string){ "" };
        if ( start + len > len( s ) ) len = len( s ) - start;
        string ret = { *s.inner, start, len };
        return ret`share;
}

string ?()( string & s, ssize_t start ) {
        if ( start < 0 ) { start += len( s ); }
        string ret = { *s.inner, start, len( s ) - start };
        return ret`share;
}

////////////////////////////////////////////////////////
// Concatenation

void ?+=?( string & s, char c ) {
        (*s.inner) += c;
}

PBOOST void ?+=?( string & s, string s2 ) {
        (*s.inner) += (*s2.inner);
}

void append( string & s, const string & s2, size_t maxlen ) {
        append( (*s.inner), (*s2.inner), maxlen );
}

void ?+=?( string & s, const char * c ) {
        (*s.inner) += c;
}

void append( string & s, const char * buffer, size_t bsize ) {
        append( (*s.inner), buffer, bsize );
}

string ?+?( string s, char c ) {
        string ret = s;
        ret += c;
        return ret;
}

string ?+?( char c, string s ) {
        string ret = c;
        ret += s;
        return ret;
}

PBOOST string ?+?( string s, string s2 ) {
        string ret = s;
        ret += s2;
        return ret;
}

string ?+?( const char * s,  char c ) {
        string ret = s;
        ret += c;
        return ret;
}

string ?+?( char c, const char * s ) {
        string ret = c;
        ret += s;
        return ret;
}

string ?+?( const char * s1, const char * s2 ) {
        string ret = s1;
        ret += s2;
        return ret;
}

string ?+?( const char * s1, string s2 ) {
        string ret = s1;
        ret += s2;
        return ret;
}

string ?+?( string s, const char * c ) {
        string ret = s;
        ret += c;
        return ret;
}

string ?+?( char c1, char c2 ) {
        string ret = c1;
        ret += c2;
        return ret;
}

////////////////////////////////////////////////////////
// Repetition

void ?*=?( string & s, strmul_factor_t factor ) {
        (*s.inner) *= factor;
}

PBOOST string ?*?( string s, strmul_factor_t factor ) {
        string ret = s;
        ret *= factor;
        return ret;
}

string ?*?( char c, strmul_factor_t factor ) {
        string ret = c;
        ret *= factor;
        return ret;
}

string ?*?( const char * s, strmul_factor_t factor ) {
        string ret = s;
        ret *= factor;
        return ret;
}

////////////////////////////////////////////////////////
// Character access

char ?[?]( const string & s, size_t index ) {
        return (*s.inner)[index];
}

string ?[?]( string & s, size_t index ) {
        string ret = { *s.inner, index, 1 };
        return ret`share;
}

////////////////////////////////////////////////////////
// Comparison

#define STRNCPY_FMT "**** Error **** strncpy: maximum length %zu is greater than string lengths %zd or %zd."

int strncmp( const string & s1, const string & s2, size_t maxlen ) {
        if ( maxlen > len( s1 ) || maxlen > len( s2 ) ) {
                abort( STRNCPY_FMT, maxlen, len( s1 ), len( s2 ) );
        } // if
        return strcmp$( s1.inner->Handle.s, maxlen, s2.inner->Handle.s, maxlen );
}

int strncmp( const string & s1, const char * s2, size_t maxlen ) {
        size_t s2len = len( s2 );
        if ( maxlen > len( s1 ) || maxlen > s2len ) {
                abort( STRNCPY_FMT, maxlen, len( s1 ), s2len );
        } // if
        return strcmp$( s1.inner->Handle.s, maxlen, s2, maxlen );
}

int strncmp( const char * s1, const string & s2, size_t maxlen ) {
        size_t s1len = len( s1 );
        if ( maxlen > s1len || maxlen > len( s2 ) ) {
                abort( STRNCPY_FMT, maxlen, s1len, len( s2 ) );
        } // if
        return strcmp$( s1, maxlen, s2.inner->Handle.s, maxlen );
}

////////////////////////////////////////////////////////
// Search

bool contains( const string & s, char ch ) {
        return contains( *s.inner, ch );
}

size_t find( const string & s, size_t start, size_t len, const string & key, size_t kstart, size_t klen ) {
        if ( start < 0 ) { start += len( s ); }
        if ( len < 0 ) { len = -len; start -= len; }
        if ( start >= len( s ) ) return 0;
        if ( start + len > len( s ) ) len = len( s ) - start;

        if ( kstart < 0 ) { kstart += len( key ); }
        if ( klen < 0 ) { klen = -klen; kstart -= klen; }
        if ( kstart >= len( key ) ) return 0;
        if ( kstart + klen > len( key ) ) klen = len( key ) - kstart;

        return findFrom( *s.inner, start, *key.inner );
}

size_t find( const string & s, char key ) {
        return find( *s.inner, key );
}

size_t find( const string & s, const string & key ) {
        return find( *s.inner, *key.inner );
}

size_t find( const string & s, const char * key ) {
        return find( *s.inner, key );
}

size_t find( const string & s, const char * key, size_t keysize ) {
        return find( *s.inner, key, keysize );
}

size_t find( const string & s, size_t start, char key ) {
        return findFrom( *s.inner, start, key );
}

size_t find( const string & s, size_t start, const char * key ) {
        return findFrom( *s.inner, start, key );
}

size_t find( const string & s, size_t start, const char * key, size_t keysize ) {
        return findFrom( *s.inner, start, key, keysize );
}

bool includes( const string & s, const string & mask ) {
        return includes( *s.inner, *mask.inner );
}

bool includes( const string & s, const char * mask ) {
        return includes( *s.inner, mask );
}

bool includes( const string & s, const char * mask, size_t masksize ) {
        return includes( *s.inner, mask, masksize );
}

bool startsWith( const string & s, const string & prefix ) {
        return startsWith( *s.inner, *prefix.inner );
}

bool startsWith( const string & s, const char * prefix ) {
        return startsWith( *s.inner, prefix );
}

bool startsWith( const string & s, const char * prefix, size_t prefixsize ) {
        return startsWith( *s.inner, prefix, prefixsize );
}

bool endsWith( const string & s, const string & suffix ) {
        return endsWith( *s.inner, *suffix.inner );
}

bool endsWith( const string & s, const char * suffix ) {
        return endsWith( *s.inner, suffix );
}

bool endsWith( const string & s, const char * suffix, size_t suffixsize ) {
        return endsWith( *s.inner, suffix, suffixsize );
}


///////////////////////////////////////////////////////////////////////////
// charclass, include, exclude

void ?{}( charclass & s, const string & chars ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, *(const string_res *)chars.inner );
}

void ?{}( charclass & s, const char * chars ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, chars );
}

void ?{}( charclass & s, const char * chars, size_t charssize ) {
        (s.inner) { malloc() };
        ?{}( *s.inner, chars, charssize );
}

void ^?{}( charclass & s ) {
        ^(*s.inner){};
        free( s.inner );
        s.inner = 0p;
}

size_t exclude( const string & s, const charclass & mask ) {
        return exclude( *s.inner, *mask.inner );
}

size_t include( const string & s, const charclass & mask ) {
        return include( *s.inner, *mask.inner );
}

size_t include( const string & s, int (*f)( int ) ) {
        size_t l = len( s );
        for ( i; l ) {
                if ( ! f( s[i] ) ) return i;
        } // for
        return l;
}

size_t exclude( const string & s, int (*f)( int ) ) {
        size_t l = len( s );
        for ( i; l ) {
                if ( f( s[i] ) ) return i;
        } // for
        return l;
}

string replace( const string & s, const string & from, const string & to ) {
        ssize_t pos;
    string r;

    pos = find( s, from );
    if ( pos < len( s ) ) {
                r = s( 0, pos ) + to + replace( s( pos + (ssize_t)len( from ) ), from, to );
    } else {
                r = s;
    } // if
    return r;
}

string translate( const string & s, int (*f)( int ) ) {
        string r = s;
        size_t l = len( r );
        for ( i; l ) {
                r[i] = (char)f( r[i] );
        } // for
        return r;
}