source: libcfa/src/collections/string_res.hfa@ 5ad6f0d

//
// 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_res -- variable-length, mutable run of text, with resource semantics
//
// Author           : Michael L. Brooks
// Created On       : Fri Sep 03 11:00:00 2021
// Last Modified By : Peter A. Buhr
// Last Modified On : Sun Apr 13 21:03:37 2025
// Update Count     : 79
//

#pragma once

#include <iostream.hfa>
#include <string.h>    // e.g. strlen


//######################### HandleNode #########################
//private

struct VbyteHeap;

struct HandleNode {
        HandleNode * flink;                                                                     // forward link
        HandleNode * blink;                                                                     // backward link
        VbyteHeap * ulink;                                                                      // upward link

        char * s;                                                                                       // pointer to byte string
        unsigned int lnth;                                                                      // length of byte string
}; // HandleNode

VbyteHeap * DEBUG_string_heap();
size_t DEBUG_string_bytes_in_heap( VbyteHeap * heap );
size_t DEBUG_string_bytes_avail_until_gc( VbyteHeap * heap );
const char * DEBUG_string_heap_start( VbyteHeap * heap );

void TUNING_set_string_heap_liveness_threshold( double val );

//######################### String #########################

// A dynamically-sized string
struct string_res {
        HandleNode Handle; // chars, start, end, global neighbours
        bool shareSet_owns_ulink;
        string_res * shareSet_prev;
        string_res * shareSet_next;
};


//######################### charclass_res #########################

struct charclass_res {
        string_res chars;
};

void ?{}( charclass_res & ) = void;
void ?{}( charclass_res &, charclass_res) = void;
charclass_res ?=?( charclass_res &, charclass_res) = void;
void ?{}( charclass_res &, const string_res & chars);
void ?{}( charclass_res &, const char * chars );
void ?{}( charclass_res &, const char * chars, size_t charssize );
void ^?{}( charclass_res & );


//######################### String #########################

// Getters
static inline size_t len( const string_res & s ) { return s.Handle.lnth; }

// Constructors, Assignment Operators, Destructor
void ?{}(string_res & s); // empty string
void ?{}(string_res & s, const char * buffer, size_t bsize); // copy specific length from buffer
static inline void ?{}(string_res & s, const char * rhs) { // copy from string literal (NULL-terminated)
        (s){ rhs, strlen(rhs) };
}
static inline void ?{}(string_res & s, char c ) {
        ?{}( s, &c, 1);
}

// Deleting the copy constructors makes the compiler reject an attempt to call/return by value
void ?{}(string_res & s, const string_res & s2) = void;
void ?{}(string_res & s, string_res & s2) = void;

enum StrResInitMode { COPY_VALUE, SHARE_EDITS };
void ?{}(string_res & s, const string_res & src, StrResInitMode, size_t start, size_t len );
static inline void ?{}(string_res & s, const string_res & src, StrResInitMode mode ) {
        ?{}( s, src, mode, 0, len(src));
}
static inline void ?{}(string_res & s, const string_res & src, StrResInitMode mode, size_t maxlen ) {
        ?{}( s, src, mode, 0, (len(src) > maxlen)?maxlen:len(src) );
}
void ?{}( string_res & s, ssize_t rhs );
void ?{}( string_res & s, size_t rhs );
void ?{}( string_res & s, double rhs );
void ?{}( string_res & s, long double rhs );
void ?{}( string_res & s, double _Complex rhs );
void ?{}( string_res & s, long double _Complex rhs );

string_res & assign(string_res & s, const string_res & src, size_t maxlen); // copy specific length from other string
string_res & assign(string_res & s, const char * buffer, size_t bsize); // copy specific length from buffer
static inline string_res & ?=?(string_res & s, const char * c) {  // copy from string literal (NULL-terminated)
        return assign(s, c, strlen( c));
}
string_res & ?=?(string_res & s, const string_res & c);
string_res & ?=?(string_res & s, string_res & c);
string_res & ?=?(string_res & s, char c);

string_res & ?=?( string_res & s, ssize_t rhs );
string_res & ?=?( string_res & s, size_t rhs );
string_res & ?=?( string_res & s, double rhs );
string_res & ?=?( string_res & s, long double rhs );
string_res & ?=?( string_res & s, double _Complex rhs );
string_res & ?=?( string_res & s, long double _Complex rhs );

void ^?{}(string_res & s);

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

struct _Istream_Rwidth {
        string_res * s;
        inline _Istream_str_base;
}; // _Istream_Rwidth

struct _Istream_Rquote {
        // string_res * s;
        // inline _Istream_str_base;
        _Istream_Rwidth rstr;
}; // _Istream_Rquote

struct _Istream_Rstr {
        string_res * s;
        inline _Istream_str_base;
//      _Istream_Rwidth rstr;
}; // _Istream_Rstr

static inline {
        // read width does not include null terminator
        _Istream_Rwidth wdi( unsigned int rwd, string_res & s ) { return (_Istream_Rwidth)@{ .s = &s, { {.scanset = 0p}, .wd = rwd, {.flags.rwd = true} } }; }
        _Istream_Rstr getline( string_res & s, const char delimiter = '\n' ) {
                return (_Istream_Rstr)@{ .s = &s, { {.delimiters = { delimiter, '\0' } }, .wd = -1, {.flags.delimiter = true} } };
        }
        _Istream_Rstr & getline( _Istream_Rwidth & f, const char delimiter = '\n' ) {
                f.delimiters[0] = delimiter; f.delimiters[1] = '\0'; f.flags.delimiter = true; return (_Istream_Rstr &)f;
        }
        _Istream_Rquote quote( string_res & s, const char Ldelimiter = '\"', const char Rdelimiter = '\0' ) {
                return (_Istream_Rquote)@{ { .s = &s, { {.delimiters = { Ldelimiter, Rdelimiter, '\0' }}, .wd = -1, {.flags.rwd = true} } } };
        }
        _Istream_Rquote & quote( _Istream_Rwidth & f, const char Ldelimiter = '"', const char Rdelimiter = '\0' ) {
                f.delimiters[0] = Ldelimiter;  f.delimiters[1] = Rdelimiter;  f.delimiters[2] = '\0';
                return (_Istream_Rquote &)f;
        }
        _Istream_Rstr incl( const char scanset[], string_res & s ) { return (_Istream_Rstr)@{ .s = &s, { {.scanset = scanset}, .wd = -1, {.flags.inex = false} } }; }
        _Istream_Rstr & incl( const char scanset[], _Istream_Rwidth & f ) { f.scanset = scanset; f.flags.inex = false; return (_Istream_Rstr &)f; }
        _Istream_Rstr excl( const char scanset[], string_res & s ) { return (_Istream_Rstr)@{ .s = &s, { {.scanset = scanset}, .wd = -1, {.flags.inex = true} } }; }
        _Istream_Rstr & excl( const char scanset[], _Istream_Rwidth & f ) { f.scanset = scanset; f.flags.inex = true; return (_Istream_Rstr &)f; }
        _Istream_Rstr ignore( string_res & s ) { return (_Istream_Rstr)@{ .s = &s, { {.scanset = 0p}, .wd = -1, {.flags.ignore = true} } }; }
        _Istream_Rstr & ignore( _Istream_Rwidth & f ) { f.flags.ignore = true; return (_Istream_Rstr &)f; }
        _Istream_Rquote & ignore( _Istream_Rquote & f ) { f.rstr.flags.ignore = true; return (_Istream_Rquote &)f; }
        _Istream_Rstr & ignore( _Istream_Rstr & f ) { f.flags.ignore = true; return (_Istream_Rstr &)f; }
} // distribution
forall( istype & | basic_istream( istype ) ) {
        istype & ?|?( istype & is, _Istream_Rquote f );
        istype & ?|?( istype & is, _Istream_Rstr f );
        static inline istype & ?|?( istype & is, _Istream_Rwidth f ) { return is | *(_Istream_Rstr *)&f; }
}

// Concatenation
void append( string_res & s, const char * buffer, size_t bsize );
void append( string_res & s, const string_res & s2, size_t maxlen );
static inline void ?+=?( string_res & s, const string_res & s2 ) { append( s, s2.Handle.s, s2.Handle.lnth ); }
static inline void ?+=?( string_res & s, char c ) { append( s, & c, 1 ); }
static inline void ?+=?( string_res & s, const char * c ) { append( s, c, strlen( c ) ); }
static inline string_res & strcat( string_res & s, const string_res & s2 ) { s += s2; return s; }
static inline string_res & strcat( string_res & s, const char * c ) { s += c; return s; }
static inline string_res & strncat( string_res & s, const string_res & s2, size_t maxlen ) { append(s, s2, maxlen); return s; }
static inline string_res & strncat( string_res & s, const char * buffer, size_t bsize ) { append(s, buffer, bsize); return s; }

// Repetition
void ?*=?(string_res & s, size_t factor);

// Character access
void assignAt( const string_res & s, size_t index, char val);
char ?[?]( const string_res & s, size_t index); // Mike changed to ret by val from Sunjay's ref, to match Peter's
//char codePointAt( const string_res & s, size_t index); // revisit under Unicode

// Comparisons
int strcmp$( const char * s1, size_t l1, const char * s2, size_t l2 );

static inline int strcmp( const string_res & s1, const string_res & s2 ) { return strcmp$( s1.Handle.s, s1.Handle.lnth, s2.Handle.s, s2.Handle.lnth ); }
static inline bool ?==?( const string_res & s1, const string_res & s2 ) { return strcmp( s1, s2 ) == 0; }
static inline bool ?!=?( const string_res & s1, const string_res & s2 ) { return strcmp( s1, s2 ) != 0; }
static inline bool ?>? ( const string_res & s1, const string_res & s2 ) { return strcmp( s1, s2 ) >  0; }
static inline bool ?>=?( const string_res & s1, const string_res & s2 ) { return strcmp( s1, s2 ) >= 0; }
static inline bool ?<=?( const string_res & s1, const string_res & s2 ) { return strcmp( s1, s2 ) <= 0; }
static inline bool ?<? ( const string_res & s1, const string_res & s2 ) { return strcmp( s1, s2 ) <  0; }

static inline int strcmp( const string_res & s1, const char * s2 ) { return strcmp$( s1.Handle.s, s1.Handle.lnth, s2, strlen( s2 ) ); }
static inline bool ?==?( const string_res & s1, const char * s2 ) { return strcmp( s1, s2 ) == 0; }
static inline bool ?!=?( const string_res & s1, const char * s2 ) { return strcmp( s1, s2 ) != 0; }
static inline bool ?>? ( const string_res & s1, const char * s2 ) { return strcmp( s1, s2 ) >  0; }
static inline bool ?>=?( const string_res & s1, const char * s2 ) { return strcmp( s1, s2 ) >= 0; }
static inline bool ?<=?( const string_res & s1, const char * s2 ) { return strcmp( s1, s2 ) <= 0; }
static inline bool ?<? ( const string_res & s1, const char * s2 ) { return strcmp( s1, s2 ) <  0; }

static inline int strcmp( const char * s1, const string_res & s2 ) { return strcmp$( s1, strlen( s1 ), s2.Handle.s, s2.Handle.lnth ); }
static inline bool ?==?( const char * s1, const string_res & s2 ) { return strcmp( s1, s2 ) == 0; }
static inline bool ?!=?( const char * s1, const string_res & s2 ) { return strcmp( s1, s2 ) != 0; }
static inline bool ?>? ( const char * s1, const string_res & s2 ) { return strcmp( s1, s2 ) >  0; }
static inline bool ?>=?( const char * s1, const string_res & s2 ) { return strcmp( s1, s2 ) >= 0; }
static inline bool ?<=?( const char * s1, const string_res & s2 ) { return strcmp( s1, s2 ) <= 0; }
static inline bool ?<? ( const char * s1, const string_res & s2 ) { return strcmp( s1, s2 ) <  0; }

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

int find$( const string_res & s, ssize_t start, ssize_t len, const string_res & key, ssize_t kstart, ssize_t klen );

int find( const string_res & s, char search);
int find( const string_res & s, const string_res & search);
int find( const string_res & s, const char * search);
int find( const string_res & s, const char * search, size_t searchsize);

int findFrom( const string_res & s, size_t fromPos, char search);
int findFrom( const string_res & s, size_t fromPos, const string_res & search);
int findFrom( const string_res & s, size_t fromPos, const char * search);
int findFrom( const string_res & s, size_t fromPos, const char * search, size_t searchsize);

bool includes( const string_res & s, const string_res & search);
bool includes( const string_res & s, const char * search);
bool includes( const string_res & s, const char * search, size_t searchsize);

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

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

int include( const string_res & s, const charclass_res & mask);
int exclude( const string_res & s, const charclass_res & mask);