amxc_string_split.c File Reference

Ambiorix string API implementation. More...

#include <stdlib.h>
#include <string.h>
#include <ctype.h>
#include <amxc/amxc_common.h>
#include <amxc/amxc_variant.h>
#include <amxc/amxc_string_split.h>
#include <amxc/amxc_utils.h>
#include <amxc/amxc_macros.h>

Go to the source code of this file.

Data Structures
struct	_string_word_flags

Macros
#define	_GNU_SOURCE

Typedefs
typedef struct _string_word_flags	amxc_string_word_flags_t
typedef int(*	amxc_string_create_part_t) (const amxc_string_t *const string, amxc_llist_t *const list, const size_t start_pos, const size_t length)
typedef bool(*	amxc_string_check_delimiter_t) (amxc_llist_t *list, const char delimiter)

Functions
static void	amxc_trim_llist (amxc_llist_t *const list)
static int	amxc_string_create_part (const amxc_string_t *const string, amxc_llist_t *const list, const size_t start_pos, const size_t length)
static bool	amxc_string_split_update_status (const amxc_string_t *const string, size_t i, amxc_string_word_flags_t *flags)
static amxc_string_split_status_t	amxc_string_split_word_is_valid (amxc_string_word_flags_t *flags, const char **reason)
static bool	amxc_need_to_add_delimiter (amxc_llist_t *list, const char delimiter)
static amxc_string_split_status_t	amxc_string_split_words_internal (const amxc_string_t *const string, amxc_llist_t *list, amxc_string_create_part_t create, amxc_string_check_delimiter_t check, const char **reason)
static amxc_string_split_status_t	amxc_build_csv_var_list (amxc_llist_t *all, amxc_var_t *csv_list)
static amxc_string_split_status_t	amxc_build_ssv_var_list (amxc_llist_t *all, amxc_var_t *ssv_list)
amxc_string_split_status_t	amxc_string_split_word (const amxc_string_t *const string, amxc_llist_t *list, const char **reason)
	Split a string in individual words or punctuation signs. More...
amxc_string_split_status_t	amxc_string_split (const amxc_string_t *const string, amxc_var_t *var, amxc_string_split_builder_t fn, const char **reason)
	Split a string in individual words or punctuation signs. More...
amxc_string_split_status_t	amxc_string_csv_to_var (const amxc_string_t *const string, amxc_var_t *var, const char **reason)
	Split a string in individual parts assuming that the string contains comma separated values. More...
amxc_string_split_status_t	amxc_string_ssv_to_var (const amxc_string_t *const string, amxc_var_t *var, const char **reason)
	Split a string in individual parts assuming that the string contains space separated values. More...
amxc_string_split_status_t	amxc_string_split_to_llist (const amxc_string_t *const string, amxc_llist_t *list, const char separator)
	Simple split function using a single character separator. More...
amxc_string_t *	amxc_string_get_from_llist (const amxc_llist_t *const llist, const unsigned int index)
	Helper function to be used with amxc_string_split_llist. More...
const char *	amxc_string_get_text_from_llist (const amxc_llist_t *const llist, const unsigned int index)
	Helper function to be used with amxc_string_split_llist. More...

Detailed Description

Ambiorix string API implementation.

Definition in file amxc_string_split.c.

Macro Definition Documentation

_GNU_SOURCE

#define _GNU_SOURCE

Definition at line 56 of file amxc_string_split.c.

Typedef Documentation

amxc_string_check_delimiter_t

typedef bool(* amxc_string_check_delimiter_t) (amxc_llist_t *list, const char delimiter)

Definition at line 90 of file amxc_string_split.c.

amxc_string_create_part_t

typedef int(* amxc_string_create_part_t) (const amxc_string_t *const string, amxc_llist_t *const list, const size_t start_pos, const size_t length)

Definition at line 85 of file amxc_string_split.c.

amxc_string_word_flags_t

typedef struct _string_word_flags amxc_string_word_flags_t

Function Documentation

amxc_build_csv_var_list()

static amxc_string_split_status_t amxc_build_csv_var_list ( amxc_llist_t *  all, amxc_var_t *  csv_list  )

static

Definition at line 347 of file amxc_string_split.c.

                                                                 {
    amxc_string_split_status_t retval = AMXC_ERROR_STRING_SPLIT_INVALID_INPUT;
    bool quotes = false;
    bool sqbrackets = false;
    bool add_empty = true;
    bool last_is_comma = false;
    amxc_string_t csv_part;
 
    amxc_var_set_type(csv_list, AMXC_VAR_ID_LIST);
 
    amxc_string_init(&csv_part, 0);
    amxc_llist_for_each(it, all) {
        amxc_string_t* part = amxc_string_from_llist_it(it);
        const char* txt_part = amxc_string_get(part, 0);
        last_is_comma = false;
        if(amxc_string_text_length(part) == 1) {
            switch(txt_part[0]) {
            case '"':
            case '\'':
                amxc_string_append(&csv_part, txt_part, amxc_string_text_length(part));
                quotes = !quotes;
                break;
            case '[':
                if(amxc_string_is_empty(&csv_part)) {
                    csv_list = amxc_var_add(amxc_llist_t, csv_list, NULL);
                    sqbrackets = !sqbrackets;
                } else {
                    amxc_string_append(&csv_part, txt_part, amxc_string_text_length(part));
                }
                break;
            case ']':
                if(sqbrackets) {
                    if(amxc_string_text_length(&csv_part) > 0) {
                        amxc_var_t* item = amxc_var_add_new(csv_list);
                        amxc_string_trim(&csv_part, NULL);
                        amxc_var_push(cstring_t, item, amxc_string_take_buffer(&csv_part));
                    }
                    if(csv_list != NULL) {
                        csv_list = amxc_container_of(csv_list->lit.llist, amxc_var_t, data);
                    }
                    sqbrackets = !sqbrackets;
                    add_empty = false;
                } else {
                    amxc_string_append(&csv_part, txt_part, amxc_string_text_length(part));
                }
                break;
            case ',':
                if(quotes) {
                    amxc_string_append(&csv_part, txt_part, amxc_string_text_length(part));
                } else {
                    if(amxc_string_text_length(&csv_part) > 0) {
                        amxc_var_t* item = amxc_var_add_new(csv_list);
                        amxc_string_trim(&csv_part, NULL);
                        amxc_var_push(cstring_t, item, amxc_string_take_buffer(&csv_part));
                    } else if(add_empty) {
                        amxc_var_t* item = amxc_var_add_new(csv_list);
                        amxc_var_set(cstring_t, item, "");
                    }
                    add_empty = true;
                }
                last_is_comma = true;
                break;
            case ' ':
                amxc_string_append(&csv_part, txt_part, amxc_string_text_length(part));
                break;
            default:
                amxc_string_append(&csv_part, txt_part, amxc_string_text_length(part));
                break;
            }
        } else {
            amxc_string_append(&csv_part, txt_part, amxc_string_text_length(part));
        }
    }
    if(amxc_string_text_length(&csv_part) > 0) {
        amxc_var_t* item = amxc_var_add_new(csv_list);
        amxc_string_trim(&csv_part, NULL);
        amxc_var_push(cstring_t, item, amxc_string_take_buffer(&csv_part));
    } else if(last_is_comma && add_empty) {
        amxc_var_add(cstring_t, csv_list, "");
    }
    amxc_string_clean(&csv_part);
 
    retval = AMXC_STRING_SPLIT_OK;
 
    return retval;
}

amxc_build_ssv_var_list()

static amxc_string_split_status_t amxc_build_ssv_var_list ( amxc_llist_t *  all, amxc_var_t *  ssv_list  )

static

Definition at line 435 of file amxc_string_split.c.

                                                                 {
    amxc_string_split_status_t retval = AMXC_ERROR_STRING_SPLIT_INVALID_INPUT;
    bool quotes = false;
    bool sqbrackets = false;
    amxc_string_t csv_part;
 
    amxc_var_set_type(ssv_list, AMXC_VAR_ID_LIST);
 
    amxc_string_init(&csv_part, 0);
    amxc_llist_for_each(it, all) {
        amxc_string_t* part = amxc_string_from_llist_it(it);
        const char* txt_part = amxc_string_get(part, 0);
        if(amxc_string_text_length(part) == 1) {
            switch(txt_part[0]) {
            case '"':
            case '\'':
                amxc_string_append(&csv_part, txt_part, amxc_string_text_length(part));
                quotes = !quotes;
                break;
            case '[':
                ssv_list = amxc_var_add(amxc_llist_t, ssv_list, NULL);
                sqbrackets = !sqbrackets;
                break;
            case ']':
                if(amxc_string_text_length(&csv_part) > 0) {
                    amxc_var_t* item = amxc_var_add_new(ssv_list);
                    amxc_var_push(cstring_t, item, amxc_string_take_buffer(&csv_part));
                }
                if(ssv_list != NULL) {
                    ssv_list = amxc_container_of(ssv_list->lit.llist, amxc_var_t, data);
                }
                sqbrackets = !sqbrackets;
                break;
            case ' ':
                if(quotes) {
                    amxc_string_append(&csv_part, txt_part, amxc_string_text_length(part));
                } else {
                    if(amxc_string_text_length(&csv_part) > 0) {
                        amxc_var_t* item = amxc_var_add_new(ssv_list);
                        amxc_var_push(cstring_t, item, amxc_string_take_buffer(&csv_part));
                    }
                    amxc_string_reset(&csv_part);
                }
                break;
            default:
                amxc_string_append(&csv_part, txt_part, amxc_string_text_length(part));
                break;
            }
        } else {
            amxc_string_append(&csv_part, txt_part, amxc_string_text_length(part));
        }
    }
    if(amxc_string_text_length(&csv_part) > 0) {
        amxc_var_t* item = amxc_var_add_new(ssv_list);
        amxc_var_push(cstring_t, item, amxc_string_take_buffer(&csv_part));
    }
    amxc_string_clean(&csv_part);
 
    retval = AMXC_STRING_SPLIT_OK;
 
    return retval;
}

amxc_need_to_add_delimiter()

static bool amxc_need_to_add_delimiter ( amxc_llist_t *  list, const char  delimiter  )

static

Definition at line 253 of file amxc_string_split.c.

                                                             {
    bool retval = false;
    amxc_llist_it_t* it = amxc_llist_get_last(list);
    amxc_string_t* str_part = NULL;
    const char* part = NULL;
 
    if(it == NULL) {
        retval = (isspace(delimiter) == 0);
        goto exit;
    }
 
    str_part = amxc_string_from_llist_it(it);
    part = amxc_string_get(str_part, 0);
 
    retval = true;
 
    if(amxc_string_text_length(str_part) == 1) {
        if(isspace(part[0]) != 0) {
            if((isspace(delimiter) != 0) && (delimiter != '\n')) {
                retval = false;
            }
        }
    }
 
exit:
    return retval;
}

amxc_string_create_part()

static int amxc_string_create_part ( const amxc_string_t *const  string, amxc_llist_t *const  list, const size_t  start_pos, const size_t  length  )

static

Definition at line 127 of file amxc_string_split.c.

                                                        {
    int retval = -1;
    amxc_string_t* part = NULL;
    char* buffer = NULL;
 
    buffer = amxc_string_dup(string, start_pos, length);
    when_null(buffer, exit);
    when_failed(amxc_string_new(&part, 0), exit);
    if((length == 1) && isspace(buffer[0])) {
        amxc_string_set_at(part, 0, buffer, 1, amxc_string_no_flags);
        free(buffer);
    } else {
        when_failed(amxc_string_push_buffer(part, buffer, length + 1), exit);
    }
    amxc_llist_append(list, &part->it);
 
    retval = 0;
 
exit:
    if(retval != 0) {
        free(buffer);
        amxc_string_delete(&part);
    }
    return retval;
}

amxc_string_split_update_status()

static bool amxc_string_split_update_status ( const amxc_string_t *const  string, size_t  i, amxc_string_word_flags_t *  flags  )

static

Definition at line 156 of file amxc_string_split.c.

                                                                             {
    bool skip = false;
    if(string->buffer[i] == '\\') {
        flags->escape = true;
        skip = true;
    }
    flags->start_quote = false;
    if(!flags->between_single_quotes && (string->buffer[i] == '"')) {
        flags->between_double_quotes = !flags->between_double_quotes;
        flags->start_quote = flags->between_double_quotes;
    }
    if(!flags->between_double_quotes && (string->buffer[i] == '\'')) {
        flags->between_single_quotes = !flags->between_single_quotes;
        flags->start_quote = flags->between_single_quotes;
    }
    if(!(flags->between_double_quotes || flags->between_single_quotes)) {
        switch(string->buffer[i]) {
        case '(':
            flags->round_brackets++;
            break;
        case ')':
            flags->round_brackets--;
            break;
        case '{':
            flags->curly_brackets++;
            break;
        case '}':
            flags->curly_brackets--;
            break;
        case '[':
            flags->square_brackets++;
            break;
        case ']':
            flags->square_brackets--;
            break;
        }
    }
 
    return skip;
}

amxc_string_split_word_is_valid()

static amxc_string_split_status_t amxc_string_split_word_is_valid ( amxc_string_word_flags_t *  flags, const char **  reason  )

static

Definition at line 200 of file amxc_string_split.c.

                                                     {
    amxc_string_split_status_t retval = AMXC_STRING_SPLIT_OK;
    const char* msg = "";
 
    if(flags->between_double_quotes) {
        retval = AMXC_ERROR_STRING_MISSING_DQUOTE;
        msg = "Missing closing double quote - \"";
        goto exit;
    }
    if(flags->between_single_quotes) {
        retval = AMXC_ERROR_STRING_MISSING_SQUOTE;
        msg = "Missing closing single quote - '";
        goto exit;
    }
    if(flags->round_brackets > 0) {
        retval = AMXC_ERROR_STRING_MISSING_RBRACKET;
        msg = "Missing closing round bracket - )";
        goto exit;
    }
    if(flags->round_brackets < 0) {
        retval = AMXC_ERROR_STRING_MISSING_RBRACKET;
        msg = "Missing open round bracket - (";
        goto exit;
    }
    if(flags->curly_brackets > 0) {
        retval = AMXC_ERROR_STRING_MISSING_CBRACKET;
        msg = "Missing closing curly bracket - }";
        goto exit;
    }
    if(flags->curly_brackets < 0) {
        retval = AMXC_ERROR_STRING_MISSING_CBRACKET;
        msg = "Missing opening curly bracket - {";
        goto exit;
    }
    if(flags->square_brackets > 0) {
        retval = AMXC_ERROR_STRING_MISSING_SBRACKET;
        msg = "Missing closing square bracket - ]";
        goto exit;
    }
    if(flags->square_brackets < 0) {
        retval = AMXC_ERROR_STRING_MISSING_SBRACKET;
        msg = "Missing opening square bracket - [";
        goto exit;
    }
 
exit:
    if(reason != NULL) {
        *reason = msg;
    }
    return retval;
}

amxc_string_split_words_internal()

static amxc_string_split_status_t amxc_string_split_words_internal ( const amxc_string_t *const  string, amxc_llist_t *  list, amxc_string_create_part_t  create, amxc_string_check_delimiter_t  check, const char **  reason  )

static

Definition at line 283 of file amxc_string_split.c.

                                                      {
    amxc_string_word_flags_t flags;
    size_t start_pos = 0;
    size_t i = 0;
    amxc_string_split_status_t retval = AMXC_ERROR_STRING_SPLIT_INVALID_INPUT;
 
    flags.start_quote = false;
    flags.between_double_quotes = false;
    flags.between_single_quotes = false;
    flags.escape = false;
    flags.round_brackets = 0;
    flags.curly_brackets = 0;
    flags.square_brackets = 0;
 
    for(i = 0; i < string->last_used; i++) {
        if(flags.escape == false) {
            if(amxc_string_split_update_status(string, i, &flags) == true) {
                continue;
            }
        }
 
        flags.escape = false;
 
        if(((flags.between_double_quotes == false) &&
            (flags.between_single_quotes == false)) ||
           ( flags.start_quote == true)) {
            if(isspace(string->buffer[i])) {
                if(i - start_pos > 0) {
                    when_failed(create(string, list, start_pos, i - start_pos), exit);
                }
                if(check(list, string->buffer[i])) {
                    when_failed(create(string, list, i, 1), exit);
                }
                start_pos = i + 1;
                continue;
            }
            if(ispunct(string->buffer[i])) {
                if(i - start_pos > 0) {
                    when_failed(create(string, list, start_pos, i - start_pos), exit);
                }
                if(check(list, string->buffer[i])) {
                    when_failed(create(string, list, i, 1), exit);
                }
                start_pos = i + 1;
                continue;
            }
        }
    }
 
    if(i - start_pos != 0) {
        when_failed(create(string, list, start_pos, i - start_pos), exit);
    }
 
    retval = amxc_string_split_word_is_valid(&flags, reason);
 
exit:
    return retval;
}

amxc_trim_llist()

static void amxc_trim_llist ( amxc_llist_t *const  list )

static

Definition at line 93 of file amxc_string_split.c.

                                                      {
    amxc_llist_for_each_reverse(it, list) {
        amxc_string_t* part = amxc_string_from_llist_it(it);
        if(amxc_string_text_length(part) != 1) {
            break;
        }
        if(isspace(part->buffer[0]) == 0) {
            break;
        }
        amxc_string_delete(&part);
    }
 
    while(true) {
        amxc_llist_it_t* first = amxc_llist_get_first(list);
        amxc_llist_it_t* last = amxc_llist_get_last(list);
 
        if((first != NULL) && (last != NULL)) {
            amxc_string_t* fpart = amxc_string_from_llist_it(first);
            amxc_string_t* lpart = amxc_string_from_llist_it(last);
 
            if((amxc_string_text_length(fpart) == 1) &&
               ( amxc_string_text_length(lpart) == 1)) {
                if((fpart->buffer[0] == '[') &&
                   ( lpart->buffer[0] == ']')) {
                    amxc_string_delete(&fpart);
                    amxc_string_delete(&lpart);
                    continue;
                }
            }
        }
        break;
    }
}