How to split a string with a delimiter larger than one single char?

Supposing I have this:

"foo bar 1 and foo bar 2"

How can I split it into:

foo bar 1
foo bar 2

?

I tried strtok() and strsep() but neither worked. They don't recognize "and" as delimiter, the开发者_运维问答y recognize "a", "n" and "d" as delimiters.

Any function to help me with this or I'll have to split by the blank space and do some string manipulation?

---

You can use strstr() to find the first " and ", and "tokenize" the string yourself by just skipping forward so many characters, and doing it again.

---

The primary problem with splitting strings in C is that it inevitably results in some dynamic memory management, and that tends to be avoided by the standard library whenever possible. That is why none of the standard C functions deal with dynamic memory allocation, only malloc/calloc/realloc do that.

But it is not too difficult to do this oneself. Let me walk you through it.

We need to return a number of strings, and the easiest way to do that is to return an array of pointers to strings, which is terminated by a NULL item. Except for the final NULL, each element in the array points to a dynamically allocated string.

First we need a couple of helper functions to deal with such arrays. The easiest one is one that computes the number of strings (elements before the final NULL):

/* Return length of a NULL-delimited array of strings. */
size_t str_array_len(char **array)
{
    size_t len;

    for (len = 0; array[len] != NULL; ++len)
        continue;
    return len;
}

Another easy one is the function for freeing the array:

/* Free a dynamic array of dynamic strings. */
void str_array_free(char **array)
{
    if (array == NULL)
        return;
    for (size_t i = 0; array[i] != NULL; ++i)
        free(array[i]);
    free(array);
}

Somewhat more complicated is the function that adds a copy of a string to the array. It needs to handle a number of special cases, such as when the array does not exist yet (the whole array is NULL). Also, it needs to handle strings that are not terminated with '\0' so that it is easier for our actual splitting function to just use parts of the input string when appending.

/* Append an item to a dynamically allocated array of strings. On failure,
   return NULL, in which case the original array is intact. The item
   string is dynamically copied. If the array is NULL, allocate a new
   array. Otherwise, extend the array. Make sure the array is always
   NULL-terminated. Input string might not be '\0'-terminated. */
char **str_array_append(char **array, size_t nitems, const char *item, 
                        size_t itemlen)
{
    /* Make a dynamic copy of the item. */
    char *copy;
    if (item == NULL)
        copy = NULL;
    else {
        copy = malloc(itemlen + 1);
        if (copy == NULL)
            return NULL;
        memcpy(copy, item, itemlen);
        copy[itemlen] = '\0';
    }

    /* Extend array with one element. Except extend it by two elements, 
       in case it did not yet exist. This might mean it is a teeny bit
       too big, but we don't care. */
    array = realloc(array, (nitems + 2) * sizeof(array[0]));
    if (array == NULL) {
        free(copy);
        return NULL;
    }

    /* Add copy of item to array, and return it. */
    array[nitems] = copy;
    array[nitems+1] = NULL;
    return array;
}

That's a moutful. For really good style, it would have been better to split off the making of a dynamic copy if the input item to its own function, but I will leave that as an excercise to the reader.

Finally, we have the actual splitting function. It, also, needs to handle some special cases:

• The input string might start or end with the separator.
• There might be separators right next to each other.
• The input string might not include the separator at all.

I have opted to add an empty string to the result if the separator is right next to the start or end of the input string, or right next to another separator. If you need something else, you need to tweak the code.

Other than the special cases, and some error handling, the splitting is now reasonably straightforward.

/* Split a string into substrings. Return dynamic array of dynamically
   allocated substrings, or NULL if there was an error. Caller is
   expected to free the memory, for example with str_array_free. */
char **str_split(const char *input, const char *sep)
{
    size_t nitems = 0;
    char **array = NULL;
    const char *start = input;
    char *next = strstr(start, sep);
    size_t seplen = strlen(sep);
    const char *item;
    size_t itemlen;

    for (;;) {
        next = strstr(start, sep);
        if (next == NULL) {
            /* Add the remaining string (or empty string, if input ends with
               separator. */
            char **new = str_array_append(array, nitems, start, strlen(start));
            if (new == NULL) {
                str_array_free(array);
                return NULL;
            }
            array = new;
            ++nitems;
            break;
        } else if (next == input) {
            /* Input starts with separator. */
            item = "";
            itemlen = 0;
        } else {
            item = start;
            itemlen = next - item;
        }
        char **new = str_array_append(array, nitems, item, itemlen);
        if (new == NULL) {
            str_array_free(array);
            return NULL;
        }
        array = new;
        ++nitems;
        start = next + seplen;
    }

    if (nitems == 0) {
        /* Input does not contain separator at all. */
        assert(array == NULL);
        array = str_array_append(array, nitems, input, strlen(input));
    }

    return array;
}

Here is the whole program in one piece. It also includes a main program to run some test cases.

#include <assert.h>
#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>


/* Append an item to a dynamically allocated array of strings. On failure,
   return NULL, in which case the original array is intact. The item
   string is dynamically copied. If the array is NULL, allocate a new
   array. Otherwise, extend the array. Make sure the array is always
   NULL-terminated. Input string might not be '\0'-terminated. */
char **str_array_append(char **array, size_t nitems, const char *item, 
                        size_t itemlen)
{
    /* Make a dynamic copy of the item. */
    char *copy;
    if (item == NULL)
        copy = NULL;
    else {
        copy = malloc(itemlen + 1);
        if (copy == NULL)
            return NULL;
        memcpy(copy, item, itemlen);
        copy[itemlen] = '\0';
    }

    /* Extend array with one element. Except extend it by two elements, 
       in case it did not yet exist. This might mean it is a teeny bit
       too big, but we don't care. */
    array = realloc(array, (nitems + 2) * sizeof(array[0]));
    if (array == NULL) {
        free(copy);
        return NULL;
    }

    /* Add copy of item to array, and return it. */
    array[nitems] = copy;
    array[nitems+1] = NULL;
    return array;
}


/* Free a dynamic array of dynamic strings. */
void str_array_free(char **array)
{
    if (array == NULL)
        return;
    for (size_t i = 0; array[i] != NULL; ++i)
        free(array[i]);
    free(array);
}


/* Split a string into substrings. Return dynamic array of dynamically
   allocated substrings, or NULL if there was an error. Caller is
   expected to free the memory, for example with str_array_free. */
char **str_split(const char *input, const char *sep)
{
    size_t nitems = 0;
    char **array = NULL;
    const char *start = input;
    char *next = strstr(start, sep);
    size_t seplen = strlen(sep);
    const char *item;
    size_t itemlen;

    for (;;) {
        next = strstr(start, sep);
        if (next == NULL) {
            /* Add the remaining string (or empty string, if input ends with
               separator. */
            char **new = str_array_append(array, nitems, start, strlen(start));
            if (new == NULL) {
                str_array_free(array);
                return NULL;
            }
            array = new;
            ++nitems;
            break;
        } else if (next == input) {
            /* Input starts with separator. */
            item = "";
            itemlen = 0;
        } else {
            item = start;
            itemlen = next - item;
        }
        char **new = str_array_append(array, nitems, item, itemlen);
        if (new == NULL) {
            str_array_free(array);
            return NULL;
        }
        array = new;
        ++nitems;
        start = next + seplen;
    }

    if (nitems == 0) {
        /* Input does not contain separator at all. */
        assert(array == NULL);
        array = str_array_append(array, nitems, input, strlen(input));
    }

    return array;
}


/* Return length of a NULL-delimited array of strings. */
size_t str_array_len(char **array)
{
    size_t len;

    for (len = 0; array[len] != NULL; ++len)
        continue;
    return len;
}


#define MAX_OUTPUT 20


int main(void)
{
    struct {
        const char *input;
        const char *sep;
        char *output[MAX_OUTPUT];
    } tab[] = {
        /* Input is empty string. Output should be a list with an empty 
           string. */
        {
            "",
            "and",
            {
                "",
                NULL,
            },
        },
        /* Input is exactly the separator. Output should be two empty 
           strings. */
        {
            "and",
            "and",
            {
                "",
                "",
                NULL,
            },
        },
        /* Input is non-empty, but does not have separator. Output should
           be the same string. */
        {
            "foo",
            "and",
            {
                "foo",
                NULL,
            },
        },
        /* Input is non-empty, and does have separator. */
        {
            "foo bar 1 and foo bar 2",
            " and ",
            {
                "foo bar 1",
                "foo bar 2",
                NULL,
            },
        },
    };
    const int tab_len = sizeof(tab) / sizeof(tab[0]);
    bool errors;

    errors = false;

    for (int i = 0; i < tab_len; ++i) {
        printf("test %d\n", i);

        char **output = str_split(tab[i].input, tab[i].sep);
        if (output == NULL) {
            fprintf(stderr, "output is NULL\n");
            errors = true;
            break;
        }
        size_t num_output = str_array_len(output);
        printf("num_output %lu\n", (unsigned long) num_output);

        size_t num_correct = str_array_len(tab[i].output);
        if (num_output != num_correct) {
            fprintf(stderr, "wrong number of outputs (%lu, not %lu)\n",
                    (unsigned long) num_output, (unsigned long) num_correct);
            errors = true;
        } else {
            for (size_t j = 0; j < num_output; ++j) {
                if (strcmp(tab[i].output[j], output[j]) != 0) {
                    fprintf(stderr, "output[%lu] is '%s' not '%s'\n",
                            (unsigned long) j, output[j], tab[i].output[j]);
                    errors = true;
                    break;
                }
            }
        }

        str_array_free(output);
        printf("\n");
    }

    if (errors)
        return EXIT_FAILURE;   
    return 0;
}

---

Here's a nice short example I just wrote up showing how to use strstr to split a string on a given string:

#include <string.h>
#include <stdio.h>

void split(char *phrase, char *delimiter)
{
    char *loc = strstr(phrase, delimiter);
    if (loc == NULL)
    {
        printf("Could not find delimiter\n");
    }
    else
    {
        char buf[256]; /* malloc would be much more robust here */
        int length = strlen(delimiter);
        strncpy(buf, phrase, loc - phrase);
        printf("Before delimiter: '%s'\n", buf);
        printf("After delimiter: '%s'\n", loc+length);
    }
}

int main()
{
    split("foo bar 1 and foo bar 2", "and");
    printf("-----\n");
    split("foo bar 1 and foo bar 2", "quux");
    return 0;
}

Output:

Before delimiter: 'foo bar 1 '
After delimiter: ' foo bar 2'
-----
Could not find delimiter

Of course, I haven't fully tested it, and it's probably vulnerable to most of the standard buffer overflow issues related to string lengths; but it's a demonstrable example at the very least.

---

If you know the type of delimiter example comma or semicolon you can try with this:

#include<stdio.h>
#include<conio.h>
int main()
{
  int i=0,temp=0,temp1=0, temp2=0;
  char buff[12]="123;456;789";
   for(i=0;buff[i]!=';',i++)
   {
     temp=temp*10+(buff[i]-48);
   }
   for(i=0;buff[i]!=';',i++)
   {
     temp1=temp1*10+(buff[i]-48);
   }
   for(i=0;buff[i],i++)
   {
     temp2=temp2*10+(buff[i]-48);
   }
    printf("temp=%d temp1=%d temp2=%d",temp,temp1,temp2);
    getch();
  return 0;
}

Output:

temp=123 temp1=456 temp2=789