For an assignment, part of what I have to do involves the use of malloc and realloc. I first create a 2D array of chars, the dimensions being the number of lines and the number of characters. I then use malloc to allocate enough memory to store input from some file. Using fgets I read one line in at a time, and store it in the array. This part works fine (or so I think). The problem comes in when I try to reallocate memory for more lines if need be. The program flow is supposed to be like this:
Create a character array of 50 lines, with 80 characters per line (working)
Use fgets to read one line at a time and save it to the array (working)
When 50 lines have been read, reallocate the array to allow for 100 lines (not working)
Keep reallocating as need be (not working)
This is what I have so far (the core of it at least, I omitted irrelevant code):
#define NUMBER_OF_LINES 50
#define CHARACTERS_PER_LINE 80
FILE *inputFile = fopen("some.text", "r");
char **lines;
lines = malloc(NUMBER_OF_LINES * sizeof(*lines));
int i;
for (i = 0; i < NUMBER_OF_LINES; i++)
*(lines+i) = malloc(CHARACTERS_PER_LINE * sizeof(char));
int linesRemaining = NUMBER_OF_LINES;
int reallocCount = 1;
i = 0;
while (!feof(inputFile)) {
if (!linesRemaining) {
reallocCount++;
lines = realloc(lines, (NUM_OF_LINES * reallocCount) * sizeof(*lines));
linesRemaining = NUM_OF_LINES;
}
fgets(*(lines+i), CHARS_PER_LINE, inputFile);
i++;
linesRemaining--;
}
My gut tells me the problem is with the realloc, so I'll explain what I think it's doing.
realloc(lines, (NUM_OF_LINES * reallocCount) * sizeof(*lines));
The first argument, lines, is the pointer I would like to reallocate a certain amount of memory. NUM_OF_LINES is the amount I would like to increase the size by. I multiply this by reallocLinesCount, which is a counter that keeps track of how many sets of 50 lines I ought to have. The sizeof(*lines) part is the size of a pointer to a char.
Thank you for reading and any help is greatly appreciated :)
EDIT: thank you all for the responses; I do not have time right now to read all of the answers right now, but all of your answers will be more thoroughly read and understood once this imminent deadline has passed :D
My motto is: "say what you mean". In your case, you MEAN to enlarge your array when it's not big enough to hold your data.
FILE *in; // you fill this in
int nlines=50; // initial value
char **buffer=malloc(nlines * sizeof *buffer);
int i=0;
for(int i=0; !feof(in); ++i)
{
if(i>=nlines)
buffer=realloc(buffer, (nlines+=50)*sizeof *buffer);
buffer[i]=malloc(80);
fgets(buffer[i], 80, in);
}
realloc() will often find out that there is not enough available room to expand the existing array in-place; in that case, it will create an entirely new array of the specified size, copy the contents of the old array to the new one, deallocate the old array, and return a pointer to the new one. So you should write
char **oldLines = lines;
lines = realloc(...);
(the purpose of oldLines is to keep the original pointer in case realloc() runs out of memory and returns NULL, as per #Brian L's tip).
This is how you should realloc:
char **new_lines = realloc(lines, (NUM_OF_LINES * ++reallocLinesCount) * sizeof(*lines));
if (new_lines)
{
lines = new_lines;
}
else
{
// Memory allocation fails. Do some error handling.
}
Read realloc reference for details.
EDIT
You need more allocation for each new lines.
You are allocating more pointers to lines but not the lines themselves. It is in your code at the beginning:
for (i = 0; i < NUMBER_OF_LINES; i++)
*(lines+i) = malloc(CHARACTERS_PER_LINE * sizeof(char));
So after you allocated your number of lines for each line you allocate the space for the line itself. You forgot to do this for the new lines when you reallocate.
Let's first see how realloc() works. It returns a pointer to new
memory on success, and NULL on failure. On failure, it doesn't
touch the old memory, and on success, it free()'s it, after copying
your data to the new place.
So, the way to use realloc() safely is:
/* allocate memory using malloc() */
ptr = malloc(N * sizeof *ptr);
/* make sure malloc succeeded */
...
/* call realloc() */
new_ptr = realloc(ptr, M * sizeof *new_ptr);
/* see if it succeeded */
if (new_ptr) {
/* okay, we can set ptr */
ptr = new_ptr;
} else {
/* realloc failed, old pointer still valid */
}
So, the first thing is that you are using realloc() incorrectly.
You should never say x = realloc(x, ...);, because if realloc()
fails, you assign x to NULL, and the old memory is lost. This is
a memory leak.
Now, on to your problem. Let's say you have successfully read
NUMBER_OF_LINES lines. Now you want to make room for an additional
NUMBER_OF_LINES lines. You would do:
char **new_lines = realloc(lines, NUMBER_OF_LINES*reallocCount*sizeof *new_lines);
if (new_lines) {
lines = new_lines;
} else {
fprintf(stderr, "realloc failed!\n");
return;
}
/* Now, lines[NUMBER_OF_LINES] to lines[2*NUMBER_OF_LINES-1] are
* available to point someplace useful. They don't point anywhere
* useful yet. We have to allocate memory for them, just like earlier */
start = NUMBER_OF_LINES*reallocCount;
for (i=0; i < NUMBER_OF_LINES; ++i) {
/* You weren't allocating memory here, and were writing to
* lines[0] through lines[NUMBER_OF_LINES-1], which is not what
* you want. */
lines[start+i] = malloc(CHARS_PER_LINE * sizeof *lines[start+i]);
/* check the result of malloc here */
}
fgets(lines[start+i], CHARS_PER_LINE, inputFile);
One final note: it's almost always wrong to use while (!feof(fp))
to read lines from a file.
Related
i am new to coding and am having a problem with the following.
I am required to read from a text file, each row will contain:
command arg1 arg2 arg3...
command arg1 arg2
command
command arg1 arg2 ... arg9
etc
What i am trying to do is read this entire file into a 2D string array called array using malloc. This way if i were to do:
array[0][0] i would access command arg1 arg2 arg3
array[1][0] i would access command arg1 arg2
and so on.
I also know there is a max of 100 rows and 256 characters per line. Below is how i attempted to declare my malloc however when trying to allocate strings to the 2d array, it only allocated single characters.
I dont quite understand how to do this, detailed explanation would be greatly appreciated
int row = 100;
int col = 256;
int **array;
array = (int**)malloc(row*sizeof(array));
if(!array){
perror("Error occured allocating memory");
exit(-1);
}
for(int i = 0; i<row;i++){
array[i] = (int*)malloc(col*sizeof(array));
}
If I got it right, you need to set up a two dimensional array of char * instead of int.
That is, you address the correct row by dereferencing once (array[the_ith_row]), and then address the correct element(command, arg1, arg2, ...) by another dereference (array[the_ith_row][the_jth_col]).
Notice: strings like "arg1" and "command" are treated as "array of chars" therefore you need to store a char * in order to access them. int could only store one char(with some extra space consumption), therefore won't work here.
So, the correct one should look like:
#include <string.h>
int row = 100;
int col = 256;
char ***array;
array = (char ***)malloc(row * sizeof(char **));
if (!array) {
perror("Error occured allocating memory");
exit(-1);
}
for (int i = 0; i < row; i++) {
array[i] = (char **)malloc(col * sizeof(char *));
}
// Do some example assignments
for (int j = 0; j < col; j++) {
array[i][j] = strcpy((char *)malloc(100), "test_string");
}
//therefore printf("%s", array[0][0]); will print test_string"
UPDATE: I missed some * here..
You are allocating using sizeof(array) which is not the correct unit of allocation that you want.
It looks like what you want are two different kinds of memory allocations or objects.
The first is an array of pointers to character strings since the file data is a series of character strings.
The second kind of memory allocation is for the memory to hold the actual character string.
The first kind of memory allocation, to an array of pointers to character strings would be:
char **pArray = malloc (100 * sizeof(char *)); // allocate the array of character string pointers
The second kind of memory allocation, to a character string which is an array of characters would be:
char *pString = malloc ((256 + 1) * sizeof(char)); // allocate a character array for up to 256 characters
The 256 + 1 is needed in order to allocate space for 256 characters plus one more for the end of string character.
So to allocate the entire needed space, you would do the following:
int iIndex;
int nMax = 100;
char **pArray = malloc (nMax, sizeof(char *)); // allocate array of rows
for (iIndex = 0; iIndex < nMax; iIndex++) {
pArray[iIndex] = malloc ((256 + 1) * sizeof (char)); // allocate a row
}
// now use the pArray to read in the lines of text from the file.
// for the first line, pArray[0], second pArray[1], etc.
Using realloc()
A question posed is using the realloc() function to adjust the size of the allocated memory.
For the second kind of memory, memory for the actual character string, the main thing is to use realloc() as normal to expand or shrink the amount of memory. However if memory is reduced, you need to consider if the text string was truncated and a new end of string terminator is provided to ensure the text string is properly terminated with and end of string indicator.
// modify size of a text string memory area for text string in pArray[i]
// memory area. use a temporary and test that realloc() worked before
// changing the pointer value in pArray[] element.
char *p = realloc (pArray[i], (nSize + 1) * sizeof (char));
if (p != NULL) {
pArray[i] = p; // valid realloc() so replace our pointer.
pArray[i][nSize] = 0; // ensure zero terminator for string
}
If you ensure that when the memory area for pArray] is set to NULL after allocating the array, you can just use the realloc() function as above without first using malloc() since if the pointer in the argument to realloc() is NULL then realloc() will just do a malloc() for the memory.
For the first kind of memory, you will need to consider freeing any memory whose pointers may be destroyed when the allocated array is shortened. This means that you will need to do a bit more management and keeping management data about the allocated memory area. If you can guarantee that you will only be increasing the size of the array and never shortening it then you don't need to do any management and you can just use the same approach as provided for the second kind of memory above.
However if the memory allocated for the first kind of memory will need to be smaller as well as larger, you need to have some idea as to the size of the memory area allocated. Probably the easiest would be to have a simple struct that would provide both a pointer to the array allocated as well as the max count of items the array can hold.
typedef struct {
size_t nCount;
char **pArray;
} ArrayObj;
Warning: the following code has not been tested or even compiled. Also note that this only works for if the memory allocation will be increased.
Then you would wrap the realloc() function within a management function. This version of the function only handles if realloc() is always to expand the array. If making it smaller you will need to handle that case in this function.
ArrayObj ArrayObjRealloc (ArrayObj obj, size_t nNewCount)
{
// make the management a bit easier by just adding one to the count
// to determine how much memory to allocate.
char **pNew = realloc (obj.pArray, (nNewCount + 1) * sizeof (char *));
if (pNew != NULL) {
size_t ix;
// realloc() worked and provided a valid pointer to the new area.
// update the management information we are going to return.
// set the new space to NULL to have it in an initial and known state.
// initializing the new space to NULL will allow for knowing which array
// elements have a valid pointer and which don't.
obj.pArray = pNew;
for (ix = nNewCount; ix >= obj.nCount; ix--) {
obj.pArray[ix] = NULL;
}
obj.nCount = nNewCount;
}
return obj;
}
and use this function something like
AnyObj obj = {0, NULL};
// allocate for the first time
obj = ArrayObjRealloc (obj, 100);
// do stuff with the array allocated
strcpy (obj.pArray[i], "some text");
// make the array larger
obj = ArrayObjRealloc (obj, 150);
Example: my array contains integers {1,2,3,4}
I want to resize my array in order to contain {2,3,4} and when I do array[0] it should give me 2.
When you shrink an array with realloc(), you can remove space at the end of the array, but not at the beginning of the array. Therefore, to achieve the result you want, you'd have to copy the portion of the array you want to keep to its new position and then resize the array:
/* setup */
int *data = malloc(4 * sizeof(int)); // Error check omitted
for (int i = 0; i < 4; i++)
data[i] = i + 1;
/* setup complete */
/* Move elements before shrinking array */
memmove(&data[0], &data[1], 3 * sizeof(int)); // Safe move
int *new_data = realloc(data, 3 * sizeof(int));
if (new_data != 0)
data = new_data;
I'm not sure whether a 'shrinking' realloc() does ever return NULL (but the standard doesn't say it can't), so the code takes no chances. Many people would write data = realloc(data, 3 * sizeof(int)); and run the risk, and they'd get away with it almost all the time. Note that a 'growing' realloc() really can return NULL and you can then leak memory if you use the old_ptr = realloc(old_ptr, new_size) idiom.
Usually, a 'shrinking' realloc() will return the original pointer (unless you shrink it to zero size, when it may return NULL). A 'growing' realloc() can often change where the data is stored.
I have an array, say, text, that contains strings read in by another function. The length of the strings is unknown and the amount of them is unknown as well. How should I try to allocate memory to an array of strings (and not to the strings themselves, which already exist as separate arrays)?
What I have set up right now seems to read the strings just fine, and seems to do the post-processing I want done correctly (I tried this with a static array). However, when I try to printf the elements of text, I get a segmentation fault. To be more precise, I get a segmentation fault when I try to print out specific elements of text, such as text[3] or text[5]. I assume this means that I'm allocating memory to text incorrectly and all the strings read are not saved to text correctly?
So far I've tried different approaches, such as allocating a set amount of some size_t=k , k*sizeof(char) at first, and then reallocating more memory (with realloc k*sizeof(char)) if cnt == (k-2), where cnt is the index of **text.
I tried to search for this, but the only similar problem I found was with a set amount of strings of unknown length.
I'd like to figure out as much as I can on my own, and didn't post the actual code because of that. However, if none of this makes any sense, I'll post it.
EDIT: Here's the code
int main(void){
char **text;
size_t k=100;
size_t cnt=1;
int ch;
size_t lng;
text=malloc(k*sizeof(char));
printf("Input:\n");
while(1) {
ch = getchar();
if (ch == EOF) {
text[cnt++]='\0';
break;
}
if (cnt == k - 2) {
k *= 2;
text = realloc(text, (k * sizeof(char))); /* I guess at least this is incorrect?*/
}
text[cnt]=readInput(ch); /* read(ch) just reads the line*/
lng=strlen(text[cnt]);
printf("%d,%d\n",lng,cnt);
cnt++;
}
text=realloc(text,cnt*sizeof(char));
print(text); /*prints all the lines*/
return 0;
}
The short answer is you can't directly allocate the memory unless you know how much to allocate.
However, there are various ways of determining how much you need to allocate.
There are two aspects to this. One is knowing how many strings you need to handle. There must be some defined way of knowing; either you're given a count, or there some specific pointer value (usually NULL) that tells you when you've reached the end.
To allocate the array of pointers to pointers, it is probably simplest to count the number of necessary pointers, and then allocate the space. Assuming a null terminated list:
size_t i;
for (i = 0; list[i] != NULL; i++)
;
char **space = malloc(i * sizeof(*space));
...error check allocation...
For each string, you can use strdup(); you assume that the strings are well-formed and hence null terminated. Or you can write your own analogue of strdup().
for (i = 0; list[i] != NULL; i++)
{
space[i] = strdup(list[i]);
...error check allocation...
}
An alternative approach scans the list of pointers once, but uses malloc() and realloc() multiple times. This is probably slower overall.
If you can't reliably tell when the list of strings ends or when the strings themselves end, you are hosed. Completely and utterly hosed.
C don't have strings. It just has pointers to (conventionally null-terminated) sequence of characters, and call them strings.
So just allocate first an array of pointers:
size_t nbelem= 10; /// number of elements
char **arr = calloc(nbelem, sizeof(char*));
You really want calloc because you really want that array to be cleared, so each pointer there is NULL. Of course, you test that calloc succeeded:
if (!arr) perror("calloc failed"), exit(EXIT_FAILURE);
At last, you fill some of the elements of the array:
arr[0] = "hello";
arr[1] = strdup("world");
(Don't forget to free the result of strdup and the result of calloc).
You could grow your array with realloc (but I don't advise doing that, because when realloc fails you could have lost your data). You could simply grow it by allocating a bigger copy, copy it inside, and redefine the pointer, e.g.
{ size_t newnbelem = 3*nbelem/2+10;
char**oldarr = arr;
char**newarr = calloc(newnbelem, sizeof(char*));
if (!newarr) perror("bigger calloc"), exit(EXIT_FAILURE);
memcpy (newarr, oldarr, sizeof(char*)*nbelem);
free (oldarr);
arr = newarr;
}
Don't forget to compile with gcc -Wall -g on Linux (improve your code till no warnings are given), and learn how to use the gdb debugger and the valgrind memory leak detector.
In c you can not allocate an array of string directly. You should stick with pointer to char array to use it as array of string. So use
char* strarr[length];
And to mentain the array of characters
You may take the approach somewhat like this:
Allocate a block of memory through a call to malloc()
Keep track of the size of input
When ever you need a increament in buffer size call realloc(ptr,size)
I need an array of strings in which the length of the array is not known at compile time. what I've done is:
char **words;
words = malloc(capacity * sizeof(char *));
char nextword[MAX_LEN + 1];
while (true) {
if (num_entries == capacity) {
capacity += 5;
realloc(words, (sizeof(char *) * capacity));
}
printf("\nEnter a word: ");
fgets (nextword, MAX_LEN, stdin);
remove_newline(nextword);
if (strlen(nextword) == 0) break;
words[num_entries] = malloc(strlen(nextword + 1));
if (words[num_entries] == NULL) {
printf("\nout of space\n");
break;
}
strcpy(words[num_entries], nextword);
num_entries++;
this seems to work to expand the size once, except that the first element after the expansion has become NULL for some reason. The second time realloc is executed I get an error:
"invalid next size".
realloc is not guaranteed to give you back the same chunk of memory, because the block that was originally allocated from the heap may not have enough space to accommodate your new requested size. In this case you will get back a new block of memory, with your old data copied to it.
You need to capture the returned value on each loop and use it to check for the data you expect, and also check for it being 0 (if the realloc cannot be done).
words = realloc(words,..)
is an antipattern - avoid this since your old memory can get lost if the realloc fails.
Your code is almost there, just need a few modifications. An important thing to remember is that realloc does not modify the value that you pass to it, and it is not required to return the pointer to the same chunk of memory that you passed to it. Here is a working example of using realloc. It is rather straightforward, so you should be able to fix your code by simply following the example.
char **more_words = realloc(words, capacity);
if (more_words) {
words = more_words;
} else {
// Do something about realloc failure
}
I've dynamically allocated a structure, conceptually very similar to a matrix, to hold a set of strings. I've encountered a problem while trying to free the memory. My code looks like this:
# include <stdio.h>
# include <string.h>
# include <malloc.h>
# define SIZE 2
typedef struct fork{
char** dataPointersArray;
char* dataArray;
}fork;
int main(int argc, char* argv[]){
fork forkDS;
int i;
char* dataArrayPtr;
unsigned char data[255] = "some data"; /* this is actually a function's output */
int PtrIndex;
/* allocate memory for the arrays */
forkDS.dataPointersArray = (char**) calloc(SIZE ,sizeof(char*));
if(forkDS.dataPointersArray == NULL){
printf("couldn't allocate memory \n");
}
forkDS.dataArray = (char*) calloc(SIZE, 255);
if( forkDS.dataArray == NULL){
free(forkDS.dataPointersArray);
printf("couldn't allocate memory \n");
}
dataArrayPtr = forkDS.dataArray;
for(i = 0; i < SIZE; i++){
/* update the dataPointers Array */
forkDS.dataPointersArray[i] = dataArrayPtr;
/* copy data into data array */
memcpy(dataArrayPtr,data,20);
dataArrayPtr[255] = '\0';
/* update the pointer of the data array */
dataArrayPtr = dataArrayPtr + 256;
}
for (PtrIndex = 0; PtrIndex < 2; PtrIndex++) {
if (*(forkDS.dataPointersArray + PtrIndex) != NULL) {
*(forkDS.dataPointersArray + PtrIndex) = NULL;
}
}
/* DEBUG comment - this 2 lines works */
free(forkDS.dataArray);
forkDS.dataArray = NULL;
/* DEBUG comment - the next line fails */
free(forkDS.dataPointersArray);
forkDS.dataPointersArray = NULL;
return 0;
}
So the structure actually contains 2 arrays, one of pointers to strings, and the other one contains the strings aligned one after the other, separated by a terminating \0.
The code works fine, and the for loop in the end works as well. The first call to free also works. The problem is that the last call to free fails.
Although trying to search all possible data on the issue, all the examples I've found regarded the case where the second array, which holds the strings, is allocated step by step in a for loop, and freed afterwards in a for loop as well.
I wanted to avoid using dynamic allocation in a loop, and therefore my code looks different.
Does anyone know what the problem is?
======================================================================================
Thanks a lot to all of you who answered me. Eventually, the bug was solved. The problem was that the dataPointersArray was filled with more than SIZE elements in some other piece of code, which seemed innocent at first, and actually caused the free call to fail.
Thanks again for the comments!
Shachar
You are allocating SIZE*255 bytes, but using SIZE * 256 bytes:
forkDS.dataArray = (char*) calloc(SIZE, 255);
dataArrayPtr = forkDS.dataArray;
//SIZE TIMES loop:
dataArrayPtr = dataArrayPtr + 256;
So when you are NULLing the pointers, you probably overwrite control data placed past the end of the array by malloc that free is looking for.
You allocated the space for an array consisting of SIZE lines with 255 characters each. The highest index on each line therefore is 254 = 255 - 1. As you write the \0 character, you write it at the beginning of the next line. After the last iteration, you would be off by SIZE bytes.
Just another detail: If any of the memory allocations failed, the program would only print its error messages, but it won't stop causing a SEGFAULT later.