I have this simple code to read the lines of a file and store them in a struct:
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
struct filedata {
char **items;
int lines;
};
struct filedata *read_file(char *filename) {
FILE* file = fopen(filename, "r");
if (file == NULL) {
printf("Can't read %s \n", filename);
exit(1);
}
char rbuff;
int nlines = 0; // amount of lines
int chr = 0; // character count
int maxlen = 0; // max line length (to create optimal buffer)
int minlen = 2; // min line length (ignores empty lines with just \n, etc)
while ((rbuff = fgetc(file) - 0) != EOF) {
if (rbuff == '\n') {
if (chr > maxlen) {
maxlen = chr + 1;
}
if (chr > minlen) {
nlines++;
}
chr = 0;
}
else {
chr++;
}
}
struct filedata *rdata = malloc(sizeof(struct filedata));
rdata->lines = nlines;
printf("lines: %d\nmax string len: %d\n\n", nlines, maxlen);
rewind(file);
char *list[nlines];
int buffsize = maxlen * sizeof(char);
char buff[buffsize];
int i = 0;
while (fgets(buff, buffsize, file)) {
if (strlen(buff) > minlen) {
list[i] = malloc(strlen(buff) * sizeof(char) + 1);
strcpy(list[i], buff);
i++;
}
}
rdata->items = (char **)list;
fclose(file);
int c = 0;
for (c; c < rdata->lines; c++) {
printf("line %d: %s\n", c + 1, rdata->items[c]);
}
printf("\n");
return rdata;
}
int main(void) {
char fname[] = "test.txt";
struct filedata *ptr = read_file(fname);
int c = 0;
for (c; c < ptr->lines; c++) {
printf("line %d: %s\n", c + 1, ptr->items[c]);
}
return 0;
}
This is the output when I run it:
lines: 2
max string len: 6
line 1: hello
line 2: world
line 1: hello
line 2: H��
For some reason when it reaches the second index in ptr->items, it prints gibberish output. But yet, if I throw some printf()'s in there to show the pointer addresses, they're exactly the same.
Valgrind also prints this when iterating over the char array the second time:
==3777== Invalid read of size 8
==3777== at 0x400AB3: main (test.c:81)
==3777== Address 0xfff000540 is on thread 1's stack
==3777== 240 bytes below stack pointer
But that really doesn't give me any clues in this case.
I'm using gcc 4.9.4 with glibc-2.24 if that matters.
list is an non-static local variable and using it after exiting its scope (returning from read_file in this case) will invoke undefined behavior because it will vanish on exiting its scope. Allocate it dynamically (typically on the heap) like
char **list = malloc(sizeof(char*) * nlines);
Adding code to check if malloc()s are successful will make your code better.
The variable list is local to read_file, but you store a pointer to list in rdata->items. When read_file returns, rdata->items is a dangling pointer, and accessing it is undefined behavior.
Related
I am reading a file that contains several lines of strings(max length 50 characters). To store those strings I created a char double-pointer using calloc. The way my code works is as it finds a line in the file it adds one new row (char *) and 50 columns (char) and then stores the value.
My understanding is that I can call this method and get this pointer with values in return. However, I was not getting the values so I check where I am losing it and I found that the memory is not persisting after while loop. I am able to print strings using print 1 statement but print 2 gives me null.
Please let me know what I am doing wrong here.
char **read_file(char *file)
{
FILE *fp = fopen(file, "r");
char line[50] = {0};
char **values = NULL;
int index = 0;
if (fp == NULL)
{
perror("Unable to open file!");
exit(1);
}
// read both sequence
while (fgets(line, 50, fp))
{
values = (char **)calloc(index + 1, sizeof(char *));
values[index] = (char *)calloc(50, sizeof(char));
values[index] = line;
printf("%s",values[index]); // print 1
index++;
}
fclose(fp);
printf("%s", values[0]); // print 2
return values;
}
line content is overwritten on each loop iteration (by fgets()).
values is overwritten (data loss) and leaks memory on each iteration index > 1.
value[index] is allocated memory on each iteration which leaks as you overwrite it with the address of line on the following line.
line is a local variable so you cannot return it to caller where it will be out of scope.
caller has no way to tell how many entries values contain.
Here is a working implementation with a few changes. On error it closes the file and frees up memory allocated and return NULL instead of exiting. Moved printf() to caller:
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#define BUF_LEN 50
char **read_file(char *file) {
FILE *fp = fopen(file, "r");
if(!fp) {
perror("Unable to open file!");
return NULL;
}
char **values = NULL;
char line[BUF_LEN];
unsigned index;
for(index = 0;; index++) {
char **values2 = realloc(values, (index + 1) * sizeof(char *));
if(!values2) {
perror("realloc failed");
goto err;
}
values = values2;
if(!fgets(line, BUF_LEN, fp)) break;
values[index] = strdup(line);
}
fclose(fp);
values[index] = NULL;
return values;
err:
fclose(fp);
for(unsigned i = 0; i < index; i++) {
free(values[i]);
}
free(values);
return NULL;
}
int main() {
char **values = read_file("test.txt");
for(unsigned i = 0; values[i]; i++) {
printf("%s", values[i]);
free(values[i]);
}
free(values);
return 0;
}
fgets() returns line ending in '\n' or at most BUF_LEN - 1 of data. This means a given value[i] may or may not be ending with a \n. You may want this behavior, or you want value[i] to be consistent and not contain any trailing \n irregardless of the input.
strdup() is _POSIX_C_SOURCE >= 200809L and not standard c,
so if you build with --std=c11 the symbol would not be defined.
I want to change my input.txt file to an integer array.
But sadly I keep missing one integer whenever new-line-character is met.
Following is my main()
int main(int args, char* argv[]) {
int *val;
char *STRING = readFile();
val = convert(STRING);
return 0;
}
Following is my file input function
char *readFile() {
int count;
FILE *fp;
fp = fopen("input.txt", "r");
if(fp==NULL) printf("File is NULL!n");
char* STRING;
char oneLine[255];
STRING = (char*)malloc(255);
assert(STRING!=NULL);
while(1){
fgets(oneLine, 255, fp);
count += strlen(oneLine);
STRING = (char*)realloc(STRING, count+1);
strcat(STRING, oneLine);
if(feof(fp)) break;
}
fclose(fp);
return STRING;
}
Following is my integer array function
int *convert(char *STRING){
int *intarr;
intarr = (int*)malloc(sizeof(int)*16);
int a=0;
char *ptr = strtok(STRING, " ");
while (ptr != NULL){
intarr[a] = atoi(ptr);
printf("number = %s\tindex = %d\n", ptr, a);
a++;
ptr = strtok(NULL, " ");
}
return intarr;
}
There are many issues.
This is a corrected version of your program, all comments are mine. Minimal error checking is done for brevity. intarr = malloc(sizeof(int) * 16); will be a problem if there are more than 16 numbers in the file, this should be handled somehow, for example by growing intarr with realloc, similar to what you're doing in readFile.
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <assert.h>
char *readFile() {
FILE *fp;
fp = fopen("input.txt", "r");
if (fp == NULL)
{
printf("File is NULL!n");
return NULL; // abort if file could not be opened
}
#define MAXLINELENGTH 255 // define a constant rather than hardcoding "255" at several places
char* STRING;
char oneLine[MAXLINELENGTH];
STRING = malloc(MAXLINELENGTH);
int count = MAXLINELENGTH; // count mus be initialized and better declare it here
assert(STRING != NULL);
STRING[0] = 0; // memory pointed by STRING must be initialized
while (fgets(oneLine, MAXLINELENGTH, fp) != NULL) // correct usage of fgets
{
count += strlen(oneLine);
STRING = realloc(STRING, count + 1);
strcat(STRING, oneLine);
}
fclose(fp);
return STRING;
}
int *convert(char *STRING, int *nbofvalues) { // nbofvalues for returning the number of values
int *intarr;
intarr = malloc(sizeof(int) * 16);
int a = 0;
char *ptr = strtok(STRING, " \n"); // strings may be separated by '\n', or ' '
*nbofvalues = 0;
while (ptr != NULL) {
intarr[a] = atoi(ptr);
printf("number = %s\tindex = %d\n", ptr, a);
a++;
ptr = strtok(NULL, " \n"); // strings are separated by '\n' or ' '
} // read the fgets documentation which
// terminates read strings by \n
*nbofvalues = a; // return number of values
return intarr;
}
int main(int args, char* argv[]) {
int *val;
char *STRING = readFile();
if (STRING == NULL)
{
printf("readFile() problem\n"); // abort if file could not be read
return 1;
}
int nbvalues;
val = convert(STRING, &nbvalues); // nbvalues contains the number of values
// print numbers
for (int i = 0; i < nbvalues; i++)
{
printf("%d: %d\n", i, val[i]);
}
free(val); // free memory
free(STRING); // free memory
return 0;
}
I'm not sure what your requirement is, but this can be simplified a lot because there is no need to read the file into memory and then convert the strings into number. You could convert the numbers on the fly as you read them. And as already mentioned in a comment, calling realloc for each line is inefficient. There is room for more improvements.
I am doing a project where I have to read in text from a file and then extract every word that is 4 characters long and allocate it into dynamic array.My approach is to create int function that will get number of 4 letter words and return that number , then create another function that will grab that number and create dynamic array consisting of that many elements. The problem with this approach is how to populate that array with words that meet the requirement.
int func1(FILE *pFile){
int counter = 0;
int words = 0;
char inputWords[length];
while(fscanf(pFile,"%s",inputWords) != EOF){
if(strlen(inputWords)==4){
#counting 4 letter words
counter++;
}
}
}
return counter;
}
int main(){
#creating pointer to a textFile
FILE *pFile = fopen("smallDictionary.txt","r");
int line = 0;
#sending pointer into a function
func1(pFile);
fclose(pFile);
return 0;
}
I would suggest reading lines of input with fgets(), and breaking each line into tokens with strtok(). As each token is found, the length can be checked, and if the token is four characters long it can be saved to an array using strdup().
In the code below, storage is allocated for pointers to char which will store the addresses of four-letter words. num_words holds the number of four-letter words found, and max_words holds the maximum number of words that can currently be stored. When a new word needs to be added, num_words is incremented, and if there is not enough storage, more space is allocated. Then strdup() is used to duplicate the token, and the address is assigned to the next pointer in words.
Note that strdup() is not in the C Standard Library, but that it is POSIX. The feature test macro in the first line of the program may be needed to enable this function. Also note that strdup() allocates memory for the duplicated string which must be freed by the caller.
#define _POSIX_C_SOURCE 200809L
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#define BUF_SZ 1000
#define ALLOC_INC 100
int main(void)
{
FILE *fp = fopen("filename.txt", "r");
if (fp == NULL) {
perror("Unable to open file");
exit(EXIT_FAILURE);
}
char buffer[BUF_SZ];
char **words = NULL;
size_t num_words = 0;
size_t max_words = 0;
char *token;
char *delims = " \t\r\n";
while (fgets(buffer, sizeof buffer, fp) != NULL) {
token = strtok(buffer, delims);
while (token != NULL) {
if (strlen(token) == 4) {
++num_words;
if (num_words > max_words) {
max_words += ALLOC_INC;
char **temp = realloc(words, sizeof *temp * max_words);
if (temp == NULL) {
perror("Unable to allocate memory");
exit(EXIT_FAILURE);
}
words = temp;
}
words[num_words-1] = strdup(token);
}
token = strtok(NULL, delims);
}
}
if (fclose(fp) != 0) {
perror("Unable to close file");
exit(EXIT_FAILURE);
}
for (size_t i = 0; i < num_words; i++) {
puts(words[i]);
}
/* Free allocated memory */
for (size_t i = 0; i < num_words; i++) {
free(words[i]);
}
free(words);
return 0;
}
Update
OP has mentioned that nonstandard functions are not permitted in solving this problem. Though strdup() is POSIX, and both common and standard in this sense, it is not always available. In such circumstances it is common to simply implement strdup(), as it is straightforward to do so. Here is the above code, modified so that now the function my_strdup() is used in place of strdup(). The code is unchanged, except that the feature test macro has been removed, the call to strdup() has been changed to my_strdup(), and of course now there is a function prototype and a definition for my_strdup():
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#define BUF_SZ 1000
#define ALLOC_INC 100
char * my_strdup(const char *);
int main(void)
{
FILE *fp = fopen("filename.txt", "r");
if (fp == NULL) {
perror("Unable to open file");
exit(EXIT_FAILURE);
}
char buffer[BUF_SZ];
char **words = NULL;
size_t num_words = 0;
size_t max_words = 0;
char *token;
char *delims = " \t\r\n";
while (fgets(buffer, sizeof buffer, fp) != NULL) {
token = strtok(buffer, delims);
while (token != NULL) {
if (strlen(token) == 4) {
++num_words;
if (num_words > max_words) {
max_words += ALLOC_INC;
char **temp = realloc(words, sizeof *temp * max_words);
if (temp == NULL) {
perror("Unable to allocate memory");
exit(EXIT_FAILURE);
}
words = temp;
}
words[num_words-1] = my_strdup(token);
}
token = strtok(NULL, delims);
}
}
if (fclose(fp) != 0) {
perror("Unable to close file");
exit(EXIT_FAILURE);
}
for (size_t i = 0; i < num_words; i++) {
puts(words[i]);
}
/* Free allocated memory */
for (size_t i = 0; i < num_words; i++) {
free(words[i]);
}
free(words);
return 0;
}
char * my_strdup(const char *str)
{
size_t sz = strlen(str) + 1;
char *dup = malloc(sizeof *dup * sz);
if (dup) {
strcpy(dup, str);
}
return dup;
}
Final Update
OP had not posted code in the question when the above solution was written. The posted code does not compile as is. In addition to missing #includes and various syntax errors (extra braces, incorrect comment syntax) there are a couple of more significant issues. In func1(), the length variable is used uninitialized. This should be large enough so that inputWords[] can hold any expected word. Also, width specifiers should be used with %s in scanf() format strings to avoid buffer overflow. And, OP code should be checking whether the file opened successfully. Finally, func1() returns a value, but the calling function does not even assign this value to a variable.
To complete the task, the value returned from func1() should be used to declare a 2d array to store the four-letter words. The file can be rewound, but this time as fscanf() retrieves words in a loop, if a word has length 4, strcpy() is used to copy the word into the array.
Here is a modified version of OP's code:
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#define MAX_WORD 100
int func1(FILE *pFile){
int counter = 0;
char inputWords[MAX_WORD];
while(fscanf(pFile,"%99s",inputWords) != EOF) {
if(strlen(inputWords) == 4) {
counter++;
}
}
return counter;
}
int main(void)
{
FILE *pFile = fopen("filename.txt","r");
if (pFile == NULL) {
perror("Unable to open file");
exit(EXIT_FAILURE);
}
char inputWords[MAX_WORD];
int num_4words = func1(pFile);
char words[num_4words][MAX_WORD];
int counter = 0;
rewind(pFile);
while(fscanf(pFile,"%99s",inputWords) != EOF) {
if(strlen(inputWords) == 4) {
strcpy(words[counter], inputWords);
counter++;
}
}
if (fclose(pFile) != 0) {
perror("Unable to close file");
}
for (int i = 0; i < num_4words; i++) {
puts(words[i]);
}
return 0;
}
The code I am working on reads in a dictionary of 45430 words and then prints to the file all the other words in the dictionary contained within each word. I am just working on getting the file MyDictionary txt file read into the char array word[45430][30] and then printing this to the words-in-words txt file. I run into a seg fault at 44946 word when I do so, but in the same while loop I am also printing to the console and all words print out properly. Why is it I am getting this seg fault for writing to the file? And why is there no seg fault writing to the console?
Code:
#include <stdio.h>
#include <stdlib.h>
#include <wchar.h>
#include <string.h>
//char ***alloc_array(int,int);
int main(void){
FILE *fr; //declare file read file pointer
FILE *fp; //declare file printed file pointer
//char array to read in up to 30 chars
char line[31];
long numwords=45430; //number of words in dictionary
int maxlength=31; // the longest string in dictionary (30 chars)
long i; //counts up to 45430
//allocate space for 45430 words at a max length of 30 chars (1 extra char for "\0")
char ***word = calloc(numwords, sizeof(char **));
for(i = 0; i != numwords; i++) {
word[i] = calloc(maxlength, sizeof(char *));
}
//Open MyDictionary.txt and determine if there is an error or not
fr = fopen ("MyDictionary.txt", "r"); // open the file for reading
if(fr==NULL){
printf("\nError! opening input file");
exit(1); //Program exits if file pointer returns NULL.
}
//Open words-within-words.txt and determine if there is an error or not
fp = fopen ("words-within-words.txt", "w"); // open the file for reading
if(fp==NULL){
printf("\nError! opening output file");
exit(1); //Program exits if file pointer returns NULL.
}
int j=0; //counts to 30 for max length
i=0;
while(fgets(line, 40, fr) != NULL){ //get a line, up to 40 chars from fr and put first . done if NULL
for(j=0;j<30;){
word[i][j]=&line[j];
j++;
}
j=0;
printf("\n%s",word[i][j]); //print out each word of dictionary to console on its own line
/*
if((i>4 && i<8)||(i>45428)){
fprintf(fp,"\nanalyze:word[i][0]=%s\tword[i][2]=%s\ti=%li",word[i][0],word[i][2],i+1);
}
*/
fprintf(fp,"%s",word[i][j]); //print out each word of dictionary to words-in-words on its own line
i++;
}
fclose(fr); //close the files prior to exiting
fclose(fp);
return 0;
} //main
char ***word = calloc(numwords, sizeof(char **));
for(i = 0; i != numwords; i++) {
word[i] = calloc(maxlength, sizeof(char *));
}
You've got one too many levels of indirection. You are storing a list of words. A word is a char *, so a list of words would be char **.
char **word = calloc(numwords, sizeof(char *));
for (i = 0; i != numwords; i++) {
word[i] = calloc(maxlength, sizeof(char));
}
This will then necessitate changes to the rest of your code. You can get rid of j entirely. This:
for(j=0;j<30;){
word[i][j]=&line[j];
j++;
}
Becomes:
strcpy(word[i], line);
And this:
j=0;
printf("\n%s",word[i][j]);
fprintf(fp,"%s",word[i][j]);
i++;
Becomes:
printf("%s\n", word[i]);
fprintf(fp, "%s\n", word[i]);
'word' should be an array of pointers, so the right type is char **, not char ***.
Each entry in the array is a pointer to a buffer of characters:
char **word = (char **)calloc(numwords, sizeof(char *));
if (!word)
// exit with error
for (i = 0; i != numwords; i++) {
word[i] = (char *)calloc(maxlength, sizeof(char)); // just allocate 31 bytes
if (!word[i])
// exit with error
}
Then a read from file can be done like this:
for (i = 0; fgets(line, 40, fr); i++) {
strncpy(word[i], line, maxlength);
printf("word %d: %s\n", i, word[i]);
}
To have one chunk of memory do allocate like this:
#include <stdio.h>
#include <stdlib.h>
int main(void)
{
int result = EXIT_SUCCESS;
size_t n = 45430;
size_t l = 30;
char (* words)[n][l + 1] = calloc(n, l + 1);
if (NULL == words)
{
result = EXIT_FAILURE;
perror("calloc() failed");
goto lblExit;
}
for (size_t i = 0; i < n; ++i)
{
strncpy((*words)[i], "test", l);
}
for (size_t i = 0; i < n; ++i)
{
printf("%zu: '%s'\n", i, (*words)[i]);
}
free(words);
lblExit:
return result;
}
I'm experimenting with C structs and I've come up with a invalid write of size 8 followed by invalid read of size 8 messages from valgrind.
My code is only looping through arguments (if argc > 1) and for each filename, it scans for a string and unsigned int indicating name and age(struct player).
This is all the code I've got so far:
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
struct player {
char name[20];
unsigned int age;
};
struct player *player_new_from_stream(FILE * stream){
struct player *new_player = (struct player*) malloc(sizeof(struct player));
char *p_name = malloc(20);
char *p_age = malloc(20);
if (stream != stdin){
if (fgets(p_name, 20, stream) != NULL){
char *p = strrchr(p_name, '\n');
if (p)
*p = '\0';
strcpy(new_player->name, p_name);
}
if (fgets(p_age, 20, stream) != NULL)
new_player->age = atoi(p_age);
}
else {
printf("enter name and age for a player\n");
gets(p_name);
gets(p_age);
strcpy(new_player->name, p_name);
new_player->age = atoi(p_age);
}
free(p_name);
free(p_age);
return new_player;
}
void player_inspect(struct player plyr, char* prefix){
printf("[%s] name: %s\n", prefix, plyr.name);
printf("[%s] age : %d\n", prefix, plyr.age);
}
int main(int argc, char* argv[]){
FILE * stream;
char* argument;
// below: trying to allocate (argc - 1) pointers
// valgrind's --show-origins=yes points here for both errors
struct player **players = malloc(sizeof(int) * (argc - 1));
int i = 1;
for (; i < argc; i++){
argument = argv[i];
if (strcmp("-", argument) != 0){
if ((stream = fopen(argument, "r")) == NULL) perror("Error opening file");
else {
// the next line emits Invalid write of size 8 in valgrind
players[i-1] = player_new_from_stream(stream);
fclose(stream);
}
} else {
players[i-1] = player_new_from_stream(stdin);
}
}
i = 0;
char buffer[15];
for (; i < argc - 1; i++){
sprintf(buffer, "%d", i);
// the next line emits Invalid read of size 8
player_inspect(*(players[i]), buffer);
free(players[i]);
}
free(players);
return 0;
}
What is wrong here? I want to return a pointer to struct player from player_new_from_stream and pack this pointer to array players in main().
This is wrong:
struct player **players = malloc(sizeof(int) * (argc - 1));
Use this instead:
struct player **players = malloc(sizeof(*players) * (argc - 1));
Note that on your system, sizeof(int) == 4 while sizeof(struct player *) == 8.
I ran it under valgrind with valid input files (player files), compiled with gcc -g and it didn't give any of these invalid read/write messages.
It also worked for using stdin.
However, when I ran it with non-existent files, it had a read error at
i = 0;
char buffer[15];
for (; i < argc - 1; i++){
sprintf(buffer, "%d", i);
player_inspect(*(players[i]), buffer); // <<HERE
free(players[i]);
}
Since the players[i] pointer was NULL due to the pointer at that array index not being set if the fopen call fails.
You need to do double allocation if you want to use array:
struct player **players = malloc(sizeof(struct player*) * (argc - 1));
for (int i=0; i<argc-1;i++)
player[i] = malloc(sizeof(struct player));