parsing into a struct - c

Question

trying to parse a csv file containing lines in this format
1001,Kauri tree,1002,-1,1001,1001
and the struct node has attributes: id, name and then paths[4] // (4 paths)
my code is not working because it seg faults. I'm on a mac so I valgrind doesn't work for me. Can anyone help with my code? or better yet give me another option for debugging? I am using the Geany IDE.
#include <stdio.h>
#include <stdlib.h>
#include <stdbool.h>
#include <ctype.h>
#include <string.h>
#define MAX_NAME_LENGTH 20
#define MAX_LINE_LENGTH 100
typedef struct node_s {
int id;
char name[MAX_NAME_LENGTH + 1];
int paths[4];
} Node;
Node readNode(FILE *infile)
{
char buffer[MAX_LINE_LENGTH];
Node *node = NULL;
char* tok;
const char comma[2] = ",";
char* inputLine = fgets(buffer, MAX_LINE_LENGTH, infile);
if (inputLine != NULL) {
tok = strtok(buffer, comma);
if (tok == NULL) {
node->id = -1;
} else {
node->id = *tok;
tok = strtok(NULL, comma);
strncpy(node->name, tok, MAX_NAME_LENGTH);
int i = 0;
while(tok != NULL || i < 4) {
tok = strtok(NULL, comma);
node->paths[i] = atoi(tok);
i++;
}
}
}
return *node;
}
int main(void)
{
FILE* infile = fopen("new_list.txt", "r");
Node node = readNode(infile);
while (node.id >= 0) {
printf("Node: id = %d, name = '%s', neighbours = [%d, %d, %d, %d]\n",
node.id, node.name,
node.paths[0], node.paths[1], node.paths[2], node.paths[3]);
node = readNode(infile);
}
}

Answer 1

Your primary problem in readNode() is your declaration of Node *node = NULL; declares a pointer that is initialized NULL and points to no valid storage. You then attempt to assign values to the memory pointed to by node (e.g. NULL) invoking Undefined Behavior and almost guaranteeing a SegFault.
You have several options to handle the issue,
provide storage for node in main() and pass as a parameter -- declare node in main() with automatic storage and pass the address of the struct as an additional parameter to readNode(), filling the values within the function (readNode can be declared as void in this case). You can also dynamically allocate node in main() and simply pass the pointer as an additional parameter,
allocate storage for node in readNode() and return a pointer -- dynamically declare storage for node within readNode() using malloc or calloc (don't forget to free the storage at the end of each loop to prevent a memory leak), or
declare readNode() as type Node -- the easiest way is simply to declare node within readNode and rely on the fact that a function can always return its own type to return a filled struct to main() as noted in the comments.
Either way will eliminate your Segmentation Fault as valid storage will be provided for node both in readNode and in main().
While not an error, the standard coding style for C avoids the use of camelCase or MixedCase variable names in favor of all lower-case while reserving upper-case names for use with macros and constants. It is a matter of style -- so it is completely up to you, but failing to follow it can lead to the wrong first impression in some circles. readnode is just a readable readNode, etc..
While it is perfectly fine to pass FILE *infile is a parameter to readnode(), it does you absolutely no good, and invites Undefined Behavior if you fail to validate that infile is open for reading before calling readnode() in main(). Library functions (and your functions should) provide a meaningful return to allow you to determine if the function call succeeded or failed -- use them, always.
Don't hard-code numbers or strings in your code (this is called using magic numbers -- don't do it). Instead, if you need a constant #define one (or more) as you have with MAX_NAME_LENGTH and MAX_LINE_LENGTH, and don't skimp on buffer sizes, e.g.
#include <stdio.h>
#include <string.h>
#define NPATHS 4
#define MAX_NAME_LENGTH 64 /* don't skimp on buffer size */
#define MAX_LINE_LENGTH 512
typedef struct node_s {
int id;
char name[MAX_NAME_LENGTH + 1];
int paths[NPATHS];
} node_t;
(note: an exception is when numeric values are required in your code such as when specifying the scanf field-width modifier, etc. where a defined constant or variable is not allowed)
Reading with fgets is a good approach, but you need to validate that a complete line was read by checking the length against MAX_LINE_LENGTH - 1 and that the last character in the buffer is the newline character. Further, while parsing with strtok is fine, when reading formatted input, calling sscanf on the buffer filled by fgets and validating the number of conversions simplifies the conversion process. It is also helpful to fill a temporary struct with values when parsing with sscanf to protect against some number of members less than all being filled and rendering your node.id >= 0 check invalid in main(). You could do something like the following:
node_t readnode (FILE *infile)
{
char buffer[MAX_LINE_LENGTH];
node_t node = {.id = -1}; /* initialize node to indicate error */
if (fgets (buffer, MAX_LINE_LENGTH, infile)) { /* validate line read */
node_t tmp = {.id = 0}; /* parse to temporary node */
size_t len = strlen (buffer); /* get length */
/* validate complete line read */
if (len == MAX_LINE_LENGTH - 1 && buffer[len-1] != '\n') {
/* handle line too long */
fputs ("error: line too long.\n", stderr);
/* discard remaining characters in line */
while (fgets (buffer, MAX_LINE_LENGTH, infile)) {
len = strlen (buffer);
if (len && buffer[len-1] == '\n')
break;
}
}
/* parse csv values using sscanf, validate return */
if (sscanf (buffer, "%d,%64[^,],%d,%d,%d,%d", &tmp.id, tmp.name,
&tmp.paths[0], &tmp.paths[1], &tmp.paths[2],
&tmp.paths[3]) == 6)
node = tmp; /* good parse, assign tmp to node */
else /* parse failed, issue error, return zeroed node */
fputs ("readnode() error: parse of line failed.\n", stderr);
}
return node; /* return filled node on success, zeroed node otherwise */
}
Putting all the pieces together in a short example based on your code, you could do something similar to to the following that will read from the filename provided as the first-argument to your program (or from stdin by default if no argument is given):
#include <stdio.h>
#include <string.h>
#define NPATHS 4
#define MAX_NAME_LENGTH 64 /* don't skimp on buffer size */
#define MAX_LINE_LENGTH 512
typedef struct node_s {
int id;
char name[MAX_NAME_LENGTH + 1];
int paths[NPATHS];
} node_t;
node_t readnode (FILE *infile)
{
char buffer[MAX_LINE_LENGTH];
node_t node = {.id = -1}; /* initialize node to indicate error */
if (fgets (buffer, MAX_LINE_LENGTH, infile)) { /* validate line read */
node_t tmp = {.id = 0}; /* parse to temporary node */
size_t len = strlen (buffer); /* get length */
/* validate complete line read */
if (len == MAX_LINE_LENGTH - 1 && buffer[len-1] != '\n') {
/* handle line too long */
fputs ("error: line too long.\n", stderr);
/* discard remaining characters in line */
while (fgets (buffer, MAX_LINE_LENGTH, infile)) {
len = strlen (buffer);
if (len && buffer[len-1] == '\n')
break;
}
}
/* parse csv values using sscanf, validate return */
if (sscanf (buffer, "%d,%64[^,],%d,%d,%d,%d", &tmp.id, tmp.name,
&tmp.paths[0], &tmp.paths[1], &tmp.paths[2],
&tmp.paths[3]) == 6)
node = tmp; /* good parse, assign tmp to node */
else /* parse failed, issue error, return zeroed node */
fputs ("readnode() error: parse of line failed.\n", stderr);
}
return node; /* return filled node on success, zeroed node otherwise */
}
int main (int argc, char **argv)
{
node_t node = {.id = 0};
FILE *infile = argc > 1 ? fopen (argv[1], "r") : stdin;
/* validate file open for reading */
if (infile == NULL) {
perror ("fopen-infile");
return 1;
}
node = readnode (infile); /* struct can be assigned */
while (node.id >= 0) {
printf ("Node: id = %4d, name = '%s',%*s"
"neighbours = [%d, % 4d, %d, %d]\n",
node.id, node.name, 11 - (int)strlen(node.name), " ",
node.paths[0], node.paths[1], node.paths[2], node.paths[3]);
node = readnode (infile);
}
if (infile != stdin) fclose (infile); /* if not stdin, close file */
return 0;
}
Example Input File
$ cat dat/struct_node.csv
1001,Kauri tree,1002,-1,1001,1001
1002,Beach tree,1003,-2,1002,1002
1003,Pine tree,1004,-10,1003,1003
1004,Elm tree,1005,-100,1004,1004
Example Use/Output
$ ./bin/struct_rd_csv_node <dat/struct_node.csv
Node: id = 1001, name = 'Kauri tree', neighbours = [1002, -1, 1001, 1001]
Node: id = 1002, name = 'Beach tree', neighbours = [1003, -2, 1002, 1002]
Node: id = 1003, name = 'Pine tree', neighbours = [1004, -10, 1003, 1003]
Node: id = 1004, name = 'Elm tree', neighbours = [1005, -100, 1004, 1004]
Look things over and let me know if you have further questions.