Read files separated by tab in c

问题:

I am really new to C, and the reading files thing drives me crazy... I want read a file including name, born place and phone number, etc. All separated by tab

The format might be like this:

Bob Jason   Los Angeles    33333333
Alice Wong  Washington DC  111-333-222

So I create a struct to record it.

typedef struct Person{
    char name[20];
    char address[30];
    char phone[20];
} Person;

I tried many ways to read this file into struct but it failed. I tired fread:

read_file = fopen("read.txt", "r");
Person temp;
fread(&temp, sizeof(Person), 100, read_file);
printf("%s %s %s \n", temp.name, temp.address, temp.phone);

But char string does not recorded into temp separated by tab, it read the whole file into temp.name and get weird output.

Then I tried fscanf and sscanf, those all not working for separating tab

fscanf(read_file, "%s %s %s", temp.name, temp.address, temp.phone);

Or

fscanf(read_file, "%s\t%s\t%s", temp.name, temp.address, temp.phone);

This separates the string by space, so I get Bob and Jason separately, while indeed, I need to get "Bob Jason" as one char string. And I did separate these format by tab when I created the text file.

Same for sscanf, I tried different ways many times...

Please help...

回答1:

I suggest:

1. Use fgets to read the text line by line.
2. Use strtok to separate the contents of the line by using tab as the delimiter.

// Use an appropriate number for LINE_SIZE
#define LINE_SIZE 200
char line[LINE_SIZE];

if ( fgets(line, sizeof(line), read_file) == NULL )
{
   // Deal with error.
}

Person temp;
char* token = strtok(line, "\t");
if ( token == NULL )
{
   // Deal with error.
}
else
{
   // Copy token at most the number of characters
   // temp.name can hold. Similar logic applies to address
   // and phone number.

   temp.name[0] = '\0';
   strncat(temp.name, token, sizeof(temp.name)-1);
}

token = strtok(NULL, "\t");
if ( token == NULL )
{
   // Deal with error.
}
else
{
   temp.address[0] = '\0';
   strncat(temp.address, token, sizeof(temp.address)-1);
}

token = strtok(NULL, "\n");
if ( token == NULL )
{
   // Deal with error.
}
else
{
   temp.phone[0] = '\0';
   strncat(temp.phone, token, sizeof(temp.phone)-1);
}

Update

Using a helper function, the code can be reduced in size. (Thanks @chux)

// The helper function.
void copyToken(char* destination,
               char* source,
               size_t maxLen;
               char const* delimiter)
{
    char* token = strtok(source, delimiter);
    if ( token != NULL )
    {
       destination[0] = '\0';
       strncat(destination, token, maxLen-1);
    }
}

// Use an appropriate number for LINE_SIZE
#define LINE_SIZE 200
char line[LINE_SIZE];

if ( fgets(line, sizeof(line), read_file) == NULL )
{
   // Deal with error.
}

Person temp;   
copyToken(temp.name, line, sizeof(temp.name), "\t");
copyToken(temp.address, NULL, sizeof(temp.address), "\t");
copyToken(temp.phone, NULL, sizeof(temp.phone), "\n");

回答2:

This is only for demonstration, there are better ways to initialize variables, but to illustrate your main question i.e. reading a file delimited by tabs, you can write a function something like this:

Assuming a strict field definition, and your struct definition you can get tokens using strtok().

//for a file with constant field definitions
void GetFileContents(char *file, PERSON *person)
{
    char line[260];
    FILE *fp;
    char *buf=0;
    char temp[80];
    int i = -1;

    fp = fopen(file, "r");
    while(fgets(line, 260, fp))
    {
        i++;
        buf = strtok(line, "\t\n");
        if(buf) strcpy(person[i].name, buf);
        buf = strtok(NULL, "\t\n");
        if(buf) strcpy(person[i].address, buf);
        buf = strtok(NULL, "\t\n");
        if(buf) strcpy(person[i].phone, buf);
        //Note:  if you have more fields, add more strtok/strcpy sections
        //Note:  This method will ONLY work for consistent number of fields.
        //If variable number of fields, suggest 2 dimensional string array. 
    }
    fclose(fp);
}

Call it in main() like this:

int main(void)
{
    //...
    PERSON person[NUM_LINES], *pPerson;  //NUM_LINES defined elsewhere
                                         //and there are better ways
                                         //this is just for illustration
    pPerson = &person[0];//initialize pointer to person

    GetFileContents(filename, pPerson);  //call function to populate person.
    //...
    return 0;
}

回答3:

First thing,

fread(&temp, sizeof(temp), 100, read_file);

will not work because the fields are not fixed width, so it will always read 20 characters for name 30 for address and so on, which is not always the correct thing to do.

You need to read one line at a time, and then parse the line, you can use any method you like to read a like, a simple one is by using fgets() like this

char   line[100];
Person persons[100];
int    index;

index = 0;
while (fgets(line, sizeof(line), read_file) != NULL)
 {
    persons[i++] = parseLineAndExtractPerson(line);
 }

Now we need a function to parse the line and store the data in you Person struct instance

char *extractToken(const char *const line, char *buffer, size_t bufferLength)
 {
    char  *pointer;
    size_t length;
    if ((line == NULL) || (buffer == NULL))
        return NULL;
    pointer = strpbrk(line, "\t");
    if (pointer == NULL)
        length = strlen(line);
    else
        length = pointer - line;
    if (length >= bufferLength) /* truncate the string if it was too long */
        length = bufferLength - 1;
    buffer[length] = '\0';
    memcpy(buffer, line, length);

    return pointer + 1;  
 }

Person parseLineAndExtractPerson(const char *line)
 {
    Person person;

    person.name[0]    = '\0';
    person.address[0] = '\0';
    person.phone[0]   = '\0';

    line = extractToken(line, person.name, sizeof(person.name));
    line = extractToken(line, person.address, sizeof(person.address));
    line = extractToken(line, person.phone, sizeof(person.phone));

    return person;
 }        

Here is a sample implementation of a loop to read at most 100 records

int main(void)
 {
    char   line[100];
    Person persons[100];
    int    index;
    FILE  *read_file;

    read_file = fopen("/path/to/the/file.type", "r");
    if (read_file == NULL)
        return -1;
    index = 0;
    while ((index < 100) && (fgets(line, sizeof(line), read_file) != NULL))
     {
        size_t length;

        /* remove the '\n' left by `fgets()'. */            
        length = strlen(line);
        if ((length > 0) && (line[length - 1] == '\n'))
            line[length - 1] = '\0';
        persons[index++] = parseLineAndExtractPerson(line);
     }
    fclose(read_file);
    while (--index >= 0)
        printf("%s: %s, %s\n", persons[index].name, persons[index].address, persons[index].phone);
    return 0;
 }

Here is a complete program that does what I think you need

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

typedef struct Person{
    char name[20];
    char address[30];
    char phone[20];
} Person;

char *extractToken(const char *const line, char *buffer, size_t bufferLength)
 {
    char  *pointer;
    size_t length;
    if ((line == NULL) || (buffer == NULL))
        return NULL;
    pointer = strpbrk(line, "\t");
    if (pointer == NULL)
        length = strlen(line);
    else
        length = pointer - line;
    if (length >= bufferLength) /* truncate the string if it was too long */
        length = bufferLength - 1;
    buffer[length] = '\0';
    memcpy(buffer, line, length);

    return pointer + 1;  
 }

Person parseLineAndExtractPerson(const char *line)
 {
    Person person;

    person.name[0]    = '\0';
    person.address[0] = '\0';
    person.phone[0]   = '\0';

    line = extractToken(line, person.name, sizeof(person.name));
    line = extractToken(line, person.address, sizeof(person.address));
    line = extractToken(line, person.phone, sizeof(person.phone));

    return person;
 }        

int main(void)
 {
    char   line[100];
    Person persons[100];
    int    index;
    FILE  *read_file;

    read_file = fopen("/home/iharob/data.dat", "r");
    if (read_file == NULL)
        return -1;
    index = 0;
    while (fgets(line, sizeof(line), read_file) != NULL)
     {
        size_t length;

        length = strlen(line);
        if (line[length - 1] == '\n')
            line[length - 1] = '\0';
        persons[index++] = parseLineAndExtractPerson(line);
     }
    fclose(read_file);
    while (--index >= 0)
        printf("%s: %s, %s\n", persons[index].name, persons[index].address, persons[index].phone);
    return 0;
 }

回答4:

Parsing strings returned by fgets can be very annoying, especially when input is truncated. In fact, fgets leaves a lot to be desired. Did you get the correct string or was there more? Is there a newline at the end? For that matter, is the end 20 bytes away or 32768 bytes away? It would be nice if you didn't need to count that many bytes twice -- once with fgets and once with strlen, just to remove a newline that you didn't want.

Things like fscanf don't necessarily work as intended in this situation unless you have C99's "scanset" feature available, and then that will automatically add a null terminator, if you have enough room. The return value of any of the scanf family is your friend in determining whether success or failure occurred.

You can avoid the null terminator by using %NNc, where NN is the width, but if there's a \t in those NN bytes, then you need to separate it and move it to the next field, except that means bytes in the next field must be moved to the field after that one, and the 90th field will need its bytes moved to the 91st field... And hopefully you only need to do that once... Obviously that isn't actually a solution either.

Given those reasons, I feel it's easier just to read until you encounter one of the expected delimiters and let you decide the behavior of the function when the size specified is too small for a null terminator, yet large enough to fill your buffer. Anyway, here's the code. I think it's pretty straightforward:

/*
 * Read a token.
 *
 * tok: The buffer used to store the token.
 * max: The maximum number of characters to store in the buffer.
 * delims: A string containing the individual delimiter bytes.
 * fileptr: The file pointer to read the token from.
 *
 * Return value:
 *   - max: The buffer is full. In this case, the string _IS NOT_ null terminated.
 *          This may or may not be a problem: it's your choice.
 *   - (size_t)-1: An I/O error occurred before the last delimiter
 *                 (just like with `fgets`, use `feof`).
 *   - any other value: The length of the token as `strlen` would return.
 *                      In this case, the string _IS_ null terminated.
 */
size_t
read_token(char *restrict tok, size_t max, const char *restrict delims,
    FILE *restrict fileptr)
{
        int c;
        size_t n;

        for (n = 0; n < max && (c = getchar()) != EOF &&
            strchr(delims, c) == NULL; ++n)
                *tok++ = c;

        if (c == EOF)
            return (size_t)-1;

        if (n == max)
            return max;

        *tok = 0;

        return n;
}

Usage is pretty straightforward as well:

#include <stdio.h>
#include <stdlib.h>

typedef struct person {
        char    name[20];
        char    address[30];
        char    phone[20];
} Person;

int
main(void)
{
        FILE *read_file;
        Person temp;
        size_t line_num;
        size_t len;
        int c;
        int exit_status = EXIT_SUCCESS;

        read_file = fopen("read.txt", "r");
        if (read_file == NULL) {
                fprintf(stderr, "Error opening read.txt\n");
                return 1;
        }

        for (line_num = 0;; ++line_num) {
                /*
                 * Used for detecting early EOF
                 * (e.g. the last line contains only a name).
                 */
                temp.name[0] = temp.phone[0] = 0;

                len = read_token(temp.name, sizeof(temp.name), "\t",
                    read_file);
                if (len == (size_t)-1)
                        break;
                if (len == max) {
                        fprintf(stderr, "Skipping bad line %zu\n", line_num + 1);
                        while ((c = getchar()) != EOF && c != '\n')
                                ;       /* nothing */
                        continue;
                }

                len = read_token(temp.address, sizeof(temp.address), "\t",
                    read_file);
                if (len == (size_t)-1)
                        break;
                if (len == max) {
                        fprintf(stderr, "Skipping bad line %zu\n", line_num + 1);
                        while ((c = getchar()) != EOF && c != '\n')
                                ;       /* nothing */
                        continue;
                }

                len = read_token(temp.phone, sizeof(temp.phone), "\t",
                    read_file);
                if (len == (size_t)-1)
                        break;
                if (len == max) {
                        fprintf(stderr, "Skipping bad line %zu\n", line_num + 1);
                        while ((c = getchar()) != EOF && c != '\n')
                                ;       /* nothing */
                        continue;
                }

                // Do something with the input here.  Example:
                printf("Entry %zu:\n"
                    "\tName:    %.*s\n"
                    "\tAddress: %.*s\n"
                    "\tPhone:   %.*s\n\n",
                    line_num + 1,
                    (int)sizeof(temp.name), temp.name,
                    (int)sizeof(temp.address), temp.address,
                    (int)sizeof(temp.phone), temp.phone);
        }

        if (ferror(read_file)) {
                fprintf(stderr, "error reading from file\n");
                exit_status = EXIT_FAILURE;
        }
        else if (feof(read_file) && temp.phone[0] == 0 && temp.name[0] != 0) {
                fprintf(stderr, "Unexpected end of file while reading entry %zu\n",
                    line_num + 1);
                exit_status = EXIT_FAILURE;
        }
        //else feof(read_file) is still true, but we parsed a full entry/record

        fclose(read_file);
        return exit_status;
 }

Notice how the exact same 8 lines of code appear in the read loop to handle the return value of read_token? Because of that, I think there's probably room for another function to call read_token and handle its return value, allowing main to simply call this "read_token handler", but I think the code above gives you the basic idea about how to work with read_token and how it can apply in your situation. You might change the behavior in some way, if you like, but the read_token function above would suit me rather well when working with delimited input like this (things would be a bit more complex when you add quoted fields into the mix, but not much more complex as far as I can tell). You can decide what happens with max being returned. I opted for it being considered an error, but you might think otherwise. You might even add an extra getchar when n == max and consider max being a successful return value and something like (size_t)-2 being the "token too large" error indicator instead.