Reading from file strange problem

[std10093]

Program has to read from a file of 10000 names in form of Surname Name(which are seperated by space or tab,fill with them a simple-BST ,print them in order and then search from a file of 1000 names in the BST.
The problem:I should find 900 people after searching.I found 825.
The reason:Some names are saved in BST with some characters that not exist. e.g. We read from file Wu Teresa and this name is saved as Wuax Teresa and Fox Jim as Fox' Jim.
I run at a linux environment.Bellow is my code,sorry for greeklish but i had write in greeklish.Here is the part that i read from stream and save into BST.Reading from search file is done with same way.

while ( fgets(buf, LINE_LENGTH, fp) != NULL) 
	{
        s=buf;
        empty_line=0;
	size_for_malloc=0;
	until=strchr(s,'\t' ); /* until points to 'tab'*/
	if(until == NULL)
	{ /* Files from uliko-3hs-askhshs may have an empty line at the end */
		    until=strchr(s,' '); /* In case names are seperated by whitespace*/
		    if(until == NULL)
		    {
				empty_line=1;
				until=strchr(s,'\n'); /* Case of empty line */
			}
			if(until == NULL)
			{
				printf("Sorry,file has to be in form Surname Name\n");
				return -1;
			}
		}
		if(empty_line) continue;
		while(strcmp(s,until)!=0)/* Go until the end of surname */
		{
			size_for_malloc++;
			s++;
		}
		counter=0;
		while( (*s==' ') || (*s=='\t') )/* Bypass spaces */
		{
                        size_for_malloc++;
			s++;
			counter++;
		}
		size_for_malloc-=counter;
		counter=1;
		line.surname=malloc( (size_for_malloc+1)*sizeof(char) );
		strncpy(line.surname,buf,size_for_malloc);
		size_for_malloc=0;
		until=strchr(s,' ' );
		if(until == NULL)
		{
                        until=strchr(s,'\t' );
			if(until == NULL)
			{
                               until=strchr(s,'\n' );
			}
		}
  		while(strcmp(s,until)!=0)
		{
			size_for_malloc++;
			s++;
		}
		s-=size_for_malloc;
		line.name=malloc( (size_for_malloc+1)*sizeof(char) );
		strncpy(line.name,s,size_for_malloc);
		count_lines++;
                Tree_eisagogi(&BulkTree , line , &error);/* Insert name to BST */
                if(error)
                {
			printf("Sorry,can not allocate memory . Read %u lines\n",count_lines);
			return -1;
		}
	}

Thanks in advance

[CommonTater]

Show me a few lines of the input file... preserve all characters and spacing in code tags, please.

[std10093]

input file

Stevenson       Christy
Riddle  Melissa
York    Joe
Shapiro Calvin
Bynum   Gene
Atkinson        Dan
Andrews Juan
Wilkerson       Francis
Hubbard Sidney
Brooks  Billy
Hansen  Shirley
Bauer   Kent
Snow    Gretchen
Patton  Jacqueline
Chen    Anna
Peele   Lester
Thomson Scott
Bernstein       Maxine
Griffith        Phyllis

search file

Black   Constance
Fox     Sarah
Hatcher Gladys
Wu      Hazel
Lloyd   Hazel
Joyce   Jerome
Welch   Vincent
Matthews        Kim
Chappell        Joseph
MacDonald       Alison
Kane    Eric
Butler  Heather
Pickett Claire
Bowman  Michele
Barton  Terry

[tabstop]

I don't see you putting \0 at the end of any of your names.

[CommonTater]

Ok... and I gather from your code you're loading these into some kind of struct...

struct name
  { char first[25];
     char last[25]; };

sort of thing...

You may want to look into the fscanf() function here... it basically does pattern matching on your file and loads variables that match the pattern...

For example...

// create the array (use malloc so you can realloc if necessary)
struct name *names = malloc(10000 * sizeof[name]);
int x = 0;   // array pointer

// open your file

// read the file
while (! feof(file))
  { if (fscanf (" %s %s",names[x]->last,names[x]->first))
      x++; }

Now this is untested and for concept only... but something along these lines will probably save you a lot of hassle, especially if you get blank lines, random numbers of spaces or multiple tabs...

[std10093]

I don't see you putting \0 at the end of any of your names.

Doesn't fgets put it automatically?

[tabstop]

Doesn't fgets put it automatically?

strncpy doesn't.

[quzah]

while( fgets( ... ) )
{
    if( sscanf( buf, "%s %s", word1, word2 ) == 2 )
    {
        sprintf( buf, "%s %s", word1, word2 );
        malloc_len = 1 + strlen( buf );
        ... malloc
        ... strcpy( mallocedplace, buf );
    }
    else /* didn't get what you expect, quit maybe? */
        ...
}

Quzah.

[std10093]

strncpy doesn't.

Yes,you are right.But why does this cause a problem?Names are read the same way from input file and from for-search file.

[quzah]

Yes,you are right.But why does this cause a problem?Names are read the same way from input file and from for-search file.

The first time through, you read in "fred", and you strncpy "fred" to a buffer. The next time you read "bob", and you strncpy "bob" to that same buffer. That buffer now looks like: bobd.

The strncpy() function copies at most len characters from src into dst.
If src is less than len characters long, the remainder of dst is filled
with `\0' characters. Otherwise, dst is not terminated.

Quzah.

[tabstop]

Yes,you are right.But why does this cause a problem?Names are read the same way from input file and from for-search file.

If you don't create the tree correctly in the first place, then searching for something later is irrelevant because the names don't actually exist in the tree.

[std10093]

The first time through, you read in "fred", and you strncpy "fred" to a buffer. The next time you read "bob", and you strncpy "bob" to that same buffer. That buffer now looks like: bobd.


Quzah.

YES!! that's it! This is what i thought it was the problem but a cannot find the solution.I tried to fill the buffer whit spaces before the 'reading' for the next name but this did not work.So the solution is to put a '\0' at the end of name?

[std10093]

However this does not explain the existance of some characters that do not exist in names.Such us

',Ψ,π

[quzah]

However this does not explain the existance of some characters that do not exist in names.Such us

',Ψ,π

Sure it does.

int main( void ) {
    char foo[BUFSIZ];
    return 0;
}

What values do the characters in foo hold?


Quzah.

[std10093]

Pseudo-random.Right? So if i figure out a way to put '\0' at the end of every name and surname then the program will be correct!Maybe i could use strcpy instead of strncpy-in case i put the '\0' .