fgetc() not working as I had hoped

Question

I hope i don't get voted down to quickly for this, but I have a project I'm working on for school in which I have to build a spell checker. I decided to use a trie, and it seems to be working, but I have a bug I can't find. I think the issue is in the following,

bool load(const char* dictionary)
{

    if (!rootNode)
    {
        rootNode = trieNodeCreate();
        if (!rootNode)
        {
            printf("could not allocate root node");
            return false;
        }
    }

    // Open the file 
    FILE* fp = fopen(dictionary, "r");

    if (fp == NULL)
    {
        printf("could not open dictioanry %s\n", dictionary);
        return false;
    }


    int index = 0;
    for (int c = fgetc(fp); c != EOF; c = fgetc(fp))
    {
        char word[LENGTH];
        if (c != '\n' )
        {
            word[index] = c;
            index++;
        }
        else
        {
            trieWordInsert(word, rootNode);
            index = 0;
                wordCount ++;

        } 

    }
    fclose(fp);

    if (wordCount)
    {
        return true;
    }
    return false;
}

but I've been unable to find it. The rest of the project can be found at

https://github.com/iMillJoe/spell-checker

Compile with all warnings and debug info (e.g. with `gcc -Wall -g`). **Use a debugger** (e.g. `gdb`) — Basile Starynkevitch, Mar 09 '14 at 18:21
Building it in X-Code, I get no warnings of any kind. Using lldb, i can see problems, but not what makes things go wrong. — Joe Million, Mar 09 '14 at 18:25
Please be more specific: what is the bug you're seeing? Cheers — MBlanc, Mar 09 '14 at 18:27
@Mblanc, the only word It seems I can determine is spelled correctly is 'a' even though my trie seems to work very well when I test it outside of the project, hence i think the bug lies here where I plug my trie into the distro code. — Joe Million, Mar 09 '14 at 18:29
@ThoAppelsin Yes, i can assume the dictionary is all lowercased words delimited by newlines. — Joe Million, Mar 09 '14 at 18:36

Utkan Gezer · Answer 1 · 2014-03-09T19:44:30.093

Declare your word[LENGTH] array outside of the loop, else it will just discard the word pointer and free the allocated at the end of each cycle, create a new one. I don't think you want that, I think you would rather want that only when the if condition does not get fulfilled.

I may not know what trieWordInsert does, but I will assume that you'll need a 0 terminator.

Declare word[LENGTH] = { 0 }; right before the for( ... ) { ... }
Add a memset( word, 0, LENGTH); inside the else block there
Include memory.h or string.h for the memset if you haven't included either one of those by far

And this should be it, I think...

Edit: After having learned how trieWordInsert more or less issues the word pushed in...

DIRECT CODE FOR EZ MODE:

bool load( const char* dictionary )
{

    if ( !rootNode )
    {
        rootNode = trieNodeCreate( );
        if ( !rootNode )
        {
            printf( "could not allocate root node" );
            return false;
        }
    }

    // Open the file 
    FILE* fp = fopen( dictionary, "r" );

    if ( fp == NULL )
    {
        printf( "could not open dictioanry %s\n", dictionary );
        return false;
    }

    int index = 0;
    char word[LENGTH];
    for ( int c = fgetc( fp ); c != EOF; c = fgetc( fp ) )
    {
        if ( c != '\n' )
        {
            word[index] = c;
            index++;
        }
        else
        {
            word[index] = 0;
            trieWordInsert( word, rootNode );
            index = 0;
            wordCount++;
        }

    }
    fclose( fp );

    if ( wordCount )
    {
        return true;
    }
    return false;
}

The `memset` is a tad overkill; just make sure the string is terminated before sending it (which should also be protected against adding empty strings, but the OP is likely parsing "correct" data "correctly" so may not be a big deal. None the less, this answer is solid, and up ticked. — WhozCraig, Mar 09 '14 at 18:46
@WhozCraig who knows... maybe he does something weird inside that `trieWordInsert`, requiring each character pointed by the `word` for `LENGTH` long to be zero. I have only wanted to make sure. — Utkan Gezer, Mar 09 '14 at 19:39

SHR · Accepted Answer · 2014-03-09T18:50:29.180

1

I think you didn't end the word with a '\0'

char word[LENGTH];
 if (c != '\n' )
 {
     word[index] = c;
     index++;
 }
 else
 {
     word[index] = '\0'; //missing this one!!!
     trieWordInsert(word, rootNode);
     index = 0;
     wordCount ++;
 }

I think you better use fscanf and read the file word by word.

edited Mar 09 '14 at 18:50

answered Mar 09 '14 at 18:44

SHR

7,940
9
38
57

Use of fscanf() was warned against in the projects specs. – Joe Million Mar 09 '14 at 18:49
Any string must have a terminating null. try to print the words before you call `trieWordInsert` and compare with the file. – SHR Mar 09 '14 at 18:51
@Joe - How does `trieWordInsert` know how long the word is that is being inserted? Surely all words are not exactly `LENGTH` long. – M.M Mar 09 '14 at 18:58
@MattMcNabb you might be onto something, trieWordInsert calls strlen() to figure out how long the string is. (LENGTH is 45, the longest word in english) If '/0' is not the end of the string, strlen() won't work. – Joe Million Mar 09 '14 at 19:03

fgetc() not working as I had hoped

2 Answers2

DIRECT CODE FOR EZ MODE: