Add new file

uniquewords.c

+
+#include <stdio.h>
+#include <string.h>
+#include <ctype.h>
+
+#define MAX_LINE 1024
+#define MAX_WORDS 50000  // Increased limit for unique words
+#define MAX_WORD_LENGTH 100
+
+// Structure to store unique words and their counts
+typedef struct {
+    char word[MAX_WORD_LENGTH];
+    int count;
+} WordCount;
+
+// Function prototypes
+void process_file(const char *filename);
+int count_unique_words(char *line, WordCount *word_counts, int *unique_word_count);
+int find_word(WordCount *word_counts, int unique_word_count, const char *word);
+
+int main() 
+{
+    process_file("raamat1184.txt");
+    return 0;
+}
+
+// Function to process the file and count unique words
+void process_file(const char *filename) 
+{
+    FILE *file = fopen(filename, "r");
+
+    if (file == NULL) 
+    {
+        printf("Error: Could not open file.\n");
+        return;
+    } 
+    else 
+    {
+        printf("File opened successfully.\n");
+    }
+
+    char line[MAX_LINE];
+    int unique_word_count = 0;
+    WordCount word_counts[MAX_WORDS] = {0}; // Array to store unique words and their counts
+
+    while (fgets(line, sizeof(line), file)) 
+    {
+        count_unique_words(line, word_counts, &unique_word_count);
+    }
+
+    fclose(file);
+    printf("File closed successfully.\n");
+    printf("There are %d different words in this text.\n", unique_word_count);
+}
+
+// Function to count unique words in a line
+int count_unique_words(char *line, WordCount *word_counts, int *unique_word_count) 
+{
+    char *word = strtok(line, " \t\n");
+
+    while (word != NULL) 
+    {
+        // Convert word to lowercase to ensure case-insensitive comparison
+        for (char *p = word; *p; ++p) *p = tolower(*p);
+
+        int index = find_word(word_counts, *unique_word_count, word);
+        if (index == -1) 
+        {
+            // New unique word
+            if (*unique_word_count < MAX_WORDS) 
+            {
+                strncpy(word_counts[*unique_word_count].word, word, MAX_WORD_LENGTH - 1);
+                word_counts[*unique_word_count].word[MAX_WORD_LENGTH - 1] = '\0'; // Ensure null-termination
+                word_counts[*unique_word_count].count = 1;
+                (*unique_word_count)++;
+            } 
+            else 
+            {
+                printf("Error: Exceeded maximum number of unique words.\n");
+                return -1;
+            }
+        } 
+        else 
+        {
+            // Existing word
+            word_counts[index].count++;
+        }
+        word = strtok(NULL, " \t\n");
+    }
+
+    return *unique_word_count;
+}
+
+// Function to find a word in the word_counts array
+int find_word(WordCount *word_counts, int unique_word_count, const char *word) 
+{
+    for (int i = 0; i < unique_word_count; i++) 
+    {
+        if (strcmp(word_counts[i].word, word) == 0) 
+        {
+            return i;
+        }
+    }
+    return -1;
+}