/**************************************************
* Section : 15
* Machine No. : N
* Roll No. : 19CS100XY
* Name : Aritra Hazra
* Assignment No : 8
* Description : Determines Statistics of Word Array
***************************************************/

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

void initializeWordsArray(char ***pwords, int wc, int wl)
{
    int i;
    (*pwords) = (char **)malloc(wc * sizeof(char *));
    for(i=0; i<wc; i++)
        (*pwords)[i] = (char *)malloc(wl * sizeof(char));
}

void scanWordsIntoWordsArray(char **words, int wc)
{
    int i;
    printf("Enter %d Words into Words-Array: ", wc);
    for(i=0; i<wc; i++)
        scanf("%s", words[i]);
}

void printWordsFromWordsArray(char **words, int wc)
{
    int i;
    printf("Sorted %d Words from Words-Array --\n", wc);
    for(i=0; i<wc; i++)
        printf(" %s,", words[i]);
    printf("\n");
}

void lexicographicSortWordsArray(char **words, int wc, int wl)
{
    int i, j, minIndex;
    char *tempWord;
    tempWord = (char *)malloc(wl * sizeof(char));
    /* selection sort */
    for(i=0; i<wc-1; i++){
        minIndex = i;
        for(j=i+1; j<wc; j++){
            if(strcasecmp(words[j],words[minIndex]) < 0)
                minIndex = j;
        }
        strcpy(tempWord, words[i]);
        strcpy(words[i], words[minIndex]);
        strcpy(words[minIndex], tempWord);
    }
}

void printWordLengthStatistics(char **words, int wc)
{
    int i, wordLen_1_2=0, wordLen_3_5=0, wordLen_6_10=0, wordLen_10_more=0;
    for(i=0; i<wc; i++){
        if(strlen(words[i]) > 10) wordLen_10_more++;  /* word having length more than 10 characters */
        else if(strlen(words[i]) > 5) wordLen_6_10++; /* word having length between 6-10 characters */
        else if(strlen(words[i]) > 2) wordLen_3_5++; /* word having length between 3-5 characters */
        else wordLen_1_2++; /* word having length between 1-2 character(s) */
    }
    printf("\n-- The Word Length Statistics --\n");
    printf("Number of Words with Length 1-2: %d\n", wordLen_1_2);
    printf("Number of Words with Length 3-5: %d\n", wordLen_3_5);
    printf("Number of Words with Length 6-10: %d\n", wordLen_6_10);
    printf("Number of Words with Length larger than 10: %d\n", wordLen_10_more);
}

void printAlphabetDistribution(char **words, int wc)
{
    int i, j, totalCharCount=0, totalVowelCount = 0, totalCapsCount = 0, charCount, charPercent;
    char ch;
    
    printf("\n-- The Alphabet Distribution --\n");
    /* finding total number of characters */
    for(i=0; i<wc; i++){
        for(j=0; j<strlen(words[i]); j++){
            totalCharCount++;
            if( isupper(words[i][j]) ) {
                totalCapsCount ++;
            }
            if( (tolower(words[i][j])=='a') || (tolower(words[i][j])=='e') || (tolower(words[i][j])=='i') || (tolower(words[i][j])=='o') || (tolower(words[i][j])=='u') ) {
                totalVowelCount++;
            }
        }
    }
    printf("Total Number of Letters Present = %d\n",totalCharCount);
    printf("Total Number of Vowels Present = %d\n",totalVowelCount);
    printf("Total Number of Consonants Present = %d\n",totalCharCount-totalVowelCount);
    printf("Total Number of Capital Letters Present = %d\n",totalCapsCount);
    printf("Total Number of Small Letters Present = %d\n",totalCharCount-totalCapsCount);

    printf("\n-- The Percentage Distribution of the Letters a-z --\n");
    for(ch='a'; ch<='z'; ch++){ /* looping for finding frequency of each character a-z */
        charCount = 0;
        for(i=0; i<wc; i++){
            for(j=0; j<strlen(words[i]); j++){
                if(tolower(words[i][j]) == ch)
                    charCount++;
            }
        }
        /* determine percentage of occurrence (rounf-off to an integer value) */
        charPercent = (int)(((charCount*100.0)/totalCharCount) + 0.5);
        printf("%c [ %3d %% ] : ", ch, charPercent);
        for(i=0; i<charPercent; i++)
            printf("=");
        printf("\n");
    }
}

int removePrintDuplicateWords(char **words, int wc)
{
    int noOfWords = wc;
    int i, j, k;
    printf("\n-- Duplicate Words --\n");
    for(i=0; i<noOfWords-1; i++){
        j=i+1; 
        while(j<noOfWords){
            if ( strcasecmp(words[i],words[j]) ) /* not a duplicate */
                break;
            else /* duplicate word */
                j++;
        }
        if(j > i+1){
            printf(" %s", words[i]);
            for(k=0; k<noOfWords-j; k++){ /* remove duplicates */
                strcpy(words[i+1+k], words[j+k]);
            }
        }
        noOfWords = noOfWords - (j-i-1); /* reduce number of words */
    }
    printf("\n");
    return noOfWords;
}

int main()
{
    char **words;
    int wordCount, wordLength, noOfUniqueWords;
    
    printf("Enter Number of Words: ");
    scanf("%d", &wordCount);
    printf("Enter Maximum Length of Words: ");
    scanf("%d", &wordLength);
    
    initializeWordsArray(&words, wordCount, wordLength);
    
    scanWordsIntoWordsArray(words, wordCount);
    
    lexicographicSortWordsArray(words, wordCount, wordLength);
    
    printf("\n+++ Word Statistics +++\n\n-- ");
    printWordsFromWordsArray(words, wordCount);
    
    printWordLengthStatistics(words, wordCount);
    
    printAlphabetDistribution(words, wordCount);
    
    noOfUniqueWords = removePrintDuplicateWords(words, wordCount);
    
    printf("-- Unique ");
    printWordsFromWordsArray(words, noOfUniqueWords);

    return 0;
}