#include "parse.h"
#include "indextext.h"
#include <stdio.h>
#include <strings.h>
#include <ctype.h>
#pragma alloca

static wordlist* currentwords;
static totalwords;

void indexword(const char* word, int wordlength, int lineNumber);

void indexbodytext(const char* bodytext, int lineNumber) {
	int charindex;
	int wordlength;
	wordlength = 0;
	for (charindex = 0; charindex < strlen(bodytext); charindex++) {
		if (isalnum(bodytext[charindex]) ||
			bodytext[charindex] == '_' ||
			bodytext[charindex] == '~') {

			wordlength++;
		} else {
			if (wordlength > 2) {
				indexword(bodytext + (charindex - wordlength),
						  wordlength, lineNumber);
			}
			wordlength = 0;
		}
	}
	if (wordlength > 2) {
		indexword(bodytext + (charindex - wordlength), wordlength, lineNumber);
	}
}

int simplify_tag(char* original_text) {
	char* tag = original_text;
	char* index;
	char* end_original = original_text + strlen(original_text);
	int result_length = 0;

	while ((isalnum(*tag) || (strspn(tag, "_./")))) {
		*tag = tolower(*tag);
		tag++;
	}
	index = tag + strspn(tag, " \t\n\r");
	if (*index == '=') {
	    *tag++ = *index++;
		index += strspn(index, " \t\n\r");
		if (*index == '"') {
			while (++index < end_original && *index != '"') {
				*tag++ = tolower(*index);
			}
		} else {
			while ((index < end_original) &&
				   ((isalnum(*index) ||
					 strspn(index, "_./")))) {
				*tag++ = tolower(*index++);
			}
		}
		*tag = '\0';
		result_length = index - original_text;
	}
	return result_length;
}

void indexcomment(const char* commenttext) {
	int tag_index = 0;
	int score = 0;
	int comment_length = strlen(commenttext);
/*	char* tag_result = (char*)malloc(comment_length);  */
	char* tag_result = (char*)alloca(comment_length);

	strcpy(tag_result, commenttext + 1 + strspn(commenttext + 1, " \t\n\r"));
	tag_index = simplify_tag(tag_result) + 2;
	if (tag_index > 0) {
		if (tag_index < comment_length) {
			const char* score_pointer = commenttext + tag_index + 1;
			score_pointer += strspn(score_pointer, " \t\n\r");
			if (score_pointer < commenttext + comment_length &&
				isdigit(*score_pointer)) {
				sscanf(score_pointer, "%i\n", &score);
			}
		}
		indexword(tag_result, strlen(tag_result), score);
	}
/*	free(tag_result);  */
}

void indexword(const char* word, int wordlength, int lineNumber) {
	wordlist** listpointer;
	for (listpointer = &currentwords;
		 *listpointer != NULL;
		 listpointer = &((*listpointer)->next)) {
		if (wordlength == strlen((*listpointer)->word)) {
			if (strncasecmp((*listpointer)->word, word, wordlength) == 0) {
				(*listpointer)->density++;
				break;
			}
		}
	}
	if ((*listpointer) == NULL) {
		int charIndex;
		*listpointer = (wordlist*)malloc(sizeof(wordlist));
		(*listpointer)->word = (char*)malloc(wordlength + 1);
		for (charIndex = 0; charIndex < wordlength; charIndex++) {
			(*listpointer)->word[charIndex] = tolower(word[charIndex]);
		}
		(*listpointer)->word[charIndex] = '\0';
		(*listpointer)->firstline = lineNumber;
		(*listpointer)->density = 1;
		(*listpointer)->next = NULL;
	}
	totalwords++;
}

void initializeindex(){
	currentwords = NULL;
	totalwords = 0;
}

wordlist* orphanindex(){
	wordlist* templist;
	for (templist = currentwords; templist!= NULL; templist = templist->next) {
	    templist->density =
			(2<<16) * (((float)templist->density) / totalwords);
	}
	templist = currentwords;
	currentwords = NULL;
	return templist;
}