#include "parse.h"
#include "indextext.h"
#include <stdio.h>
#include <string.h>
#include <errno.h>
#include <stdlib.h>
#include "database.h"

void yyerror(const char* errorMessage) {
	fprintf(stderr, "YYERROR DURING PARSING LINE %i: %s\n", currentLine,
			errorMessage);
}

int is_binary_file(const char* file_path) {
	int is_binary = 0;
	
	if (strlen(file_path) > 4) {
		const char* suffix = file_path + (strlen(file_path) - 4);
		if ((strcasecmp(suffix, ".gif") == 0) ||
			(strcasecmp(suffix, ".xbm") == 0) ||
			(strcasecmp(suffix + 2, ".Z") == 0) ||
			(strcasecmp(suffix + 1, ".gz") == 0) ||
			(strcasecmp(suffix, ".tar") == 0) ||
			(strcasecmp(suffix, ".jpg") == 0)) {
			is_binary = 1;
		}
	}
	return is_binary;
}


int index_next_file(FILE* list_file) {
	static int number_of_files = 0;
	static char input_line[1024];
	char* additional_terms;
	FILE* infile;
	wordlist* words;
	wordlist* nextwords;
	int wordct;
	FileID fileID;
	int success = 0;

	if (fgets(input_line, 1024, list_file) != 0) {
		input_line[strlen(input_line) - 1] = '\0';
		additional_terms = strchr(input_line, ' ');
		if (additional_terms != 0) {
			*additional_terms++ = '\0';
		}
		if ((infile = fopen(input_line, "r")) == (FILE*)0) {
			perror(strerror(errno));
		} else {
			initializeindex();
			fileID = add_file(input_line);
			printf("INDEXING %s\n", input_line);
			if (!is_binary_file(input_line)) {
				if (number_of_files++ == 0) {
					lex_init(infile);
				} else {
					lex_restart(infile);
					/* pack up the database once for every 128 files indexed */
					if ((number_of_files % 128) == 0) {
						pack_database();
					}
				}
				yyparse();
				if (strlen(currentTitle) > 0) {
					add_file_property(fileID, "title", currentTitle);
				}
			}
			if (additional_terms != 0) {
				int tag_length;
				while ((tag_length = simplify_tag(additional_terms)) > 0) {
/* 					printf("additional_terms: %s\n", additional_terms); */
					add_word(additional_terms, fileID,5000,1);
					additional_terms += tag_length + 1;
				}
			}
			for (wordct = 0, words = orphanindex(); words != NULL; wordct++) {
				add_word(words->word, fileID, words->density,words->firstline);
				nextwords = words->next;
				free(words->word);
				free(words);
				words = nextwords;
			}
			fclose(infile);
			success = 1;
		}
	}
	return success;
}

void
main(int argc, char* argv[])
{
	if (argc != 2) {
		fprintf(stderr, "\tSyntax: %s <dbfilename>\n", argv[0]);
		exit(1);
	}
	if (open_database(argv[1], database_write) == 0) {
		fprintf(stderr, "\tUnable to open database %s\n", argv[1]);
		exit(1);
	}
	fprintf(stderr, "Reading file names from stdin...\n");
	while (index_next_file(stdin));
	close_database();
	printf("All files indexed.  Exiting normally.\n");
	exit(0);
}