cirilisp/read.c

#define _POSIX_C_SOURCE 200809L
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <regex.h>

#include "util.h"
#include "read.h"
#include "print.h"

typedef enum
{
	undefinedToken,
	numberToken,
	symbolToken,
	lParenthesisToken,
	rParenthesisToken
} tokenType;

typedef struct _Token
{
	tokenType type;
	char *lexeme;
	struct _Token *next;
} token;

int completeSExpr(token **tokenList);
char *readline();
void append(token **head, token *appendix);
token *lexLine(char *input);
object parseObject(token **inputList);

token *tokenList = NULL;

object read(char *prompt)
{
	printf("%s", prompt);

	while (!completeSExpr(&tokenList))
	{
		char *input = readline();
		if (input == NULL) /* унесен је EOF сигнал */
		{
			printf("\nКрај улазног стрима.\n");
			exit(0);
		}
		append(&tokenList, lexLine(input));
	}

	return parseObject(&tokenList);
}

int completeSExpr(token **tokenList)
{
	int result = 0, indentLevel = 0;
	token *current = *tokenList;

	while (current != NULL)
	{
		if (current->type == lParenthesisToken)
		{
			++indentLevel;
		}
		else if (current->type == rParenthesisToken)
		{
			if (indentLevel == 0)
			{
				token **deleteParen = tokenList;
				while (*deleteParen != current)
				{
					deleteParen = &(*deleteParen)->next;
				}
				*deleteParen = current->next;
				free(current);
				current = *deleteParen;
				continue;
			}
			else
			{
				if (indentLevel == 1)
				{
					result = 1;
				}
				--indentLevel;
			}
		}
		else
		{
			if (indentLevel == 0)
			{
				result = 1;
			}
		}
		current = current->next;
	}
	return result;
}

ssize_t bytesRead;
size_t nbytes = 2048;
char *buffer = NULL;

char *readline()
{
	if (buffer == NULL)
	{
		buffer = (char *) malloc(nbytes + 1);
	}
	bytesRead = getline(&buffer, &nbytes, stdin);
	if (bytesRead == -1)
	{
		return NULL;
	}
	
	char *cpy = malloc(strlen(buffer)+1);
	strcpy(cpy, buffer);
	cpy[strlen(cpy)-1] = '\0';
	// Уклања завршни њу-лајн или ЕОФ у стрингу и копира га на ново место

	return cpy;
}

void append(token **head, token *tail)
{
	token **current = head;

	while (*current != NULL)
	{
		current = &(*current)->next;
	}
	*current = tail;
}

regex_t regNumber, regSymbol, regLParenthesis, regRParenthesis, regSpace;

token *lex1Token(char *input, int *i)
{
	token *result = malloc(sizeof(token));
	result->next = NULL;

	regcomp(&regSpace, "^[[:space:]]*", REG_EXTENDED);

	regcomp(&regNumber, "^[-+]?[[:digit:]]+", REG_EXTENDED);
	regcomp(&regSymbol, "^[-+/*]", REG_EXTENDED);
/* за сада подржава само симболе -, +, * и / */
	regcomp(&regLParenthesis, "^\\(", REG_EXTENDED);
	regcomp(&regRParenthesis, "^\\)", REG_EXTENDED);

	const int nmatches = 1;
	regmatch_t a[nmatches];

	regexec(&regSpace, input + *i, nmatches, a, 0);
	*i += a[0].rm_eo;
/* помера индекс да би се игнорисали почетни "вајт-спејс" карактери */

	if (!regexec(&regNumber, input + *i, nmatches, a, 0))
	{
		result->type = numberToken;
	}
	else if (!regexec(&regSymbol, input + *i, nmatches, a, 0))
	{
		result->type = symbolToken;
	}
	else if (!regexec(&regLParenthesis, input + *i, nmatches, a, 0))
	{
		result->type = lParenthesisToken;
	}
	else if (!regexec(&regRParenthesis, input + *i, nmatches, a, 0))
	{
		result->type = rParenthesisToken;
	}
	else
	{
		result->type = undefinedToken;
		result->lexeme = NULL;
		goto skipStringCopy;
	}
	result->lexeme = malloc((a[0].rm_eo + 1) * sizeof(char));
	strncpy(result->lexeme, input + *i, a[0].rm_eo);
	result->lexeme[a[0].rm_eo] = '\0';
	*i += a[0].rm_eo;

	regexec(&regSpace, input + *i, nmatches, a, 0);
	*i += a[0].rm_eo;
/* игнорисање крајњих вајт-спејс карактера */

	skipStringCopy:
	regfree(&regSpace);
	regfree(&regNumber);
	regfree(&regSymbol);
	regfree(&regLParenthesis);
	regfree(&regRParenthesis);

	return result;
}

token *lexLine(char *input)
{
	int i = 0, n;
	n = strlen(input);
	token *root = NULL, **new;
	new = &root;
	while (i < n)
	{
		*new = lex1Token(input, &i);
		if ((*new)->type == undefinedToken)
		{
/* уколико се у реду нађе токен који је лексички погрешан, штампа се место тог
токена у реду и бришу се сви токени нађени у реду, функција враћа NULL*/
			fprintf(stderr, "Невалидан токен на месту %d\n", i);
			new = &root;
			while (*new != NULL)
			{
				free(root->lexeme);
				new = &((*new)->next);
				free(root);
				root = *new;
			}
			return NULL;
		}
		new = &((*new)->next);
	}
	return root;
}

object parseObject(token **inputList)
{
	object result;

	token input = **inputList;
	free(*inputList);
	*inputList = input.next;

	if (input.type == numberToken)
	{
		result.type = numberObject;
		result.address = malloc(sizeof(long long int));
		*((long long *) result.address) = atoll(input.lexeme);
		return result;
	}
	else if (input.type == symbolToken)
	{
		result.type = symbolObject;
		result.address = malloc((strlen(input.lexeme) + 1)
					 * sizeof(char));
		strcpy((char *) result.address, input.lexeme);
		return result;
	}
	else if (input.type == lParenthesisToken)
	{
		object *listCurrent = &result;

		while ((*inputList)->type != rParenthesisToken)
		{
			listCurrent->type = consObject;
			listCurrent->address = malloc(sizeof(cons));

			((cons *) listCurrent->address)->car =
				parseObject(inputList);

			listCurrent = &(((cons *) listCurrent->address)->cdr);
		}

		(*listCurrent).type = nilObject;
		(*listCurrent).address = NULL;

		input = **inputList;
		free(*inputList);
		*inputList = input.next;
	}

	return result;
}