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

#define STRIP_WHITESPACE(text) while (isspace(*(text))) (text)++

#define IS_KEY_CHAR(c) (\
	'a' <= (c) && (c) <= 'z' ||\
	'A' <= (c) && (c) <= 'Z' ||\
	'0' <= (c) && (c) <= '9' ||\
	(c) == '_' || (c) == '-'\
)

#define IS_DIGIT(c) (\
	'0' <= (c) && (c) <= '9'\
)

int cudl_err = CUDL_OK;

static char *fread_all(FILE *file) {
	size_t size;
	char *buffer;
	fseek(file, 0, SEEK_END);
	size = ftell(file);
	rewind(file);
	clearerr(file);
	if ((buffer = malloc(size + 1)) == NULL)
		return NULL;
	if (fread(buffer, 1, size, file) != size) {
		free(buffer);
		return NULL;
	}
	buffer[size] = '\0';
	return buffer;
}

void cudl_debug(struct cudl_value value) {
	int i;
	switch (value.tag) {
		case CUDL_TAG_NULL:
			printf("%%null");
			break;
		case CUDL_TAG_BOOL:
			if (value.data.boolean)
				printf("%%true");
			else
				printf("%%false");
			break;
		case CUDL_TAG_NUMBER:
			printf("%lf", value.data.number);
			break;
		case CUDL_TAG_STRING:
			printf("\"%s\"", value.data.string);
			break;;
		case CUDL_TAG_ARRAY:
			printf("[");
			for (i = 0; i < value.data.array.length; i++) {
				if (i != 0)
					printf(" ");
				cudl_debug(value.data.array.values[i]);
			}
			printf("]");
			break;
		case CUDL_TAG_MAP:
			printf("{");
			for (i = 0; i < value.data.map.length; i++) {
				if (i != 0)
					printf(" ");
				printf("\"%s\": ", value.data.map.fields[i].key);
				cudl_debug(value.data.map.fields[i].value);
			}
			printf("}");
			break;
		default:
			printf("UNKNOWN");
			break;
	}
}

/* Free all children of the value, not the value itself */
void cudl_deinit_value(struct cudl_value value) {
	int i;
	switch (value.tag) {
		case CUDL_TAG_ARRAY:
			for (i = 0; i < value.data.array.length; i++) {
				cudl_deinit_value(value.data.array.values[i]);
			}
			free(value.data.array.values);
			break;
		case CUDL_TAG_NULL:
		default:
			break;
	}
}

/* Parse a value from input and store it in value.
 * Return the number of bytes consumed.
 * Input must end with a null byte */
static size_t parse_value(char *input, struct cudl_value *value);

static size_t parse_bool_or_null(char *input, struct cudl_value *value) {
	if (strncmp(input, "null", 4) == 0) {
		value->tag = CUDL_TAG_NULL;
		return 4;
	}
	if (strncmp(input, "true", 4) == 0) {
		value->tag = CUDL_TAG_BOOL;
		value->data.boolean = 1;
		return 4;
	}
	if (strncmp(input, "false", 5) == 0) {
		value->tag = CUDL_TAG_BOOL;
		value->data.boolean = 0;
		return 5;
	}
	cudl_err = CUDL_ERR_EXPECTED_BOOL_OR_NULL;
	return 0;
}

static size_t parse_number(char *input, struct cudl_value *value) {
	double number;
	size_t i, exponentStart;
	int exponent, otherExponent;
	int exponentUsed;
	exponentUsed = 0;
	number = 0;
	i = input[0] == '-';
	for (;; i++) {
		if (IS_DIGIT(input[i])) {
			number = number * 10 + (input[i] - '0');
			exponent++;
			continue;
		} else if (input[i] == '.') {
			exponent = 0;
			exponentUsed = 1;
			continue;
		}
		break;
	}
	if (input[0] == '-')
		number = 0 - number;
	if (!exponentUsed)
		exponent = 0;
	otherExponent = 0;
	if (input[i] == 'e' && (IS_DIGIT(input[i+1]) || (input[i+1] == '-' && IS_DIGIT(input[i+2])))) {
		i++;
		exponentStart = i;
		i += input[i] == '-';
		for (;; i++) {
			if (IS_DIGIT(input[i]))
				otherExponent = otherExponent * 10 + (input[i] - '0');
			else
				break;
		}
		if (input[exponentStart] == '-')
			otherExponent = 0 - otherExponent;
	}
	exponent = exponent - otherExponent;
	for (; exponent > 0; exponent--) {
		number /= 10;
	}
	for (; exponent < 0; exponent++) {
		number *= 10;
	}
	value->tag = CUDL_TAG_NUMBER;
	value->data.number = number;
	return i;
}

/* Convert UCS character to utf-8 bytes.
 * Return number of bytes generated.
 * Sets cudl_error on error.
 * Shamelessly lifted from https://github.com/cktan/tomc99 */
static size_t cudl_ucs_to_utf8(int64_t ucs, char utf8[6]) {
	if (
		0xd800 <= ucs && ucs <= 0xdfff ||
		0xfffe <= ucs && ucs <= 0xffff ||
		ucs < 0
	) {
		cudl_err = CUDL_ERR_UNRECOGNISED_UNICODE;
		return 0;
	}

	/* 0x00000000 - 0x0000007F:
	   0xxxxxxx
	*/
	if (ucs <= 0x7F) {
		utf8[0] = (unsigned char) ucs;
		return 1;
	}

	/* 0x00000080 - 0x000007FF:
	   110xxxxx 10xxxxxx
	*/
	if (ucs <= 0x000007FF) {
		utf8[0] = 0xc0 | (ucs >> 6);
		utf8[1] = 0x80 | (ucs & 0x3f);
		return 2;
	}

	/* 0x00000800 - 0x0000FFFF:
	   1110xxxx 10xxxxxx 10xxxxxx
	*/
	if (ucs <= 0x0000FFFF) {
		utf8[0] = 0xe0 | (ucs >> 12);
		utf8[1] = 0x80 | ((ucs >> 6) & 0x3f);
		utf8[2] = 0x80 | (ucs & 0x3f);
		return 3;
	}

	/* 0x00010000 - 0x001FFFFF:
	   11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
	*/
	if (ucs <= 0x001FFFFF) {
		utf8[0] = 0xf0 | (ucs >> 18);
		utf8[1] = 0x80 | ((ucs >> 12) & 0x3f);
		utf8[2] = 0x80 | ((ucs >> 6) & 0x3f);
		utf8[3] = 0x80 | (ucs & 0x3f);
		return 4;
	}

	/* 0x00200000 - 0x03FFFFFF:
	   111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
	*/
	if (ucs <= 0x03FFFFFF) {
		utf8[0] = 0xf8 | (ucs >> 24);
		utf8[1] = 0x80 | ((ucs >> 18) & 0x3f);
		utf8[2] = 0x80 | ((ucs >> 12) & 0x3f);
		utf8[3] = 0x80 | ((ucs >> 6) & 0x3f);
		utf8[4] = 0x80 | (ucs & 0x3f);
		return 5;
	}

	/* 0x04000000 - 0x7FFFFFFF:
	   1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
	*/
	if (ucs <= 0x7FFFFFFF) {
		utf8[0] = 0xfc | (ucs >> 30);
		utf8[1] = 0x80 | ((ucs >> 24) & 0x3f);
		utf8[2] = 0x80 | ((ucs >> 18) & 0x3f);
		utf8[3] = 0x80 | ((ucs >> 12) & 0x3f);
		utf8[4] = 0x80 | ((ucs >> 6) & 0x3f);
		utf8[5] = 0x80 | (ucs & 0x3f);
		return 6;
	}

	cudl_err = CUDL_ERR_UNRECOGNISED_UNICODE;
	return 0;
}

/* Parse a string starting after the opening quote.
 * Set string to be the contents of the string.
 * No memory is allocated if an error occurs. */
static size_t parse_quoted_string(char *input, char **string) {
	size_t length, capacity;
	char *original_input, *newstring;
	int64_t ucs;
	int ucs_length, i;

	length = 0;
	capacity = 32;
	original_input = input;
	if ((*string = malloc(capacity)) == NULL) {
		cudl_err = CUDL_ERR_OUT_OF_MEMORY;
		return 0;
	}
	for (;;) {
		if (*input == '\0') {
			cudl_err = CUDL_ERR_UNMATCHED_QUOTE;
			free(*string);
			return 0;
		}
		if (*input == '"') {
			if ((newstring = realloc(*string, length + 1)) == NULL) {
				cudl_err = CUDL_ERR_OUT_OF_MEMORY;
				free(*string);
				return 0;
			}
			*string = newstring;
			(*string)[length] = '\0';
			input++;
			return input - original_input;
		}
		if (length >= capacity) {
			if ((newstring = realloc(*string, capacity * 2)) == NULL) {
				cudl_err = CUDL_ERR_OUT_OF_MEMORY;
				free(*string);
				return 0;
			}
			*string = newstring;
			capacity *= 2;
		}
		if (*input == '\\') {
			input++;
			switch (*input) {
				case '\0':
					cudl_err = CUDL_ERR_EXPECTED_ESCAPE_SEQUENCE;
					free(*string);
					return 0;
				case 'b':
					(*string)[length++] = '\b';
					input++;
					break;
				case 't':
					(*string)[length++] = '\t';
					input++;
					break;
				case 'n':
					(*string)[length++] = '\n';
					input++;
					break;
				case 'r':
					(*string)[length++] = '\r';
					input++;
					break;
				case '"':
					(*string)[length++] = '"';
					input++;
					break;
				case '\\':
					(*string)[length++] = '\\';
					input++;
					break;
				case 'u':
				case 'U':
					ucs = 0;
					ucs_length = (*input == 'u') ? 4 : 8;
					input++;
					for (i = 0; i < ucs_length; i++) {
						if (input[i] == '\0') {
							cudl_err = CUDL_ERR_EXPECTED_ESCAPE_SEQUENCE;
							free(*string);
							return 0;
						}
						if ('0' <= input[i] && input[i] <= '9') {
							ucs = (ucs << 4) + (input[i] - '0');
						} else if ('a' <= input[i] && input[i] <= 'z') {
							ucs = (ucs << 4) + (input[i] - 'a' + 10);
						} else if ('A' <= input[i] && input[i] <= 'Z') {
							ucs = (ucs << 4) + (input[i] - 'A' + 10);
						} else {
							cudl_err = CUDL_ERR_EXPECTED_ESCAPE_SEQUENCE;
							free(*string);
							return 0;
						}
					}
					if (length + 6 > capacity) {
						if ((newstring = realloc(*string, capacity * 2)) == NULL) {
							cudl_err = CUDL_ERR_OUT_OF_MEMORY;
							free(*string);
							return 0;
						}
						*string = newstring;
						capacity *= 2;
					}
					length += cudl_ucs_to_utf8(ucs, (*string) + length);
					if (cudl_err) {
						free(*string);
						return 0;
					}
					input += ucs_length;
					break;
				default:
					(*string)[length++] = *input;
					input++;
					break;
			}
		} else {
			(*string)[length++] = *(input++);
		}
	}
}

static size_t parse_array(char *input, struct cudl_value *value) {
	size_t length, capacity;
	struct cudl_value *values, *newvalues;
	int i;
	char *original_input;

	original_input = input;
	value->tag = CUDL_TAG_ARRAY;
	length = 0;
	capacity = 8;
	if ((values = malloc(capacity * sizeof(struct cudl_value))) == NULL) {
		cudl_err = CUDL_ERR_OUT_OF_MEMORY;
		return 0;
	}

	STRIP_WHITESPACE(input);
	for (;;) {
		if (*input == '\0') {
			cudl_err = CUDL_ERR_UNMATCHED_BRACK;
			for (i = 0; i < length; i++)
				cudl_deinit_value(values[i]);
			free(values);
			return 0;
		} else if (*input == ']') {
			input++;
			values = realloc(values, length * sizeof(struct cudl_value));
			value->data.array.length = length;
			value->data.array.values = values;
			return input - original_input;
		}
		if (length >= capacity) {
			if ((newvalues = realloc(values, 2 * capacity * sizeof(struct cudl_value))) == NULL) {
				cudl_err = CUDL_ERR_OUT_OF_MEMORY;
				for (i = 0; i < length; i++)
					cudl_deinit_value(values[i]);
				free(values);
				return 0;
			}
			values = newvalues;
			capacity *= 2;
		}
		input += parse_value(input, values + length);
		if (cudl_err) {
			for (i = 0; i < length; i++)
				cudl_deinit_value(values[i]);
			free(values);
			return 0;
		}
		length++;
	}
}

static size_t parse_map_key(char *input, char **key) {
	char *original_input;
	switch (*input) {
		case '\0':
			cudl_err = CUDL_ERR_EXPECTED_MAP_KEY;
			return 0;
		case '"':
			input++;
			return parse_quoted_string(input, key) + 1;
		default:
			original_input = input;
			while (IS_KEY_CHAR(*input))
				input++;
			if (input == original_input) {
				cudl_err = CUDL_ERR_EXPECTED_MAP_KEY;
				return 0;
			}
			if ((*key = malloc(input - original_input + 1)) == NULL) {
				cudl_err = CUDL_ERR_OUT_OF_MEMORY;
				return 0;
			}
			memcpy(*key, original_input, input - original_input);
			(*key)[input - original_input] = '\0';
			return input - original_input;
	}
}

static size_t parse_map(char *input, struct cudl_value *value, char end_char) {
	char *original_input;
	int i;
	struct cudl_map_field *fields, *newfields;
	size_t length, capacity;

	original_input = input;
	value->tag = CUDL_TAG_MAP;
	length = 0;
	capacity = 8;
	if ((fields = malloc(capacity * sizeof(struct cudl_map_field))) == NULL) {
		cudl_err = CUDL_ERR_OUT_OF_MEMORY;
		return 0;
	}

	STRIP_WHITESPACE(input);
	for (;;) {
		if (*input == end_char) {
			input++;
			fields = realloc(fields, length * sizeof(struct cudl_map_field));
			value->data.map.length = length;
			value->data.map.fields = fields;
			return input - original_input;
		}
		if (*input == '\0') {
			cudl_err = CUDL_ERR_UNMATCHED_BRACE;
			for (i = 0; i < length; i++) {
				cudl_deinit_value(fields[i].value);
				free(fields[i].key);
			}
			free(fields);
			return 0;
		}
		if (length >= capacity) {
			if ((newfields = realloc(fields,  2 * capacity * sizeof(struct cudl_map_field))) == NULL) {
				cudl_err = CUDL_ERR_OUT_OF_MEMORY;
				for (i = 0; i < length; i++) {
					cudl_deinit_value(fields[i].value);
					free(fields[i].key);
				}
				free(fields);
				return 0;
			}
			fields = newfields;
			capacity *= 2;
		}
		input += parse_map_key(input, &fields[length].key);
		if (cudl_err) {
			for (i = 0; i < length; i++) {
				cudl_deinit_value(fields[i].value);
				free(fields[i].key);
			}
			free(fields);
			return 0;
		}
		STRIP_WHITESPACE(input);
		if (*input != ':') {
			cudl_err = CUDL_ERR_EXPECTED_COLON;
			for (i = 0; i < length; i++) {
				cudl_deinit_value(fields[i].value);
				free(fields[i].key);
			}
			free(fields[length].key);
			free(fields);
			return 0;
		}
		input++;
		STRIP_WHITESPACE(input);
		input += parse_value(input, &fields[length].value);
		if (cudl_err) {
			for (i = 0; i < length; i++) {
				cudl_deinit_value(fields[i].value);
				free(fields[i].key);
			}
			free(fields[length].key);
			free(fields);
			return 0;
		}
		length++;
	}
}

static size_t _parse_value(char *input, struct cudl_value *value) {
	if (*input == '%')
		return parse_bool_or_null(++input, value) + 1;
	if (*input == '[')
		return parse_array(++input, value) + 1;
	if (*input == '{')
		return parse_map(++input, value, '}') + 1;
	if (*input == '"') {
		value->tag = CUDL_TAG_STRING;
		return parse_quoted_string(++input, &value->data.string) + 1;
	}
	if (IS_DIGIT(*input) || *input == '-')
		return parse_number(input, value);
	cudl_err = CUDL_ERR_UNRECOGNISED_VALUE;
	return 0;
}

static size_t parse_value(char *input, struct cudl_value *value) {
	char *original_input;
	original_input = input;
	input += _parse_value(input, value);
	STRIP_WHITESPACE(input);
	return input - original_input;
}

void cudl_parse_from_file(FILE *file, struct cudl_value *value) {
	char *input, *original_input;
	if ((original_input = input = fread_all(file)) == NULL) {
		if (ferror(file))
			cudl_err = CUDL_ERR_READING;
		else
			cudl_err = CUDL_ERR_OUT_OF_MEMORY;
		return;
	}
	input += cudl_parse(input, value);
	if (*input != '\0')
		cudl_deinit_value(*value);
	free(original_input);
}

size_t cudl_parse(char *input, struct cudl_value *value) {
	STRIP_WHITESPACE(input);
	return parse_value(input, value);
}