dev/c/tex_parser/tex_parser.c

#include "tex_parser.h"

#include <string.h>
#include <ctype.h>
#include "tex_parser_tags.h"
#include "tex_parser_const.h"
#include "zalloc_ext.h"
#include "zalloc.h"
#include "c_const.h"
#include "xerror.h"
#include "zerror.h"

/* only for debug */
#define __TEX_PARSER_DEBUG
#include <stdio.h>


void tex_parse(const char *source, size_t len, struct tex_struct_s **tex_struct, struct zerror_s **error)
{
	if (!len)
		len = strlen(source);

	enum where_e
	{
		IN_UNDEF = 0, // undefined place
		IN_COMMENT, // any comment
		IN_TAG, // any tag
		IN_TAG_UNKNOWN, // unknown tag
		IN_SPACE, // space, \tag  { for example
		IN_TAGPARM, // in \tag{} curly braces
		IN_TAG_BEGIN,
		IN_TAG_END,
		IN_TAG_TABULARNEWLINE,
		IN_TABLE,
		IN_TABLE_PARAMS, // column properties devided by '>'
		IN_TABLE_TAG_HLINE,
		IN_TABLE_TAG_MULTICOLUMN,

		IN_TABLE_COMMENT,
		IN_TABLE_TAG,
		IN_TABLE_TAG_UNKNOWN,
		IN_TABLE_SPACE,
		IN_TABLE_TAGPARM,
		IN_TABLE_TAG_CLINE,

		IN_TABLE_TAG_ENDFOOT,
		IN_TABLE_TAG_ENDHEAD,
		IN_TABLE_TAG_HSPACE,

		IN_TABLE_TAG_MULTIROW,
		IN_FORMULA, // in $...$

	} where_stack[MAX_TEX_STACK_LEVEL] = {IN_UNDEF};

	char *tag = NULL, *param = NULL;
	size_t i = 0;
	size_t tex_level = 0;

	for (i = 0; i < len; i++) {

		/* stack checking */
		if (tex_level + 1 == MAX_TEX_STACK_LEVEL) {
			z_set_error(error, TEX_PARSER_DOMAIN, TEX_PARSER_ERROR_STACK,
			            "tex_parse(): stack overflow, symbol %c(0x%2.2x) at position %d",
			            source[i], (u_int8_t)source[i], i);
			return;
		}

		/* read one character from input stream */
		char c = source[i];
		#ifdef __TEX_PARSER_DEBUG
			printf("tex_level = %lu, c = %c\n", tex_level, c);
		#endif

		/* looking at where_stack and encountered character */
		switch (where_stack[tex_level]) {
		case IN_UNDEF:
			if (c == '\\') {
				zclear(&tag);
				where_stack[++tex_level] = IN_TAG;


			} else if (isspace(c)) {
				where_stack[++tex_level] = IN_SPACE;

			} else if (c == '%') {
				where_stack[++tex_level] = IN_COMMENT;

			} else if (isgraph(c) || c < 0) {

			} else {
				z_set_error(error, TEX_PARSER_DOMAIN, TEX_PARSER_ERROR_UNEXPECTED_SYMBOL,
				            "tex_parse(): IN_UNDEF unexpected symbol %c(0x%2.2x) at position %d",
				            source[i], (u_int8_t)source[i], i);
				return;
			}
		break;

		case IN_COMMENT:
			if (c == '\r' || c == '\n') {
				where_stack[tex_level--] = IN_UNDEF;
			}
		break;

		case IN_TAG:
			if (isalnum(c)) {
				tag = zalloc_append8_str(tag, c);

			} else if (!tag && (c == '\\' || c == ','
			           || c == '%' || c == '_')) {
				where_stack[tex_level--] = IN_UNDEF;

			} else if (!tag) {
				z_set_error(error, TEX_PARSER_DOMAIN, TEX_PARSER_ERROR_UNEXPECTED_SYMBOL,
				            "tex_parse(): IN_TAG empty tag, symbol %c(0x%2.2x) at position %d",
				            source[i], (u_int8_t)source[i], i);
				return;

			} else {
				if (!strcmp(tag, TEX_TAG_BEGIN))
					where_stack[tex_level] = IN_TAG_BEGIN;

				else if (!strcmp(tag, TEX_TAG_END))
					where_stack[tex_level] = IN_TAG_END;

				else if (!strcmp(tag, TEX_TAG_TABULARNEWLINE))
					where_stack[tex_level] = IN_TAG_TABULARNEWLINE;

				else
					where_stack[tex_level] = IN_TAG_UNKNOWN;

				#ifdef __TEX_PARSER_DEBUG
					printf("\\%s\n", tag);
				#endif
				zclear(&tag);

				i--;
			}
		break;

		case IN_SPACE:
			if (!isspace(c)) {
				where_stack[tex_level--] = IN_UNDEF;
				i--;
			}
		break;

		case IN_TAGPARM:
			if (c == '{' || c == '[') { // tag params
				zclear(&param);
				where_stack[++tex_level] = IN_TAGPARM;

			} else if (c == '}' || c == ']') { // end tag params
				#ifdef __TEX_PARSER_DEBUG
				if (param)
					printf("IN_TAGPARM: {%s}\n", param);
				#endif
				where_stack[tex_level--] = IN_UNDEF;

				if (tex_level && where_stack[tex_level - 1] == IN_TAG_BEGIN
				    && !strcmp(param, TEX_PARAM_TABLE))
					where_stack[tex_level++] = IN_TABLE;//??????????????????????

			} else if (c == '\\') { // new tag
				zclear(&tag);
				where_stack[++tex_level] = IN_TAG;

			} else if (c == '%') {
				where_stack[++tex_level] = IN_COMMENT;

			} else if (isgraph(c) || isspace(c) || c < 0) {
				param = zalloc_append8_str(param, c);

			} else {
				z_set_error(error, TEX_PARSER_DOMAIN, TEX_PARSER_ERROR_UNEXPECTED_SYMBOL,
				            "tex_parse(): IN_TAGPARM unexpected symbol %c(0x%2.2x) at position %d",
				            source[i], (u_int8_t)source[i], i);
				return;
			}
		break;

		case IN_TAG_UNKNOWN:
		case IN_TAG_BEGIN:
		case IN_TAG_END:
		case IN_TAG_TABULARNEWLINE:
			if (c == '{' || c == '[') {
				zclear(&param);
				where_stack[++tex_level] = IN_TAGPARM;

			} else if (c == '}' || c == ']') {
				#ifdef __TEX_PARSER_DEBUG
				if (param)
					printf("IN_TAGPARM: {%s}\n", param);
				#endif
				where_stack[tex_level--] = IN_UNDEF;
				i--;

			} else if (c == '\\') {
				zclear(&tag);
				where_stack[tex_level] = IN_TAG;

			} else if (isspace(c)) {
				where_stack[++tex_level] = IN_SPACE;

			} else if (c == '%')
				where_stack[++tex_level] = IN_COMMENT;

			else if (isgraph(c) || c < 0) {

			} else {
				z_set_error(error, TEX_PARSER_DOMAIN, TEX_PARSER_ERROR_UNEXPECTED_SYMBOL,
				            "tex_parse(): IN_%d unexpected symbol %c(0x%2.2x) at position %d",
				            where_stack[tex_level], source[i], (u_int8_t)source[i], i);
				return;
			}
		break;

		case IN_FORMULA:

		break;

		default:
			z_set_error(error, TEX_PARSER_DOMAIN, TEX_PARSER_ERROR_PLACE_UNKNOWN,
			            "tex_parse(): IN_%d unknown place, symbol %c(0x%2.2x) at position %d",
			            where_stack[tex_level], source[i], (u_int8_t)source[i], i);
			return;
		break;
		}
	}
}