table.c

一个很有名的浏览器

	while (span--) {
		struct table_column *column = &table->columns[table->columns_count++];

		column->align = align;
		column->valign = valign;
		column->width = width;
		column->group = group;
		group = 0;
	}
}

static void
set_td_width(struct table *table, int col, int width, int force)
{
	if (col >= table->cols_x_count) {
		int n = table->cols_x_count;
		int i;
		int *new_cols_x;

		n = smart_raise(col + 1, n, sizeof(*new_cols_x), SMART_RAISE_LIMIT);
		if (!n && table->cols_x_count) return;
		if (!n) n = col + 1;

		new_cols_x = mem_realloc(table->cols_x, n * sizeof(*new_cols_x));
		if (!new_cols_x) return;

		for (i = table->cols_x_count; i < n; i++)
			new_cols_x[i] = WIDTH_AUTO;
		table->cols_x_count = n;
		table->cols_x = new_cols_x;
	}

	if (force || table->cols_x[col] == WIDTH_AUTO) {
		table->cols_x[col] = width;
		return;
	}

	if (width == WIDTH_AUTO) return;

	if (width < 0 && table->cols_x[col] >= 0) {
		table->cols_x[col] = width;
		return;
	}

	if (width >= 0 && table->cols_x[col] < 0) return;
	table->cols_x[col] = (table->cols_x[col] + width) >> 1;
}

static unsigned char *
skip_table(unsigned char *html, unsigned char *eof)
{
	int level = 1;

	while (1) {
		unsigned char *name;
		int namelen;

		while (html < eof
		       && (*html != '<'
			    || parse_element(html, eof, &name, &namelen, NULL,
					    &html)))
			html++;

		if (html >= eof) return eof;

		if (!strlcasecmp(name, namelen, "TABLE", 5)) {
			level++;
		} else if (!strlcasecmp(name, namelen, "/TABLE", 6)) {
			level--;
			if (!level) return html;
		}
	}
}

struct table *
parse_table(unsigned char *html, unsigned char *eof, unsigned char **end,
	    unsigned char *attr, int sh)
{
	struct table *table;
	struct table_cell *cell;
	unsigned char *t_name, *t_attr, *en;
	unsigned char *l_fragment_id = NULL;
	color_t last_bgcolor;
	int t_namelen;
	int in_cell = 0;
	int l_al = ALIGN_LEFT;
	int l_val = VALIGN_MIDDLE;
	int colspan, rowspan;
	int group = 0;
	int i, j, k;
	int c_al = ALIGN_TR, c_val = VALIGN_TR, c_width = WIDTH_AUTO, c_span = 0;
	int cols, rows;
	int col = 0, row = -1;
	int maxj;

	*end = html;

	table = new_table();
	if (!table) return NULL;

	parse_table_attributes(table, attr, sh);
	last_bgcolor = table->bgcolor;

se:
	en = html;

see:
	html = en;
	if (!in_cell) {
		add_table_bad_html_start(table, html);
	}

	while (html < eof && *html != '<') html++;

	if (html >= eof) {
		if (in_cell) CELL(table, col, row)->end = html;
		add_table_bad_html_end(table, html);
		goto scan_done;
	}

	if (html + 2 <= eof && (html[1] == '!' || html[1] == '?')) {
		html = skip_comment(html, eof);
		goto se;
	}

	if (parse_element(html, eof, &t_name, &t_namelen, &t_attr, &en)) {
		html++;
		goto se;
	}

	if (!strlcasecmp(t_name, t_namelen, "TABLE", 5)) {
		en = skip_table(en, eof);
		goto see;
	}

	if (!strlcasecmp(t_name, t_namelen, "/TABLE", 6)) {
		if (c_span) new_columns(table, c_span, c_width, c_al, c_val, 1);
		if (in_cell) CELL(table, col, row)->end = html;

		add_table_bad_html_end(table, html);

		goto scan_done;
	}

	if (!strlcasecmp(t_name, t_namelen, "CAPTION", 7)) {
		add_table_bad_html_end(table, html);
		if (!table->caption.start)
			table->caption.start = html;
		goto see;
	}

	if (!strlcasecmp(t_name, t_namelen, "/CAPTION", 8)) {
		if (table->caption.start && !table->caption.end)
			table->caption.end = html;
		goto see;
	}

	if (!strlcasecmp(t_name, t_namelen, "COLGROUP", 8)) {
		if (c_span) new_columns(table, c_span, c_width, c_al, c_val, 1);

		add_table_bad_html_end(table, html);

		c_al = ALIGN_TR;
		c_val = VALIGN_TR;
		c_width = WIDTH_AUTO;
		get_align(t_attr, &c_al);
		get_valign(t_attr, &c_val);
		get_column_width(t_attr, &c_width, sh);
		c_span = get_num(t_attr, "span");
		if (c_span == -1) c_span = 1;
		else if (c_span > HTML_MAX_COLSPAN) c_span = HTML_MAX_COLSPAN;
		goto see;
	}

	if (!strlcasecmp(t_name, t_namelen, "/COLGROUP", 9)) {
		if (c_span) new_columns(table, c_span, c_width, c_al, c_val, 1);

		add_table_bad_html_end(table, html);

		c_span = 0;
		c_al = ALIGN_TR;
		c_val = VALIGN_TR;
		c_width = WIDTH_AUTO;
		goto see;
	}

	if (!strlcasecmp(t_name, t_namelen, "COL", 3)) {
		int sp, width, al, val;

		add_table_bad_html_end(table, html);

		sp = get_num(t_attr, "span");
		if (sp == -1) sp = 1;
		else if (sp > HTML_MAX_COLSPAN) sp = HTML_MAX_COLSPAN;

		width = c_width;
		al = c_al;
		val = c_val;
		get_align(t_attr, &al);
		get_valign(t_attr, &val);
		get_column_width(t_attr, &width, sh);
		new_columns(table, sp, width, al, val, !!c_span);
		c_span = 0;
		goto see;
	}

	/* /TR /TD /TH */
	if (t_namelen == 3
	    && t_name[0] == '/'
	    && toupper(t_name[1]) == 'T') {
		unsigned char c = toupper(t_name[2]);

		if (c == 'R' || c == 'D' || c == 'H') {
			if (c_span)
				new_columns(table, c_span, c_width, c_al, c_val, 1);

			if (in_cell) {
				CELL(table, col, row)->end = html;
				in_cell = 0;
			}

			add_table_bad_html_end(table, html);
		}
	}

	/* All following tags have T as first letter. */
	if (toupper(t_name[0]) != 'T') goto see;

	/* TR */
	if (t_namelen == 2 && toupper(t_name[1]) == 'R') {
		if (c_span) new_columns(table, c_span, c_width, c_al, c_val, 1);

		if (in_cell) {
			CELL(table, col, row)->end = html;
			in_cell = 0;
		}

		add_table_bad_html_end(table, html);

		if (group) group--;
		l_al = ALIGN_LEFT;
		l_val = VALIGN_MIDDLE;
		last_bgcolor = table->bgcolor;
		get_align(t_attr, &l_al);
		get_valign(t_attr, &l_val);
		get_bgcolor(t_attr, &last_bgcolor);
		mem_free_set(&l_fragment_id, get_attr_val(t_attr, "id"));
		row++;
		col = 0;
		goto see;
	}

	/* THEAD TBODY TFOOT */
	if (t_namelen == 5
	    && ((!strncasecmp(&t_name[1], "HEAD", 4)) ||
		(!strncasecmp(&t_name[1], "BODY", 4)) ||
		(!strncasecmp(&t_name[1], "FOOT", 4)))) {
		if (c_span) new_columns(table, c_span, c_width, c_al, c_val, 1);

		add_table_bad_html_end(table, html);

		group = 2;
	}

	/* TD TH */
	if (t_namelen != 2
	    || (toupper(t_name[1]) != 'D'
		&& toupper(t_name[1]) != 'H'))
		goto see;

	if (c_span) new_columns(table, c_span, c_width, c_al, c_val, 1);

	add_table_bad_html_end(table, html);

	if (in_cell) {
		CELL(table, col, row)->end = html;
		in_cell = 0;
	}

	if (row == -1) {
		row = 0;
		col = 0;
	}

	for (;;col++) {
		cell = new_cell(table, col, row);
		if (!cell) goto see;

		if (!cell->is_used) break;
		if (cell->colspan == -1) goto see;
	}

	in_cell = 1;

	cell->col = col;
	cell->row = row;
	cell->is_used = 1;
	cell->start = en;

	cell->align = l_al;
	cell->valign = l_val;
	cell->fragment_id = get_attr_val(t_attr, "id");
	if (!cell->fragment_id && l_fragment_id) {
		cell->fragment_id = l_fragment_id;
		l_fragment_id = NULL;
	}

	cell->is_header = (toupper(t_name[1]) == 'H');
	if (cell->is_header) cell->align = ALIGN_CENTER;

	if (group == 1) cell->is_group = 1;

	if (col < table->columns_count) {
		if (table->columns[col].align != ALIGN_TR)
			cell->align = table->columns[col].align;
		if (table->columns[col].valign != VALIGN_TR)
			cell->valign = table->columns[col].valign;
	}

	cell->bgcolor = last_bgcolor;

	get_align(t_attr, &cell->align);
	get_valign(t_attr, &cell->valign);
	get_bgcolor(t_attr, &cell->bgcolor);

	colspan = get_num(t_attr, "colspan");
	if (colspan == -1) colspan = 1;
	else if (!colspan) colspan = -1;
	else if (colspan > HTML_MAX_COLSPAN) colspan = HTML_MAX_COLSPAN;

	rowspan = get_num(t_attr, "rowspan");
	if (rowspan == -1) rowspan = 1;
	else if (!rowspan) rowspan = -1;
	else if (rowspan > HTML_MAX_ROWSPAN) rowspan = HTML_MAX_ROWSPAN;

	cell->colspan = colspan;
	cell->rowspan = rowspan;

	if (colspan == 1) {
		int width = WIDTH_AUTO;

		get_column_width(t_attr, &width, sh);
		if (width != WIDTH_AUTO)
			set_td_width(table, col, width, 0);
	}

	cols = table->cols;
	for (i = 1; colspan != -1 ? i < colspan : i < cols; i++) {
		struct table_cell *span_cell = new_cell(table, col + i, row);

		if (!span_cell)
			goto abort;

		if (span_cell->is_used) {
			colspan = i;
			for (k = 0; k < i; k++)
				CELL(table, col + k, row)->colspan = colspan;
			break;
		}

		span_cell->is_used = span_cell->is_spanned = 1;
		span_cell->rowspan = rowspan;
		span_cell->colspan = colspan;
		span_cell->col = col;
		span_cell->row = row;
	}

	rows = table->rows;
	maxj = rowspan != -1 ? rowspan : rows;
	/* Out of memory prevention, limit allocated memory to HTML_MAX_CELLS_MEMORY.
	 * Not perfect but better than nothing. */
	if (maxj * i > HTML_MAX_CELLS_MEMORY / sizeof(*cell))
		goto abort;

	for (j = 1; j < maxj; j++) {
		for (k = 0; k < i; k++) {
			struct table_cell *span_cell = new_cell(table, col + k, row + j);

			if (!span_cell)
				goto abort;

			if (span_cell->is_used) {
				int l, m;

				if (span_cell->col == col
				    && span_cell->row == row)
					continue;

				for (l = 0; l < k; l++)
					memset(CELL(table, col + l, row + j), 0,
					       sizeof(*span_cell));

				rowspan = j;

				for (l = 0; l < i; l++)
					for (m = 0; m < j; m++)
						CELL(table, col + l, row + m)->rowspan = j;
				goto see;
			}

			span_cell->is_used = span_cell->is_spanned = 1;
			span_cell->rowspan = rowspan;
			span_cell->colspan = colspan;
			span_cell->col = col;
			span_cell->row = row;
		}
	}

	goto see;

scan_done:
	*end = html;

	mem_free_if(l_fragment_id);

	for (col = 0; col < table->cols; col++) for (row = 0; row < table->rows; row++) {
		struct table_cell *cell = CELL(table, col, row);

		if (!cell->is_spanned) {
			if (cell->colspan == -1) cell->colspan = table->cols - col;
			if (cell->rowspan == -1) cell->rowspan = table->rows - row;
		}
	}

	if (table->rows) {
		table->rows_heights = mem_calloc(table->rows, sizeof(*table->rows_heights));
		if (!table->rows_heights)
			goto abort;
	} else {
		table->rows_heights = NULL;
	}

	for (col = 0; col < table->columns_count; col++)
		if (table->columns[col].width != WIDTH_AUTO)
			set_td_width(table, col, table->columns[col].width, 1);
	set_td_width(table, table->cols, WIDTH_AUTO, 0);

	return table;

abort:
	*end = eof;
	free_table(table);
	return NULL;
}