Renamed libs

Cleaner and more flexible.

7 files changed, 954 insertions, 0 deletions

diff --git a/libs/libtxt/Makefile b/libs/libtxt/Makefile
new file mode 100644
index 0000000..93a7993
--- /dev/null
+++ b/libs/libtxt/Makefile
@@ -0,0 +1,20 @@
+# MIT License, Copyright (c) 2020 Marvin Borner
+
+COBJS = keymap.o #xml.o html.o
+CC = ccache ../../cross/opt/bin/i686-elf-gcc
+LD = ccache ../../cross/opt/bin/i686-elf-ld
+AR = ccache ../../cross/opt/bin/i686-elf-ar
+
+CFLAGS = $(CFLAGS_DEFAULT) -I../ -I../libc/inc/ -Duserspace
+
+all: libtxt
+
+%.o: %.c
+	@$(CC) -c $(CFLAGS) $< -o $@
+
+libtxt: $(COBJS)
+	@mkdir -p ../../build/
+	@$(AR) rcs ../../build/libtxt.a $+
+
+clean:
+	@find . -name "*.o" -type f -delete
diff --git a/libs/libtxt/html.c b/libs/libtxt/html.c
new file mode 100644
index 0000000..9295e17
--- /dev/null
+++ b/libs/libtxt/html.c
@@ -0,0 +1,238 @@
+// MIT License, Copyright (c) 2020 Marvin Borner
+// HTML parsing is mainly based on the XML parser
+
+#include <assert.h>
+#include <libgui/gui.h>
+#include <libtxt/html.h>
+#include <libtxt/xml.h>
+#include <list.h>
+#include <mem.h>
+#include <print.h>
+#include <str.h>
+
+static int is_self_closing(const char *tag)
+{
+	const char *void_elements[] = { "area",	 "base", "br",	 "col",	  "embed",  "hr",    "img",
+					"input", "link", "meta", "param", "source", "track", "wbr" };
+
+	for (u32 i = 0; i < sizeof(void_elements) / sizeof(void_elements[0]); ++i) {
+		if (!strcmp(void_elements[i], tag))
+			return 1;
+	}
+	return 0;
+}
+
+static char *normalize_tag_name(char *tag)
+{
+	for (char *p = tag; *p; ++p)
+		*p = *p > 0x40 && *p < 0x5b ? *p | 0x60 : *p;
+	return tag;
+}
+
+static struct dom *new_object(const char *tag, struct dom *parent)
+{
+	struct dom *object = malloc(sizeof(*object));
+	object->tag = strdup(tag);
+	object->parent = parent;
+	object->content = NULL;
+	object->children = list_new();
+	return object;
+}
+
+static void print_dom(struct dom *dom, u32 level)
+{
+	struct node *iterator = dom->children->head;
+	while (iterator != NULL) {
+		struct dom *obj = iterator->data;
+		for (u32 i = 0; i < level; i++)
+			print("\t");
+		printf("'%s': '%s'\n", obj->tag, obj->content ? obj->content : "");
+		if (obj->children->head)
+			print_dom(obj, level + 1);
+		iterator = iterator->next;
+	}
+}
+
+static struct dom *generate_dom(char *data, u32 length)
+{
+	struct xml_token tokens[128];
+	struct xml parser;
+	xml_init(&parser);
+	void *buffer = data;
+	enum xml_error error = xml_parse(&parser, buffer, length, tokens, 128);
+
+	if (error != XML_SUCCESS && error != XML_ERROR_BUFFERDRY) {
+		printf("XML parse error: %d\n", err);
+		printf("DATA: '%s'\n", data);
+		return NULL;
+	}
+
+	struct dom *root = new_object("root", NULL);
+	struct dom *current = root;
+
+	static char name[256] = { 0 };
+	for (u32 i = 0; i < parser.ntokens; i++) {
+		const struct xml_token *token = tokens + i;
+		name[0] = '\0';
+		switch (token->type) {
+		case XML_START_TAG:
+			memcpy(&name, (u8 *)buffer + token->start_pos,
+			       token->end_pos - token->start_pos);
+			name[token->end_pos - token->start_pos] = '\0';
+			normalize_tag_name(name);
+			current = new_object(name, current);
+			printf("Adding %s to %s\n", current->tag, current->parent->tag);
+			list_add(current->parent->children, current);
+			if (is_self_closing(name))
+				current = current->parent;
+			break;
+		case XML_END_TAG:
+			memcpy(&name, (u8 *)buffer + token->start_pos,
+			       token->end_pos - token->start_pos);
+			name[token->end_pos - token->start_pos] = '\0';
+			normalize_tag_name(name);
+
+			if (is_self_closing(name))
+				break;
+
+			if (!current || !current->parent || strcmp(name, current->tag))
+				return NULL;
+
+			current = current->parent;
+			break;
+		case XML_CHARACTER:
+			if (!current)
+				continue;
+
+			if (token->end_pos == token->start_pos + 2) {
+				const char *ptr = (char *)buffer + token->start_pos;
+
+				if (ptr[0] == '\r' && ptr[1] == '\n')
+					continue;
+			}
+			memcpy(&name, (u8 *)buffer + token->start_pos,
+			       token->end_pos - token->start_pos);
+			name[token->end_pos - token->start_pos] = '\0';
+			char *clean_name = name;
+			for (char *p = name; *p; p++) {
+				if (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t') {
+					clean_name++;
+				} else {
+					break;
+				}
+			}
+			if (!strlen(clean_name))
+				break;
+			current->content = strdup(clean_name);
+			break;
+		default:
+			break;
+		}
+
+		i += token->size;
+	}
+
+	assert(root);
+	print("GENERATED!\n");
+	print_dom(root, 0);
+	return root;
+}
+
+static struct html_element *new_html_element(struct element *container, struct dom *dom)
+{
+	struct html_element *elem = malloc(sizeof(*elem));
+	elem->x_offset = 0;
+	elem->y_offset = 0;
+	elem->dom = dom;
+	elem->obj = container;
+	return elem;
+}
+
+// TODO: Better structure?
+// TODO: Less code duplication (e.g. for headings)
+#define CMP(tag, tag_string) (!strcmp((tag), (tag_string)))
+static struct html_element *render_object(struct html_element *container, struct dom *dom)
+{
+	char *tag = dom->tag;
+
+	assert(container);
+	if (CMP(tag, "html")) {
+		struct element *obj =
+			gui_add_container(container->obj, 0, 0, 100, 100, COLOR_WHITE);
+		return new_html_element(obj, dom);
+	} else if (CMP(tag, "body")) {
+		struct element *obj =
+			gui_add_container(container->obj, 0, 0, 100, 100, COLOR_WHITE);
+		return new_html_element(obj, dom);
+	} else if (CMP(tag, "h1")) {
+		struct element *obj =
+			gui_add_label(container->obj, container->x_offset, container->y_offset,
+				      FONT_32, dom->content, COLOR_WHITE, COLOR_BLACK);
+		container->x_offset = 0;
+		container->y_offset += obj->ctx->size.y;
+		return new_html_element(obj, dom);
+	} else if (CMP(tag, "h2")) {
+		struct element *obj =
+			gui_add_label(container->obj, container->x_offset, container->y_offset,
+				      FONT_24, dom->content, COLOR_WHITE, COLOR_BLACK);
+		container->x_offset = 0;
+		container->y_offset += obj->ctx->size.y;
+		return new_html_element(obj, dom);
+	} else if (CMP(tag, "h3")) {
+		struct element *obj =
+			gui_add_label(container->obj, container->x_offset, container->y_offset,
+				      FONT_16, dom->content, COLOR_WHITE, COLOR_BLACK);
+		container->x_offset = 0;
+		container->y_offset += obj->ctx->size.y;
+		return new_html_element(obj, dom);
+	} else if (CMP(tag, "p")) {
+		struct element *obj =
+			gui_add_label(container->obj, container->x_offset, container->y_offset,
+				      FONT_16, dom->content, COLOR_WHITE, COLOR_BLACK);
+		container->x_offset = 0;
+		container->y_offset += obj->ctx->size.y;
+		return new_html_element(obj, dom);
+	} else if (CMP(tag, "hr")) {
+		gfx_draw_rectangle(container->obj->ctx,
+				   vec2(container->x_offset, container->y_offset),
+				   vec2(container->obj->ctx->size.x - container->x_offset,
+					container->y_offset + 2),
+				   COLOR_BLACK);
+		container->x_offset = 0;
+		container->y_offset += 2;
+		return container;
+	} else if (CMP(tag, "head") || CMP(tag, "meta") || CMP(tag, "title")) {
+		return container;
+	} else {
+		printf("UNKNOWN %s\n", tag);
+		if (dom->content && strlen(dom->content) > 0) {
+			struct element *obj = gui_add_label(container->obj, container->x_offset,
+							    container->y_offset, FONT_16,
+							    dom->content, COLOR_WHITE, COLOR_BLACK);
+			container->x_offset = 0;
+			container->y_offset += obj->ctx->size.y;
+			return new_html_element(obj, dom);
+		}
+		return container;
+	}
+}
+
+int html_render_dom(struct html_element *container, struct dom *dom)
+{
+	struct node *iterator = dom->children->head;
+	while (iterator != NULL) {
+		struct dom *obj = iterator->data;
+		struct html_element *rendered = render_object(container, obj);
+		if (obj->children->head && rendered)
+			html_render_dom(rendered, obj);
+		iterator = iterator->next;
+	}
+	return 1;
+}
+
+int html_render(struct element *container, char *data, u32 length)
+{
+	struct dom *dom = generate_dom(data, length);
+	struct html_element *obj = new_html_element(container, dom);
+	return dom && obj && html_render_dom(obj, dom);
+}
diff --git a/libs/libtxt/html.h b/libs/libtxt/html.h
new file mode 100644
index 0000000..c1b29f2
--- /dev/null
+++ b/libs/libtxt/html.h
@@ -0,0 +1,26 @@
+// MIT License, Copyright (c) 2020 Marvin Borner
+// HTML parsing is mainly based on the XML parser
+
+#ifndef HTML_H
+#define HTML_H
+
+#include <def.h>
+#include <list.h>
+
+struct dom {
+	char *tag;
+	char *content;
+	struct dom *parent;
+	struct list *children;
+};
+
+struct html_element {
+	u32 x_offset;
+	u32 y_offset;
+	struct dom *dom;
+	struct element *obj;
+};
+
+int html_render(struct element *container, char *data, u32 length);
+
+#endif
diff --git a/libs/libtxt/keymap.c b/libs/libtxt/keymap.c
new file mode 100644
index 0000000..67054f6
--- /dev/null
+++ b/libs/libtxt/keymap.c
@@ -0,0 +1,89 @@
+// MIT License, Copyright (c) 2020 Marvin Borner
+
+#include <def.h>
+#include <libtxt/keymap.h>
+#include <mem.h>
+#include <print.h>
+#include <sys.h>
+
+static void map(struct keymap *keymap, int line, char ch, int index)
+{
+	switch (line) {
+	case 0:
+		keymap->map[index] = ch;
+		break;
+	case 1:
+		keymap->shift_map[index] = ch;
+		break;
+	case 2:
+		keymap->alt_map[index] = ch;
+		break;
+	default:
+		break;
+	}
+}
+
+// Very ugly code but it should work for now
+struct keymap *keymap_parse(const char *path)
+{
+	char *keymap_src = sread(path);
+	if (!keymap_src)
+		return NULL;
+	struct keymap *keymap = malloc(sizeof(*keymap));
+
+	int index = 0;
+	int ch_index = 0;
+	char ch;
+	int escaped = 0;
+	int line = 0;
+	int skip = 0;
+	while ((ch = keymap_src[index]) != '\0' || escaped) {
+		if (ch == ' ' && !skip) {
+			skip = 1;
+			index++;
+			continue;
+		} else if (ch == '\n') {
+			ch_index = 0;
+			index++;
+			line++;
+			continue;
+		} else if (ch == '\\' && !escaped) {
+			escaped = 1;
+			index++;
+			continue;
+		}
+		skip = 0;
+
+		if (ch == ' ' && !escaped)
+			ch = 0;
+
+		ch_index++;
+		if (escaped) {
+			switch (ch) {
+			case 'b':
+				ch = '\b';
+				break;
+			case 't':
+				ch = '\t';
+				break;
+			case 'n':
+				ch = '\n';
+				break;
+			case '\\':
+				ch = '\\';
+				break;
+			case ' ':
+				ch = ' ';
+				break;
+			default:
+				print("Unknown escape!\n");
+			}
+			escaped = 0;
+		}
+
+		map(keymap, line, ch, ch_index);
+		index++;
+	}
+
+	return keymap;
+}
diff --git a/libs/libtxt/keymap.h b/libs/libtxt/keymap.h
new file mode 100644
index 0000000..9f1966e
--- /dev/null
+++ b/libs/libtxt/keymap.h
@@ -0,0 +1,16 @@
+// MIT License, Copyright (c) 2020 Marvin Borner
+
+#ifndef KEYMAP_H
+#define KEYMAP_H
+
+#define KEYMAP_LENGTH 90
+
+struct keymap {
+	char map[KEYMAP_LENGTH];
+	char shift_map[KEYMAP_LENGTH];
+	char alt_map[KEYMAP_LENGTH];
+};
+
+struct keymap *keymap_parse(const char *path);
+
+#endif
diff --git a/libs/libtxt/xml.c b/libs/libtxt/xml.c
new file mode 100644
index 0000000..9a5fd76
--- /dev/null
+++ b/libs/libtxt/xml.c
@@ -0,0 +1,514 @@
+// Inspired by sxml (capmar)
+// MIT License, Copyright (c) 2020 Marvin Borner
+
+#include <assert.h>
+#include <libtxt/xml.h>
+#include <mem.h>
+#include <str.h>
+
+static const char *str_findchr(const char *start, const char *end, int c)
+{
+	const char *it;
+
+	assert(start <= end);
+	assert(0 <= c && c <= 127);
+
+	it = (const char *)memchr((void *)start, c, end - start);
+	return (it != NULL) ? it : end;
+}
+
+static const char *str_findstr(const char *start, const char *end, const char *needle)
+{
+	u32 needlelen;
+	int first;
+	assert(start <= end);
+
+	needlelen = strlen(needle);
+	assert(0 < needlelen);
+	first = (u8)needle[0];
+
+	while (start + needlelen <= end) {
+		const char *it =
+			(const char *)memchr((void *)start, first, (end - start) - (needlelen - 1));
+		if (it == NULL)
+			break;
+
+		if (memcmp(it, needle, needlelen) == 0)
+			return it;
+
+		start = it + 1;
+	}
+
+	return end;
+}
+
+static int str_starts_with(const char *start, const char *end, const char *prefix)
+{
+	long nbytes;
+	assert(start <= end);
+
+	nbytes = strlen(prefix);
+	if (end - start < nbytes)
+		return 0;
+
+	return memcmp(prefix, start, nbytes) == 0;
+}
+
+static int white_space(int c)
+{
+	switch (c) {
+	case ' ':
+	case '\t':
+	case '\r':
+	case '\n':
+		return 1;
+	}
+
+	return 0;
+}
+
+static int name_start_char(int c)
+{
+	if (0x80 <= c)
+		return 1;
+
+	return c == ':' || ('A' <= c && c <= 'Z') || c == '_' || ('a' <= c && c <= 'z');
+}
+
+static int name_char(int c)
+{
+	return name_start_char(c) || c == '-' || c == '.' || ('0' <= c && c <= '9') || c == 0xB7 ||
+	       (0x0300 <= c && c <= 0x036F) || (0x203F <= c && c <= 0x2040);
+}
+
+#define is_space(c) (white_space(((u8)(c))))
+#define is_alpha(c) (name_start_char(((u8)(c))))
+#define is_alnum(c) (name_char(((u8)(c))))
+
+static const char *str_ltrim(const char *start, const char *end)
+{
+	const char *it;
+	assert(start <= end);
+
+	for (it = start; it != end && is_space(*it); it++)
+		;
+
+	return it;
+}
+
+static const char *str_rtrim(const char *start, const char *end)
+{
+	const char *it, *prev;
+	assert(start <= end);
+
+	for (it = end; start != it; it = prev) {
+		prev = it - 1;
+		if (!is_space(*prev))
+			return it;
+	}
+
+	return start;
+}
+
+static const char *str_find_notalnum(const char *start, const char *end)
+{
+	const char *it;
+	assert(start <= end);
+
+	for (it = start; it != end && is_alnum(*it); it++)
+		;
+
+	return it;
+}
+
+#define buffer_from_offset(args, i) ((args)->buffer + (i))
+#define buffer_tooffset(args, ptr) (unsigned)((ptr) - (args)->buffer)
+#define buffer_getend(args) ((args)->buffer + (args)->buffer_length)
+
+static int state_push_token(struct xml *state, struct xml_args *args, enum xml_type type,
+			    const char *start, const char *end)
+{
+	struct xml_token *token;
+	u32 i;
+	if (args->num_tokens <= state->ntokens)
+		return 0;
+
+	i = state->ntokens++;
+	token = &args->tokens[i];
+	token->type = type;
+	token->start_pos = buffer_tooffset(args, start);
+	token->end_pos = buffer_tooffset(args, end);
+	token->size = 0;
+
+	switch (type) {
+	case XML_START_TAG:
+		state->tag_level++;
+		break;
+
+	case XML_END_TAG:
+		assert(0 < state->tag_level);
+		state->tag_level--;
+		break;
+
+	default:
+		break;
+	}
+
+	return 1;
+}
+
+static enum xml_error state_set_pos(struct xml *state, const struct xml_args *args, const char *ptr)
+{
+	state->buffer_pos = buffer_tooffset(args, ptr);
+	return (state->ntokens <= args->num_tokens) ? XML_SUCCESS : XML_ERROR_TOKENSFULL;
+}
+
+#define state_commit(dest, src) memcpy((dest), (src), sizeof(struct xml))
+
+#define XML_ERROR_STRICT XML_ERROR_INVALID
+#define ENTITY_MAXLEN 8
+#define MIN(a, b) ((a) < (b) ? (a) : (b))
+
+static enum xml_error parse_characters(struct xml *state, struct xml_args *args, const char *end)
+{
+	const char *start = buffer_from_offset(args, state->buffer_pos);
+	const char *limit, *colon, *ampr = str_findchr(start, end, '&');
+	assert(end <= buffer_getend(args));
+
+	if (ampr != start)
+		state_push_token(state, args, XML_CHARACTER, start, ampr);
+
+	if (ampr == end)
+		return state_set_pos(state, args, ampr);
+
+	limit = MIN(ampr + ENTITY_MAXLEN, end);
+	colon = str_findchr(ampr, limit, ';');
+	if (colon == limit)
+		return (limit == end) ? XML_ERROR_BUFFERDRY : XML_ERROR_INVALID;
+
+	start = colon + 1;
+	state_push_token(state, args, XML_CHARACTER, ampr, start);
+	return state_set_pos(state, args, start);
+}
+
+static enum xml_error parse_attrvalue(struct xml *state, struct xml_args *args, const char *end)
+{
+	while (buffer_from_offset(args, state->buffer_pos) != end) {
+		enum xml_error error = parse_characters(state, args, end);
+		if (error != XML_SUCCESS)
+			return error;
+	}
+
+	return XML_SUCCESS;
+}
+
+static enum xml_error parse_attributes(struct xml *state, struct xml_args *args)
+{
+	const char *start = buffer_from_offset(args, state->buffer_pos);
+	const char *end = buffer_getend(args);
+	const char *name = str_ltrim(start, end);
+
+	u32 ntokens = state->ntokens;
+	assert(0 < ntokens);
+
+	while (name != end && is_alpha(*name)) {
+		const char *eq, *space, *quot, *value;
+		enum xml_error error;
+
+		eq = str_findchr(name, end, '=');
+		if (eq == end)
+			return XML_ERROR_BUFFERDRY;
+
+		space = str_rtrim(name, eq);
+		state_push_token(state, args, XML_CDATA, name, space);
+
+		quot = str_ltrim(eq + 1, end);
+		if (quot == end)
+			return XML_ERROR_BUFFERDRY;
+		else if (*quot != '\'' && *quot != '"')
+			return XML_ERROR_INVALID;
+
+		value = quot + 1;
+		quot = str_findchr(value, end, *quot);
+		if (quot == end)
+			return XML_ERROR_BUFFERDRY;
+
+		state_set_pos(state, args, value);
+		error = parse_attrvalue(state, args, quot);
+		if (error != XML_SUCCESS)
+			return error;
+
+		name = str_ltrim(quot + 1, end);
+	}
+
+	{
+		struct xml_token *token = args->tokens + (ntokens - 1);
+		token->size = (u16)(state->ntokens - ntokens);
+	}
+
+	return state_set_pos(state, args, name);
+}
+
+#define TAG_LEN(str) (sizeof(str) - 1)
+#define TAG_MINSIZE 1
+
+static enum xml_error parse_comment(struct xml *state, struct xml_args *args)
+{
+	static const char START_TAG[] = "<!--";
+	static const char END_TAG[] = "-->";
+
+	const char *dash;
+	const char *start = buffer_from_offset(args, state->buffer_pos);
+	const char *end = buffer_getend(args);
+	if (end - start < (int)TAG_LEN(START_TAG))
+		return XML_ERROR_BUFFERDRY;
+
+	if (!str_starts_with(start, end, START_TAG))
+		return XML_ERROR_INVALID;
+
+	start += TAG_LEN(START_TAG);
+	dash = str_findstr(start, end, END_TAG);
+	if (dash == end)
+		return XML_ERROR_BUFFERDRY;
+
+	state_push_token(state, args, XML_COMMENT, start, dash);
+	return state_set_pos(state, args, dash + TAG_LEN(END_TAG));
+}
+
+static enum xml_error parse_instruction(struct xml *state, struct xml_args *args)
+{
+	static const char START_TAG[] = "<?";
+	static const char END_TAG[] = "?>";
+
+	enum xml_error error;
+	const char *quest, *space;
+	const char *start = buffer_from_offset(args, state->buffer_pos);
+	const char *end = buffer_getend(args);
+	assert(TAG_MINSIZE <= end - start);
+
+	if (!str_starts_with(start, end, START_TAG))
+		return XML_ERROR_INVALID;
+
+	start += TAG_LEN(START_TAG);
+	space = str_find_notalnum(start, end);
+	if (space == end)
+		return XML_ERROR_BUFFERDRY;
+
+	state_push_token(state, args, XML_INSTRUCTION, start, space);
+
+	state_set_pos(state, args, space);
+	error = parse_attributes(state, args);
+	if (error != XML_SUCCESS)
+		return error;
+
+	quest = buffer_from_offset(args, state->buffer_pos);
+	if (end - quest < (int)TAG_LEN(END_TAG))
+		return XML_ERROR_BUFFERDRY;
+
+	if (!str_starts_with(quest, end, END_TAG))
+		return XML_ERROR_INVALID;
+
+	return state_set_pos(state, args, quest + TAG_LEN(END_TAG));
+}
+
+static enum xml_error parse_doctype(struct xml *state, struct xml_args *args)
+{
+	static const char START_TAG[] = "<!DOCTYPE";
+	static const char END_TAG[] = ">";
+
+	const char *bracket;
+	const char *start = buffer_from_offset(args, state->buffer_pos);
+	const char *end = buffer_getend(args);
+	if (end - start < (int)TAG_LEN(START_TAG))
+		return XML_ERROR_BUFFERDRY;
+
+	if (!str_starts_with(start, end, START_TAG))
+		return XML_ERROR_BUFFERDRY;
+
+	start += TAG_LEN(START_TAG);
+	bracket = str_findstr(start, end, END_TAG);
+	if (bracket == end)
+		return XML_ERROR_BUFFERDRY;
+
+	state_push_token(state, args, XML_DOCTYPE, start, bracket);
+	return state_set_pos(state, args, bracket + TAG_LEN(END_TAG));
+}
+
+static enum xml_error parse_start(struct xml *state, struct xml_args *args)
+{
+	enum xml_error error;
+	const char *gt, *name, *space;
+	const char *start = buffer_from_offset(args, state->buffer_pos);
+	const char *end = buffer_getend(args);
+	assert(TAG_MINSIZE <= end - start);
+
+	if (!(start[0] == '<' && is_alpha(start[1])))
+		return XML_ERROR_INVALID;
+
+	name = start + 1;
+	space = str_find_notalnum(name, end);
+	if (space == end)
+		return XML_ERROR_BUFFERDRY;
+
+	state_push_token(state, args, XML_START_TAG, name, space);
+
+	state_set_pos(state, args, space);
+	error = parse_attributes(state, args);
+	if (error != XML_SUCCESS)
+		return error;
+
+	gt = buffer_from_offset(args, state->buffer_pos);
+
+	if (gt != end && *gt == '/') {
+		state_push_token(state, args, XML_END_TAG, name, space);
+		gt++;
+	}
+
+	if (gt == end)
+		return XML_ERROR_BUFFERDRY;
+
+	if (*gt != '>')
+		return XML_ERROR_INVALID;
+
+	return state_set_pos(state, args, gt + 1);
+}
+
+static enum xml_error parse_end(struct xml *state, struct xml_args *args)
+{
+	const char *gt, *space;
+	const char *start = buffer_from_offset(args, state->buffer_pos);
+	const char *end = buffer_getend(args);
+	assert(TAG_MINSIZE <= end - start);
+
+	if (!(str_starts_with(start, end, "</") && is_alpha(start[2])))
+		return XML_ERROR_INVALID;
+
+	start += 2;
+	gt = str_findchr(start, end, '>');
+	if (gt == end)
+		return XML_ERROR_BUFFERDRY;
+
+	space = str_find_notalnum(start, gt);
+	if (str_ltrim(space, gt) != gt)
+		return XML_ERROR_STRICT;
+
+	state_push_token(state, args, XML_END_TAG, start, space);
+	return state_set_pos(state, args, gt + 1);
+}
+
+static enum xml_error parse_cdata(struct xml *state, struct xml_args *args)
+{
+	static const char START_TAG[] = "<![CDATA[";
+	static const char END_TAG[] = "]]>";
+
+	const char *bracket;
+	const char *start = buffer_from_offset(args, state->buffer_pos);
+	const char *end = buffer_getend(args);
+	if (end - start < (int)TAG_LEN(START_TAG))
+		return XML_ERROR_BUFFERDRY;
+
+	if (!str_starts_with(start, end, START_TAG))
+		return XML_ERROR_INVALID;
+
+	start += TAG_LEN(START_TAG);
+	bracket = str_findstr(start, end, END_TAG);
+	if (bracket == end)
+		return XML_ERROR_BUFFERDRY;
+
+	state_push_token(state, args, XML_CDATA, start, bracket);
+	return state_set_pos(state, args, bracket + TAG_LEN(END_TAG));
+}
+
+void xml_init(struct xml *state)
+{
+	state->buffer_pos = 0;
+	state->ntokens = 0;
+	state->tag_level = 0;
+}
+
+#define ROOT_FOUND(state) (0 < (state)->tag_level)
+#define ROOT_PARSED(state) ((state)->tag_level == 0)
+
+enum xml_error xml_parse(struct xml *state, const char *buffer, u32 buffer_length,
+			 struct xml_token tokens[], u32 num_tokens)
+{
+	struct xml temp = *state;
+	const char *end = buffer + buffer_length;
+
+	struct xml_args args;
+	args.buffer = buffer;
+	args.buffer_length = buffer_length;
+	args.tokens = tokens;
+	args.num_tokens = num_tokens;
+
+	while (!ROOT_FOUND(&temp)) {
+		enum xml_error error;
+		const char *start = buffer_from_offset(&args, temp.buffer_pos);
+		const char *lt = str_ltrim(start, end);
+		state_set_pos(&temp, &args, lt);
+		state_commit(state, &temp);
+
+		if (end - lt < TAG_MINSIZE)
+			return XML_ERROR_BUFFERDRY;
+
+		if (*lt != '<')
+			return XML_ERROR_INVALID;
+
+		switch (lt[1]) {
+		case '?':
+			error = parse_instruction(&temp, &args);
+			break;
+		case '!':
+			error = (lt[2] == '-') ? parse_comment(&temp, &args) :
+						       parse_doctype(&temp, &args);
+			break;
+		default:
+			error = parse_start(&temp, &args);
+			break;
+		}
+
+		if (error != XML_SUCCESS)
+			return error;
+
+		state_commit(state, &temp);
+	}
+
+	while (!ROOT_PARSED(&temp)) {
+		enum xml_error error;
+		const char *start = buffer_from_offset(&args, temp.buffer_pos);
+		const char *lt = str_findchr(start, end, '<');
+		while (buffer_from_offset(&args, temp.buffer_pos) != lt) {
+			error = parse_characters(&temp, &args, lt);
+			if (error != XML_SUCCESS)
+				return error;
+
+			state_commit(state, &temp);
+		}
+
+		if (end - lt < TAG_MINSIZE)
+			return XML_ERROR_BUFFERDRY;
+
+		switch (lt[1]) {
+		case '?':
+			error = parse_instruction(&temp, &args);
+			break;
+		case '/':
+			error = parse_end(&temp, &args);
+			break;
+		case '!':
+			error = (lt[2] == '-') ? parse_comment(&temp, &args) :
+						       parse_cdata(&temp, &args);
+			break;
+		default:
+			error = parse_start(&temp, &args);
+			break;
+		}
+
+		if (error != XML_SUCCESS)
+			return error;
+
+		state_commit(state, &temp);
+	}
+
+	return XML_SUCCESS;
+}
diff --git a/libs/libtxt/xml.h b/libs/libtxt/xml.h
new file mode 100644
index 0000000..43a8005
--- /dev/null
+++ b/libs/libtxt/xml.h
@@ -0,0 +1,51 @@
+// Inspired by sxml (capmar)
+// MIT License, Copyright (c) 2020 Marvin Borner
+
+#ifndef XML_H
+#define XML_H
+
+#include <def.h>
+
+enum xml_error {
+	XML_ERROR_INVALID = -1,
+	XML_SUCCESS = 0,
+	XML_ERROR_BUFFERDRY = 1,
+	XML_ERROR_TOKENSFULL = 2
+};
+
+struct xml_token {
+	u16 type;
+	u16 size;
+	u32 start_pos;
+	u32 end_pos;
+};
+
+struct xml_args {
+	const char *buffer;
+	u32 buffer_length;
+	struct xml_token *tokens;
+	u32 num_tokens;
+};
+
+enum xml_type {
+	XML_START_TAG,
+	XML_END_TAG,
+	XML_CHARACTER,
+	XML_CDATA,
+	XML_INSTRUCTION,
+	XML_DOCTYPE,
+	XML_COMMENT
+};
+
+struct xml {
+	u32 buffer_pos;
+	u32 ntokens;
+	u32 tag_level;
+};
+
+enum xml_error xml_parse(struct xml *parser, const char *buffer, u32 buffer_length,
+			 struct xml_token *tokens, u32 num_tokens);
+
+void xml_init(struct xml *parser);
+
+#endif