diff options
| -rw-r--r-- | README.md | 1 | ||||
| -rw-r--r-- | apps/browser.c | 65 | ||||
| -rw-r--r-- | apps/server.c | 3 | ||||
| -rw-r--r-- | libc/inc/mem.h | 1 | ||||
| -rw-r--r-- | libc/mem.c | 12 | ||||
| -rw-r--r-- | libgui/gfx.c | 3 | ||||
| -rw-r--r-- | libtxt/Makefile | 2 | ||||
| -rw-r--r-- | libtxt/inc/xml.h | 51 | ||||
| -rw-r--r-- | libtxt/xml.c | 512 | ||||
| -rw-r--r-- | res/www/index.html | 2 |
10 files changed, 645 insertions, 7 deletions
@@ -91,3 +91,4 @@ Resources: Libraries: - [upng (heavily modified lodepng)](https://github.com/elanthis/upng) - [ZLIB License](https://github.com/lvandeve/lodepng/blob/7fdcc96a5e5864eee72911c3ca79b1d9f0d12292/LICENSE) +- [sxml](https://github.com/capmar/sxml) - [Unlicense License](https://github.com/capmar/sxml/blob/91176b4c62ef7c6342804e02fc440b2e82326469/UNLICENSE) diff --git a/apps/browser.c b/apps/browser.c index bdc7b14..179acb1 100644 --- a/apps/browser.c +++ b/apps/browser.c @@ -9,6 +9,7 @@ #include <net.h> #include <print.h> #include <str.h> +#include <xml.h> #define WIDTH 640 #define HEIGHT 400 @@ -45,6 +46,62 @@ u32 status_color(char *http_code) return c; } +void parse(void *data, u32 len, char **out) +{ + struct xml_token tokens[128]; + + struct xml parser; + xml_init(&parser); + void *buffer = data; + len = strlen(data); + enum xml_error err = xml_parse(&parser, buffer, len, tokens, 128); + printf("%s\n", data); + if (err != XML_SUCCESS) { + printf("ERROR %d\n", err); + return; + } + + *out[0] = '\0'; + + u32 pos = 0; + u32 indent = 0; + char name[16] = { 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: + for (u32 j = 0; j < indent; j++) + strcat(*out, "\t"); + indent++; + memcpy(&name, (u8 *)buffer + token->start_pos, + token->end_pos - token->start_pos); + name[token->end_pos - token->start_pos] = '\0'; + printf("%s\n", name); + print("START TAG\n"); + strcat(*out, name); + strcat(*out, "\n"); + break; + case XML_END_TAG: + indent--; + for (u32 j = 0; j < indent; j++) + strcat(*out, "\t"); + memcpy(&name, (u8 *)buffer + token->start_pos, + token->end_pos - token->start_pos); + name[token->end_pos - token->start_pos] = '\0'; + print("END TAG\n"); + strcat(*out, name); + strcat(*out, "/\n"); + break; + default: + break; + } + + i += token->size; + } + printf("%s\n", *out); +} + void on_submit(void *event, struct element *box) { (void)event; @@ -66,11 +123,13 @@ void on_submit(void *event, struct element *box) struct element_label *c = code_label->data; struct socket *socket = net_open(S_TCP); - if (socket && net_connect(socket, ip, 80)) { + if (socket && net_connect(socket, ip, 8000)) { net_send(socket, query, strlen(query)); char buf[4096] = { 0 }; + char parsed[4096] = { 0 }; net_receive(socket, buf, 4096); - l->text = http_data(buf); + parse(http_data(buf), 4096, (char **)&parsed); + l->text = http_data(parsed); c->text = http_code(buf); c->color_fg = status_color(c->text); } else { @@ -90,8 +149,10 @@ int main() code_label = gui_add_label(root, 0, 0, FONT_24, "000", COLOR_BLACK, COLOR_WHITE); struct element *text_input = gui_add_text_input(root, LABEL_WIDTH, 0, 100, FONT_24, COLOR_WHITE, COLOR_BLACK); + memcpy(((struct element_text_input *)text_input->data)->text, strdup("127.0.0.1"), 10); output = gui_add_text_box(root, 0, FONT_HEIGHT + 2, 100, 100, FONT_16, "Enter URL and press Enter :)", COLOR_WHITE, COLOR_BLACK); + gui_sync(root, text_input); text_input->event.on_submit = on_submit; diff --git a/apps/server.c b/apps/server.c index fe2c17c..4ab3369 100644 --- a/apps/server.c +++ b/apps/server.c @@ -7,9 +7,8 @@ #include <str.h> #define PORT 8000 -#define FILE "/res/www/index.html" #define PATH "/res/www" -#define ERROR "/res/www/404.html" +#define ERROR PATH "/404.html" int main() { diff --git a/libc/inc/mem.h b/libc/inc/mem.h index 8977a48..964b24b 100644 --- a/libc/inc/mem.h +++ b/libc/inc/mem.h @@ -22,6 +22,7 @@ void free(void *ptr); void *memcpy(void *dest, const void *src, u32 n); void *memset(void *dest, int val, u32 n); +void *memchr(void *src, int c, u32 n); int memcmp(const void *s1, const void *s2, u32 n); #endif @@ -43,6 +43,18 @@ void *memset(void *dest, int val, u32 n) return dest; } +void *memchr(void *src, int c, u32 n) +{ + const u8 *s = (const u8 *)src; + + while (n-- > 0) { + if (*s == c) + return (void *)s; + s++; + } + return NULL; +} + int memcmp(const void *s1, const void *s2, u32 n) { const u8 *a = (const u8 *)s1; diff --git a/libgui/gfx.c b/libgui/gfx.c index 3d3785e..6d54550 100644 --- a/libgui/gfx.c +++ b/libgui/gfx.c @@ -121,7 +121,10 @@ void gfx_write(struct context *ctx, int x, int y, enum font_type font_type, u32 cnt = 0; } else if (text[i] == '\n') { cnt = 0; + x = 0; y += font->height; + } else if (text[i] == '\t') { + x += 4 * font->width; } else { // TODO: Overflow on single line input if ((cnt + 1) * font->width > ctx->width) { diff --git a/libtxt/Makefile b/libtxt/Makefile index 436cb3b..543a90d 100644 --- a/libtxt/Makefile +++ b/libtxt/Makefile @@ -1,6 +1,6 @@ # MIT License, Copyright (c) 2020 Marvin Borner -COBJS = keymap.o +COBJS = keymap.o xml.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 diff --git a/libtxt/inc/xml.h b/libtxt/inc/xml.h new file mode 100644 index 0000000..43a8005 --- /dev/null +++ b/libtxt/inc/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 diff --git a/libtxt/xml.c b/libtxt/xml.c new file mode 100644 index 0000000..3a69ea7 --- /dev/null +++ b/libtxt/xml.c @@ -0,0 +1,512 @@ +// Inspired by sxml (capmar) +// MIT License, Copyright (c) 2020 Marvin Borner + +#include <assert.h> +#include <mem.h> +#include <str.h> +#include <xml.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 = state->ntokens++; + if (args->num_tokens < state->ntokens) + return 0; + + 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 err = parse_characters(state, args, end); + if (err != XML_SUCCESS) + return err; + } + + 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 err; + + 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); + err = parse_attrvalue(state, args, quot); + if (err != XML_SUCCESS) + return err; + + 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 3 + +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 err; + 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); + err = parse_attributes(state, args); + if (err != XML_SUCCESS) + return err; + + 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 err; + 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); + err = parse_attributes(state, args); + if (err != XML_SUCCESS) + return err; + + 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 err; + 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 '?': + err = parse_instruction(&temp, &args); + break; + case '!': + err = parse_doctype(&temp, &args); + break; + default: + err = parse_start(&temp, &args); + break; + } + + if (err != XML_SUCCESS) + return err; + + state_commit(state, &temp); + } + + while (!ROOT_PARSED(&temp)) { + enum xml_error err; + 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) { + err = parse_characters(&temp, &args, lt); + if (err != XML_SUCCESS) + return err; + + state_commit(state, &temp); + } + + if (end - lt < TAG_MINSIZE) + return XML_ERROR_BUFFERDRY; + + switch (lt[1]) { + case '?': + err = parse_instruction(&temp, &args); + break; + case '/': + err = parse_end(&temp, &args); + break; + case '!': + err = (lt[2] == '-') ? parse_comment(&temp, &args) : + parse_cdata(&temp, &args); + break; + default: + err = parse_start(&temp, &args); + break; + } + + if (err != XML_SUCCESS) + return err; + + state_commit(state, &temp); + } + + return XML_SUCCESS; +} diff --git a/res/www/index.html b/res/www/index.html index 49e1b07..15a9a7b 100644 --- a/res/www/index.html +++ b/res/www/index.html @@ -1,5 +1,3 @@ -<!DOCTYPE html> -<!-- MIT License, Copyright (c) 2020 Marvin Borner --> <html> <head> <meta charset="UTF-8" /> |