diff options
Diffstat (limited to 'kit/xml.c')
| -rw-r--r-- | kit/xml.c | 483 |
1 files changed, 483 insertions, 0 deletions
diff --git a/kit/xml.c b/kit/xml.c new file mode 100644 index 0000000..f61bfc1 --- /dev/null +++ b/kit/xml.c @@ -0,0 +1,483 @@ +#include "xml.h" + +#include "input_buffer.h" +#include <assert.h> + +typedef struct { + ib_token_t last; + str_builder_t text; + kit_da_xml_t tags; +} kit_xml_intermediate_t; + +static s32 kit_xml_alloc_and_unescape_(str_builder_t *dst, str_t str, + kit_allocator_t *alloc) { + assert(dst != NULL); + assert(str.size == 0 || str.values != NULL); + + if (dst == NULL) + return KIT_ERROR_INTERNAL; + if (str.size != 0 && str.values == NULL) + return KIT_ERROR_INTERNAL; + + DA_INIT(*dst, str.size, alloc); + + if (dst->size != str.size) + return KIT_ERROR_OUT_OF_MEMORY; + + dst->size = 0; + + for (i64 i = 0; i < str.size; i++) + if (str.values[i] != '&') + dst->values[dst->size++] = str.values[i]; + else { + i64 n = 1; + while (i + n < str.size && str.values[i + n] != ';') + n++; + if (i + n >= str.size) { + DA_DESTROY(*dst); + return KIT_PARSING_FAILED; + } + if (n == 3 && memcmp(str.values + i, "<", 4) == 0) + dst->values[dst->size++] = '<'; + else if (n == 3 && memcmp(str.values + i, ">", 4) == 0) + dst->values[dst->size++] = '>'; + else if (n == 4 && memcmp(str.values + i, "&", 5) == 0) + dst->values[dst->size++] = '&'; + else if (n == 5 && memcmp(str.values + i, """, 6) == 0) + dst->values[dst->size++] = '"'; + else if (n == 5 && memcmp(str.values + i, "'", 6) == 0) + dst->values[dst->size++] = '\''; + else if (n - 2 <= 8 && str.values[i + 1] == '#' && + str.values[i + 2] == 'x') { + // hex encoding + // + + c8 buf[8]; + u64 x = 0; + + memcpy(buf, str.values + (i + 3), n - 2); + + for (i64 k = 0; k < n - 2; k++) { + c8 c = str.values[i + 3 + k]; + x <<= 8; + if (c >= '0' && c <= '9') + x |= (c - '0'); + else if (c >= 'a' && c <= 'f') + x |= 10 + (c - 'a'); + else if (c >= 'A' && c <= 'F') + x |= 10 + (c - 'A'); + else { + x = 0; + break; + } + } + + if (x == 0 || x > 255u) { + // TODO + // UTF-8 encoding + + DA_DESTROY(*dst); + return KIT_PARSING_FAILED; + } + + dst->values[dst->size++] = (c8) x; + } else if (n - 1 <= 20 && str.values[i + 1] == '#') { + // dec encoding + // + + c8 buf[20]; + u64 x = 0; + + memcpy(buf, str.values + (i + 2), n - 2); + + for (i64 k = 0; k < n - 1; k++) { + c8 c = str.values[i + 2 + k]; + x *= 10; + if (c >= '0' && c <= '9') + x += (c - '0'); + else { + x = 0; + break; + } + } + + if (x == 0 || x > 255u) { + // TODO + // UTF-8 encoding + + DA_DESTROY(*dst); + return KIT_PARSING_FAILED; + } + + dst->values[dst->size++] = (c8) x; + } else { + DA_DESTROY(*dst); + return KIT_PARSING_FAILED; + } + i += n; + } + + return KIT_OK; +} + +static ib_token_t kit_xml_parse_text_(ib_token_t begin, + str_builder_t *dst) { + ib_token_t last = ib_until(begin, SZ("<")); + + DA_RESIZE(*dst, last.size); + + assert(dst->size == last.size); + if (dst->size != last.size) + last.status |= KIT_ERROR_OUT_OF_MEMORY; + else if (last.size > 0) + memcpy(dst->values, ib_str(last).values, last.size); + + for (;;) { + ib_token_t comment_open = ib_exact(last, SZ("<!--")); + + if (comment_open.status != KIT_OK) + break; + + ib_token_t comment_text = ib_until(comment_open, SZ("-->")); + ib_token_t comment_close = ib_exact(comment_text, SZ("-->")); + ib_token_t next_text = ib_until(comment_close, SZ("<")); + + if (next_text.status == KIT_OK && next_text.size > 0) { + i64 n = dst->size; + DA_RESIZE(*dst, n + next_text.size); + + assert(dst->size == n + next_text.size); + if (dst->size != n + next_text.size) + next_text.status |= KIT_ERROR_OUT_OF_MEMORY; + else + memcpy(dst->values + n, ib_str(next_text).values, + ib_str(next_text).size); + } + + last = next_text; + } + + return last; +} + +static ib_token_t kit_xml_parse_string_(ib_token_t begin, + ib_token_t *value) { + assert(value != NULL); + if (value == NULL) { + begin.status |= KIT_ERROR_INTERNAL; + return begin; + } + + ib_token_t quotes_open = ib_exact(begin, SZ("\"")); + ib_token_t apostr_open = ib_exact(begin, SZ("'")); + + ib_token_t open = quotes_open.status == KIT_OK ? quotes_open + : apostr_open; + + *value = ib_until(open, ib_str(open)); + ib_token_t close = ib_exact(*value, ib_str(open)); + + return close; +} + +static kit_xml_intermediate_t kit_xml_parse_buf_( + ib_token_t begin, kit_allocator_t *alloc) { + kit_xml_intermediate_t res; + memset(&res, 0, sizeof res); + + ib_token_t last, spaces; + memset(&last, 0, sizeof last); + memset(&spaces, 0, sizeof spaces); + + str_builder_t tag_text_string; + str_builder_t tag_tail_string; + DA_INIT(tag_text_string, 0, alloc); + DA_INIT(tag_tail_string, 0, alloc); + + ib_token_t tag_text = kit_xml_parse_text_(begin, &tag_text_string); + last = tag_text; + + DA_INIT(res.tags, 0, alloc); + + for (;;) { + ib_token_t tagend_open = ib_exact(last, SZ("</")); + if (tagend_open.status == KIT_OK) + break; + + ib_token_t tag_open = ib_exact(last, SZ("<")); + + if (tag_open.status != KIT_OK) + break; + + xml_t tag; + memset(&tag, 0, sizeof tag); + + ib_token_t decl_open = ib_exact(tag_open, SZ("?")); + + if (decl_open.status == KIT_OK) { + tag.is_declaration = 1; + last = decl_open; + } else + last = tag_open; + + spaces = ib_any(last, SZ(" \t\r\n")); + ib_token_t tag_name = ib_none(spaces, SZ(" \t\r\n/>")); + + DA_INIT(tag.properties, 0, alloc); + + last = tag_name; + + for (;;) { + spaces = ib_any(last, SZ(" \t\r\n")); + ib_token_t property = ib_none(spaces, SZ(" \t\r\n=?/>")); + + if (property.status != KIT_OK || property.size == 0) + break; + + spaces = ib_any(property, SZ(" \t\r\n")); + ib_token_t equals = ib_exact(spaces, SZ("=")); + spaces = ib_any(equals, SZ(" \t\r\n")); + + ib_token_t value; + last = kit_xml_parse_string_(spaces, &value); + + if (last.status == KIT_OK) { + i64 n = tag.properties.size; + DA_RESIZE(tag.properties, n + 1); + + assert(tag.properties.size == n + 1); + if (tag.properties.size != n + 1) { + last.status |= KIT_ERROR_OUT_OF_MEMORY; + DA_DESTROY(tag.properties); + } else { + last.status |= kit_xml_alloc_and_unescape_( + &tag.properties.values[n].name, ib_str(property), + alloc); + last.status |= kit_xml_alloc_and_unescape_( + &tag.properties.values[n].value, ib_str(value), alloc); + } + } + } + + if (tag.is_declaration) { + ib_token_t tag_decl_close = ib_exact(spaces, SZ("?>")); + + last = tag_decl_close; + + DA_INIT(tag.text, 0, alloc); + DA_INIT(tag.children, 0, alloc); + } else { + ib_token_t tag_close = ib_exact(spaces, SZ(">")); + ib_token_t tag_close_empty = ib_exact(spaces, SZ("/>")); + + if (tag_close.status == KIT_OK) { + kit_xml_intermediate_t im = kit_xml_parse_buf_(tag_close, + alloc); + + tag.text = im.text; + tag.children = im.tags; + + tagend_open = ib_exact(im.last, SZ("</")); + spaces = ib_any(tagend_open, SZ(" \t\r\n")); + ib_token_t tagend_name = ib_exact(spaces, ib_str(tag_name)); + spaces = ib_any(tagend_name, SZ(" \t\r\n")); + ib_token_t tagend_close = ib_exact(spaces, SZ(">")); + + last = tagend_close; + + } else if (tag_close_empty.status == KIT_OK) { + last = tag_close_empty; + + DA_INIT(tag.text, 0, alloc); + DA_INIT(tag.children, 0, alloc); + } else + last.status |= KIT_PARSING_FAILED; + } + + ib_token_t tag_tail = kit_xml_parse_text_(last, &tag_tail_string); + + last = tag_tail; + + if (last.status == KIT_OK) { + i64 n = res.tags.size; + DA_RESIZE(res.tags, n + 1); + + assert(res.tags.size == n + 1); + if (res.tags.size != n + 1) { + last.status |= KIT_ERROR_OUT_OF_MEMORY; + xml_destroy(&tag); + } else { + last.status |= kit_xml_alloc_and_unescape_( + &tag.tag, ib_str(tag_name), alloc); + last.status |= kit_xml_alloc_and_unescape_( + &tag.tail, WRAP_STR(tag_tail_string), alloc); + + res.tags.values[n] = tag; + } + } else + xml_destroy(&tag); + } + + if (last.status != KIT_OK) { + for (i64 i = 0; i < res.tags.size; i++) + xml_destroy(res.tags.values + i); + DA_DESTROY(res.text); + DA_DESTROY(res.tags); + } else + last.status |= kit_xml_alloc_and_unescape_( + &res.text, WRAP_STR(tag_text_string), alloc); + + DA_DESTROY(tag_text_string); + DA_DESTROY(tag_tail_string); + + res.last = last; + return res; +} + +kit_xml_parse_result_t kit_xml_parse(is_handle_t is, + kit_allocator_t *alloc) { + input_buffer_t ib = ib_init(is, alloc); + kit_xml_intermediate_t im = kit_xml_parse_buf_(ib_token(&ib), + alloc); + + kit_xml_parse_result_t res; + memset(&res, 0, sizeof res); + + res.status = im.last.status; + + if (res.status != KIT_OK) { + ib_destroy(&ib); + return res; + } + + if (im.text.size == 0 && im.tags.size == 1) { + res.xml = im.tags.values[0]; + DA_DESTROY(im.text); + DA_DESTROY(im.tags); + ib_destroy(&ib); + return res; + } + + DA_INIT(res.xml.tag, 0, alloc); + DA_INIT(res.xml.tail, 0, alloc); + DA_INIT(res.xml.properties, 0, alloc); + + res.xml.text = im.text; + res.xml.children = im.tags; + + ib_destroy(&ib); + return res; +} + +kit_xml_text_t kit_xml_print(kit_xml_t *xml, kit_allocator_t *alloc) { + // TODO + // + + assert(xml != NULL); + + xml_text_t result; + memset(&result, 0, sizeof result); + + (void) alloc; + + result.status = KIT_ERROR_NOT_IMPLEMENTED; + return result; +} + +static s32 kit_xml_append_text_(str_builder_t *buf, xml_t *xml) { + assert(buf != NULL); + assert(xml != NULL); + + i64 n = buf->size; + DA_RESIZE(*buf, n + xml->text.size); + + assert(buf->size == n + xml->text.size); + if (buf->size != n + xml->text.size) + return KIT_ERROR_OUT_OF_MEMORY; + + if (xml->text.size > 0) + memcpy(buf->values + n, xml->text.values, xml->text.size); + + for (i64 i = 0; i < xml->children.size; i++) { + s32 s = kit_xml_append_text_(buf, xml->children.values + i); + if (s != KIT_OK) + return s; + + str_t tail = WRAP_STR(xml->children.values[i].tail); + + if (tail.size <= 0) + continue; + + n = buf->size; + DA_RESIZE(*buf, n + tail.size); + + assert(buf->size == n + tail.size); + if (buf->size != n + tail.size) + return KIT_ERROR_OUT_OF_MEMORY; + + if (tail.size > 0) + memcpy(buf->values + n, tail.values, tail.size); + } + + return KIT_OK; +} + +kit_xml_text_t kit_xml_full_text(kit_xml_t *xml, + kit_allocator_t *alloc) { + kit_xml_text_t res; + res.status = KIT_OK; + DA_INIT(res.text, 0, alloc); + + if (xml != NULL) + res.status = kit_xml_append_text_(&res.text, xml); + else + res.status = KIT_ERROR_INVALID_ARGUMENT; + + return res; +} + +b8 kit_xml_has_property(kit_xml_t *xml, kit_str_t name) { + assert(xml != NULL); + if (xml == NULL) + return 0; + + for (i64 i = 0; i < xml->properties.size; i++) + if (AR_EQUAL(xml->properties.values[i].name, name)) + return 1; + + return 0; +} + +str_t kit_xml_property(kit_xml_t *xml, str_t name) { + assert(xml != NULL); + if (xml == NULL) + return str(0, NULL); + + for (i64 i = 0; i < xml->properties.size; i++) + if (AR_EQUAL(xml->properties.values[i].name, name)) + return WRAP_STR(xml->properties.values[i].value); + + assert(0); + return str(0, NULL); +} + +void kit_xml_destroy(kit_xml_t *xml) { + assert(xml != NULL); + if (xml == NULL) + return; + + for (i64 i = 0; i < xml->properties.size; i++) { + DA_DESTROY(xml->properties.values[i].name); + DA_DESTROY(xml->properties.values[i].value); + } + + for (i64 i = 0; i < xml->children.size; i++) + kit_xml_destroy(xml->children.values + i); + + DA_DESTROY(xml->tag); + DA_DESTROY(xml->text); + DA_DESTROY(xml->tail); + + DA_DESTROY(xml->properties); + DA_DESTROY(xml->children); +} |