// Scan ***, **, or * and return number scanned, or 0.
// Advances position.
static int
scan_delims(subject* subj, unsigned char c, bool * can_open, bool * can_close)
{
int numdelims = 0;
int before_char_pos;
int32_t after_char = 0;
int32_t before_char = 0;
int len;
bool left_flanking, right_flanking;
if (subj->pos == 0) {
before_char = 10;
} else {
before_char_pos = subj->pos - 1;
// walk back to the beginning of the UTF_8 sequence:
while (peek_at(subj, before_char_pos) >> 6 == 2 &&
before_char_pos > 0) {
before_char_pos -= 1;
}
len = utf8proc_iterate(subj->input.data + before_char_pos,
subj->pos - before_char_pos, &before_char);
if (len == -1) {
before_char = 10;
}
}
while (peek_char(subj) == c) {
numdelims++;
advance(subj);
}
len = utf8proc_iterate(subj->input.data + subj->pos,
subj->input.len - subj->pos, &after_char);
if (len == -1) {
after_char = 10;
}
left_flanking = numdelims > 0 && !utf8proc_is_space(after_char) &&
!(utf8proc_is_punctuation(after_char) &&
!utf8proc_is_space(before_char) &&
!utf8proc_is_punctuation(before_char));
right_flanking = numdelims > 0 && !utf8proc_is_space(before_char) &&
!(utf8proc_is_punctuation(before_char) &&
!utf8proc_is_space(after_char) &&
!utf8proc_is_punctuation(after_char));
if (c == '_') {
*can_open = left_flanking && !right_flanking;
*can_close = right_flanking && !left_flanking;
} else {
*can_open = left_flanking;
*can_close = right_flanking;
}
return numdelims;
}
void MultiLineTextView::typesetGlyphs(const std::string& str,
const zeus::CColor& defaultColor,
unsigned wrap)
{
if (wrap)
{
typesetGlyphs(LineWrap(str, wrap), defaultColor);
return;
}
m_width = 0;
m_lines.clear();
size_t rem = str.size() + 1;
const utf8proc_uint8_t* it = reinterpret_cast<const utf8proc_uint8_t*>(str.data());
size_t lineCount = 0;
while (rem)
{
utf8proc_int32_t ch;
utf8proc_ssize_t sz = utf8proc_iterate(it, -1, &ch);
if (sz < 0)
Log.report(logvisor::Fatal, "invalid UTF-8 char");
if (ch == '\n' || ch == '\0')
++lineCount;
rem -= sz;
it += sz;
}
m_lines.reserve(lineCount);
rem = str.size() + 1;
it = reinterpret_cast<const utf8proc_uint8_t*>(str.data());
const utf8proc_uint8_t* beginIt = it;
while (rem)
{
utf8proc_int32_t ch;
utf8proc_ssize_t sz = utf8proc_iterate(it, -1, &ch);
if (ch == '\n' || ch == '\0')
{
m_lines.emplace_back(new TextView(m_viewSystem, *this, m_fontAtlas, m_align, m_lineCapacity));
m_lines.back()->typesetGlyphs(std::string((char*)beginIt, it - beginIt), defaultColor);
m_width = std::max(m_width, m_lines.back()->nominalWidth());
beginIt = it + 1;
}
rem -= sz;
it += sz;
}
updateSize();
}
SSliceStatus sslice_get_first_rune(SSlice *s, rune *r) {
rune r2;
ssize_t bytes_read;
if (sslice_empty(s)) {
return SSLICE_END;
}
bytes_read = utf8proc_iterate(
(const unsigned char *)s->data, s->len, &r2
);
if (bytes_read < 1) {
switch (bytes_read) {
case UTF8PROC_ERROR_NOMEM:
return SSLICE_MEMORY_EXHAUSTED;
case UTF8PROC_ERROR_OVERFLOW:
return SSLICE_OVERFLOW;
case UTF8PROC_ERROR_INVALIDUTF8:
return SSLICE_INVALID_UTF8;
case UTF8PROC_ERROR_NOTASSIGNED:
return SSLICE_NOT_ASSIGNED;
case UTF8PROC_ERROR_INVALIDOPTS:
return SSLICE_INVALID_OPTS;
default:
return SSLICE_ERROR_UNKNOWN;
}
}
*r = r2;
return SSLICE_OK;
}
string_script_t get_string_script(char *str, size_t len) {
int32_t ch;
script_t last_script = SCRIPT_UNKNOWN;
script_t script = SCRIPT_UNKNOWN;
uint8_t *ptr = (uint8_t *)str;
size_t script_len = 0;
size_t idx = 0;
bool is_ascii = true;
while (idx < len) {
ssize_t char_len = utf8proc_iterate(ptr, len, &ch);
if (ch == 0) break;
script = get_char_script((uint32_t)ch);
if (script == SCRIPT_COMMON && last_script != SCRIPT_UNKNOWN) {
script = last_script;
}
if (last_script != script && last_script != SCRIPT_UNKNOWN && last_script != SCRIPT_COMMON) {
if (script_len < len) {
while (true) {
char_len = utf8proc_iterate_reversed((const uint8_t *)str, idx, &ch);
if (ch == 0) break;
script = get_char_script((uint32_t)ch);
if (script != SCRIPT_COMMON) {
break;
}
script_len -= char_len;
ptr -= char_len;
idx -= char_len;
}
}
break;
}
is_ascii = is_ascii && ch < MAX_ASCII;
ptr += char_len;
idx += char_len;
script_len += char_len;
if (script != SCRIPT_UNKNOWN) {
last_script = script;
}
}
return (string_script_t) {last_script, script_len, is_ascii};
}
inline int sustream_prev_char(struct ustream *ustrm)
{
struct sustream* sstrm = (struct sustream*)(ustrm->stream);
sstrm->cur -= sstrm->octcnt;
if (sstrm->cur < sstrm->str)
return STRM_END;
sstrm->octcnt = utf8proc_iterate(sstrm->cur, 4, &ustrm->point);
if (sstrm->octcnt < 0)
return WRONG_ENCODING;
return OK;
}
unsigned int utf8_glypheme_length(const char *p)
{
int32_t cp;
int len = utf8proc_iterate((uint8_t *) p, -1, &cp);
if (len == UTF8PROC_ERROR_INVALIDUTF8)
return 1;
// Check for combining characters
const char *p2 = p + len;
while (*p2)
{
int comb_len = utf8proc_iterate((uint8_t *) p2, -1, &cp);
if (comb_len == UTF8PROC_ERROR_INVALIDUTF8)
return 1;
const utf8proc_property_t* prop = utf8proc_get_property(cp);
if (!prop->combining_class)
break;
len += comb_len;
p2 += comb_len;
}
return len;
}
int
utf8proc_mbtowc(wchar_t *pwc, const char *s, size_t n)
{
utf8proc_ssize_t slen;
if (s == NULL)
return (0);
/*
* *pwc == -1 indicates invalid codepoint
* slen < 0 indicates an error
*/
slen = utf8proc_iterate(s, n, pwc);
if (*pwc == (wchar_t)-1 || slen < 0)
return (-1);
return (slen);
}
static void
test_utf8proc_iterate ( void )
{
char utf8_str_simple[] = "The quick brown.fox";
uint8_t *utf8_str_simple_ptr = (uint8_t*)(&utf8_str_simple[0]);
size_t length_simple_string = strlen ( utf8_str_simple );
int32_t unicode_char;
unsigned int index;
ssize_t bytes_read;
for (index = 0; index < length_simple_string; ++index) {
bytes_read = utf8proc_iterate (
&utf8_str_simple_ptr[index],
length_simple_string - index,
&unicode_char );
rtems_test_assert ( bytes_read == 1 );
rtems_test_assert ( (uint8_t)unicode_char == utf8_str_simple_ptr[index]);
}
}
SSliceStatus sslice_base_assign(SSlice *s, char *cs, bool validate) {
size_t slen = strlen(cs);
if (validate) {
SSlice cursor;
int32_t cc;
ssize_t bytes_read;
cursor.data = cs;
cursor.len = slen;
while (cursor.len > 0) {
bytes_read = utf8proc_iterate(
(const unsigned char *)cursor.data, cursor.len, &cc
);
if (bytes_read < 1) {
switch (bytes_read) {
case UTF8PROC_ERROR_NOMEM:
return SSLICE_MEMORY_EXHAUSTED;
case UTF8PROC_ERROR_OVERFLOW:
return SSLICE_OVERFLOW;
case UTF8PROC_ERROR_INVALIDUTF8:
return SSLICE_INVALID_UTF8;
case UTF8PROC_ERROR_NOTASSIGNED:
return SSLICE_NOT_ASSIGNED;
case UTF8PROC_ERROR_INVALIDOPTS:
return SSLICE_INVALID_OPTS;
default:
return SSLICE_ERROR_UNKNOWN;
}
}
cursor.data += bytes_read;
cursor.len -= bytes_read;
}
}
s->data = cs;
s->len = slen;
return SSLICE_OK;
}
SSliceStatus sslice_advance_rune(SSlice *s) {
rune r;
ssize_t bytes_read;
if (sslice_empty(s)) {
return SSLICE_END;
}
bytes_read = utf8proc_iterate(
(const unsigned char *)s->data, s->len, &r
);
if (bytes_read < 1) {
switch (bytes_read) {
case UTF8PROC_ERROR_NOMEM:
return SSLICE_MEMORY_EXHAUSTED;
case UTF8PROC_ERROR_OVERFLOW:
return SSLICE_OVERFLOW;
case UTF8PROC_ERROR_INVALIDUTF8:
return SSLICE_INVALID_UTF8;
case UTF8PROC_ERROR_NOTASSIGNED:
return SSLICE_NOT_ASSIGNED;
case UTF8PROC_ERROR_INVALIDOPTS:
return SSLICE_INVALID_OPTS;
default:
return SSLICE_ERROR_UNKNOWN;
}
}
if (bytes_read < (ssize_t)s->len) {
s->data += bytes_read;
s->len -= bytes_read;
return SSLICE_OK;
}
sslice_clear(s);
return SSLICE_END;
}
string_tree_t *regex_string_tree(char *regex, size_t len) {
uint8_t *char_ptr = (uint8_t *)regex;
bool in_set = false;
bool in_brackets = false;
int32_t codepoint;
int32_t last_codepoint = 0;
ssize_t char_len;
size_t bracket_start;
size_t bracket_len;
char temp_char[MAX_UTF8_CHAR_SIZE];
ssize_t temp_char_len;
string_tree_t *tree = string_tree_new();
if (len == 0) {
// Single token with zero-length
string_tree_add_string_len(tree, regex, len);
string_tree_finalize_token(tree);
return tree;
}
uint32_array *char_set = uint32_array_new();
size_t idx = 0;
int i, j;
bool add_to_index = false;
while (idx < len) {
char_len = utf8proc_iterate(char_ptr, len, &codepoint);
if (char_len <= 0) {
uint32_array_destroy(char_set);
string_tree_destroy(tree);
return NULL;
}
if (!(utf8proc_codepoint_valid(codepoint))) {
idx += char_len;
char_ptr += char_len;
continue;
}
add_to_index = true;
if (codepoint == LSQUARE_CODEPOINT && last_codepoint != BACKSLASH_CODEPOINT) {
log_debug("begin set\n");
in_set = true;
codepoint = BEGIN_SET_CODEPOINT;
uint32_array_clear(char_set);
} else if (codepoint == RSQUARE_CODEPOINT && last_codepoint != BACKSLASH_CODEPOINT && in_set) {
log_debug("end set");
for (j = 0; j < char_set->n; j++) {
temp_char_len = utf8proc_encode_char(char_set->a[j], (uint8_t *)temp_char);
log_debug("Adding string %.*s\n", (int)temp_char_len, temp_char);
string_tree_add_string_len(tree, temp_char, temp_char_len);
}
string_tree_finalize_token(tree);
uint32_array_clear(char_set);
// Add a special codepoint to the sequence to distinguish from an escaped square bracket
codepoint = END_SET_CODEPOINT;
in_set = false;
} else if (codepoint == LCURLY_CODEPOINT && last_codepoint != BACKSLASH_CODEPOINT) {
in_brackets = true;
bracket_start = idx + char_len;
bracket_len = 0;
add_to_index = false;
} else if (codepoint == RCURLY_CODEPOINT && last_codepoint != BACKSLASH_CODEPOINT && in_brackets) {
log_debug("Adding bracketed string: %.*s\n", (int) bracket_len, regex + bracket_start);
string_tree_add_string_len(tree, regex + bracket_start, bracket_len);
in_brackets = false;
} else if ((codepoint == LPAREN_CODEPOINT || codepoint == RPAREN_CODEPOINT) && last_codepoint != BACKSLASH_CODEPOINT) {
log_debug("group\n");
add_to_index = false;
} else if (in_set) {
log_debug("in set\n");
// Queue node, we'll add them to the trie
uint32_array_push(char_set, codepoint);
add_to_index = false;
} else if (in_brackets) {
add_to_index = false;
bracket_len += char_len;
} else if (codepoint == BACKSLASH_CODEPOINT && last_codepoint != BACKSLASH_CODEPOINT) {
add_to_index = false;
}
log_debug("codepoint = %d\n", codepoint);
if (add_to_index) {
temp_char_len = utf8proc_encode_char(codepoint, (uint8_t *)temp_char);
log_debug("char = %.*s\n", (int)temp_char_len, temp_char);
string_tree_add_string_len(tree, temp_char, temp_char_len);
string_tree_finalize_token(tree);
}
idx += char_len;
char_ptr += char_len;
}
uint32_array_destroy(char_set);
return tree;
}
group_capture_array *parse_groups(char *regex, size_t len) {
uint8_t *char_ptr = (uint8_t *)regex;
char last_ch = '\0';
bool in_group = false;
bool in_set = false;
int32_t codepoint, last_codepoint = 0;
ssize_t char_len;
char temp_char[MAX_UTF8_CHAR_SIZE];
ssize_t temp_char_len;
if (len == 0) {
return NULL;
}
group_capture_array *groups = group_capture_array_new_size(1);
size_t idx = 0;
size_t pos = 0;
size_t group_start = 0;
size_t chars_in_group = 0;
while (idx < len) {
char_len = utf8proc_iterate(char_ptr, len, &codepoint);
if (char_len <= 0) {
log_error("char %s had len=%zd\n", char_ptr, char_len);
return NULL;
}
if (!(utf8proc_codepoint_valid(codepoint))) {
idx += char_len;
char_ptr += char_len;
pos++;
continue;
}
if (codepoint == LSQUARE_CODEPOINT && last_codepoint != BACKSLASH_CODEPOINT) {
log_debug("begin set\n");
in_set = true;
} else if (codepoint == RSQUARE_CODEPOINT && last_codepoint != BACKSLASH_CODEPOINT) {
log_debug("end set");
pos++;
in_set = false;
} else if (codepoint == LPAREN_CODEPOINT && last_codepoint != BACKSLASH_CODEPOINT) {
log_debug("begin group\n");
in_group = true;
group_start = pos;
} else if (codepoint == RPAREN_CODEPOINT && last_codepoint != BACKSLASH_CODEPOINT) {
log_debug("close group\n");
in_group = false;
group_capture_array_push(groups, (group_capture_t){group_start, pos - group_start});
} else if (!in_set) {
log_debug("other char\n");
pos++;
}
idx += char_len;
char_ptr += char_len;
}
return groups;
}
static int __lua_font_wrap_text ( lua_State *_l ) {
ex_font_t *font;
const char *text, *whitespace;
int maxWidth;
const char *str, *nextstr, *word_start;
char *newtext, *newtext_p, *last_newtext;
int ch, next_ch, len, newlen, cpylen;
uint ft_index, prev_ft_index;
int cur_x, word_start_x;
ex_glyph_t *glyph;
bool linebreak, beginningOfLine, trimWhitespace, skipcpy;
bool wrapword, collapseSpace, collapseLinebreak;
// get lua arguments
ex_lua_check_nargs(_l,5);
text = luaL_checkstring(_l,1);
luaL_checktype( _l, 2, LUA_TLIGHTUSERDATA );
font = lua_touserdata(_l,2);
whitespace = luaL_checkstring(_l,3);
maxWidth = luaL_checkint(_l,4);
trimWhitespace = beginningOfLine = (luaL_checkint(_l,5) == 1);
//
len = strlen(text);
str = nextstr = word_start = text;
newtext_p = newtext = last_newtext = ex_malloc( len * sizeof(char) );
prev_ft_index = -1;
// get wrapMode
wrapword = false;
collapseSpace = collapseLinebreak = false;
if ( !strncmp( whitespace, "pre-wrap", 8 ) ) {
wrapword = true;
collapseSpace = false;
collapseLinebreak = false;
}
else if ( !strncmp( whitespace, "pre-line", 8 ) ) {
wrapword = true;
collapseSpace = true;
collapseLinebreak = false;
}
else if ( !strncmp( whitespace, "normal", 6 ) ) {
wrapword = true;
collapseSpace = true;
collapseLinebreak = true;
}
else if ( !strncmp( whitespace, "nowrap", 6 ) ) {
wrapword = false;
collapseSpace = true;
collapseLinebreak = true;
}
else if ( !strncmp( whitespace, "pre", 3 ) ) {
wrapword = false;
collapseSpace = false;
collapseLinebreak = false;
}
// process text
cur_x = word_start_x = 0;
linebreak = false;
while ( *str ) {
skipcpy = false;
nextstr += utf8proc_iterate ((const uint8_t *)str, -1, &ch);
// if this is line-break
if ( ch == '\n' || ch == '\r' ) {
if ( collapseLinebreak ) {
ch = ' '; // turn it to space
}
else {
linebreak = true;
}
}
// if this is space
if ( ch == ' ' || ch == '\t' || ch == '\f' ) {
if ( collapseSpace ) {
const char * nextnextstr = nextstr;
while ( *nextnextstr ) {
nextstr = nextnextstr;
nextnextstr += utf8proc_iterate ((const uint8_t *)nextnextstr, -1, &next_ch);
// if next_ch is white-space, then collapse this char
if ( next_ch == ' ' || next_ch == '\t' || next_ch == '\f' ) {
str = nextstr;
continue;
}
// if next_ch is line-break and collapseLinebreak is true, then collapse this char
if ( next_ch == '\n' || next_ch == '\r' ) {
if ( collapseLinebreak ) {
str = nextstr;
continue;
}
}
//
break;
}
// skip first-time collapse
if ( trimWhitespace ) {
trimWhitespace = false;
str = nextstr;
continue;
}
// yes, must turn it to space to make sure only one space
ch = ' ';
}
}
//
trimWhitespace = false;
// process word-break, word-wrap
if ( wrapword ) {
word_start = str;
word_start_x = cur_x;
// if this character can break
if ( nextstr == NULL || __can_word_break (ch) ) {
// advanced character
ft_index = ex_font_get_index ( font, ch );
glyph = ex_font_get_glyph ( font, ft_index );
cur_x += ex_font_get_kerning( font, prev_ft_index, ft_index );
cur_x += glyph->advance_x;
prev_ft_index = ft_index;
// check if the word exceed content width
if ( cur_x > maxWidth ) {
if ( !beginningOfLine ) {
linebreak = true;
// skip copy the white-space if it is at the end of the wrap
if ( ch == ' ' || ch == '\t' || ch == '\f' ) {
skipcpy = true;
}
else {
nextstr = word_start;
cur_x = word_start_x;
}
}
}
beginningOfLine = false;
}
else {
// advanced current character
ft_index = ex_font_get_index ( font, ch );
glyph = ex_font_get_glyph ( font, ft_index );
cur_x += ex_font_get_kerning( font, prev_ft_index, ft_index );
cur_x += glyph->advance_x;
prev_ft_index = ft_index;
const char * nextnextstr = nextstr;
while ( *nextnextstr ) {
nextstr = nextnextstr;
nextnextstr += utf8proc_iterate ((const uint8_t *)nextnextstr, -1, &next_ch);
// if this character can break
if ( __can_word_break (next_ch) ) {
break;
}
// advanced character
ft_index = ex_font_get_index ( font, next_ch );
glyph = ex_font_get_glyph ( font, ft_index );
cur_x += ex_font_get_kerning( font, prev_ft_index, ft_index );
cur_x += glyph->advance_x;
prev_ft_index = ft_index;
// TODO: process word-break
// check if the word exceed content width
if ( cur_x > maxWidth ) {
if ( !beginningOfLine ) {
linebreak = true;
nextstr = word_start;
cur_x = word_start_x;
skipcpy = true;
break;
}
}
}
}
}
else {
// advanced character
ft_index = ex_font_get_index ( font, ch );
glyph = ex_font_get_glyph ( font, ft_index );
cur_x += ex_font_get_kerning( font, prev_ft_index, ft_index );
cur_x += glyph->advance_x;
prev_ft_index = ft_index;
}
// copy character to newtext_p
if ( !skipcpy ) {
cpylen = nextstr - str;
if ( cpylen > 0 ) {
strncpy( newtext_p, str, cpylen);
newtext_p += cpylen;
}
}
// step
str = nextstr;
if ( linebreak ) {
break;
}
}
// text1
newlen = newtext_p-newtext;
if ( newlen > 0 ) {
lua_pushlstring(_l, newtext, newlen);
}
else {
lua_pushnil(_l);
}
// text2
if ( linebreak && *str ) {
lua_pushstring(_l, str );
}
else {
lua_pushnil(_l);
}
lua_pushinteger(_l,cur_x); // width
lua_pushboolean(_l,linebreak); // line-break
//
ex_free(newtext);
return 4; // text1(can be nil), text2(can be nil), width of text1, linebreak
}
static
void S_out(cmark_renderer *renderer,
const char *source,
bool wrap,
cmark_escaping escape)
{
int length = cmark_strbuf_safe_strlen(source);
unsigned char nextc;
int32_t c;
int i = 0;
int len;
cmark_chunk remainder = cmark_chunk_literal("");
int k = renderer->buffer->size - 1;
wrap = wrap && !renderer->no_wrap;
if (renderer->in_tight_list_item && renderer->need_cr > 1) {
renderer->need_cr = 1;
}
while (renderer->need_cr) {
if (k < 0 || renderer->buffer->ptr[k] == '\n') {
k -= 1;
} else {
cmark_strbuf_putc(renderer->buffer, '\n');
if (renderer->need_cr > 1) {
cmark_strbuf_put(renderer->buffer, renderer->prefix->ptr,
renderer->prefix->size);
}
}
renderer->column = 0;
renderer->begin_line = true;
renderer->need_cr -= 1;
}
while (i < length) {
if (renderer->begin_line) {
cmark_strbuf_put(renderer->buffer, renderer->prefix->ptr,
renderer->prefix->size);
// note: this assumes prefix is ascii:
renderer->column = renderer->prefix->size;
}
len = utf8proc_iterate((const uint8_t *)source + i, length - i, &c);
if (len == -1) { // error condition
return; // return without rendering rest of string
}
nextc = source[i + len];
if (c == 32 && wrap) {
if (!renderer->begin_line) {
cmark_strbuf_putc(renderer->buffer, ' ');
renderer->column += 1;
renderer->begin_line = false;
renderer->last_breakable = renderer->buffer->size -
1;
// skip following spaces
while (source[i + 1] == ' ') {
i++;
}
}
} else if (c == 10) {
cmark_strbuf_putc(renderer->buffer, '\n');
renderer->column = 0;
renderer->begin_line = true;
renderer->last_breakable = 0;
} else if (escape == LITERAL) {
cmark_render_code_point(renderer, c);
renderer->begin_line = false;
} else {
(renderer->outc)(renderer, escape, c, nextc);
renderer->begin_line = false;
}
// If adding the character went beyond width, look for an
// earlier place where the line could be broken:
if (renderer->width > 0 &&
renderer->column > renderer->width &&
!renderer->begin_line &&
renderer->last_breakable > 0) {
// copy from last_breakable to remainder
cmark_chunk_set_cstr(&remainder, (char *) renderer->buffer->ptr + renderer->last_breakable + 1);
// truncate at last_breakable
cmark_strbuf_truncate(renderer->buffer, renderer->last_breakable);
// add newline, prefix, and remainder
cmark_strbuf_putc(renderer->buffer, '\n');
cmark_strbuf_put(renderer->buffer, renderer->prefix->ptr,
renderer->prefix->size);
cmark_strbuf_put(renderer->buffer, remainder.data, remainder.len);
renderer->column = renderer->prefix->size + remainder.len;
cmark_chunk_free(&remainder);
renderer->last_breakable = 0;
renderer->begin_line = false;
}
i += len;
}
}
bool trie_search_from_index(trie_t *self, char *text, uint32_t start_node_id, phrase_array **phrases) {
if (text == NULL) return false;
ssize_t len, remaining;
int32_t unich = 0;
unsigned char ch = '\0';
const uint8_t *ptr = (const uint8_t *)text;
const uint8_t *fail_ptr = ptr;
uint32_t node_id = start_node_id;
trie_node_t node = trie_get_node(self, node_id), last_node = node;
uint32_t next_id;
bool match = false;
uint32_t index = 0;
uint32_t phrase_len = 0;
uint32_t phrase_start = 0;
uint32_t data;
trie_search_state_t state = SEARCH_STATE_BEGIN, last_state = SEARCH_STATE_BEGIN;
bool advance_index = true;
while(1) {
len = utf8proc_iterate(ptr, -1, &unich);
remaining = len;
if (len <= 0) return false;
if (!(utf8proc_codepoint_valid(unich))) return false;
int cat = utf8proc_category(unich);
bool is_letter = utf8_is_letter(cat);
// If we're in the middle of a word and the first letter was not a match, skip the word
if (is_letter && state == SEARCH_STATE_NO_MATCH) {
log_debug("skipping\n");
ptr += len;
index += len;
last_state = state;
continue;
}
// Match in the middle of a word
if (is_letter && last_state == SEARCH_STATE_MATCH) {
log_debug("last_state == SEARCH_STATE_MATCH && is_letter\n");
// Only set match to false so we don't callback
match = false;
}
for (int i=0; remaining > 0; remaining--, i++, ptr++, last_node=node, last_state=state, node_id=next_id) {
ch = (unsigned char) *ptr;
log_debug("char=%c\n", ch);
next_id = trie_get_transition_index(self, node, *ptr);
node = trie_get_node(self, next_id);
if (node.check != node_id) {
state = is_letter ? SEARCH_STATE_NO_MATCH : SEARCH_STATE_BEGIN;
if (match) {
log_debug("match is true and state==SEARCH_STATE_NO_MATCH\n");
if (*phrases == NULL) {
*phrases = phrase_array_new_size(1);
}
phrase_array_push(*phrases, (phrase_t){phrase_start, phrase_len, data});
index = phrase_start + phrase_len;
advance_index = false;
// Set the text back to the end of the last phrase
ptr = (const uint8_t *)text + index;
len = utf8proc_iterate(ptr, -1, &unich);
log_debug("ptr=%s\n", ptr);
} else {
ptr += remaining;
log_debug("done with char, now at %s\n", ptr);
}
fail_ptr = ptr;
node_id = start_node_id;
last_node = node = trie_get_node(self, node_id);
phrase_start = phrase_len = 0;
last_state = state;
match = false;
break;
} else {
log_debug("node.check == node_id\n");
state = SEARCH_STATE_PARTIAL_MATCH;
if (last_state == SEARCH_STATE_NO_MATCH || last_state == SEARCH_STATE_BEGIN) {
log_debug("phrase_start=%u\n", index);
phrase_start = index;
fail_ptr = ptr + remaining;
}
if (node.base < 0) {
int32_t data_index = -1*node.base;
trie_data_node_t data_node = self->data->a[data_index];
unsigned char *current_tail = self->tail->a + data_node.tail;
size_t tail_len = strlen((char *)current_tail);
char *query_tail = (char *)(*ptr ? ptr + 1 : ptr);
size_t query_tail_len = strlen((char *)query_tail);
log_debug("next node tail: %s\n", current_tail);
log_debug("query node tail: %s\n", query_tail);
if (tail_len <= query_tail_len && strncmp((char *)current_tail, query_tail, tail_len) == 0) {
state = SEARCH_STATE_MATCH;
log_debug("Tail matches\n");
last_state = state;
data = data_node.data;
log_debug("%u, %d, %zu\n", index, phrase_len, tail_len);
ptr += tail_len;
index += tail_len;
advance_index = false;
phrase_len = index + 1 - phrase_start;
match = true;
} else if (match) {
log_debug("match is true and longer phrase tail did not match\n");
log_debug("phrase_start=%d, phrase_len=%d\n", phrase_start, phrase_len);
if (*phrases == NULL) {
*phrases = phrase_array_new_size(1);
}
phrase_array_push(*phrases, (phrase_t){phrase_start, phrase_len, data});
ptr = fail_ptr;
match = false;
index = phrase_start + phrase_len;
advance_index = false;
}
}
if (ch != '\0') {
trie_node_t terminal_node = trie_get_transition(self, node, '\0');
if (terminal_node.check == next_id) {
log_debug("Transition to NUL byte matched\n");
state = SEARCH_STATE_MATCH;
match = true;
phrase_len = index + (uint32_t)len - phrase_start;
if (terminal_node.base < 0) {
int32_t data_index = -1*terminal_node.base;
trie_data_node_t data_node = self->data->a[data_index];
data = data_node.data;
}
log_debug("Got match with len=%d\n", phrase_len);
fail_ptr = ptr;
}
}
}
}
if (unich == 0) {
if (last_state == SEARCH_STATE_MATCH) {
log_debug("Found match at the end\n");
if (*phrases == NULL) {
*phrases = phrase_array_new_size(1);
}
phrase_array_push(*phrases, (phrase_t){phrase_start, phrase_len, data});
}
break;
}
if (advance_index) index += len;
advance_index = true;
log_debug("index now %u\n", index);
} // while
return true;
}
phrase_t trie_search_prefixes_from_index(trie_t *self, char *word, size_t len, uint32_t start_node_id) {
log_debug("Call to trie_search_prefixes_from_index\n");
uint32_t node_id = start_node_id, last_node_id = node_id;
trie_node_t node = trie_get_node(self, node_id), last_node = node;
log_debug("last_node_id = %d\n", last_node_id);
uint32_t value = 0, phrase_start = 0, phrase_len = 0;
uint8_t *ptr = (uint8_t *)word;
ssize_t char_len = 0;
uint32_t idx = 0;
size_t separator_char_len = 0;
int32_t codepoint = 0;
bool first_char = true;
trie_data_node_t data_node;
trie_node_t terminal_node;
while (idx < len) {
char_len = utf8proc_iterate(ptr, len, &codepoint);
log_debug("char_len = %zu, char=%d\n", char_len, codepoint);
if (char_len <= 0) break;
if (!(utf8proc_codepoint_valid(codepoint))) break;
bool is_hyphen = utf8_is_hyphen(codepoint);
int cat = utf8proc_category(codepoint);
bool is_space = utf8_is_separator(cat);
uint8_t *char_ptr = ptr;
size_t i = 0;
bool skip_char = false;
bool break_out = false;
for (i = 0; i < char_len; i++) {
node_id = trie_get_transition_index(self, last_node, *char_ptr);
node = trie_get_node(self, node_id);
log_debug("At idx=%zu, char=%.*s\n", i, (int)char_len, char_ptr);
if (node.check != last_node_id) {
log_debug("node.check = %d and last_node_id = %d\n", node.check, last_node_id);
if (is_hyphen || (is_space && *ptr != ' ')) {
log_debug("Got hyphen or other separator, trying space instead\n");
node_id = trie_get_transition_index(self, last_node, ' ');
node = trie_get_node(self, node_id);
}
if (is_hyphen && node.check != last_node_id) {
log_debug("No space transition\n");
ptr += char_len;
idx += char_len;
separator_char_len = char_len;
node_id = last_node_id;
node = trie_get_node(self, node_id);
skip_char = true;
break;
} else if (node.check != last_node_id) {
break_out = true;
log_debug("Breaking\n");
break;
}
break;
}
if (first_char) {
phrase_start = idx;
first_char = false;
}
if (node.base < 0) {
log_debug("Searching tail\n");
data_node = trie_get_data_node(self, node);
uint32_t current_tail_pos = data_node.tail;
unsigned char *current_tail = self->tail->a + current_tail_pos;
log_debug("comparing tail: %s vs %s\n", current_tail, char_ptr + 1);
size_t current_tail_len = strlen((char *)current_tail);
size_t match_len = i + 1;
size_t offset = i + 1;
size_t tail_pos = 0;
log_debug("offset=%zu\n", offset);
if (char_len > 1) {
log_debug("char_len = %zu\n", char_len);
log_debug("Doing strncmp: (%zu) %s vs %s\n", char_len - offset, current_tail, char_ptr + 1);
if (strncmp((char *)ptr + offset, (char *)current_tail, char_len - offset) == 0) {
match_len += char_len - offset;
tail_pos = char_len - offset;
log_debug("in char match_len = %zu\n", match_len);
} else {
return NULL_PHRASE;
}
}
size_t tail_match_len = utf8_common_prefix_len((char *)ptr + char_len, (char *)current_tail + tail_pos, current_tail_len - tail_pos);
match_len += tail_match_len;
log_debug("match_len=%zu\n", match_len);
if (tail_match_len == current_tail_len - tail_pos) {
if (first_char) phrase_start = idx;
phrase_len = (uint32_t)(idx + match_len) - phrase_start;
log_debug("tail match! phrase_len=%u\n", phrase_len);
value = data_node.data;
return (phrase_t){phrase_start, phrase_len, value};
} else {
return NULL_PHRASE;
}
} else if (node.check == last_node_id) {
terminal_node = trie_get_transition(self, node, '\0');
log_debug("Trying link from %d to terminal node\n", last_node_id);
if (terminal_node.check == node_id) {
log_debug("Transition to NUL byte matched\n");
if (terminal_node.base < 0) {
phrase_len = (uint32_t)(idx + char_len) - phrase_start;
data_node = trie_get_data_node(self, terminal_node);
value = data_node.data;
}
log_debug("Got match with len=%d\n", phrase_len);
}
}
last_node = node;
last_node_id = node_id;
log_debug("last_node_id = %d\n", last_node_id);
char_ptr++;
}
if (break_out) {
break;
} else if (skip_char) {
continue;
}
log_debug("Incrementing index\n");
idx += char_len;
ptr += char_len;
}
log_debug("exited while loop\n");
if (phrase_len == 0) return NULL_PHRASE;
return (phrase_t) {phrase_start, phrase_len, value};
}
