static ParserElem*
S_consume_text(StringIterator *iter) {
StringIterator *temp = StrIter_Clone(iter);
while (1) {
int32_t code_point = StrIter_Next(temp);
if (code_point == '\\') {
code_point = StrIter_Next(temp);
if (code_point == STR_OOB) {
break;
}
}
else if (code_point == STR_OOB) {
break;
}
else if (StrHelp_is_whitespace(code_point)
|| code_point == '"'
|| code_point == '('
|| code_point == ')'
) {
StrIter_Recede(temp, 1);
break;
}
}
String *text = StrIter_crop(iter, temp);
StrIter_Assign(iter, temp);
DECREF(temp);
return ParserElem_new(TOKEN_STRING, (Obj*)text);
}
static ParserElem*
S_consume_keyword(StringIterator *iter, const char *keyword,
size_t keyword_len, int type) {
if (!StrIter_Starts_With_Utf8(iter, keyword, keyword_len)) {
return NULL;
}
StringIterator *temp = StrIter_Clone(iter);
StrIter_Advance(temp, keyword_len);
int32_t lookahead = StrIter_Next(temp);
if (lookahead == STR_OOB) {
DECREF(temp);
return NULL;
}
if (StrHelp_is_whitespace(lookahead)
|| lookahead == '"'
|| lookahead == '('
|| lookahead == ')'
|| lookahead == '+'
|| lookahead == '-'
) {
StrIter_Recede(temp, 1);
StrIter_Assign(iter, temp);
DECREF(temp);
return ParserElem_new(type, NULL);
}
DECREF(temp);
return NULL;
}
int32_t
Str_Code_Point_From_IMP(String *self, size_t tick) {
if (tick == 0) { return STR_OOB; }
StringIterator *iter = STACK_ITER(self, self->size);
StrIter_Recede(iter, tick - 1);
return StrIter_Prev(iter);
}
static int64_t
S_find(String *string, String *substring) {
StringIterator *iter = Str_Find(string, substring);
if (iter == NULL) { return -1; }
size_t tick = StrIter_Recede(iter, SIZE_MAX);
DECREF(iter);
return (int64_t)tick;
}
static ParserElem*
S_consume_field(StringIterator *iter) {
StringIterator *temp = StrIter_Clone(iter);
// Field names constructs must start with a letter or underscore.
int32_t code_point = StrIter_Next(temp);
if (code_point == STR_OOB) {
DECREF(temp);
return NULL;
}
if (!(isalpha(code_point) || code_point == '_')) {
DECREF(temp);
return NULL;
}
// Only alphanumerics and underscores are allowed in field names.
while (':' != (code_point = StrIter_Next(temp))) {
if (code_point == STR_OOB) {
DECREF(temp);
return NULL;
}
if (!(isalnum(code_point) || code_point == '_')) {
DECREF(temp);
return NULL;
}
}
// Field name constructs must be followed by something sensible.
int32_t lookahead = StrIter_Next(temp);
if (lookahead == STR_OOB) {
DECREF(temp);
return NULL;
}
if (!(isalnum(lookahead)
|| lookahead == '_'
|| lookahead > 127
|| lookahead == '"'
|| lookahead == '('
)
) {
DECREF(temp);
return NULL;
}
// Consume string data.
StrIter_Recede(temp, 2); // Back up over lookahead and colon.
String *field = StrIter_crop(iter, temp);
StrIter_Advance(temp, 1); // Skip colon.
StrIter_Assign(iter, temp);
DECREF(temp);
return ParserElem_new(TOKEN_FIELD, (Obj*)field);
}
Query*
QParser_Expand_Leaf_IMP(QueryParser *self, Query *query) {
QueryParserIVARS *const ivars = QParser_IVARS(self);
LeafQuery *leaf_query = (LeafQuery*)query;
Schema *schema = ivars->schema;
bool is_phrase = false;
bool ambiguous = false;
// Determine whether we can actually process the input.
if (!Query_is_a(query, LEAFQUERY)) { return NULL; }
String *full_text = LeafQuery_Get_Text(leaf_query);
if (!Str_Get_Size(full_text)) { return NULL; }
// If quoted, always generate PhraseQuery.
StringIterator *top = Str_Top(full_text);
StringIterator *tail = Str_Tail(full_text);
StrIter_Skip_Next_Whitespace(top);
StrIter_Skip_Prev_Whitespace(tail);
if (StrIter_Starts_With_Utf8(top, "\"", 1)) {
is_phrase = true;
StrIter_Advance(top, 1);
if (StrIter_Ends_With_Utf8(tail, "\"", 1)
&& !StrIter_Ends_With_Utf8(tail, "\\\"", 2)
) {
StrIter_Recede(tail, 1);
}
}
String *source_text = StrIter_substring(top, tail);
// Either use LeafQuery's field or default to Parser's list.
Vector *fields;
if (LeafQuery_Get_Field(leaf_query)) {
fields = Vec_new(1);
Vec_Push(fields, INCREF(LeafQuery_Get_Field(leaf_query)));
}
else {
fields = (Vector*)INCREF(ivars->fields);
}
CharBuf *unescape_buf = CB_new(Str_Get_Size(source_text));
Vector *queries = Vec_new(Vec_Get_Size(fields));
for (uint32_t i = 0, max = Vec_Get_Size(fields); i < max; i++) {
String *field = (String*)Vec_Fetch(fields, i);
Analyzer *analyzer = ivars->analyzer
? ivars->analyzer
: Schema_Fetch_Analyzer(schema, field);
if (!analyzer) {
Vec_Push(queries,
(Obj*)QParser_Make_Term_Query(self, field,
(Obj*)source_text));
}
else {
// Extract token texts.
String *split_source = S_unescape(self, source_text, unescape_buf);
Vector *maybe_texts = Analyzer_Split(analyzer, split_source);
uint32_t num_maybe_texts = Vec_Get_Size(maybe_texts);
Vector *token_texts = Vec_new(num_maybe_texts);
// Filter out zero-length token texts.
for (uint32_t j = 0; j < num_maybe_texts; j++) {
String *token_text = (String*)Vec_Fetch(maybe_texts, j);
if (Str_Get_Size(token_text)) {
Vec_Push(token_texts, INCREF(token_text));
}
}
if (Vec_Get_Size(token_texts) == 0) {
/* Query might include stop words. Who knows? */
ambiguous = true;
}
// Add either a TermQuery or a PhraseQuery.
if (is_phrase || Vec_Get_Size(token_texts) > 1) {
Vec_Push(queries, (Obj*)
QParser_Make_Phrase_Query(self, field, token_texts));
}
else if (Vec_Get_Size(token_texts) == 1) {
Vec_Push(queries,
(Obj*)QParser_Make_Term_Query(self, field, Vec_Fetch(token_texts, 0)));
}
DECREF(token_texts);
DECREF(maybe_texts);
DECREF(split_source);
}
}
Query *retval;
if (Vec_Get_Size(queries) == 0) {
retval = (Query*)NoMatchQuery_new();
if (ambiguous) {
NoMatchQuery_Set_Fails_To_Match((NoMatchQuery*)retval, false);
}
}
else if (Vec_Get_Size(queries) == 1) {
retval = (Query*)INCREF(Vec_Fetch(queries, 0));
}
else {
retval = QParser_Make_OR_Query(self, queries);
}
// Clean up.
DECREF(unescape_buf);
DECREF(queries);
DECREF(fields);
DECREF(source_text);
DECREF(tail);
DECREF(top);
return retval;
}
Vector*
QueryLexer_Tokenize_IMP(QueryLexer *self, String *query_string) {
QueryLexerIVARS *const ivars = QueryLexer_IVARS(self);
Vector *elems = Vec_new(0);
if (!query_string) { return elems; }
StringIterator *iter = Str_Top(query_string);
while (StrIter_Has_Next(iter)) {
ParserElem *elem = NULL;
if (StrIter_Skip_Whitespace(iter)) {
// Fast-forward past whitespace.
continue;
}
if (ivars->heed_colons) {
ParserElem *elem = S_consume_field(iter);
if (elem) {
Vec_Push(elems, (Obj*)elem);
}
}
int32_t code_point = StrIter_Next(iter);
switch (code_point) {
case '(':
elem = ParserElem_new(TOKEN_OPEN_PAREN, NULL);
break;
case ')':
elem = ParserElem_new(TOKEN_CLOSE_PAREN, NULL);
break;
case '+':
if (StrIter_Has_Next(iter)
&& !StrIter_Skip_Whitespace(iter)
) {
elem = ParserElem_new(TOKEN_PLUS, NULL);
}
else {
elem = ParserElem_new(TOKEN_STRING, (Obj*)Str_newf("+"));
}
break;
case '-':
if (StrIter_Has_Next(iter)
&& !StrIter_Skip_Whitespace(iter)
) {
elem = ParserElem_new(TOKEN_MINUS, NULL);
}
else {
elem = ParserElem_new(TOKEN_STRING, (Obj*)Str_newf("-"));
}
break;
case '"':
StrIter_Recede(iter, 1);
elem = S_consume_quoted_string(iter);
break;
case 'O':
StrIter_Recede(iter, 1);
elem = S_consume_keyword(iter, "OR", 2, TOKEN_OR);
if (!elem) {
elem = S_consume_text(iter);
}
break;
case 'A':
StrIter_Recede(iter, 1);
elem = S_consume_keyword(iter, "AND", 3, TOKEN_AND);
if (!elem) {
elem = S_consume_text(iter);
}
break;
case 'N':
StrIter_Recede(iter, 1);
elem = S_consume_keyword(iter, "NOT", 3, TOKEN_NOT);
if (!elem) {
elem = S_consume_text(iter);
}
break;
default:
StrIter_Recede(iter, 1);
elem = S_consume_text(iter);
break;
}
Vec_Push(elems, (Obj*)elem);
}
DECREF(iter);
return elems;
}
static void
test_iterator_substring(TestBatchRunner *runner) {
String *string = Str_newf("a%sb%sc%sd", smiley, smiley, smiley);
StringIterator *start = Str_Top(string);
StringIterator *end = Str_Tail(string);
{
String *substring = StrIter_crop(start, end);
TEST_TRUE(runner, Str_Equals(substring, (Obj*)string),
"StrIter_crop whole string");
DECREF(substring);
}
StrIter_Advance(start, 2);
StrIter_Recede(end, 2);
{
String *substring = StrIter_crop(start, end);
String *wanted = Str_newf("b%sc", smiley);
TEST_TRUE(runner, Str_Equals(substring, (Obj*)wanted),
"StrIter_crop");
TEST_TRUE(runner, StrIter_Starts_With(start, wanted),
"Starts_With returns true");
TEST_TRUE(runner, StrIter_Ends_With(end, wanted),
"Ends_With returns true");
DECREF(wanted);
DECREF(substring);
}
{
String *short_str = Str_newf("b%sx", smiley);
TEST_FALSE(runner, StrIter_Starts_With(start, short_str),
"Starts_With returns false");
TEST_FALSE(runner, StrIter_Ends_With(start, short_str),
"Ends_With returns false");
String *long_str = Str_newf("b%sxxxxxxxxxxxx%sc", smiley, smiley);
TEST_FALSE(runner, StrIter_Starts_With(start, long_str),
"Starts_With long string returns false");
TEST_FALSE(runner, StrIter_Ends_With(end, long_str),
"Ends_With long string returns false");
DECREF(short_str);
DECREF(long_str);
}
{
String *substring = StrIter_crop(end, NULL);
String *wanted = Str_newf("%sd", smiley);
TEST_TRUE(runner, Str_Equals(substring, (Obj*)wanted),
"StrIter_crop with NULL tail");
DECREF(wanted);
DECREF(substring);
}
{
String *substring = StrIter_crop(NULL, start);
String *wanted = Str_newf("a%s", smiley);
TEST_TRUE(runner, Str_Equals(substring, (Obj*)wanted),
"StrIter_crop with NULL top");
DECREF(wanted);
DECREF(substring);
}
DECREF(start);
DECREF(end);
DECREF(string);
}
static void
test_iterator(TestBatchRunner *runner) {
static const int32_t code_points[] = {
0x41,
0x7F,
0x80,
0x7FF,
0x800,
0xFFFF,
0x10000,
0x10FFFF
};
static size_t num_code_points
= sizeof(code_points) / sizeof(code_points[0]);
CharBuf *buf = CB_new(0);
for (size_t i = 0; i < num_code_points; ++i) {
CB_Cat_Char(buf, code_points[i]);
}
String *string = CB_To_String(buf);
{
StringIterator *iter = Str_Top(string);
TEST_TRUE(runner, StrIter_Equals(iter, (Obj*)iter),
"StringIterator equal to self");
TEST_FALSE(runner, StrIter_Equals(iter, (Obj*)CFISH_TRUE),
"StringIterator not equal non-iterators");
DECREF(iter);
}
{
StringIterator *top = Str_Top(string);
StringIterator *tail = Str_Tail(string);
TEST_INT_EQ(runner, StrIter_Compare_To(top, (Obj*)tail), -1,
"Compare_To top < tail");
TEST_INT_EQ(runner, StrIter_Compare_To(tail, (Obj*)top), 1,
"Compare_To tail > top");
TEST_INT_EQ(runner, StrIter_Compare_To(top, (Obj*)top), 0,
"Compare_To top == top");
StringIterator *clone = StrIter_Clone(top);
TEST_TRUE(runner, StrIter_Equals(clone, (Obj*)top), "Clone");
StrIter_Assign(clone, tail);
TEST_TRUE(runner, StrIter_Equals(clone, (Obj*)tail), "Assign");
String *other = Str_newf("Other string");
StringIterator *other_iter = Str_Top(other);
TEST_FALSE(runner, StrIter_Equals(other_iter, (Obj*)tail),
"Equals returns false for different strings");
StrIter_Assign(clone, other_iter);
TEST_TRUE(runner, StrIter_Equals(clone, (Obj*)other_iter),
"Assign iterator with different string");
DECREF(other);
DECREF(other_iter);
DECREF(clone);
DECREF(top);
DECREF(tail);
}
{
StringIterator *iter = Str_Top(string);
for (size_t i = 0; i < num_code_points; ++i) {
TEST_TRUE(runner, StrIter_Has_Next(iter), "Has_Next %d", i);
int32_t code_point = StrIter_Next(iter);
TEST_INT_EQ(runner, code_point, code_points[i], "Next %d", i);
}
TEST_TRUE(runner, !StrIter_Has_Next(iter),
"Has_Next at end of string");
TEST_INT_EQ(runner, StrIter_Next(iter), STR_OOB,
"Next at end of string");
StringIterator *tail = Str_Tail(string);
TEST_TRUE(runner, StrIter_Equals(iter, (Obj*)tail), "Equals tail");
DECREF(tail);
DECREF(iter);
}
{
StringIterator *iter = Str_Tail(string);
for (size_t i = num_code_points; i--;) {
TEST_TRUE(runner, StrIter_Has_Prev(iter), "Has_Prev %d", i);
int32_t code_point = StrIter_Prev(iter);
TEST_INT_EQ(runner, code_point, code_points[i], "Prev %d", i);
}
TEST_TRUE(runner, !StrIter_Has_Prev(iter),
"Has_Prev at end of string");
TEST_INT_EQ(runner, StrIter_Prev(iter), STR_OOB,
"Prev at start of string");
StringIterator *top = Str_Top(string);
TEST_TRUE(runner, StrIter_Equals(iter, (Obj*)top), "Equals top");
DECREF(top);
DECREF(iter);
}
{
StringIterator *iter = Str_Top(string);
StrIter_Next(iter);
TEST_INT_EQ(runner, StrIter_Advance(iter, 2), 2,
"Advance returns number of code points");
TEST_INT_EQ(runner, StrIter_Next(iter), code_points[3],
"Advance works");
TEST_INT_EQ(runner,
StrIter_Advance(iter, 1000000), num_code_points - 4,
"Advance past end of string");
StrIter_Prev(iter);
TEST_INT_EQ(runner, StrIter_Recede(iter, 2), 2,
"Recede returns number of code points");
TEST_INT_EQ(runner, StrIter_Prev(iter), code_points[num_code_points-4],
"Recede works");
TEST_INT_EQ(runner, StrIter_Recede(iter, 1000000), num_code_points - 4,
"Recede past start of string");
DECREF(iter);
}
DECREF(string);
DECREF(buf);
}
// Find an ending boundary before the current position given by the iterator.
// Skip up to max_skip code points plus potential whitespace. Update the
// iterator and return number of code points skipped. Return true if a
// ending edge (sentence) was found.
bool
S_find_ending_boundary(StringIterator *tail, uint32_t max_skip,
uint32_t *num_skipped_ptr) {
int32_t code_point;
// Check if we're at an ending boundary already. Don't check for a word
// boundary because we need space for a trailing ellipsis.
StringIterator *iter = StrIter_Clone(tail);
do {
code_point = StrIter_Next(iter);
if (code_point == STRITER_DONE) {
// Skip remaining whitespace.
*num_skipped_ptr = StrIter_Skip_Prev_Whitespace(tail);
DECREF(iter);
return true;
}
} while (StrHelp_is_whitespace(code_point));
// Keep track of the first word boundary.
StringIterator *word = NULL;
uint32_t word_offset = 0;
StrIter_Assign(iter, tail);
for (uint32_t i = 0;
STRITER_DONE != (code_point = StrIter_Prev(iter));
++i)
{
if (code_point == '.') {
StrIter_Assign(tail, iter);
StrIter_Advance(tail, 1); // Include period.
*num_skipped_ptr = i;
DECREF(word);
DECREF(iter);
return true;
}
if (StrHelp_is_whitespace(code_point)) {
if (word == NULL) {
word = StrIter_Clone(iter);
word_offset = i + 1;
}
}
else if (i >= max_skip) {
// Break only at non-whitespace to allow another sentence
// boundary to be found.
break;
}
}
if (word == NULL) {
// Make space for ellipsis.
*num_skipped_ptr = StrIter_Recede(tail, 1);
}
else {
// Use word boundary if no sentence boundary was found.
StrIter_Assign(tail, word);
// Strip whitespace and punctuation that collides with an ellipsis.
while (STRITER_DONE != (code_point = StrIter_Prev(tail))) {
if (!StrHelp_is_whitespace(code_point)
&& code_point != '.'
&& code_point != ','
&& code_point != ';'
&& code_point != ':'
&& code_point != ':'
&& code_point != '?'
&& code_point != '!'
) {
StrIter_Advance(tail, 1); // Back up.
break;
}
++word_offset;
}
*num_skipped_ptr = word_offset;
}
DECREF(word);
DECREF(iter);
return false;
}
