@@ -118,21 +118,21 @@ class GrammarMatcherForTokenMaskCache : public EarleyParser {
118118 );
119119
120120 /* ! \brief Check each token in a given interval.
121- \param tokenizer_info The tokenizer info.
122- \param interval The interval to check.
123- \param speculative_calculation_applied Whether to use speculative calculation.
124- \param speculative_mask The speculative mask for speculative calculation.
125- \param definite_accepted_bitset The definite accepted bitset for TagDispatch rules.
126- \param is_root_rule Whether to consider the parent rule. If false, there will be
127- no uncertain tokens. Useful for the root rule.
128- \param fill_reject_indices Whether to fill the rejected indices.
129- \param last_rejected_range The last rejected subtree range. If a token's index is less than
130- this value, it will be rejected directly.
131- \param prev_token The previous token parsed in the parser.
132- \param prev_matched_size The matched size of the previous token.
133- \return True if the rejected indices are filled as usual, False otherwise.
134- \note All the tokens in the given interval will be divided into accepted, rejected and
135- uncertain tokens.
121+ * \param tokenizer_info The tokenizer info.
122+ * \param interval The interval to check.
123+ * \param speculative_calculation_applied Whether to use speculative calculation.
124+ * \param speculative_mask The speculative mask for speculative calculation.
125+ * \param definite_accepted_bitset The definite accepted bitset for TagDispatch rules.
126+ * \param is_root_rule Whether to consider the parent rule. If false, there will be
127+ * no uncertain tokens. Useful for the root rule.
128+ * \param fill_reject_indices Whether to fill the rejected indices.
129+ * \param last_rejected_range The last rejected subtree range. If a token's index is less than
130+ * this value, it will be rejected directly.
131+ * \param prev_token The previous token parsed in the parser.
132+ * \param prev_matched_size The matched size of the previous token.
133+ * \return True if the rejected indices are filled as usual, False otherwise.
134+ * \note All the tokens in the given interval will be divided into accepted, rejected and
135+ * uncertain tokens.
136136 */
137137 bool CheckTokensInInterval (
138138 const TokenizerInfo& tokenizer_info,
@@ -148,11 +148,11 @@ class GrammarMatcherForTokenMaskCache : public EarleyParser {
148148 );
149149
150150 /* ! \brief Apply speculative calculation for a token.
151- \param token The token to check.
152- \param index The index of the token in the vocabulary.
153- \param speculative_mask The speculative mask for speculative calculation.
154- \param definite_accepted_bitset The definite accepted bitset for TagDispatch rules.
155- \return True if the token is accepted by speculative calculation, False otherwise.
151+ * \param token The token to check.
152+ * \param index The index of the token in the vocabulary.
153+ * \param speculative_mask The speculative mask for speculative calculation.
154+ * \param definite_accepted_bitset The definite accepted bitset for TagDispatch rules.
155+ * \return True if the token is accepted by speculative calculation, False otherwise.
156156 */
157157 bool ApplySpeculativeCalculation (
158158 const std::string& token,
@@ -162,10 +162,10 @@ class GrammarMatcherForTokenMaskCache : public EarleyParser {
162162 );
163163
164164 /* ! \brief Find the common prefix size with the previous token.
165- \param token The current token.
166- \param prev_token The previous token.
167- \param prev_matched_size The matched size of the previous token.
168- \param accepted Whether the current token is accepted.
165+ * \param token The current token.
166+ * \param prev_token The previous token.
167+ * \param prev_matched_size The matched size of the previous token.
168+ * \param accepted Whether the current token is accepted.
169169 */
170170 void FindCommonPrefixWithPreviousToken (
171171 const std::string& token,
@@ -174,21 +174,21 @@ class GrammarMatcherForTokenMaskCache : public EarleyParser {
174174 bool * accepted
175175 );
176176
177- // The id of the initial rule.
177+ /* ! \brief The id of the initial rule. */
178178 int32_t init_rule_id;
179179
180- // The initial state of the parser.
180+ /* ! \brief The initial state of the parser. */
181181 ParserState initial_state;
182182
183183 /* !
184- \brief This is a mapping from TagDispatch rule id to the bitset used for second slicing.
185- \note If a rule is a TagDispatch rule, then there will be an AC automaton for its triggers.
186- Which means that it can accept a lot of tokens. However, it will be slow to check a lot of
187- tokens. The DynamicBitset here is used to do a second slicing: if a token's substr(1, n - 1)
188- can be accepted by the start state of the AC automaton, then it will be True in the bitset.
189- When we check a token, we first check if its first character can transit to the start state.
190- If yes, then we check if it is in the bitset. If yes, then we accept it directly.
191- */
184+ * \brief This is a mapping from TagDispatch rule id to the bitset used for second slicing.
185+ * \note If a rule is a TagDispatch rule, then there will be an AC automaton for its triggers.
186+ * Which means that it can accept a lot of tokens. However, it will be slow to check a lot of
187+ * tokens. The DynamicBitset here is used to do a second slicing: if a token's substr(1, n - 1)
188+ * can be accepted by the start state of the AC automaton, then it will be True in the bitset.
189+ * When we check a token, we first check if its first character can transit to the start state.
190+ * If yes, then we check if it is in the bitset. If yes, then we accept it directly.
191+ */
192192 const std::unordered_map<int32_t , DynamicBitset>& tag_dispatch_rule_id_to_second_slicing_bitset;
193193
194194 // Temporary data for GetAdaptiveTokenMask.
@@ -802,7 +802,6 @@ class GrammarCompilerNoCache {
802802 const TokenizerInfo tokenizer_info_;
803803 /* ! \brief The maximum number of threads to use. */
804804 const int max_threads_;
805- /* ! \brief Mapping from the rule_id to the definite accepted token mask. */
806805};
807806
808807CompiledGrammar GrammarCompilerNoCache::MultiThreadCompileGrammar (Grammar grammar_unoptimized) {
0 commit comments