MDEV-32578 row_merge_fts_doc_tokenize() handles parser plugin inconsi…

…stently

When mysql/mysql-server@0c954c2
added a plugin interface for FULLTEXT INDEX tokenization to MySQL 5.7,
fts_tokenize_ctx::processed_len got a second meaning, which is only
partly implemented in row_merge_fts_doc_tokenize().

This inconsistency could cause a crash when using FULLTEXT...WITH PARSER.
A test case that would crash MySQL 8.0 when using an n-gram parser and
single-character words would fail to crash in MySQL 5.7, because the
buf_full condition in row_merge_fts_doc_tokenize() was not met.

This change is inspired by
mysql/mysql-server@38e9a07
that appeared in MySQL 5.7.44.

storage/innobase/include/row0ftsort.h

@@ -108,7 +108,10 @@ typedef UT_LIST_BASE_NODE_T(row_fts_token_t)     fts_token_list_t;
 
 /** Structure stores information from string tokenization operation */
 struct fts_tokenize_ctx {
-	ulint			processed_len;  /*!< processed string length */
+	/** the processed string length in bytes
+	(when using the built-in tokenizer),
+	or the number of row_merge_fts_doc_tokenize_by_parser() calls */
+	ulint			processed_len;
 	ulint			init_pos;       /*!< doc start position */
 	ulint			buf_used;       /*!< the sort buffer (ID) when
 						tokenization stops, which
@@ -119,6 +122,7 @@ struct fts_tokenize_ctx {
 	ib_rbt_t*		cached_stopword;/*!< in: stopword list */
 	dfield_t		sort_field[FTS_NUM_FIELDS_SORT];
 						/*!< in: sort field */
+	/** parsed tokens (when using an external parser) */
 	fts_token_list_t	fts_token_list;
 
 	fts_tokenize_ctx() :

storage/innobase/row/row0ftsort.cc

@@ -506,7 +506,10 @@ row_merge_fts_doc_tokenize(
 
 	/* Tokenize the data and add each word string, its corresponding
 	doc id and position to sort buffer */
-	while (t_ctx->processed_len < doc->text.f_len) {
+	while (parser
+               ? (!t_ctx->processed_len
+                  || UT_LIST_GET_LEN(t_ctx->fts_token_list))
+               : t_ctx->processed_len < doc->text.f_len) {
 		ulint		idx = 0;
 		ulint		cur_len;
 		doc_id_t	write_doc_id;
@@ -847,7 +850,8 @@ DECLARE_THREAD(fts_parallel_tokenization)(
 			/* Not yet finish processing the "doc" on hand,
 			continue processing it */
 			ut_ad(doc.text.f_str);
-			ut_ad(t_ctx.processed_len < doc.text.f_len);
+			ut_ad(buf[0]->index->parser
+			      || t_ctx.processed_len < doc.text.f_len);
 		}
 
 		processed = row_merge_fts_doc_tokenize(
@@ -857,7 +861,8 @@ DECLARE_THREAD(fts_parallel_tokenization)(
 
 		/* Current sort buffer full, need to recycle */
 		if (!processed) {
-			ut_ad(t_ctx.processed_len < doc.text.f_len);
+			ut_ad(buf[0]->index->parser
+			      || t_ctx.processed_len < doc.text.f_len);
 			ut_ad(t_ctx.rows_added[t_ctx.buf_used]);
 			break;
 		}