Update keywords (#15443) (#15593)

pingcap · Dec 5, 2023 · 49d2067 · 49d2067
1 parent 887ecd1
commit 49d2067
Show file tree

Hide file tree

Showing 2 changed files with 69 additions and 37 deletions.
diff --git a/keywords.md b/keywords.md
@@ -66,7 +66,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - ACCOUNT
 - ACTION
 - ADD (R)
-- ADMIN (R)
+- ADMIN
 - ADVISE
 - AFTER
 - AGAINST
@@ -96,6 +96,8 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - BACKEND
 - BACKUP
 - BACKUPS
+- BATCH
+- BDR
 - BEGIN
 - BERNOULLI
 - BETWEEN (R)
@@ -112,8 +114,8 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - BOOLEAN
 - BOTH (R)
 - BTREE
-- BUCKETS (R)
-- BUILTINS (R)
+- BUCKETS
+- BUILTINS
 - BY (R)
 - BYTE
 
@@ -122,8 +124,9 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - CACHE
 - CALIBRATE
 - CALL (R)
-- CANCEL (R)
+- CANCEL
 - CAPTURE
+- CARDINALITY
 - CASCADE (R)
 - CASCADED
 - CASE (R)
@@ -143,12 +146,13 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - CLOSE
 - CLUSTER
 - CLUSTERED
-- CMSKETCH (R)
+- CMSKETCH
 - COALESCE
 - COLLATE (R)
 - COLLATION
 - COLUMN (R)
 - COLUMN_FORMAT
+- COLUMN_STATS_USAGE
 - COLUMNS
 - COMMENT
 - COMMIT
@@ -165,6 +169,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - CONTEXT
 - CONTINUE (R)
 - CONVERT (R)
+- CORRELATION
 - CPU
 - CREATE (R)
 - CROSS (R)
@@ -197,7 +202,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - DAY_MICROSECOND (R)
 - DAY_MINUTE (R)
 - DAY_SECOND (R)
-- DDL (R)
+- DDL
 - DEALLOCATE
 - DECIMAL (R)
 - DECLARE
@@ -207,7 +212,8 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - DELAYED (R)
 - DELETE (R)
 - DENSE_RANK (R-Window)
-- DEPTH (R)
+- DEPENDENCY
+- DEPTH
 - DESC (R)
 - DESCRIBE (R)
 - DIGEST
@@ -221,8 +227,9 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - DIV (R)
 - DO
 - DOUBLE (R)
-- DRAINER (R)
+- DRAINER
 - DROP (R)
+- DRY
 - DUAL (R)
 - DUPLICATE
 - DYNAMIC
@@ -302,6 +309,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - HELP
 - HIGH_PRIORITY (R)
 - HISTOGRAM
+- HISTOGRAMS_IN_FLIGHT
 - HISTORY
 - HOSTS
 - HOUR
@@ -350,8 +358,8 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 
 <a id="J" class="letter" href="#J">J</a>
 
-- JOB (R)
-- JOBS (R)
+- JOB
+- JOBS
 - JOIN (R)
 - JSON
 
@@ -384,6 +392,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - LIST
 - LOAD (R)
 - LOCAL
+- LOCAL_ONLY
 - LOCALTIME (R)
 - LOCALTIMESTAMP (R)
 - LOCATION
@@ -439,8 +448,8 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - NOCACHE
 - NOCYCLE
 - NODEGROUP
-- NODE_ID (R)
-- NODE_STATE (R)
+- NODE_ID
+- NODE_STATE
 - NOMAXVALUE
 - NOMINVALUE
 - NONCLUSTERED
@@ -468,7 +477,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - ONLINE
 - ONLY
 - OPEN
-- OPTIMISTIC (R)
+- OPTIMISTIC
 - OPTIMIZE (R)
 - OPTION (R)
 - OPTIONAL
@@ -496,7 +505,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - PERCENT_RANK (R-Window)
 - PER_DB
 - PER_TABLE
-- PESSIMISTIC (R)
+- PESSIMISTIC
 - PLACEMENT (S)
 - PLUGINS
 - POINT
@@ -514,7 +523,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - PROFILE
 - PROFILES
 - PROXY
-- PUMP (R)
+- PUMP
 - PURGE
 
 <a id="Q" class="letter" href="#Q">Q</a>
@@ -537,8 +546,8 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - REDUNDANT
 - REFERENCES (R)
 - REGEXP (R)
-- REGION (R)
-- REGIONS (R)
+- REGION
+- REGIONS
 - RELEASE (R)
 - RELOAD
 - REMOVE
@@ -553,6 +562,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - REPLICATION
 - REQUIRE (R)
 - REQUIRED
+- RESET
 - RESOURCE
 - RESPECT
 - RESTART
@@ -575,14 +585,17 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - ROW_NUMBER (R-Window)
 - ROWS (R-Window)
 - RTREE
+- RUN
 
 <a id="S" class="letter" href="#S">S</a>
 
-- SAMPLES (R)
+- SAMPLERATE
+- SAMPLES
 - SAN
 - SAVEPOINT
 - SECOND
 - SECOND_MICROSECOND (R)
+- SECONDARY
 - SECONDARY_ENGINE
 - SECONDARY_LOAD
 - SECONDARY_UNLOAD
@@ -594,6 +607,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - SERIAL
 - SERIALIZABLE
 - SESSION
+- SESSION_STATES
 - SET (R)
 - SETVAL
 - SHARD_ROW_ID_BITS
@@ -612,7 +626,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - SOME
 - SOURCE
 - SPATIAL (R)
-- SPLIT (R)
+- SPLIT
 - SQL (R)
 - SQL_BIG_RESULT (R)
 - SQL_BUFFER_RESULT
@@ -634,19 +648,22 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - SSL (R)
 - START
 - STARTING (R)
-- STATS (R)
+- STATISTICS
+- STATS
 - STATS_AUTO_RECALC
-- STATS_BUCKETS (R)
+- STATS_BUCKETS
 - STATS_COL_CHOICE
 - STATS_COL_LIST
 - STATS_EXTENDED (R)
-- STATS_HEALTHY (R)
-- STATS_HISTOGRAMS (R)
-- STATS_META (R)
+- STATS_HEALTHY
+- STATS_HISTOGRAMS
+- STATS_LOCKED
+- STATS_META
 - STATS_OPTIONS
 - STATS_PERSISTENT
 - STATS_SAMPLE_PAGES
 - STATS_SAMPLE_RATE
+- STATS_TOPN
 - STATUS
 - STORAGE
 - STORED (R)
@@ -668,15 +685,17 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - TABLESAMPLE (R)
 - TABLESPACE
 - TABLE_CHECKSUM
+- TELEMETRY
+- TELEMETRY_ID
 - TEMPORARY
 - TEMPTABLE
 - TERMINATED (R)
 - TEXT
 - THAN
 - THEN (R)
-- TIDB (R)
+- TIDB
 - TiDB_CURRENT_TSO (R)
-- TIFLASH (R)
+- TIFLASH
 - TIKV_IMPORTER
 - TIME
 - TIMESTAMP
@@ -685,7 +704,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - TINYTEXT (R)
 - TO (R)
 - TOKEN_ISSUER
-- TOPN (R)
+- TOPN
 - TPCC
 - TPCH_10
 - TRACE
@@ -696,6 +715,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - TRIGGERS
 - TRUE (R)
 - TRUNCATE
+- TSO
 - TTL
 - TTL_ENABLE
 - TTL_JOB_INTERVAL
@@ -745,7 +765,7 @@ The following list shows the keywords in TiDB. Reserved keywords are marked with
 - WHEN (R)
 - WHERE (R)
 - WHILE (R)
-- WIDTH (R)
+- WIDTH
 - WINDOW (R-Window)
 - WITH (R)
 - WITHOUT

diff --git a/scripts/check-keywords.sh b/scripts/check-keywords.sh
@@ -16,33 +16,45 @@ keywords = kwdocs.read_text()
 errors = 0
 section = "Unknown"
 for line in parser.read_text().split("\n"):
-    if line.find("The following tokens belong to ReservedKeyword") >= 0:
+    if line == "":
+        section = "NotKeywordToken"
+
+    elif line.find("The following tokens belong to ReservedKeyword") >= 0:
         section = "ReservedKeyword"
 
-    if line.find("The following tokens belong to UnReservedKeyword") >= 0:
+    elif line.find("The following tokens belong to UnReservedKeyword") >= 0:
         section = "UnReservedKeyword"
 
-    if line.find("The following tokens belong to NotKeywordToken") >= 0:
+    elif line.find("The following tokens belong to TiDBKeyword") >= 0:
+        section = "TiDBKeyword"
+
+    elif line.find("The following tokens belong to NotKeywordToken") >= 0:
         section = "NotKeywordToken"
 
     if section == "ReservedKeyword":
         if m := re.match(r'^\t\w+\s+"(\w+)"$', line):
             kw = m.groups()[0]
             if not (
-                kwm := re.search(f"^- {kw} \((R|R-Window)\)$", keywords, re.MULTILINE)
+                kwm := re.search(f"^- {kw} \\((R|R-Window)\\)$", keywords, re.MULTILINE)
             ):
                 if kwm := re.search(f"^- {kw}$", keywords, re.MULTILINE):
-                    print(f"Keyword not labeled as reserved: {kw}")
-                    errors += 1
+                    print(f"Reserved keyword not labeled as reserved: {kw}")
                 else:
                     print(f"Missing docs for reserved keyword: {kw}")
-                    errors += 1
+                errors += 1
 
-    if section == "UnReservedKeyword":
+    if section in ["UnReservedKeyword", "TiDBKeyword"]:
         if m := re.match(r'^\t\w+\s+"(\w+)"$', line):
             kw = m.groups()[0]
             if not (kwm := re.search(f"^- {kw}$", keywords, re.MULTILINE)):
-                print(f"Missing docs for non-reserved keyword: {kw}")
+                if kwm := re.search(
+                    f"^- {kw} \\((R|R-Window)\\)$", keywords, re.MULTILINE
+                ):
+                    print(
+                        f"Non-reserved keyword from {section} labeled as reserved: {kw}"
+                    )
+                else:
+                    print(f"Missing docs for non-reserved keyword from {section}: {kw}")
                 errors += 1
 
 sys.exit(errors)