Update docspell check

utils/check-style/aspell-ignore/en/caseinsensitive.txt

@@ -37,5 +37,19 @@ clickstream
 cmake
 codebase
 codec
-config
+configs?
 cpp
+csv
+filesystem
+json[a-z]*
+latencies
+ng
+noop
+nullable
+num
+obfuscator
+preemptable
+substring
+tabseparated[a-z]*
+utils?
+whitespaces?

utils/check-style/aspell-ignore/en/casesensitive.txt

@@ -1,20 +1,218 @@
 ACLs
-AMQP
-APL
-ASLR
-AST
 ASan
 Actian
 AddressSanitizer
 AppleClang
 ArrowStream
 AvroConfluent
-BMI
-CCTOOLS
-CFQ
 CLion
 CMakeLists
+CPUs
+CSVWithNames
+CSVWithNamesAndTypes
 CamelCase
 CentOS
+Contrib
+Ctrl
+CustomSeparated
+CustomSeparatedWithNames
+CustomSeparatedWithNamesAndTypes
+DBMSs
+DateTime
+DockerHub
+Doxygen
+Encodings
+Enum
+Eoan
+Fuzzer
+Fuzzers
+GTest
+Gb
+Gcc
+GoogleTest
+HDDs
+Heredoc
+Homebrew
+Homebrew's
+Hostname
+IPv
+IntN
+Integrations
+LowCardinality
+MacOS
+Memcheck
+MemorySanitizer
+MergeTree
+MessagePack
+MiB
+Ok
+OpenSUSE
+OpenStack
+OpenTelemetry
+Postgres
+Precompiled
+PrettyCompact
+PrettyCompactMonoBlock
+PrettyCompactNoEscapes
+PrettyNoEscapes
+PrettySpace
+PrettySpaceNoEscapes
+Protobuf
+ProtobufSingle
+QTCreator
+RawBLOB
+RedHat
+RowBinary
+RowBinaryWithNames
+RowBinaryWithNamesAndTypes
+Runtime
+Stateful
+Submodules
+Subqueries
+TSVRaw
+TSan
+TargetSpecific
+TemplateIgnoreSpaces
+Testflows
+Tgz
+Toolset
+Tradeoff
+UBSan
+UInt
+UIntN
+UPDATEs
+Uint
+Updatable
+Valgrind
+Vectorized
+VirtualBox
+Werror
+Woboq
+WriteBuffer
+WriteBuffers
+XCode
+Zipkin
+ZooKeeper
+ZooKeeper's
 blockSize
 camelCase
+coroutines
+cppkafka
+cpu
+cronjob
+csvwithnames
+csvwithnamesandtypes
+customseparated
+customseparatedwithnames
+dbms
+ddl
+ok
+postfix
+postfixes
+postgresql
+pre
+prebuild
+prebuilt
+preloaded
+preprocessed
+preprocessor
+prestable
+prettycompact
+prettycompactmonoblock
+prettycompactnoescapes
+prettynoescapes
+prettyspace
+prettyspacenoescapes
+prlimit
+prometheus
+proto
+protobuf
+protobufsingle
+psql
+ptrs
+py
+rapidjson
+rawblob
+readahead
+readline
+readme
+readonly
+rebalanced
+replxx
+repo
+rethrow
+risc
+ro
+rocksdb
+rowNumberInBlock
+rowbinary
+rowbinarywithnames
+rowbinarywithnamesandtypes
+rsync
+runningAccumulate
+russian
+rw
+simdjson
+sql
+src
+stacktraces
+statbox
+stateful
+stderr
+stdin
+stdout
+strtod
+strtoll
+strtoull
+structs
+subdirectories
+subexpressions
+submodule
+submodules
+subpattern
+subpatterns
+subqueries
+subquery
+subseconds
+subtree
+sudo
+symlink
+symlinks
+syntaxes
+systemd
+tcp
+templateignorespaces
+tgz
+th
+tmp
+tokenization
+toml
+toolset
+tskv
+tsv
+tui
+turbostat
+txt
+unary
+unencrypted
+unixodbc
+url
+userspace
+usr
+variadic
+varint
+vectorized
+wchc
+wchs
+webpage
+webserver
+wget
+wrt
+xcode
+xml
+xz
+zLib
+zkcopy
+zlib
+znodes
+zstd

utils/check-style/aspell-ignore/en/todo.txt

@@ -1,227 +1,27 @@
-CPUs
-CSV
-CSVWithNames
-CSVWithNamesAndTypes
-Config
-Configs
-Contrib
-Ctrl
-CustomSeparated
-CustomSeparatedWithNames
-CustomSeparatedWithNamesAndTypes
-DBMSs
-DDL
-DNS
-DSF
-DateTime
-DockerHub
-Dockerhub
-Doxygen
-ECC
-ENUM
-EOF
-Encodings
-Enum
-Eoan
-Filesystem
 FixedString
 FreeBSD
-Fuzzer
-Fuzzers
-GDB
-GLIBC
-GTest
-Gb
-Gcc
-Git's
-GoogleTest
-HDDs
-HDFS
-HTTPS
-Heredoc
-Homebrew
-Homebrew's
-Hostname
-IEC
-IPv
-IntN
-Integrations
-JDBC
-JSON
-JSONAsString
-JSONColumns
-JSONColumnsWithMetadata
-JSONCompact
-JSONCompactColumns
-JSONCompactEachRow
-JSONCompactEachRowWithNames
-JSONCompactEachRowWithNamesAndTypes
-JSONCompactStrings
-JSONCompactStringsEachRow
-JSONCompactStringsEachRowWithNames
-JSONCompactStringsEachRowWithNamesAndTypes
-JSONEachRow
-JSONEachRowWithProgress
-JSONStrings
-JSONStringsEachRow
-JSONStringsEachRowWithProgress
-JSONs
-JVM
 Jaeger
 Jemalloc
 Jepsen
 KDevelop
 Kx
-LEB
-LF
-LGPL
-LOCALTIME
-LOCALTIMESTAMP
-LSM
-LVM
 LibFuzzer
 LineAsString
-LowCardinality
-MEMTABLE
-MF
 MSan
-MULTITARGET
-MacOS
-Memcheck
-MemorySanitizer
-MergeTree
-MessagePack
-MiB
 MsgPack
 Multiline
 Multithreading
 MySQLDump
-NCQ
-NEKUDOTAYIM
-NNNNNNNNNN
-NULLIF
-NVME
 NuRaft
-Nullable
-OBFUSCATOR
-ODBC
-OSS
-Ok
-OpenSUSE
-OpenStack
-OpenTelemetry
-PAAMAYIM
-PCLMULQDQ
-POCO
-POPCNT
 Parsers
-Postgres
-Precompiled
-PrettyCompact
-PrettyCompactMonoBlock
-PrettyCompactNoEscapes
-PrettyNoEscapes
-PrettySpace
-PrettySpaceNoEscapes
-Protobuf
-ProtobufSingle
-QTCreator
-RBAC
-README
-RawBLOB
-RedHat
-RowBinary
-RowBinaryWithNames
-RowBinaryWithNamesAndTypes
-Runtime
-SATA
-SDK
-SERIALIZABLE
-SHA
-SIMD
-SMALLINT
-SQLSTATE
-SSD
-SSL
-SSSE
-SUBSTRING
 Schemas
-Splitted
-Stateful
-Submodules
-Subqueries
-TCP
-THINLTO
-TLS
-TSKV
-TSV
-TSVRaw
-TSan
-TabSeparated
-TabSeparatedRaw
-TabSeparatedRawWithNames
-TabSeparatedRawWithNamesAndTypes
-TabSeparatedWithNames
-TabSeparatedWithNamesAndTypes
-TargetSpecific
-TemplateIgnoreSpaces
-Testflows
-Tgz
-Toolset
-Tradeoff
-UB
-UBSan
-UI
-UInt
-UIntN
-UPDATEs
-UPLOADER
-UTF
-Uint
-Updatable
-Util
-Valgrind
-Vectorized
-VirtualBox
-WERROR
-WIP
-Werror
-Woboq
-WriteBuffer
-WriteBuffers
-XCode
-XFS
-Xcode
-Xcode's
-YAML
-YYYY
-ZAB
-ZFS
-Zipkin
-ZooKeeper
-ZooKeeper's
 comparising
 conf
-config
-configs
 const
 contrib
 convenvtional
-convered
-convertion
-convertions
-coroutines
-cpp
-cppkafka
-cpu
 croaring
-cronjob
 crst
-csv
-csvwithnames
-csvwithnamesandtypes
-customseparated
-customseparatedwithnames
 customseparatedwithnamesandtypes
 cyrus
 datacenter
@@ -229,18 +29,15 @@ datafiles
 dataset
 datasets
 datetime
-dbms
-ddl
 deallocation
 debian
 decompressor
-delimeters
 denormals
 deserialization
 deserialized
 destructor
 destructors
-directy
+directly
 dirs
 dmesg
 dont
@@ -249,11 +46,9 @@ durations
 endian
 enum
 envi
-exeption
 extern
 fastops
 fcoverage
-filesystem
 filesystems
 flatbuffers
 fmtlib
@@ -277,7 +72,6 @@ hardlinks
 hdfs
 heredoc
 heredocs
-hh
 homebrew
 http
 https
@@ -299,33 +93,12 @@ io
 isro
 jdbc
 jemalloc
-json
-jsonasstring
-jsoncolumns
-jsoncolumnsmonoblock
-jsoncompact
-jsoncompactcolumns
-jsoncompacteachrow
-jsoncompacteachrowwithnames
-jsoncompacteachrowwithnamesandtypes
-jsoncompactstrings
-jsoncompactstringseachrow
-jsoncompactstringseachrowwithnames
-jsoncompactstringseachrowwithnamesandtypes
-jsoneachrow
-jsoneachrowwithprogress
-jsonstrings
-jsonstringseachrow
-jsonstringseachrowwithprogress
 kafka
 kafkacat
 konsole
 krb
-lables
-latencies
 lexicographically
 libFuzzer
-libary
 libc
 libcpuid
 libcxx
@@ -371,13 +144,7 @@ mysqljs
 namespace
 nan
 nc
-neccessary
-ng
-noop
-num
-obfuscator
 odbc
-ok
 openldap
 opentelemetry
 os
@@ -393,144 +160,15 @@ perf
 performant
 poco
 popcnt
-postfix
-postfixes
-postgresql
-pre
-prebuild
-prebuilt
-preemptable
-preloaded
-prematurally
-preprocessed
-preprocessor
-presense
 presentational
-prestable
-prettycompact
-prettycompactmonoblock
-prettycompactnoescapes
-prettynoescapes
-prettyspace
-prettyspacenoescapes
-prioritied
-prlimit
-prometheus
-proto
-protobuf
-protobufsingle
-psql
-ptrs
-py
-rapidjson
-rawblob
-readahead
-readline
-readme
-readonly
-rebalanced
-replxx
-repo
-repo’s
 representable
 requestor
 resultset
-rethrow
-risc
-ro
-rocksdb
-rowNumberInBlock
-rowbinary
-rowbinarywithnames
-rowbinarywithnamesandtypes
-rsync
-runningAccumulate
-runnnig
 runtime
 ruok
-russian
-rw
 sasl
 schemas
-simdjson
 skippingerrors
 sparsehash
-splitted
-sql
-src
 srst
 srvr
-ss
-stacktraces
-statbox
-stateful
-stderr
-stdin
-stdout
-strtod
-strtoll
-strtoull
-structs
-subdirectories
-subexpressions
-submodule
-submodules
-subpattern
-subpatterns
-subqueries
-subquery
-subseconds
-substring
-subtree
-sudo
-symlink
-symlinks
-syntaxes
-systemd
-tabseparated
-tabseparatedraw
-tabseparatedrawwithnames
-tabseparatedrawwithnamesandtypes
-tabseparatedwithnames
-tabseparatedwithnamesandtypes
-tcp
-templated
-templateignorespaces
-tgz
-th
-tmp
-tokenization
-toml
-toolset
-tskv
-tsv
-tui
-turbostat
-txt
-unary
-unencrypted
-unixodbc
-unsecure
-url
-userspace
-usr
-utils
-variadic
-varint
-vectorized
-wchc
-wchs
-webpage
-webserver
-wget
-whitespace
-whitespaces
-wrt
-xcode
-xml
-xz
-zLib
-zkcopy
-zlib
-znodes
-zstd

utils/check-style/check-doc-aspell

@@ -13,9 +13,10 @@ ASPELL_IGNORE_PATH="${ROOT_PATH}/utils/check-style/aspell-ignore/${CHECK_LANG}"
 STATUS=0
 for fname in ${ROOT_PATH}/docs/${CHECK_LANG}/**/*.md; do
     errors=$(aspell list --mode=markdown --lang=${CHECK_LANG} < "$fname" \
-        | grep -v -f "${ASPELL_IGNORE_PATH}/todo.txt" \
+        | grep -Ewv -f "${ASPELL_IGNORE_PATH}/todo.txt" \
-        | grep -vi -f "${ASPELL_IGNORE_PATH}/caseinsensitive.txt" \
+        | grep -Ewvi -f "${ASPELL_IGNORE_PATH}/caseinsensitive.txt" \
-        | grep -v -f "${ASPELL_IGNORE_PATH}/casesensitive.txt" \
+        | grep -Ewv -f "${ASPELL_IGNORE_PATH}/casesensitive.txt" \
+        | grep -Ewv "[A-Z]+" \
         | sort | uniq)
     if [ ! -z "$errors" ]; then
         STATUS=1

utils/check-style/check-typos

@@ -5,7 +5,7 @@
 ROOT_PATH=$(git rev-parse --show-toplevel)
 
 codespell \
-    --skip '*generated*,*gperf*,*.bin,*.mrk*,*.idx,checksums.txt,*.dat,*.pyc,*.kate-swp,*obfuscateQueries.cpp' \
+    --skip '*generated*,*gperf*,*.bin,*.mrk*,*.idx,checksums.txt,*.dat,*.pyc,*.kate-swp,*obfuscateQueries.cpp,aspell-ignore/' \
     --ignore-words "${ROOT_PATH}/utils/check-style/codespell-ignore-words.list" \
     --exclude-file "${ROOT_PATH}/utils/check-style/codespell-ignore-lines.list" \
     --quiet-level 2 \