Add new wordlists and add some words to existing lists' updaters

README.md

@@ -15,24 +15,26 @@ Included word lists
 Currently the plugin comes bundled with following word lists, compiled into one
 big list at installation time:
 
-| List name         | Example words                                 |
-|-------------------|-----------------------------------------------|
-| `acronyms`        | API, JSON, PaaS                               |
-| `algorithms`      | quicksort, subtree, memoization               |
-| `brands`          | GitHub, AdSense, DynamoDB                     |
-| `comments`        | TODO, FIXME, XXX                              |
-| `cpp`             | const, enum, lvalue                           |
-| `docker`          | Dockerfile, cgroups, ENTRYPOINT               |
-| `file-extensions` | .js, .yaml, .lua                              |
-| `git`             | submodule, worktree, rebase                   |
-| `html`            | h1, iframe, fieldset                          |
-| `kubernetes`      | ConfigMap, CSIDriver, HorizontalPodAutoscaler |
-| `lorem-ipsum`     | Lorem, ipsum, dolor                           |
-| `nerd-fonts`      |  ,  , &nbsp;                               | <!-- a non-breaking space is needed to render the last symbol correctly on GitHub -->
-| `prometheus`      | PromQL, Alertmanager, Pushgateway             |
-| `python`          | docstring, iterable, awaitable                |
-| `unix`            | SIGTERM, chroot, grep                         |
-| `versions`        | v1, v2, v3                                    |
+| List name         | Example words                                  |
+|-------------------|------------------------------------------------|
+| `acronyms`        | API, JSON, PaaS                                |
+| `algorithms`      | quicksort, subtree, memoization                |
+| `brands`          | GitHub, AdSense, DynamoDB                      |
+| `comments`        | TODO, FIXME, XXX                               |
+| `cpp`             | const, enum, lvalue                            |
+| `docker`          | Dockerfile, cgroups, ENTRYPOINT                |
+| `file-extensions` | .js, .yaml, .lua                               |
+| `git`             | submodule, worktree, rebase                    |
+| `html`            | h1, iframe, fieldset                           |
+| `jargon`          | sanitizers, indistinguishability, microservice |
+| `kubernetes`      | ConfigMap, CSIDriver, HorizontalPodAutoscaler  |
+| `lorem-ipsum`     | Lorem, ipsum, dolor                            |
+| `nerd-fonts`      |  ,  , &nbsp;                                | <!-- a non-breaking space is needed to render the last symbol correctly on GitHub -->
+| `names`           | WannaCry, PuTTY, Merkle                        |
+| `prometheus`      | PromQL, Alertmanager, Pushgateway              |
+| `python`          | docstring, iterable, awaitable                 |
+| `unix`            | SIGTERM, chroot, grep                          |
+| `versions`        | v1, v2, v3                                     |
 
 You can disable lists you don't want to use. See the _customization_ section
 for details.

scripts/update-acronyms

@@ -2,9 +2,56 @@
 source "$(dirname $0)/common.sh"
 
 emit_header "Common IT acronyms and abbreviations"
-curl -s "https://en.wikipedia.org/wiki/List_of_computing_and_IT_abbreviations" \
-  | egrep -o '<li><a.+title.+>[a-zA-Z0-9\.]+<' \
-  | grep -v '(page does not exist)' \
-  | egrep -o '>[a-zA-Z0-9\.]+<' \
-  | tr -d '<>' \
-  | emit_words
+{
+  curl -s "https://en.wikipedia.org/wiki/List_of_computing_and_IT_abbreviations" \
+    | egrep -o '<li><a.+title.+>[a-zA-Z0-9\.]+<' \
+    | grep -v '(page does not exist)' \
+    | egrep -o '>[a-zA-Z0-9\.]+<' \
+    | tr -d '<>'
+  cat <<heretag
+3DES
+APIs
+APK
+AXFR
+BSSID
+CAs
+CCMP
+CSPRNG
+CVSS
+CWE
+DMARC
+DNSSEC
+DSGVO
+ECB
+EDNS
+GC
+GDPR
+GNSS
+HKLM
+HMAC
+HSTS
+HTA
+IAM
+IBAN
+IBANs
+IDN
+JWT
+KDF
+LTS
+MD5
+MITM
+MTAs
+OCSP
+OSINT
+PEM
+R2
+RC4
+RCE
+SNI
+SSRF
+TKIP
+TOTP
+U2F
+XHR
+heretag
+} | emit_words

scripts/update-brands

@@ -2,8 +2,66 @@
 source "$(dirname $0)/common.sh"
 
 emit_header "Popular (mostly IT-related) brands"
-curl -s "https://raw.githubusercontent.com/simple-icons/simple-icons/master/_data/simple-icons.json" \
-  | jq -r '.icons | map(.title)[]' \
-  | tr ' ' '\n' \
-  | grep '[a-zA-Z]' \
-  | emit_words
+{
+  curl -s "https://raw.githubusercontent.com/simple-icons/simple-icons/master/_data/simple-icons.json" \
+    | jq -r '.icons | map(.title)[]' \
+    | tr ' ' '\n' \
+    | grep '[a-zA-Z]'
+  cat <<heretag
+Acronis
+AMD64
+AngularJS
+Artifactory
+BitLocker
+Chromebook
+CloudFormation
+CloudTrail
+Coverity
+Cppcheck
+Defender
+DigiCert
+Duqu
+Entra ID
+FortiGate
+FOSSGIS
+Foxit
+GitLab's
+GlobalSign
+Hoek
+iPhones
+jsoup
+Karaf
+Log4j
+Lync
+Mailchimp
+MailHog
+Nessus
+Netgear
+Nvidia
+Office
+OkHttp
+OneDrive's
+OpenPhish
+OSTIF
+PKZIP
+PowerDNS
+PrivateLink
+Protection
+PRTG
+PSFTPd
+reCAPTCHA
+reCAPTCHA's
+Semgrep
+SendGrid
+Sophos
+Splashtop
+Threema's
+TinyMCE
+TopAccess
+Vertica
+Wazuh
+YubiKey
+YubiKey's
+YubiKeys
+heretag
+} | emit_words

scripts/update-unix

@@ -17,4 +17,12 @@ emit_header "Common UNIX commands"
   | cut -d '>' -f3 \
   | cut -d '<' -f1
   busybox --list
+  cat <<heretag
+cron
+gpg
+podman
+rsync
+sudo
+tcpdump
+heretag
 } | emit_words