From e74b3e04a9baf674a07694731886218bb6098c0a Mon Sep 17 00:00:00 2001
From: Krypton <root@krypton.ninja>
Date: Sun, 21 Jul 2024 13:16:44 +0200
Subject: [PATCH] feat: Generate JSON file with the wordlists

---
 .github/workflows/update-json-file.yml | 32 +++++++++++
 tools/make_json.py                     | 80 ++++++++++++++++++++++++++
 2 files changed, 112 insertions(+)
 create mode 100644 .github/workflows/update-json-file.yml
 create mode 100644 tools/make_json.py

diff --git a/.github/workflows/update-json-file.yml b/.github/workflows/update-json-file.yml
new file mode 100644
index 0000000..78ae001
--- /dev/null
+++ b/.github/workflows/update-json-file.yml
@@ -0,0 +1,32 @@
+name: Update JSON file
+
+on:
+  push:
+    branches:
+      - main
+  workflow_dispatch:
+
+jobs:
+  update_readme_file:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v2
+        with:
+          token: ${{ secrets.PAT }}
+      - name: Setup Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: "3.11"
+      - name: Make JSON file
+        run: python tools/make_json.py
+      - name: Check if there are any changes
+        id: verify_diff
+        run: |
+          git diff --quiet . || echo "changed=true" >> $GITHUB_OUTPUT
+      - name: Update README file
+        if: steps.verify_diff.outputs.changed == 'true'
+        run: |
+          git config --global user.name 'Krypton'
+          git config --global user.email 'root@krypton.ninja'
+          git commit -am "chore: Automatic wordlists.json file update with accurate wordlists [skip ci]"
+          git push
diff --git a/tools/make_json.py b/tools/make_json.py
new file mode 100644
index 0000000..0bb711f
--- /dev/null
+++ b/tools/make_json.py
@@ -0,0 +1,80 @@
+# This code is bad, I even lost myself in it - but it works 😎 🤓
+
+import glob
+import json
+import os
+import re
+
+
+keywords = {
+    "Sql": "SQL",
+    "Jsp": "JSP",
+    "Lfi": "LFI",
+    "Tlds": "TLDs",
+    "Wp": "WP",
+    "Vpn": "VPN",
+    "Wpa": "WPA",
+    "Us": "US",
+    "Uri": "URI",
+    "Http": "HTTP",
+    "Os": "OS",
+    "Beos": "BeOS",
+    "Chromeos": "ChromeOS",
+    "Ios": "iOS",
+    "Hp": "HP",
+    "Openbsd": "OpenBSD",
+    "Sunos": "SunOS",
+    "Webos": "webOS",
+    "Itunes": "iTunes",
+    "Xml": "XML",
+    "Xss": "XSS",
+}
+
+
+class Wordlist:
+    def __init__(self, name: str, href: str, lines: int, tags: list[type[str]]):
+        self.name = name
+        self.href = href
+        self.lines = lines
+        self.tags = tags
+
+
+wordlists: list[type[Wordlist]] = []
+
+for filename in glob.iglob("./wordlists/**/*", recursive=True):
+    if os.path.isdir(filename):
+        continue
+
+    rel_path = os.path.relpath(filename)
+    levels = rel_path.split(os.sep)
+    folders = levels[1 : len(levels) - 1]
+    name = levels[len(levels) - 1]
+    name = name[: len(name) - 4].replace("_", " ").title()
+
+    for keyword in keywords:
+        name = re.sub(rf"\b{keyword}\b", keywords[keyword], name)
+
+    href = "/".join(levels)
+
+    lines = 0
+    if name == "Rockyou":  # Can't count correctly for that one as it's a ZIP file
+        lines = 14344392
+    else:
+        with open(rel_path, "rb") as f:
+            lines = sum(1 for _ in f)
+
+    tags: list[type[str]] = []
+    for folder in folders:
+        tags.append(folder.replace("_", "-"))
+
+    wordlist = Wordlist(name, href, lines, tags)
+    wordlists.append(wordlist)
+
+with open("wordlists.json", "w", encoding="utf-8") as f:
+    f.write(
+        json.dumps(
+            [wordlist.__dict__ for wordlist in sorted(wordlists, key=lambda w: w.name)],
+            sort_keys=True,
+            indent=4,
+        )
+    )