File size: 11,861 Bytes
27f6252
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
"""
In-process cheat sheet: CWE ↔ CAPEC ↔ MITRE ATT&CK taxonomy.

Loaded once at startup from the existing on-disk JSON files. All public lookups
are O(1) dict access. Fuzzy CWE-name matching uses rapidfuzz (WRatio scorer);
falls back to substring match if rapidfuzz is unavailable.

Sources:
  data/CTI/raw/cwe_capec_mitre_mapping.json  — CWE → {capecs, mitre_techniques}
  data/CTI/docs/attack_mappings.json          — tactic ↔ technique taxonomy
  data/knowledge_base/rag_exports/cwe_chunks.json — CWE human-readable names
"""

from __future__ import annotations

import json
import logging
import re
from pathlib import Path
from typing import Optional

logger = logging.getLogger(__name__)

_RE_CVE   = re.compile(r"\bCVE-\d{4}-\d{4,}\b", re.I)
_RE_CWE   = re.compile(r"\bCWE[-\s]?(\d+)\b", re.I)
_RE_CAPEC = re.compile(r"\bCAPEC[-\s]?(\d+)\b", re.I)
_RE_TECH  = re.compile(r"\bT(\d{4}(?:\.\d{3})?)\b")
_RE_TAC   = re.compile(r"\bTA(\d{4})\b")

_FUZZY_SCORE_CUTOFF = 90   # partial_ratio threshold; matches "sql injection" → CWE-89
_FUZZY_TOP_K = 3


class CheatSheet:
    """
    In-memory CWE/CAPEC/MITRE taxonomy with rapidfuzz fuzzy CWE-name matching.

    Never instantiate directly — use init_cheat_sheet() / get_cheat_sheet().
    """

    def __init__(self) -> None:
        # CWE-ID (uppercase, e.g. "CWE-79") → {capecs, mitre_techniques, mitre_description}
        self.cwe_map: dict[str, dict] = {}
        # TA0001 → [T1190, T1133, ...]
        self.tactic_to_techs: dict[str, list[str]] = {}
        # T1059 → [TA0002, ...]
        self.tech_to_tactics: dict[str, list[str]] = {}
        # "collection" → "TA0009"
        self.tactic_name_to_id: dict[str, str] = {}
        # CWE-79 → "Improper Neutralization of Input During Web Page Generation"
        self.cwe_names: dict[str, str] = {}
        # lower-name → CWE-ID  (for fuzzy / substring matching)
        self._name_to_cwe: dict[str, str] = {}
        self._fuzzy_available = False

    # ── Loaders ──────────────────────────────────────────────────────────────

    def load(self, project_root: Path) -> "CheatSheet":
        self._load_cwe_map(project_root / "data/CTI/raw/cwe_capec_mitre_mapping.json")
        self._load_attack_mappings(project_root / "data/CTI/docs/attack_mappings.json")
        self._load_cwe_names(project_root / "data/knowledge_base/rag_exports/cwe_chunks.json")
        self._try_import_rapidfuzz()
        logger.info(
            "CheatSheet loaded: %d CWEs, %d tactics, %d techniques, %d CWE names, fuzzy=%s",
            len(self.cwe_map), len(self.tactic_to_techs), len(self.tech_to_tactics),
            len(self.cwe_names), self._fuzzy_available,
        )
        return self

    def _load_cwe_map(self, path: Path) -> None:
        if not path.exists():
            logger.warning("cwe_capec_mitre_mapping.json not found at %s", path)
            return
        raw: dict = json.loads(path.read_text())
        # Keys may be "CWE-79" or "79" — normalise to "CWE-XX"
        for k, v in raw.items():
            cid = k if k.upper().startswith("CWE-") else f"CWE-{k}"
            self.cwe_map[cid.upper()] = v

    def _load_attack_mappings(self, path: Path) -> None:
        if not path.exists():
            logger.warning("attack_mappings.json not found at %s", path)
            return
        data = json.loads(path.read_text())
        self.tactic_to_techs   = {k.upper(): v for k, v in data.get("tactic_to_techniques", {}).items()}
        self.tech_to_tactics   = {k.upper(): v for k, v in data.get("technique_to_tactics", {}).items()}
        self.tactic_name_to_id = {k.lower(): v.upper()
                                   for k, v in data.get("tactic_name_to_id", {}).items()}

    def _load_cwe_names(self, path: Path) -> None:
        if not path.exists():
            logger.warning("cwe_chunks.json not found at %s", path)
            return
        chunks: list = json.loads(path.read_text())
        for c in chunks:
            p = c.get("payload", {})
            cid = p.get("cwe_id", "")
            name = p.get("name", "")
            if cid and name:
                self.cwe_names[cid.upper()] = name
                self._name_to_cwe[name.lower()] = cid.upper()

    def _try_import_rapidfuzz(self) -> None:
        try:
            import rapidfuzz  # noqa: F401
            self._fuzzy_available = True
        except ImportError:
            logger.warning("rapidfuzz not available — falling back to substring CWE matching")

    # ── Public lookups (O(1)) ────────────────────────────────────────────────

    def lookup_techniques_for_cwe(self, cwe_id: str) -> list[str]:
        return self.cwe_map.get(cwe_id.upper(), {}).get("mitre_techniques", [])

    def lookup_capecs_for_cwe(self, cwe_id: str) -> list[str]:
        return self.cwe_map.get(cwe_id.upper(), {}).get("capecs", [])

    def lookup_techniques_for_tactic(self, tactic_id: str) -> list[str]:
        return self.tactic_to_techs.get(tactic_id.upper(), [])

    def lookup_tactics_for_technique(self, tech_id: str) -> list[str]:
        return self.tech_to_tactics.get(tech_id.upper(), [])

    def cwe_name(self, cwe_id: str) -> str:
        return self.cwe_names.get(cwe_id.upper(), "")

    # ── Fuzzy CWE-name matching ───────────────────────────────────────────────

    def fuzzy_match_cwe(
        self,
        query: str,
        k: int = _FUZZY_TOP_K,
        score_cutoff: int = _FUZZY_SCORE_CUTOFF,
    ) -> list[dict]:
        """
        Match CWE names against arbitrary query text using rapidfuzz WRatio.

        Returns [{"cwe_id": "CWE-79", "name": "...", "score": 92.3}, ...]
        sorted by score desc. Falls back to substring match when rapidfuzz
        is unavailable.
        """
        if not self._name_to_cwe:
            return []

        query_lower = query.lower()

        if self._fuzzy_available:
            from rapidfuzz import process as fz_process, fuzz

            matches = fz_process.extract(
                query_lower,
                self._name_to_cwe.keys(),
                scorer=fuzz.partial_ratio,
                limit=k,
                score_cutoff=score_cutoff,
            )
            return [
                {"cwe_id": self._name_to_cwe[m[0]], "name": m[0], "score": m[1]}
                for m in matches
            ]

        # Fallback: substring match
        results = []
        for name, cid in self._name_to_cwe.items():
            if len(name) >= 5 and name in query_lower:
                results.append({"cwe_id": cid, "name": name, "score": 75.0})
                if len(results) >= k:
                    break
        return results

    # ── Entity extraction from free text ────────────────────────────────────

    def entity_to_hints(self, text: str, max_items: int = 6) -> dict:
        """
        NER-lite + fuzzy matching over arbitrary user query text.

        Returns dict with keys: cves, cwes, capecs, techniques, tactics, name_hits.
        name_hits is a list of {"cwe_id", "name", "score"} from fuzzy CWE matching.
        """
        if not text:
            return {"cves": [], "cwes": [], "capecs": [], "techniques": [], "tactics": [], "name_hits": []}

        cves       = list(dict.fromkeys(_RE_CVE.findall(text)))[:max_items]
        cwes       = [f"CWE-{n}"  for n in dict.fromkeys(_RE_CWE.findall(text))][:max_items]
        capecs     = [f"CAPEC-{n}" for n in dict.fromkeys(_RE_CAPEC.findall(text))][:max_items]
        techniques = [f"T{n}" for n in dict.fromkeys(_RE_TECH.findall(text))][:max_items]
        tactics    = [f"TA{n}" for n in dict.fromkeys(_RE_TAC.findall(text))][:max_items]

        # Fuzzy match against sliding window of 2-5 words so partial_ratio works
        name_hits = self._fuzzy_match_windows(text, k=max_items)

        return {
            "cves":       cves,
            "cwes":       cwes,
            "capecs":     capecs,
            "techniques": techniques,
            "tactics":    tactics,
            "name_hits":  name_hits,
        }

    def _fuzzy_match_windows(self, text: str, k: int = 3) -> list[dict]:
        """
        Slide a 2-5-word window over text and fuzzy-match each window against
        CWE names. Deduplicates by cwe_id, keeps highest score per CWE.
        """
        words = re.sub(r"[^\w\s]", " ", text.lower()).split()
        seen: dict[str, dict] = {}  # cwe_id → best hit
        for size in range(2, 6):    # 2-word to 5-word windows
            for i in range(len(words) - size + 1):
                window = " ".join(words[i : i + size])
                for hit in self.fuzzy_match_cwe(window, k=k):
                    cid = hit["cwe_id"]
                    if cid not in seen or hit["score"] > seen[cid]["score"]:
                        seen[cid] = hit
        # Sort by score desc, return top-k
        return sorted(seen.values(), key=lambda h: h["score"], reverse=True)[:k]

    # ── Prompt rendering ─────────────────────────────────────────────────────

    def render_hint_block(self, hints: dict, max_cwes: int = 3, max_techs_per_cwe: int = 4) -> str:
        """
        Compact hint block for injection into rewriter / translate-symptom prompts.

        Hard-capped to avoid prompt bloat (<400 chars target).
        """
        lines: list[str] = []
        seen_cwes: set[str] = set()

        # Explicit CWE IDs from regex
        for cwe in hints.get("cwes", [])[:max_cwes]:
            self._append_cwe_line(cwe, lines, seen_cwes, max_techs_per_cwe)

        # Fuzzy name hits (if not already covered above)
        for hit in hints.get("name_hits", []):
            if len(seen_cwes) >= max_cwes:
                break
            self._append_cwe_line(hit["cwe_id"], lines, seen_cwes, max_techs_per_cwe)

        # Technique → tactic mapping
        for tech in hints.get("techniques", [])[:4]:
            tacs = self.lookup_tactics_for_technique(tech)[:2]
            if tacs:
                lines.append(f"  {tech}: tactics={tacs}")

        if not lines:
            return ""
        return "KG taxonomy hints:\n" + "\n".join(lines)

    def _append_cwe_line(
        self,
        cwe: str,
        lines: list[str],
        seen: set[str],
        max_techs: int,
    ) -> None:
        if cwe in seen:
            return
        seen.add(cwe)
        nm     = self.cwe_names.get(cwe, "")
        techs  = self.lookup_techniques_for_cwe(cwe)[:max_techs]
        capecs = self.lookup_capecs_for_cwe(cwe)[:3]
        parts  = []
        if techs:
            parts.append(f"techniques={techs}")
        if capecs:
            parts.append(f"capecs={capecs}")
        if parts:
            label = f"{cwe}({nm})" if nm else cwe
            lines.append(f"  {label}: {', '.join(parts)}")


# ── Singleton management ──────────────────────────────────────────────────────

_singleton: Optional[CheatSheet] = None


def init_cheat_sheet(project_root: Path) -> CheatSheet:
    """Load and cache the global CheatSheet singleton."""
    global _singleton
    _singleton = CheatSheet().load(Path(project_root))
    return _singleton


def get_cheat_sheet() -> Optional[CheatSheet]:
    """Return the loaded CheatSheet, or None if init_cheat_sheet was never called."""
    return _singleton