|
1
|
""" |
|
2
|
Puppet manifest parser — extracts classes, defined types, node definitions, |
|
3
|
resource declarations, includes, and parameters from .pp files using tree-sitter. |
|
4
|
""" |
|
5
|
|
|
6
|
import logging |
|
7
|
from pathlib import Path |
|
8
|
|
|
9
|
from navegador.graph.schema import EdgeType, NodeLabel |
|
10
|
from navegador.graph.store import GraphStore |
|
11
|
from navegador.ingestion.parser import LanguageParser |
|
12
|
|
|
13
|
logger = logging.getLogger(__name__) |
|
14
|
|
|
15
|
|
|
16
|
def _get_puppet_language(): |
|
17
|
try: |
|
18
|
import tree_sitter_puppet as tspuppet # type: ignore[import] |
|
19
|
from tree_sitter import Language |
|
20
|
|
|
21
|
return Language(tspuppet.language()) |
|
22
|
except ImportError as e: |
|
23
|
raise ImportError("Install tree-sitter-puppet: pip install tree-sitter-puppet") from e |
|
24
|
|
|
25
|
|
|
26
|
def _node_text(node, source: bytes) -> str: |
|
27
|
return source[node.start_byte : node.end_byte].decode("utf-8", errors="replace") |
|
28
|
|
|
29
|
|
|
30
|
def _class_identifier_text(node, source: bytes) -> str: |
|
31
|
"""Join identifier children of a class_identifier with '::'.""" |
|
32
|
parts = [_node_text(child, source) for child in node.children if child.type == "identifier"] |
|
33
|
return "::".join(parts) if parts else _node_text(node, source) |
|
34
|
|
|
35
|
|
|
36
|
class PuppetParser(LanguageParser): |
|
37
|
"""Parses Puppet manifest files into the navegador graph.""" |
|
38
|
|
|
39
|
def __init__(self) -> None: |
|
40
|
from tree_sitter import Parser # type: ignore[import] |
|
41
|
|
|
42
|
self._parser = Parser(_get_puppet_language()) |
|
43
|
|
|
44
|
def parse_file(self, path: Path, repo_root: Path, store: GraphStore) -> dict[str, int]: |
|
45
|
source = path.read_bytes() |
|
46
|
tree = self._parser.parse(source) |
|
47
|
rel_path = str(path.relative_to(repo_root)) |
|
48
|
|
|
49
|
store.create_node( |
|
50
|
NodeLabel.File, |
|
51
|
{ |
|
52
|
"name": path.name, |
|
53
|
"path": rel_path, |
|
54
|
"language": "puppet", |
|
55
|
"line_count": source.count(b"\n"), |
|
56
|
}, |
|
57
|
) |
|
58
|
|
|
59
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
60
|
self._walk(tree.root_node, source, rel_path, store, stats) |
|
61
|
return stats |
|
62
|
|
|
63
|
# ── AST walker ──────────────────────────────────────────────────────────── |
|
64
|
|
|
65
|
def _walk(self, node, source: bytes, file_path: str, store: GraphStore, stats: dict) -> None: |
|
66
|
if node.type == "class_definition": |
|
67
|
self._handle_class(node, source, file_path, store, stats) |
|
68
|
return |
|
69
|
if node.type == "defined_resource_type": |
|
70
|
self._handle_defined_type(node, source, file_path, store, stats) |
|
71
|
return |
|
72
|
if node.type == "node_definition": |
|
73
|
self._handle_node(node, source, file_path, store, stats) |
|
74
|
return |
|
75
|
if node.type == "include_statement": |
|
76
|
self._handle_include(node, source, file_path, store, stats) |
|
77
|
return |
|
78
|
for child in node.children: |
|
79
|
self._walk(child, source, file_path, store, stats) |
|
80
|
|
|
81
|
# ── Handlers ────────────────────────────────────────────────────────────── |
|
82
|
|
|
83
|
def _handle_class( |
|
84
|
self, node, source: bytes, file_path: str, store: GraphStore, stats: dict |
|
85
|
) -> None: |
|
86
|
name = self._extract_class_identifier(node, source) |
|
87
|
if not name: |
|
88
|
return |
|
89
|
|
|
90
|
store.create_node( |
|
91
|
NodeLabel.Class, |
|
92
|
{ |
|
93
|
"name": name, |
|
94
|
"file_path": file_path, |
|
95
|
"line_start": node.start_point[0] + 1, |
|
96
|
"line_end": node.end_point[0] + 1, |
|
97
|
"docstring": "", |
|
98
|
"semantic_type": "puppet_class", |
|
99
|
}, |
|
100
|
) |
|
101
|
store.create_edge( |
|
102
|
NodeLabel.File, |
|
103
|
{"path": file_path}, |
|
104
|
EdgeType.CONTAINS, |
|
105
|
NodeLabel.Class, |
|
106
|
{"name": name, "file_path": file_path}, |
|
107
|
) |
|
108
|
stats["classes"] += 1 |
|
109
|
stats["edges"] += 1 |
|
110
|
|
|
111
|
self._extract_parameters(node, source, file_path, name, store, stats) |
|
112
|
self._extract_resources(node, source, file_path, name, store, stats) |
|
113
|
|
|
114
|
def _handle_defined_type( |
|
115
|
self, node, source: bytes, file_path: str, store: GraphStore, stats: dict |
|
116
|
) -> None: |
|
117
|
name = self._extract_class_identifier(node, source) |
|
118
|
if not name: |
|
119
|
return |
|
120
|
|
|
121
|
store.create_node( |
|
122
|
NodeLabel.Class, |
|
123
|
{ |
|
124
|
"name": name, |
|
125
|
"file_path": file_path, |
|
126
|
"line_start": node.start_point[0] + 1, |
|
127
|
"line_end": node.end_point[0] + 1, |
|
128
|
"docstring": "", |
|
129
|
"semantic_type": "puppet_defined_type", |
|
130
|
}, |
|
131
|
) |
|
132
|
store.create_edge( |
|
133
|
NodeLabel.File, |
|
134
|
{"path": file_path}, |
|
135
|
EdgeType.CONTAINS, |
|
136
|
NodeLabel.Class, |
|
137
|
{"name": name, "file_path": file_path}, |
|
138
|
) |
|
139
|
stats["classes"] += 1 |
|
140
|
stats["edges"] += 1 |
|
141
|
|
|
142
|
self._extract_parameters(node, source, file_path, name, store, stats) |
|
143
|
self._extract_resources(node, source, file_path, name, store, stats) |
|
144
|
|
|
145
|
def _handle_node( |
|
146
|
self, node, source: bytes, file_path: str, store: GraphStore, stats: dict |
|
147
|
) -> None: |
|
148
|
name = self._extract_node_name(node, source) |
|
149
|
if not name: |
|
150
|
return |
|
151
|
|
|
152
|
store.create_node( |
|
153
|
NodeLabel.Class, |
|
154
|
{ |
|
155
|
"name": name, |
|
156
|
"file_path": file_path, |
|
157
|
"line_start": node.start_point[0] + 1, |
|
158
|
"line_end": node.end_point[0] + 1, |
|
159
|
"docstring": "", |
|
160
|
"semantic_type": "puppet_node", |
|
161
|
}, |
|
162
|
) |
|
163
|
store.create_edge( |
|
164
|
NodeLabel.File, |
|
165
|
{"path": file_path}, |
|
166
|
EdgeType.CONTAINS, |
|
167
|
NodeLabel.Class, |
|
168
|
{"name": name, "file_path": file_path}, |
|
169
|
) |
|
170
|
stats["classes"] += 1 |
|
171
|
stats["edges"] += 1 |
|
172
|
|
|
173
|
self._extract_resources(node, source, file_path, name, store, stats) |
|
174
|
|
|
175
|
def _handle_include( |
|
176
|
self, node, source: bytes, file_path: str, store: GraphStore, stats: dict |
|
177
|
) -> None: |
|
178
|
ident_node = None |
|
179
|
for child in node.children: |
|
180
|
if child.type == "class_identifier": |
|
181
|
ident_node = child |
|
182
|
break |
|
183
|
if not ident_node: |
|
184
|
return |
|
185
|
|
|
186
|
module = _class_identifier_text(ident_node, source) |
|
187
|
store.create_node( |
|
188
|
NodeLabel.Import, |
|
189
|
{ |
|
190
|
"name": module, |
|
191
|
"file_path": file_path, |
|
192
|
"line_start": node.start_point[0] + 1, |
|
193
|
"module": module, |
|
194
|
"semantic_type": "puppet_include", |
|
195
|
}, |
|
196
|
) |
|
197
|
store.create_edge( |
|
198
|
NodeLabel.File, |
|
199
|
{"path": file_path}, |
|
200
|
EdgeType.IMPORTS, |
|
201
|
NodeLabel.Import, |
|
202
|
{"name": module, "file_path": file_path}, |
|
203
|
) |
|
204
|
stats["edges"] += 1 |
|
205
|
|
|
206
|
# ── Extractors ──────────────────────────────────────────────────────────── |
|
207
|
|
|
208
|
def _extract_class_identifier(self, node, source: bytes) -> str | None: |
|
209
|
"""Find and return the class_identifier text from a class/define node.""" |
|
210
|
for child in node.children: |
|
211
|
if child.type == "class_identifier": |
|
212
|
return _class_identifier_text(child, source) |
|
213
|
return None |
|
214
|
|
|
215
|
def _extract_node_name(self, node, source: bytes) -> str | None: |
|
216
|
"""Extract the node name from a node_definition (string child of node_name).""" |
|
217
|
for child in node.children: |
|
218
|
if child.type == "node_name": |
|
219
|
for grandchild in child.children: |
|
220
|
if grandchild.type == "string": |
|
221
|
return _node_text(grandchild, source).strip("'\"") |
|
222
|
return _node_text(child, source).strip("'\"") |
|
223
|
return None |
|
224
|
|
|
225
|
def _extract_parameters( |
|
226
|
self, |
|
227
|
node, |
|
228
|
source: bytes, |
|
229
|
file_path: str, |
|
230
|
class_name: str, |
|
231
|
store: GraphStore, |
|
232
|
stats: dict, |
|
233
|
) -> None: |
|
234
|
"""Extract parameters from a parameter_list inside a class/define.""" |
|
235
|
for child in node.children: |
|
236
|
if child.type != "parameter_list": |
|
237
|
continue |
|
238
|
for param in child.children: |
|
239
|
if param.type != "parameter": |
|
240
|
continue |
|
241
|
var_node = None |
|
242
|
for pc in param.children: |
|
243
|
if pc.type == "variable": |
|
244
|
var_node = pc |
|
245
|
break |
|
246
|
if not var_node: |
|
247
|
continue |
|
248
|
var_name = _node_text(var_node, source).lstrip("$") |
|
249
|
store.create_node( |
|
250
|
NodeLabel.Variable, |
|
251
|
{ |
|
252
|
"name": var_name, |
|
253
|
"file_path": file_path, |
|
254
|
"line_start": param.start_point[0] + 1, |
|
255
|
"semantic_type": "puppet_parameter", |
|
256
|
}, |
|
257
|
) |
|
258
|
store.create_edge( |
|
259
|
NodeLabel.Class, |
|
260
|
{"name": class_name, "file_path": file_path}, |
|
261
|
EdgeType.CONTAINS, |
|
262
|
NodeLabel.Variable, |
|
263
|
{"name": var_name, "file_path": file_path}, |
|
264
|
) |
|
265
|
stats["edges"] += 1 |
|
266
|
|
|
267
|
def _extract_resources( |
|
268
|
self, |
|
269
|
node, |
|
270
|
source: bytes, |
|
271
|
file_path: str, |
|
272
|
class_name: str, |
|
273
|
store: GraphStore, |
|
274
|
stats: dict, |
|
275
|
) -> None: |
|
276
|
"""Walk the block of a class/define/node to find resource declarations.""" |
|
277
|
for child in node.children: |
|
278
|
if child.type == "block": |
|
279
|
self._walk_block_for_resources(child, source, file_path, class_name, store, stats) |
|
280
|
break |
|
281
|
|
|
282
|
def _walk_block_for_resources( |
|
283
|
self, |
|
284
|
node, |
|
285
|
source: bytes, |
|
286
|
file_path: str, |
|
287
|
class_name: str, |
|
288
|
store: GraphStore, |
|
289
|
stats: dict, |
|
290
|
) -> None: |
|
291
|
"""Recursively find resource_declaration nodes inside a block.""" |
|
292
|
if node.type == "resource_declaration": |
|
293
|
self._handle_resource(node, source, file_path, class_name, store, stats) |
|
294
|
return |
|
295
|
for child in node.children: |
|
296
|
self._walk_block_for_resources(child, source, file_path, class_name, store, stats) |
|
297
|
|
|
298
|
def _handle_resource( |
|
299
|
self, |
|
300
|
node, |
|
301
|
source: bytes, |
|
302
|
file_path: str, |
|
303
|
class_name: str, |
|
304
|
store: GraphStore, |
|
305
|
stats: dict, |
|
306
|
) -> None: |
|
307
|
"""Handle a resource_declaration: first identifier = type, first string = title.""" |
|
308
|
res_type = None |
|
309
|
res_title = None |
|
310
|
for child in node.children: |
|
311
|
if child.type == "identifier" and res_type is None: |
|
312
|
res_type = _node_text(child, source) |
|
313
|
if child.type == "string" and res_title is None: |
|
314
|
res_title = _node_text(child, source).strip("'\"") |
|
315
|
if not res_type: |
|
316
|
return |
|
317
|
|
|
318
|
name = f"{res_type}[{res_title}]" if res_title else res_type |
|
319
|
store.create_node( |
|
320
|
NodeLabel.Function, |
|
321
|
{ |
|
322
|
"name": name, |
|
323
|
"file_path": file_path, |
|
324
|
"line_start": node.start_point[0] + 1, |
|
325
|
"line_end": node.end_point[0] + 1, |
|
326
|
"docstring": "", |
|
327
|
"class_name": class_name, |
|
328
|
"semantic_type": "puppet_resource", |
|
329
|
}, |
|
330
|
) |
|
331
|
store.create_edge( |
|
332
|
NodeLabel.Class, |
|
333
|
{"name": class_name, "file_path": file_path}, |
|
334
|
EdgeType.CONTAINS, |
|
335
|
NodeLabel.Function, |
|
336
|
{"name": name, "file_path": file_path}, |
|
337
|
) |
|
338
|
stats["functions"] += 1 |
|
339
|
stats["edges"] += 1 |
|
340
|
|