|
1
|
""" |
|
2
|
C++ AST parser — extracts classes, structs, namespaces, functions, methods, |
|
3
|
and #include directives from .cpp, .hpp, .cc, and .cxx 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_cpp_language(): |
|
17
|
try: |
|
18
|
import tree_sitter_cpp as tscpp # type: ignore[import] |
|
19
|
from tree_sitter import Language |
|
20
|
|
|
21
|
return Language(tscpp.language()) |
|
22
|
except ImportError as e: |
|
23
|
raise ImportError("Install tree-sitter-cpp: pip install tree-sitter-cpp") 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
|
class CppParser(LanguageParser): |
|
31
|
"""Parses C++ source files into the navegador graph.""" |
|
32
|
|
|
33
|
def __init__(self) -> None: |
|
34
|
from tree_sitter import Parser # type: ignore[import] |
|
35
|
|
|
36
|
self._parser = Parser(_get_cpp_language()) |
|
37
|
|
|
38
|
def parse_file(self, path: Path, repo_root: Path, store: GraphStore) -> dict[str, int]: |
|
39
|
source = path.read_bytes() |
|
40
|
tree = self._parser.parse(source) |
|
41
|
rel_path = str(path.relative_to(repo_root)) |
|
42
|
|
|
43
|
store.create_node( |
|
44
|
NodeLabel.File, |
|
45
|
{ |
|
46
|
"name": path.name, |
|
47
|
"path": rel_path, |
|
48
|
"language": "cpp", |
|
49
|
"line_count": source.count(b"\n"), |
|
50
|
}, |
|
51
|
) |
|
52
|
|
|
53
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
54
|
self._walk(tree.root_node, source, rel_path, store, stats, class_name=None) |
|
55
|
return stats |
|
56
|
|
|
57
|
# ── AST walker ──────────────────────────────────────────────────────────── |
|
58
|
|
|
59
|
def _walk( |
|
60
|
self, |
|
61
|
node, |
|
62
|
source: bytes, |
|
63
|
file_path: str, |
|
64
|
store: GraphStore, |
|
65
|
stats: dict, |
|
66
|
class_name: str | None, |
|
67
|
) -> None: |
|
68
|
if node.type in ("class_specifier", "struct_specifier"): |
|
69
|
self._handle_class(node, source, file_path, store, stats) |
|
70
|
return |
|
71
|
if node.type == "function_definition": |
|
72
|
self._handle_function(node, source, file_path, store, stats, class_name) |
|
73
|
return |
|
74
|
if node.type == "preproc_include": |
|
75
|
self._handle_include(node, source, file_path, store, stats) |
|
76
|
return |
|
77
|
if node.type == "namespace_definition": |
|
78
|
# Recurse into namespace body |
|
79
|
body = node.child_by_field_name("body") |
|
80
|
if not body: |
|
81
|
body = next((c for c in node.children if c.type == "declaration_list"), None) |
|
82
|
if body: |
|
83
|
for child in body.children: |
|
84
|
self._walk(child, source, file_path, store, stats, class_name) |
|
85
|
return |
|
86
|
for child in node.children: |
|
87
|
self._walk(child, source, file_path, store, stats, class_name) |
|
88
|
|
|
89
|
# ── Handlers ────────────────────────────────────────────────────────────── |
|
90
|
|
|
91
|
def _handle_class( |
|
92
|
self, node, source: bytes, file_path: str, store: GraphStore, stats: dict |
|
93
|
) -> None: |
|
94
|
name_node = node.child_by_field_name("name") |
|
95
|
if not name_node: |
|
96
|
name_node = next((c for c in node.children if c.type == "type_identifier"), None) |
|
97
|
if not name_node: |
|
98
|
return |
|
99
|
name = _node_text(name_node, source) |
|
100
|
|
|
101
|
store.create_node( |
|
102
|
NodeLabel.Class, |
|
103
|
{ |
|
104
|
"name": name, |
|
105
|
"file_path": file_path, |
|
106
|
"line_start": node.start_point[0] + 1, |
|
107
|
"line_end": node.end_point[0] + 1, |
|
108
|
"docstring": "", |
|
109
|
}, |
|
110
|
) |
|
111
|
store.create_edge( |
|
112
|
NodeLabel.File, |
|
113
|
{"path": file_path}, |
|
114
|
EdgeType.CONTAINS, |
|
115
|
NodeLabel.Class, |
|
116
|
{"name": name, "file_path": file_path}, |
|
117
|
) |
|
118
|
stats["classes"] += 1 |
|
119
|
stats["edges"] += 1 |
|
120
|
|
|
121
|
# Base classes |
|
122
|
base_clause = node.child_by_field_name("base_clause") |
|
123
|
if not base_clause: |
|
124
|
base_clause = next((c for c in node.children if c.type == "base_class_clause"), None) |
|
125
|
if base_clause: |
|
126
|
for child in base_clause.children: |
|
127
|
if child.type == "type_identifier": |
|
128
|
parent_name = _node_text(child, source) |
|
129
|
store.create_edge( |
|
130
|
NodeLabel.Class, |
|
131
|
{"name": name, "file_path": file_path}, |
|
132
|
EdgeType.INHERITS, |
|
133
|
NodeLabel.Class, |
|
134
|
{"name": parent_name, "file_path": file_path}, |
|
135
|
) |
|
136
|
stats["edges"] += 1 |
|
137
|
|
|
138
|
# Walk class body for member functions |
|
139
|
body = node.child_by_field_name("body") |
|
140
|
if not body: |
|
141
|
body = next((c for c in node.children if c.type == "field_declaration_list"), None) |
|
142
|
if body: |
|
143
|
for child in body.children: |
|
144
|
if child.type == "function_definition": |
|
145
|
self._handle_function(child, source, file_path, store, stats, class_name=name) |
|
146
|
|
|
147
|
def _handle_function( |
|
148
|
self, |
|
149
|
node, |
|
150
|
source: bytes, |
|
151
|
file_path: str, |
|
152
|
store: GraphStore, |
|
153
|
stats: dict, |
|
154
|
class_name: str | None, |
|
155
|
) -> None: |
|
156
|
declarator = node.child_by_field_name("declarator") |
|
157
|
name = self._extract_function_name(declarator, source) |
|
158
|
if not name: |
|
159
|
return |
|
160
|
|
|
161
|
label = NodeLabel.Method if class_name else NodeLabel.Function |
|
162
|
store.create_node( |
|
163
|
label, |
|
164
|
{ |
|
165
|
"name": name, |
|
166
|
"file_path": file_path, |
|
167
|
"line_start": node.start_point[0] + 1, |
|
168
|
"line_end": node.end_point[0] + 1, |
|
169
|
"docstring": "", |
|
170
|
"class_name": class_name or "", |
|
171
|
}, |
|
172
|
) |
|
173
|
|
|
174
|
container_label = NodeLabel.Class if class_name else NodeLabel.File |
|
175
|
container_key = ( |
|
176
|
{"name": class_name, "file_path": file_path} if class_name else {"path": file_path} |
|
177
|
) |
|
178
|
store.create_edge( |
|
179
|
container_label, |
|
180
|
container_key, |
|
181
|
EdgeType.CONTAINS, |
|
182
|
label, |
|
183
|
{"name": name, "file_path": file_path}, |
|
184
|
) |
|
185
|
stats["functions"] += 1 |
|
186
|
stats["edges"] += 1 |
|
187
|
|
|
188
|
self._extract_calls(node, source, file_path, name, label, store, stats) |
|
189
|
|
|
190
|
def _extract_function_name(self, declarator, source: bytes) -> str | None: |
|
191
|
"""Recursively dig through C++ declarator nodes to find the function name.""" |
|
192
|
if declarator is None: |
|
193
|
return None |
|
194
|
if declarator.type == "identifier": |
|
195
|
return _node_text(declarator, source) |
|
196
|
if declarator.type == "qualified_identifier": |
|
197
|
# MyClass::method — take the last component |
|
198
|
name_node = declarator.child_by_field_name("name") |
|
199
|
if name_node: |
|
200
|
return _node_text(name_node, source) |
|
201
|
# fallback: last identifier child |
|
202
|
ids = [c for c in declarator.children if c.type == "identifier"] |
|
203
|
if ids: |
|
204
|
return _node_text(ids[-1], source) |
|
205
|
if declarator.type in ("function_declarator", "pointer_declarator", "reference_declarator"): |
|
206
|
inner = declarator.child_by_field_name("declarator") |
|
207
|
return self._extract_function_name(inner, source) |
|
208
|
# Fallback: first identifier child |
|
209
|
for child in declarator.children: |
|
210
|
if child.type == "identifier": |
|
211
|
return _node_text(child, source) |
|
212
|
return None |
|
213
|
|
|
214
|
def _handle_include( |
|
215
|
self, node, source: bytes, file_path: str, store: GraphStore, stats: dict |
|
216
|
) -> None: |
|
217
|
path_node = node.child_by_field_name("path") |
|
218
|
if not path_node: |
|
219
|
path_node = next( |
|
220
|
(c for c in node.children if c.type in ("string_literal", "system_lib_string")), |
|
221
|
None, |
|
222
|
) |
|
223
|
if not path_node: |
|
224
|
return |
|
225
|
module = _node_text(path_node, source).strip('<>"') |
|
226
|
store.create_node( |
|
227
|
NodeLabel.Import, |
|
228
|
{ |
|
229
|
"name": module, |
|
230
|
"file_path": file_path, |
|
231
|
"line_start": node.start_point[0] + 1, |
|
232
|
"module": module, |
|
233
|
}, |
|
234
|
) |
|
235
|
store.create_edge( |
|
236
|
NodeLabel.File, |
|
237
|
{"path": file_path}, |
|
238
|
EdgeType.IMPORTS, |
|
239
|
NodeLabel.Import, |
|
240
|
{"name": module, "file_path": file_path}, |
|
241
|
) |
|
242
|
stats["edges"] += 1 |
|
243
|
|
|
244
|
def _extract_calls( |
|
245
|
self, |
|
246
|
fn_node, |
|
247
|
source: bytes, |
|
248
|
file_path: str, |
|
249
|
fn_name: str, |
|
250
|
fn_label: str, |
|
251
|
store: GraphStore, |
|
252
|
stats: dict, |
|
253
|
) -> None: |
|
254
|
def walk(node): |
|
255
|
if node.type == "call_expression": |
|
256
|
func = node.child_by_field_name("function") |
|
257
|
if not func: |
|
258
|
func = next( |
|
259
|
( |
|
260
|
c |
|
261
|
for c in node.children |
|
262
|
if c.type in ("identifier", "qualified_identifier", "field_expression") |
|
263
|
), |
|
264
|
None, |
|
265
|
) |
|
266
|
if func: |
|
267
|
callee = _node_text(func, source).split("::")[-1].split(".")[-1].split("->")[-1] |
|
268
|
store.create_edge( |
|
269
|
fn_label, |
|
270
|
{"name": fn_name, "file_path": file_path}, |
|
271
|
EdgeType.CALLS, |
|
272
|
NodeLabel.Function, |
|
273
|
{"name": callee, "file_path": file_path}, |
|
274
|
) |
|
275
|
stats["edges"] += 1 |
|
276
|
for child in node.children: |
|
277
|
walk(child) |
|
278
|
|
|
279
|
body = fn_node.child_by_field_name("body") |
|
280
|
if not body: |
|
281
|
body = next((c for c in fn_node.children if c.type == "compound_statement"), None) |
|
282
|
if body: |
|
283
|
walk(body) |
|
284
|
|