|
1
|
""" |
|
2
|
C AST parser — extracts functions, structs, typedefs, and #include directives |
|
3
|
from .c and .h 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_c_language(): |
|
17
|
try: |
|
18
|
import tree_sitter_c as tsc # type: ignore[import] |
|
19
|
from tree_sitter import Language |
|
20
|
|
|
21
|
return Language(tsc.language()) |
|
22
|
except ImportError as e: |
|
23
|
raise ImportError("Install tree-sitter-c: pip install tree-sitter-c") 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 CParser(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_c_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": "c", |
|
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) |
|
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
|
) -> None: |
|
67
|
if node.type == "function_definition": |
|
68
|
self._handle_function(node, source, file_path, store, stats) |
|
69
|
return |
|
70
|
if node.type in ("struct_specifier", "union_specifier", "enum_specifier"): |
|
71
|
self._handle_struct(node, source, file_path, store, stats) |
|
72
|
return |
|
73
|
if node.type == "preproc_include": |
|
74
|
self._handle_include(node, source, file_path, store, stats) |
|
75
|
return |
|
76
|
for child in node.children: |
|
77
|
self._walk(child, source, file_path, store, stats) |
|
78
|
|
|
79
|
# ── Handlers ────────────────────────────────────────────────────────────── |
|
80
|
|
|
81
|
def _handle_function( |
|
82
|
self, node, source: bytes, file_path: str, store: GraphStore, stats: dict |
|
83
|
) -> None: |
|
84
|
# function_definition: type declarator body |
|
85
|
# declarator may be a function_declarator with a name field |
|
86
|
declarator = node.child_by_field_name("declarator") |
|
87
|
name = self._extract_function_name(declarator, source) |
|
88
|
if not name: |
|
89
|
return |
|
90
|
|
|
91
|
store.create_node( |
|
92
|
NodeLabel.Function, |
|
93
|
{ |
|
94
|
"name": name, |
|
95
|
"file_path": file_path, |
|
96
|
"line_start": node.start_point[0] + 1, |
|
97
|
"line_end": node.end_point[0] + 1, |
|
98
|
"docstring": "", |
|
99
|
"class_name": "", |
|
100
|
}, |
|
101
|
) |
|
102
|
store.create_edge( |
|
103
|
NodeLabel.File, |
|
104
|
{"path": file_path}, |
|
105
|
EdgeType.CONTAINS, |
|
106
|
NodeLabel.Function, |
|
107
|
{"name": name, "file_path": file_path}, |
|
108
|
) |
|
109
|
stats["functions"] += 1 |
|
110
|
stats["edges"] += 1 |
|
111
|
|
|
112
|
self._extract_calls(node, source, file_path, name, store, stats) |
|
113
|
|
|
114
|
def _extract_function_name(self, declarator, source: bytes) -> str | None: |
|
115
|
"""Recursively dig through declarator nodes to find the function name.""" |
|
116
|
if declarator is None: |
|
117
|
return None |
|
118
|
if declarator.type == "identifier": |
|
119
|
return _node_text(declarator, source) |
|
120
|
if declarator.type == "function_declarator": |
|
121
|
inner = declarator.child_by_field_name("declarator") |
|
122
|
return self._extract_function_name(inner, source) |
|
123
|
if declarator.type == "pointer_declarator": |
|
124
|
inner = declarator.child_by_field_name("declarator") |
|
125
|
return self._extract_function_name(inner, source) |
|
126
|
# Fallback: look for identifier child |
|
127
|
for child in declarator.children: |
|
128
|
if child.type == "identifier": |
|
129
|
return _node_text(child, source) |
|
130
|
return None |
|
131
|
|
|
132
|
def _handle_struct( |
|
133
|
self, node, source: bytes, file_path: str, store: GraphStore, stats: dict |
|
134
|
) -> None: |
|
135
|
name_node = node.child_by_field_name("name") |
|
136
|
if not name_node: |
|
137
|
name_node = next((c for c in node.children if c.type == "type_identifier"), None) |
|
138
|
if not name_node: |
|
139
|
return |
|
140
|
name = _node_text(name_node, source) |
|
141
|
|
|
142
|
kind = ( |
|
143
|
"struct" |
|
144
|
if node.type == "struct_specifier" |
|
145
|
else ("union" if node.type == "union_specifier" else "enum") |
|
146
|
) |
|
147
|
store.create_node( |
|
148
|
NodeLabel.Class, |
|
149
|
{ |
|
150
|
"name": name, |
|
151
|
"file_path": file_path, |
|
152
|
"line_start": node.start_point[0] + 1, |
|
153
|
"line_end": node.end_point[0] + 1, |
|
154
|
"docstring": kind, |
|
155
|
}, |
|
156
|
) |
|
157
|
store.create_edge( |
|
158
|
NodeLabel.File, |
|
159
|
{"path": file_path}, |
|
160
|
EdgeType.CONTAINS, |
|
161
|
NodeLabel.Class, |
|
162
|
{"name": name, "file_path": file_path}, |
|
163
|
) |
|
164
|
stats["classes"] += 1 |
|
165
|
stats["edges"] += 1 |
|
166
|
|
|
167
|
def _handle_include( |
|
168
|
self, node, source: bytes, file_path: str, store: GraphStore, stats: dict |
|
169
|
) -> None: |
|
170
|
path_node = node.child_by_field_name("path") |
|
171
|
if not path_node: |
|
172
|
path_node = next( |
|
173
|
(c for c in node.children if c.type in ("string_literal", "system_lib_string")), |
|
174
|
None, |
|
175
|
) |
|
176
|
if not path_node: |
|
177
|
return |
|
178
|
module = _node_text(path_node, source).strip('<>"') |
|
179
|
store.create_node( |
|
180
|
NodeLabel.Import, |
|
181
|
{ |
|
182
|
"name": module, |
|
183
|
"file_path": file_path, |
|
184
|
"line_start": node.start_point[0] + 1, |
|
185
|
"module": module, |
|
186
|
}, |
|
187
|
) |
|
188
|
store.create_edge( |
|
189
|
NodeLabel.File, |
|
190
|
{"path": file_path}, |
|
191
|
EdgeType.IMPORTS, |
|
192
|
NodeLabel.Import, |
|
193
|
{"name": module, "file_path": file_path}, |
|
194
|
) |
|
195
|
stats["edges"] += 1 |
|
196
|
|
|
197
|
def _extract_calls( |
|
198
|
self, |
|
199
|
fn_node, |
|
200
|
source: bytes, |
|
201
|
file_path: str, |
|
202
|
fn_name: str, |
|
203
|
store: GraphStore, |
|
204
|
stats: dict, |
|
205
|
) -> None: |
|
206
|
def walk(node): |
|
207
|
if node.type == "call_expression": |
|
208
|
func = node.child_by_field_name("function") |
|
209
|
if not func: |
|
210
|
func = next((c for c in node.children if c.type == "identifier"), None) |
|
211
|
if func: |
|
212
|
callee = _node_text(func, source) |
|
213
|
store.create_edge( |
|
214
|
NodeLabel.Function, |
|
215
|
{"name": fn_name, "file_path": file_path}, |
|
216
|
EdgeType.CALLS, |
|
217
|
NodeLabel.Function, |
|
218
|
{"name": callee, "file_path": file_path}, |
|
219
|
) |
|
220
|
stats["edges"] += 1 |
|
221
|
for child in node.children: |
|
222
|
walk(child) |
|
223
|
|
|
224
|
body = fn_node.child_by_field_name("body") |
|
225
|
if not body: |
|
226
|
body = next((c for c in fn_node.children if c.type == "compound_statement"), None) |
|
227
|
if body: |
|
228
|
walk(body) |
|
229
|
|