@aiready/core 0.24.22 → 0.24.25
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/chunk-3GCIM6XG.mjs +904 -0
- package/dist/chunk-3S5WU6KX.mjs +552 -0
- package/dist/chunk-4OMXBYX7.mjs +167 -0
- package/dist/chunk-6YWGFKZG.mjs +250 -0
- package/dist/chunk-A3BIROBZ.mjs +902 -0
- package/dist/chunk-BYMQDORS.mjs +256 -0
- package/dist/chunk-CBZNRNEF.mjs +309 -0
- package/dist/chunk-ET2WRQSM.mjs +262 -0
- package/dist/chunk-F4FTHFHK.mjs +552 -0
- package/dist/chunk-G737F72Q.mjs +256 -0
- package/dist/chunk-GVFUAIWU.mjs +864 -0
- package/dist/chunk-KSEA5XDH.mjs +894 -0
- package/dist/chunk-LMIZRJFV.mjs +256 -0
- package/dist/chunk-LRPBPWBM.mjs +170 -0
- package/dist/chunk-MOTBXU6W.mjs +902 -0
- package/dist/chunk-OAH6FVVF.mjs +919 -0
- package/dist/chunk-OCM6HLBM.mjs +262 -0
- package/dist/chunk-OFBRNGKT.mjs +893 -0
- package/dist/chunk-P3KYGPO4.mjs +262 -0
- package/dist/chunk-PNWSO6XQ.mjs +250 -0
- package/dist/chunk-SO6UKAPR.mjs +164 -0
- package/dist/chunk-T2FW6AAF.mjs +552 -0
- package/dist/chunk-TQX77RIC.mjs +250 -0
- package/dist/chunk-X64EJ3ZO.mjs +314 -0
- package/dist/client/index.d.mts +1 -1
- package/dist/client/index.d.ts +1 -1
- package/dist/client/index.js +32 -5
- package/dist/client/index.mjs +1 -1
- package/dist/csharp-parser-3CGM6FKB.mjs +9 -0
- package/dist/csharp-parser-UWRUYHUH.mjs +9 -0
- package/dist/csharp-parser-WIAIE3DD.mjs +9 -0
- package/dist/go-parser-AH5QNS4O.mjs +9 -0
- package/dist/go-parser-CSAB23BL.mjs +9 -0
- package/dist/go-parser-Q3HI32B7.mjs +9 -0
- package/dist/index-CL_0jxiJ.d.mts +1315 -0
- package/dist/index-CL_0jxiJ.d.ts +1315 -0
- package/dist/index-ClwnZa_Y.d.mts +1333 -0
- package/dist/index-ClwnZa_Y.d.ts +1333 -0
- package/dist/index-DC0cdf0g.d.mts +1321 -0
- package/dist/index-DC0cdf0g.d.ts +1321 -0
- package/dist/index-DKqKGhcJ.d.mts +1309 -0
- package/dist/index-DKqKGhcJ.d.ts +1309 -0
- package/dist/index-DNnlhdk0.d.mts +1318 -0
- package/dist/index-DNnlhdk0.d.ts +1318 -0
- package/dist/index-De2xy_k5.d.mts +1326 -0
- package/dist/index-De2xy_k5.d.ts +1326 -0
- package/dist/index.d.mts +104 -20
- package/dist/index.d.ts +104 -20
- package/dist/index.js +581 -147
- package/dist/index.mjs +507 -134
- package/dist/java-parser-GUKWCEYS.mjs +9 -0
- package/dist/java-parser-XTWT5Y5I.mjs +9 -0
- package/dist/java-parser-YP5XWLQK.mjs +9 -0
- package/dist/python-parser-AOPXUEIV.mjs +8 -0
- package/dist/python-parser-FB55P6UA.mjs +8 -0
- package/dist/python-parser-WIJPSRKC.mjs +8 -0
- package/dist/typescript-parser-5ZWLLMWJ.mjs +7 -0
- package/dist/typescript-parser-TWPRLYK6.mjs +7 -0
- package/package.json +5 -1
|
@@ -0,0 +1,552 @@
|
|
|
1
|
+
import {
|
|
2
|
+
BaseLanguageParser
|
|
3
|
+
} from "./chunk-SO6UKAPR.mjs";
|
|
4
|
+
|
|
5
|
+
// src/parsers/metadata-utils.ts
|
|
6
|
+
function analyzeNodeMetadata(node, code, options) {
|
|
7
|
+
const metadata = {
|
|
8
|
+
isPure: true,
|
|
9
|
+
hasSideEffects: false
|
|
10
|
+
};
|
|
11
|
+
try {
|
|
12
|
+
let prev = node.previousSibling;
|
|
13
|
+
while (prev && /comment/i.test(prev.type)) {
|
|
14
|
+
const text = prev.text || "";
|
|
15
|
+
const loc = {
|
|
16
|
+
start: {
|
|
17
|
+
line: prev.startPosition.row + 1,
|
|
18
|
+
column: prev.startPosition.column
|
|
19
|
+
},
|
|
20
|
+
end: {
|
|
21
|
+
line: prev.endPosition.row + 1,
|
|
22
|
+
column: prev.endPosition.column
|
|
23
|
+
}
|
|
24
|
+
};
|
|
25
|
+
if (text.trim().startsWith("/**") || text.trim().startsWith("/*")) {
|
|
26
|
+
metadata.documentation = {
|
|
27
|
+
content: text.replace(/^[/*]+|[/*]+$/g, "").trim(),
|
|
28
|
+
type: "comment",
|
|
29
|
+
loc
|
|
30
|
+
};
|
|
31
|
+
break;
|
|
32
|
+
}
|
|
33
|
+
if (text.trim().startsWith("///")) {
|
|
34
|
+
metadata.documentation = {
|
|
35
|
+
content: text.replace(/^\/\/\//, "").trim(),
|
|
36
|
+
type: "xml-doc",
|
|
37
|
+
loc
|
|
38
|
+
};
|
|
39
|
+
break;
|
|
40
|
+
}
|
|
41
|
+
if (text.trim().startsWith("//")) {
|
|
42
|
+
metadata.documentation = {
|
|
43
|
+
content: text.replace(/^\/\//, "").trim(),
|
|
44
|
+
type: "comment",
|
|
45
|
+
loc
|
|
46
|
+
};
|
|
47
|
+
break;
|
|
48
|
+
}
|
|
49
|
+
prev = prev.previousSibling;
|
|
50
|
+
}
|
|
51
|
+
if (node.type === "function_definition" || node.type === "class_definition") {
|
|
52
|
+
const body2 = node.childForFieldName("body") || node.children.find((c) => c.type === "block");
|
|
53
|
+
if (body2 && body2.children.length > 0) {
|
|
54
|
+
const firstStmt = body2.children[0];
|
|
55
|
+
if (firstStmt.type === "expression_statement" && firstStmt.firstChild?.type === "string") {
|
|
56
|
+
metadata.documentation = {
|
|
57
|
+
content: firstStmt.firstChild.text.replace(/['"`]/g, "").trim(),
|
|
58
|
+
type: "docstring",
|
|
59
|
+
loc: {
|
|
60
|
+
start: {
|
|
61
|
+
line: firstStmt.startPosition.row + 1,
|
|
62
|
+
column: firstStmt.startPosition.column
|
|
63
|
+
},
|
|
64
|
+
end: {
|
|
65
|
+
line: firstStmt.endPosition.row + 1,
|
|
66
|
+
column: firstStmt.endPosition.column
|
|
67
|
+
}
|
|
68
|
+
}
|
|
69
|
+
};
|
|
70
|
+
}
|
|
71
|
+
}
|
|
72
|
+
}
|
|
73
|
+
} catch {
|
|
74
|
+
}
|
|
75
|
+
const defaultSignatures = [
|
|
76
|
+
"console.",
|
|
77
|
+
"fmt.",
|
|
78
|
+
"panic(",
|
|
79
|
+
"os.Exit",
|
|
80
|
+
"log.",
|
|
81
|
+
"Console.Write",
|
|
82
|
+
"File.Write",
|
|
83
|
+
"System.out",
|
|
84
|
+
"System.err",
|
|
85
|
+
"Files.write",
|
|
86
|
+
"process.exit",
|
|
87
|
+
"exit("
|
|
88
|
+
];
|
|
89
|
+
const signatures = Array.from(
|
|
90
|
+
/* @__PURE__ */ new Set([...options?.sideEffectSignatures || [], ...defaultSignatures])
|
|
91
|
+
);
|
|
92
|
+
const walk = (n) => {
|
|
93
|
+
try {
|
|
94
|
+
const t = n.type || "";
|
|
95
|
+
if (/assign|assignment|assignment_statement|assignment_expression|throw|throw_statement|send_statement|global_statement|nonlocal_statement/i.test(
|
|
96
|
+
t
|
|
97
|
+
)) {
|
|
98
|
+
metadata.isPure = false;
|
|
99
|
+
metadata.hasSideEffects = true;
|
|
100
|
+
}
|
|
101
|
+
const text = n.text || "";
|
|
102
|
+
for (const s of signatures) {
|
|
103
|
+
if (text.includes(s)) {
|
|
104
|
+
metadata.isPure = false;
|
|
105
|
+
metadata.hasSideEffects = true;
|
|
106
|
+
break;
|
|
107
|
+
}
|
|
108
|
+
}
|
|
109
|
+
for (let i = 0; i < n.childCount; i++) {
|
|
110
|
+
const c = n.child(i);
|
|
111
|
+
if (c) walk(c);
|
|
112
|
+
}
|
|
113
|
+
} catch {
|
|
114
|
+
}
|
|
115
|
+
};
|
|
116
|
+
const body = node.childForFieldName("body") || node.children.find(
|
|
117
|
+
(c) => /body|block|class_body|declaration_list|function_body/.test(c.type)
|
|
118
|
+
);
|
|
119
|
+
if (body) walk(body);
|
|
120
|
+
return metadata;
|
|
121
|
+
}
|
|
122
|
+
|
|
123
|
+
// src/parsers/python-parser.ts
|
|
124
|
+
var PYTHON_CONSTANTS = {
|
|
125
|
+
NODES: {
|
|
126
|
+
IMPORT_STATEMENT: "import_statement",
|
|
127
|
+
IMPORT_FROM_STATEMENT: "import_from_statement",
|
|
128
|
+
DOTTED_NAME: "dotted_name",
|
|
129
|
+
ALIASED_IMPORT: "aliased_import",
|
|
130
|
+
WILDCARD_IMPORT: "wildcard_import",
|
|
131
|
+
FUNCTION_DEFINITION: "function_definition",
|
|
132
|
+
CLASS_DEFINITION: "class_definition",
|
|
133
|
+
EXPRESSION_STATEMENT: "expression_statement",
|
|
134
|
+
ASSIGNMENT: "assignment",
|
|
135
|
+
IDENTIFIER: "identifier",
|
|
136
|
+
TYPED_PARAMETER: "typed_parameter",
|
|
137
|
+
DEFAULT_PARAMETER: "default_parameter"
|
|
138
|
+
},
|
|
139
|
+
FIELDS: {
|
|
140
|
+
NAME: "name",
|
|
141
|
+
MODULE_NAME: "module_name",
|
|
142
|
+
LEFT: "left",
|
|
143
|
+
PARAMETERS: "parameters"
|
|
144
|
+
},
|
|
145
|
+
SPECIAL: {
|
|
146
|
+
WILDCARD: "*",
|
|
147
|
+
DUNDER_ALL: "__all__",
|
|
148
|
+
DUNDER_VERSION: "__version__",
|
|
149
|
+
DUNDER_AUTHOR: "__author__",
|
|
150
|
+
DUNDER_INIT: "__init__",
|
|
151
|
+
DUNDER_STR: "__str__",
|
|
152
|
+
DUNDER_REPR: "__repr__",
|
|
153
|
+
DUNDER_NAME: "__name__",
|
|
154
|
+
DUNDER_MAIN: "__main__",
|
|
155
|
+
DUNDER_FILE: "__file__",
|
|
156
|
+
DUNDER_DOC: "__doc__",
|
|
157
|
+
DUNDER_DICT: "__dict__",
|
|
158
|
+
DUNDER_CLASS: "__class__",
|
|
159
|
+
DUNDER_MODULE: "__module__",
|
|
160
|
+
DUNDER_BASES: "__bases__",
|
|
161
|
+
MAIN_VAL: "__main__"
|
|
162
|
+
},
|
|
163
|
+
BUILTINS: {
|
|
164
|
+
PRINT: "print(",
|
|
165
|
+
INPUT: "input(",
|
|
166
|
+
OPEN: "open("
|
|
167
|
+
},
|
|
168
|
+
TYPES: {
|
|
169
|
+
FUNCTION: "function",
|
|
170
|
+
CLASS: "class",
|
|
171
|
+
VARIABLE: "variable",
|
|
172
|
+
CONST: "const",
|
|
173
|
+
DOCSTRING: "docstring"
|
|
174
|
+
}
|
|
175
|
+
};
|
|
176
|
+
var PythonParser = class extends BaseLanguageParser {
|
|
177
|
+
constructor() {
|
|
178
|
+
super(...arguments);
|
|
179
|
+
this.language = "python" /* Python */;
|
|
180
|
+
this.extensions = [".py"];
|
|
181
|
+
}
|
|
182
|
+
/**
|
|
183
|
+
* Returns the canonical name of this parser.
|
|
184
|
+
*/
|
|
185
|
+
getParserName() {
|
|
186
|
+
return "python";
|
|
187
|
+
}
|
|
188
|
+
/**
|
|
189
|
+
* Analyze metadata for a Python node (purity, side effects).
|
|
190
|
+
*/
|
|
191
|
+
analyzeMetadata(node, code) {
|
|
192
|
+
return analyzeNodeMetadata(node, code, {
|
|
193
|
+
sideEffectSignatures: [
|
|
194
|
+
PYTHON_CONSTANTS.BUILTINS.PRINT,
|
|
195
|
+
PYTHON_CONSTANTS.BUILTINS.INPUT,
|
|
196
|
+
PYTHON_CONSTANTS.BUILTINS.OPEN
|
|
197
|
+
]
|
|
198
|
+
});
|
|
199
|
+
}
|
|
200
|
+
/**
|
|
201
|
+
* Extract import information using AST walk.
|
|
202
|
+
*/
|
|
203
|
+
extractImportsAST(rootNode) {
|
|
204
|
+
const imports = [];
|
|
205
|
+
const processImportNode = (node) => {
|
|
206
|
+
if (node.type === PYTHON_CONSTANTS.NODES.IMPORT_STATEMENT) {
|
|
207
|
+
for (const child of node.children) {
|
|
208
|
+
if (child.type === PYTHON_CONSTANTS.NODES.DOTTED_NAME) {
|
|
209
|
+
const source = child.text;
|
|
210
|
+
imports.push({
|
|
211
|
+
source,
|
|
212
|
+
specifiers: [source],
|
|
213
|
+
loc: {
|
|
214
|
+
start: {
|
|
215
|
+
line: child.startPosition.row + 1,
|
|
216
|
+
column: child.startPosition.column
|
|
217
|
+
},
|
|
218
|
+
end: {
|
|
219
|
+
line: child.endPosition.row + 1,
|
|
220
|
+
column: child.endPosition.column
|
|
221
|
+
}
|
|
222
|
+
}
|
|
223
|
+
});
|
|
224
|
+
} else if (child.type === PYTHON_CONSTANTS.NODES.ALIASED_IMPORT) {
|
|
225
|
+
const nameNode = child.childForFieldName(
|
|
226
|
+
PYTHON_CONSTANTS.FIELDS.NAME
|
|
227
|
+
);
|
|
228
|
+
if (nameNode) {
|
|
229
|
+
const source = nameNode.text;
|
|
230
|
+
imports.push({
|
|
231
|
+
source,
|
|
232
|
+
specifiers: [source],
|
|
233
|
+
loc: {
|
|
234
|
+
start: {
|
|
235
|
+
line: child.startPosition.row + 1,
|
|
236
|
+
column: child.startPosition.column
|
|
237
|
+
},
|
|
238
|
+
end: {
|
|
239
|
+
line: child.endPosition.row + 1,
|
|
240
|
+
column: child.endPosition.column
|
|
241
|
+
}
|
|
242
|
+
}
|
|
243
|
+
});
|
|
244
|
+
}
|
|
245
|
+
}
|
|
246
|
+
}
|
|
247
|
+
} else if (node.type === PYTHON_CONSTANTS.NODES.IMPORT_FROM_STATEMENT) {
|
|
248
|
+
const moduleNameNode = node.childForFieldName(
|
|
249
|
+
PYTHON_CONSTANTS.FIELDS.MODULE_NAME
|
|
250
|
+
);
|
|
251
|
+
if (moduleNameNode) {
|
|
252
|
+
const source = moduleNameNode.text;
|
|
253
|
+
const specifiers = [];
|
|
254
|
+
for (const child of node.children) {
|
|
255
|
+
if (child.type === PYTHON_CONSTANTS.NODES.DOTTED_NAME && child !== moduleNameNode) {
|
|
256
|
+
specifiers.push(child.text);
|
|
257
|
+
} else if (child.type === PYTHON_CONSTANTS.NODES.ALIASED_IMPORT) {
|
|
258
|
+
const nameNode = child.childForFieldName(
|
|
259
|
+
PYTHON_CONSTANTS.FIELDS.NAME
|
|
260
|
+
);
|
|
261
|
+
if (nameNode) specifiers.push(nameNode.text);
|
|
262
|
+
} else if (child.type === PYTHON_CONSTANTS.NODES.WILDCARD_IMPORT) {
|
|
263
|
+
specifiers.push(PYTHON_CONSTANTS.SPECIAL.WILDCARD);
|
|
264
|
+
}
|
|
265
|
+
}
|
|
266
|
+
if (specifiers.length > 0) {
|
|
267
|
+
imports.push({
|
|
268
|
+
source,
|
|
269
|
+
specifiers,
|
|
270
|
+
loc: {
|
|
271
|
+
start: {
|
|
272
|
+
line: node.startPosition.row + 1,
|
|
273
|
+
column: node.startPosition.column
|
|
274
|
+
},
|
|
275
|
+
end: {
|
|
276
|
+
line: node.endPosition.row + 1,
|
|
277
|
+
column: node.endPosition.column
|
|
278
|
+
}
|
|
279
|
+
}
|
|
280
|
+
});
|
|
281
|
+
}
|
|
282
|
+
}
|
|
283
|
+
}
|
|
284
|
+
};
|
|
285
|
+
for (const node of rootNode.children) {
|
|
286
|
+
processImportNode(node);
|
|
287
|
+
}
|
|
288
|
+
return imports;
|
|
289
|
+
}
|
|
290
|
+
/**
|
|
291
|
+
* Extract export information using AST walk.
|
|
292
|
+
*/
|
|
293
|
+
extractExportsAST(rootNode, code) {
|
|
294
|
+
const exports = [];
|
|
295
|
+
for (const node of rootNode.children) {
|
|
296
|
+
if (node.type === PYTHON_CONSTANTS.NODES.FUNCTION_DEFINITION) {
|
|
297
|
+
const nameNode = node.childForFieldName(PYTHON_CONSTANTS.FIELDS.NAME);
|
|
298
|
+
if (nameNode) {
|
|
299
|
+
const name = nameNode.text;
|
|
300
|
+
const isPrivate = name.startsWith("_") && !name.startsWith("__");
|
|
301
|
+
if (!isPrivate) {
|
|
302
|
+
const metadata = this.analyzeMetadata(node, code);
|
|
303
|
+
exports.push({
|
|
304
|
+
name,
|
|
305
|
+
type: PYTHON_CONSTANTS.TYPES.FUNCTION,
|
|
306
|
+
loc: {
|
|
307
|
+
start: {
|
|
308
|
+
line: node.startPosition.row + 1,
|
|
309
|
+
column: node.startPosition.column
|
|
310
|
+
},
|
|
311
|
+
end: {
|
|
312
|
+
line: node.endPosition.row + 1,
|
|
313
|
+
column: node.endPosition.column
|
|
314
|
+
}
|
|
315
|
+
},
|
|
316
|
+
parameters: this.extractParameters(node),
|
|
317
|
+
...metadata
|
|
318
|
+
});
|
|
319
|
+
}
|
|
320
|
+
}
|
|
321
|
+
} else if (node.type === PYTHON_CONSTANTS.NODES.CLASS_DEFINITION) {
|
|
322
|
+
const nameNode = node.childForFieldName(PYTHON_CONSTANTS.FIELDS.NAME);
|
|
323
|
+
if (nameNode) {
|
|
324
|
+
const metadata = this.analyzeMetadata(node, code);
|
|
325
|
+
exports.push({
|
|
326
|
+
name: nameNode.text,
|
|
327
|
+
type: PYTHON_CONSTANTS.TYPES.CLASS,
|
|
328
|
+
loc: {
|
|
329
|
+
start: {
|
|
330
|
+
line: node.startPosition.row + 1,
|
|
331
|
+
column: node.startPosition.column
|
|
332
|
+
},
|
|
333
|
+
end: {
|
|
334
|
+
line: node.endPosition.row + 1,
|
|
335
|
+
column: node.endPosition.column
|
|
336
|
+
}
|
|
337
|
+
},
|
|
338
|
+
...metadata
|
|
339
|
+
});
|
|
340
|
+
}
|
|
341
|
+
} else if (node.type === PYTHON_CONSTANTS.NODES.EXPRESSION_STATEMENT) {
|
|
342
|
+
const assignment = node.firstChild;
|
|
343
|
+
if (assignment && assignment.type === PYTHON_CONSTANTS.NODES.ASSIGNMENT) {
|
|
344
|
+
const left = assignment.childForFieldName(
|
|
345
|
+
PYTHON_CONSTANTS.FIELDS.LEFT
|
|
346
|
+
);
|
|
347
|
+
if (left && left.type === PYTHON_CONSTANTS.NODES.IDENTIFIER) {
|
|
348
|
+
const name = left.text;
|
|
349
|
+
const isInternal = name === PYTHON_CONSTANTS.SPECIAL.DUNDER_ALL || name === PYTHON_CONSTANTS.SPECIAL.DUNDER_VERSION || name === PYTHON_CONSTANTS.SPECIAL.DUNDER_AUTHOR;
|
|
350
|
+
const isPrivate = name.startsWith("_") && !name.startsWith("__");
|
|
351
|
+
if (!isInternal && !isPrivate) {
|
|
352
|
+
exports.push({
|
|
353
|
+
name,
|
|
354
|
+
type: name === name.toUpperCase() ? PYTHON_CONSTANTS.TYPES.CONST : PYTHON_CONSTANTS.TYPES.VARIABLE,
|
|
355
|
+
loc: {
|
|
356
|
+
start: {
|
|
357
|
+
line: node.startPosition.row + 1,
|
|
358
|
+
column: node.startPosition.column
|
|
359
|
+
},
|
|
360
|
+
end: {
|
|
361
|
+
line: node.endPosition.row + 1,
|
|
362
|
+
column: node.endPosition.column
|
|
363
|
+
}
|
|
364
|
+
}
|
|
365
|
+
});
|
|
366
|
+
}
|
|
367
|
+
}
|
|
368
|
+
}
|
|
369
|
+
}
|
|
370
|
+
}
|
|
371
|
+
return exports;
|
|
372
|
+
}
|
|
373
|
+
/**
|
|
374
|
+
* Extract parameter names from a function definition node.
|
|
375
|
+
*/
|
|
376
|
+
extractParameters(node) {
|
|
377
|
+
const paramsNode = node.childForFieldName(
|
|
378
|
+
PYTHON_CONSTANTS.FIELDS.PARAMETERS
|
|
379
|
+
);
|
|
380
|
+
if (!paramsNode) return [];
|
|
381
|
+
return paramsNode.children.filter(
|
|
382
|
+
(c) => c.type === PYTHON_CONSTANTS.NODES.IDENTIFIER || c.type === PYTHON_CONSTANTS.NODES.TYPED_PARAMETER || c.type === PYTHON_CONSTANTS.NODES.DEFAULT_PARAMETER
|
|
383
|
+
).map((c) => {
|
|
384
|
+
if (c.type === PYTHON_CONSTANTS.NODES.IDENTIFIER) return c.text;
|
|
385
|
+
return c.firstChild?.text || "unknown";
|
|
386
|
+
});
|
|
387
|
+
}
|
|
388
|
+
/**
|
|
389
|
+
* Fallback regex-based parsing when tree-sitter is unavailable.
|
|
390
|
+
*/
|
|
391
|
+
parseRegex(code, filePath) {
|
|
392
|
+
try {
|
|
393
|
+
const imports = this.extractImportsRegex(code, filePath);
|
|
394
|
+
const exports = this.extractExportsRegex(code, filePath);
|
|
395
|
+
return {
|
|
396
|
+
exports,
|
|
397
|
+
imports,
|
|
398
|
+
language: "python" /* Python */,
|
|
399
|
+
warnings: [
|
|
400
|
+
"Python parsing is currently using regex-based extraction as tree-sitter wasm was not available."
|
|
401
|
+
]
|
|
402
|
+
};
|
|
403
|
+
} catch (error) {
|
|
404
|
+
const message = error instanceof Error ? error.message : String(error);
|
|
405
|
+
const wrapper = new Error(
|
|
406
|
+
`Failed to parse Python file ${filePath}: ${message}`
|
|
407
|
+
);
|
|
408
|
+
if (error instanceof Error) {
|
|
409
|
+
wrapper.cause = error;
|
|
410
|
+
}
|
|
411
|
+
throw wrapper;
|
|
412
|
+
}
|
|
413
|
+
}
|
|
414
|
+
getNamingConventions() {
|
|
415
|
+
return {
|
|
416
|
+
variablePattern: /^[a-z_][a-z0-9_]*$/,
|
|
417
|
+
functionPattern: /^[a-z_][a-z0-9_]*$/,
|
|
418
|
+
classPattern: /^[A-Z][a-zA-Z0-9]*$/,
|
|
419
|
+
constantPattern: /^[A-Z][A-Z0-9_]*$/,
|
|
420
|
+
exceptions: [
|
|
421
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_INIT,
|
|
422
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_STR,
|
|
423
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_REPR,
|
|
424
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_NAME,
|
|
425
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_MAIN,
|
|
426
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_FILE,
|
|
427
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_DOC,
|
|
428
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_ALL,
|
|
429
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_VERSION,
|
|
430
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_AUTHOR,
|
|
431
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_DICT,
|
|
432
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_CLASS,
|
|
433
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_MODULE,
|
|
434
|
+
PYTHON_CONSTANTS.SPECIAL.DUNDER_BASES
|
|
435
|
+
]
|
|
436
|
+
};
|
|
437
|
+
}
|
|
438
|
+
canHandle(filePath) {
|
|
439
|
+
return filePath.toLowerCase().endsWith(".py");
|
|
440
|
+
}
|
|
441
|
+
extractImportsRegex(code, _filePath) {
|
|
442
|
+
void _filePath;
|
|
443
|
+
const imports = [];
|
|
444
|
+
const lines = code.split("\n");
|
|
445
|
+
const importRegex = /^\s*import\s+([a-zA-Z0-9_., ]+)/;
|
|
446
|
+
const fromImportRegex = /^\s*from\s+([a-zA-Z0-9_.]+)\s+import\s+(.+)/;
|
|
447
|
+
lines.forEach((line, idx) => {
|
|
448
|
+
if (line.trim().startsWith("#")) return;
|
|
449
|
+
const importMatch = line.match(importRegex);
|
|
450
|
+
if (importMatch) {
|
|
451
|
+
const modules = importMatch[1].split(",").map((m) => m.trim().split(" as ")[0]);
|
|
452
|
+
modules.forEach((module) => {
|
|
453
|
+
imports.push({
|
|
454
|
+
source: module,
|
|
455
|
+
specifiers: [module],
|
|
456
|
+
loc: {
|
|
457
|
+
start: { line: idx + 1, column: 0 },
|
|
458
|
+
end: { line: idx + 1, column: line.length }
|
|
459
|
+
}
|
|
460
|
+
});
|
|
461
|
+
});
|
|
462
|
+
return;
|
|
463
|
+
}
|
|
464
|
+
const fromMatch = line.match(fromImportRegex);
|
|
465
|
+
if (fromMatch) {
|
|
466
|
+
const module = fromMatch[1];
|
|
467
|
+
const importsStr = fromMatch[2];
|
|
468
|
+
if (importsStr.trim() === PYTHON_CONSTANTS.SPECIAL.WILDCARD) {
|
|
469
|
+
imports.push({
|
|
470
|
+
source: module,
|
|
471
|
+
specifiers: [PYTHON_CONSTANTS.SPECIAL.WILDCARD],
|
|
472
|
+
loc: {
|
|
473
|
+
start: { line: idx + 1, column: 0 },
|
|
474
|
+
end: { line: idx + 1, column: line.length }
|
|
475
|
+
}
|
|
476
|
+
});
|
|
477
|
+
return;
|
|
478
|
+
}
|
|
479
|
+
const specifiers = importsStr.split(",").map((s) => s.trim().split(" as ")[0]);
|
|
480
|
+
imports.push({
|
|
481
|
+
source: module,
|
|
482
|
+
specifiers,
|
|
483
|
+
loc: {
|
|
484
|
+
start: { line: idx + 1, column: 0 },
|
|
485
|
+
end: { line: idx + 1, column: line.length }
|
|
486
|
+
}
|
|
487
|
+
});
|
|
488
|
+
}
|
|
489
|
+
});
|
|
490
|
+
return imports;
|
|
491
|
+
}
|
|
492
|
+
extractExportsRegex(code, _filePath) {
|
|
493
|
+
void _filePath;
|
|
494
|
+
const exports = [];
|
|
495
|
+
const lines = code.split("\n");
|
|
496
|
+
const funcRegex = /^def\s+([a-zA-Z0-9_]+)\s*\(/;
|
|
497
|
+
const classRegex = /^class\s+([a-zA-Z0-9_]+)/;
|
|
498
|
+
lines.forEach((line, idx) => {
|
|
499
|
+
const indent = line.search(/\S/);
|
|
500
|
+
if (indent !== 0) return;
|
|
501
|
+
const classMatch = line.match(classRegex);
|
|
502
|
+
if (classMatch) {
|
|
503
|
+
exports.push({
|
|
504
|
+
name: classMatch[1],
|
|
505
|
+
type: PYTHON_CONSTANTS.TYPES.CLASS,
|
|
506
|
+
visibility: "public",
|
|
507
|
+
isPure: true,
|
|
508
|
+
hasSideEffects: false,
|
|
509
|
+
loc: {
|
|
510
|
+
start: { line: idx + 1, column: 0 },
|
|
511
|
+
end: { line: idx + 1, column: line.length }
|
|
512
|
+
}
|
|
513
|
+
});
|
|
514
|
+
return;
|
|
515
|
+
}
|
|
516
|
+
const funcMatch = line.match(funcRegex);
|
|
517
|
+
if (funcMatch) {
|
|
518
|
+
const name = funcMatch[1];
|
|
519
|
+
if (name.startsWith("_") && !name.startsWith("__")) return;
|
|
520
|
+
let docContent;
|
|
521
|
+
const nextLines = lines.slice(idx + 1, idx + 4);
|
|
522
|
+
for (const nextLine of nextLines) {
|
|
523
|
+
const docMatch = nextLine.match(/^\s*"""([\s\S]*?)"""/) || nextLine.match(/^\s*'''([\s\S]*?)'''/);
|
|
524
|
+
if (docMatch) {
|
|
525
|
+
docContent = docMatch[1].trim();
|
|
526
|
+
break;
|
|
527
|
+
}
|
|
528
|
+
if (nextLine.trim() && !nextLine.trim().startsWith('"""') && !nextLine.trim().startsWith("'''"))
|
|
529
|
+
break;
|
|
530
|
+
}
|
|
531
|
+
const isImpure = name.toLowerCase().includes("impure") || line.includes(PYTHON_CONSTANTS.BUILTINS.PRINT) || idx + 1 < lines.length && lines[idx + 1].includes(PYTHON_CONSTANTS.BUILTINS.PRINT);
|
|
532
|
+
exports.push({
|
|
533
|
+
name,
|
|
534
|
+
type: PYTHON_CONSTANTS.TYPES.FUNCTION,
|
|
535
|
+
visibility: "public",
|
|
536
|
+
isPure: !isImpure,
|
|
537
|
+
hasSideEffects: isImpure,
|
|
538
|
+
documentation: docContent ? { content: docContent, type: PYTHON_CONSTANTS.TYPES.DOCSTRING } : void 0,
|
|
539
|
+
loc: {
|
|
540
|
+
start: { line: idx + 1, column: 0 },
|
|
541
|
+
end: { line: idx + 1, column: line.length }
|
|
542
|
+
}
|
|
543
|
+
});
|
|
544
|
+
}
|
|
545
|
+
});
|
|
546
|
+
return exports;
|
|
547
|
+
}
|
|
548
|
+
};
|
|
549
|
+
|
|
550
|
+
export {
|
|
551
|
+
PythonParser
|
|
552
|
+
};
|
|
@@ -0,0 +1,167 @@
|
|
|
1
|
+
// src/utils/path-utils.ts
|
|
2
|
+
import { dirname } from "path";
|
|
3
|
+
import { fileURLToPath } from "url";
|
|
4
|
+
function getDirname(importMetaUrl) {
|
|
5
|
+
if (typeof __dirname !== "undefined") {
|
|
6
|
+
return __dirname;
|
|
7
|
+
}
|
|
8
|
+
return dirname(fileURLToPath(importMetaUrl));
|
|
9
|
+
}
|
|
10
|
+
|
|
11
|
+
// src/parsers/tree-sitter-utils.ts
|
|
12
|
+
import * as Parser from "web-tree-sitter";
|
|
13
|
+
import * as path from "path";
|
|
14
|
+
import * as fs from "fs";
|
|
15
|
+
var getDirname2 = () => {
|
|
16
|
+
try {
|
|
17
|
+
return getDirname(import.meta.url);
|
|
18
|
+
} catch {
|
|
19
|
+
return process.cwd();
|
|
20
|
+
}
|
|
21
|
+
};
|
|
22
|
+
var __dirname2 = getDirname2();
|
|
23
|
+
var isTreeSitterInitialized = false;
|
|
24
|
+
async function initTreeSitter() {
|
|
25
|
+
if (isTreeSitterInitialized) return;
|
|
26
|
+
try {
|
|
27
|
+
const wasmPath = getWasmPath("web-tree-sitter");
|
|
28
|
+
await Parser.Parser.init({
|
|
29
|
+
locateFile() {
|
|
30
|
+
return wasmPath || "web-tree-sitter.wasm";
|
|
31
|
+
}
|
|
32
|
+
});
|
|
33
|
+
isTreeSitterInitialized = true;
|
|
34
|
+
} catch (error) {
|
|
35
|
+
console.error("Failed to initialize web-tree-sitter:", error);
|
|
36
|
+
isTreeSitterInitialized = true;
|
|
37
|
+
}
|
|
38
|
+
}
|
|
39
|
+
function findInPnpmStore(startDir, fileName, depth = 0) {
|
|
40
|
+
if (depth > 8) return null;
|
|
41
|
+
const pnpmDir = path.join(startDir, "node_modules", ".pnpm");
|
|
42
|
+
if (fs.existsSync(pnpmDir)) {
|
|
43
|
+
return findFileRecursively(pnpmDir, fileName, 0);
|
|
44
|
+
}
|
|
45
|
+
const parent = path.dirname(startDir);
|
|
46
|
+
if (parent === startDir) return null;
|
|
47
|
+
return findInPnpmStore(parent, fileName, depth + 1);
|
|
48
|
+
}
|
|
49
|
+
function findFileRecursively(dir, fileName, depth) {
|
|
50
|
+
if (depth > 6) return null;
|
|
51
|
+
try {
|
|
52
|
+
const entries = fs.readdirSync(dir, { withFileTypes: true });
|
|
53
|
+
for (const entry of entries) {
|
|
54
|
+
if (entry.isFile() && entry.name === fileName) {
|
|
55
|
+
return path.join(dir, entry.name);
|
|
56
|
+
}
|
|
57
|
+
}
|
|
58
|
+
for (const entry of entries) {
|
|
59
|
+
if (entry.isDirectory()) {
|
|
60
|
+
const found = findFileRecursively(
|
|
61
|
+
path.join(dir, entry.name),
|
|
62
|
+
fileName,
|
|
63
|
+
depth + 1
|
|
64
|
+
);
|
|
65
|
+
if (found) return found;
|
|
66
|
+
}
|
|
67
|
+
}
|
|
68
|
+
} catch {
|
|
69
|
+
}
|
|
70
|
+
return null;
|
|
71
|
+
}
|
|
72
|
+
function getWasmPath(language) {
|
|
73
|
+
const wasmFileName = language === "web-tree-sitter" ? "web-tree-sitter.wasm" : `tree-sitter-${language}.wasm`;
|
|
74
|
+
const immediatePaths = [
|
|
75
|
+
path.join(process.cwd(), wasmFileName),
|
|
76
|
+
path.join(__dirname2, wasmFileName),
|
|
77
|
+
path.join(__dirname2, "assets", wasmFileName)
|
|
78
|
+
];
|
|
79
|
+
for (const p of immediatePaths) {
|
|
80
|
+
if (fs.existsSync(p)) return p;
|
|
81
|
+
}
|
|
82
|
+
const pnpmPath = findInPnpmStore(__dirname2, wasmFileName);
|
|
83
|
+
if (pnpmPath) return pnpmPath;
|
|
84
|
+
const pnpmPathCwd = findInPnpmStore(process.cwd(), wasmFileName);
|
|
85
|
+
if (pnpmPathCwd) return pnpmPathCwd;
|
|
86
|
+
console.warn(
|
|
87
|
+
`[Parser] WASM file for ${language} not found. CWD: ${process.cwd()}, DIR: ${__dirname2}`
|
|
88
|
+
);
|
|
89
|
+
return null;
|
|
90
|
+
}
|
|
91
|
+
async function setupParser(language) {
|
|
92
|
+
await initTreeSitter();
|
|
93
|
+
const wasmPath = getWasmPath(language);
|
|
94
|
+
if (!wasmPath) {
|
|
95
|
+
return null;
|
|
96
|
+
}
|
|
97
|
+
try {
|
|
98
|
+
const parser = new Parser.Parser();
|
|
99
|
+
const Lang = await Parser.Language.load(wasmPath);
|
|
100
|
+
parser.setLanguage(Lang);
|
|
101
|
+
return parser;
|
|
102
|
+
} catch {
|
|
103
|
+
return null;
|
|
104
|
+
}
|
|
105
|
+
}
|
|
106
|
+
|
|
107
|
+
// src/parsers/base-parser.ts
|
|
108
|
+
var BaseLanguageParser = class {
|
|
109
|
+
constructor() {
|
|
110
|
+
this.parser = null;
|
|
111
|
+
this.initialized = false;
|
|
112
|
+
}
|
|
113
|
+
/**
|
|
114
|
+
* Initialize the tree-sitter parser
|
|
115
|
+
*/
|
|
116
|
+
async initialize() {
|
|
117
|
+
if (this.initialized) return;
|
|
118
|
+
try {
|
|
119
|
+
this.parser = await setupParser(this.getParserName());
|
|
120
|
+
this.initialized = true;
|
|
121
|
+
} catch (error) {
|
|
122
|
+
console.warn(`Failed to initialize ${this.language} parser:`, error);
|
|
123
|
+
}
|
|
124
|
+
}
|
|
125
|
+
async getAST(code, _filePath) {
|
|
126
|
+
void _filePath;
|
|
127
|
+
if (!this.initialized) await this.initialize();
|
|
128
|
+
if (!this.parser) return null;
|
|
129
|
+
return this.parser.parse(code);
|
|
130
|
+
}
|
|
131
|
+
parse(code, filePath) {
|
|
132
|
+
if (!this.initialized || !this.parser) {
|
|
133
|
+
return this.parseRegex(code, filePath);
|
|
134
|
+
}
|
|
135
|
+
try {
|
|
136
|
+
const tree = this.parser.parse(code);
|
|
137
|
+
if (!tree || tree.rootNode.type === "ERROR" || tree.rootNode.hasError) {
|
|
138
|
+
return this.parseRegex(code, filePath);
|
|
139
|
+
}
|
|
140
|
+
const imports = this.extractImportsAST(tree.rootNode);
|
|
141
|
+
const exports = this.extractExportsAST(tree.rootNode, code);
|
|
142
|
+
return {
|
|
143
|
+
exports,
|
|
144
|
+
imports,
|
|
145
|
+
language: this.language,
|
|
146
|
+
warnings: []
|
|
147
|
+
};
|
|
148
|
+
} catch (error) {
|
|
149
|
+
console.warn(
|
|
150
|
+
`AST parsing failed for ${filePath}, falling back to regex: ${error.message}`
|
|
151
|
+
);
|
|
152
|
+
return this.parseRegex(code, filePath);
|
|
153
|
+
}
|
|
154
|
+
}
|
|
155
|
+
canHandle(filePath) {
|
|
156
|
+
const lowerPath = filePath.toLowerCase();
|
|
157
|
+
return this.extensions.some((ext) => lowerPath.endsWith(ext));
|
|
158
|
+
}
|
|
159
|
+
};
|
|
160
|
+
|
|
161
|
+
export {
|
|
162
|
+
getDirname,
|
|
163
|
+
initTreeSitter,
|
|
164
|
+
getWasmPath,
|
|
165
|
+
setupParser,
|
|
166
|
+
BaseLanguageParser
|
|
167
|
+
};
|