|
1
|
""" |
|
2
|
DiffAnalyzer — maps git diff output to affected knowledge-graph nodes. |
|
3
|
|
|
4
|
Given uncommitted changes in a repository, this module tells you: |
|
5
|
- which files changed |
|
6
|
- which functions/classes/methods fall within the changed line ranges |
|
7
|
- which knowledge nodes (Concept, Rule, Decision, WikiPage) are linked |
|
8
|
to those symbols via ANNOTATES, IMPLEMENTS, or GOVERNS edges |
|
9
|
- a single impact_summary dict bundling all of the above |
|
10
|
|
|
11
|
Usage:: |
|
12
|
|
|
13
|
from pathlib import Path |
|
14
|
from navegador.diff import DiffAnalyzer |
|
15
|
from navegador.graph.store import GraphStore |
|
16
|
|
|
17
|
store = GraphStore.sqlite(".navegador/graph.db") |
|
18
|
analyzer = DiffAnalyzer(store, Path(".")) |
|
19
|
print(analyzer.impact_summary()) |
|
20
|
""" |
|
21
|
|
|
22
|
from __future__ import annotations |
|
23
|
|
|
24
|
import json |
|
25
|
import subprocess |
|
26
|
from pathlib import Path |
|
27
|
from typing import Any |
|
28
|
|
|
29
|
from navegador.graph.store import GraphStore |
|
30
|
from navegador.vcs import GitAdapter |
|
31
|
|
|
32
|
# ── Cypher helpers ──────────────────────────────────────────────────────────── |
|
33
|
|
|
34
|
# All symbols (Function / Class / Method) in a given file with their line ranges |
|
35
|
_SYMBOLS_IN_FILE = """ |
|
36
|
MATCH (n) |
|
37
|
WHERE (n:Function OR n:Class OR n:Method) |
|
38
|
AND n.file_path = $file_path |
|
39
|
RETURN labels(n)[0] AS type, |
|
40
|
n.name AS name, |
|
41
|
n.file_path AS file_path, |
|
42
|
n.line_start AS line_start, |
|
43
|
n.line_end AS line_end |
|
44
|
ORDER BY n.line_start |
|
45
|
""" |
|
46
|
|
|
47
|
# Knowledge nodes reachable from a code symbol in one hop via cross-layer edges. |
|
48
|
# Handles: |
|
49
|
# (symbol)-[:ANNOTATES|IMPLEMENTS]->(knowledge) — code → knowledge |
|
50
|
# (knowledge)-[:GOVERNS|ANNOTATES]->(symbol) — knowledge → code |
|
51
|
_KNOWLEDGE_FOR_SYMBOL = """ |
|
52
|
MATCH (sym) |
|
53
|
WHERE (sym:Function OR sym:Class OR sym:Method) |
|
54
|
AND sym.name = $name |
|
55
|
AND sym.file_path = $file_path |
|
56
|
OPTIONAL MATCH (sym)-[:ANNOTATES|IMPLEMENTS]->(k1) |
|
57
|
WHERE (k1:Concept OR k1:Rule OR k1:Decision OR k1:WikiPage) |
|
58
|
OPTIONAL MATCH (k2)-[:GOVERNS|ANNOTATES]->(sym) |
|
59
|
WHERE (k2:Concept OR k2:Rule OR k2:Decision OR k2:WikiPage) |
|
60
|
WITH collect(DISTINCT k1) + collect(DISTINCT k2) AS knowledge_nodes |
|
61
|
UNWIND knowledge_nodes AS k |
|
62
|
RETURN DISTINCT |
|
63
|
labels(k)[0] AS type, |
|
64
|
k.name AS name, |
|
65
|
coalesce(k.description, '') AS description, |
|
66
|
coalesce(k.domain, '') AS domain, |
|
67
|
coalesce(k.status, '') AS status |
|
68
|
""" |
|
69
|
|
|
70
|
|
|
71
|
def _parse_unified_diff_hunks(diff_output: str) -> dict[str, list[tuple[int, int]]]: |
|
72
|
""" |
|
73
|
Parse the output of ``git diff -U0 HEAD`` and return a mapping of |
|
74
|
file path → list of (start_line, end_line) changed ranges. |
|
75
|
|
|
76
|
Only new/modified lines (``+`` prefix) are tracked; deleted-only hunks |
|
77
|
contribute the surrounding context line instead (line before deletion). |
|
78
|
""" |
|
79
|
result: dict[str, list[tuple[int, int]]] = {} |
|
80
|
current_file: str | None = None |
|
81
|
current_new_start = 0 |
|
82
|
current_new_count = 0 |
|
83
|
|
|
84
|
for line in diff_output.splitlines(): |
|
85
|
# New file header: +++ b/path/to/file |
|
86
|
if line.startswith("+++ b/"): |
|
87
|
current_file = line[6:] |
|
88
|
if current_file not in result: |
|
89
|
result[current_file] = [] |
|
90
|
elif line.startswith("+++ /dev/null"): |
|
91
|
current_file = None # deleted file — skip |
|
92
|
elif line.startswith("@@ "): |
|
93
|
# Hunk header: @@ -old_start[,old_count] +new_start[,new_count] @@ |
|
94
|
try: |
|
95
|
new_info = line.split("+")[1].split("@@")[0].strip() |
|
96
|
if "," in new_info: |
|
97
|
new_start_str, new_count_str = new_info.split(",", 1) |
|
98
|
current_new_start = int(new_start_str) |
|
99
|
current_new_count = int(new_count_str) |
|
100
|
else: |
|
101
|
current_new_start = int(new_info) |
|
102
|
current_new_count = 1 |
|
103
|
if current_file and current_new_count > 0: |
|
104
|
end = current_new_start + max(current_new_count - 1, 0) |
|
105
|
result.setdefault(current_file, []).append((current_new_start, end)) |
|
106
|
except (ValueError, IndexError): |
|
107
|
pass |
|
108
|
|
|
109
|
return result |
|
110
|
|
|
111
|
|
|
112
|
def _lines_overlap( |
|
113
|
changed_ranges: list[tuple[int, int]], |
|
114
|
line_start: int | None, |
|
115
|
line_end: int | None, |
|
116
|
) -> bool: |
|
117
|
"""Return True when any changed range overlaps with [line_start, line_end].""" |
|
118
|
if line_start is None: |
|
119
|
return False |
|
120
|
sym_start = line_start |
|
121
|
sym_end = line_end if line_end is not None else line_start |
|
122
|
for r_start, r_end in changed_ranges: |
|
123
|
if r_start <= sym_end and r_end >= sym_start: |
|
124
|
return True |
|
125
|
return False |
|
126
|
|
|
127
|
|
|
128
|
# ── DiffAnalyzer ───────────────────────────────────────────────────────────── |
|
129
|
|
|
130
|
|
|
131
|
class DiffAnalyzer: |
|
132
|
"""Maps git diff output to affected graph nodes. |
|
133
|
|
|
134
|
Parameters |
|
135
|
---------- |
|
136
|
store: |
|
137
|
An open :class:`~navegador.graph.store.GraphStore`. |
|
138
|
repo_path: |
|
139
|
Root of the git repository to inspect. |
|
140
|
""" |
|
141
|
|
|
142
|
def __init__(self, store: GraphStore, repo_path: Path) -> None: |
|
143
|
self.store = store |
|
144
|
self.repo_path = Path(repo_path) |
|
145
|
self._git = GitAdapter(self.repo_path) |
|
146
|
|
|
147
|
# ------------------------------------------------------------------ |
|
148
|
# Public API |
|
149
|
# ------------------------------------------------------------------ |
|
150
|
|
|
151
|
def changed_files(self) -> list[str]: |
|
152
|
"""Return paths of files with uncommitted changes (staged + unstaged). |
|
153
|
|
|
154
|
Delegates to :class:`~navegador.vcs.GitAdapter` which runs |
|
155
|
``git diff HEAD --name-only``. |
|
156
|
""" |
|
157
|
return self._git.changed_files() |
|
158
|
|
|
159
|
def changed_lines(self) -> dict[str, list[tuple[int, int]]]: |
|
160
|
"""Return a mapping of file path → list of (start, end) changed line ranges. |
|
161
|
|
|
162
|
Runs ``git diff -U0 HEAD`` and parses the unified diff hunk headers. |
|
163
|
Falls back to the whole file (line 1 to a large sentinel) when the |
|
164
|
diff cannot be parsed precisely — ensuring callers always get a result. |
|
165
|
""" |
|
166
|
result = subprocess.run( |
|
167
|
["git", "diff", "-U0", "HEAD"], |
|
168
|
cwd=self.repo_path, |
|
169
|
capture_output=True, |
|
170
|
text=True, |
|
171
|
check=False, |
|
172
|
) |
|
173
|
if result.returncode != 0 or not result.stdout.strip(): |
|
174
|
# No HEAD yet (initial commit) or empty diff — treat all changed |
|
175
|
# files as fully-changed using a wide sentinel range. |
|
176
|
return {f: [(1, 999_999)] for f in self.changed_files()} |
|
177
|
|
|
178
|
parsed = _parse_unified_diff_hunks(result.stdout) |
|
179
|
|
|
180
|
# Files returned by changed_files() but missing from the parsed diff |
|
181
|
# (e.g. binary files, new untracked files added with --intent-to-add) |
|
182
|
# get a full-file sentinel range. |
|
183
|
for f in self.changed_files(): |
|
184
|
if f not in parsed: |
|
185
|
parsed[f] = [(1, 999_999)] |
|
186
|
|
|
187
|
return parsed |
|
188
|
|
|
189
|
def affected_symbols(self) -> list[dict[str, Any]]: |
|
190
|
"""Return functions/classes/methods whose line ranges overlap changed lines. |
|
191
|
|
|
192
|
Each entry is a dict with the keys: |
|
193
|
``type``, ``name``, ``file_path``, ``line_start``, ``line_end``. |
|
194
|
""" |
|
195
|
line_map = self.changed_lines() |
|
196
|
symbols: list[dict[str, Any]] = [] |
|
197
|
seen: set[tuple[str, str]] = set() |
|
198
|
|
|
199
|
for file_path, changed_ranges in line_map.items(): |
|
200
|
rows = self.store.query(_SYMBOLS_IN_FILE, {"file_path": file_path}) |
|
201
|
if not rows or not rows.result_set: |
|
202
|
continue |
|
203
|
for row in rows.result_set: |
|
204
|
sym_type, name, fp, line_start, line_end = row |
|
205
|
key = (name, fp or file_path) |
|
206
|
if key in seen: |
|
207
|
continue |
|
208
|
if _lines_overlap(changed_ranges, line_start, line_end): |
|
209
|
seen.add(key) |
|
210
|
symbols.append( |
|
211
|
{ |
|
212
|
"type": sym_type, |
|
213
|
"name": name, |
|
214
|
"file_path": fp or file_path, |
|
215
|
"line_start": line_start, |
|
216
|
"line_end": line_end, |
|
217
|
} |
|
218
|
) |
|
219
|
|
|
220
|
return symbols |
|
221
|
|
|
222
|
def affected_knowledge(self) -> list[dict[str, Any]]: |
|
223
|
"""Return knowledge nodes linked to affected code symbols. |
|
224
|
|
|
225
|
Traverses ANNOTATES, IMPLEMENTS (code → knowledge) and |
|
226
|
GOVERNS, ANNOTATES (knowledge → code) edges from each affected symbol. |
|
227
|
|
|
228
|
Each entry is a dict with the keys: |
|
229
|
``type``, ``name``, ``description``, ``domain``, ``status``. |
|
230
|
""" |
|
231
|
symbols = self.affected_symbols() |
|
232
|
knowledge: list[dict[str, Any]] = [] |
|
233
|
seen: set[str] = set() |
|
234
|
|
|
235
|
for sym in symbols: |
|
236
|
rows = self.store.query( |
|
237
|
_KNOWLEDGE_FOR_SYMBOL, |
|
238
|
{"name": sym["name"], "file_path": sym["file_path"]}, |
|
239
|
) |
|
240
|
if not rows or not rows.result_set: |
|
241
|
continue |
|
242
|
for row in rows.result_set: |
|
243
|
k_type, k_name, k_desc, k_domain, k_status = row |
|
244
|
if k_name in seen: |
|
245
|
continue |
|
246
|
seen.add(k_name) |
|
247
|
knowledge.append( |
|
248
|
{ |
|
249
|
"type": k_type, |
|
250
|
"name": k_name, |
|
251
|
"description": k_desc, |
|
252
|
"domain": k_domain, |
|
253
|
"status": k_status, |
|
254
|
} |
|
255
|
) |
|
256
|
|
|
257
|
return knowledge |
|
258
|
|
|
259
|
def impact_summary(self) -> dict[str, Any]: |
|
260
|
"""Return a full impact summary: files, symbols, and knowledge nodes. |
|
261
|
|
|
262
|
The returned dict has the keys: |
|
263
|
``files`` — list of changed file paths |
|
264
|
``symbols`` — list of affected symbol dicts |
|
265
|
``knowledge`` — list of affected knowledge node dicts |
|
266
|
``counts`` — sub-dict with ``files``, ``symbols``, ``knowledge`` counts |
|
267
|
""" |
|
268
|
files = self.changed_files() |
|
269
|
symbols = self.affected_symbols() |
|
270
|
knowledge = self.affected_knowledge() |
|
271
|
|
|
272
|
return { |
|
273
|
"files": files, |
|
274
|
"symbols": symbols, |
|
275
|
"knowledge": knowledge, |
|
276
|
"counts": { |
|
277
|
"files": len(files), |
|
278
|
"symbols": len(symbols), |
|
279
|
"knowledge": len(knowledge), |
|
280
|
}, |
|
281
|
} |
|
282
|
|
|
283
|
# ------------------------------------------------------------------ |
|
284
|
# Formatters |
|
285
|
# ------------------------------------------------------------------ |
|
286
|
|
|
287
|
def to_json(self) -> str: |
|
288
|
"""Serialise impact_summary() as a JSON string.""" |
|
289
|
return json.dumps(self.impact_summary(), indent=2, default=str) |
|
290
|
|
|
291
|
def to_markdown(self) -> str: |
|
292
|
"""Render impact_summary() as a human-readable Markdown string.""" |
|
293
|
summary = self.impact_summary() |
|
294
|
lines: list[str] = ["# Diff Impact Summary\n"] |
|
295
|
|
|
296
|
# Files |
|
297
|
lines.append(f"## Changed Files ({summary['counts']['files']})\n") |
|
298
|
if summary["files"]: |
|
299
|
for f in summary["files"]: |
|
300
|
lines.append(f"- `{f}`") |
|
301
|
else: |
|
302
|
lines.append("_No changed files._") |
|
303
|
lines.append("") |
|
304
|
|
|
305
|
# Symbols |
|
306
|
lines.append(f"## Affected Symbols ({summary['counts']['symbols']})\n") |
|
307
|
if summary["symbols"]: |
|
308
|
for sym in summary["symbols"]: |
|
309
|
loc = f"line {sym['line_start']}" if sym.get("line_start") else "" |
|
310
|
lines.append( |
|
311
|
f"- **{sym['type']}** `{sym['name']}` — `{sym['file_path']}`" |
|
312
|
+ (f" ({loc})" if loc else "") |
|
313
|
) |
|
314
|
else: |
|
315
|
lines.append("_No affected symbols found in graph._") |
|
316
|
lines.append("") |
|
317
|
|
|
318
|
# Knowledge |
|
319
|
lines.append(f"## Affected Knowledge ({summary['counts']['knowledge']})\n") |
|
320
|
if summary["knowledge"]: |
|
321
|
for k in summary["knowledge"]: |
|
322
|
desc = f" — {k['description']}" if k.get("description") else "" |
|
323
|
lines.append(f"- **{k['type']}** `{k['name']}`{desc}") |
|
324
|
else: |
|
325
|
lines.append("_No linked knowledge nodes._") |
|
326
|
|
|
327
|
return "\n".join(lines) |
|
328
|
|