notebooklm
VerifiedInstall, authenticate, troubleshoot, and operate Gemini Notebook through the notebooklm-py CLI or typed async Python API. Use for notebook and source management, grounded chat and research, and artifact generation or download when the user mentions Gemini Notebook, notebooklm-py, the notebooklm CLI, or its Python API. Do not use for the generic Gemini API or unrelated content creation.
scripts/decode_mobile_grpc.py
View on GitHub#!/usr/bin/env python3
"""Recover NotebookLM Android gRPC message *shapes* from captured protobuf bodies.
Companion to ``scripts/capture_mobile_grpc.js`` and
``docs/android/capture.md``. Reads the ``.pb`` files a capture session wrote
(gRPC envelope already stripped) and prints a merged schema per method+direction:
field number, wire type, whether it repeats, and — for length-delimited fields —
whether the payload is a nested message, a string, or opaque bytes.
Privacy: string/bytes *values* are never printed, only their length. Varint and
fixed32/64 values ARE printed because they are flags, enums, page sizes, and
``google.protobuf.Timestamp`` seconds/nanos — not private text. Even so, do not
commit this tool's output verbatim; summarise shapes in docs instead (see
``docs/android/endpoints.md``).
Usage:
python scripts/decode_mobile_grpc.py <capture-dir> [method-substring]
The wire format is walked with a hand-rolled varint reader so the script has no
third-party dependency (the project does not ship protobuf). Length-delimited
payloads are speculatively parsed as nested messages; this over-eagerly parses
UTF-8 text that happens to be valid wire format, so treat very deep trees under
free-text fields (chat turns, source/artifact content) as approximate.
"""
from __future__ import annotations
import collections
import glob
import os
import struct
import sys
WIRE = {0: "varint", 1: "i64", 2: "len", 5: "i32"}
def read_varint(b: bytes, i: int) -> tuple[int, int]:
shift = 0
val = 0
while True:
if i >= len(b):
raise ValueError("truncated varint")
x = b[i]
i += 1
val |= (x & 0x7F) << shift
if not x & 0x80:
return val, i
shift += 7
if shift > 70:
raise ValueError("varint too long")
def parse_message(b: bytes) -> list[tuple[int, int, object]]:
"""Parse ``b`` as a protobuf message, consuming it fully or raising."""
fields: list[tuple[int, int, object]] = []
i, n = 0, len(b)
while i < n:
key, i = read_varint(b, i)
field, wt = key >> 3, key & 7
if field == 0:
raise ValueError("field 0")
if wt == 0:
val, i = read_varint(b, i)
fields.append((field, wt, val))
elif wt == 1:
if i + 8 > n:
raise ValueError("truncated i64")
fields.append((field, wt, b[i : i + 8]))
i += 8
elif wt == 5:
if i + 4 > n:
raise ValueError("truncated i32")
fields.append((field, wt, b[i : i + 4]))
i += 4
elif wt == 2:
ln, i = read_varint(b, i)
if i + ln > n:
raise ValueError("truncated len")
fields.append((field, wt, b[i : i + ln]))
i += ln
else:
raise ValueError(f"bad wire type {wt}")
return fields
def classify_len(payload: bytes, depth: int) -> tuple[str, object]:
if not payload:
return "empty", None
if depth < 12:
try:
sub = parse_message(payload)
if sub and all(1 <= f[0] <= 536_870_911 for f in sub):
return "message", sub
except ValueError:
pass
try:
payload.decode("utf-8")
return "string", len(payload)
except UnicodeDecodeError:
return "bytes", len(payload)
class Node:
__slots__ = ("wtypes", "max_repeat", "child", "scalars", "str_lens", "kinds")
def __init__(self) -> None:
self.wtypes: set[str] = set()
self.max_repeat = 0
self.child: dict[int, Node] | None = None
self.scalars: set[int] = set()
self.str_lens: set[int] = set()
self.kinds: set[str] = set()
def merge(fields: list[tuple[int, int, object]], schema: dict[int, Node], depth: int) -> None:
per_field = collections.Counter(f[0] for f in fields)
seen: set[int] = set()
for field, wt, payload in fields:
node = schema.setdefault(field, Node())
node.wtypes.add(WIRE[wt])
if field not in seen:
node.max_repeat = max(node.max_repeat, per_field[field])
seen.add(field)
if wt == 0 and isinstance(payload, int):
if payload < 1_000_000_000_000 and len(node.scalars) < 6:
node.scalars.add(payload)
elif wt == 5 and isinstance(payload, (bytes, bytearray)):
node.scalars.add(struct.unpack("<I", payload)[0])
elif wt == 1 and isinstance(payload, (bytes, bytearray)):
# fixed64. WIRE declares this type, so omitting it here silently
# dropped every i64 sample from the reported scalars.
node.scalars.add(struct.unpack("<Q", payload)[0])
elif wt == 2 and isinstance(payload, (bytes, bytearray)):
kind, info = classify_len(bytes(payload), depth)
node.kinds.add(kind)
if kind == "message":
node.child = node.child or {}
merge(info, node.child, depth + 1) # type: ignore[arg-type]
elif kind in ("string", "bytes") and isinstance(info, int):
node.str_lens.add(info)
def render(schema: dict[int, Node], indent: int = 1) -> list[str]:
out: list[str] = []
pad = " " * indent
for field in sorted(schema):
node = schema[field]
wt = "/".join(sorted(node.wtypes))
rep = " (repeated)" if node.max_repeat > 1 else ""
line = f"{pad}#{field} [{wt}]{rep}"
if node.kinds:
line += " " + "/".join(sorted(node.kinds))
lens = sorted(node.str_lens)
if lens and node.kinds & {"string", "bytes"}:
line += f" len={lens[0] if len(lens) == 1 else f'{lens[0]}..{lens[-1]}'}"
if node.scalars and "message" not in node.kinds:
line += f" = {sorted(node.scalars)[:6]}"
out.append(line)
if node.child:
out.extend(render(node.child, indent + 1))
return out
def main() -> None:
if len(sys.argv) < 2:
sys.exit(f"usage: {sys.argv[0]} <capture-dir> [method-substring]")
directory = sys.argv[1]
filt = sys.argv[2].lower() if len(sys.argv) > 2 else ""
groups: dict[tuple[str, str], list[str]] = collections.defaultdict(list)
for path in sorted(glob.glob(os.path.join(directory, "*.pb"))):
method, direction, _idx = (
os.path.basename(path).split("_", 1)[1].rsplit(".pb", 1)[0].split(".")
)
if filt and filt not in method.lower():
continue
groups[(method, direction)].append(path)
for method, direction in sorted(groups):
files = groups[(method, direction)]
schema: dict[int, Node] = {}
sizes, parsed = [], 0
for path in files:
with open(path, "rb") as fh:
body = fh.read()
sizes.append(len(body))
try:
merge(parse_message(body), schema, 0)
parsed += 1
except ValueError as exc:
print(f" !! {os.path.basename(path)}: {exc}")
print(
f"\n### {method}.{direction} "
f"(n={len(files)}, parsed={parsed}, bytes={min(sizes)}..{max(sizes)})"
)
print("\n".join(render(schema)))
if __name__ == "__main__":
main()
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| deprecation-and-migrationaddyosmani/agent-skillsManages deprecation and migration. Use when removing old systems, APIs, or features. Use when migrating users from one implementation to another. Use when migrating a database schema in production, such as renaming or dropping a column without downtime (expand/contract). Use when deciding whether to maintain or sunset existing code. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| performance-optimizationaddyosmani/agent-skillsOptimizes application performance across frontend, backend, queries, and databases. Use when performance requirements exist, when you suspect performance regressions, when Core Web Vitals or load times need improvement, when N+1 query patterns need fixing, or when profiling reveals bottlenecks. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |