gitlytics 0.4.0__tar.gz → 0.4.2__tar.gz

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.
Files changed (53) hide show
  1. {gitlytics-0.4.0 → gitlytics-0.4.2}/PKG-INFO +1 -1
  2. {gitlytics-0.4.0 → gitlytics-0.4.2}/pyproject.toml +1 -1
  3. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/__init__.py +1 -1
  4. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/api.py +116 -36
  5. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/automation.py +14 -5
  6. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/cli.py +15 -3
  7. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/core.py +72 -62
  8. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/process.py +38 -10
  9. gitlytics-0.4.2/src/gitlytics/static/_headers +6 -0
  10. gitlytics-0.4.2/src/gitlytics/static/_redirects +1 -0
  11. gitlytics-0.4.0/src/gitlytics/static/assets/CardModal-DmHoAnrK.js → gitlytics-0.4.2/src/gitlytics/static/assets/CardModal-BnM9BB99.js +1 -1
  12. gitlytics-0.4.0/src/gitlytics/static/assets/ExportModal-BF_sosAP.js → gitlytics-0.4.2/src/gitlytics/static/assets/ExportModal-B7bldVwi.js +1 -1
  13. gitlytics-0.4.0/src/gitlytics/static/assets/exports-CrU6Ta4Q.js → gitlytics-0.4.2/src/gitlytics/static/assets/exports-BM_YcnUf.js +3 -3
  14. gitlytics-0.4.2/src/gitlytics/static/assets/index-BRI0yQ1K.css +1 -0
  15. gitlytics-0.4.2/src/gitlytics/static/assets/index-DO5EQgEU.js +124 -0
  16. gitlytics-0.4.0/src/gitlytics/static/assets/index.es-ComYkcii.js → gitlytics-0.4.2/src/gitlytics/static/assets/index.es-DjwqRfc7.js +1 -1
  17. gitlytics-0.4.0/src/gitlytics/static/assets/jspdf.es.min-vktmDsgQ.js → gitlytics-0.4.2/src/gitlytics/static/assets/jspdf.es.min-B8eoAQ72.js +3 -3
  18. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/index.html +26 -25
  19. gitlytics-0.4.2/src/gitlytics/static/robots.txt +4 -0
  20. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics.egg-info/PKG-INFO +1 -1
  21. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics.egg-info/SOURCES.txt +9 -7
  22. {gitlytics-0.4.0 → gitlytics-0.4.2}/tests/test_api.py +25 -0
  23. {gitlytics-0.4.0 → gitlytics-0.4.2}/tests/test_cli.py +1 -2
  24. {gitlytics-0.4.0 → gitlytics-0.4.2}/tests/test_process.py +30 -3
  25. {gitlytics-0.4.0 → gitlytics-0.4.2}/tests/test_username.py +6 -9
  26. gitlytics-0.4.0/src/gitlytics/static/assets/index-D5qdeNvu.css +0 -1
  27. gitlytics-0.4.0/src/gitlytics/static/assets/index-hb6NMx5Y.js +0 -119
  28. gitlytics-0.4.0/src/gitlytics/static/robots.txt +0 -5
  29. {gitlytics-0.4.0 → gitlytics-0.4.2}/LICENSE +0 -0
  30. {gitlytics-0.4.0 → gitlytics-0.4.2}/README.md +0 -0
  31. {gitlytics-0.4.0 → gitlytics-0.4.2}/setup.cfg +0 -0
  32. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/__main__.py +0 -0
  33. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/android-chrome-192x192.png +0 -0
  34. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/android-chrome-512x512.png +0 -0
  35. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/apple-touch-icon.png +0 -0
  36. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/assets/html2canvas-pro.esm-9xys3ejh.js +0 -0
  37. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/assets/html2canvas.esm-DXEQVQnt.js +0 -0
  38. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/assets/purify.es-CC4Brkr_.js +0 -0
  39. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/favicon-16x16.png +0 -0
  40. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/favicon-32x32.png +0 -0
  41. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/favicon-48x48.png +0 -0
  42. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/favicon.ico +0 -0
  43. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/gitlytics-logo.png +0 -0
  44. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/google-search-icon-48x48.png +0 -0
  45. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/google-search-icon.png +0 -0
  46. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics/static/sitemap.xml +0 -0
  47. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics.egg-info/dependency_links.txt +0 -0
  48. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics.egg-info/entry_points.txt +0 -0
  49. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics.egg-info/requires.txt +0 -0
  50. {gitlytics-0.4.0 → gitlytics-0.4.2}/src/gitlytics.egg-info/top_level.txt +0 -0
  51. {gitlytics-0.4.0 → gitlytics-0.4.2}/tests/test_automation.py +0 -0
  52. {gitlytics-0.4.0 → gitlytics-0.4.2}/tests/test_core.py +0 -0
  53. {gitlytics-0.4.0 → gitlytics-0.4.2}/tests/test_star_history.py +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: gitlytics
3
- Version: 0.4.0
3
+ Version: 0.4.2
4
4
  Summary: Monitor and automate your GitHub repository traffic analytics.
5
5
  Author-email: Ameya Chopade <ameyaccod171@gmail.com>
6
6
  License: Apache-2.0
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
4
4
 
5
5
  [project]
6
6
  name = "gitlytics"
7
- version = "0.4.0"
7
+ version = "0.4.2"
8
8
  description = "Monitor and automate your GitHub repository traffic analytics."
9
9
  readme = "README.md"
10
10
  requires-python = ">=3.9"
@@ -9,7 +9,7 @@ from pathlib import Path
9
9
 
10
10
  # Single source of truth for the package version.
11
11
  # Mirrors the version in pyproject.toml — keep them in sync.
12
- __version__ = "0.4.0"
12
+ __version__ = "0.4.2"
13
13
 
14
14
  __all__ = [
15
15
  "fetch_traffic",
@@ -5,13 +5,16 @@ Powers the FastAPI backend — serves traffic data and the React dashboard to th
5
5
  import hashlib
6
6
  import logging
7
7
  import os
8
+ import secrets
8
9
  import shutil
10
+ import threading as _threading
9
11
  import time as _time
10
12
  import uuid
13
+ from collections import OrderedDict
11
14
  from pathlib import Path
12
15
 
13
16
  import pandas as pd
14
- from fastapi import FastAPI, HTTPException, Body, File, Header, Response, UploadFile
17
+ from fastapi import FastAPI, HTTPException, Body, File, Header, Request, Response, UploadFile, Depends
15
18
  from fastapi.responses import FileResponse, JSONResponse
16
19
  from fastapi.middleware.cors import CORSMiddleware
17
20
  from fastapi.staticfiles import StaticFiles
@@ -21,13 +24,14 @@ from gitlytics.core import (
21
24
  get_user_profile,
22
25
  get_public_user,
23
26
  get_public_repos,
27
+ validate_github_username,
24
28
  fetch_traffic_data,
25
29
  fetch_deep_stats_for_top,
26
30
  fetch_star_history,
27
31
  GitHubRateLimitError,
28
32
  StarHistoryFetchError,
29
33
  )
30
- from gitlytics.process import process_uploaded_csv, build_react_payload
34
+ from gitlytics.process import process_uploaded_csv, build_react_payload, _to_bool
31
35
 
32
36
  logger = logging.getLogger(__name__)
33
37
 
@@ -38,9 +42,11 @@ app = FastAPI(title="GitHub Traffic API")
38
42
  _ALLOWED_ORIGINS = [
39
43
  "http://localhost",
40
44
  "http://localhost:3000",
45
+ "http://localhost:4173",
41
46
  "http://localhost:8000",
42
47
  "http://127.0.0.1",
43
48
  "http://127.0.0.1:3000",
49
+ "http://127.0.0.1:4173",
44
50
  "http://127.0.0.1:8000",
45
51
  ]
46
52
 
@@ -55,13 +61,62 @@ app.add_middleware(
55
61
  )
56
62
 
57
63
 
58
- _auth_cache: dict = {} # sha256_prefix -> (valid, username, expires_at)
59
- _AUTH_CACHE_TTL = 300 # 5 minutes
64
+ # In-process short-term cache for token validation. Cleared when the
65
+ # Python process exits — same lifetime as the CLI / FastAPI worker.
66
+ # OrderedDict + insert-order eviction gives us a bounded LRU without
67
+ # pulling in functools.lru_cache (which can't be cleared on TTL).
68
+ _auth_cache: "OrderedDict[str, tuple]" = OrderedDict()
69
+ _auth_cache_lock = _threading.Lock()
70
+ _AUTH_CACHE_TTL = 300 # 5 minutes — applied only to positive (valid=True) results.
71
+ _AUTH_NEG_TTL = 10 # short TTL for negative results so a transient 5xx doesn't lock the user out.
72
+ _AUTH_CACHE_MAX = 256 # cap entries so a long-running worker can't grow unbounded
60
73
 
61
74
  # Hard cap on CSV upload size — prevents DoS via /api/upload-csv.
62
75
  _MAX_UPLOAD_BYTES = 25 * 1024 * 1024
63
76
 
64
77
 
78
+ class _SecurityHeadersMiddleware:
79
+ def __init__(self, app):
80
+ self.app = app
81
+
82
+ async def __call__(self, scope, receive, send):
83
+ if scope["type"] != "http":
84
+ await self.app(scope, receive, send)
85
+ return
86
+
87
+ async def send_wrapper(message):
88
+ if message["type"] == "http.response.start":
89
+ headers = list(message.get("headers", []))
90
+ headers.extend([
91
+ (b"x-content-type-options", b"nosniff"),
92
+ (b"x-frame-options", b"DENY"),
93
+ (b"referrer-policy", b"strict-origin-when-cross-origin"),
94
+ ])
95
+ message = {**message, "headers": headers}
96
+ await send(message)
97
+
98
+ await self.app(scope, receive, send_wrapper)
99
+
100
+
101
+ app.add_middleware(_SecurityHeadersMiddleware)
102
+
103
+
104
+ def _require_upload_access(
105
+ request: Request,
106
+ x_upload_key: str | None = Header(default=None),
107
+ ) -> None:
108
+ client_host = (request.client.host if request.client else "").lower()
109
+ if client_host in ("127.0.0.1", "::1", "testclient", "testserver"):
110
+ return
111
+ expected = (os.environ.get("GITLYTICS_UPLOAD_KEY") or "").strip()
112
+ if not expected:
113
+ raise HTTPException(status_code=403, detail="Upload not allowed from this host.")
114
+ got = (x_upload_key or "").strip()
115
+ if got and secrets.compare_digest(got, expected):
116
+ return
117
+ raise HTTPException(status_code=403, detail="Upload not allowed from this host.")
118
+
119
+
65
120
  def _get_token(token: str = None) -> str:
66
121
  # C-2: explicit empty string must not fall through to the env token
67
122
  if token and token.strip():
@@ -71,15 +126,28 @@ def _get_token(token: str = None) -> str:
71
126
 
72
127
  def _validate_token_cached(token: str):
73
128
  # M-1: cache validation results to avoid a double HTTP round-trip on every /api/traffic call
74
- from gitlytics.core import validate_token as _validate_token
75
129
  key = hashlib.sha256(token.encode()).hexdigest()[:16]
76
130
  now = _time.time()
77
- if key in _auth_cache:
78
- valid, username, expires = _auth_cache[key]
79
- if now < expires:
80
- return valid, username
81
- valid, username = _validate_token(token)
82
- _auth_cache[key] = (valid, username, now + _AUTH_CACHE_TTL)
131
+ # Lock guards the read-then-write so two threads can't both miss the
132
+ # cache and double-fetch the same token (FastAPI runs sync routes in
133
+ # a worker threadpool, so this race is reachable under load).
134
+ with _auth_cache_lock:
135
+ cached = _auth_cache.get(key)
136
+ if cached is not None:
137
+ valid, username, expires = cached
138
+ if now < expires:
139
+ # Refresh insertion order so an active key isn't evicted FIFO-style.
140
+ _auth_cache.move_to_end(key)
141
+ return valid, username
142
+ valid, username = validate_token(token)
143
+ # Negative results get a short TTL so a transient upstream 5xx doesn't
144
+ # lock the user out for 5 minutes; positive results cache fully.
145
+ ttl = _AUTH_CACHE_TTL if valid else _AUTH_NEG_TTL
146
+ with _auth_cache_lock:
147
+ _auth_cache[key] = (valid, username, now + ttl)
148
+ _auth_cache.move_to_end(key)
149
+ while len(_auth_cache) > _AUTH_CACHE_MAX:
150
+ _auth_cache.popitem(last=False)
83
151
  return valid, username
84
152
 
85
153
 
@@ -119,29 +187,19 @@ def auth(token: str = Body("", embed=True)):
119
187
 
120
188
  @app.post("/api/username")
121
189
  def get_username_data(username: str = Body("", embed=True)):
122
- """Fetches public profile and repos for any GitHub username — no token required."""
123
190
  if not username or not username.strip():
124
191
  raise HTTPException(status_code=400, detail="Username is required.")
125
192
  try:
126
- profile = get_public_user(username.strip())
127
- # Distinguish "user found" (login matches the request) from "GitHub failed
128
- # and we returned a stub". Returning 200 for the latter would silently mask
129
- # upstream outages.
130
- if profile.get("login") == username.strip() and profile.get("html_url"):
131
- pass
132
- elif profile.get("login") == username.strip():
133
- # Found but GitHub didn't include html_url — still treat as success.
134
- pass
135
- else:
136
- raise HTTPException(status_code=502, detail="GitHub did not return a profile.")
137
- repos = get_public_repos(username.strip())
193
+ clean = validate_github_username(username.strip())
194
+ profile = get_public_user(clean)
195
+ repos = get_public_repos(clean)
138
196
  return {"profile": profile, "repos": repos}
139
197
  except ValueError as exc:
140
198
  raise HTTPException(status_code=404, detail=str(exc))
141
- except HTTPException:
142
- raise
199
+ except RuntimeError:
200
+ raise HTTPException(status_code=502, detail="GitHub is temporarily unavailable.")
143
201
  except Exception as exc:
144
- logger.warning(f"Username fetch failed for {username}: {exc}")
202
+ logger.warning("Username fetch failed for %s: %s", username, exc)
145
203
  raise HTTPException(status_code=500, detail="Failed to fetch GitHub data.")
146
204
 
147
205
 
@@ -189,6 +247,13 @@ def get_traffic(token: str = Body("", embed=True)):
189
247
  if not df.empty:
190
248
  df = df.drop_duplicates(subset=["date", "repository"], keep="last")
191
249
 
250
+ # Normalize `is_private` to a real bool. CSV round-trip + pd.concat
251
+ # widens the dtype to object (mix of bool, "True"/"False" strings, and
252
+ # NaN); without this, downstream code calling `bool("False")` would
253
+ # label every public repo as private.
254
+ if not df.empty and "is_private" in df.columns:
255
+ df["is_private"] = df["is_private"].map(_to_bool).astype(bool)
256
+
192
257
  df = df.replace([float('inf'), float('-inf')], None).where(pd.notnull(df), None)
193
258
 
194
259
  # Build a quick view-sum map to find the top 20 repos for deep fetching.
@@ -237,12 +302,20 @@ def get_star_history_endpoint(
237
302
  # sense than 429 here so the SPA can distinguish "GitHub is throttling us"
238
303
  # from "your request was malformed".
239
304
  raise HTTPException(status_code=400, detail=str(exc))
305
+ except ValueError as exc:
306
+ # fetch_star_history raises ValueError for empty owner / repo / trailing slash;
307
+ # returning 400 is more honest than letting Starlette surface a 500.
308
+ raise HTTPException(status_code=400, detail=str(exc))
240
309
  response.headers["Cache-Control"] = "public, max-age=86400"
241
310
  return {"owner": owner, "repo": repo, "points": points}
242
311
 
243
312
 
244
313
  @app.post("/api/upload-csv")
245
- def upload_csv(file: UploadFile = File(...)):
314
+ def upload_csv(
315
+ request: Request,
316
+ file: UploadFile = File(...),
317
+ _: None = Depends(_require_upload_access),
318
+ ):
246
319
  try:
247
320
  data_dir = os.environ.get("GITLYTICS_DATA_DIR")
248
321
  if data_dir:
@@ -277,7 +350,8 @@ def upload_csv(file: UploadFile = File(...)):
277
350
  except HTTPException:
278
351
  raise
279
352
  except Exception as e:
280
- raise HTTPException(status_code=400, detail=str(e))
353
+ logger.warning("CSV upload failed: %s", e)
354
+ raise HTTPException(status_code=400, detail="Failed to process CSV upload.")
281
355
 
282
356
 
283
357
  # Static file serving
@@ -299,21 +373,27 @@ def serve_index():
299
373
  def serve_spa_fallback(full_path: str):
300
374
  """SPA catch-all — returns index.html so React Router handles navigation."""
301
375
  # Any unhandled /api/* path is a real 404, not a SPA route.
302
- if full_path.startswith("api/"):
376
+ if full_path.startswith("api/") or full_path.startswith("assets/"):
303
377
  return JSONResponse(status_code=404, content={"error": "Not found."})
304
378
 
305
- asset_file = frontend_dir / full_path
306
- if asset_file.exists() and asset_file.is_file():
379
+ # Confine the resolved path to frontend_dir — otherwise a request like
380
+ # `GET /..%2F..%2Fapp.py` would let `asset_file` escape the static
381
+ # directory and serve arbitrary files the worker can read.
382
+ frontend_root = frontend_dir.resolve()
383
+ try:
384
+ asset_file = (frontend_dir / full_path).resolve()
385
+ except (OSError, ValueError):
386
+ # `resolve()` can raise on broken symlinks or odd encodings.
387
+ return JSONResponse(status_code=404, content={"error": "Not found."})
388
+ is_inside = asset_file == frontend_root or asset_file.is_relative_to(frontend_root)
389
+ if is_inside and asset_file.is_file():
307
390
  return FileResponse(asset_file)
308
391
 
309
392
  index_file = frontend_dir / "index.html"
310
393
  if index_file.exists():
311
394
  return FileResponse(index_file)
312
395
 
313
- return JSONResponse(
314
- status_code=503,
315
- content={"error": "Dashboard assets not found in the package installation."}
316
- )
396
+ return JSONResponse(status_code=404, content={"error": "Not found."})
317
397
 
318
398
 
319
399
  assets_dir = frontend_dir / "assets"
@@ -177,11 +177,15 @@ def run_sync_cycle(token: str, repo_names=None, data_dir="./data", output_mode="
177
177
 
178
178
  final_rows.sort(key=lambda x: (str(x.get("date", "")), str(x.get("repository", ""))))
179
179
 
180
- # Write everything back to the CSV file
181
- with open(csv_path, "w", newline="", encoding="utf-8") as f:
180
+ # Atomic write: dump to a sibling .tmp file then os.replace() it into
181
+ # place. A crash mid-write leaves the original CSV intact, so a single
182
+ # interrupted sync never corrupts the user's whole traffic history.
183
+ tmp_path = csv_path + ".tmp"
184
+ with open(tmp_path, "w", newline="", encoding="utf-8") as f:
182
185
  writer = csv.DictWriter(f, fieldnames=existing_fields, extrasaction='ignore')
183
186
  writer.writeheader()
184
187
  writer.writerows(final_rows)
188
+ os.replace(tmp_path, csv_path)
185
189
 
186
190
  logger.info(f"Successfully processed traffic data. Added {new_records_added} new daily records to {csv_path}")
187
191
 
@@ -234,19 +238,24 @@ def run_sync(token: str, repo_names=None, data_dir="./data", output_mode="monthl
234
238
  # Always re-validate the token before fetching to catch expiry early
235
239
  is_valid, msg = validate_token(token)
236
240
  if not is_valid:
237
- # Distinguish between a dead token (stop forever) and a network blip (retry next cycle)
241
+ # Distinguish between a dead token (stop forever) and a network
242
+ # blip (retry next cycle). Both 401 and 403 are terminal —
243
+ # 401 means the token is gone; 403 means the token is fine but
244
+ # lacks permissions, which is also a "never coming back" state.
238
245
  msg_lower = msg.lower()
239
246
  is_auth_failure = (
240
247
  "401" in msg
248
+ or "403" in msg
241
249
  or "authentication failed" in msg_lower
242
250
  or "invalid token" in msg_lower
243
251
  or "revoked" in msg_lower
244
252
  or "bad credentials" in msg_lower
253
+ or "insufficient permissions" in msg_lower
245
254
  )
246
255
  if is_auth_failure:
247
256
  logger.critical(
248
- "FATAL: Token is expired, revoked, or invalid (401 Unauthorized). "
249
- "Terminating daemon to prevent zombie process."
257
+ "FATAL: Token is expired, revoked, or lacks permissions "
258
+ "(401/403). Terminating daemon to prevent zombie process."
250
259
  )
251
260
  sys.exit(1)
252
261
  else:
@@ -72,7 +72,12 @@ def main():
72
72
 
73
73
  # ── DASHBOARD subcommand ──────────────────────────────────────────────────
74
74
  dash_parser = subparsers.add_parser("dashboard", help="Serve the local React dashboard.")
75
- dash_parser.add_argument("--host", default="127.0.0.1", help="Host IP.")
75
+ dash_parser.add_argument("--host", default="127.0.0.1", help="Host IP (default 127.0.0.1).")
76
+ dash_parser.add_argument(
77
+ "--public-bind",
78
+ action="store_true",
79
+ help="Allow binding to 0.0.0.0 (insecure without reverse-proxy auth).",
80
+ )
76
81
  dash_parser.add_argument("--port", type=int, default=8000, help="Port to bind.")
77
82
  dash_parser.add_argument("-t", "--token", help="GitHub token for headless TV display.")
78
83
  dash_parser.add_argument("--data-dir", help="Inject historical CSV database.")
@@ -151,9 +156,16 @@ def main():
151
156
  sys.exit(2)
152
157
 
153
158
  elif args.command == "dashboard":
154
- print(f"\n[Gitlytics] Starting Gitlytics Dashboard on http://{args.host}:{args.port}\n")
159
+ host = args.host
160
+ if host == "0.0.0.0" and not args.public_bind:
161
+ print("❌ Error: Use --public-bind to listen on 0.0.0.0 (default is 127.0.0.1).")
162
+ sys.exit(2)
163
+ if args.public_bind:
164
+ host = "0.0.0.0"
165
+ print("⚠️ Warning: dashboard bound to all interfaces — use only behind a trusted network or reverse proxy.")
166
+ print(f"\n[Gitlytics] Starting Gitlytics Dashboard on http://{host}:{args.port}\n")
155
167
  serve_dashboard(
156
- host=args.host,
168
+ host=host,
157
169
  port=args.port,
158
170
  token=token,
159
171
  data_dir=args.data_dir
@@ -8,6 +8,7 @@ Handles fetching traffic and deep metadata from the GitHub API.
8
8
  """
9
9
  import json
10
10
  import logging
11
+ import re
11
12
 
12
13
  import requests
13
14
  import pandas as pd
@@ -17,16 +18,14 @@ from datetime import datetime, timezone, timedelta
17
18
  logger = logging.getLogger(__name__)
18
19
 
19
20
  BASE = "https://api.github.com"
21
+ GITHUB_USERNAME_RE = re.compile(r"^[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,37}[a-zA-Z0-9])?$", re.IGNORECASE)
20
22
  # Single source of truth for the GitHub media-type + API version headers.
21
23
  _GITHUB_BASE_HEADERS = {
22
24
  "Accept": "application/vnd.github+json",
23
25
  "X-GitHub-Api-Version": "2022-11-28",
24
26
  }
25
27
  _PUBLIC_HEADERS = dict(_GITHUB_BASE_HEADERS)
26
- _MAX_RELEASE_PAGES = 10 # hard cap when walking /releases
27
- _MAX_PR_PAGES = 10 # hard cap when walking /pulls
28
- _MAX_COMMIT_PAGES = 50 # hard cap when counting /commits (50 * 100 = 5000)
29
- _PER_PAGE = 100 # request size for paginated endpoints
28
+ _PER_PAGE = 100 # page size for paginated endpoints
30
29
 
31
30
 
32
31
  class GitHubRateLimitError(Exception):
@@ -52,7 +51,8 @@ def validate_token(token: str) -> tuple[bool, str]:
52
51
  except requests.exceptions.ConnectionError:
53
52
  return False, "No internet connection."
54
53
  except Exception as e:
55
- return False, str(e)
54
+ logger.warning("Token validation error: %s", e)
55
+ return False, "Authentication failed."
56
56
 
57
57
  if r.status_code == 200:
58
58
  data = r.json()
@@ -84,8 +84,16 @@ def get_user_profile(token: str) -> dict:
84
84
  return {"login": "", "name": "", "avatar_url": "", "followers": 0, "following": 0}
85
85
 
86
86
 
87
+ def validate_github_username(username: str) -> str:
88
+ u = (username or "").strip()
89
+ if not u or not GITHUB_USERNAME_RE.match(u):
90
+ raise ValueError("Invalid GitHub username.")
91
+ return u
92
+
93
+
87
94
  def get_public_user(username: str) -> dict:
88
95
  """Fetches a public GitHub user profile — no PAT required."""
96
+ username = validate_github_username(username)
89
97
  try:
90
98
  r = requests.get(f"{BASE}/users/{username}", headers=_PUBLIC_HEADERS, timeout=10)
91
99
  if r.status_code == 200:
@@ -106,15 +114,15 @@ def get_public_user(username: str) -> dict:
106
114
  }
107
115
  if r.status_code == 404:
108
116
  raise ValueError(f"User '{username}' not found.")
109
- # 5xx and other unexpected codes — log a warning and return a stub so the
110
- # caller (the API endpoint) can return a 502 without leaking the failure.
111
117
  logger.warning(f"Failed to fetch user {username}: HTTP {r.status_code}")
118
+ raise RuntimeError(f"GitHub returned HTTP {r.status_code}.")
112
119
  except ValueError:
113
- # Re-raise user-not-found as-is.
120
+ raise
121
+ except RuntimeError:
114
122
  raise
115
123
  except Exception as exc:
116
124
  logger.warning(f"Could not fetch public user profile: {exc}")
117
- return {"login": username, "name": username, "avatar_url": ""}
125
+ raise RuntimeError("GitHub is unreachable.") from exc
118
126
 
119
127
 
120
128
  def _normalise_repo(repo: dict) -> dict:
@@ -176,6 +184,7 @@ def _fetch_public_repos_by_stars(username: str, per_page: int) -> list:
176
184
 
177
185
 
178
186
  def get_public_repos(username: str, max_repos: int = 50) -> list:
187
+ username = validate_github_username(username)
179
188
  """
180
189
  Fetches up to max_repos best public repos for a username using two calls:
181
190
  1. Most recently updated (catches active projects)
@@ -227,29 +236,33 @@ def _safe_get(url: str, headers: dict, params: dict = None) -> tuple:
227
236
  return {}, -1
228
237
 
229
238
 
230
- def _walk_paginated_count(url: str, headers: dict, params: dict, max_pages: int) -> int:
231
- """
232
- Count items across pages. Returns the total count, capped at max_pages * per_page.
233
- The GitHub REST API for /pulls and /releases does not return a total_count,
234
- so we have to walk pages.
239
+ def _count_via_link_header(url: str, headers: dict, params: dict = None) -> int:
240
+ """Get item count in exactly 1 API call using the Link header pagination trick.
241
+
242
+ Requests per_page=1 so GitHub includes a Link header with the last page
243
+ number, which equals the total count. Falls back to len(response) when
244
+ the repository has only one page of results (no Link header present).
235
245
  """
236
- total = 0
237
- per_page = int(params.get("per_page", _PER_PAGE)) if params else _PER_PAGE
238
- page_params = dict(params or {})
239
- page_params["per_page"] = per_page
240
- for page in range(1, max_pages + 1):
241
- page_params["page"] = page
242
- data, status = _safe_get(url, headers, page_params)
243
- if status != 200 or not isinstance(data, list) or not data:
244
- break
245
- total += len(data)
246
- if len(data) < per_page:
247
- break
248
- return total
246
+ p = dict(params or {})
247
+ p["per_page"] = 1
248
+ try:
249
+ r = requests.get(url, headers=headers, params=p, timeout=10)
250
+ if r.status_code != 200:
251
+ return 0
252
+ link = r.headers.get("Link", "")
253
+ match = re.search(r'[?&]page=(\d+)[^>]*>;\s*rel="last"', link)
254
+ if match:
255
+ return int(match.group(1))
256
+ # No Link header = all results fit on one page
257
+ data = r.json()
258
+ return len(data) if isinstance(data, list) else 0
259
+ except Exception as exc:
260
+ logger.warning("Link-header count failed for %s: %s", url, exc)
261
+ return 0
249
262
 
250
263
 
251
264
  def get_deep_repo_stats(token: str, full_name: str) -> dict:
252
- """Fetches commit activity, open PRs, releases, and community health for one repo."""
265
+ """Fetch commit count, open PRs, releases, and community health in exactly 4 API calls."""
253
266
  h = make_headers(token)
254
267
  stats = {
255
268
  "total_commits": None,
@@ -262,60 +275,55 @@ def get_deep_repo_stats(token: str, full_name: str) -> dict:
262
275
  "has_code_of_conduct": None,
263
276
  }
264
277
 
265
- # Commit activity — GitHub computes this async and returns 202 when not ready.
266
- # NOTE: stats/commit_activity only reports the trailing 52-week window, so for
267
- # repos where the user expects to see lifetime history we count via
268
- # /repos/{full_name}/commits?per_page=100 and walk pages. GitHub hard-caps
269
- # /commits at ~1000 results on a single unauthenticated/traffic endpoint, so
270
- # the lifetime count returned here is the lifetime visible to the API.
271
- ca_url = f"{BASE}/repos/{full_name}/stats/commit_activity"
272
- ca_data, status = _safe_get(ca_url, h)
273
- if status == 202:
274
- logger.info(f"Commit activity not ready yet (202) for {full_name}; will populate on next fetch.")
278
+ # Call 1: Commit activity (52-week window). GitHub computes this asynchronously
279
+ # and returns 202 while it's being prepared. When ready it gives us a weekly
280
+ # breakdown which also powers the CommitActivity chart in the dashboard.
281
+ ca_data, ca_status = _safe_get(f"{BASE}/repos/{full_name}/stats/commit_activity", h)
282
+ if ca_status == 202:
283
+ logger.info("Commit activity computing (202) for %s — will populate on next fetch.", full_name)
275
284
  if isinstance(ca_data, list) and ca_data:
276
285
  stats["total_commits"] = sum(week.get("total", 0) for week in ca_data)
277
286
  else:
278
- # Fall back to a lifetime count when stats/commit_activity is empty or
279
- # not yet computed (202 / empty list). Walking /commits is slower but
280
- # always returns the real lifetime history for repos under ~10k commits.
281
- commits_url = f"{BASE}/repos/{full_name}/commits"
287
+ # Call 1b (fallback): commit_activity not ready — use Link header trick to
288
+ # count commits in 1 request instead of walking up to 50 pages.
282
289
  try:
283
- stats["total_commits"] = _walk_paginated_count(
284
- commits_url, h, {"per_page": _PER_PAGE}, _MAX_COMMIT_PAGES
290
+ stats["total_commits"] = _count_via_link_header(
291
+ f"{BASE}/repos/{full_name}/commits", h
285
292
  )
286
293
  except Exception as exc:
287
- logger.warning(f"Could not fetch commit count for {full_name}: {exc}")
294
+ logger.warning("Could not fetch commit count for %s: %s", full_name, exc)
288
295
 
289
- # Open PRs — walk pages because GitHub's /pulls endpoint doesn't return a total.
290
- pr_url = f"{BASE}/repos/{full_name}/pulls"
296
+ # Call 2: Open PRs — Link header trick replaces up to 10 paginated calls.
291
297
  try:
292
- stats["open_prs"] = _walk_paginated_count(pr_url, h, {"state": "open"}, _MAX_PR_PAGES)
298
+ stats["open_prs"] = _count_via_link_header(
299
+ f"{BASE}/repos/{full_name}/pulls", h, {"state": "open"}
300
+ )
293
301
  except Exception as exc:
294
- logger.warning(f"Could not fetch open PRs for {full_name}: {exc}")
302
+ logger.warning("Could not fetch open PRs for %s: %s", full_name, exc)
295
303
 
296
- # Community health profile — README, license, contributing, CoC
304
+ # Call 3: Community health (README, license, contributing, CoC) — 1 call, unchanged.
297
305
  cp_data, _ = _safe_get(f"{BASE}/repos/{full_name}/community/profile", h)
298
306
  if isinstance(cp_data, dict) and "files" in cp_data:
299
- files = cp_data.get("files", {})
307
+ files = cp_data["files"]
300
308
  stats["has_readme"] = bool(files.get("readme"))
301
309
  stats["has_license"] = bool(files.get("license"))
302
310
  stats["has_contributing"] = bool(files.get("contributing"))
303
311
  stats["has_code_of_conduct"] = bool(files.get("code_of_conduct"))
304
312
 
305
- # Releases — walk pages to get the real total, then read the latest's date from
306
- # the first page (GitHub returns releases sorted newest-first, so page=1 has it).
313
+ # Call 4: Releases — Link header trick for count + per_page=1 returns the
314
+ # newest release (GitHub sorts newest-first) so we get last_release_at too.
307
315
  try:
308
316
  rel_url = f"{BASE}/repos/{full_name}/releases"
309
- first_page, status = _safe_get(rel_url, h, {"per_page": _PER_PAGE, "page": 1})
310
- if status == 200 and isinstance(first_page, list):
311
- stats["total_releases"] = _walk_paginated_count(
312
- rel_url, h, {"per_page": _PER_PAGE}, _MAX_RELEASE_PAGES
313
- )
314
- if first_page:
315
- latest = first_page[0]
316
- stats["last_release_at"] = latest.get("published_at") or latest.get("created_at")
317
+ r = requests.get(rel_url, headers=h, params={"per_page": 1}, timeout=10)
318
+ if r.status_code == 200:
319
+ link = r.headers.get("Link", "")
320
+ match = re.search(r'[?&]page=(\d+)[^>]*>;\s*rel="last"', link)
321
+ stats["total_releases"] = int(match.group(1)) if match else len(r.json())
322
+ data = r.json()
323
+ if isinstance(data, list) and data:
324
+ stats["last_release_at"] = data[0].get("published_at") or data[0].get("created_at")
317
325
  except Exception as exc:
318
- logger.warning(f"Could not fetch releases for {full_name}: {exc}")
326
+ logger.warning("Could not fetch releases for %s: %s", full_name, exc)
319
327
 
320
328
  return stats
321
329
 
@@ -408,6 +416,8 @@ def fetch_star_history(owner: str, repo: str, token: str | None = None) -> list[
408
416
  if status in (403, 429):
409
417
  # GitHub's rate-limit guard returns 403 with a 'rate limit' message in body.
410
418
  raise GitHubRateLimitError("GitHub rate limit reached, try again later")
419
+ if status == 404:
420
+ raise StarHistoryFetchError("user or repo does not exist")
411
421
  if status != 200 or not isinstance(meta, dict):
412
422
  raise StarHistoryFetchError(f"Failed to fetch repo metadata (HTTP {status})")
413
423
  total_stars = int(meta.get("stargazers_count", 0) or 0)