misc/collect.py

   1 #!/usr/bin/env python3
   2 """
   3 Tool to create overview.json files and update the config.js.
   4 """
   5
   6 from pathlib import Path
   7 import urllib.request
   8 import tempfile
   9 import argparse
  10 import json
  11 import glob
  12 import sys
  13 import os
  14 import re
  15
  16 SUPPORTED_METADATA_VERSION = 1
  17
  18 assert sys.version_info >= (3, 5), "Python version too old. Python >=3.5.0 needed."
  19
  20
  21 # accepts {<file-path>: <file-content>}
  22 def merge_profiles(profiles, download_url):
  23     # json output data
  24     output = {}
  25
  26     def get_title(title):
  27         if "title" in title:
  28             return title["title"]
  29         else:
  30             return "{} {} {}".format(
  31                 title.get("vendor", ""), title["model"], title.get("variant", "")
  32             ).strip()
  33
  34     def add_profile(path, id, target, profile, code=None):
  35         images = []
  36         for image in profile["images"]:
  37             images.append({"name": image["name"], "type": image["type"]})
  38
  39         if target is None:
  40             target = profile["target"]
  41
  42         for entry in profile["titles"]:
  43             title = get_title(entry)
  44
  45             if len(title) == 0:
  46                 sys.stderr.write(
  47                     "Empty title. Skip title for {} in {}\n".format(id, path)
  48                 )
  49                 continue
  50
  51             # device is a duplicate, try to differentiate by target
  52             if title in output["models"]:
  53                 title = "{} ({})".format(title, target)
  54
  55             output["models"][title] = {"id": id, "target": target, "images": images}
  56
  57             if code is not None:
  58                 output["models"][title]["code"] = code
  59
  60     for path, content in profiles.items():
  61         obj = json.loads(content)
  62
  63         if obj["metadata_version"] != SUPPORTED_METADATA_VERSION:
  64             sys.stderr.write(
  65                 "{} has unsupported metadata version: {} => skip\n".format(
  66                     path, obj["metadata_version"]
  67                 )
  68             )
  69             continue
  70
  71         code = obj.get("version_code", obj.get("version_commit"))
  72
  73         if "version_code" not in output:
  74             output = {"version_code": code, "download_url": download_url, "models": {}}
  75
  76         # if we have mixed codes/commits, store in device object
  77         if output["version_code"] == code:
  78             code = None
  79
  80         try:
  81             if "profiles" in obj:
  82                 for id in obj["profiles"]:
  83                     add_profile(path, id, obj.get("target"), obj["profiles"][id], code)
  84             else:
  85                 add_profile(path, obj["id"], obj["target"], obj, code)
  86         except json.decoder.JSONDecodeError as e:
  87             sys.stderr.write("Skip {}\n   {}\n".format(path, e))
  88         except KeyError as e:
  89             sys.stderr.write("Abort on {}\n   Missing key {}\n".format(path, e))
  90             exit(1)
  91
  92     return output
  93
  94
  95 def update_config(config_path, versions):
  96     content = ""
  97     with open(str(config_path), "r", encoding="utf-8") as file:
  98         content = file.read()
  99
 100     content = re.sub("versions:[\\s]*{[^}]*}", "versions: {}".format(versions), content)
 101     with open(str(config_path), "w+") as file:
 102         file.write(content)
 103
 104
 105 """
 106 Scrape profiles.json using links like https://downloads.openwrt.org/releases/19.07.3/targets/?json
 107 Merge into overview.json files.
 108 Update config.json.
 109 """
 110
 111
 112 def scrape(args):
 113     url = args.domain
 114     www_path = args.www_path
 115     config_path = "{}/config.js".format(www_path)
 116     data_path = "{}/data".format(www_path)
 117     versions = {}
 118
 119     def handle_release(target):
 120         profiles = {}
 121         with urllib.request.urlopen("{}/?json".format(target)) as file:
 122             array = json.loads(file.read().decode("utf-8"))
 123             for profile in filter(lambda x: x.endswith("/profiles.json"), array):
 124                 with urllib.request.urlopen("{}/{}".format(target, profile)) as file:
 125                     profiles["{}/{}".format(target, profile)] = file.read().decode(
 126                         "utf-8"
 127                     )
 128         return profiles
 129
 130     if not os.path.isfile(config_path):
 131         print("file not found: {}".format(config_path))
 132         exit(1)
 133
 134     # fetch release URLs
 135     with urllib.request.urlopen(url) as infile:
 136         for path in re.findall(r"href=[\"']?([^'\" >]+)", str(infile.read())):
 137             if not path.startswith("/") and path.endswith("targets/"):
 138                 release = path.strip("/").split("/")[-2]
 139                 download_url = "{}/{}/{{target}}".format(url, path)
 140
 141                 profiles = handle_release("{}/{}".format(url, path))
 142                 output = merge_profiles(profiles, download_url)
 143                 if len(output) > 0:
 144                     os.makedirs("{}/{}".format(data_path, release), exist_ok=True)
 145                     # write overview.json
 146                     with open(
 147                         "{}/{}/overview.json".format(data_path, release), "w"
 148                     ) as outfile:
 149                         if args.formatted:
 150                             json.dump(output, outfile, indent="  ", sort_keys=True)
 151                         else:
 152                             json.dump(output, outfile, sort_keys=True)
 153
 154                     versions[release] = "data/{}/overview.json".format(release)
 155
 156     update_config(config_path, versions)
 157
 158
 159 """
 160 Scrape profiles.json using wget (slower but more generic).
 161 Merge into overview.json files.
 162 Update config.json.
 163 """
 164
 165
 166 def scrape_wget(args):
 167     url = args.domain
 168     www_path = args.www_path
 169     config_path = "{}/config.js".format(www_path)
 170     data_path = "{}/data".format(www_path)
 171     versions = {}
 172
 173     with tempfile.TemporaryDirectory() as tmp_dir:
 174         # download all profiles.json files
 175         os.system(
 176             "wget -c -r -P {} -A 'profiles.json' --reject-regex 'kmods|packages' --no-parent {}".format(
 177                 tmp_dir, url
 178             )
 179         )
 180
 181         # delete empty folders
 182         os.system("find {}/* -type d -empty -delete".format(tmp_dir))
 183
 184         # create overview.json files
 185         for path in glob.glob("{}/*/snapshots".format(tmp_dir)) + glob.glob(
 186             "{}/*/releases/*".format(tmp_dir)
 187         ):
 188             release = os.path.basename(path)
 189             base = path[len(tmp_dir) + 1 :]
 190
 191             profiles = {}
 192             for ppath in Path(path).rglob("profiles.json"):
 193                 with open(str(ppath), "r", encoding="utf-8") as file:
 194                     profiles[ppath] = file.read()
 195
 196             if len(profiles) == 0:
 197                 continue
 198
 199             versions[release] = "data/{}/overview.json".format(release)
 200
 201             output = merge_profiles(
 202                 profiles, "https://{}/targets/{{target}}".format(base)
 203             )
 204             os.makedirs("{}/{}".format(data_path, release), exist_ok=True)
 205
 206             # write overview.json
 207             with open("{}/{}/overview.json".format(data_path, release), "w") as outfile:
 208                 if args.formatted:
 209                     json.dump(output, outfile, indent="  ", sort_keys=True)
 210                 else:
 211                     json.dump(output, outfile, sort_keys=True)
 212
 213         update_config(config_path, versions)
 214
 215
 216 """
 217 Find and merge json files for a single release.
 218 """
 219
 220
 221 def merge(args):
 222     input_paths = args.input_path
 223     # OpenWrt JSON device files
 224     profiles = {}
 225
 226     def add_path(path):
 227         with open(str(path), "r", encoding="utf-8") as file:
 228             profiles[path] = file.read()
 229
 230     for path in input_paths:
 231         if os.path.isdir(path):
 232             for filepath in Path(path).rglob("*.json"):
 233                 add_path(filepath)
 234         else:
 235             if not path.endswith(".json"):
 236                 sys.stderr.write("Folder does not exists: {}\n".format(path))
 237                 exit(1)
 238             add_path(path)
 239
 240     output = merge_profiles(profiles, args.download_url)
 241
 242     if args.formatted:
 243         json.dump(output, sys.stdout, indent="  ", sort_keys=True)
 244     else:
 245         json.dump(output, sys.stdout, sort_keys=True)
 246
 247
 248 """
 249 Scan local directory for releases with profiles.json.
 250 Merge into overview.json files.
 251 Update config.json.
 252 """
 253
 254
 255 def scan(args):
 256     # firmware selector config
 257     config_path = "{}/config.js".format(args.www_path)
 258     # the overview.json files are placed here
 259     data_path = "{}/data".format(args.www_path)
 260     versions = {}
 261
 262     # args.images_path => args.releases_path
 263     releases = {}
 264     for path in Path(args.images_path).rglob("profiles.json"):
 265         with open(str(path), "r", encoding="utf-8") as file:
 266             content = file.read()
 267             obj = json.loads(content)
 268             release = obj["version_number"]
 269             releases.setdefault(release, {})[path] = content
 270
 271     for release, profiles in releases.items():
 272         output = merge_profiles(profiles, args.download_url)
 273
 274         versions[release] = "data/{}/overview.json".format(release)
 275         os.makedirs("{}/{}".format(data_path, release), exist_ok=True)
 276
 277         # write overview.json
 278         with open("{}/{}/overview.json".format(data_path, release), "w") as outfile:
 279             if args.formatted:
 280                 json.dump(output, outfile, indent="  ", sort_keys=True)
 281             else:
 282                 json.dump(output, outfile, sort_keys=True)
 283
 284     update_config(config_path, versions)
 285
 286
 287 def main():
 288     parser = argparse.ArgumentParser()
 289     parser.add_argument(
 290         "--formatted", action="store_true", help="Output formatted JSON data."
 291     )
 292     subparsers = parser.add_subparsers(dest="action")
 293     subparsers.required = True
 294
 295     parser_merge = subparsers.add_parser(
 296         "merge", help="Search for profiles.json files and output an overview.json."
 297     )
 298     parser_merge.add_argument(
 299         "input_path",
 300         nargs="+",
 301         help="Input folder that is traversed for OpenWrt JSON device files.",
 302     )
 303     parser_merge.add_argument(
 304         "--download-url",
 305         action="store",
 306         default="",
 307         help="Link to get the image from. May contain {target} (replaced by e.g. ath79/generic), {version} (replace by the version key from config.js) and {commit} (git commit in hex notation).",
 308     )
 309
 310     parser_scrape = subparsers.add_parser("scrape", help="Scrape webpage for releases.")
 311     parser_scrape.add_argument(
 312         "domain", help="Domain to scrape. E.g. https://downloads.openwrt.org"
 313     )
 314     parser_scrape.add_argument("www_path", help="Path the config.js file is in.")
 315     parser_scrape.add_argument(
 316         "--use-wget", action="store_true", help="Use wget to scrape the site."
 317     )
 318
 319     parser_scan = subparsers.add_parser("scan", help="Scan directory for releases.")
 320     parser_scan.add_argument(
 321         "download_url", help="Download for images. E.g. https://downloads.openwrt.org"
 322     )
 323     parser_scan.add_argument("images_path", help="Directory to scan for releases.")
 324     parser_scan.add_argument("www_path", help="Path the config.js file is in.")
 325
 326     args = parser.parse_args()
 327
 328     if args.action == "merge":
 329         merge(args)
 330
 331     if args.action == "scan":
 332         scan(args)
 333
 334     if args.action == "scrape":
 335         if args.use_wget:
 336             scrape_wget(args)
 337         else:
 338             scrape(args)
 339
 340
 341 if __name__ == "__main__":
 342     main()