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