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