contrib/process_linux_firmware.py

   1 #!/usr/bin/python3
   2 import os
   3 import time
   4 import urllib.request
   5 import sqlite3
   6 import feedparser
   7 import argparse
   8 import logging
   9 import email
  10 import email.utils
  11 import smtplib
  12 import subprocess
  13 import sys
  14 from datetime import date
  15 from enum import Enum
  16
  17 URL = "https://lore.kernel.org/linux-firmware/new.atom"
  18
  19
  20 class ContentType(Enum):
  21     REPLY = 1
  22     PATCH = 2
  23     PULL_REQUEST = 3
  24     SPAM = 4
  25
  26
  27 content_types = {
  28     "diff --git": ContentType.PATCH,
  29     "Signed-off-by:": ContentType.PATCH,
  30     "are available in the Git repository at": ContentType.PULL_REQUEST,
  31 }
  32
  33
  34 def classify_content(content):
  35     # load content into the email library
  36     msg = email.message_from_string(content)
  37
  38     # check the subject
  39     subject = msg["Subject"]
  40     if "Re:" in subject:
  41         return ContentType.REPLY
  42     if "PATCH" in subject:
  43         return ContentType.PATCH
  44
  45     for part in msg.walk():
  46         if part.get_content_type() == "text/plain":
  47             try:
  48                 body = part.get_payload(decode=True).decode("utf-8")
  49                 for key in content_types.keys():
  50                     if key in body:
  51                         return content_types[key]
  52                 break
  53             except UnicodeDecodeError as e:
  54                 logging.warning("Failed to decode email: %s, treating as SPAM" % e)
  55                 break
  56     return ContentType.SPAM
  57
  58
  59 def fetch_url(url):
  60     with urllib.request.urlopen(url) as response:
  61         return response.read().decode("utf-8")
  62
  63
  64 def quiet_cmd(cmd):
  65     logging.debug("Running {}".format(cmd))
  66     output = subprocess.check_output(cmd, stderr=subprocess.STDOUT)
  67     logging.debug(output)
  68
  69
  70 def reply_email(content, branch):
  71     if "SMTP_USER" in os.environ:
  72         user = os.environ["SMTP_USER"]
  73     if "SMTP_PASS" in os.environ:
  74         password = os.environ["SMTP_PASS"]
  75     if "SMTP_SERVER" in os.environ:
  76         server = os.environ["SMTP_SERVER"]
  77     if "SMTP_PORT" in os.environ:
  78         port = os.environ["SMTP_PORT"]
  79     if not user or not password or not server or not port:
  80         logging.debug("Missing SMTP configuration, not sending email")
  81         return
  82
  83     reply = email.message.EmailMessage()
  84
  85     orig = email.message_from_string(content)
  86     reply["To"] = ", ".join(
  87         email.utils.formataddr(t)
  88         for t in email.utils.getaddresses(
  89             orig.get_all("from", []) + orig.get_all("to", []) + orig.get_all("cc", [])
  90         )
  91     )
  92
  93     reply["From"] = "linux-firmware@kernel.org"
  94     reply["Subject"] = "Re: {}".format(orig["Subject"])
  95     reply["In-Reply-To"] = orig["Message-Id"]
  96     reply["References"] = orig["Message-Id"]
  97     reply["Thread-Topic"] = orig["Thread-Topic"]
  98     reply["Thread-Index"] = orig["Thread-Index"]
  99
 100     content = (
 101         "Your request has been forwarded by the Linux Firmware Kernel robot.\n"
 102         "Please follow up at https://gitlab.com/kernel-firmware/linux-firmware/-/merge_requests to ensure it gets merged\n"
 103         "Your request is '{}'".format(branch)
 104     )
 105     reply.set_content(content)
 106
 107     mailserver = smtplib.SMTP(server, port)
 108     mailserver.ehlo()
 109     mailserver.starttls()
 110     mailserver.ehlo()
 111     mailserver.login(user, password)
 112     mailserver.sendmail(reply["From"], reply["To"], reply.as_string())
 113     mailserver.quit()
 114
 115
 116 def create_pr(remote, branch):
 117     cmd = [
 118         "git",
 119         "push",
 120         "-u",
 121         remote,
 122         branch,
 123         "-o",
 124         "merge_request.create",
 125         "-o",
 126         "merge_request.remove_source_branch",
 127         "-o",
 128         "merge_request.target=main",
 129         "-o",
 130         "merge_request.title={}".format(branch),
 131     ]
 132     quiet_cmd(cmd)
 133
 134
 135 def refresh_branch():
 136     quiet_cmd(["git", "checkout", "main"])
 137     quiet_cmd(["git", "pull"])
 138
 139
 140 def delete_branch(branch):
 141     quiet_cmd(["git", "checkout", "main"])
 142     quiet_cmd(["git", "branch", "-D", branch])
 143
 144
 145 def process_pr(url, num, remote):
 146     branch = "robot/pr-{}-{}".format(num, int(time.time()))
 147     cmd = ["b4", "pr", "-b", branch, url]
 148     try:
 149         quiet_cmd(cmd)
 150     except subprocess.CalledProcessError:
 151         logging.warning("Failed to apply PR")
 152         return None
 153
 154     # determine if it worked (we can't tell unfortunately by return code)
 155     cmd = ["git", "branch", "--list", branch]
 156     logging.debug("Running {}".format(cmd))
 157     result = subprocess.check_output(cmd)
 158
 159     if result:
 160         logging.info("Forwarding PR for {}".format(branch))
 161         if remote:
 162             create_pr(remote, branch)
 163         delete_branch(branch)
 164         return branch
 165     return None
 166
 167
 168 def process_patch(mbox, num, remote):
 169     # create a new branch for the patch
 170     branch = "robot/patch-{}-{}".format(num, int(time.time()))
 171     cmd = ["git", "checkout", "-b", branch]
 172     quiet_cmd(cmd)
 173
 174     # apply the patch
 175     cmd = ["git", "am"]
 176     logging.debug("Running {}".format(cmd))
 177     p = subprocess.Popen(
 178         cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE
 179     )
 180     stdout, stderr = p.communicate(mbox.encode("utf-8"))
 181     for line in stdout.splitlines():
 182         logging.debug(line.decode("utf-8"))
 183     for line in stderr.splitlines():
 184         logging.debug(line.decode("utf-8"))
 185     if p.returncode != 0:
 186         quiet_cmd(["git", "am", "--abort"])
 187     else:
 188         logging.info("Opening PR for {}".format(branch))
 189         if remote:
 190             create_pr(remote, branch)
 191
 192     delete_branch(branch)
 193     if p.returncode == 0:
 194         return branch
 195     return None
 196
 197
 198 def update_database(conn, url):
 199     c = conn.cursor()
 200
 201     c.execute(
 202         """CREATE TABLE IF NOT EXISTS firmware (url text, processed integer default 0, spam integer default 0)"""
 203     )
 204
 205     # local file
 206     if os.path.exists(url):
 207         with open(url, "r") as f:
 208             atom = f.read()
 209     # remote file
 210     else:
 211         logging.info("Fetching {}".format(url))
 212         atom = fetch_url(url)
 213
 214     # Parse the atom and extract the URLs
 215     feed = feedparser.parse(atom)
 216
 217     # Insert the URLs into the database (oldest first)
 218     feed["entries"].reverse()
 219     for entry in feed["entries"]:
 220         c.execute("SELECT url FROM firmware WHERE url = ?", (entry.link,))
 221         if c.fetchone():
 222             continue
 223         c.execute("INSERT INTO firmware VALUES (?, ?, ?)", (entry.link, 0, 0))
 224
 225     # Commit the changes and close the connection
 226     conn.commit()
 227
 228
 229 def process_database(conn, remote):
 230     c = conn.cursor()
 231
 232     # get all unprocessed urls that aren't spam
 233     c.execute("SELECT url FROM firmware WHERE processed = 0 AND spam = 0")
 234     num = 0
 235     msg = ""
 236
 237     rows = c.fetchall()
 238
 239     if not rows:
 240         logging.info("No new entries")
 241         return
 242
 243     refresh_branch()
 244
 245     # loop over all unprocessed urls
 246     for row in rows:
 247
 248         branch = None
 249         msg = "Processing ({}%)".format(round(num / len(rows) * 100))
 250         print(msg, end="\r", flush=True)
 251
 252         url = "{}raw".format(row[0])
 253         logging.debug("Processing {}".format(url))
 254         mbox = fetch_url(url)
 255         classification = classify_content(mbox)
 256
 257         if classification == ContentType.PATCH:
 258             logging.debug("Processing patch ({})".format(row[0]))
 259             branch = process_patch(mbox, num, remote)
 260
 261         if classification == ContentType.PULL_REQUEST:
 262             logging.debug("Processing PR ({})".format(row[0]))
 263             branch = process_pr(row[0], num, remote)
 264
 265         if classification == ContentType.SPAM:
 266             logging.debug("Marking spam ({})".format(row[0]))
 267             c.execute("UPDATE firmware SET spam = 1 WHERE url = ?", (row[0],))
 268
 269         if classification == ContentType.REPLY:
 270             logging.debug("Ignoring reply ({})".format(row[0]))
 271
 272         c.execute("UPDATE firmware SET processed = 1 WHERE url = ?", (row[0],))
 273         num += 1
 274         print(" " * len(msg), end="\r", flush=True)
 275
 276         # commit changes
 277         conn.commit()
 278
 279         # send any emails
 280         if branch:
 281             reply_email(mbox, branch)
 282
 283     logging.info("Finished processing {} new entries".format(len(rows)))
 284
 285
 286 if __name__ == "__main__":
 287     parser = argparse.ArgumentParser(description="Process linux-firmware mailing list")
 288     parser.add_argument("--url", default=URL, help="URL to get ATOM feed from")
 289     parser.add_argument(
 290         "--database",
 291         default=os.path.join("contrib", "linux_firmware.db"),
 292         help="sqlite database to store entries in",
 293     )
 294     parser.add_argument("--dry", action="store_true", help="Don't open pull requests")
 295     parser.add_argument(
 296         "--debug", action="store_true", help="Enable debug logging to console"
 297     )
 298     parser.add_argument("--remote", default="origin", help="Remote to push to")
 299     parser.add_argument(
 300         "--refresh-cycle", default=0, help="How frequently to run (in minutes)"
 301     )
 302     args = parser.parse_args()
 303
 304     if not os.path.exists("WHENCE"):
 305         logging.critical(
 306             "Please run this script from the root of the linux-firmware repository"
 307         )
 308         sys.exit(1)
 309
 310     log = os.path.join(
 311         "contrib",
 312         "{prefix}-{date}.{suffix}".format(
 313             prefix="linux_firmware", suffix="txt", date=date.today()
 314         ),
 315     )
 316     logging.basicConfig(
 317         format="%(asctime)s %(levelname)s:\t%(message)s",
 318         filename=log,
 319         filemode="w",
 320         level=logging.DEBUG,
 321     )
 322
 323     # set a format which is simpler for console use
 324     console = logging.StreamHandler()
 325     if args.debug:
 326         console.setLevel(logging.DEBUG)
 327     else:
 328         console.setLevel(logging.INFO)
 329     formatter = logging.Formatter("%(asctime)s : %(levelname)s : %(message)s")
 330     console.setFormatter(formatter)
 331     logging.getLogger("").addHandler(console)
 332
 333     while True:
 334         conn = sqlite3.connect(args.database)
 335         # update the database
 336         update_database(conn, args.url)
 337
 338         if args.dry:
 339             remote = ""
 340         else:
 341             remote = args.remote
 342
 343         # process the database
 344         process_database(conn, remote)
 345
 346         conn.close()
 347
 348         if args.refresh_cycle:
 349             logging.info("Sleeping for {} minutes".format(args.refresh_cycle))
 350             time.sleep(int(args.refresh_cycle) * 60)
 351         else:
 352             break