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