Exemplo n.º 1
0
async def html_app(request, app_name):
    app = Repo.select().where(Repo.name == app_name)

    if app.count == 0:
        raise NotFound()

    return {"app": app[0], 'relative_path_to_root': '../../', 'path': request.path}
Exemplo n.º 2
0
async def launch_monthly_job():
    today = date.today().day

    for repo in Repo.select().where(Repo.random_job_day == today):
        task_logger.info(
            f"Launch monthly job for {repo.name} on day {today} of the month ")
        await create_job(repo.name, repo.url)
Exemplo n.º 3
0
    def get(self, username, reponame):
        try:
            repo = (Repo.select().join(User).alias("user").where(
                (User.name == username) & (Repo.name == reponame)).get())
            title = repo.user.name + "/" + repo.name

            timemap = self.get_query_argument("timemap", "false") == "true"
            datetime = self.get_query_argument("datetime", None)
            key = self.get_query_argument("key", None)

            if key and not timemap:
                self.render("repo/memento.html",
                            repo=repo,
                            key=key,
                            datetime=datetime)
            elif key and timemap:
                self.render("repo/history.html", repo=repo, key=key)
            else:
                cs = (CSet.select(fn.distinct(
                    CSet.hkey)).where(CSet.repo == repo).limit(5).alias("cs"))
                samples = (HMap.select(HMap.val).join(
                    cs, on=(HMap.sha == cs.c.hkey_id)))
                self.render("repo/show.html",
                            title=title,
                            repo=repo,
                            samples=list(samples))
        except Repo.DoesNotExist:
            raise HTTPError(404)
Exemplo n.º 4
0
    def get(self, username, reponame):
        try:
            repo = (Repo.select().join(User).alias("user")
                .where((User.name == username) & (Repo.name == reponame))
                .get())
            title = repo.user.name + "/" + repo.name

            timemap = self.get_query_argument("timemap", "false") == "true"
            datetime = self.get_query_argument("datetime", None)
            key = self.get_query_argument("key", None)

            if key and not timemap:
                self.render("repo/memento.html", repo=repo, key=key,
                    datetime=datetime)
            elif key and timemap:
                self.render("repo/history.html", repo=repo, key=key)
            else:
                cs = (CSet.select(fn.distinct(CSet.hkey))
                    .where(CSet.repo == repo).limit(5).alias("cs"))
                samples = (HMap.select(HMap.val)
                    .join(cs, on=(HMap.sha == cs.c.hkey_id)))
                self.render("repo/show.html", title=title, repo=repo,
                    samples=list(samples))
        except Repo.DoesNotExist:
            raise HTTPError(404)
Exemplo n.º 5
0
    def get(self, username, reponame):
        try:
            repo = (Repo.select().join(User).alias("user")
                .where((User.name == username) & (Repo.name == reponame))
                .get())
            title = repo.user.name + "/" + repo.name

            timemap = self.get_query_argument("timemap", "false") == "true"
            datetime = self.get_query_argument("datetime", None)
            key = self.get_query_argument("key", None)
            index = self.get_query_argument("index", "false") == "true"

            if self.get_query_argument("datetime", None):
                datestr = self.get_query_argument("datetime")
                try:
                    ts = date(datestr, QSDATEFMT)
                except ValueError:
                    raise HTTPError(reason="Invalid format of datetime param", status_code=400)
            elif "Accept-Datetime" in self.request.headers:
                datestr = self.request.headers.get("Accept-Datetime")
                ts = date(datestr, RFC1123DATEFMT)
            else:
                ts = now()
            if key and not timemap:
                chain = revision_logic.get_chain_at_ts(repo, key, ts)
                # use ts of cset instead of now(), to make prev work
                if len(chain) != 0:
                    ts = chain[-1].time

                cs_prev = revision_logic.get_cset_prev_before_ts(repo, key, ts)
                cs_next = revision_logic.get_cset_next_after_ts(repo, key, ts)
                if cs_prev:
                    cs_prev_str = self.request.protocol + "://" + self.request.host + self.request.path + "?key=" + key + "&datetime=" + cs_prev.time.strftime(QSDATEFMT)
                else:
                    cs_prev_str = ""
                if cs_next:
                    cs_next_str = self.request.protocol + "://" + self.request.host + self.request.path + "?key=" + key + "&datetime=" + cs_next.time.strftime(QSDATEFMT)
                else:
                    cs_next_str = "" 
                commit_message = revision_logic.get_commit_message(repo, key, ts)

                self.render("repo/memento.html", repo=repo, key=key, datetime=datetime, cs_next_str=cs_next_str, cs_prev_str=cs_prev_str, commit_message=commit_message)
            elif key and timemap:
                self.render("repo/history.html", repo=repo, key=key)
            elif index:
                cs = (CSet.select(fn.distinct(CSet.hkey)).where((CSet.repo == repo) & (CSet.time <= ts)).alias("cs"))
                key_count = (HMap.select(HMap.val).join(cs, on=(HMap.sha == cs.c.hkey_id))).count()

                page = int(self.get_query_argument("page", "1"))

                hm = revision_logic.get_repo_index(repo, ts, page)
                
                self.render("repo/index.html", repo=repo, title=title, key_count=key_count, page_size=revision_logic.INDEX_PAGE_SIZE, hm=hm, current_page=page)
            else:
                hm = list(revision_logic.get_repo_index(repo, ts, 1, 5))
                # cs = (CSet.select(fn.distinct(CSet.hkey)).where(CSet.repo == repo).limit(5).alias("cs"))
                # samples = (HMap.select(HMap.val).join(cs, on=(HMap.sha == cs.c.hkey_id)))
                self.render("repo/show.html", title=title, repo=repo, hm=hm)
        except Repo.DoesNotExist:
            raise HTTPError(reason="Repo not found.", status_code=404)
Exemplo n.º 6
0
def set_random_day_for_monthy_job():
    for repo in Repo.select().where((Repo.random_job_day == None)):
        repo.random_job_day = random.randint(1, 28)
        task_logger.info(
            f"set random day for monthly job of repo '{repo.name}' at '{repo.random_job_day}'"
        )
        repo.save()
Exemplo n.º 7
0
 def get(self, username, reponame):
     try:
         repo = Repo.select().join(User).alias("user").where((User.name == username) & (Repo.name == reponame)).get()
         if not repo.private:
             self._get(repo)
         else:
             self._getAuth(repo)
     except Repo.DoesNotExist:
         raise HTTPError(reason="Repo not found.", status_code=404)
Exemplo n.º 8
0
    def get(self):
        query = tornado.escape.url_unescape(self.get_argument("q", ""))

        if query:
            pattern = "%" + query + "%"
            repos = Repo.select().join(User).alias("user").where(Repo.name ** pattern, Repo.private == False)
            users = User.select().where(User.name ** pattern)
        else:
            repos = []
            users = []

        self.render("search/show.html", query=query, repos=repos, users=users)
Exemplo n.º 9
0
async def launch_monthly_job(type):
    # XXX DRY
    job_command_last_part = ""
    if type == "arm":
        job_command_last_part = " (~ARM~)"
    elif type == "testing-unstable":
        job_command_last_part = [" (testing)", " (unstable)"]

    today = date.today().day

    for repo in Repo.select().where(Repo.random_job_day == today):
        task_logger.info(f"Launch monthly job for {repo.name} on day {today} of the month ")
        await create_job(repo.name, repo.app_list, repo, job_command_last_part)
Exemplo n.º 10
0
    def get(self):
        query = tornado.escape.url_unescape(self.get_argument("q", ""))

        if query:
            pattern = "%" + query + "%"
            repos = (Repo.select().join(User).alias("user").where(
                Repo.name**pattern))
            users = User.select().where(User.name**pattern)
        else:
            repos = []
            users = []

        self.render("search/show.html", query=query, repos=repos, users=users)
Exemplo n.º 11
0
async def ws_app(request, websocket, app_name):
    # XXX I don't check if the app exists because this websocket is supposed to
    # be only loaded from the app page which does this job already
    app = Repo.select().where(Repo.name == app_name)[0]

    subscribe(websocket, f"app-jobs-{app.url}")

    await websocket.send(ujson.dumps({
        "action": "init_jobs",
        "data": Job.select().where(Job.url_or_path ==
                                   app.url).order_by(-Job.id),
    }))

    await websocket.wait_closed()
Exemplo n.º 12
0
async def html_job(request, job_id):
    job = Job.select().where(Job.id == job_id)

    if job.count == 0:
        raise NotFound()

    job = job[0]

    app = Repo.select().where(Repo.url == job.url_or_path)
    app = app[0] if app else None

    return {
        "job": job,
        'app': app,
        'relative_path_to_root': '../',
        'path': request.path
    }
Exemplo n.º 13
0
    def delete(self, username, reponame):
        # Check whether the key exists and if maybe the last change already is
        # a delete, else insert a `CSet.DELETE` entry without any blob data.

        key = self.get_query_argument("key")

        if username != self.current_user.name:
            raise HTTPError(403)

        if not key:
            raise HTTPError(400)

        datestr = self.get_query_argument("datetime", None)
        ts = datestr and date(datestr, QSDATEFMT) or now()

        try:
            repo = (Repo.select(Repo.id).join(
                User).where((User.name == username)
                            & (Repo.name == reponame)).naive().get())
        except Repo.DoesNotExist:
            raise HTTPError(404)

        sha = shasum(key.encode("utf-8"))

        try:
            last = (CSet.select(
                CSet.time,
                CSet.type).where((CSet.repo == repo)
                                 & (CSet.hkey == sha)).order_by(
                                     CSet.time.desc()).limit(1).naive().get())
        except CSet.DoesNotExist:
            # No changeset was found for the given key -
            # the resource does not exist.
            raise HTTPError(400)

        if not ts > last.time:
            # Appended timestamps must be monotonically increasing!
            raise HTTPError(400)

        if last.type == CSet.DELETE:
            # The resource was deleted already, return instantly.
            return self.finish()

        # Insert the new "delete" change.
        CSet.create(repo=repo, hkey=sha, time=ts, type=CSet.DELETE, len=0)
Exemplo n.º 14
0
    def test_commits_create(self, get_commits):
        fixtures = self.data.buffer.read()
        jsn = json.loads(fixtures.decode())

        request = HTTPRequest(self.TEST_REPO['href'])
        response = HTTPResponse(request, client.OK, buffer=io.BytesIO(fixtures))

        future = Future()
        future.set_result(response)
        get_commits.return_value = future

        body = {'href': self.TEST_REPO['href']}
        response = self.fetch(self.get_app().reverse_url('create'), method='POST',
                              body=urlencode(body).encode())

        self.assertIn('всего {}'.format(len(jsn)), response.body.decode())
        self.assertEqual(len(jsn), Commit.select().count())
        self.assertEqual(1, Repo.select().count())
Exemplo n.º 15
0
    def get(self, username):
        try:
            user = User.select().where(User.name == username).get()
        except User.DoesNotExist:
            raise HTTPError(reason="User not found.", status_code=404)
        
        repos = Repo.select().where(Repo.user == user)
        reposit = repos.iterator()

        # TODO: Paginate?

        first = None
        try:
            first = reposit.next()
        except StopIteration:
            # No repos for user
            # No need to raise an error, just return empty list in json
            pass
            

        accept = self.request.headers.get("Accept", "")
        user_url = (self.request.protocol + "://" + self.request.host)

        if "application/json" in accept or "*/*" in accept:
            self.set_header("Content-Type", "application/json")

            self.write('{"username": '******', "repositories": {"list":[')

            m = ('{{"name": "{0}", "uri": "' + user_url +
                 '/'+username+'/{0}"}}')

            if first:
                self.write(m.format(first.name))

            for repo in reposit:
                self.write(', ' + m.format(repo.name))

            self.write(']}')
            self.write('}')
Exemplo n.º 16
0
def get_repo_count():
	return Repo.select().count()
Exemplo n.º 17
0
    def get(self, username, reponame):
        timemap = self.get_query_argument("timemap", "false") == "true"
        index = self.get_query_argument("index", "false") == "true"
        key = self.get_query_argument("key", None)

        if (index and timemap) or (index and key) or (timemap and not key):
            raise HTTPError(400)

        if self.get_query_argument("datetime", None):
            datestr = self.get_query_argument("datetime")
            ts = date(datestr, QSDATEFMT)
        elif "Accept-Datetime" in self.request.headers:
            datestr = self.request.headers.get("Accept-Datetime")
            ts = date(datestr, RFC1123DATEFMT)
        else:
            ts = now()

        try:
            repo = (Repo.select(Repo.id).join(
                User).where((User.name == username)
                            & (Repo.name == reponame)).naive().get())
        except Repo.DoesNotExist:
            raise HTTPError(404)

        if key and not timemap:
            # Recreate the resource for the given key in its latest state -
            # if no `datetime` was provided - or in the state it was in at
            # the time indicated by the passed `datetime` argument.

            self.set_header("Content-Type", "application/n-quads")
            self.set_header("Vary", "accept-datetime")

            sha = shasum(key.encode("utf-8"))

            # Fetch all relevant changes from the last "non-delta" onwards,
            # ordered by time. The returned delta-chain consists of either:
            # a snapshot followed by 0 or more deltas, or
            # a single delete.
            chain = list(
                CSet.select(CSet.time, CSet.type).where((CSet.repo == repo) & (
                    CSet.hkey == sha) & (CSet.time <= ts) & (CSet.time >= SQL(
                        "COALESCE((SELECT time FROM cset "
                        "WHERE repo_id = %s "
                        "AND hkey_id = %s "
                        "AND time <= %s "
                        "AND type != %s "
                        "ORDER BY time DESC "
                        "LIMIT 1), 0)", repo.id, sha, ts, CSet.DELTA))).
                order_by(CSet.time).naive())

            if len(chain) == 0:
                # A resource does not exist for the given key.
                raise HTTPError(404)

            timegate_url = (self.request.protocol + "://" + self.request.host +
                            self.request.path)
            timemap_url = (self.request.protocol + "://" + self.request.host +
                           self.request.uri + "&timemap=true")

            self.set_header(
                "Link", '<%s>; rel="original"'
                ', <%s>; rel="timegate"'
                ', <%s>; rel="timemap"' % (key, timegate_url, timemap_url))

            self.set_header("Memento-Datetime",
                            chain[-1].time.strftime(RFC1123DATEFMT))

            if chain[0].type == CSet.DELETE:
                # The last change was a delete. Return a 404 response with
                # appropriate "Link" and "Memento-Datetime" headers.
                raise HTTPError(404)

            # Load the data required in order to restore the resource state.
            blobs = (Blob.select(Blob.data).where(
                (Blob.repo == repo) & (Blob.hkey == sha)
                & (Blob.time << map(lambda e: e.time, chain))).order_by(
                    Blob.time).naive())

            if len(chain) == 1:
                # Special case, where we can simply return
                # the blob data of the snapshot.
                snap = blobs.first().data
                return self.finish(decompress(snap))

            stmts = set()

            for i, blob in enumerate(blobs.iterator()):
                data = decompress(blob.data)

                if i == 0:
                    # Base snapshot for the delta chain
                    stmts.update(data.splitlines())
                else:
                    for line in data.splitlines():
                        mode, stmt = line[0], line[2:]
                        if mode == "A":
                            stmts.add(stmt)
                        else:
                            stmts.discard(stmt)

            self.write(join(stmts, "\n"))
        elif key and timemap:
            # Generate a timemap containing historic change information
            # for the requested key. The timemap is in the default link-format
            # or as JSON (http://mementoweb.org/guide/timemap-json/).

            sha = shasum(key.encode("utf-8"))

            csets = (CSet.select(
                CSet.time).where((CSet.repo == repo)
                                 & (CSet.hkey == sha)).order_by(
                                     CSet.time.desc()).naive())

            # TODO: Paginate?

            csit = csets.iterator()

            try:
                first = csit.next()
            except StopIteration:
                # Resource for given key does not exist.
                raise HTTPError(404)

            req = self.request
            base = req.protocol + "://" + req.host + req.path

            accept = self.request.headers.get("Accept", "")

            if "application/json" in accept or "*/*" in accept:
                self.set_header("Content-Type", "application/json")

                self.write('{"original_uri": ' + json_encode(key))
                self.write(', "mementos": {"list":[')

                m = ('{{"datetime": "{0}", "uri": "' + base + '?key=' +
                     url_escape(key) + '&datetime={1}"}}')

                self.write(
                    m.format(first.time.isoformat(),
                             first.time.strftime(QSDATEFMT)))

                for cs in csit:
                    self.write(', ' + m.format(cs.time.isoformat(),
                                               cs.time.strftime(QSDATEFMT)))

                self.write(']}')
                self.write('}')
            else:
                m = (',\n'
                     '<' + base + '?key=' + url_escape(key) + '&datetime={0}>'
                     '; rel="memento"'
                     '; datetime="{1}"'
                     '; type="application/n-quads"')

                self.set_header("Content-Type", "application/link-format")

                self.write('<' + key + '>; rel="original"')
                self.write(
                    m.format(first.time.strftime(QSDATEFMT),
                             first.time.strftime(RFC1123DATEFMT)))

                for cs in csit:
                    self.write(
                        m.format(cs.time.strftime(QSDATEFMT),
                                 cs.time.strftime(RFC1123DATEFMT)))
        elif index:
            # Generate an index of all URIs contained in the dataset at the
            # provided point in time or in its current state.

            self.set_header("Vary", "accept-datetime")
            self.set_header("Content-Type", "text/plain")

            page = int(self.get_query_argument("page", "1"))

            # Subquery for selecting max. time per hkey group
            mx = (CSet.select(
                CSet.hkey,
                fn.Max(CSet.time).alias("maxtime")).where(
                    (CSet.repo == repo) & (CSet.time <= ts)).group_by(
                        CSet.hkey).order_by(CSet.hkey).paginate(
                            page, INDEX_PAGE_SIZE).alias("mx"))

            # Query for all the relevant csets (those with max. time values)
            cs = (CSet.select(CSet.hkey, CSet.time).join(
                mx,
                on=((CSet.hkey == mx.c.hkey_id) & (CSet.time == mx.c.maxtime)
                    )).where((CSet.repo == repo)
                             & (CSet.type != CSet.DELETE)).alias("cs"))

            # Join with the hmap table to retrieve the plain key values
            hm = (HMap.select(HMap.val).join(
                cs, on=(HMap.sha == cs.c.hkey_id)).naive())

            for h in hm.iterator():
                self.write(h.val + "\n")
        else:
            raise HTTPError(400)
Exemplo n.º 18
0
async def api_list_app(request):
    query = Repo.select()

    return response.json([model_to_dict(x) for x in query.order_by(Repo.name)])
Exemplo n.º 19
0
async def ws_apps(request, websocket):
    subscribe(websocket, "jobs")
    subscribe(websocket, "apps")

    # I need to do this because peewee strangely f**k up on join and remove the
    # subquery fields which breaks everything
    repos = Repo.raw('''
    SELECT
        "id",
        "name",
        "url",
        "revision",
        "state",
        "random_job_day",
        "job_id",
        "job_name",
        "job_state",
        "created_time",
        "started_time",
        "end_time"
    FROM
        "repo" AS "t1"
    INNER JOIN (
        SELECT
            "t1"."id" as "job_id",
            "t1"."name" as "job_name",
            "t1"."url_or_path",
            "t1"."state" as "job_state",
            "t1"."created_time",
            "t1"."started_time",
            "t1"."end_time"
        FROM
            "job" AS "t1"
        INNER JOIN (
            SELECT
                Max("t2"."id") AS "max_id"
            FROM
                "job" AS "t2"
            GROUP BY
                "t2"."url_or_path"
        )
        AS
            "t3"
        ON
            ("t1"."id" = "t3"."max_id")
    ) AS
        "t5"
    ON
        ("t5"."url_or_path" = "t1"."url")
    ORDER BY
        "name"
    ''')

    repos = [{
        "id":
        x.id,
        "name":
        x.name,
        "url":
        x.url,
        "revision":
        x.revision,
        "state":
        x.state,
        "random_job_day":
        x.random_job_day,
        "job_id":
        x.job_id,
        "job_name":
        x.job_name,
        "job_state":
        x.job_state,
        "created_time":
        datetime.strptime(x.created_time.split(".")[0], '%Y-%m-%d %H:%M:%S')
        if x.created_time else None,
        "started_time":
        datetime.strptime(x.started_time.split(".")[0], '%Y-%m-%d %H:%M:%S')
        if x.started_time else None,
        "end_time":
        datetime.strptime(x.end_time.split(".")[0], '%Y-%m-%d %H:%M:%S')
        if x.end_time else None,
    } for x in repos]

    # add apps without jobs
    selected_repos = {x["id"] for x in repos}
    for repo in Repo.select().where(Repo.id.not_in(selected_repos)):
        repos.append({
            "id": repo.id,
            "name": repo.name,
            "url": repo.url,
            "revision": repo.revision,
            "state": repo.state,
            "random_job_day": repo.random_job_day,
            "job_id": None,
            "job_name": None,
            "job_state": None,
            "created_time": None,
            "started_time": None,
            "end_time": None,
        })

    repos = sorted(repos, key=lambda x: x["name"])

    await websocket.send(ujson.dumps({
        "action": "init_apps",
        "data": repos,
    }))

    await websocket.wait_closed()
Exemplo n.º 20
0
    def put(self, username, reponame):
        # Create a new revision of the resource specified by `key`.

        fmt = self.request.headers.get("Content-Type", "application/n-triples")
        key = self.get_query_argument("key", None)

        if username != self.current_user.name:
            raise HTTPError(403)

        if not key:
            raise HTTPError(400)

        datestr = self.get_query_argument("datetime", None)
        ts = datestr and date(datestr, QSDATEFMT) or now()

        try:
            repo = (Repo.select(Repo.id).join(
                User).where((User.name == username)
                            & (Repo.name == reponame)).naive().get())
        except Repo.DoesNotExist:
            raise HTTPError(404)

        sha = shasum(key.encode("utf-8"))

        chain = list(
            CSet.select(CSet.time, CSet.type, CSet.len).where(
                (CSet.repo == repo) & (CSet.hkey == sha) & (CSet.time >= SQL(
                    "COALESCE((SELECT time FROM cset "
                    "WHERE repo_id = %s "
                    "AND hkey_id = %s "
                    "AND type != %s "
                    "ORDER BY time DESC "
                    "LIMIT 1), 0)", repo.id, sha, CSet.DELTA))).order_by(
                        CSet.time).naive())

        if len(chain) > 0 and not ts > chain[-1].time:
            # Appended timestamps must be monotonically increasing!
            raise HTTPError(400)

        if len(chain) == 0:
            # Mapping for `key` likely does not exist:
            # Store the SHA-to-KEY mapping in HMap,
            # looking out for possible collisions.
            try:
                HMap.create(sha=sha, val=key)
            except IntegrityError:
                val = HMap.select(HMap.val).where(HMap.sha == sha).scalar()
                if val != key:
                    raise HTTPError(500)

        # Parse and normalize into a set of N-Quad lines
        stmts = parse(self.request.body, fmt)
        snapc = compress(join(stmts, "\n"))

        if len(chain) == 0 or chain[0].type == CSet.DELETE:
            # Provide dummy value for `patch` which is never stored.
            # If we get here, we always store a snapshot later on!
            patch = ""
        else:
            # Reconstruct the previous state of the resource
            prev = set()

            blobs = (Blob.select(Blob.data).where(
                (Blob.repo == repo) & (Blob.hkey == sha)
                & (Blob.time << map(lambda e: e.time, chain))).order_by(
                    Blob.time).naive())

            for i, blob in enumerate(blobs.iterator()):
                data = decompress(blob.data)

                if i == 0:
                    # Base snapshot for the delta chain
                    prev.update(data.splitlines())
                else:
                    for line in data.splitlines():
                        mode, stmt = line[0], line[2:]
                        if mode == "A":
                            prev.add(stmt)
                        else:
                            prev.discard(stmt)

            if stmts == prev:
                # No changes, nothing to be done. Bail out.
                return self.finish()

            patch = compress(
                join(
                    map(lambda s: "D " + s, prev - stmts) +
                    map(lambda s: "A " + s, stmts - prev), "\n"))

        # Calculate the accumulated size of the delta chain including
        # the (potential) patch from the previous to the pushed state.
        acclen = reduce(lambda s, e: s + e.len, chain[1:], 0) + len(patch)

        blen = len(chain) > 0 and chain[0].len or 0  # base length

        if (len(chain) == 0 or chain[0].type == CSet.DELETE
                or len(snapc) <= len(patch) or SNAPF * blen <= acclen):
            # Store the current state as a new snapshot
            Blob.create(repo=repo, hkey=sha, time=ts, data=snapc)
            CSet.create(repo=repo,
                        hkey=sha,
                        time=ts,
                        type=CSet.SNAPSHOT,
                        len=len(snapc))
        else:
            # Store a directed delta between the previous and current state
            Blob.create(repo=repo, hkey=sha, time=ts, data=patch)
            CSet.create(repo=repo,
                        hkey=sha,
                        time=ts,
                        type=CSet.DELTA,
                        len=len(patch))
Exemplo n.º 21
0
async def monitor_apps_lists(monitor_git=False,
                             monitor_only_good_quality_apps=False):
    "parse apps lists every hour or so to detect new apps"

    # only support github for now :(
    async def get_master_commit_sha(url):
        command = await asyncio.create_subprocess_shell(
            f"git ls-remote {url} master",
            stdout=asyncio.subprocess.PIPE,
            stderr=asyncio.subprocess.PIPE)
        data = await command.stdout.read()
        commit_sha = data.decode().strip().replace("\t", " ").split(" ")[0]
        return commit_sha

    async with aiohttp.ClientSession() as session:
        task_logger.info(f"Downloading applist...")
        async with session.get(APPS_LIST) as resp:
            data = await resp.json()
            data = data["apps"]

    repos = {x.name: x for x in Repo.select()}

    for app_id, app_data in data.items():
        commit_sha = await get_master_commit_sha(app_data["git"]["url"])

        if app_data["state"] != "working":
            task_logger.debug(
                f"skip {app_id} because state is {app_data['state']}")
            continue

        if monitor_only_good_quality_apps:
            if app_data.get("level") in [None, "?"] or app_data["level"] <= 4:
                task_logger.debug(
                    f"skip {app_id} because app is not good quality")
                continue

        # already know, look to see if there is new commits
        if app_id in repos:
            repo = repos[app_id]

            # but first check if the URL has changed
            if repo.url != app_data["git"]["url"]:
                task_logger.info(
                    f"Application {app_id} has changed of url from {repo.url} to {app_data['git']['url']}"
                )

                repo.url = app_data["git"]["url"]
                repo.save()

                await broadcast(
                    {
                        "action": "update_app",
                        "data": model_to_dict(repo),
                    }, "apps")

                # change the url of all jobs that used to have this URL I
                # guess :/
                # this isn't perfect because that could overwrite added by
                # hand jobs but well...
                for job in Job.select().where(Job.url_or_path == repo.url,
                                              Job.state == "scheduled"):
                    job.url_or_path = repo.url
                    job.save()

                    task_logger.info(
                        f"Updating job {job.name} #{job.id} for {app_id} to {repo.url} since the app has changed of url"
                    )

                    await broadcast(
                        {
                            "action": "update_job",
                            "data": model_to_dict(job),
                        }, [
                            "jobs", f"job-{job.id}",
                            f"app-jobs-{job.url_or_path}"
                        ])

            # we don't want to do anything else
            if not monitor_git:
                continue

            repo_is_updated = False
            if repo.revision != commit_sha:
                task_logger.info(
                    f"Application {app_id} has new commits on github "
                    f"({repo.revision} → {commit_sha}), schedule new job")
                repo.revision = commit_sha
                repo.save()
                repo_is_updated = True

                await create_job(app_id, repo.url)

            repo_state = "working" if app_data[
                "state"] == "working" else "other_than_working"

            if repo.state != repo_state:
                repo.state = repo_state
                repo.save()
                repo_is_updated = True

            if repo.random_job_day is None:
                repo.random_job_day = random.randint(1, 28)
                repo.save()
                repo_is_updated = True

            if repo_is_updated:
                await broadcast(
                    {
                        "action": "update_app",
                        "data": model_to_dict(repo),
                    }, "apps")

        # new app
        elif app_id not in repos:
            task_logger.info(f"New application detected: {app_id} " +
                             (", scheduling a new job" if monitor_git else ""))
            repo = Repo.create(
                name=app_id,
                url=app_data["git"]["url"],
                revision=commit_sha,
                state="working"
                if app_data["state"] == "working" else "other_than_working",
                random_job_day=random.randint(1, 28),
            )

            await broadcast(
                {
                    "action": "new_app",
                    "data": model_to_dict(repo),
                }, "apps")

            if monitor_git:
                await create_job(app_id, repo.url)

        await asyncio.sleep(1)

    # delete apps removed from the list
    unseen_repos = set(repos.keys()) - set(data.keys())

    for repo_name in unseen_repos:
        repo = repos[repo_name]

        # delete scheduled jobs first
        task_logger.info(
            f"Application {repo_name} has been removed from the app list, start by removing its scheduled job if there are any..."
        )
        for job in Job.select().where(Job.url_or_path == repo.url,
                                      Job.state == "scheduled"):
            await api_stop_job(None, job.id)  # not sure this is going to work
            job_id = job.id

            task_logger.info(
                f"Delete scheduled job {job.name} #{job.id} for application {repo_name} because the application is being deleted."
            )

            data = model_to_dict(job)
            job.delete_instance()

            await broadcast({
                "action": "delete_job",
                "data": data,
            }, ["jobs", f"job-{job_id}", f"app-jobs-{job.url_or_path}"])

        task_logger.info(
            f"Delete application {repo_name} because it has been removed from the apps list."
        )
        data = model_to_dict(repo)
        repo.delete_instance()

        await broadcast({
            "action": "delete_app",
            "data": data,
        }, "apps")