xref: /openbmc/openbmc/poky/meta/lib/oe/distro_check.py (revision 92b42cb3)
1#
2# Copyright OpenEmbedded Contributors
3#
4# SPDX-License-Identifier: GPL-2.0-only
5#
6
7def create_socket(url, d):
8    import urllib
9    from bb.utils import export_proxies
10
11    export_proxies(d)
12    return urllib.request.urlopen(url)
13
14def get_links_from_url(url, d):
15    "Return all the href links found on the web location"
16
17    from bs4 import BeautifulSoup, SoupStrainer
18
19    soup = BeautifulSoup(create_socket(url,d), "html.parser", parse_only=SoupStrainer("a"))
20    hyperlinks = []
21    for line in soup.find_all('a', href=True):
22        hyperlinks.append(line['href'].strip('/'))
23    return hyperlinks
24
25def find_latest_numeric_release(url, d):
26    "Find the latest listed numeric release on the given url"
27    max=0
28    maxstr=""
29    for link in get_links_from_url(url, d):
30        try:
31            # TODO use bb.utils.vercmp_string_op()
32            release = float(link)
33        except:
34            release = 0
35        if release > max:
36            max = release
37            maxstr = link
38    return maxstr
39
40def is_src_rpm(name):
41    "Check if the link is pointing to a src.rpm file"
42    return name.endswith(".src.rpm")
43
44def package_name_from_srpm(srpm):
45    "Strip out the package name from the src.rpm filename"
46
47    # ca-certificates-2016.2.7-1.0.fc24.src.rpm
48    # ^name           ^ver     ^release^removed
49    (name, version, release) = srpm.replace(".src.rpm", "").rsplit("-", 2)
50    return name
51
52def get_source_package_list_from_url(url, section, d):
53    "Return a sectioned list of package names from a URL list"
54
55    bb.note("Reading %s: %s" % (url, section))
56    links = get_links_from_url(url, d)
57    srpms = filter(is_src_rpm, links)
58    names_list = map(package_name_from_srpm, srpms)
59
60    new_pkgs = set()
61    for pkgs in names_list:
62       new_pkgs.add(pkgs + ":" + section)
63    return new_pkgs
64
65def get_source_package_list_from_url_by_letter(url, section, d):
66    import string
67    from urllib.error import HTTPError
68    packages = set()
69    for letter in (string.ascii_lowercase + string.digits):
70        # Not all subfolders may exist, so silently handle 404
71        try:
72            packages |= get_source_package_list_from_url(url + "/" + letter, section, d)
73        except HTTPError as e:
74            if e.code != 404: raise
75    return packages
76
77def get_latest_released_fedora_source_package_list(d):
78    "Returns list of all the name os packages in the latest fedora distro"
79    latest = find_latest_numeric_release("http://archive.fedoraproject.org/pub/fedora/linux/releases/", d)
80    package_names = get_source_package_list_from_url_by_letter("http://archive.fedoraproject.org/pub/fedora/linux/releases/%s/Everything/source/tree/Packages/" % latest, "main", d)
81    package_names |= get_source_package_list_from_url_by_letter("http://archive.fedoraproject.org/pub/fedora/linux/updates/%s/SRPMS/" % latest, "updates", d)
82    return latest, package_names
83
84def get_latest_released_opensuse_source_package_list(d):
85    "Returns list of all the name os packages in the latest opensuse distro"
86    latest = find_latest_numeric_release("http://download.opensuse.org/source/distribution/leap", d)
87
88    package_names = get_source_package_list_from_url("http://download.opensuse.org/source/distribution/leap/%s/repo/oss/suse/src/" % latest, "main", d)
89    package_names |= get_source_package_list_from_url("http://download.opensuse.org/update/leap/%s/oss/src/" % latest, "updates", d)
90    return latest, package_names
91
92def get_latest_released_clear_source_package_list(d):
93    latest = find_latest_numeric_release("https://download.clearlinux.org/releases/", d)
94    package_names = get_source_package_list_from_url("https://download.clearlinux.org/releases/%s/clear/source/SRPMS/" % latest, "main", d)
95    return latest, package_names
96
97def find_latest_debian_release(url, d):
98    "Find the latest listed debian release on the given url"
99
100    releases = [link.replace("Debian", "")
101                for link in get_links_from_url(url, d)
102                if link.startswith("Debian")]
103    releases.sort()
104    try:
105        return releases[-1]
106    except:
107        return "_NotFound_"
108
109def get_debian_style_source_package_list(url, section, d):
110    "Return the list of package-names stored in the debian style Sources.gz file"
111    import gzip
112
113    package_names = set()
114    for line in gzip.open(create_socket(url, d), mode="rt"):
115        if line.startswith("Package:"):
116            pkg = line.split(":", 1)[1].strip()
117            package_names.add(pkg + ":" + section)
118    return package_names
119
120def get_latest_released_debian_source_package_list(d):
121    "Returns list of all the name of packages in the latest debian distro"
122    latest = find_latest_debian_release("http://ftp.debian.org/debian/dists/", d)
123    url = "http://ftp.debian.org/debian/dists/stable/main/source/Sources.gz"
124    package_names = get_debian_style_source_package_list(url, "main", d)
125    url = "http://ftp.debian.org/debian/dists/stable-proposed-updates/main/source/Sources.gz"
126    package_names |= get_debian_style_source_package_list(url, "updates", d)
127    return latest, package_names
128
129def find_latest_ubuntu_release(url, d):
130    """
131    Find the latest listed Ubuntu release on the given ubuntu/dists/ URL.
132
133    To avoid matching development releases look for distributions that have
134    updates, so the resulting distro could be any supported release.
135    """
136    url += "?C=M;O=D" # Descending Sort by Last Modified
137    for link in get_links_from_url(url, d):
138        if "-updates" in link:
139            distro = link.replace("-updates", "")
140            return distro
141    return "_NotFound_"
142
143def get_latest_released_ubuntu_source_package_list(d):
144    "Returns list of all the name os packages in the latest ubuntu distro"
145    latest = find_latest_ubuntu_release("http://archive.ubuntu.com/ubuntu/dists/", d)
146    url = "http://archive.ubuntu.com/ubuntu/dists/%s/main/source/Sources.gz" % latest
147    package_names = get_debian_style_source_package_list(url, "main", d)
148    url = "http://archive.ubuntu.com/ubuntu/dists/%s-updates/main/source/Sources.gz" % latest
149    package_names |= get_debian_style_source_package_list(url, "updates", d)
150    return latest, package_names
151
152def create_distro_packages_list(distro_check_dir, d):
153    import shutil
154
155    pkglst_dir = os.path.join(distro_check_dir, "package_lists")
156    bb.utils.remove(pkglst_dir, True)
157    bb.utils.mkdirhier(pkglst_dir)
158
159    per_distro_functions = (
160                            ("Debian", get_latest_released_debian_source_package_list),
161                            ("Ubuntu", get_latest_released_ubuntu_source_package_list),
162                            ("Fedora", get_latest_released_fedora_source_package_list),
163                            ("openSUSE", get_latest_released_opensuse_source_package_list),
164                            ("Clear", get_latest_released_clear_source_package_list),
165                           )
166
167    for name, fetcher_func in per_distro_functions:
168        try:
169            release, package_list = fetcher_func(d)
170        except Exception as e:
171            bb.warn("Cannot fetch packages for %s: %s" % (name, e))
172        bb.note("Distro: %s, Latest Release: %s, # src packages: %d" % (name, release, len(package_list)))
173        if len(package_list) == 0:
174            bb.error("Didn't fetch any packages for %s %s" % (name, release))
175
176        package_list_file = os.path.join(pkglst_dir, name + "-" + release)
177        with open(package_list_file, 'w') as f:
178            for pkg in sorted(package_list):
179                f.write(pkg + "\n")
180
181def update_distro_data(distro_check_dir, datetime, d):
182    """
183    If distro packages list data is old then rebuild it.
184    The operations has to be protected by a lock so that
185    only one thread performes it at a time.
186    """
187    if not os.path.isdir (distro_check_dir):
188        try:
189            bb.note ("Making new directory: %s" % distro_check_dir)
190            os.makedirs (distro_check_dir)
191        except OSError:
192            raise Exception('Unable to create directory %s' % (distro_check_dir))
193
194
195    datetime_file = os.path.join(distro_check_dir, "build_datetime")
196    saved_datetime = "_invalid_"
197    import fcntl
198    try:
199        if not os.path.exists(datetime_file):
200            open(datetime_file, 'w+').close() # touch the file so that the next open won't fail
201
202        f = open(datetime_file, "r+")
203        fcntl.lockf(f, fcntl.LOCK_EX)
204        saved_datetime = f.read()
205        if saved_datetime[0:8] != datetime[0:8]:
206            bb.note("The build datetime did not match: saved:%s current:%s" % (saved_datetime, datetime))
207            bb.note("Regenerating distro package lists")
208            create_distro_packages_list(distro_check_dir, d)
209            f.seek(0)
210            f.write(datetime)
211
212    except OSError as e:
213        raise Exception('Unable to open timestamp: %s' % e)
214    finally:
215        fcntl.lockf(f, fcntl.LOCK_UN)
216        f.close()
217
218def compare_in_distro_packages_list(distro_check_dir, d):
219    if not os.path.isdir(distro_check_dir):
220        raise Exception("compare_in_distro_packages_list: invalid distro_check_dir passed")
221
222    localdata = bb.data.createCopy(d)
223    pkglst_dir = os.path.join(distro_check_dir, "package_lists")
224    matching_distros = []
225    pn = recipe_name = d.getVar('PN')
226    bb.note("Checking: %s" % pn)
227
228    if pn.find("-native") != -1:
229        pnstripped = pn.split("-native")
230        localdata.setVar('OVERRIDES', "pn-" + pnstripped[0] + ":" + d.getVar('OVERRIDES'))
231        recipe_name = pnstripped[0]
232
233    if pn.startswith("nativesdk-"):
234        pnstripped = pn.split("nativesdk-")
235        localdata.setVar('OVERRIDES', "pn-" + pnstripped[1] + ":" + d.getVar('OVERRIDES'))
236        recipe_name = pnstripped[1]
237
238    if pn.find("-cross") != -1:
239        pnstripped = pn.split("-cross")
240        localdata.setVar('OVERRIDES', "pn-" + pnstripped[0] + ":" + d.getVar('OVERRIDES'))
241        recipe_name = pnstripped[0]
242
243    if pn.find("-initial") != -1:
244        pnstripped = pn.split("-initial")
245        localdata.setVar('OVERRIDES', "pn-" + pnstripped[0] + ":" + d.getVar('OVERRIDES'))
246        recipe_name = pnstripped[0]
247
248    bb.note("Recipe: %s" % recipe_name)
249
250    distro_exceptions = dict({"OE-Core":'OE-Core', "OpenedHand":'OpenedHand', "Intel":'Intel', "Upstream":'Upstream', "Windriver":'Windriver', "OSPDT":'OSPDT Approved', "Poky":'poky'})
251    tmp = localdata.getVar('DISTRO_PN_ALIAS') or ""
252    for str in tmp.split():
253        if str and str.find("=") == -1 and distro_exceptions[str]:
254            matching_distros.append(str)
255
256    distro_pn_aliases = {}
257    for str in tmp.split():
258        if "=" in str:
259            (dist, pn_alias) = str.split('=')
260            distro_pn_aliases[dist.strip().lower()] = pn_alias.strip()
261
262    for file in os.listdir(pkglst_dir):
263        (distro, distro_release) = file.split("-")
264        f = open(os.path.join(pkglst_dir, file), "r")
265        for line in f:
266            (pkg, section) = line.split(":")
267            if distro.lower() in distro_pn_aliases:
268                pn = distro_pn_aliases[distro.lower()]
269            else:
270                pn = recipe_name
271            if pn == pkg:
272                matching_distros.append(distro + "-" + section[:-1]) # strip the \n at the end
273                f.close()
274                break
275        f.close()
276
277    for item in tmp.split():
278        matching_distros.append(item)
279    bb.note("Matching: %s" % matching_distros)
280    return matching_distros
281
282def create_log_file(d, logname):
283    logpath = d.getVar('LOG_DIR')
284    bb.utils.mkdirhier(logpath)
285    logfn, logsuffix = os.path.splitext(logname)
286    logfile = os.path.join(logpath, "%s.%s%s" % (logfn, d.getVar('DATETIME'), logsuffix))
287    if not os.path.exists(logfile):
288            slogfile = os.path.join(logpath, logname)
289            if os.path.exists(slogfile):
290                    os.remove(slogfile)
291            open(logfile, 'w+').close()
292            os.symlink(logfile, slogfile)
293            d.setVar('LOG_FILE', logfile)
294    return logfile
295
296
297def save_distro_check_result(result, datetime, result_file, d):
298    pn = d.getVar('PN')
299    logdir = d.getVar('LOG_DIR')
300    if not logdir:
301        bb.error("LOG_DIR variable is not defined, can't write the distro_check results")
302        return
303    bb.utils.mkdirhier(logdir)
304
305    line = pn
306    for i in result:
307        line = line + "," + i
308    f = open(result_file, "a")
309    import fcntl
310    fcntl.lockf(f, fcntl.LOCK_EX)
311    f.seek(0, os.SEEK_END) # seek to the end of file
312    f.write(line + "\n")
313    fcntl.lockf(f, fcntl.LOCK_UN)
314    f.close()
315