xref: /trunk/test/odf-validate/pkgcheck.py (revision 0e16a31d78a652c56c5c0fc79c752f48b9048df9)
1*0e16a31dSJim Jagielski#**************************************************************
2*0e16a31dSJim Jagielski#
3*0e16a31dSJim Jagielski#  Licensed to the Apache Software Foundation (ASF) under one
4*0e16a31dSJim Jagielski#  or more contributor license agreements.  See the NOTICE file
5*0e16a31dSJim Jagielski#  distributed with this work for additional information
6*0e16a31dSJim Jagielski#  regarding copyright ownership.  The ASF licenses this file
7*0e16a31dSJim Jagielski#  to you under the Apache License, Version 2.0 (the
8*0e16a31dSJim Jagielski#  "License"); you may not use this file except in compliance
9*0e16a31dSJim Jagielski#  with the License.  You may obtain a copy of the License at
10*0e16a31dSJim Jagielski#
11*0e16a31dSJim Jagielski#    http://www.apache.org/licenses/LICENSE-2.0
12*0e16a31dSJim Jagielski#
13*0e16a31dSJim Jagielski#  Unless required by applicable law or agreed to in writing,
14*0e16a31dSJim Jagielski#  software distributed under the License is distributed on an
15*0e16a31dSJim Jagielski#  "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
16*0e16a31dSJim Jagielski#  KIND, either express or implied.  See the License for the
17*0e16a31dSJim Jagielski#  specific language governing permissions and limitations
18*0e16a31dSJim Jagielski#  under the License.
19*0e16a31dSJim Jagielski#
20*0e16a31dSJim Jagielski#**************************************************************
21*0e16a31dSJim Jagielski
22*0e16a31dSJim Jagielski# usage: pkgcheck.py <outdir> <pkg.json>
23*0e16a31dSJim Jagielski# ODF 1.2 Part 3 2.2.1, 3.2, 3.3 and 3.4.1 on every package in <outdir>. Each finding is
24*0e16a31dSJim Jagielski# "shall: [rule] ..." or "should: [rule] ...". Schema validity of the manifest is validate.py's job.
25*0e16a31dSJim Jagielskiimport sys, os, re, zipfile, json, struct, subprocess, shutil, warnings
26*0e16a31dSJim Jagielskifrom xml.etree import ElementTree as ET
27*0e16a31dSJim Jagielski
28*0e16a31dSJim JagielskiD = os.path.dirname(os.path.abspath(__file__))
29*0e16a31dSJim JagielskiJING = ["java", "-jar", os.environ.get("ODFV_JING", "/Applications/XMLauthor/lib/jing.jar"), "-i"]
30*0e16a31dSJim JagielskiDSIG_SCHEMA = D + "/schema/OpenDocument-v1.2-os-dsig-schema.rng"
31*0e16a31dSJim JagielskiMANI = "{urn:oasis:names:tc:opendocument:xmlns:manifest:1.0}"
32*0e16a31dSJim JagielskiDSIG = "{urn:oasis:names:tc:opendocument:xmlns:digitalsignature:1.0}"
33*0e16a31dSJim JagielskiERR = re.compile(r"^(.*?):(\d+):(\d+): (error|fatal|warning): (.*)$")
34*0e16a31dSJim Jagielski
35*0e16a31dSJim Jagielski
36*0e16a31dSJim Jagielskidef check_package(path, work):
37*0e16a31dSJim Jagielski    """Returns [(entry, level, rule, message)] for one package."""
38*0e16a31dSJim Jagielski    out = []
39*0e16a31dSJim Jagielski
40*0e16a31dSJim Jagielski    def add(entry, level, rule, msg):
41*0e16a31dSJim Jagielski        out.append((entry, level, rule, msg))
42*0e16a31dSJim Jagielski    z = zipfile.ZipFile(path)
43*0e16a31dSJim Jagielski    infos = z.infolist()
44*0e16a31dSJim Jagielski    names = [i.filename for i in infos]
45*0e16a31dSJim Jagielski    for n in sorted({n for n in names if names.count(n) > 1}):
46*0e16a31dSJim Jagielski        add(n, "shall", "zip-duplicate", "zip entry name occurs more than once")
47*0e16a31dSJim Jagielski    for i in infos:
48*0e16a31dSJim Jagielski        if i.compress_type not in (zipfile.ZIP_STORED, zipfile.ZIP_DEFLATED):
49*0e16a31dSJim Jagielski            add(i.filename, "shall", "zip-method", "compression method %d is neither STORED nor DEFLATED (2.2.1 A)" % i.compress_type)
50*0e16a31dSJim Jagielski        if i.filename.endswith("/"):
51*0e16a31dSJim Jagielski            add(i.filename, "should", "zip-dir-entry", "directories have no zip entries (4.3)")
52*0e16a31dSJim Jagielski    files = {n for n in names if not n.endswith("/")}
53*0e16a31dSJim Jagielski
54*0e16a31dSJim Jagielski    # mimetype (3.3)
55*0e16a31dSJim Jagielski    mime = None
56*0e16a31dSJim Jagielski    if "mimetype" not in files:
57*0e16a31dSJim Jagielski        add("mimetype", "should", "mimetype-missing", "no mimetype file (2.2.1 C)")
58*0e16a31dSJim Jagielski    else:
59*0e16a31dSJim Jagielski        mi = z.getinfo("mimetype")
60*0e16a31dSJim Jagielski        first = min(infos, key=lambda i: i.header_offset)
61*0e16a31dSJim Jagielski        if first.filename != "mimetype" or mi.header_offset != 0:
62*0e16a31dSJim Jagielski            add("mimetype", "shall", "mimetype-first", "mimetype is not the first file of the zip (3.3)")
63*0e16a31dSJim Jagielski        with open(path, "rb") as f:
64*0e16a31dSJim Jagielski            f.seek(mi.header_offset)
65*0e16a31dSJim Jagielski            hdr = f.read(30)
66*0e16a31dSJim Jagielski        if len(hdr) == 30 and hdr[:4] == b"PK\x03\x04":
67*0e16a31dSJim Jagielski            method, extralen = struct.unpack("<H", hdr[8:10])[0], struct.unpack("<H", hdr[28:30])[0]
68*0e16a31dSJim Jagielski            if method != zipfile.ZIP_STORED:
69*0e16a31dSJim Jagielski                add("mimetype", "shall", "mimetype-compressed", "mimetype is compressed, local header method %d (3.3)" % method)
70*0e16a31dSJim Jagielski            if extralen:
71*0e16a31dSJim Jagielski                add("mimetype", "shall", "mimetype-extra", "mimetype has a %d byte extra field in its local header (3.3)" % extralen)
72*0e16a31dSJim Jagielski        else:
73*0e16a31dSJim Jagielski            add("mimetype", "shall", "mimetype-first", "no local file header at the recorded offset")
74*0e16a31dSJim Jagielski        if mi.compress_type != zipfile.ZIP_STORED:
75*0e16a31dSJim Jagielski            add("mimetype", "shall", "mimetype-compressed", "mimetype is compressed, central directory method %d (3.3)" % mi.compress_type)
76*0e16a31dSJim Jagielski        raw = z.read("mimetype")
77*0e16a31dSJim Jagielski        try:
78*0e16a31dSJim Jagielski            mime = raw.decode("ascii")
79*0e16a31dSJim Jagielski        except UnicodeDecodeError:
80*0e16a31dSJim Jagielski            add("mimetype", "shall", "mimetype-ascii", "mimetype content is not ASCII (3.3)")
81*0e16a31dSJim Jagielski
82*0e16a31dSJim Jagielski    # META-INF contents (2.2.1 D, E)
83*0e16a31dSJim Jagielski    for n in sorted(files):
84*0e16a31dSJim Jagielski        if n.startswith("META-INF/") and n != "META-INF/manifest.xml" and "signatures" not in n.rsplit("/", 1)[-1]:
85*0e16a31dSJim Jagielski            add(n, "shall", "meta-inf-file", "file other than manifest.xml and signature files in META-INF (2.2.1 E)")
86*0e16a31dSJim Jagielski
87*0e16a31dSJim Jagielski    # manifest (2.2.1 B, 3.2)
88*0e16a31dSJim Jagielski    entries = None
89*0e16a31dSJim Jagielski    if "META-INF/manifest.xml" not in files:
90*0e16a31dSJim Jagielski        add("META-INF/manifest.xml", "shall", "manifest-missing", "no META-INF/manifest.xml (2.2.1 B)")
91*0e16a31dSJim Jagielski    else:
92*0e16a31dSJim Jagielski        try:
93*0e16a31dSJim Jagielski            root = ET.fromstring(z.read("META-INF/manifest.xml"))
94*0e16a31dSJim Jagielski            if root.tag != MANI + "manifest":
95*0e16a31dSJim Jagielski                add("META-INF/manifest.xml", "shall", "manifest-root", "root element is not manifest:manifest (2.2.1 B.2)")
96*0e16a31dSJim Jagielski            else:
97*0e16a31dSJim Jagielski                entries = [e for e in root if e.tag == MANI + "file-entry"]
98*0e16a31dSJim Jagielski        except ET.ParseError as e:
99*0e16a31dSJim Jagielski            add("META-INF/manifest.xml", "shall", "manifest-malformed", "not well-formed: %s (2.2.1 B.1)" % e)
100*0e16a31dSJim Jagielski    if entries is not None:
101*0e16a31dSJim Jagielski        paths = [e.get(MANI + "full-path") for e in entries]
102*0e16a31dSJim Jagielski        for p in sorted({p for p in paths if paths.count(p) > 1}, key=str):
103*0e16a31dSJim Jagielski            add("META-INF/manifest.xml", "shall", "manifest-duplicate", "more than one file-entry for %s (3.2)" % p)
104*0e16a31dSJim Jagielski        for p in paths:
105*0e16a31dSJim Jagielski            if p in ("META-INF/manifest.xml", "mimetype"):
106*0e16a31dSJim Jagielski                add("META-INF/manifest.xml", "shall", "manifest-self", "file-entry for %s (3.2)" % p)
107*0e16a31dSJim Jagielski            elif p and p.endswith("/") and p != "/":
108*0e16a31dSJim Jagielski                if p + "content.xml" not in files and p + "styles.xml" not in files:
109*0e16a31dSJim Jagielski                    add("META-INF/manifest.xml", "should", "manifest-dir-entry", "file-entry for %s, a directory without a document (4.3)" % p)
110*0e16a31dSJim Jagielski            elif p not in files and p != "/":
111*0e16a31dSJim Jagielski                add("META-INF/manifest.xml", "should", "manifest-dangling", "file-entry for %s, which is not in the package (3.2)" % p)
112*0e16a31dSJim Jagielski        for n in sorted(files):
113*0e16a31dSJim Jagielski            if n != "mimetype" and not n.startswith("META-INF/") and n not in paths:
114*0e16a31dSJim Jagielski                add(n, "shall", "manifest-coverage", "no file-entry in the manifest (3.2)")
115*0e16a31dSJim Jagielski        root_entry = [e for e in entries if e.get(MANI + "full-path") == "/"]
116*0e16a31dSJim Jagielski        if "mimetype" in files and not root_entry:
117*0e16a31dSJim Jagielski            add("META-INF/manifest.xml", "shall", "manifest-root-entry", 'no file-entry for "/" although the package has a mimetype file (3.2)')
118*0e16a31dSJim Jagielski        for e in root_entry:
119*0e16a31dSJim Jagielski            mt = e.get(MANI + "media-type")
120*0e16a31dSJim Jagielski            if "mimetype" not in files:
121*0e16a31dSJim Jagielski                add("mimetype", "shall", "mimetype-missing", 'file-entry for "/" but no mimetype file (3.3)')
122*0e16a31dSJim Jagielski            elif mime is not None and mt != mime:
123*0e16a31dSJim Jagielski                add("mimetype", "shall", "mimetype-mismatch", 'mimetype is "%s" but the "/" media-type is "%s" (3.3)' % (mime, mt))
124*0e16a31dSJim Jagielski        # encrypted entries (3.4.1)
125*0e16a31dSJim Jagielski        for e in entries:
126*0e16a31dSJim Jagielski            if e.find(MANI + "encryption-data") is None:
127*0e16a31dSJim Jagielski                continue
128*0e16a31dSJim Jagielski            p = e.get(MANI + "full-path")
129*0e16a31dSJim Jagielski            size = e.get(MANI + "size")
130*0e16a31dSJim Jagielski            if size is None or not re.fullmatch(r"[0-9]+", size):
131*0e16a31dSJim Jagielski                add(p, "shall", "encrypted-size", "encrypted entry without a valid manifest:size (3.4.1, 4.8.13)")
132*0e16a31dSJim Jagielski            if p in files and z.getinfo(p).compress_type != zipfile.ZIP_STORED:
133*0e16a31dSJim Jagielski                add(p, "shall", "encrypted-method", "encrypted entry is not flagged STORED in the zip (3.4.1)")
134*0e16a31dSJim Jagielski
135*0e16a31dSJim Jagielski    # signature files (2.2.1 D)
136*0e16a31dSJim Jagielski    sigs = sorted(n for n in files if n.startswith("META-INF/") and "signatures" in n.rsplit("/", 1)[-1])
137*0e16a31dSJim Jagielski    for n in sigs:
138*0e16a31dSJim Jagielski        try:
139*0e16a31dSJim Jagielski            root = ET.fromstring(z.read(n))
140*0e16a31dSJim Jagielski        except ET.ParseError as e:
141*0e16a31dSJim Jagielski            add(n, "shall", "signature-malformed", "not well-formed: %s (2.2.1 D.1)" % e)
142*0e16a31dSJim Jagielski            continue
143*0e16a31dSJim Jagielski        if root.tag != DSIG + "document-signatures":
144*0e16a31dSJim Jagielski            add(n, "shall", "signature-root", "root element is not dsig:document-signatures (2.2.1 D.2)")
145*0e16a31dSJim Jagielski            continue
146*0e16a31dSJim Jagielski        p = os.path.join(work, os.path.basename(path), n)
147*0e16a31dSJim Jagielski        os.makedirs(os.path.dirname(p), exist_ok=True)
148*0e16a31dSJim Jagielski        open(p, "wb").write(z.read(n))
149*0e16a31dSJim Jagielski        r = subprocess.run(JING + [DSIG_SCHEMA, p], capture_output=True, text=True)
150*0e16a31dSJim Jagielski        for line in (r.stdout + r.stderr).splitlines():
151*0e16a31dSJim Jagielski            m = ERR.match(line)
152*0e16a31dSJim Jagielski            if m:
153*0e16a31dSJim Jagielski                add(n, "shall", "signature-schema", "%s:%s %s (2.2.1 D.3)" % (m.group(2), m.group(3), m.group(5)))
154*0e16a31dSJim Jagielski            elif line.strip():
155*0e16a31dSJim Jagielski                add(n, "shall", "signature-schema", line.strip())
156*0e16a31dSJim Jagielski    return out
157*0e16a31dSJim Jagielski
158*0e16a31dSJim Jagielski
159*0e16a31dSJim Jagielskidef main():
160*0e16a31dSJim Jagielski    outdir, pkgjson = sys.argv[1], sys.argv[2]
161*0e16a31dSJim Jagielski    if not os.path.abspath(pkgjson).startswith(os.path.join(D, "runs") + os.sep):
162*0e16a31dSJim Jagielski        sys.exit("pkgcheck.py: <pkg.json> must live under " + os.path.join(D, "runs"))
163*0e16a31dSJim Jagielski    work = os.path.abspath(pkgjson[:-5] + ".xml")
164*0e16a31dSJim Jagielski    shutil.rmtree(work, ignore_errors=True)
165*0e16a31dSJim Jagielski    errors, bad_zip = [], []
166*0e16a31dSJim Jagielski    pkgs = sorted(p for p in os.listdir(outdir) if not p.startswith("."))
167*0e16a31dSJim Jagielski    for pkg in pkgs:
168*0e16a31dSJim Jagielski        try:
169*0e16a31dSJim Jagielski            found = check_package(os.path.join(outdir, pkg), work)
170*0e16a31dSJim Jagielski        except zipfile.BadZipFile:
171*0e16a31dSJim Jagielski            bad_zip.append(pkg); continue
172*0e16a31dSJim Jagielski        for entry, level, rule, msg in found:
173*0e16a31dSJim Jagielski            errors.append([pkg + "/" + entry, 0, 0, "%s: [%s] %s" % (level, rule, msg)])
174*0e16a31dSJim Jagielski    json.dump({"packages": pkgs, "bad_zip": bad_zip, "errors": errors}, open(pkgjson, "w"))
175*0e16a31dSJim Jagielski    shall = [e for e in errors if e[3].startswith("shall:")]
176*0e16a31dSJim Jagielski    print(f"package check: {len(shall)} shall and {len(errors) - len(shall)} should findings in "
177*0e16a31dSJim Jagielski          f"{len({e[0].split('/')[0] for e in errors})} of {len(pkgs)} packages, bad zips: {bad_zip}")
178*0e16a31dSJim Jagielski    rules = {}
179*0e16a31dSJim Jagielski    for e in errors:
180*0e16a31dSJim Jagielski        k = e[3].split("]")[0] + "]"
181*0e16a31dSJim Jagielski        rules[k] = rules.get(k, 0) + 1
182*0e16a31dSJim Jagielski    for k in sorted(rules):
183*0e16a31dSJim Jagielski        print(f"  {rules[k]:6d} {k}")
184*0e16a31dSJim Jagielski
185*0e16a31dSJim Jagielski
186*0e16a31dSJim Jagielskimain()
187