xref: /openbmc/openbmc/poky/meta/lib/oe/spdx_common.py (revision 8460358c) 
1#
2# Copyright OpenEmbedded Contributors
3#
4# SPDX-License-Identifier: GPL-2.0-only
5#
6
7import bb
8import collections
9import json
10import oe.packagedata
11import re
12import shutil
13
14from pathlib import Path
15
16
17LIC_REGEX = re.compile(
18    rb"^\W*SPDX-License-Identifier:\s*([ \w\d.()+-]+?)(?:\s+\W*)?$",
19    re.MULTILINE,
20)
21
22
23def extract_licenses(filename):
24    """
25    Extract SPDX License identifiers from a file
26    """
27    try:
28        with open(filename, "rb") as f:
29            size = min(15000, os.stat(filename).st_size)
30            txt = f.read(size)
31            licenses = re.findall(LIC_REGEX, txt)
32            if licenses:
33                ascii_licenses = [lic.decode("ascii") for lic in licenses]
34                return ascii_licenses
35    except Exception as e:
36        bb.warn(f"Exception reading {filename}: {e}")
37    return []
38
39
40def is_work_shared_spdx(d):
41    return '/work-shared/' in d.getVar('S')
42
43
44def load_spdx_license_data(d):
45    with open(d.getVar("SPDX_LICENSES"), "r") as f:
46        data = json.load(f)
47        # Transform the license array to a dictionary
48        data["licenses"] = {l["licenseId"]: l for l in data["licenses"]}
49
50    return data
51
52
53def process_sources(d):
54    """
55    Returns True if the sources for this recipe should be included in the SPDX
56    or False if not
57    """
58    pn = d.getVar("PN")
59    assume_provided = (d.getVar("ASSUME_PROVIDED") or "").split()
60    if pn in assume_provided:
61        for p in d.getVar("PROVIDES").split():
62            if p != pn:
63                pn = p
64                break
65
66    # glibc-locale: do_fetch, do_unpack and do_patch tasks have been deleted,
67    # so avoid archiving source here.
68    if pn.startswith("glibc-locale"):
69        return False
70    if d.getVar("PN") == "libtool-cross":
71        return False
72    if d.getVar("PN") == "libgcc-initial":
73        return False
74    if d.getVar("PN") == "shadow-sysroot":
75        return False
76
77    return True
78
79
80Dep = collections.namedtuple("Dep", ["pn", "hashfn", "in_taskhash"])
81
82
83def collect_direct_deps(d, dep_task):
84    """
85    Find direct dependencies of current task
86
87    Returns the list of recipes that have a dep_task that the current task
88    depends on
89    """
90    current_task = "do_" + d.getVar("BB_CURRENTTASK")
91    pn = d.getVar("PN")
92
93    taskdepdata = d.getVar("BB_TASKDEPDATA", False)
94
95    for this_dep in taskdepdata.values():
96        if this_dep[0] == pn and this_dep[1] == current_task:
97            break
98    else:
99        bb.fatal(f"Unable to find this {pn}:{current_task} in taskdepdata")
100
101    deps = set()
102
103    for dep_name in this_dep.deps:
104        dep_data = taskdepdata[dep_name]
105        if dep_data.taskname == dep_task and dep_data.pn != pn:
106            deps.add((dep_data.pn, dep_data.hashfn, dep_name in this_dep.taskhash_deps))
107
108    return sorted(deps)
109
110
111def get_spdx_deps(d):
112    """
113    Reads the SPDX dependencies JSON file and returns the data
114    """
115    spdx_deps_file = Path(d.getVar("SPDXDEPS"))
116
117    deps = []
118    with spdx_deps_file.open("r") as f:
119        for d in json.load(f):
120            deps.append(Dep(*d))
121    return deps
122
123
124def collect_package_providers(d):
125    """
126    Returns a dictionary where each RPROVIDES is mapped to the package that
127    provides it
128    """
129    deploy_dir_spdx = Path(d.getVar("DEPLOY_DIR_SPDX"))
130
131    providers = {}
132
133    deps = collect_direct_deps(d, "do_create_spdx")
134    deps.append((d.getVar("PN"), d.getVar("BB_HASHFILENAME"), True))
135
136    for dep_pn, dep_hashfn, _ in deps:
137        localdata = d
138        recipe_data = oe.packagedata.read_pkgdata(dep_pn, localdata)
139        if not recipe_data:
140            localdata = bb.data.createCopy(d)
141            localdata.setVar("PKGDATA_DIR", "${PKGDATA_DIR_SDK}")
142            recipe_data = oe.packagedata.read_pkgdata(dep_pn, localdata)
143
144        for pkg in recipe_data.get("PACKAGES", "").split():
145            pkg_data = oe.packagedata.read_subpkgdata_dict(pkg, localdata)
146            rprovides = set(
147                n
148                for n, _ in bb.utils.explode_dep_versions2(
149                    pkg_data.get("RPROVIDES", "")
150                ).items()
151            )
152            rprovides.add(pkg)
153
154            if "PKG" in pkg_data:
155                pkg = pkg_data["PKG"]
156                rprovides.add(pkg)
157
158            for r in rprovides:
159                providers[r] = (pkg, dep_hashfn)
160
161    return providers
162
163
164def get_patched_src(d):
165    """
166    Save patched source of the recipe in SPDX_WORKDIR.
167    """
168    spdx_workdir = d.getVar("SPDXWORK")
169    spdx_sysroot_native = d.getVar("STAGING_DIR_NATIVE")
170    pn = d.getVar("PN")
171
172    workdir = d.getVar("WORKDIR")
173
174    try:
175        # The kernel class functions require it to be on work-shared, so we dont change WORKDIR
176        if not is_work_shared_spdx(d):
177            # Change the WORKDIR to make do_unpack do_patch run in another dir.
178            d.setVar("WORKDIR", spdx_workdir)
179            # Restore the original path to recipe's native sysroot (it's relative to WORKDIR).
180            d.setVar("STAGING_DIR_NATIVE", spdx_sysroot_native)
181
182            # The changed 'WORKDIR' also caused 'B' changed, create dir 'B' for the
183            # possibly requiring of the following tasks (such as some recipes's
184            # do_patch required 'B' existed).
185            bb.utils.mkdirhier(d.getVar("B"))
186
187            bb.build.exec_func("do_unpack", d)
188
189            if d.getVar("SRC_URI") != "":
190                bb.build.exec_func("do_patch", d)
191
192        # Copy source from work-share to spdx_workdir
193        if is_work_shared_spdx(d):
194            share_src = d.getVar('S')
195            d.setVar("WORKDIR", spdx_workdir)
196            d.setVar("STAGING_DIR_NATIVE", spdx_sysroot_native)
197            # Copy source to ${SPDXWORK}, same basename dir of ${S};
198            src_dir = (
199                spdx_workdir
200                + "/"
201                + os.path.basename(share_src)
202            )
203            # For kernel souce, rename suffix dir 'kernel-source'
204            # to ${BP} (${BPN}-${PV})
205            if bb.data.inherits_class("kernel", d):
206                src_dir = spdx_workdir + "/" + d.getVar('BP')
207
208            bb.note(f"copyhardlinktree {share_src} to {src_dir}")
209            oe.path.copyhardlinktree(share_src, src_dir)
210
211        # Some userland has no source.
212        if not os.path.exists(spdx_workdir):
213            bb.utils.mkdirhier(spdx_workdir)
214    finally:
215        d.setVar("WORKDIR", workdir)
216
217
218def has_task(d, task):
219    return bool(d.getVarFlag(task, "task", False)) and not bool(d.getVarFlag(task, "noexec", False))
220
221
222def fetch_data_to_uri(fd, name):
223    """
224    Translates a bitbake FetchData to a string URI
225    """
226    uri = fd.type
227    # Map gitsm to git, since gitsm:// is not a valid URI protocol
228    if uri == "gitsm":
229        uri = "git"
230    proto = getattr(fd, "proto", None)
231    if proto is not None:
232        uri = uri + "+" + proto
233    uri = uri + "://" + fd.host + fd.path
234
235    if fd.method.supports_srcrev():
236        uri = uri + "@" + fd.revisions[name]
237
238    return uri
239