zephyr/scripts/west_commands/zspdx/getincludes.py
Steve Winslow fd31b9b4ac west: spdx: Generate SPDX 2.2 tag-value documents
This adds support to generate SPDX 2.2 tag-value documents via the
new west spdx command. The CMake file-based APIs are leveraged to
create relationships from source files to the corresponding
generated build files. SPDX-License-Identifier comments in source
files are scanned and filled into the SPDX documents.

Before `west build` is run, a specific file must be created in the
build directory so that the CMake API reply will run. This can be
done by running:

    west spdx --init -d BUILD_DIR

After `west build` is run, SPDX generation is then activated by
calling `west spdx`; currently this requires passing the build
directory as a parameter again:

    west spdx -d BUILD_DIR

This will generate three SPDX documents in `BUILD_DIR/spdx/`:

1) `app.spdx`: This contains the bill-of-materials for the
application source files used for the build.

2) `zephyr.spdx`: This contains the bill-of-materials for the
specific Zephyr source code files that are used for the build.

3) `build.spdx`: This contains the bill-of-materials for the built
output files.

Each file in the bill-of-materials is scanned, so that its hashes
(SHA256 and SHA1) can be recorded, along with any detected licenses
if an `SPDX-License-Identifier` appears in the file.

SPDX Relationships are created to indicate dependencies between
CMake build targets; build targets that are linked together; and
source files that are compiled to generate the built library files.

`west spdx` can be called with optional parameters for further
configuration:

* `-n PREFIX`: specifies a prefix for the Document Namespaces that
will be included in the generated SPDX documents. See SPDX spec 2.2
section 2.5 at
https://spdx.github.io/spdx-spec/2-document-creation-information/.
If -n is omitted, a default namespace will be generated according
to the default format described in section 2.5 using a random UUID.

* `-s SPDX_DIR`: specifies an alternate directory where the SPDX
documents should be written. If not specified, they will be saved
in `BUILD_DIR/spdx/`.

* `--analyze-includes`: in addition to recording the compiled
source code files (e.g. `.c`, `.S`) in the bills-of-materials, if
this flag is specified, `west spdx` will attempt to determine the
specific header files that are included for each `.c` file. This
will take longer, as it performs a dry run using the C compiler
for each `.c` file (using the same arguments that were passed to it
for the actual build).

* `--include-sdk`: if `--analyze-includes` is used, then adding
`--include-sdk` will create a fourth SPDX document, `sdk.spdx`,
which will list any header files included from the SDK.

Signed-off-by: Steve Winslow <steve@swinslow.net>
2021-05-05 11:14:06 -04:00

66 lines
2.3 KiB
Python

# Copyright (c) 2021 The Linux Foundation
#
# SPDX-License-Identifier: Apache-2.0
from subprocess import run, PIPE
from west import log
# Given a path to the applicable C compiler, a C source file, and the
# corresponding TargetCompileGroup, determine which include files would
# be used.
# Arguments:
# 1) path to applicable C compiler
# 2) C source file being analyzed
# 3) TargetCompileGroup for the current target
# Returns: list of paths to include files, or [] on error or empty findings.
def getCIncludes(compilerPath, srcFile, tcg):
log.dbg(f" - getting includes for {srcFile}")
# prepare fragments
fragments = [fr for fr in tcg.compileCommandFragments if fr.strip() != ""]
# prepare include arguments
includes = ["-I" + incl.path for incl in tcg.includes]
# prepare defines
defines = ["-D" + d.define for d in tcg.defines]
# prepare command invocation
cmd = [compilerPath, "-E", "-H"] + fragments + includes + defines + [srcFile]
cp = run(cmd, stdout=PIPE, stderr=PIPE, universal_newlines=True)
if cp.returncode != 0:
log.dbg(f" - calling {compilerPath} failed with error code {cp.returncode}")
return []
else:
# response will be in cp.stderr, not cp.stdout
return extractIncludes(cp.stderr)
# Parse the response from the CC -E -H call, to extract the include file paths
def extractIncludes(resp):
includes = set()
# lines we want will start with one or more periods, followed by
# a space and then the include file path, e.g.:
# .... /home/steve/programming/zephyr/zephyrproject/zephyr/include/kernel.h
# the number of periods indicates the depth of nesting (for transitively-
# included files), but here we aren't going to care about that. We'll
# treat everything as tied to the corresponding source file.
# once we hit the line "Multiple include guards may be useful for:",
# we're done; ignore everything after that
for rline in resp.splitlines():
if rline.startswith("Multiple include guards"):
break
if rline[0] == ".":
sline = rline.split(" ", maxsplit=1)
if len(sline) != 2:
continue
includes.add(sline[1])
includesList = list(includes)
includesList.sort()
return includesList