# Running Linters It is common to run some form of linters over the code base. It is desirable to also use our build tool for this to have the benefit of parallel (or even remote) build and sound caching. Additionally, this also allows the lint tools to see the file layout as it occurs in the actual compile action, including generated files. Remember that even for source files this layout does not have to coincide with the layout of files in the source repository. Conveniently, our build rules have support for collecting the relevant information needed for linting built in. If a target is built with the configuration variable `LINT` set to a true value, lint information is provided for the transitive sources; as [third-party dependencies](third-party-software.md) are typically exported without `LINT` among the flexible variables, that naturally forms a boundary of the "own" code (to be linted, as opposed to third-party code). So, continuing the [third-party tutorial](third-party-software.md), we can obtain abstract nodes for our sources (`main.cpp`, `greet/greet.hpp`, `greet/greet.cpp`). ``` sh $ just-mr analyse -D '{"LINT": true}' --dump-nodes - INFO: Performing repositories setup INFO: Found 5 repositories involved INFO: Setup finished, exec ["just","analyse","-C","...","-D","{\"LINT\": true}","--dump-nodes","-"] INFO: Requested target is [["@","tutorial","","helloworld"],{"LINT":true}] INFO: Analysed target [["@","tutorial","","helloworld"],{"LINT":true}] INFO: Export targets found: 0 cached, 1 uncached, 0 not eligible for caching INFO: Result of target [["@","tutorial","","helloworld"],{"LINT":true}]: { "artifacts": { "helloworld": {"data":{"id":"1154ef311dff82653bc6a1a92bfc6152bc116cb86652b4b5218385fe39054391","path":"work/helloworld"},"type":"ACTION"} }, "provides": { "debug-hdrs": { }, "debug-srcs": { }, "dwarf-pkg": { }, "lint": [ {"id":"aec2651a9f3b869554e3a2e5cc3ab85d86610a79aae388a4f7839396dc2167d2","type":"NODE"}, {"id":"509e506f8d0c2ebe4fca63fe7cc528be17165bae464410ac5fba97a6ed92930d","type":"NODE"}, {"id":"cb3bd5f3934e199ddf5d43eae6866f8c6ba449f060213a244c93f0826c070c3f","type":"NODE"} ], "package": { "to_bin": true }, "run-libs": { } }, "runfiles": { } } INFO: Target nodes of target [["@","tutorial","","helloworld"],{"LINT":true}]: { "509e506f8d0c2ebe4fca63fe7cc528be17165bae464410ac5fba97a6ed92930d": { "result": { "artifact_stage": { "include": { "data": { "id": "a882ca13d51c70aa6b02d4996aa426ee6e21bf85881ec0f6dbe0a278f5a27b7b" }, "type": "TREE" }, "work/greet/greet.hpp": { "data": { "path": "greet/greet.hpp", "repository": "tutorial" }, "type": "LOCAL" } }, "provides": { "cmd": ["c++","-O2","-Wall","-I","work","-isystem","include","-E","work/greet/greet.hpp"], "direct deps artifact names": ["include/fmt","work/greet/greet.hpp"], "extra outs": [], "src": "work/greet/greet.hpp" }, "runfiles": { } }, "type": "VALUE_NODE" }, "aec2651a9f3b869554e3a2e5cc3ab85d86610a79aae388a4f7839396dc2167d2": { "result": { "artifact_stage": { "include": { "data": { "id": "29d6c7fef3c48a1a3b15edd770b33d073d5c9cd1e6e9fb22917831fcdd762ebb" }, "type": "TREE" }, "work/main.cpp": { "data": { "path": "main.cpp", "repository": "tutorial" }, "type": "LOCAL" } }, "provides": { "cmd": ["c++","-O2","-Wall","-I","work","-isystem","include","-c","work/main.cpp","-o","work/main.o"], "direct deps artifact names": ["include/greet/greet.hpp"], "extra outs": [], "src": "work/main.cpp" }, "runfiles": { } }, "type": "VALUE_NODE" }, "cb3bd5f3934e199ddf5d43eae6866f8c6ba449f060213a244c93f0826c070c3f": { "result": { "artifact_stage": { "include": { "data": { "id": "a882ca13d51c70aa6b02d4996aa426ee6e21bf85881ec0f6dbe0a278f5a27b7b" }, "type": "TREE" }, "work/greet/greet.cpp": { "data": { "path": "greet/greet.cpp", "repository": "tutorial" }, "type": "LOCAL" }, "work/greet/greet.hpp": { "data": { "path": "greet/greet.hpp", "repository": "tutorial" }, "type": "LOCAL" } }, "provides": { "cmd": ["c++","-O2","-Wall","-I","work","-isystem","include","-c","work/greet/greet.cpp","-o","work/greet/greet.o"], "direct deps artifact names": ["include/fmt","work/greet/greet.hpp"], "extra outs": [], "src": "work/greet/greet.cpp" }, "runfiles": { } }, "type": "VALUE_NODE" } } ``` We find the sources in correct staging, together with the respective compile command (or preprocessing, in case of headers) provided. The latter is important, to find the correct include files and to know the correct defines to be used. Of course, those abstract nodes are just an implementation detail and there is a rule to define linting for the collected sources. It takes two programs (targets consisting of a single artifact), - the `linter` for running the lint task on a single file, and - the `summarizer` for summarizing the lint results; additionally, arbitrary `config` data can be given to have config files available, but also to use a linter built from source. As for every rule, the details can be obtained with the `describe` subcommand. ``` sh $ just-mr --main rules-cc describe --rule lint targets INFO: Performing repositories setup INFO: Found 2 repositories involved INFO: Setup finished, exec ["just","describe","-C","...","--rule","lint","targets"] | Run a given linter on the lint information provided by the given targets. ... Target fields - "linter" | Single artifact running the lint checks. | | This program is invoked with ... ``` Let's go through these programs we have to provide one by one. The first one is supposed to call the actual linter; as many linters, including `clang-tidy` which we use as an example, prefer to obtain the command information through a [compilation database](https://clang.llvm.org/docs/JSONCompilationDatabase.html) there is actually some work to do, especially as the directory entry has to be an absolute path. We also move the configuration file `.clang-tidy` from the configuration directory (located in a directory given to us through the environment variable `CONFIG`) to the position expected by `clang-tidy`. ``` {.python srcname="run_clang_tidy.py"} #!/usr/bin/env python3 import json import os import shutil import subprocess import sys def dump_meta(src, cmd): OUT = os.environ.get("OUT") with open(os.path.join(OUT, "config.json"), "w") as f: json.dump({"src": src, "cmd": cmd}, f) def run_lint(src, cmd): dump_meta(src, cmd) config = os.environ.get("CONFIG") shutil.copyfile(os.path.join(config, ".clang-tidy"), ".clang-tidy") db = [ {"directory": os.getcwd(), "arguments": cmd, "file": src}] with open("compile_commands.json", "w") as f: json.dump(db,f) new_cmd = [ "clang-tidy", src ] return subprocess.run(new_cmd).returncode if __name__ == "__main__": sys.exit(run_lint(sys.argv[1], sys.argv[2:])) ``` The actual information on success or failure is provided through the exit code and information on the problems discovered (if any) is reported on stdout or stderr. Additionally, our launcher also writes the meta data in a file `config.json` in the directory for additional (usually machine-readable) diagnose output; the location of this directory is given to us by the environment variable `OUT`. We use a pretty simple `.clang-tidy` for demonstration purpose. ``` {.md srcname=".clang-tidy"} Checks: 'clang-analyzer-*,misc-*,-misc-include-*' WarningsAsErrors: 'clang-analyzer-*,misc-*,-misc-include-*' ``` Computing a summary of the individual lint results (given to the summarizer as subdirectories of the current working directory) is straight forward: the overall linting passed if all individual checks passed and for the failed tests we format stdout and stderr in some easy-to-read way; additionally, we also provide a machine-readable summary of the failures. ``` {.python srcname="summary.py"} #!/usr/bin/env python3 import json import os import sys FAILED = {} for lint in sorted(os.listdir()): if os.path.isdir(lint): with open(os.path.join(lint, "result")) as f: result = f.read().strip() if result != "PASS": record = {} with open(os.path.join(lint, "out/config.json")) as f: record["config"] = json.load(f) with open(os.path.join(lint, "stdout")) as f: log = f.read() with open(os.path.join(lint, "stderr")) as f: log += f.read() record["log"] = log FAILED[lint] = record with open(os.path.join(os.environ.get("OUT"), "failures.json"), "w") as f: json.dump(FAILED, f) failures = list(FAILED.keys()) for f in failures: src = FAILED[f]["config"]["src"] log = FAILED[f]["log"] print("%s %s" % (f, src)) print("".join([" " + line + "\n" for line in log.splitlines()])) if failures: sys.exit(1) ``` Of course, our launcher and summarizer have to be executable ``` sh $ chmod 755 run_clang_tidy.py summary.py ``` Now we can define our lint target. ``` {.jsonc srcname="TARGETS"} ... , "lint": { "type": ["@", "rules", "lint", "targets"] , "targets": ["helloworld"] , "linter": ["run_clang_tidy.py"] , "summarizer": ["summary.py"] , "config": [".clang-tidy"] } ... ``` As most rules, the lint rules also have a `"defaults"` target, which allows to set `PATH` appropriately for all lint actions. This can be useful if the linters are installed in a non-standard directory. ``` sh $ mkdir -p tutorial-defaults/lint $ echo '{"defaults": {"type": "defaults", "PATH": ["'"${TOOLCHAIN_PATH}"'"]}}' > tutorial-defaults/lint/TARGETS $ git add tutorial-defaults $ git commit -m 'add lint defaults' ``` We now can build our lint report in the same way as any test report. ``` sh $ just-mr build lint -P report INFO: Performing repositories setup INFO: Found 5 repositories involved INFO: Setup finished, exec ["just","build","-C","...","lint","-P","report"] INFO: Requested target is [["@","tutorial","","lint"],{}] INFO: Analysed target [["@","tutorial","","lint"],{}] INFO: Export targets found: 0 cached, 1 uncached, 0 not eligible for caching INFO: Target tainted ["lint"]. INFO: Discovered 11 actions, 0 tree overlays, 7 trees, 0 blobs INFO: Building [["@","tutorial","","lint"],{}]. INFO: Processed 7 actions, 3 cache hits. INFO: Artifacts built, logical paths are: out [a90a9e3a8ac23526eb31ae46c80434cfd5810ed5:41:t] report [e69de29bb2d1d6434b8b29ae775ad8c2e48c5391:0:f] result [7ef22e9a431ad0272713b71fdc8794016c8ef12f:5:f] work [52b9cfc07b53c59fb066bc95329f4ca6457e7338:111:t] INFO: Backing up artifacts of 1 export targets INFO: Target tainted ["lint"]. ``` Note the difference between the discovered and processed actions. This is inherent to the way linting works: in order to know the precise command line with which a source file is compiled, the respective target has to be analysed. However, to lint the source files, it is not necessary to actually build the binary whose source code should be linted. Of course, if a combined test target that includes linting as well as end-to-end tests is considered, the binary has to be built (and hence the actions processed) for other reasons. To see that some real linting is going on, let's modify one of our source files. Say, we'll make the greeting independent of the recipient. ``` {.cpp srcname="greet/greet.cpp"} #include "greet.hpp" #include void greet(std::string const& s) { fmt::print("Hello!\n"); } ``` Building succeeds without any warning. ``` sh $ just-mr build helloworld INFO: Performing repositories setup INFO: Found 5 repositories involved INFO: Setup finished, exec ["just","build","-C","...","helloworld"] INFO: Requested target is [["@","tutorial","","helloworld"],{}] INFO: Analysed target [["@","tutorial","","helloworld"],{}] INFO: Export targets found: 1 cached, 0 uncached, 0 not eligible for caching INFO: Discovered 4 actions, 0 tree overlays, 2 trees, 0 blobs INFO: Building [["@","tutorial","","helloworld"],{}]. INFO: Processed 4 actions, 1 cache hits. INFO: Artifacts built, logical paths are: helloworld [2cb87c743e9fd3d18543732945df3ef9ca084be6:132736:x] ``` However, the linter reports it. ``` sh $ just-mr build lint -P report || : INFO: Performing repositories setup INFO: Found 5 repositories involved INFO: Setup finished, exec ["just","build","-C","...","lint","-P","report"] INFO: Requested target is [["@","tutorial","","lint"],{}] INFO: Analysed target [["@","tutorial","","lint"],{}] INFO: Export targets found: 1 cached, 0 uncached, 0 not eligible for caching INFO: Target tainted ["lint"]. INFO: Discovered 8 actions, 0 tree overlays, 6 trees, 0 blobs INFO: Building [["@","tutorial","","lint"],{}]. WARN (action:415a94f0c74ec937e2504b9ef5f94696232ff2c57eb2bec00c226896e2eb8be6): lint failed for work/greet/greet.cpp (exit code 1); outputs: - "out" [caf25f0a518d21909625f9a7974002796f6d8b5f:39:t] - "result" [94e1707e853c36f514de3876408c09a0e0ca6fc4:5:f] - "stderr" [ffc377e8898697782ab96419f6ab82c60985c752:235:f] - "stdout" [17975e013bd2cc3f66509e11737e1e169f1bd162:231:f] INFO: Processed 4 actions, 2 cache hits. INFO: Artifacts built, logical paths are: out [c298959107421711f8d87a2b96e95858c065b9b9:41:t] FAILED report [0b0ab9eb90c28ece0f14a13a6ae5c97da4a32170:531:f] FAILED result [94e1707e853c36f514de3876408c09a0e0ca6fc4:5:f] FAILED work [007eec6bad8b691c067dd2c54165ac2912711474:111:t] FAILED INFO: Failed artifacts: out [c298959107421711f8d87a2b96e95858c065b9b9:41:t] FAILED report [0b0ab9eb90c28ece0f14a13a6ae5c97da4a32170:531:f] FAILED result [94e1707e853c36f514de3876408c09a0e0ca6fc4:5:f] FAILED work [007eec6bad8b691c067dd2c54165ac2912711474:111:t] FAILED 0000000002 work/greet/greet.cpp work/greet/greet.cpp:4:31: error: parameter 's' is unused [misc-unused-parameters,-warnings-as-errors] 4 | void greet(std::string const& s) { | ^ | /*s*/ 287 warnings generated. Suppressed 286 warnings (286 in non-user code). Use -header-filter=.* to display errors from all non-system headers. Use -system-headers to display errors from system headers as well. 1 warning treated as error INFO: Target tainted ["lint"]. WARN: Build result contains failed artifacts. ```