semgrep{,-core}: 0.112.1 -> 1.0.0origin/backport-195838-to-release-22.11

Added update script

(cherry picked from commit f6ca99b86d2e75b4fc416f04ad138aef3c747aae)

4 files changed, 224 insertions, 52 deletions

diff --git a/pkgs/tools/security/semgrep/common.nix b/pkgs/tools/security/semgrep/common.nix
index a721448124e4..0b3524dca567 100644
--- a/pkgs/tools/security/semgrep/common.nix
+++ b/pkgs/tools/security/semgrep/common.nix
@@ -1,41 +1,55 @@
 { lib, fetchFromGitHub, fetchzip, stdenv }:
 
 rec {
-  version = "0.112.1";
+  version = "1.0.0";
 
   src = fetchFromGitHub {
     owner = "returntocorp";
     repo = "semgrep";
     rev = "v${version}";
-    sha256 = "sha256-SZtxZz4x6YUKw1uO5HQTU4lRY989SoCNsPQphJr+L0Y=";
+    sha256 = "sha256-4fNBpokHKCtMB3P0ot1TzcuzOs5hlyH8nIw+bCGqThA=";
   };
 
   # submodule dependencies
   # these are fetched so we:
   #   1. don't fetch the many submodules we don't need
   #   2. avoid fetchSubmodules since it's prone to impurities
-  langsSrc = fetchFromGitHub {
-    owner = "returntocorp";
-    repo = "semgrep-langs";
-    rev = "91e288062eb794e8a5e6967d1009624237793491";
-    sha256 = "sha256-z2t2bTRyj5zu9h/GBg2YeRFimpJsd3dA7dK8VBaKzHo=";
-  };
-
-  interfacesSrc = fetchFromGitHub {
-    owner = "returntocorp";
-    repo = "semgrep-interfaces";
-    rev = "7bc457a32e088ef21adf1529fa0ddeea634b9131";
-    sha256 = "sha256-xN8Qm1/YLa49k9fZKDoPPmHASI2ipI3mkKlwEK2ajO4=";
+  submodules = {
+    "cli/src/semgrep/lang" = fetchFromGitHub {
+      owner = "returntocorp";
+      repo = "semgrep-langs";
+      rev = "65cb2ed80e31e01b122f893fef8428d14432da75";
+      sha256 = "sha256-HdPJdOlMM1l7vNSATkEu5KrCkpt2feEAH8LFDU84KUM=";
+    };
+    "cli/src/semgrep/semgrep_interfaces" = fetchFromGitHub {
+      owner = "returntocorp";
+      repo = "semgrep-interfaces";
+      rev = "c69e30a4cf39f11cab5378700f5e193e8282079e";
+      sha256 = "sha256-Wr3/TWx/LHiTFCoGY4sqdsn3dHvMsEIVYA3RGiv88xQ=";
+    };
   };
 
   # fetch pre-built semgrep-core since the ocaml build is complex and relies on
   # the opam package manager at some point
-  coreRelease = if stdenv.isDarwin then fetchzip {
-      url = "https://github.com/returntocorp/semgrep/releases/download/v${version}/semgrep-v${version}-osx.zip";
-      sha256 = "sha256-JiOH39vMDL6r9WKuPO0CDkRwGZtzl/GIFoSegVddFpw=";
-  } else fetchzip {
-      url = "https://github.com/returntocorp/semgrep/releases/download/v${version}/semgrep-v${version}-ubuntu-16.04.tgz";
-      sha256 = "sha256-V6r+VQrgz8uVSbRa2AmW4lnLxovk63FL7LqVKD46RBw=";
+  core = rec {
+    data = {
+      x86_64-linux = {
+        suffix = "-ubuntu-16.04.tgz";
+        sha256 = "sha256-SsaAuhcDyO3nr6H2xOtdxzOoEQd6aIe0mlpehvDWzU0=";
+      };
+      x86_64-darwin = {
+        suffix = "-osx.zip";
+        sha256 = "sha256-DAcAB/q6XeljCp4mVljIJB4AUjUuzMSRMFzIuyjWMew=";
+      };
+    };
+    src = let
+      inherit (stdenv.hostPlatform) system;
+      selectSystemData = data: data.${system} or (throw "Unsupported system: ${system}");
+      inherit (selectSystemData data) suffix sha256;
+    in fetchzip {
+      url = "https://github.com/returntocorp/semgrep/releases/download/v${version}/semgrep-v${version}${suffix}";
+      inherit sha256;
+    };
   };
 
   meta = with lib; {
diff --git a/pkgs/tools/security/semgrep/default.nix b/pkgs/tools/security/semgrep/default.nix
index 1df7c515e305..c97d877b483e 100644
--- a/pkgs/tools/security/semgrep/default.nix
+++ b/pkgs/tools/security/semgrep/default.nix
@@ -15,12 +15,26 @@ let
 in
 buildPythonApplication rec {
   pname = "semgrep";
-  inherit (common) version;
-  src = "${common.src}/cli";
+  inherit (common) src version;
 
-  SEMGREP_CORE_BIN = "${semgrep-core}/bin/semgrep-core";
+  postPatch = (lib.concatStringsSep "\n" (lib.mapAttrsToList (
+    path: submodule: ''
+      # substitute ${path}
+      # remove git submodule placeholder
+      rm -r ${path}
+      # link submodule
+      ln -s ${submodule}/ ${path}
+    ''
+  ) common.submodules)) + ''
+    cd cli
+  '';
 
   nativeBuildInputs = [ pythonRelaxDepsHook ];
+  # tell cli/setup.py to not copy semgrep-core into the result
+  # this means we can share a copy of semgrep-core and avoid an issue where it
+  # copies the binary but doesn't retain the executable bit
+  SEMGREP_SKIP_BIN = true;
+
   pythonRelaxDeps = [
     "attrs"
     "boltons"
@@ -28,13 +42,26 @@ buildPythonApplication rec {
     "typing-extensions"
   ];
 
-  postPatch = ''
-    # remove git submodule placeholders
-    rm -r ./src/semgrep/{lang,semgrep_interfaces}
-    # link submodule dependencies
-    ln -s ${common.langsSrc}/ ./src/semgrep/lang
-    ln -s ${common.interfacesSrc}/ ./src/semgrep/semgrep_interfaces
-  '';
+  propagatedBuildInputs = with pythonPackages; [
+    attrs
+    boltons
+    colorama
+    click
+    click-option-group
+    glom
+    requests
+    ruamel-yaml
+    tqdm
+    packaging
+    jsonschema
+    wcmatch
+    peewee
+    defusedxml
+    urllib3
+    typing-extensions
+    python-lsp-jsonrpc
+    tomli
+  ];
 
   doCheck = true;
   checkInputs = [ git pytestCheckHook ] ++ (with pythonPackages; [
@@ -45,7 +72,9 @@ buildPythonApplication rec {
   ]);
   disabledTests = [
     # requires networking
-    "tests/unit/test_metric_manager.py"
+    "test_send"
+    # requires networking
+    "test_parse_exclude_rules_auto"
   ];
   preCheck = ''
     # tests need a home directory
@@ -59,25 +88,16 @@ buildPythonApplication rec {
       --replace 'addopts = "--splitting-algorithm=least_duration"' ""
   '';
 
-  propagatedBuildInputs = with pythonPackages; [
-    attrs
-    boltons
-    colorama
-    click
-    click-option-group
-    glom
-    requests
-    ruamel-yaml
-    tqdm
-    packaging
-    jsonschema
-    wcmatch
-    peewee
-    defusedxml
-    urllib3
-    typing-extensions
-    python-lsp-jsonrpc
-  ];
+  # since we stop cli/setup.py from finding semgrep-core and copying it into
+  # the result we need to provide it on the PATH
+  preFixup = ''
+    makeWrapperArgs+=(--prefix PATH : ${lib.makeBinPath [ semgrep-core ]})
+  '';
+
+  passthru = {
+    inherit common;
+    updateScript = ./update.sh;
+  };
 
   meta = common.meta // {
     description = common.meta.description + " - cli";
diff --git a/pkgs/tools/security/semgrep/semgrep-core.nix b/pkgs/tools/security/semgrep/semgrep-core.nix
index 3a9c904ad733..e5ce941298a4 100644
--- a/pkgs/tools/security/semgrep/semgrep-core.nix
+++ b/pkgs/tools/security/semgrep/semgrep-core.nix
@@ -6,8 +6,7 @@ in
 stdenvNoCC.mkDerivation rec {
   pname = "semgrep-core";
   inherit (common) version;
-
-  src = common.coreRelease;
+  inherit (common.core) src;
 
   installPhase = ''
     runHook preInstall
diff --git a/pkgs/tools/security/semgrep/update.sh b/pkgs/tools/security/semgrep/update.sh
new file mode 100755
index 000000000000..2b464d602018
--- /dev/null
+++ b/pkgs/tools/security/semgrep/update.sh
@@ -0,0 +1,139 @@
+#!/usr/bin/env nix-shell
+#!nix-shell -i bash -p curl gnused jq
+
+set -euxo pipefail
+
+# provide a github token so you don't get rate limited
+# if you use gh cli you can use:
+#     `export GITHUB_TOKEN="$(cat ~/.config/gh/config.yml | yq '.hosts."github.com".oauth_token' -r)"`
+# or just set your token by hand:
+#     `read -s -p "Enter your token: " GITHUB_TOKEN; export GITHUB_TOKEN`
+#     (we use read so it doesn't show in our shell history and in secret mode so the token you paste isn't visible)
+if [ -z "${GITHUB_TOKEN:-}" ]; then
+    echo "no GITHUB_TOKEN provided - you could meet API request limiting" >&2
+fi
+
+ROOT="$(dirname "$(readlink -f "$0")")"
+NIXPKGS_ROOT="$ROOT/../../../.."
+NIX_DRV="$ROOT/default.nix"
+
+COMMON_FILE="$ROOT/common.nix"
+
+instantiateClean() {
+    nix-instantiate -A "$1" --eval --strict | cut -d\" -f2
+}
+
+# get latest version
+NEW_VERSION=$(
+  curl -s -H
+    "Accept: application/vnd.github.v3+json" \
+    ${GITHUB_TOKEN:+ -H "Authorization: bearer $GITHUB_TOKEN"} \
+    https://api.github.com/repos/returntocorp/semgrep/releases/latest \
+  | jq -r '.tag_name'
+)
+# trim v prefix
+NEW_VERSION="${NEW_VERSION:1}"
+OLD_VERSION="$(instantiateClean semgrep.common.version)"
+
+if [[ "$OLD_VERSION" == "$NEW_VERSION" ]]; then
+    echo "Already up to date"
+    exit
+fi
+
+replace() {
+    sed -i "s@$1@$2@g" "$3"
+}
+
+fetchgithub() {
+    set +eo pipefail
+    nix-build -A "$1" 2>&1 >/dev/null | grep "got:" | cut -d':' -f2 | sed 's| ||g'
+    set -eo pipefail
+}
+
+fetchzip() {
+    set +eo pipefail
+    nix-build -E "with import $NIXPKGS_ROOT {}; fetchzip {url = \"$1\"; sha256 = lib.fakeSha256; }" 2>&1 >/dev/null | grep "got:" | cut -d':' -f2 | sed 's| ||g'
+    set -eo pipefail
+}
+
+replace "$OLD_VERSION" "$NEW_VERSION" "$COMMON_FILE"
+
+echo "Updating src"
+
+OLD_HASH="$(instantiateClean semgrep.common.src.outputHash)"
+echo "Old hash $OLD_HASH"
+TMP_HASH="sha256-AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA="
+replace "$OLD_HASH" "$TMP_HASH" "$COMMON_FILE"
+NEW_HASH="$(fetchgithub semgrep.common.src)"
+echo "New hash $NEW_HASH"
+replace "$TMP_HASH" "$NEW_HASH" "$COMMON_FILE"
+
+echo "Updated src"
+
+# loop through platforms for core
+nix-instantiate -E "with import $NIXPKGS_ROOT {}; builtins.attrNames semgrep.common.core.data" --eval --strict --json \
+| jq '.[]' -r \
+| while read -r PLATFORM; do
+    echo "Updating core for $PLATFORM"
+    SUFFIX=$(instantiateClean semgrep.common.core.data."$1".suffix "$PLATFORM")
+    OLD_HASH=$(instantiateClean semgrep.common.core.data."$1".sha256 "$PLATFORM")
+    echo "Old hash $OLD_HASH"
+
+    NEW_URL="https://github.com/returntocorp/semgrep/releases/download/v$NEW_VERSION/semgrep-v$NEW_VERSION$SUFFIX"
+    NEW_HASH="$(fetchzip "$NEW_URL")"
+    echo "New hash $NEW_HASH"
+
+    replace "$OLD_HASH" "$NEW_HASH" "$COMMON_FILE"
+
+    echo "Updated core for $PLATFORM"
+done
+
+OLD_PWD=$PWD
+TMPDIR="$(mktemp -d)"
+# shallow clone to check submodule commits, don't actually need the submodules
+git clone https://github.com/returntocorp/semgrep "$TMPDIR/semgrep" --depth 1 --branch "v$NEW_VERSION"
+
+get_submodule_commit() {
+    OLD_PWD=$PWD
+    (
+        cd "$TMPDIR/semgrep"
+        git ls-tree --object-only HEAD "$1"
+        cd "$OLD_PWD"
+    )
+}
+
+# loop through submodules
+nix-instantiate -E "with import $NIXPKGS_ROOT {}; builtins.attrNames semgrep.passthru.common.submodules" --eval --strict --json \
+| jq '.[]' -r \
+| while read -r SUBMODULE; do
+    echo "Updating $SUBMODULE"
+    OLD_REV=$(instantiateClean semgrep.passthru.common.submodules."$SUBMODULE".rev)
+    echo "Old commit $OLD_REV"
+    OLD_HASH=$(instantiateClean semgrep.passthru.common.submodules."$SUBMODULE".outputHash)
+    echo "Old hash $OLD_HASH"
+
+    NEW_REV=$(get_submodule_commit "$SUBMODULE")
+    echo "New commit $NEW_REV"
+
+    if [[ "$OLD_REV" == "$NEW_REV" ]]; then
+      echo "$SUBMODULE already up to date"
+      continue
+    fi
+
+    NEW_URL=$(instantiateClean semgrep.passthru.common.submodules."$SUBMODULE".url | sed "s@$OLD_REV@$NEW_REV@g")
+    NEW_HASH=$(nix --experimental-features nix-command hash to-sri "sha256:$(nix-prefetch-url "$NEW_URL")")
+
+    TMP_HASH="sha256-ABAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA="
+    replace "$OLD_REV" "$NEW_REV" "$COMMON_FILE"
+    replace "$OLD_HASH" "$TMP_HASH" "$COMMON_FILE"
+    NEW_HASH="$(fetchgithub semgrep.passthru.common.submodules."$SUBMODULE")"
+    echo "New hash $NEW_HASH"
+    replace "$TMP_HASH" "$NEW_HASH" "$COMMON_FILE"
+
+    echo "Updated $SUBMODULE"
+done
+
+rm -rf "$TMPDIR"
+
+echo "Finished"
+