Merge pull request #293 from nix-community/fix-index-apps

Improve indexer architecture

src/apps/translate-index/default.nix

@@ -1,15 +1,26 @@
 {
+  callNixWithD2N,
   utils,
   translate,
   coreutils,
   jq,
   parallel,
+  python3,
   writeScript,
   ...
 }: let
-  script = writeScript "run-translate" ''
-    ${translate}/bin/translate $1 $targetDir || echo "Failed to translate $1"
-  '';
+  script =
+    utils.writePureShellScript
+    [coreutils translate jq python3]
+    ''
+      job_nr=$2
+      time=$(date +%s)
+      runtime=$(($time - $start_time))
+      average_runtime=$(python3 -c "print($runtime / $job_nr)")
+      total_remaining_time=$(python3 -c "print($average_runtime * ($num_jobs - $job_nr))")
+      echo "starting job nr. $job_nr; average job runtime: $average_runtime sec; remaining time: $total_remaining_time sec"
+      translate $1 $targetDir || echo "Failed to translate $1"
+    '';
 in
   utils.writePureShellScriptBin
   "translate-index"
@@ -31,6 +42,26 @@ in
 
     export targetDir
 
-    parallel --halt now,fail=1 -j$(nproc) --delay 1 -a <(jq '.[]' -c -r $index) ${script}
+    export num_jobs=$(jq 'length' -c -r $index)
+    seq $num_jobs > $TMPDIR/job_numbers
 
+    JOBS=''${JOBS:-$(nproc)}
+
+    # build translator executables
+    export TRANSLATOR_DIR=$TMPDIR/translators
+    for translator in $(jq '.[] | .translator' -c -r libraries-io/index.json); do
+      bin="$TRANSLATOR_DIR/$translator"
+      if [ ! -e "$bin" ]; then
+        echo "building executable for translator $translator"
+        ${callNixWithD2N} build -o "$bin" "
+          dream2nix.framework.translatorInstances.$translator.translateBin
+        "
+      fi
+    done
+
+    export start_time=$(date +%s)
+    parallel --halt now,fail=1 -j$JOBS --link -a <(jq '.[]' -c -r $index) -a $TMPDIR/job_numbers ${script}
+
+    runtime=$(($(date +%s) - $start_time))
+    echo "FINISHED! Executed $num_jobs jobs in $runtime seconds"
   ''

src/apps/translate/default.nix

@@ -46,9 +46,14 @@ utils.writePureShellScriptBin
 
   echo -e "\nTranslating:: $name (translator: $translator) (lock path: $dreamLockPath)"
 
-  translateBin=$(${callNixWithD2N} build --print-out-paths --no-link "
-    dream2nix.framework.translatorInstances.$translator.translateBin
-  ")
+  # allow pre-built translator executables to avoid the `nix build` on each run
+  if [ -n "$TRANSLATOR_DIR" ]; then
+    translateBin="$TRANSLATOR_DIR/$translator"
+  else
+    translateBin=$(${callNixWithD2N} build --print-out-paths --no-link "
+      dream2nix.framework.translatorInstances.$translator.translateBin
+    ")
+  fi
 
   echo "
     {

src/subsystems/nodejs/translators/npm/default.nix

@@ -20,6 +20,7 @@
     moreutils,
     nodePackages,
     openssh,
+    python3,
     writeScriptBin,
     ...
   }:
@@ -32,6 +33,7 @@
       moreutils
       nodePackages.npm
       openssh
+      python3
     ]
     ''
       # accroding to the spec, the translator reads the input from a json file
@@ -42,6 +44,9 @@
       name=$(jq '.project.name' -c -r $jsonInput)
       version=$(jq '.project.version' -c -r $jsonInput)
       npmArgs=$(jq '.project.subsystemInfo.npmArgs' -c -r $jsonInput)
+      if [ "$npmArgs" == "null" ]; then
+        npmArgs=
+      fi
 
       if [ "$version" = "null" ]; then
         candidate="$name"
@@ -49,7 +54,6 @@
         candidate="$name@$version"
       fi
 
-
       pushd $TMPDIR
       newSource=$(pwd)
 
@@ -62,26 +66,14 @@
       # call package-lock translator
       ${subsystems.nodejs.translators.package-lock.translateBin} $TMPDIR/newJsonInput
 
-      # generate source info for main package
-      url=$(npm view $candidate dist.tarball)
-      hash=$(npm view $candidate dist.integrity)
-      echo "
-        {
-          \"type\": \"http\",
-          \"url\": \"$url\",
-          \"hash\": \"$hash\"
-        }
-      " > $TMPDIR/sourceInfo.json
+      # get resolved package version
+      export version=$(npm view $candidate version)
+
+      # set correct package version under `packages`
+      cat $outputFile \
+        | python3 ${./fixup-dream-lock.py} $TMPDIR/sourceInfo.json \
+        | sponge $outputFile
 
-      # add main package source info to dream-lock.json
-      ${apps.callNixWithD2N} eval --json "
-        with dream2nix.utils.dreamLock;
-        replaceRootSources {
-          dreamLock = l.fromJSON (l.readFile \"$outputFile\");
-          newSourceRoot = l.fromJSON (l.readFile \"$TMPDIR/sourceInfo.json\");
-        }
-      " \
-        | sponge "$outputFile"
     '';
 
   # inherit options from package-lock translator

src/subsystems/nodejs/translators/npm/fixup-dream-lock.py

@@ -0,0 +1,14 @@
+import json
+import os
+import sys
+
+lock = json.load(sys.stdin)
+version = os.environ.get('version')
+
+# set default package version correctly
+defaultPackage = lock['_generic']['defaultPackage']
+lock['_generic']['packages'] = {
+  defaultPackage: version
+}
+
+print(json.dumps(lock, indent=2))

src/utils/default.nix

@@ -34,7 +34,7 @@
 
   translatorUtils = callPackageDream ./translator.nix {};
 
-  indexUtils = callPackageDream ./index.nix {};
+  indexUtils = callPackageDream ./index {};
 
   poetry2nixSemver = import "${externalSources.poetry2nix}/semver.nix" {
     inherit lib;

src/utils/index/build-script.py

@@ -0,0 +1,59 @@
+import json
+import sys
+import subprocess as sp
+from pathlib import Path
+
+
+def store_error(attrPath, category, text, name=None):
+    with open(f"errors/{attrPath.replace('/', '--')}", 'w') as f:
+        json.dump(
+            dict(
+                attrPath=attrPath,
+                during=category,
+                error=text,
+                name=name,
+            ),
+            f,
+        )
+
+
+input = json.loads(sys.argv[1])
+attr = input['attr']
+attrPath = '.'.join(input['attrPath'])
+
+# handle eval error
+if "error" in input:
+    error = input['error']
+    print(
+        f"Evaluation failed. attr: {attr} attrPath: {attrPath}\n"
+        "Error:\n{error}",
+        file=sys.stderr
+    )
+    store_error(attrPath, 'eval', error)
+# try to build package
+else:
+    name = input['name']
+    drvPath = input['drvPath']
+    print(
+        f"Building {name} attr: {attr} attrPath: {attrPath} "
+        f"drvPath: ({drvPath})",
+        file=sys.stderr
+    )
+    try:
+        proc = sp.run(
+            ['nix', 'build', '-L', drvPath],
+            capture_output=True,
+            check=True,
+        )
+        print(
+            f"Finished {name}. attr: {attr} attrPath: {attrPath}",
+            file=sys.stderr
+        )
+    # handle build error
+    except sp.CalledProcessError as error:
+        Path('errors').mkdir(exist_ok=True)
+        print(
+            f"Error while building {name}. attr: {attr} attrPath: {attrPath}",
+            file=sys.stderr
+        )
+        store_error(attrPath, 'build', error.stderr.decode(), name)

src/utils/index/default.nix

@@ -80,6 +80,7 @@ in rec {
       '';
     in
       mkApp script;
+
     mkTranslateApp = name:
       mkApp (
         pkgs.writers.writeBash "translate-${name}" ''
@@ -88,27 +89,41 @@ in rec {
             ${name}/index.json ${name}/locks
         ''
       );
-    mkCiJobApp = name: input:
+
+    mkCiAppWith = commands:
       mkApp (
         utils.writePureShellScript
         (with pkgs; [
           coreutils
           git
           gnugrep
+          openssh
         ])
         ''
-          mainBranch=$(git branch | grep -E '(master)|(main)')
-          git branch data || :
-          git checkout data
+          flake=$(cat flake.nix)
+          flakeLock=$(cat flake.lock)
+          set -x
+          git fetch origin data || :
+          git checkout origin/data || :
+          git branch -D data || :
+          git checkout -b data
           # the flake should always be the one from the current main branch
-          git checkout $mainBranch flake.nix
-          git checkout $mainBranch flake.lock
-          ${(mkIndexApp name input).program}
-          ${(mkTranslateApp name).program}
+          rm -rf ./*
+          echo "$flake" > flake.nix
+          echo "$flakeLock" > flake.lock
+          ${commands}
           git add .
-          git commit "automatic update - $(date --rfc-3339=seconds)"
+          git commit -m "automatic update - $(date --rfc-3339=seconds)"
         ''
       );
+
+    mkCiJobApp = name: input:
+      mkCiAppWith
+      ''
+        ${(mkIndexApp name input).program}
+        ${(mkTranslateApp name).program}
+      '';
+
     translateApps = l.listToAttrs (
       l.map
       (
@@ -119,6 +134,7 @@ in rec {
       )
       indexNames
     );
+
     indexApps = l.listToAttrs (
       l.mapAttrsToList
       (
@@ -129,6 +145,7 @@ in rec {
       )
       indexes
     );
+
     ciJobApps = l.listToAttrs (
       l.mapAttrsToList
       (
@@ -140,6 +157,50 @@ in rec {
       indexes
     );
 
+    ciJobAllApp =
+      mkCiAppWith
+      ''
+        ${lib.concatStringsSep "\n" (l.mapAttrsToList (_: app: app.program) indexApps)}
+        ${lib.concatStringsSep "\n" (l.mapAttrsToList (_: app: app.program) translateApps)}
+      '';
+
+    buildAllApp = let
+      buildScript =
+        pkgs.writers.writePython3 "build-job" {}
+        ./build-script.py;
+      statsScript =
+        pkgs.writers.writePython3 "build-job" {}
+        ./make-stats.py;
+    in
+      mkApp (
+        utils.writePureShellScript
+        (with pkgs; [
+          coreutils
+          git
+          parallel
+          nix
+          nix-eval-jobs
+        ])
+        ''
+          rm -rf ./errors
+          mkdir -p ./errors
+          JOBS=''${JOBS:-$(nproc)}
+          EVAL_JOBS=''${EVAL_JOBS:-1}
+          LIMIT=''${LIMIT:-0}
+          if [ "$LIMIT" -gt "0" ]; then
+            limit="head -n $LIMIT"
+          else
+            limit="cat"
+          fi
+          echo "settings: JOBS $JOBS; EVAL_JOBS: $EVAL_JOBS; LIMIT $LIMIT"
+          parallel --halt now,fail=1 -j$JOBS --link \
+            -a <(nix-eval-jobs --gc-roots-dir $TMPDIR/gcroot --flake "$(realpath .)#packages.x86_64-linux" --workers $EVAL_JOBS --max-memory-size 3000 | $limit) \
+            ${buildScript}
+          ${statsScript}
+          rm -r ./errors
+        ''
+      );
+
     mkIndexOutputs = name: let
       src = "${toString source}/${name}/locks";
     in
@@ -167,7 +228,9 @@ in rec {
     outputs = {
       packages = allPackages;
       apps =
-        indexApps
+        {ci-job-all = ciJobAllApp;}
+        // {build-all = buildAllApp;}
+        // indexApps
         // translateApps
         // ciJobApps;
     };

src/utils/index/make-stats.py

@@ -0,0 +1,32 @@
+import json
+import os
+
+error_files = os.listdir('errors')
+eval_errors = 0
+build_errors = 0
+all_errors = {}
+
+for file in error_files:
+    with open(file) as f:
+        error = json.load(f)
+    # add error to all_errors
+    all_errors[error['attrPath']] = error
+    # count error types
+    if error['category'] == 'eval':
+        eval_errors += 1
+    else:
+        build_errors += 1
+
+num_errors = eval_errors + build_errors
+
+stats = dict(
+    errors=num_errors,
+    errors_eval=eval_errors,
+    errors_build=build_errors,
+)
+
+with open("errors.json", 'w') as f:
+    json.dump(all_errors, f)
+
+with open('stats.json', 'w') as f:
+    json.dump(stats, f)