initial commit

script/file_sizes.txt

@@ -0,0 +1,5 @@
+src/grammar.json    	0.2MB	     11156
+src/node-types.json 	0.1MB	      7916
+src/parser.c        	41.8MB	   1311934
+src/scanner.c       	0.0MB	        37
+total               	42.2MB	   1331043

script/known_failures.txt

@@ -0,0 +1,3 @@
+examples/Newtonsoft.Json/Src/Newtonsoft.Json.Tests/Serialization/SerializationErrorHandlingTests.cs
+examples/Newtonsoft.Json/Src/Newtonsoft.Json.Tests/TestFixtureBase.cs
+examples/Newtonsoft.Json/Src/Newtonsoft.Json/JsonReader.cs

script/parse-examples

@@ -0,0 +1,45 @@
+#!/bin/bash
+
+set -eu
+
+cd "$(dirname "$0")/.."
+
+function clone_repo {
+  owner=$1
+  name=$2
+  sha=$3
+
+  path=examples/$name
+  if [ ! -d "$path" ]; then
+    echo "Cloning $owner/$name"
+    git clone "https://github.com/$owner/$name" "$path"
+  fi
+
+  pushd "$path" > /dev/null
+  actual_sha=$(git rev-parse HEAD)
+  if [ "$actual_sha" != "$sha"  ]; then
+    echo "Updating $owner/$name to $sha"
+    git fetch
+    git reset --hard $sha
+  fi
+  popd > /dev/null
+}
+
+clone_repo JamesNK Newtonsoft.Json 7c3d7f8da7e35dde8fa74188b0decff70f8f10e3
+clone_repo nunit nunit             ad49f27294bd0f2677d8699756c6ccb10df600f8
+clone_repo OrchardCMS orchard      0a82721968232b07354edcaac63a9ccea02220c6
+
+known_failures="$(cat script/known_failures.txt)"
+
+tree-sitter parse -q \
+  'examples/**/*.cs' \
+  $(for failure in $known_failures; do echo "!${failure}"; done)
+
+example_count=$(find examples -name '*.cs' | wc -l)
+failure_count=$(wc -w <<< "$known_failures")
+success_count=$(( $example_count - $failure_count ))
+success_percent=$(bc -l <<< "100*${success_count}/${example_count}")
+
+printf \
+  "Successfully parsed %d of %d example files (%.1f%%)\n" \
+  $success_count $example_count $success_percent

script/update-file-sizes

@@ -0,0 +1,3 @@
+#!/bin/bash
+
+wc -l -m src/*.* | awk '{printf("%-20s\t%0.1fMB\t%10s\n", $3, $2/1048576, $1)}' > script/file_sizes.txt

script/update-known-failures

@@ -0,0 +1,7 @@
+#!/bin/bash
+
+set -eu
+
+find examples -name '*.cs' -print0      | \
+  xargs -0 -n 1000 tree-sitter parse -q | \
+  tee >(cut -d' ' -f1 | sort > script/known-failures.txt)