feat: implement parser for type derivation programs

.github/workflows/misc.yml

@@ -28,6 +28,15 @@ jobs:
     - name: Check version update patch file
       run: python3 ci/version.py check
 
+  verify-antlr:
+    name: Verify ANTLR-generated code
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v2
+      - name: Check
+        working-directory: rs/antlr
+        run: python3 generate.py --ci
+
   commitlint:
     name: Lint commits for semantic-release
     runs-on: ubuntu-latest

proto/substrait/validator/validator.proto

@@ -73,6 +73,10 @@ message Node {
     // represents the parse result of the referred file.
     string resolved_uri = 9;
 
+    // This node represents an abstract syntax tree node, used for representing
+    // complex YAML string parse results.
+    google.protobuf.Empty ast_node = 10;
+
     // No longer used. The more generic ResolvedUri type is used instead.
     YamlReference yaml_reference = 5 [deprecated = true];
   }

rs/Cargo.toml

@@ -46,6 +46,9 @@ serde_json = "1"
 # being useful elsewhere too).
 regex = "1.5"
 
+# Used for the type derivation DSL.
+antlr-rust = "0.3.0-beta"
+
 # Used for checking URI syntax.
 uriparse = "0.6"
 

rs/antlr/.gitignore

@@ -0,0 +1,2 @@
+antlr.jar
+__pycache__

rs/antlr/README.md

@@ -0,0 +1,10 @@
+# ANTLR code generation logic
+
+The validator includes a parser for type expressions based on an ANTLR grammar.
+Unfortunately, the ANTLR code generator is written in Java, and would thus add
+a huge build dependency (a JRE) to the validator build environment. This is
+especially problematic for the distribution of Cargo crates, which are
+fundamentally source distributions that should not depend on anything other
+than other Rust crates. Therefore, the generated files are checked in to git
+and distributed with the crate, and regeneration must thus be done manually.
+Call the generate.py script to do so.

rs/antlr/antlr.py

@@ -0,0 +1,179 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: Apache-2.0
+
+"""Wrapper script to download and run a suitable version of ANTLR for
+generating or verifying the Rust bindings for a given grammar."""
+
+import urllib.request
+import os
+import sys
+import hashlib
+import logging
+import tempfile
+import shutil
+import filecmp
+import subprocess
+import difflib
+import argparse
+
+
+# NOTE: the Rust bindings for ANTLR are not (yet) official, so we need to
+# download a forked ANTLR build.
+ANTLR_URL = "https://github.com/rrevenantt/antlr4rust/releases/download/antlr4-4.8-2-Rust0.3.0-beta/antlr4-4.8-2-SNAPSHOT-complete.jar"
+ANTLR_SHA1 = "775d24ac1ad5df1eb0ed0e802f0fb2a5aeace43c"
+
+
+class Failure(Exception):
+    """Used for fatal errors."""
+
+
+def fail(msg):
+    """Logs and throws an error message."""
+    logging.error(msg)
+    raise Failure(msg)
+
+
+def download_file(fname, url):
+    """Downloads a file if it does not already exist."""
+    if not os.path.isfile(fname):
+        logging.info(f"Downloading {fname}...")
+        urllib.request.urlretrieve(ANTLR_URL, fname)
+
+
+def verify_file_hash(fname, hash_str):
+    """Verifies the hash of a (downloaded) file."""
+    logging.info(f"Verifying {fname}...")
+    with open(fname, "rb") as f:
+        file_hash = hashlib.sha1()
+        while chunk := f.read(8192):
+            file_hash.update(chunk)
+    actual = file_hash.hexdigest()
+    if hash_str != actual:
+        fail(f"Verification failed; hash should be {hash_str} but was {actual}")
+
+
+def verify_file_identical(new, old):
+    """Verifies that two text files are identical, printing a diff if not."""
+    logging.info(f"Verifying {new} against {old}...")
+    if not os.path.isfile(new):
+        fail(f"{new} does not exist")
+    if not os.path.isfile(old):
+        fail(f"{old} does not exist")
+    if not filecmp.cmp(new, old, shallow=False):
+        with open(new, "r") as f:
+            new_data = f.readlines()
+        with open(old, "r") as f:
+            old_data = f.readlines()
+        sys.stdout.writelines(difflib.unified_diff(old_data, new_data, old, new))
+        fail(f"{new} is different, see diff")
+
+
+def run_antlr(antlr, grammar, output_dir, verify=False, java="java"):
+    """Runns the given ANTLR JAR on the given grammar, sending outputs to
+    output_dir. If verify is set, instead of copying the newly-generated files,
+    this checks that there are no differences between the newly and previously
+    generated files."""
+    logging.info("Running ANTLR...")
+
+    # Determine the names of the generated files that we're interested in.
+    name = os.path.basename(grammar).split(".")[0].lower()
+    expected_files = [f"{name}lexer.rs", f"{name}parser.rs", f"{name}listener.rs"]
+
+    # Run in a temporary directory, because ANTLR spams random files we didn't
+    # ask for in its working directory.
+    with tempfile.TemporaryDirectory() as generate_dir:
+        shutil.copyfile(grammar, os.path.join(generate_dir, os.path.basename(grammar)))
+        subprocess.run(
+            [
+                java,
+                "-jar",
+                os.path.realpath(antlr),
+                "-Dlanguage=Rust",
+                os.path.basename(grammar),
+            ],
+            cwd=generate_dir,
+        )
+
+        logging.info("Copying/verifying output files...")
+        for expected_file in expected_files:
+            src = os.path.join(generate_dir, expected_file)
+            dest = os.path.join(output_dir, expected_file)
+            if not os.path.isfile(src):
+                fail(f"ANTLR failed to generate {expected_file}")
+            with open(src, "r+") as f:
+                data = f.read()
+                data = (
+                    "// SPDX-License-Identifier: Apache-2.0\n"
+                    "#![allow(clippy::all)]\n"
+                    "#![cfg_attr(rustfmt, rustfmt_skip)]\n"
+                    f"{data}"
+                )
+                f.seek(0)
+                f.write(data)
+            if verify:
+                verify_file_identical(src, dest)
+            else:
+                if os.path.exists(dest):
+                    os.unlink(dest)
+                shutil.copyfile(src, dest)
+
+
+def main(*args):
+    """Utility to generate Rust bindings for an ANTLR grammar."""
+    parser = argparse.ArgumentParser(description=main.__doc__)
+    parser.add_argument(
+        "--antlr",
+        metavar="antlr.jar",
+        default=os.path.join(os.path.dirname(os.path.realpath(__file__)), "antlr.jar"),
+        help="alternate location for the ANTLR jar",
+    )
+    parser.add_argument(
+        "--no-download",
+        action="store_true",
+        help="don't attempt to download the ANTLR jar",
+    )
+    parser.add_argument(
+        "--no-verify",
+        action="store_true",
+        help="don't attempt to verify the hash of the ANTLR jar",
+    )
+    parser.add_argument(
+        "--java", default="java", help="path to java executable to call ANTLR with"
+    )
+    parser.add_argument(
+        "--ci-check",
+        action="store_true",
+        help="instead of regenerating the files, assert that the files do not need to be regenerated",
+    )
+    parser.add_argument("grammar", help="the .g4 grammar file to generate")
+    parser.add_argument(
+        "dest_dir", default=".", nargs="?", help="where to copy the generated files to"
+    )
+    args = parser.parse_args(args)
+
+    logging.basicConfig(level=logging.INFO)
+
+    # Acquire ANTLR jar.
+    if args.no_download:
+        if not os.path.isfile(args.antlr):
+            parser.error(f"{args.antlr} does not exist and auto-download is disabled")
+    else:
+        download_file(args.antlr, ANTLR_URL)
+    if not args.no_verify:
+        verify_file_hash(args.antlr, ANTLR_SHA1)
+
+    # Run ANTLR.
+    if not os.path.isfile(args.grammar):
+        parser.error(f"{args.grammar} does not exist")
+    run_antlr(
+        args.antlr, args.grammar, args.dest_dir, verify=args.ci_check, java=args.java
+    )
+
+
+if __name__ == "__main__":
+    try:
+        main(*sys.argv[1:])
+        logging.info("Done")
+    except Failure:
+        logging.info("Returning failure exit status")
+        sys.exit(1)

rs/antlr/generate.py

@@ -0,0 +1,38 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: Apache-2.0
+
+import sys
+import logging
+import argparse
+import antlr
+
+"""Script for regenerating or verifying all the ANTLR-generated files of the
+validator."""
+
+
+def main(*args):
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument(
+        "--ci",
+        action="store_true",
+        help="instead of regenerating, verify that the files don't need to be regenerated",
+    )
+    args = parser.parse_args(args)
+
+    logging.basicConfig(level=logging.INFO)
+
+    ci = ["--ci-check"] if args.ci else []
+    antlr.main(
+        "../src/parse/extensions/simple/derivations/SubstraitType.g4",
+        "../src/parse/extensions/simple/derivations",
+        *ci,
+    )
+
+
+if __name__ == "__main__":
+    try:
+        main(*sys.argv[1:])
+        logging.info("Done")
+    except antlr.Failure:
+        logging.info("Returning failure exit status")
+        sys.exit(1)

rs/src/export/html/mod.rs

@@ -524,6 +524,7 @@ fn format_node_tree(
         tree::NodeType::YamlMap => format!("{brief} {}", format_span("type", "YAML map")),
         tree::NodeType::YamlArray => format!("{brief} {}", format_span("type", "YAML array")),
         tree::NodeType::YamlPrimitive(data) => format!("= {}{brief}", format_span("value", data)),
+        tree::NodeType::AstNode => format!("{brief} {}", format_span("type", "AST node")),
     };
     let header = format!(
         "{} {value} {}",

rs/src/export/proto.rs

@@ -188,6 +188,7 @@ impl From<&tree::NodeType> for validator::node::NodeType {
                 validator::node::NodeType::YamlPrimitive(data.into())
             }
             tree::NodeType::ResolvedUri(uri) => validator::node::NodeType::ResolvedUri(uri.clone()),
+            tree::NodeType::AstNode => validator::node::NodeType::AstNode(()),
         }
     }
 }