Improve java install robustness (#13)

* Improve java install robustness

Done in a hurry, could use some refactoring.

* Make java_path a Path, not string

* Additional Path vs. string type fixes.

* Flake and typing fixes

* Bump version to 0.0.20

interface_gen/generate.py

@@ -46,7 +46,8 @@ def find_in_path(self, path: str | None = None) -> 'Schemas':
         self.schemas = list(start_path.rglob("*.avsc"))
         return self
 
-    def from_avro_idl(self, output_dir: Path) -> 'Schemas':
+    def from_avro_idl(self, env: dict[str, str], avro_cmd: list[str],
+                      output_dir: Path) -> 'Schemas':
         """ Generate all schemas from main Avro IDL files, and select all
             schemas for future operations on this object. """
         protocol_path = self.proto_dir
@@ -55,9 +56,8 @@ def from_avro_idl(self, output_dir: Path) -> 'Schemas':
             print(f"--> Generating schema(s) for {avdl}")
             version_dir = avdl.parent.name
             sdir = output_dir / version_dir / "schema"
-            script = toolchain.script_dir().parent / "avro" / "bin" / "avro-idl.sh"
-            subprocess.check_call([str(script), str(avdl), str(sdir)])
-        return self.find_in_path(output_dir)
+            subprocess.check_call(avro_cmd + [str(avdl), str(sdir)], env=env)
+        return self.find_in_path(str(output_dir))
 
     # --> Action methods
 
@@ -68,8 +68,9 @@ def gen_proto3(self):
             pb_dir.mkdir(parents=True, exist_ok=True)
             print(f"--> Generating proto3 for {str(schema_file)} in {pb_dir}")
             convert_avro_to_proto(schema_file, pb_dir)
-            # workaround: avrotize func. above always names file <namespace>.proto,
-            #  which causes all except the last schema to be overwritten. Rename that
+            # workaround: avrotize func. above always names file
+            # <namespace>.proto, which causes all except the last schema to be
+            # overwritten. Rename that
             #  output file here, until we can fix the avrotize lib.
             ns = namespace(schema_file)
             if ns:
@@ -87,20 +88,19 @@ def main():
 Output will be written to the output directory (-o): Avro schemas, proto3
 definitions, markdown documentation.
 """
-    parser = argparse.ArgumentParser(prog="generate.py",
-                                     description="Generate docs and definitions from Avro IDL.",
+    desc = "Generate docs and definitions from Avro IDL."
+    parser = argparse.ArgumentParser(prog="generate.py", description=desc,
                                      epilog=epilog)
     parser.add_argument('-p', '--protocol-dir',
                         help="Path w/ Avro IDL files in version subdirs")
     parser.add_argument('-o', '--output-dir', required=True,
                         help="Path where to generate markdown docs")
     parser.add_argument('-i', '--install-toolchain',
-                        help="Install toolchain dependencies if needed.",
+                        help="(deprecated: always attempts toolchain install)",
                         action="store_true")
     args = parser.parse_args()
 
-    if args.install_toolchain:
-        toolchain.install()
+    (env, avro_cmd) = toolchain.install()
 
     if args.protocol_dir:
         proto_dir = Path(args.protocol_dir)
@@ -113,7 +113,7 @@ def main():
         out_dir = proto_dir.parent / "generated"
 
     print("--> Generating Avro schemas..")
-    schemas = Schemas(proto_dir).from_avro_idl(out_dir)
+    schemas = Schemas(proto_dir).from_avro_idl(env, avro_cmd, out_dir)
     print(f"--> Found schemas: {schemas.schemas}")
 
     print("--> Generating proto3 definitions for all schemas")

interface_gen/toolchain.py

@@ -1,4 +1,5 @@
 import jdk
+import os
 from pathlib import Path
 import subprocess
 import shutil
@@ -20,17 +21,34 @@ def avro_jar(version: str) -> Path:
     return Path(f"avro-tools-{version}.jar")
 
 
-def ensure_jre():
-    """ Ensure that JRE is installed. """
+def find_java(root: Path) -> Path | None:
+    java_files = list(root.rglob("java"))
+    if len(java_files) == 0:
+        print(f"Java not found in {str(root)}")
+        return None
+    print(f"found java: {list(map(lambda p: str(p), java_files))}")
+    java_path = Path(java_files[0])
+    env = java_env(java_path)
+    subprocess.check_output([str(java_path), '-version'], env=env)
+    return java_path
+
+
+def ensure_jre() -> Path | None:
+    """ Ensure that JRE is installed, and return its path. """
+    jre_dir = Path.home() / '.jre'
     # See if jdk / jre already installedd
     jre = shutil.which('java')
+    if not jre:
+        jre = find_java(jre_dir)
     if jre:
         print(f"(Java is already installed: {jre})")
-        return
+        return Path(jre)
 
-    print("--> Installing OpenJDK 22 in $HOME/.jre/")
+    print(f"--> Installing OpenJDK 22 in {jre_dir}")
     jdk.install('22', jre=True)
 
+    return find_java(jre_dir)
+
 
 def install_avro_tools(target_dir: Path):
     global AVRO_VERSION
@@ -44,25 +62,36 @@ def install_avro_tools(target_dir: Path):
         download_file(url, target_dir)
 
 
-def create_avro_script(target_dir: Path):
-    target_dir.mkdir(parents=True, exist_ok=True)
-    script = target_dir / "avro-idl.sh"
-    print(f"--> Generating IDL->Schema script {script}")
-    with script.open("w") as f:
-        f.write(f"""\
-#!/bin/bash
+def java_env(java_path: Path) -> dict[str, str]:
+    """ Return environment with path, etc. set for java command """
+    bin_dir = java_path.parent
+    java_home = bin_dir.parent
+    lib_dir = java_home / 'lib'
+    new_path = os.environ['PATH'] + f":{bin_dir}"
+    env = os.environ.copy()
+    env['PATH'] = new_path
+    env['LD_LIBRARY_PATH'] = str(lib_dir)
+    return env
+
 
-# This script is generated by toolchain.py
-java -jar {target_dir}/{avro_jar(AVRO_VERSION)} idl2schemata "$@"
-""")
-    script.chmod(0o755)
+# TODO make this file a class that retains toolchain paths, etc. and provides a
+# method to run the command
+def create_avro_cmd(java_path: Path, target_dir: Path) \
+        -> tuple[dict[str, str], list[str]]:
+    env = java_env(java_path)
+    return (env, [str(java_path), '-jar',
+                  f'{target_dir}/{avro_jar(AVRO_VERSION)}',
+                  'idl2schemata'])
 
 
-def install():
-    """ Install the toolchain: tools needed for build and code generation. """
+def install() -> tuple[dict[str, str], list[str]]:
+    """ Install the toolchain and return the command needed to generate
+        schemas and code. """
     sdir = script_dir()
     avro_bin = sdir.parent / "avro" / "bin"
 
-    ensure_jre()
+    java_path = ensure_jre()
+    if not java_path:
+        raise Exception("Failed to install java.")
     install_avro_tools(avro_bin)
-    create_avro_script(avro_bin)
+    return create_avro_cmd(java_path, avro_bin)

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "interface_gen"
-version = "0.0.19"
+version = "0.0.20"
 authors = [{ name = "Aaron Fabbri", email = "[email protected]" }]
 description = "Define data format schemas once and generate code for multiple languages and serialization frameworks."
 readme = "dist-readme.md"