[metadata] Implement encoder for purls (part 1)

maintainers/tools/purl_tables/main.py

@@ -0,0 +1,109 @@
+#!/usr/bin/env python3
+
+"""Tool for updating the tables for PURL."""
+
+import argparse
+import os
+import sys
+
+from typing import List
+from urllib.parse import quote
+
+
+def _percent_encode(s: str) -> str:
+    """
+    Percent-encodes a character according to the [PURL specification](https://github.com/package-url/purl-spec/blob/main/PURL-SPECIFICATION.rst#character-encoding).
+    """
+
+    try:
+        return quote(s, encoding=None).replace("%3A", ":")
+    except:
+        return None
+
+
+def _run(args) -> int:
+    lc_all = os.environ.get("LC_ALL")
+    if lc_all != "en_US.UTF-8":
+        print("Your environment settings will reorder the file badly.")
+        print("Please rerun as:")
+        print('  LC_ALL="en_US.UTF-8"', " ".join(sys.argv))
+        return 0
+
+    with open(args.output, "w") as f:
+        f.write("""# Generated by maintainers/tools/purl_tables/main.py. DO NOT EDIT.
+
+visibility([
+    "//purl/...",
+])
+
+""")
+        f.write("encode_byte = {\n")
+        for i in range(0, 256, 1):
+          f.write("""  %d: "%s",\n""" % (i, _percent_encode(bytes([i]))))
+        f.write("}\n")
+
+    percent_encoding_tests = [
+        "foo",
+        "Hello, World!",
+        "path: /foo",
+        # German
+        "München",
+        "Köln",
+        # Swedish
+        "Småland",
+        # French
+        "française",
+        # Spanish
+        "¡Hola Mundo!",
+        # Arabic
+        "مرحبا بالعالم!",
+        # Chinese
+        "你好世界！",
+        # Japanese
+        "こんにちは世界！",
+        # Emoji,
+        "🙎",
+        "🙊",
+        # Emoji with modifiers.
+        str(
+            b"\xf0\x9f\x99\x8e\xf0\x9f\x8f\xbe\xe2\x80\x8d\xe2\x99\x80\xef\xb8\x8f",
+            "utf8",
+        ),
+    ]
+
+    test_cases = {s: _percent_encode(s) for s in percent_encoding_tests}
+
+    if args.output.endswith(".bzl"):
+        test_data = args.output.replace(".bzl", "_test.bzl")
+        with open(test_data, "w") as f:
+            f.write("""# Generated by maintainers/tools/purl_tables/main.py. DO NOT EDIT.
+
+visibility([
+    "//purl/...",
+])
+""")
+            f.write("test_cases = {\n")
+            for s in percent_encoding_tests:
+                f.write("  \"%s\": \"%s\",\n" % (s, _percent_encode(s)))
+            f.write("}\n")
+
+    return 0
+
+
+def main(argv: List[str]) -> int:
+    """Main program.
+    Args:
+      argv: command-line arguments, such as sys.argv (including the program name
+      in argv[0]).
+    Returns:
+      Zero on successful program termination, non-zero otherwise.
+    """
+
+    parser = argparse.ArgumentParser(description="Update tables for PURL.")
+    parser.add_argument("--output", required=True, help="The .bzl file to write to.")
+
+    return _run(parser.parse_args(argv[1:]))
+
+
+if __name__ == "__main__":
+    sys.exit(main(sys.argv))

metadata/purl/BUILD

@@ -1,3 +1,6 @@
+load("//purl:percent_encoding_test.bzl", "percent_encoding_test")
+load("//purl/private:tables_test.bzl", "test_cases")
+
 exports_files(
     [
         "purl.bzl",
@@ -12,3 +15,8 @@ filegroup(
     ],
     visibility = ["//visibility:public"],
 )
+
+percent_encoding_test(
+    name = "percent_encoding_test",
+    cases = test_cases,
+)

metadata/purl/percent_encoding.bzl

@@ -0,0 +1,37 @@
+"""Utils for [purl](https://github.com/package-url/purl-spec)'s `percent encoding`.
+
+Spec: https://github.com/package-url/purl-spec/blob/main/PURL-SPECIFICATION.rst#character-encoding
+"""
+
+load("//purl:string.bzl", "string")
+load("//purl/private:tables.bzl", "encode_byte")
+
+visibility([
+    "//purl/...",
+])
+
+def _encode_byte(b):
+    """Encodes a single byte.
+
+    Args:
+      c: The byte to encode.
+    Returns:
+      The encoded string.
+    """
+
+    encoded = encode_byte.get(b, None)
+    if not encoded:
+        fail("Cannot encode {} (type={})".format(b, type(b)))
+
+    return encoded
+
+def percent_encode(value):
+    """Encodes the provided string.
+
+    Args:
+      value (string): The string to encode.
+    Returns:
+      The encoded string.
+    """
+
+    return "".join([_encode_byte(b) for b in string.to_bytes(value)])

metadata/purl/percent_encoding_test.bzl

@@ -0,0 +1,53 @@
+"""Rule for testing `percent_encode` and `percent_decode`."""
+
+load("//purl:percent_encoding.bzl", "percent_encode")
+
+visibility([
+    "//purl/...",
+])
+
+_bash_executable = """
+#!/usr/bin/env bash
+
+echo "All tests passed"
+""".strip()
+
+_bat_executable = """
+echo "Hello World"
+""".strip()
+
+def _percent_encoding_test_impl(ctx):
+    for decoded, encoded in ctx.attr.cases.items():
+        actual_encoded = percent_encode(decoded)
+        if encoded != actual_encoded:
+            fail("Error encoding {}: expected {}, got {}".format(decoded, encoded, actual_encoded))
+
+    # Unix does not care about the file extension, so always use `.bat` so it
+    # also works on Windows.
+    executable = ctx.actions.declare_file("{}.bat".format(ctx.attr.name))
+    ctx.actions.write(
+        output = executable,
+        content = _bash_executable if (ctx.configuration.host_path_separator == ":") else _bat_executable,
+        is_executable = True,
+    )
+
+    return [
+        DefaultInfo(
+            files = depset(
+                direct = [
+                    executable,
+                ],
+            ),
+            executable = executable,
+        ),
+    ]
+
+percent_encoding_test = rule(
+    implementation = _percent_encoding_test_impl,
+    attrs = {
+        "cases": attr.string_dict(
+            mandatory = True,
+        ),
+    },
+    test = True,
+)