RyanJson/scripts/tools/gen_rfc8259_embedded.py

#!/usr/bin/env python3
import os
import re
from pathlib import Path


def repo_root_from_script() -> Path:
    return Path(__file__).resolve().parents[2]


def escape_c_byte(b: int) -> str:
    if b == 0x22:
        return r"\""
    if b == 0x5C:
        return r"\\"
    if b == 0x0A:
        return r"\n"
    if b == 0x0D:
        return r"\r"
    if b == 0x09:
        return r"\t"
    if b == 0x08:
        return r"\b"
    if b == 0x0C:
        return r"\f"
    if 0x20 <= b <= 0x7E:
        return chr(b)
    return f"\\{b:03o}"


def format_c_string(data: bytes, indent: str = "    ", max_line_len: int = 120) -> str:
    if data is None:
        return indent + "\"\"\n"

    lines = []
    current = indent + "\""
    for b in data:
        chunk = escape_c_byte(b)
        if len(current) + len(chunk) >= max_line_len:
            current += "\"\n"
            lines.append(current)
            current = indent + "\"" + chunk
        else:
            current += chunk
    current += "\"\n"
    lines.append(current)
    return "".join(lines)


def main() -> None:
    repo_root = repo_root_from_script()
    data_dir = repo_root / "test" / "data" / "rfc8259"
    out_dir = repo_root / "test" / "unityTest" / "cases" / "RFC8259"
    out_c = out_dir / "rfc8259Embedded.c"
    out_h = out_dir / "rfc8259Embedded.h"

    names = sorted([p.name for p in data_dir.iterdir() if p.is_file() and p.name.endswith(".json")])
    if not names:
        raise SystemExit("No RFC8259 json files found.")

    out_dir.mkdir(parents=True, exist_ok=True)

    # Header
    header = []
    header.append("#ifndef RYANJSON_RFC8259_EMBEDDED_H\n")
    header.append("#define RYANJSON_RFC8259_EMBEDDED_H\n\n")
    header.append("#include <stdint.h>\n\n")
    header.append("typedef struct\n{")
    header.append("\tconst char *name;\n")
    header.append("\tconst unsigned char *data;\n")
    header.append("\tuint32_t len;\n")
    header.append("} rfc8259EmbeddedFile_t;\n\n")
    header.append("extern const rfc8259EmbeddedFile_t gRfc8259EmbeddedFiles[];\n")
    header.append("extern const uint32_t gRfc8259EmbeddedFileCount;\n\n")
    header.append("#endif // RYANJSON_RFC8259_EMBEDDED_H\n")
    out_h.write_text("".join(header), encoding="utf-8")

    # Source
    src = []
    src.append("// Auto-generated by scripts/tools/gen_rfc8259_embedded.py. Do not edit by hand.\n")
    src.append("#include \"rfc8259Embedded.h\"\n\n")

    entries = []
    for idx, name in enumerate(names):
        path = data_dir / name
        data = path.read_bytes()
        stem = Path(name).stem
        parts = [p for p in re.split(r"[^a-zA-Z0-9]+", stem) if p]
        if not parts:
            camel = f"case{idx:03d}"
        else:
            first = parts[0].lower()
            rest = "".join(p[:1].upper() + p[1:] for p in parts[1:])
            camel = first + rest
        var_name = f"rfc8259Embedded{idx:03d}{camel}Json"
        src.append(f"static const unsigned char {var_name}[] =\n")
        src.append(format_c_string(data))
        src.append(";\n\n")
        entries.append((name, var_name, len(data)))

    src.append("const rfc8259EmbeddedFile_t gRfc8259EmbeddedFiles[] = {\n")
    for name, var_name, length in entries:
        src.append(f"\t{{\"{name}\", {var_name}, {length}U}},\n")
    src.append("};\n\n")
    src.append("const uint32_t gRfc8259EmbeddedFileCount = ")
    src.append("(uint32_t)(sizeof(gRfc8259EmbeddedFiles) / sizeof(gRfc8259EmbeddedFiles[0]));\n")

    out_c.write_text("".join(src), encoding="utf-8")


if __name__ == "__main__":
    main()