uno-q-audio-synth/pd-validator/pd2json.py

#!/usr/bin/env python3
"""Parse a .pd file into JSON for Conftest/Rego validation."""
import json, sys, re
from pathlib import Path

def parse_pd(filepath):
    with open(filepath) as f:
        text = f.read()

    lines = text.split("\n")
    records = []
    x_index = 0
    errors = []

    for i, raw in enumerate(lines):
        line = raw.strip()
        if not line or line == ";":
            continue
        if not line.startswith("#"):
            errors.append(f"line {i+1}: expected # prefix")
            continue

        raw = line[1:].strip()
        if not raw.endswith(";"):
            errors.append(f"line {i+1}: missing terminating semicolon")
            continue
        body = raw[:-1].strip()

        if line.startswith("#N"):
            nbody = body[1:].strip()  # remove 'N'
            parts = nbody.split(None, 1)
            if parts[0] == "canvas":
                tokens = parts[1].split()
                rec = {
                    "type": "canvas",
                    "x": safe_int(tokens[0]),
                    "y": safe_int(tokens[1]),
                    "width": safe_int(tokens[2]),
                    "height": safe_int(tokens[3]),
                    "name": tokens[4] if len(tokens) > 4 else "",
                }
                records.append(rec)
            else:
                records.append({"type": "new", "data": rest})

        elif line.startswith("#A"):
            values = [safe_float(v) for v in rest.split()]
            records.append({"type": "array_data", "values": values})

        elif line.startswith("#X connect"):
            xbody = body[1:].strip()
            elems = xbody.split()
            if len(elems) >= 5:
                rec = {
                    "type": "connect",
                    "x_index": x_index,
                    "src": safe_int(elems[1]),
                    "outlet": safe_int(elems[2]),
                    "dst": safe_int(elems[3]),
                    "inlet": safe_int(elems[4]),
                }
                records.append(rec)
                # Connect records do NOT increment Pd's object index

        elif line.startswith("#X obj"):
            xbody = body[1:].strip()
            # xbody = "obj 20 20 notein"
            elem, x_str, y_str, rest = xbody.split(None, 3)
            obj_parts = rest.split()
            rec = {
                "type": "obj",
                "x_index": x_index,
                "x": safe_int(x_str),
                "y": safe_int(y_str),
                "name": obj_parts[0] if obj_parts else "",
                "args": obj_parts[1:],
            }
            records.append(rec)
            x_index += 1

        elif line.startswith("#X msg"):
            xbody = body[1:].strip()
            elem, x_str, y_str, content = xbody.split(None, 3)
            records.append({
                "type": "msg",
                "x_index": x_index,
                "x": safe_int(x_str),
                "y": safe_int(y_str),
                "content": content,
            })
            x_index += 1

        elif line.startswith("#X floatatom"):
            xbody = body[1:].strip()
            elems = xbody.split()
            rec = {"type": "floatatom", "x_index": x_index, "x": safe_int(elems[1]), "y": safe_int(elems[2])}
            if len(elems) > 3: rec["width"] = safe_int(elems[3])
            if len(elems) > 4: rec["min"] = safe_float(elems[4])
            if len(elems) > 5: rec["max"] = safe_float(elems[5])
            records.append(rec)
            x_index += 1

        elif line.startswith("#X symbolatom"):
            records.append({"type": "symbolatom", "x_index": x_index})
            x_index += 1

        elif line.startswith("#X text"):
            xbody = body[1:].strip()
            records.append({"type": "text", "x_index": x_index, "content": xbody})
            x_index += 1

        elif line.startswith("#X restore"):
            xbody = body[1:].strip()
            records.append({"type": "restore", "x_index": x_index, "data": xbody})
            x_index += 1

        elif line.startswith("#X coords"):
            xbody = body[1:].strip()
            records.append({"type": "coords", "x_index": x_index, "data": xbody})
            x_index += 1

        elif line.startswith("#X array"):
            xbody = body[1:].strip()
            records.append({"type": "array", "x_index": x_index, "data": xbody})
            x_index += 1

        else:
            errors.append(f"line {i+1}: unknown record type: {line[:60]}")

    return {
        "file": str(filepath),
        "x_record_count": x_index,
        "records": records,
        "parse_errors": errors,
    }

def safe_int(s):
    try:
        # Handle Pd special values like "-" for empty
        if s == "-":
            return 0
        return int(s)
    except (ValueError, IndexError):
        return 0

def safe_float(s):
    try:
        if s == "-":
            return 0.0
        return float(s)
    except (ValueError, IndexError):
        return 0.0

if __name__ == "__main__":
    if len(sys.argv) < 2:
        print("Usage: pd2json.py <file.pd> [file2.pd ...]")
        sys.exit(1)
    results = []
    for arg in sys.argv[1:]:
        results.append(parse_pd(arg))
    print(json.dumps(results if len(results) > 1 else results[0], indent=2))