Spaces:
Running on Zero
Running on Zero
File size: 3,186 Bytes
cdc4405 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 | # Copyright (c) 2026 Scenema AI
# https://scenema.ai
# SPDX-License-Identifier: MIT
"""XML prompt validation for Scenema Audio.
Validates the <speak> XML format:
<speak voice="..." scene="..." language="...">
<action>delivery/stage direction</action>
Speech text here.
<action>more direction</action>
More speech text.
</speak>
Only <speak> root with <action> children allowed. All content is freeform.
"""
import xml.etree.ElementTree as ET
from dataclasses import dataclass, field
ALLOWED_CHILD_TAGS = {"action", "sound"}
@dataclass
class ValidationResult:
valid: bool
errors: list[str] = field(default_factory=list)
voice: str | None = None
scene: str | None = None
language: str | None = None
def validate_prompt(xml_string: str) -> ValidationResult:
"""Validate a Scenema Audio XML prompt.
Checks for valid XML structure, required <speak> root element,
required voice attribute, and only <action> child elements.
Args:
xml_string: Raw XML string to validate.
Returns:
ValidationResult with parsed attributes if valid,
or a list of errors if invalid.
"""
errors: list[str] = []
if not xml_string or not xml_string.strip():
return ValidationResult(valid=False, errors=["Prompt is empty"])
try:
root = ET.fromstring(xml_string)
except ET.ParseError as e:
return ValidationResult(valid=False, errors=[f"Invalid XML: {e}"])
if root.tag != "speak":
errors.append(f"Root element must be <speak>, got <{root.tag}>")
return ValidationResult(valid=False, errors=errors)
voice = root.get("voice")
if not voice or not voice.strip():
errors.append("Missing required 'voice' attribute on <speak>")
gender = root.get("gender")
if not gender or gender.strip() not in ("male", "female"):
errors.append(
"Missing or invalid 'gender' attribute on <speak>. Must be 'male' or 'female'"
)
scene = root.get("scene")
language = root.get("language", "en")
allowed_attrs = {"voice", "scene", "language", "gender", "shot"}
for attr in root.attrib:
if attr not in allowed_attrs:
errors.append(f"Unknown attribute '{attr}' on <speak>")
for child in root:
if child.tag not in ALLOWED_CHILD_TAGS:
errors.append(
f"Unsupported tag <{child.tag}>. Only <action> and <sound> are allowed inside <speak>"
)
if len(list(child)) > 0:
errors.append(f"<{child.tag}> must contain only text, no nested elements")
has_text = False
if root.text and root.text.strip():
has_text = True
for child in root:
if child.tail and child.tail.strip():
has_text = True
break
if not has_text:
errors.append("Prompt must contain at least one speech text node")
if errors:
return ValidationResult(valid=False, errors=errors)
return ValidationResult(
valid=True,
voice=voice.strip() if voice else None,
scene=scene.strip() if scene else None,
language=language.strip() if language else None,
)
|