Android-pyApi/android_pyapi/org/xml_pull.py

"""
Java org.xmlpull.v1 APIs implemented in Python.
org.xmlpull.v1 package - XmlPullParser, XmlSerializer, XmlPullParserFactory.
Wraps Python xml.etree.ElementTree.
"""

from __future__ import annotations

from .._core import Stub, AndroidOnlyError
import xml.etree.ElementTree as _ET
from io import StringIO as _StringIO
from typing import List, Optional

__all__ = [
    "XmlPullParser", "XmlSerializer", "XmlPullParserFactory", "Factory",
]


# ============================================================
# XmlPullParser
# ============================================================

class XmlPullParser(Stub):
    """
    org.xmlpull.v1.XmlPullParser
    An XML pull parser interface.
    """
    # Event types
    START_TAG: int = 2
    END_TAG: int = 3
    TEXT: int = 4
    CDSECT: int = 5
    ENTITY_REF: int = 6
    IGNORABLE_WHITESPACE: int = 7
    PROCESSING_INSTRUCTION: int = 8
    COMMENT: int = 9
    DOCDECL: int = 10
    START_DOCUMENT: int = 0
    END_DOCUMENT: int = 1
    TYPES: List[str] = [
        "START_DOCUMENT", "END_DOCUMENT", "START_TAG", "END_TAG",
        "TEXT", "CDSECT", "ENTITY_REF", "IGNORABLE_WHITESPACE",
        "PROCESSING_INSTRUCTION", "COMMENT", "DOCDECL",
    ]
    NO_TYPES: List[str] = []
    # Features
    FEATURE_PROCESS_NAMESPACES: str = "http://xmlpull.org/v1/doc/features.html#process-ns"
    FEATURE_REPORT_NAMESPACE_ATTRIBUTES: str = "http://xmlpull.org/v1/doc/features.html#report-ns-attributes"
    FEATURE_PROCESS_DOCDECL: str = "http://xmlpull.org/v1/doc/features.html#process-docdecl"
    FEATURE_VALIDATION: str = "http://xmlpull.org/v1/doc/features.html#validation"

    class _Event:
        def __init__(self, type_=0, name="", text="", namespace="", prefix="",
                     attributes=None, depth=0, line=0, column=0):
            self.type = type_
            self.name = name
            self.text = text
            self.namespace = namespace
            self.prefix = prefix
            self.attributes = attributes or []
            self.depth = depth
            self.line = line
            self.column = column

    def __init__(self):
        super().__init__()
        self._events = []
        self._pos = -1
        self._depth = 0
        self._features = {
            self.FEATURE_PROCESS_NAMESPACES: True,
            self.FEATURE_REPORT_NAMESPACE_ATTRIBUTES: False,
            self.FEATURE_PROCESS_DOCDECL: False,
            self.FEATURE_VALIDATION: False,
        }
        self._properties = {}
        self._input_encoding = "UTF-8"
        self._namespace_stack = []

    def setInput(self, source: Any, input_encoding: str = None):
        """Set the input source for parsing."""
        if input_encoding:
            self._input_encoding = input_encoding

        if isinstance(source, str):
            # Parse directly from XML string
            self._parse_string(source)
        elif hasattr(source, 'read'):
            # File-like object
            data = source.read()
            if isinstance(data, bytes):
                data = data.decode(input_encoding or "UTF-8")
            self._parse_string(data)
        else:
            self._parse_string(str(source))

    def _parse_string(self, xml_str: str):
        """Parse XML string into a list of events."""
        self._events = []
        self._pos = -1

        try:
            root = _ET.fromstring(xml_str)
        except _ET.ParseError as e:
            # Create minimal events if parsing fails
            self._events.append(XmlPullParser._Event(XmlPullParser.START_DOCUMENT))
            self._events.append(XmlPullParser._Event(XmlPullParser.END_DOCUMENT))
            return

        self._events.append(XmlPullParser._Event(XmlPullParser.START_DOCUMENT))
        self._build_events(root, 0)
        self._events.append(XmlPullParser._Event(XmlPullParser.END_DOCUMENT))

    def _build_events(self, element: _ET.Element, depth: int):
        """Recursively build events from ElementTree."""
        attrs = []
        for k, v in element.attrib.items():
            ns = ""
            name = k
            prefix = ""
            if "}" in k:
                ns = k[1:k.index("}")]
                name = k[k.index("}") + 1:]
            attrs.append({"namespace": ns, "name": name, "prefix": prefix, "value": v})

        # START_TAG
        tag = element.tag
        ns = ""
        prefix = ""
        if "}" in tag:
            ns = tag[1:tag.index("}")]
            tag = tag[tag.index("}") + 1:]

        self._events.append(XmlPullParser._Event(
            XmlPullParser.START_TAG, tag, "", ns, prefix, attrs, depth))

        # Text content
        if element.text and element.text.strip():
            self._events.append(XmlPullParser._Event(
                XmlPullParser.TEXT, "", element.text, "", "", [], depth + 1))

        # Child elements
        for child in element:
            self._build_events(child, depth + 1)

        # END_TAG
        self._events.append(XmlPullParser._Event(
            XmlPullParser.END_TAG, tag, "", ns, prefix, [], depth))

        # Tail text
        if element.tail and element.tail.strip():
            self._events.append(XmlPullParser._Event(
                XmlPullParser.TEXT, "", element.tail, "", "", [], depth))

    # --- Event navigation ---

    def getEventType(self) -> int:
        if 0 <= self._pos < len(self._events):
            return self._events[self._pos].type
        return self.END_DOCUMENT

    def next(self) -> int:
        """Move to next event."""
        if self._pos < len(self._events) - 1:
            self._pos += 1
        return self.getEventType()

    def nextToken(self) -> int:
        """Move to next token (same as next for this impl)."""
        return self.next()

    def nextText(self) -> str:
        """Returns the text content of the current element."""
        if self.getEventType() != self.START_TAG:
            return ""
        self.next()
        text = self.getText()
        self.next()  # Move to END_TAG
        return text or ""

    def nextTag(self) -> int:
        """Move to next tag event."""
        self.next()
        while self.getEventType() not in (self.START_TAG, self.END_TAG, self.END_DOCUMENT):
            self.next()
        return self.getEventType()

    def require(self, type_: int, namespace: str, name: str):
        """Check that the current event matches expectations."""
        if self.getEventType() != type_:
            raise AndroidOnlyError(f"Expected event {type_}, got {self.getEventType()}")
        if namespace is not None and self.getNamespace() != namespace:
            raise AndroidOnlyError(f"Expected namespace '{namespace}', got '{self.getNamespace()}'")
        if name is not None and self.getName() != name:
            raise AndroidOnlyError(f"Expected name '{name}', got '{self.getName()}'")

    # --- Position info ---

    def getDepth(self) -> int:
        if 0 <= self._pos < len(self._events):
            return self._events[self._pos].depth
        return 0

    def getLineNumber(self) -> int:
        if 0 <= self._pos < len(self._events):
            return self._events[self._pos].line
        return -1

    def getColumnNumber(self) -> int:
        if 0 <= self._pos < len(self._events):
            return self._events[self._pos].column
        return -1

    def getInputEncoding(self) -> str:
        return self._input_encoding

    # --- Current event data ---

    def getName(self) -> str:
        if 0 <= self._pos < len(self._events):
            return self._events[self._pos].name
        return ""

    def getNamespace(self) -> str:
        if 0 <= self._pos < len(self._events):
            return self._events[self._pos].namespace
        return ""

    def getPrefix(self) -> str:
        if 0 <= self._pos < len(self._events):
            return self._events[self._pos].prefix
        return ""

    def getText(self) -> Optional[str]:
        if 0 <= self._pos < len(self._events):
            return self._events[self._pos].text
        return None

    def getTextCharacters(self, holder_for_start: List[int]) -> List[str]:
        """Returns text characters and fills start/length in holder."""
        text = self.getText() or ""
        if len(holder_for_start) >= 2:
            holder_for_start[0] = 0
            holder_for_start[1] = len(text)
        return list(text)

    # --- Attributes ---

    def getAttributeCount(self) -> int:
        if 0 <= self._pos < len(self._events):
            return len(self._events[self._pos].attributes)
        return 0

    def getAttributeName(self, index: int) -> str:
        if 0 <= self._pos < len(self._events):
            attrs = self._events[self._pos].attributes
            if 0 <= index < len(attrs):
                return attrs[index]["name"]
        return ""

    def getAttributeNamespace(self, index: int) -> str:
        if 0 <= self._pos < len(self._events):
            attrs = self._events[self._pos].attributes
            if 0 <= index < len(attrs):
                return attrs[index]["namespace"]
        return ""

    def getAttributePrefix(self, index: int) -> str:
        if 0 <= self._pos < len(self._events):
            attrs = self._events[self._pos].attributes
            if 0 <= index < len(attrs):
                return attrs[index]["prefix"]
        return ""

    def getAttributeType(self, index: int) -> str:
        return "CDATA"

    def getAttributeValue(self, index_or_name: Any, namespace: str = None) -> str:
        if isinstance(index_or_name, int):
            if 0 <= self._pos < len(self._events):
                attrs = self._events[self._pos].attributes
                if 0 <= index_or_name < len(attrs):
                    return attrs[index_or_name]["value"]
        else:
            # Lookup by name
            if 0 <= self._pos < len(self._events):
                attrs = self._events[self._pos].attributes
                for attr in attrs:
                    if attr["name"] == index_or_name:
                        if namespace is None or attr["namespace"] == namespace:
                            return attr["value"]
        return ""

    def isAttributeDefault(self, index: int) -> bool:
        return False

    # --- Element state ---

    def isEmptyElementTag(self) -> bool:
        """Returns True if the current event is an empty element tag."""
        if self.getEventType() == self.START_TAG:
            if self._pos + 1 < len(self._events):
                next_event = self._events[self._pos + 1]
                return next_event.type == self.END_TAG and next_event.name == self.getName()
        return False

    def isWhitespace(self) -> bool:
        text = self.getText() or ""
        return text.strip() == ""

    # --- Namespace support ---

    def getNamespaceCount(self, depth: int) -> int:
        return len(self._namespace_stack)

    def getNamespacePrefix(self, index: int) -> str:
        if 0 <= index < len(self._namespace_stack):
            return self._namespace_stack[index][0]
        return ""

    def getNamespaceUri(self, prefix: str) -> str:
        for p, uri in self._namespace_stack:
            if p == prefix:
                return uri
        return ""

    # --- Features / Properties ---

    def getFeature(self, name: str) -> bool:
        return self._features.get(name, False)

    def setFeature(self, name: str, value: bool):
        if name in self._features:
            self._features[name] = value
        else:
            raise AndroidOnlyError(f"Unknown feature: {name}")

    def getProperty(self, name: str):
        return self._properties.get(name)

    def setProperty(self, name: str, value: Any):
        self._properties[name] = value

    def toString(self) -> str:
        return f"XmlPullParser@{hex(id(self))}[event={self.getEventType()}]"

    def equals(self, other) -> bool:
        return self is other

    def hashCode(self) -> int:
        return hash(self) & 0x7FFFFFFF


# ============================================================
# XmlSerializer
# ============================================================

class XmlSerializer(Stub):
    """
    org.xmlpull.v1.XmlSerializer
    Interface for serializing XML documents.
    """

    def __init__(self):
        super().__init__()
        self._output = []
        self._depth = 0
        self._features = {}
        self._properties = {}
        self._prefixes = {}
        self._output_stream = None
        self._encoding = "UTF-8"

    def setOutput(self, output_stream, encoding: str = None):
        """Sets the output writer/stream."""
        self._output_stream = output_stream
        self._encoding = encoding or "UTF-8"

    def getFeature(self, name: str) -> bool:
        return self._features.get(name, False)

    def setFeature(self, name: str, value: bool):
        self._features[name] = value

    def getProperty(self, name: str):
        return self._properties.get(name)

    def setProperty(self, name: str, value: Any):
        self._properties[name] = value

    def setPrefix(self, prefix: str, namespace_uri: str):
        """Sets the prefix for a namespace."""
        self._prefixes[namespace_uri] = prefix
        if prefix:
            self._write(f' xmlns:{prefix}="{namespace_uri}"')
        else:
            self._write(f' xmlns="{namespace_uri}"')

    def startTag(self, namespace: str, name: str) -> "XmlSerializer":
        """Writes a start tag."""
        self._write(f"<{name}")
        self._depth += 1
        return self

    def attribute(self, namespace: str, name: str, value: str) -> "XmlSerializer":
        """Writes an attribute."""
        if value is not None:
            escaped = value.replace("&", "&amp;").replace("<", "&lt;").replace(">", "&gt;").replace('"', "&quot;")
            self._write(f' {name}="{escaped}"')
        return self

    def endTag(self, namespace: str, name: str) -> "XmlSerializer":
        """Writes an end tag."""
        self._depth -= 1
        self._write(f"</{name}>")
        return self

    def text(self, text_or_buf: Any, start: int = None, length: int = None) -> "XmlSerializer":
        """Writes text content."""
        if isinstance(text_or_buf, str):
            escaped = text_or_buf.replace("&", "&amp;").replace("<", "&lt;").replace(">", "&gt;")
            self._write(escaped)
        elif isinstance(text_or_buf, (list, tuple)) and start is not None and length is not None:
            chars = text_or_buf[start:start + length]
            self._write("".join(chars))
        return self

    def cdsect(self, text: str) -> "XmlSerializer":
        """Writes a CDATA section."""
        self._write(f"<![CDATA[{text}]]>")
        return self

    def entityRef(self, name: str) -> "XmlSerializer":
        """Writes an entity reference."""
        self._write(f"&{name};")
        return self

    def processingInstruction(self, text: str) -> "XmlSerializer":
        """Writes a processing instruction."""
        self._write(f"<?{text}?>")
        return self

    def comment(self, text: str) -> "XmlSerializer":
        """Writes a comment."""
        self._write(f"<!--{text}-->")
        return self

    def docdecl(self, text: str) -> "XmlSerializer":
        """Writes a document declaration."""
        self._write(f"<!{text}>")
        return self

    def startDocument(self, encoding: str = None, standalone: bool = None):
        """Writes the XML declaration."""
        encoding = encoding or self._encoding
        decl = f'<?xml version="1.0" encoding="{encoding}"'
        if standalone is not None:
            decl += f' standalone="{"yes" if standalone else "no"}"'
        decl += "?>"
        self._write(decl)

    def endDocument(self):
        """Finishes the document."""
        self.flush()

    def flush(self):
        """Flushes the output."""
        content = "".join(self._output)
        if self._output_stream:
            if hasattr(self._output_stream, 'write'):
                self._output_stream.write(content)
            self._output = []

    def getDepth(self) -> int:
        return self._depth

    def getName(self) -> str:
        return ""

    def getNamespace(self) -> str:
        return ""

    def getPrefix(self, namespace: str, prefix: bool = False) -> str:
        return self._prefixes.get(namespace, "")

    def _write(self, s: str):
        self._output.append(s)

    def verbatim(self, text: str) -> "XmlSerializer":
        """Writes raw text without escaping."""
        self._write(text)
        return self

    def write(self, text: str) -> "XmlSerializer":
        """Writes text."""
        return self.text(text)

    def toString(self) -> str:
        return f"XmlSerializer@{hex(id(self))}"

    def equals(self, other) -> bool:
        return self is other

    def hashCode(self) -> int:
        return hash(self) & 0x7FFFFFFF


# ============================================================
# XmlPullParserFactory
# ============================================================

class XmlPullParserFactory(Stub):
    """
    org.xmlpull.v1.XmlPullParserFactory
    Factory for creating XmlPullParser and XmlSerializer instances.
    """

    def __init__(self):
        super().__init__()
        self._features = {}
        self._properties = {}

    @staticmethod
    def newInstance(unused: Any = None, unused_cls: Any = None) -> "XmlPullParserFactory":
        return XmlPullParserFactory()

    def setFeature(self, name: str, value: bool):
        self._features[name] = value

    def getFeature(self, name: str) -> bool:
        return self._features.get(name, False)

    def setProperty(self, name: str, value: Any):
        self._properties[name] = value

    def getProperty(self, name: str):
        return self._properties.get(name)

    def newPullParser(self) -> XmlPullParser:
        """Creates a new XmlPullParser."""
        return XmlPullParser()

    def newSerializer(self) -> XmlSerializer:
        """Creates a new XmlSerializer."""
        return XmlSerializer()

    def isNamespaceAware(self) -> bool:
        return self.getFeature(XmlPullParser.FEATURE_PROCESS_NAMESPACES)

    def isValidating(self) -> bool:
        return self.getFeature(XmlPullParser.FEATURE_VALIDATION)

    def toString(self) -> str:
        return f"XmlPullParserFactory@{hex(id(self))}"

    def equals(self, other) -> bool:
        return self is other

    def hashCode(self) -> int:
        return hash(self) & 0x7FFFFFFF


# ============================================================
# Factory (alias)
# ============================================================

class Factory(XmlPullParserFactory):
    """
    Factory alias for XmlPullParserFactory.
    """
    def __init__(self):
        super().__init__()

    @staticmethod
    def newInstance(unused: Any = None, unused_cls: Any = None) -> "Factory":
        return Factory()

    @staticmethod
    def getInstance() -> "Factory":
        return Factory()

    def toString(self) -> str:
        return f"Factory@{hex(id(self))}"

    def equals(self, other) -> bool:
        return isinstance(other, Factory)

    def hashCode(self) -> int:
        return hash("Factory") & 0x7FFFFFFF