src/manifest_parser.py

"""
APK Deep Diff Analyzer - AndroidManifest.xml Parser
"""
import xml.etree.ElementTree as ET
from typing import Dict, List, Any, Optional
import os


class ManifestAnalyzer:
    """Analyze AndroidManifest.xml content."""

    def __init__(self, manifest_path: str):
        self.manifest_path = manifest_path
        self.tree = None
        self.root = None
        self._parse()

    def _parse(self):
        """Parse the manifest XML."""
        try:
            self.tree = ET.parse(self.manifest_path)
            self.root = self.tree.getroot()
        except Exception:
            self.tree = None
            self.root = None

    def _get_attr(self, element: ET.Element, attr: str, namespace: str = "android") -> Optional[str]:
        """Get attribute value with namespace handling."""
        ns = "http://schemas.android.com/apk/res/android"
        val = element.get(f"{{{ns}}}{attr}")
        if val is None:
            val = element.get(attr)
        return val

    def get_package_info(self) -> Dict[str, Any]:
        """Get package and application info."""
        if self.root is None:
            return {}

        info = {
            "package": self.root.get("package", ""),
            "versionCode": self._get_attr(self.root, "versionCode"),
            "versionName": self._get_attr(self.root, "versionName"),
        }

        app = self.root.find("application")
        if app is not None:
            info["application"] = {
                "label": self._get_attr(app, "label"),
                "icon": self._get_attr(app, "icon"),
                "theme": self._get_attr(app, "theme"),
                "debuggable": self._get_attr(app, "debuggable"),
                "networkSecurityConfig": self._get_attr(app, "networkSecurityConfig"),
                "usesCleartextTraffic": self._get_attr(app, "usesCleartextTraffic"),
                "allowBackup": self._get_attr(app, "allowBackup"),
                "extractNativeLibs": self._get_attr(app, "extractNativeLibs"),
            }

        # SDK info
        uses_sdk = self.root.find("uses-sdk")
        if uses_sdk is not None:
            info["minSdkVersion"] = self._get_attr(uses_sdk, "minSdkVersion")
            info["targetSdkVersion"] = self._get_attr(uses_sdk, "targetSdkVersion")
            info["maxSdkVersion"] = self._get_attr(uses_sdk, "maxSdkVersion")

        return info

    def get_permissions(self) -> List[str]:
        """Get all permissions."""
        if self.root is None:
            return []

        permissions = []
        for perm in self.root.findall("uses-permission"):
            name = self._get_attr(perm, "name")
            if name:
                permissions.append(name)
        return sorted(set(permissions))

    def get_activities(self) -> List[Dict[str, Any]]:
        """Get all activities."""
        if self.root is None:
            return []

        activities = []
        app = self.root.find("application")
        if app is not None:
            for activity in app.findall("activity"):
                name = self._get_attr(activity, "name")
                if name:
                    acts = {
                        "name": name,
                        "exported": self._get_attr(activity, "exported"),
                        "permission": self._get_attr(activity, "permission"),
                        "intent_filters": [],
                    }
                    for intent in activity.findall("intent-filter"):
                        actions = [a.get("name", "") for a in intent.findall("action")]
                        categories = [c.get("name", "") for c in intent.findall("category")]
                        acts["intent_filters"].append({
                            "actions": actions,
                            "categories": categories,
                        })
                    activities.append(acts)
        return activities

    def get_services(self) -> List[Dict[str, Any]]:
        """Get all services."""
        if self.root is None:
            return []

        services = []
        app = self.root.find("application")
        if app is not None:
            for service in app.findall("service"):
                name = self._get_attr(service, "name")
                if name:
                    services.append({
                        "name": name,
                        "exported": self._get_attr(service, "exported"),
                        "permission": self._get_attr(service, "permission"),
                    })
        return services

    def get_receivers(self) -> List[Dict[str, Any]]:
        """Get all broadcast receivers."""
        if self.root is None:
            return []

        receivers = []
        app = self.root.find("application")
        if app is not None:
            for receiver in app.findall("receiver"):
                name = self._get_attr(receiver, "name")
                if name:
                    receivers.append({
                        "name": name,
                        "exported": self._get_attr(receiver, "exported"),
                        "permission": self._get_attr(receiver, "permission"),
                    })
        return receivers

    def get_providers(self) -> List[Dict[str, Any]]:
        """Get all content providers."""
        if self.root is None:
            return []

        providers = []
        app = self.root.find("application")
        if app is not None:
            for provider in app.findall("provider"):
                name = self._get_attr(provider, "name")
                if name:
                    providers.append({
                        "name": name,
                        "authorities": self._get_attr(provider, "authorities"),
                        "exported": self._get_attr(provider, "exported"),
                    })
        return providers

    def get_metadata(self) -> List[Dict[str, str]]:
        """Get application metadata."""
        if self.root is None:
            return []

        metadata = []
        app = self.root.find("application")
        if app is not None:
            for meta in app.findall("meta-data"):
                name = self._get_attr(meta, "name")
                value = self._get_attr(meta, "value")
                if name:
                    metadata.append({"name": name, "value": value or ""})
        return metadata

    def get_full_analysis(self) -> Dict[str, Any]:
        """Get complete manifest analysis."""
        return {
            "package_info": self.get_package_info(),
            "permissions": self.get_permissions(),
            "activities": self.get_activities(),
            "services": self.get_services(),
            "receivers": self.get_receivers(),
            "providers": self.get_providers(),
            "metadata": self.get_metadata(),
        }


def compare_manifests(orig_manifest: str, mod_manifest: str) -> Dict[str, Any]:
    """Compare two AndroidManifest.xml files."""
    orig = ManifestAnalyzer(orig_manifest)
    mod = ManifestAnalyzer(mod_manifest)

    orig_data = orig.get_full_analysis()
    mod_data = mod.get_full_analysis()

    result = {
        "original": orig_data,
        "modified": mod_data,
        "changes": {},
    }

    # Package info comparison
    orig_pkg = orig_data.get("package_info", {})
    mod_pkg = mod_data.get("package_info", {})
    pkg_changes = {}

    for key in set(list(orig_pkg.keys()) + list(mod_pkg.keys())):
        if key == "application":
            continue
        o_val = orig_pkg.get(key)
        m_val = mod_pkg.get(key)
        if o_val != m_val:
            pkg_changes[key] = {"original": o_val, "modified": m_val}

    # Application attributes
    orig_app = orig_pkg.get("application", {})
    mod_app = mod_pkg.get("application", {})
    app_changes = {}
    for key in set(list(orig_app.keys()) + list(mod_app.keys())):
        o_val = orig_app.get(key)
        m_val = mod_app.get(key)
        if o_val != m_val:
            app_changes[key] = {"original": o_val, "modified": m_val}

    if pkg_changes or app_changes:
        result["changes"]["package_info"] = {**pkg_changes, **app_changes}

    # Permissions
    orig_perms = set(orig_data.get("permissions", []))
    mod_perms = set(mod_data.get("permissions", []))
    result["changes"]["permissions"] = {
        "added": sorted(list(mod_perms - orig_perms)),
        "removed": sorted(list(orig_perms - mod_perms)),
        "unchanged": sorted(list(orig_perms & mod_perms)),
    }

    # Activities
    orig_acts = {a["name"]: a for a in orig_data.get("activities", [])}
    mod_acts = {a["name"]: a for a in mod_data.get("activities", [])}
    act_changes = {
        "added": [],
        "removed": [],
        "modified": [],
    }
    for name, data in mod_acts.items():
        if name not in orig_acts:
            act_changes["added"].append(data)
        elif data != orig_acts[name]:
            act_changes["modified"].append({
                "name": name,
                "original": orig_acts[name],
                "modified": data,
            })
    for name, data in orig_acts.items():
        if name not in mod_acts:
            act_changes["removed"].append(data)
    result["changes"]["activities"] = act_changes

    # Services
    orig_svcs = {s["name"]: s for s in orig_data.get("services", [])}
    mod_svcs = {s["name"]: s for s in mod_data.get("services", [])}
    svc_changes = {"added": [], "removed": [], "modified": []}
    for name, data in mod_svcs.items():
        if name not in orig_svcs:
            svc_changes["added"].append(data)
        elif data != orig_svcs[name]:
            svc_changes["modified"].append({"name": name, "original": orig_svcs[name], "modified": data})
    for name, data in orig_svcs.items():
        if name not in mod_svcs:
            svc_changes["removed"].append(data)
    result["changes"]["services"] = svc_changes

    # Receivers
    orig_recs = {r["name"]: r for r in orig_data.get("receivers", [])}
    mod_recs = {r["name"]: r for r in mod_data.get("receivers", [])}
    rec_changes = {"added": [], "removed": [], "modified": []}
    for name, data in mod_recs.items():
        if name not in orig_recs:
            rec_changes["added"].append(data)
        elif data != orig_recs[name]:
            rec_changes["modified"].append({"name": name, "original": orig_recs[name], "modified": data})
    for name, data in orig_recs.items():
        if name not in mod_recs:
            rec_changes["removed"].append(data)
    result["changes"]["receivers"] = rec_changes

    # Providers
    orig_provs = {p["name"]: p for p in orig_data.get("providers", [])}
    mod_provs = {p["name"]: p for p in mod_data.get("providers", [])}
    prov_changes = {"added": [], "removed": [], "modified": []}
    for name, data in mod_provs.items():
        if name not in orig_provs:
            prov_changes["added"].append(data)
        elif data != orig_provs[name]:
            prov_changes["modified"].append({"name": name, "original": orig_provs[name], "modified": data})
    for name, data in orig_provs.items():
        if name not in mod_provs:
            prov_changes["removed"].append(data)
    result["changes"]["providers"] = prov_changes

    # Metadata
    orig_meta = {m["name"]: m["value"] for m in orig_data.get("metadata", [])}
    mod_meta = {m["name"]: m["value"] for m in mod_data.get("metadata", [])}
    meta_changes = {
        "added": [{"name": k, "value": v} for k, v in mod_meta.items() if k not in orig_meta],
        "removed": [{"name": k, "value": v} for k, v in orig_meta.items() if k not in mod_meta],
        "modified": [{"name": k, "original": orig_meta[k], "modified": mod_meta[k]} 
                     for k in orig_meta if k in mod_meta and orig_meta[k] != mod_meta[k]],
    }
    result["changes"]["metadata"] = meta_changes

    return result