another.im-ios/AnotherXMPP/modules/parsing/XMLParser.swift

import Foundation
import libxml2

enum XMLParserError: Error {
    case xmlDeclarationInside(Int, Int)
    case xmlUnknown(Int)
}

enum XMLParserEvent {
    case streamStarted(attributes: [String: String])
    case streamEnded
    case element(XMLElement)
    case parserError(XMLParserError)
}

final class XMLParser {
    let (element, elementContinuation) = AsyncStream.makeStream(of: XMLParserEvent.self, bufferingPolicy: .unbounded)

    private var ctx: xmlParserCtxtPtr?
    private var stack: [XMLElement] = []
    private var xmlnss: [String: String] = [:]

    init() {
        ctx = xmlCreatePushParserCtxt(&saxHandler, Mem.bridge(self), nil, 0, nil)
    }

    deinit {
        elementContinuation.finish()
        xmlFreeParserCtxt(ctx)
        stack = []
        xmlnss = [:]
        ctx = nil
    }

    func restart() {
        stack = []
        xmlnss = [:]
        xmlFreeParserCtxt(ctx)
        ctx = xmlCreatePushParserCtxt(&saxHandler, Mem.bridge(self), nil, 0, nil)
    }

    func parse(data: Data) {
        data.withUnsafeBytes { [weak self] ptr in
            if let addr = ptr.baseAddress {
                let err = xmlParseChunk(self?.ctx, addr.assumingMemoryBound(to: CChar.self), Int32(data.count), 0)
                if err > 0 {
                    if err == 64 {
                        let rng = Data("<?xml ".utf8)
                        if let position = data.range(of: rng)?.lowerBound {
                            self?.elementContinuation.yield(.parserError(.xmlDeclarationInside(Int(err), position)))
                        }
                    }
                    self?.elementContinuation.yield(.parserError(.xmlUnknown(Int(err))))
                }
            }
        }
    }
}

// MARK: Private part, parsing
private extension XMLParser {
    func startElement(elementName: String, prefix: String?, namespaces: [String: String]?, attributes: [String: String]) {
        // for session start
        if elementName == "stream", prefix == "stream" {
            var attrs = attributes
            if let namespaces {
                for (key, value) in namespaces {
                    attrs[key] = value
                }
            }
            elementContinuation.yield(.streamStarted(attributes: attrs))
            return
        }

        // for others elements
        if let namespaces {
            for (key, value) in namespaces where !key.isEmpty {
                xmlnss[key] = value
            }
        }

        var xmlns: String?
        if let prefix {
            xmlns = xmlnss[prefix] ?? namespaces?[""]
        } else {
            xmlns = namespaces?[""]
        }

        var name: String
        if let prefix, xmlns == nil {
            name = prefix + ":" + elementName
        } else {
            name = elementName
        }

        var element = XMLElement(name: name, xmlns: xmlns, attributes: attributes, content: nil, nodes: [])
        if !stack.isEmpty {
            element = element.updateXmlns(stack.last?.xmlns)
        }
        if xmlns != nil {
            element = element.updateXmlns(xmlns)
        }

        stack.append(element)
    }

    func endElement(elementName: String, prefix: String?) {
        if elementName == "stream", prefix == "stream" {
            elementContinuation.yield(.streamEnded)
            return
        }
        if let element = stack.popLast() {
            if stack.isEmpty {
                elementContinuation.yield(.element(element))
            } else {
                // because we just checked it
                // swiftlint:disable:next force_unwrapping
                var parent = stack.popLast()!
                parent = parent.addNode(element)
                stack.append(parent)
            }
        }
    }

    func charactersFound(_ line: String) {
        guard var exists = stack.popLast() else { return }
        exists = exists.updateContent(line)
        stack.append(exists)
    }
}

// =========================================================
// libxml2-specific stuff
private struct Attr {
    var name: UnsafePointer<UInt8>?
    var prefix: UnsafePointer<UInt8>?
    var attrUri: UnsafePointer<UInt8>?
    var valueBegin: UnsafePointer<UInt8>?
    var valueEnd: UnsafePointer<UInt8>?
}

private struct Nss {
    var prefix: UnsafePointer<UInt8>?
    var uri: UnsafePointer<UInt8>?
}

private func strFromCUtf8(_ ptr: UnsafePointer<UInt8>?) -> String? {
    if let ptr {
        return String(cString: ptr)
    }
    return nil
}

private var saxHandler = xmlSAXHandler(
    internalSubset: nil,
    isStandalone: nil,
    hasInternalSubset: nil,
    hasExternalSubset: nil,
    resolveEntity: nil,
    getEntity: nil,
    entityDecl: nil,
    notationDecl: nil,
    attributeDecl: nil,
    elementDecl: nil,
    unparsedEntityDecl: nil,
    setDocumentLocator: nil,
    startDocument: nil,
    endDocument: nil,
    startElement: nil,
    endElement: nil,
    reference: nil,
    characters: SAX_charactersFound,
    ignorableWhitespace: nil,
    processingInstruction: nil,
    comment: nil,
    warning: nil,
    error: nil, // unsafeBitCast(SAX_error, to: errorSAXFunc.self),
    fatalError: nil,
    getParameterEntity: nil,
    cdataBlock: nil,
    externalSubset: nil,
    initialized: XML_SAX2_MAGIC,
    _private: nil,
    startElementNs: SAX_startElement,
    endElementNs: SAX_endElement,
    serror: nil
)

private let SAX_charactersFound: charactersSAXFunc = { ctx_, chars_, len_ in
    guard let ctx_, let chars_ else {
        return
    }
    let data = Data(bytes: UnsafePointer<UInt8>(chars_), count: Int(len_))
    let chars = String(decoding: data, as: UTF8.self)
    let parser = unsafeBitCast(ctx_, to: XMLParser.self)
    parser.charactersFound(chars)
}

private let SAX_startElement: startElementNsSAX2Func = { ctx_, localName, prefix_, _, nb_namespaces, namespaces_, nb_attributes, _, attributes_ in
    guard let name = strFromCUtf8(localName), let ctx_ else {
        return
    }
    let prefix = strFromCUtf8(prefix_)
    var attributes: [String: String] = [:]
    var indx = 0
    let parser = unsafeBitCast(ctx_, to: XMLParser.self)

    // attributes
    if let attributes_ {
        attributes_.withMemoryRebound(to: Attr.self, capacity: Int(nb_attributes)) {
            var attrsPtr = $0
            while indx < Int(nb_attributes) {
                if let name = strFromCUtf8(attrsPtr.pointee.name), let beginPtr = attrsPtr.pointee.valueBegin, let endPtr = attrsPtr.pointee.valueEnd {
                    let data = Data(bytes: UnsafePointer<UInt8>(beginPtr), count: endPtr - beginPtr)
                    var value = String(decoding: data, as: UTF8.self).unescaped
                    if let prefix = strFromCUtf8(attrsPtr.pointee.prefix) {
                        attributes[prefix + ":" + name] = value
                    } else {
                        attributes[name] = value
                    }
                }
                attrsPtr = attrsPtr.successor()
                indx += 1
            }
        }
    }

    // namespaces
    if nb_namespaces > 0, let namespaces_ {
        var namespaces: [String: String] = [:]
        namespaces_.withMemoryRebound(to: Nss.self, capacity: Int(nb_namespaces)) {
            var nsPtr = $0
            indx = 0
            while indx < Int(nb_namespaces) {
                let prefix = strFromCUtf8(nsPtr.pointee.prefix) ?? ""
                if var uri = strFromCUtf8(nsPtr.pointee.uri) {
                    uri = uri.unescaped
                    namespaces[prefix] = uri
                }
                nsPtr = nsPtr.successor()
                indx += 1
            }
        }
        parser.startElement(elementName: name, prefix: prefix, namespaces: namespaces, attributes: attributes)
    } else {
        parser.startElement(elementName: name, prefix: prefix, namespaces: nil, attributes: attributes)
    }
}

private let SAX_endElement: endElementNsSAX2Func = { ctx_, localName, prefix_, _ in
    guard let name = strFromCUtf8(localName), let ctx_ else {
        return
    }
    let prefix = strFromCUtf8(prefix_)
    let parser = unsafeBitCast(ctx_, to: XMLParser.self)
    parser.endElement(elementName: name, prefix: prefix)
}