Initial Commit

Sources/HRW/HTMLParser.swift

@@ -0,0 +1,210 @@
+//
+//  HTMLParser.swift
+//
+//
+//  Created by Isaac Paul on 7/31/24.
+//
+
+
+/*
+public class HTMLParser  {
+    func idk() {
+        
+    }
+    
+    static public func read(_ html:String) throws -> HTML {
+        guard var xmlReader = XMLParser(str: html) else { throw EmptyStringError() }
+        while let node = try xmlReader.readToken() {
+            print("\(node)")
+            if case XMLToken.tag_start(let name, let attributes) = node {
+                if (name == "html") {
+                    let document = try HTML(attributes, xmlReader)
+                }
+            }
+        }
+        throw EmptyStringError()
+    }
+}
+*/
+//TODO: Not really an html node..
+public class HTMLText : HTMLNode, IFlowContent {
+    
+    public var content: String
+    
+    public init(content: String) {
+        self.content = content
+        try! super.init( expectedAttributes: [:])
+    }
+
+    override public func toString(_ depth:Int = 0, spacingStrat:SpacingStrat = .tabs) -> (Int, String) {
+        return (depth, content)
+    }
+}
+
+public class XMLText : XMLNode {
+    
+    public var content: String
+    
+    public init(content: String) {
+        self.content = content
+        super.init()
+    }
+}
+
+extension XMLParser {
+    public func readObjects(endTag:String? = nil, validTags:[String]? = nil) throws -> [HTMLNode] {
+        var allItems:[HTMLNode] = []
+        while let obj = try self.readObject(endTag: endTag, validTags: validTags, xmlToHtmlMapper) {
+            allItems.append(obj)
+        }
+        return allItems
+    }
+    
+    public func readObject(endTag:String? = nil, validTags:[String]? = nil, _ mapper:(_ name:String, _ attributes:[String:String], _ parser:XMLParser?)  throws -> HTMLNode) throws -> HTMLNode? {
+        let tag = try readToken()
+        switch (tag) {
+            case let .data(unicode):
+                var result = ""
+                result.unicodeScalars.append(contentsOf: unicode)
+                return HTMLText(content: result)
+            case .instruction:
+                throw AppError("Unexpected Instruction")
+            case let .tag_empty(name, attributes):
+                if let validTags = validTags {
+                    if (validTags.firstIndex(of: name) == nil) {
+                        throw AppError("Invalid tag: \(name) for element \(endTag ?? "root")")
+                    }
+                }
+                let mapped = try mapper(name, attributes, nil)
+                return mapped
+            case let .tag_end(name):
+                if let tag = endTag,
+                    name != tag {
+                    throw AppError("Unexpected end tag: \(name) for element \(tag)")
+                }
+                return nil
+            case let .tag_start(name, attributes):
+                if let validTags = validTags {
+                    if (validTags.firstIndex(of: name) == nil) {
+                        throw AppError("Invalid tag: \(name) for element \(endTag ?? "root")")
+                    }
+                }
+                return try mapper(name, attributes, self)
+            case .none:
+                if (endTag == nil) {
+                    return nil
+                } else {
+                    throw AppError("Reached end of document before closing tag: \(endTag ?? "root")")
+                }
+        }
+    }
+    
+    public func readObjectXml(endTag:String? = nil, validTags:[String]? = nil) throws -> XMLNode? {
+        let tag = try readToken()
+        switch (tag) {
+            case let .data(unicode):
+                var result = ""
+                result.unicodeScalars.append(contentsOf: unicode)
+                return XMLText(content: result)
+            case .instruction:
+                throw AppError("Unexpected Instruction")
+            case let .tag_empty(name, attributes):
+                if let validTags = validTags {
+                    if (validTags.firstIndex(of: name) == nil) {
+                        throw AppError("Invalid tag: \(name) for element  \(endTag ?? "root")")
+                    }
+                }
+                return try GenericXMLNode(name, attributes, nil)
+            case let .tag_end(name):
+                if let tag = endTag,
+                    name != tag {
+                    throw AppError("Unexpected end tag: \(name) for element \(tag)")
+                }
+                return nil
+            case let .tag_start(name, attributes):
+                if let validTags = validTags {
+                    if (validTags.firstIndex(of: name) == nil) {
+                        throw AppError("Invalid tag: \(name) for element \(endTag ?? "root")")
+                    }
+                }
+                return try GenericXMLNode(name, attributes, self)
+            case .none:
+                if (endTag == nil) {
+                    return nil
+                } else {
+                    throw AppError("Reached end of document before closing tag: \(endTag ?? "root")")
+                }
+        }
+    }
+    
+    public func readDataAsString(endTag:String) throws -> String? {
+        let tag = try readToken()
+        switch (tag) {
+            case let .data(unicode):
+                var result = ""
+                result.unicodeScalars.append(contentsOf: unicode)
+                return result
+            case .instruction:
+                throw AppError("Unexpected Instruction")
+            case let .tag_empty(name, attributes):
+                throw AppError("Unexpected empty tag: \(name)")
+            case let .tag_end(name):
+                if (name != endTag) {
+                    throw AppError("Unexpected end tag: \(name)")
+                }
+                return nil
+            case let .tag_start(name, attributes):
+                throw AppError("Unexpected tag start: \(name)")
+            case .none:
+                throw AppError("Reached end of document before closing tag: \(endTag)")
+        }
+    }
+}
+
+/*
+public func xmlToHtmlMapper(_ name:String, _ attributes:[String:String], _ parser:XMLParser?) throws -> HTMLNode {
+    switch name {
+        case "head":
+            return try Head(attributes, parser)
+        case "body":
+            return try Body(attributes, parser)
+        case "meta":
+            if (attributes["itemprop"] != nil) {
+                return try MetaProp(attributes, parser)
+            } else {
+                return try Meta(attributes, parser)
+            }
+        case "noscript":
+            return try NoScriptHead(attributes, parser)
+        case "base":
+            return try Base(attributes, parser)
+        case "title":
+            return try Title(attributes, parser)
+        case "Style":
+            return try Title(attributes, parser)
+        case "script":
+            let src = attributes["src"]
+            let type = attributes["type"] ?? ""
+            switch (type) {
+                case "":
+                    if let _ = src {
+                        return try Script.ClassicRemote(attributes, parser)
+                    } else {
+                        return try Script.Classic(attributes, parser)
+                    }
+                case "module":
+                    if let _ = src {
+                        return try Script.ModuleRemote(attributes, parser)
+                    } else {
+                        return try Script.Module(attributes, parser)
+                    }
+                case "importmap":
+                    return try Script.ImportMap(attributes, parser)
+                default:
+                    return try Script.DataBlock(attributes, parser)
+            }
+        default:
+            throw AppError("Unknown or unhandled xml tag: \(name)")
+    }
+}
+*/