xml-sax-ts 0.4.0 → 0.5.0

package/dist/index.cjs

@@ -44,7 +44,6 @@ function decodeEntities(input, onError) {
     const semi = input.indexOf(";", amp + 1);
     if (semi === -1) {
       const err = new XmlSaxError("Unterminated entity", amp, 0, 0);
-      onError?.(err);
       throw err;
     }
     let decoded;
@@ -58,7 +57,6 @@ function decodeEntities(input, onError) {
     if (decoded === void 0) {
       const entity = input.slice(amp + 1, semi);
       const err = new XmlSaxError(`Unknown entity: &${entity};`, amp, 0, 0);
-      onError?.(err);
       throw err;
     }
     result += decoded;
@@ -169,12 +167,71 @@ function splitTextForEntities(text) {
   return { emit: text, carry: "" };
 }
 
+// src/tokens.ts
+var XmlToken = class {
+  constructor(kind, position) {
+    this.kind = kind;
+    this.position = position;
+  }
+};
+var OpenTagToken = class extends XmlToken {
+  constructor(tag, depth, path, position) {
+    super("open-tag", position);
+    this.tag = tag;
+    this.depth = depth;
+    this.path = path;
+  }
+};
+var CloseTagToken = class extends XmlToken {
+  constructor(tag, depth, path, position) {
+    super("close-tag", position);
+    this.tag = tag;
+    this.depth = depth;
+    this.path = path;
+  }
+};
+var TextToken = class extends XmlToken {
+  constructor(text, position) {
+    super("text", position);
+    this.text = text;
+  }
+};
+var CdataToken = class extends XmlToken {
+  constructor(text, position) {
+    super("cdata", position);
+    this.text = text;
+  }
+};
+var CommentToken = class extends XmlToken {
+  constructor(text, position) {
+    super("comment", position);
+    this.text = text;
+  }
+};
+var ProcessingInstructionToken = class extends XmlToken {
+  constructor(processingInstruction, position) {
+    super("processing-instruction", position);
+    this.processingInstruction = processingInstruction;
+  }
+};
+var DoctypeToken = class extends XmlToken {
+  constructor(doctype, position) {
+    super("doctype", position);
+    this.doctype = doctype;
+  }
+};
+var EndToken = class extends XmlToken {
+  constructor(position) {
+    super("end", position);
+  }
+};
+
 // src/parser.ts
 var DEFAULT_OPTIONS = {
   xmlns: true,
   includeNamespaceAttributes: false,
   allowDoctype: true,
-  coalesceText: false,
+  coalesceText: true,
   trackPosition: true
 };
 var XML_NAMESPACE_URI = "http://www.w3.org/XML/1998/namespace";
@@ -205,6 +262,8 @@ var XmlSaxParser = class {
     this.offset = 0;
     this.line = 1;
     this.column = 1;
+    this.pathStack = [];
+    this.tokenQueue = [];
     this.elementStack = [];
     this.nsStack = [
       Object.assign(/* @__PURE__ */ Object.create(null), {
@@ -222,28 +281,21 @@ var XmlSaxParser = class {
     this.allowDoctype = resolved.allowDoctype;
     this.coalesceText = resolved.coalesceText;
     this.trackPosition = resolved.trackPosition;
-    this.onOpenTag = resolved.onOpenTag;
-    this.onCloseTag = resolved.onCloseTag;
-    this.onText = resolved.onText;
-    this.onCdata = resolved.onCdata;
-    this.onComment = resolved.onComment;
-    this.onProcessingInstruction = resolved.onProcessingInstruction;
-    this.onDoctype = resolved.onDoctype;
-    this.onError = resolved.onError;
   }
   feed(chunk) {
     if (this.closed) {
       this._error("Parser is closed");
     }
     if (!chunk) {
-      return;
+      return this.drainTokens();
     }
     this.buffer += chunk;
     this._parseBuffer(false);
+    return this.drainTokens();
   }
   close() {
     if (this.closed) {
-      return;
+      return this.drainTokens();
     }
     this._parseBuffer(true);
     this._flushPendingCR();
@@ -255,6 +307,30 @@ var XmlSaxParser = class {
       this._error("Unclosed tag(s) remaining");
     }
     this.closed = true;
+    this._pushToken(new EndToken(this._position()));
+    return this.drainTokens();
+  }
+  drainTokens() {
+    if (this.tokenQueue.length === 0) {
+      return [];
+    }
+    return this.tokenQueue.splice(0, this.tokenQueue.length);
+  }
+  *[Symbol.iterator]() {
+    while (this.tokenQueue.length > 0) {
+      const token = this.tokenQueue.shift();
+      if (token) {
+        yield token;
+      }
+    }
+  }
+  async *iterateChunks(chunks) {
+    for await (const chunk of chunks) {
+      this.feed(chunk);
+      yield* this;
+    }
+    this.close();
+    yield* this;
   }
   _parseBuffer(final) {
     let i = 0;
@@ -317,7 +393,7 @@ var XmlSaxParser = class {
       const data = split === -1 ? "" : body.slice(split).trim();
       const pi = { target, body: data };
       this._flushTextBuffer();
-      this.onProcessingInstruction?.(pi);
+      this._pushToken(new ProcessingInstructionToken(pi, this._position()));
       return end + 2 - start;
     }
     if (secondCode === 33) {
@@ -332,7 +408,7 @@ var XmlSaxParser = class {
         }
         const comment = this.buffer.slice(start + 4, end);
         this._flushTextBuffer();
-        this.onComment?.(comment);
+        this._pushToken(new CommentToken(comment, this._position()));
         return end + 3 - start;
       }
       if (thirdCode === 91 && this.buffer.startsWith("<![CDATA[", start)) {
@@ -347,7 +423,7 @@ var XmlSaxParser = class {
         const normalized = this._normalizeText(cdata, false);
         if (normalized.length > 0) {
           this._flushTextBuffer();
-          this.onCdata?.(normalized);
+          this._pushToken(new CdataToken(normalized, this._position()));
         }
         return end + 3 - start;
       }
@@ -365,7 +441,7 @@ var XmlSaxParser = class {
         const raw = this.buffer.slice(start + 9, end).trim();
         const doctype = { raw };
         this._flushTextBuffer();
-        this.onDoctype?.(doctype);
+        this._pushToken(new DoctypeToken(doctype, this._position()));
         return end + 1 - start;
       }
     }
@@ -411,15 +487,18 @@ var XmlSaxParser = class {
         attributes: attributes2,
         isSelfClosing: selfClosing
       };
-      this.onOpenTag?.(tag2);
+      const openPath2 = Object.freeze([...this.pathStack, plainName]);
+      const depth2 = openPath2.length;
+      this._pushToken(new OpenTagToken(tag2, depth2, openPath2, this._position()));
       if (selfClosing) {
-        this.onCloseTag?.({ name: plainName });
+        this._pushToken(new CloseTagToken({ name: plainName }, depth2, openPath2, this._position()));
         return;
       }
       this.elementStack.push({
         rawName: parsed.name,
         closeTag: { name: plainName }
       });
+      this.pathStack.push(plainName);
       return;
     }
     const parentNs = this._currentNs();
@@ -462,14 +541,23 @@ var XmlSaxParser = class {
       attributes,
       isSelfClosing: selfClosing
     };
-    this.onOpenTag?.(tag);
+    const openPath = Object.freeze([...this.pathStack, resolvedName.name]);
+    const depth = openPath.length;
+    this._pushToken(new OpenTagToken(tag, depth, openPath, this._position()));
     if (selfClosing) {
-      this.onCloseTag?.({
-        name: resolvedName.name,
-        prefix: resolvedName.prefix,
-        local: resolvedName.local,
-        uri: resolvedName.uri
-      });
+      this._pushToken(
+        new CloseTagToken(
+          {
+            name: resolvedName.name,
+            prefix: resolvedName.prefix,
+            local: resolvedName.local,
+            uri: resolvedName.uri
+          },
+          depth,
+          openPath,
+          this._position()
+        )
+      );
       return;
     }
     this.elementStack.push({
@@ -481,6 +569,7 @@ var XmlSaxParser = class {
         uri: resolvedName.uri
       }
     });
+    this.pathStack.push(resolvedName.name);
     this.nsStack.push(ns);
   }
   _parseStartTagRange(start, end) {
@@ -524,7 +613,7 @@ var XmlSaxParser = class {
       }
       const rawValue = this.buffer.slice(i, valueEnd);
       const normalized = rawValue.includes("\r") ? rawValue.replace(CRLF_RE, "\n") : rawValue;
-      const value = !normalized.includes("&") ? normalized : decodeEntities(normalized, this.onError);
+      const value = !normalized.includes("&") ? normalized : decodeEntities(normalized);
       attributes.push({ name: attrName.name, value });
       i = valueEnd + 1;
     }
@@ -540,7 +629,12 @@ var XmlSaxParser = class {
     if (entry.rawName !== rawName) {
       this._error(`Mismatched closing tag: expected </${entry.rawName}>`);
     }
-    this.onCloseTag?.(entry.closeTag);
+    const closePath = Object.freeze([...this.pathStack]);
+    const depth = closePath.length;
+    if (depth > 0) {
+      this.pathStack.pop();
+    }
+    this._pushToken(new CloseTagToken(entry.closeTag, depth, closePath, this._position()));
   }
   _emitText(text, allowPendingCR) {
     const normalized = this._normalizeText(text, allowPendingCR);
@@ -551,14 +645,14 @@ var XmlSaxParser = class {
       this._emitDecodedText(normalized);
       return;
     }
-    const decoded = decodeEntities(normalized, this.onError);
+    const decoded = decodeEntities(normalized);
     if (decoded.length > 0) {
       this._emitDecodedText(decoded);
     }
   }
   _emitDecodedText(text) {
     if (!this.coalesceText) {
-      this.onText?.(text);
+      this._pushToken(new TextToken(text, this._position()));
       return;
     }
     this.pendingTextParts.push(text);
@@ -570,7 +664,7 @@ var XmlSaxParser = class {
     const first = this.pendingTextParts[0];
     const text = this.pendingTextParts.length === 1 && first !== void 0 ? first : this.pendingTextParts.join("");
     this.pendingTextParts.length = 0;
-    this.onText?.(text);
+    this._pushToken(new TextToken(text, this._position()));
   }
   _resolveName(rawName, ns) {
     if (!this.xmlns) {
@@ -807,55 +901,88 @@ var XmlSaxParser = class {
     const line = this.trackPosition ? this.line : 0;
     const column = this.trackPosition ? this.column : 0;
     const error = new XmlSaxError(message, this.offset, line, column);
-    this.onError?.(error);
     throw error;
   }
+  _position() {
+    return {
+      offset: this.offset,
+      line: this.trackPosition ? this.line : 0,
+      column: this.trackPosition ? this.column : 0
+    };
+  }
+  _pushToken(token) {
+    this.tokenQueue.push(token);
+  }
 };
+function tokenizeXml(xml, options = {}) {
+  const parser = new XmlSaxParser(options);
+  parser.feed(xml);
+  return parser.close();
+}
+async function* tokenizeXmlAsync(chunks, options = {}) {
+  const parser = new XmlSaxParser(options);
+  yield* parser.iterateChunks(chunks);
+}
 
 // src/tree.ts
 var TreeBuilder = class {
   constructor() {
     this.stack = [];
     this.root = null;
-    this.onOpenTag = (tag) => {
-      const node = {
-        name: tag.name,
-        attributes: Object.fromEntries(
-          Object.entries(tag.attributes).map(([key, attr]) => [key, typeof attr === "string" ? attr : attr.value])
-        ),
-        children: []
-      };
-      const parent = this.stack[this.stack.length - 1];
-      if (parent) {
-        parent.children?.push(node);
-      } else {
-        this.root = node;
-      }
-      this.stack.push(node);
-    };
-    this.onText = (text) => {
-      if (!this.stack.length) {
-        return;
-      }
-      const node = this.stack[this.stack.length - 1];
-      if (!node) {
-        return;
-      }
-      const children = node.children ?? [];
-      const last = children[children.length - 1];
-      if (typeof last === "string") {
-        children[children.length - 1] = last + text;
-      } else {
-        children.push(text);
-      }
-      node.children = children;
-    };
-    this.onCdata = (text) => {
-      this.onText(text);
-    };
-    this.onCloseTag = () => {
-      this.stack.pop();
+  }
+  onOpenTag(tag) {
+    const node = {
+      name: tag.name,
+      attributes: Object.fromEntries(
+        Object.entries(tag.attributes).map(([key, attr]) => [key, typeof attr === "string" ? attr : attr.value])
+      ),
+      children: []
     };
+    const parent = this.stack[this.stack.length - 1];
+    if (parent) {
+      parent.children?.push(node);
+    } else {
+      this.root = node;
+    }
+    this.stack.push(node);
+  }
+  onText(text) {
+    if (!this.stack.length) {
+      return;
+    }
+    const node = this.stack[this.stack.length - 1];
+    if (!node) {
+      return;
+    }
+    const children = node.children ?? [];
+    const last = children[children.length - 1];
+    if (typeof last === "string") {
+      children[children.length - 1] = last + text;
+    } else {
+      children.push(text);
+    }
+    node.children = children;
+  }
+  onCdata(text) {
+    this.onText(text);
+  }
+  onCloseTag() {
+    this.stack.pop();
+  }
+  consume(token) {
+    if (token instanceof OpenTagToken) {
+      this.onOpenTag(token.tag);
+      return;
+    }
+    if (token instanceof TextToken) {
+      this.onText(token.text);
+      return;
+    }
+    if (token instanceof CdataToken) {
+      this.onCdata(token.text);
+      return;
+    }
+    this.onCloseTag();
   }
   getRoot() {
     if (!this.root) {
@@ -866,15 +993,17 @@ var TreeBuilder = class {
 };
 function parseXmlString(xml, options = {}) {
   const builder = new TreeBuilder();
-  const parser = new XmlSaxParser({
-    ...options,
-    onOpenTag: builder.onOpenTag,
-    onText: builder.onText,
-    onCdata: builder.onCdata,
-    onCloseTag: builder.onCloseTag
-  });
-  parser.feed(xml);
-  parser.close();
+  const parser = new XmlSaxParser(options);
+  for (const token of parser.feed(xml)) {
+    if (token instanceof OpenTagToken || token instanceof TextToken || token instanceof CdataToken || token instanceof CloseTagToken) {
+      builder.consume(token);
+    }
+  }
+  for (const token of parser.close()) {
+    if (token instanceof OpenTagToken || token instanceof TextToken || token instanceof CdataToken || token instanceof CloseTagToken) {
+      builder.consume(token);
+    }
+  }
   return builder.getRoot();
 }
 
@@ -1000,47 +1129,62 @@ var ObjectBuilder = class {
     this.stack = [];
     this.root = null;
     this.rootName = null;
-    this.onOpenTag = (tag) => {
-      const name = normalizeName(tag.name, this.options);
-      const attributes = normalizeAttributes(tag.attributes, this.options);
-      const state = {
-        name,
-        attributes,
-        textParts: [],
-        children: /* @__PURE__ */ Object.create(null)
-      };
-      this.rootName ?? (this.rootName = name);
-      this.stack.push(state);
-    };
-    this.onText = (text) => {
-      if (!text) {
-        return;
-      }
-      const current = this.stack[this.stack.length - 1];
-      if (!current) {
-        return;
-      }
-      current.textParts.push(text);
-    };
-    this.onCdata = (text) => {
-      this.onText(text);
-    };
-    this.onCloseTag = () => {
-      const state = this.stack.pop();
-      if (!state) {
-        return;
-      }
-      const value = finalizeElement(state, this.options);
-      const parent = this.stack[this.stack.length - 1];
-      if (!parent) {
-        this.root = value;
-        return;
-      }
-      const path = this.stack.map((entry) => entry.name);
-      addChild(parent.children, state.name, value, this.options, path);
-    };
     this.options = buildSettings(options);
   }
+  onOpenTag(tag) {
+    const name = normalizeName(tag.name, this.options);
+    const attributes = normalizeAttributes(tag.attributes, this.options);
+    const state = {
+      name,
+      attributes,
+      textParts: [],
+      children: /* @__PURE__ */ Object.create(null)
+    };
+    this.rootName ?? (this.rootName = name);
+    this.stack.push(state);
+  }
+  onText(text) {
+    if (!text) {
+      return;
+    }
+    const current = this.stack[this.stack.length - 1];
+    if (!current) {
+      return;
+    }
+    current.textParts.push(text);
+  }
+  onCdata(text) {
+    this.onText(text);
+  }
+  onCloseTag() {
+    const state = this.stack.pop();
+    if (!state) {
+      return;
+    }
+    const value = finalizeElement(state, this.options);
+    const parent = this.stack[this.stack.length - 1];
+    if (!parent) {
+      this.root = value;
+      return;
+    }
+    const path = this.stack.map((entry) => entry.name);
+    addChild(parent.children, state.name, value, this.options, path);
+  }
+  consume(token) {
+    if (token instanceof OpenTagToken) {
+      this.onOpenTag(token.tag);
+      return;
+    }
+    if (token instanceof TextToken) {
+      this.onText(token.text);
+      return;
+    }
+    if (token instanceof CdataToken) {
+      this.onCdata(token.text);
+      return;
+    }
+    this.onCloseTag();
+  }
   getResult() {
     if (this.root === null) {
       throw new Error("No root element found");
@@ -1302,10 +1446,19 @@ function finalizeElement(state, options) {
   return result;
 }
 
+exports.CdataToken = CdataToken;
+exports.CloseTagToken = CloseTagToken;
+exports.CommentToken = CommentToken;
+exports.DoctypeToken = DoctypeToken;
+exports.EndToken = EndToken;
 exports.ObjectBuilder = ObjectBuilder;
+exports.OpenTagToken = OpenTagToken;
+exports.ProcessingInstructionToken = ProcessingInstructionToken;
+exports.TextToken = TextToken;
 exports.TreeBuilder = TreeBuilder;
 exports.XmlSaxError = XmlSaxError;
 exports.XmlSaxParser = XmlSaxParser;
+exports.XmlToken = XmlToken;
 exports.buildObject = buildObject;
 exports.buildXmlNode = buildXmlNode;
 exports.objectToXml = objectToXml;
@@ -1313,5 +1466,7 @@ exports.parseXmlString = parseXmlString;
 exports.resolveName = resolveName;
 exports.serializeXml = serializeXml;
 exports.stripNamespace = stripNamespace;
+exports.tokenizeXml = tokenizeXml;
+exports.tokenizeXmlAsync = tokenizeXmlAsync;
 //# sourceMappingURL=index.cjs.map
 //# sourceMappingURL=index.cjs.map