rewrite using Saxophone parser

please note that TextDecoderStream needs to be used on binary streams this version is using web streams instead of node streams using TransformStream allows direct usage in web browser using in node is also possible since node's pipe accepts web streams
pirxpilot · Sep 15, 2023 · a4cbc8d · a4cbc8d
1 parent 8674624
commit a4cbc8d
Show file tree

Hide file tree

Showing 5 changed files with 330 additions and 324 deletions.
diff --git a/Makefile b/Makefile
@@ -4,6 +4,6 @@ lint:
 	./node_modules/.bin/jshint *.js lib test
 
 test:
-	node --require should --test
+	node --test
 
 .PHONY: check lint test
diff --git a/lib/stream.js b/lib/stream.js
@@ -1,52 +1,45 @@
-const { Transform } = require('stream');
-const { StringDecoder } = require('string_decoder');
-const sax = require('sax');
+const Saxophone = require('@pirxpilot/saxophone');
+const stack = require('./stack');
+
 const debug = require('debug')('sax-super-stream');
 
-const stack = require('./stack');
+/* global TransformStream, EventTarget */
+
 
 module.exports = initParser;
 
 // used to mark all tags that we want to skip
-const IGNORE = Object.create(null);
+const IGNORE = Symbol('ignore');
 
 function stripPrefix(name) {
-  const index = name.indexOf(':');
-  return index < 0 ? name : name.slice(index + 1);
+  const prefixed = name.split(':', 2);
+  return prefixed[1] ? prefixed : [undefined, prefixed[0]];
 }
 
-function handlers(parserConfig, fn) {
+function makeHandlers(parserConfig, emit) {
+  const ns = Object.create(null);
+  let defaultNS;
+  const waiting = [];
   const items = stack();
   const parsers = stack(parserConfig);
   const context = {};
-  let cdata;
 
-  function onopentag(node) {
-    const tag = node.local;
-    const tagParser = verifyNS(parsers.top()[tag]);
-    let elem;
+  const target = new EventTarget();
+  target.addEventListener('tagopen', ontagopen);
+  target.addEventListener('tagclose', ontagclose);
+  target.addEventListener('text', ontext);
+  target.addEventListener('cdata', ontext);
+  return target;
 
-    function parseWith(tp) {
-      elem = tp.call(items, node, items.top(), context);
-      if (elem) {
-        items.push(elem, tag);
-      }
-    }
+  function ontagopen({ detail: { name, attrs, isSelfClosing } }) {
+    const [prefix, tag] = stripPrefix(name);
+    const attrsObj = Saxophone.parseAttrs(attrs);
+    updateNamespaces(attrsObj);
+    const uri = (prefix && ns[prefix]) ?? defaultNS;
 
-    // if parser specifies namespace, it has to much naode namespace
-    function verifyNS(tp) {
-      if (!tp) {
-        return tp;
-      }
-      if (!tp.$uri) {
-        return tp;
-      }
-      if (tp.$uri === node.uri) {
-        return tp;
-      }
-    }
+    const tagParser = getParser(tag, uri);
 
-    debug('onopentag', tag);
+    debug('tagopen', tag, attrs);
     if (!tagParser) {
       parsers.push(IGNORE, tag);
       return;
@@ -59,136 +52,111 @@ function handlers(parserConfig, fn) {
       }
       parsers.push(tagParser, tag);
     }
-  }
+    if (isSelfClosing) {
+      ontagclose({ detail: { name } });
+    }
 
-  function onclosetag(tag) {
-    let parser;
-    let top;
+    function parseWith(tp) {
+      const attributes = Object.fromEntries(
+        Object.entries(attrsObj).map(
+          ([name, value]) => [
+            name,
+            {
+              name,
+              value: Saxophone.parseEntities(value)
+            }
+          ]
+        )
+      );
+      const node = { prefix, tag, attributes };
+      const elem = tp.call(items, node, items.top(), context);
+      if (elem) {
+        items.push(elem, tag);
+      }
+    }
+  }
 
-    tag = stripPrefix(tag);
+  function ontagclose({ detail: { name } }) {
+    const [, tag] = stripPrefix(name);
 
-    debug('closetag', tag);
+    debug('tagclose', tag);
 
-    parser = parsers.pop(tag);
-    if (parser && typeof parser.$after === 'function') {
+    const parser = parsers.pop(tag);
+    if (typeof parser?.$after === 'function') {
+      debug('$after', tag);
       parser.$after.call(items, items.top(), context);
     }
     if (parser !== IGNORE) {
-      top = items.pop(tag);
+      const top = items.pop(tag);
       // if nothing on the stack emit result
       if (top !== undefined && items.empty()) {
-        fn(null, top);
+        emit(top);
       }
     }
   }
 
-  function ontext(value) {
-    const textParser = parsers.top().$text;
-    if (textParser) {
-      textParser.call(items, value, items.top(), context);
-    }
-  }
+  function ontext({ detail: { contents } }) {
+    debug('text', contents);
 
-  function onopencdata() {
     const textParser = parsers.top().$text;
-    if (textParser) {
-      cdata = [];
-    }
+    textParser?.call(items, Saxophone.parseEntities(contents), items.top(), context);
   }
 
-  function oncdata(value) {
-    if (cdata) {
-      cdata.push(value);
+  function updateNamespaces(attrsObj) {
+    Object.entries(attrsObj).forEach(xmlns);
+
+    function xmlns([name, value]) {
+      if (name === 'xmlns') {
+        defaultNS = value;
+      } else if (name.startsWith('xmlns:')) {
+        const prefix = name.slice(6); // 'xmlns:'.length === 6
+        ns[prefix] = value;
+      }
     }
   }
 
-  function onclosecdata() {
-    if (!cdata) {
+  function getParser(tag, uri) {
+    const top = parsers.top();
+    const tp = top[tag];
+    if (!tp) {
       return;
     }
-    const textParser = parsers.top().$text;
-    textParser.call(items, cdata.join(''), items.top(), context);
-    cdata = undefined;
-  }
-
-  function onerror(err) {
-    debug('Detected error', err);
-    // mark error as handled
-    this.error = null;
-    fn(err);
+    if (!tp.$uri) {
+      return tp;
+    }
+    // if parser specifies namespace, it has to match node namespace
+    if (tp.$uri === uri) {
+      return tp;
+    }
   }
-
-  return {
-    onopentag,
-    onclosetag,
-    ontext,
-    onopencdata,
-    oncdata,
-    onclosecdata,
-    onerror
-  };
 }
 
 
-function initParser(parserConfig, saxOptions) {
+function initParser(config) {
 
-  saxOptions = Object.assign({
-    trim: true,
-    normalize: true,
-    lowercase: false,
-    xmlns: true,
-    position: false,
-    strictEntities: true,
-    noscript: true
-  }, saxOptions);
+  const results = [];
 
-  const parser = sax.parser(true, saxOptions);
-  const decoder = new StringDecoder('utf8');
-  let results = [];
-  let parserError;
+  const target = makeHandlers(config, (...objs) => results.push(...objs));
+  const sax = new Saxophone(target);
+  const writer = sax.getWriter();
 
-  Object.assign(parser, handlers(parserConfig, (err, obj) => {
-    if (!err) {
-      results.push(obj);
-    } else {
-      // only report the first error
-      parserError = parserError || err;
-    }
-  }));
-
-  const ts = new Transform({
-    readableObjectMode: true,
-    flush(next) {
-      parser.close();
-      if (parserError) {
-        return next(parserError);
-      }
-      flush(this);
-      next();
+  return new TransformStream({
+    async flush(controller) {
+      await writer.ready;
+      await writer.close();
+      flush(controller);
+      controller.terminate();
     },
-    transform(chunk, encoding, next) {
-      if (parserError) {
-        return next(parserError);
-      }
-      write(chunk);
-      flush(this);
-      next();
+    async transform(chunk, controller) {
+      debug('writing', chunk);
+      await writer.ready;
+      await writer.write(chunk);
+      flush(controller);
     }
   });
-  return ts;
-
-  function write(chunk) {
-    const str = decoder.write(chunk);
-    parser.write(str);
-  }
 
-  function flush(stream) {
-    if (!results.length) {
-      return;
-    }
-    results.forEach(r => {
-      stream.push(r);
-    });
-    results = [];
+  function flush(controller) {
+    results.forEach(r => controller.enqueue(r));
+    results.length = 0;
   }
 }
diff --git a/package.json b/package.json
@@ -16,12 +16,11 @@
     "object"
   ],
   "dependencies": {
-    "debug": "~2 || ~3 || ~4",
-    "sax": "^1.2.1"
+    "@pirxpilot/saxophone": "^1.0.0",
+    "debug": "~2 || ~3 || ~4"
   },
   "devDependencies": {
-    "jshint": "~2",
-    "should": "~13"
+    "jshint": "~2"
   },
   "scripts": {
     "test": "make check"
@@ -30,4 +29,4 @@
     "index.js",
     "lib"
   ]
-}
+}
diff --git a/test/fixtures/nested.xml b/test/fixtures/nested.xml
@@ -0,0 +1,12 @@
+<?xml version="1.0" encoding="utf-8"?>
+<nested>
+  <item>
+    <a>abc</a>
+    <item>
+      <a>def</a>
+      <item>
+        <a>ghi</a>
+      </item>
+      </item>
+  </item>
+</nested>