optimize Pattern Query Tree for size of its JSON.stringify()'ed representation

1 files changed, 59 insertions, 54 deletions

diff --git a/common/patterns_query_tree.js b/common/patterns_query_tree.js
index ea3607e..8d01cc1 100644
--- a/common/patterns_query_tree.js
+++ b/common/patterns_query_tree.js
@@ -3,7 +3,7 @@
  *
  * Function: Data structure to query items by URL patterns.
  *
- * Copyright (C) 2021 Wojtek Kosior
+ * Copyright (C) 2021, 2022 Wojtek Kosior <koszko@koszko.org>
  *
  * This program is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -41,8 +41,6 @@
  * proprietary program, I am not going to enforce this in court.
  */
 
-// TODO! Modify the code to use `Object.create(null)` instead of `{}`.
-
 #FROM common/patterns.js IMPORT deconstruct_url
 
 /* "Pattern Tree" is how we refer to the data structure used for querying
@@ -53,44 +51,48 @@
 const pattern_tree_make = () => ({})
 #EXPORT  pattern_tree_make  AS make
 
-function empty_node() {
-    return {
-	wildcard_matches: [null, null, null],
-	literal_match:    null,
-	children:         {}
-    };
-}
+const empty_node = () => ({});
 
 function is_empty_node(tree_node) {
-    const children = tree_node.children;
-    for (const key in children) {
-	if (children.hasOwnProperty(key))
-	    return false;
-    }
-
-    if (tree_node.wildcard_matches.reduce((a, b) => b && a !== null, 1))
+    for (const key in tree_node)
 	return false;
 
-    return tree_node.literal_match === null;
+    return true;
 }
 
+const wildcard_matches = node => [node["*"], node["**"], node["***"]];
+
 const is_wildcard = segment => ["*", "**", "***"].lastIndexOf(segment) >= 0;
 
 /*
+ * Remove reference to given child fron node and leave the node in consistent
+ * state afterwards, i.e. remove the "c" property if no childs are left.
+ */
+function delete_child(node, child_name) {
+    if (node.c) {
+	delete node.c[child_name];
+
+	for (const key in node.c)
+	    return;
+
+	delete node.c;
+    }
+}
+
+/*
  * Yields all matches of this segments sequence against the tree that starts at
  * this node. Results are produces in order from greatest to lowest pattern
  * specificity.
  */
-function* search_sequence(tree_node, segments)
-{
+function* search_sequence(tree_node, segments) {
     const nodes = [tree_node];
 
-    for (const segment of segments) {
-	const next_node = nodes[nodes.length - 1].children[segment];
-	if (next_node === undefined)
+    for (const current_segment of segments) {
+	const children = nodes[nodes.length - 1].c || {};
+	if (!Object.hasOwnProperty.call(children, current_segment))
 	    break;
 
-	nodes.push(next_node);
+	nodes.push(children[current_segment]);
     }
 
     const nsegments = segments.length;
@@ -106,10 +108,11 @@ function* search_sequence(tree_node, segments)
 
     while (nodes.length) {
 	const node = nodes.pop();
-	const items = [node.literal_match, ...node.wildcard_matches];
+	const literal_match = node.l;
+	const items = [literal_match, ...wildcard_matches(node)];
 
-	for (let i = 0; i < 4; i++) {
-	    if (items[i] !== null && conds[i]())
+	for (let i = 0; i < items.length; i++) {
+	    if (items[i] !== undefined && conds[i]())
 		yield items[i];
 	}
     }
@@ -129,53 +132,60 @@ function* search_sequence(tree_node, segments)
  * segments and value returned by item_modifier is not `null`, make the value
  * queryable by this path.
  */
-function modify_sequence(tree_node, segments, item_modifier)
-{
+function modify_sequence(tree_node, segments, item_modifier) {
     const nodes = [tree_node];
     let removed = true;
 
     for (var current_segment of segments) {
-	const child = tree_node.children[current_segment] || empty_node();
-	tree_node.children[current_segment] = child;
+	const children = tree_node.c || {};
+	tree_node.c = children;
+
+	const child = Object.hasOwnProperty.call(children, current_segment) ?
+	      children[current_segment] : empty_node();
+	children[current_segment] = child;
+
 	tree_node = child;
 	nodes.push(tree_node);
     }
 
-    tree_node.literal_match = item_modifier(tree_node.literal_match);
-    if (tree_node.literal_match !== null)
+    tree_node.l = item_modifier(tree_node.l || null);
+    if (tree_node.l === null)
+	delete tree_node.l;
+    else
 	removed = false;
 
     let i = segments.length;
 
     if (is_wildcard(current_segment)) {
-	const asterisks = current_segment.length - 1;
-	const wildcards = nodes[i - 1].wildcard_matches;
-	wildcards[asterisks] = item_modifier(wildcards[asterisks]);
-	if (wildcards[asterisks] !== null)
+	nodes[i - 1][current_segment] =
+	    item_modifier(nodes[i - 1][current_segment] || null);
+	if (nodes[i - 1][current_segment] === null)
+	    delete nodes[i - 1][current_segment];
+	else
 	    removed = false;
     }
 
-    while (!removed)
+    if (!removed)
 	return;
 
     while (i > 0) {
 	tree_node = nodes[i--];
 	if (is_empty_node(tree_node))
-	    delete nodes[i].children[segments[i]];
+	    delete_child(nodes[i], segments[i]);
+	else
+	    break;
     }
 }
 
 /* Helper function for modify_tree(). */
-function modify_path(tree_node, deco, item_modifier)
-{
+function modify_path(tree_node, deco, item_modifier) {
     tree_node = tree_node || empty_node();
     modify_sequence(tree_node, deco.path, item_modifier);
     return is_empty_node(tree_node) ? null : tree_node;
 }
 
 /* Helper function for modify_tree(). */
-function modify_domain(tree_node, deco, item_modifier)
-{
+function modify_domain(tree_node, deco, item_modifier) {
     const path_modifier = branch => modify_path(branch, deco, item_modifier);
     tree_node = tree_node || empty_node();
     /* We need an array of domain labels ordered most-significant-first. */
@@ -184,8 +194,7 @@ function modify_domain(tree_node, deco, item_modifier)
 }
 
 /* Helper function for pattern_tree_register() and pattern_tree_deregister(). */
-function modify_tree(patterns_by_proto, pattern, item_modifier)
-{
+function modify_tree(patterns_by_proto, pattern, item_modifier) {
     /*
      * We pass 'false' to disable length limits on URL parts. Length limits are
      * mostly useful in case of iteration over all patterns matching given URL.
@@ -208,8 +217,7 @@ function modify_tree(patterns_by_proto, pattern, item_modifier)
  * Make item queryable through the Pattern Tree that starts with the protocols
  * dictionary object passed in the first argument.
  */
-function pattern_tree_register(patterns_by_proto, pattern, item_name, item)
-{
+function pattern_tree_register(patterns_by_proto, pattern, item_name, item) {
     const key_prefix = pattern[pattern.length - 1] === '/' ? '/' : '_';
     item_name = key_prefix + item_name;
     const add_item = obj => Object.assign(obj || {}, {[item_name]: item});
@@ -218,8 +226,7 @@ function pattern_tree_register(patterns_by_proto, pattern, item_name, item)
 #EXPORT  pattern_tree_register  AS register
 
 /* Helper function for pattern_tree_deregister(). */
-function _remove_item(obj, item_name)
-{
+function _remove_item(obj, item_name) {
     obj = obj || {};
     delete obj[item_name];
     for (const key in obj)
@@ -233,8 +240,7 @@ function _remove_item(obj, item_name)
  * should be pattern and name that have been earlier passed to
  * pattern_tree_register().
  */
-function pattern_tree_deregister(patterns_by_proto, pattern, item_name)
-{
+function pattern_tree_deregister(patterns_by_proto, pattern, item_name) {
     const key_prefix = pattern[pattern.length - 1] === '/' ? '/' : '_';
     item_name = key_prefix + item_name;
     const remove_item = obj => _remove_item(obj, item_name);
@@ -244,12 +250,11 @@ function pattern_tree_deregister(patterns_by_proto, pattern, item_name)
 
 /*
  * Yield registered items that match url. Each yielded value is an object with
- * key being matched item names and values being the items. One such object
+ * keys being matched item names and values being the items. One such object
  * shall contain all items matched with given pattern specificity. Objects are
  * yielded in order from greatest to lowest pattern specificity.
  */
-function* pattern_tree_search(patterns_by_proto, url)
-{
+function* pattern_tree_search(patterns_by_proto, url) {
     const deco = deconstruct_url(url, false);
 
     const tree_for_proto = patterns_by_proto[deco.proto] || empty_node();