Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

CORE: Validate Type for String Settings #33503

Merged
merged 5 commits into from
Sep 10, 2018

Conversation

original-brownbear
Copy link
Member

  • Special handling for String in the generic Setting class (I think this is justified though a little dirty because we parse raw values String values for all types ... I don't see how else we can do this without changing the APIs either):
    • If we parse out a String type setting, validate that it came from a scalar raw value
  • Closes String settings leniently accept lists #33135

* Special handling for `String` in the generic
Setting class because we parse raw values
String values for all types:
  * If we parse out a `String` setting validate
that it came from a scalar raw value
* Closes elastic#33135
@elasticmachine
Copy link
Collaborator

Pinging @elastic/es-core-infra

Copy link
Member

@rjernst rjernst left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Instead of special casing this inside Setting, I think the root problem is Settings.toString used by Settings.get serializes whatever the value is to a String no matter the type. The same Map/List logic could be moved to toString to not allow any implicit conversion of object/array to String?

@rjernst
Copy link
Member

rjernst commented Sep 8, 2018

Actually I wonder if toString should only work on Numeric objects? Maybe that or the value already being a String are ok, and anything else should be an error?

@original-brownbear
Copy link
Member Author

@rjernst the problem is that we simply always expect a string here for the parsers:

e.g. all list type settings will use this (and similar) :

    public static <T> Setting<List<T>> listSetting(String key, Function<Settings, List<String>> defaultStringValue,
                                                   Function<String, T> singleValueParser, Property... properties) {
        if (defaultStringValue.apply(Settings.EMPTY) == null) {
            throw new IllegalArgumentException("default value function must not return null");
        }
        Function<String, List<T>> parser = (s) ->
                parseableStringToList(s).stream().map(singleValueParser).collect(Collectors.toList());

        return new ListSetting<>(key, defaultStringValue, parser, properties);
    }

=> If we parse an actual list out of a YAML then we currently have to serialize it to String before hitting the parser (in the method I added the change to).
=> Since we're always parsing all kinds of settings out of a string API wise at the moment + don't have have a reference to the type of the setting in the generic method I don't see how we can even add any logic based on output type that would act before that toString call. This is precisely why I first parse and then check the type of what I parsed.

=> This is really dirty :( This is the best fix I came up with that doesn't require a big change though. I could think about a bigger but cleaner change if we're willing to live with more noise here?

@original-brownbear
Copy link
Member Author

... though any change that gives us the type inside of the setting get method and before parsing would literally mean changing 100% of Setting<T> instantiations because we'd have to pass in the class of the result there ... seems a little crazy? :)

@rjernst
Copy link
Member

rjernst commented Sep 8, 2018

Ok, I understand the problem. Though, I think this hack is just leading us further down a rabbit hole of the current problem, which is that we assume all values are strings. We did already change the concrete type for lists to be a List instead of String (see #26878), but the concrete Setting parser functions all take String. Perhaps they should take Object? Then getRaw can return object instead of String, and each setting parser can decide whether to call toString() on the object (or have a helper method since it is the common case, and that helper method can error when the type is not convertible to String like with Map or List).
/cc @s1monw @jasontedor

@original-brownbear
Copy link
Member Author

original-brownbear commented Sep 8, 2018

@rjernst

Perhaps they should take Object?

Just that would probably solve the issue best I think (with the slightly annoying caveat that we'd maybe have to worry about mutability of that Object within the parser when it comes to List<?> and such ... especially when that list can itself contain more mutable things ... but I guess we have some experience in cleverly wrapping that kind of structure now :P ). That's where the String serialization round-trip was kind of handy :)

=> I think either we change all the parsers (big change but nicer :) ... apart from the mutability thing) or we pass the return type class to Setting<T> (smaller change but not as nice).

@original-brownbear
Copy link
Member Author

@jasontedor adjusted PR to reflect what we discussed just now :) Take a look if this is what you were looking for, from my end it's a fine solution to #33135.

@jasontedor
Copy link
Member

I discussed an approach with @original-brownbear where we identity which settings are list settings; we think this will give us a path forward here and also let us fix upgrading of list settings. We can discuss longer-term fixes in the future, but we need an approach to unblock some in-flight work for now.

Copy link
Member

@jasontedor jasontedor left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I left a couple of comments.

}
} else if (isList) {
throw new IllegalArgumentException(
"Expected list type value for setting [" + setting + "] but found [" + value.getClass() + ']'
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think that will fail compilation? 😇

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Compiles just fine for me? :) What am I missing?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oh, never mind, I misread. Sorry for that. 😄

*/
public String get(String setting, String defaultValue, boolean isList) {
Object value = settings.get(setting);
if(value != null) {
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Nit: if( -> if ( (add space)

@original-brownbear
Copy link
Member Author

@jasontedor Missing space added :)

@jasontedor
Copy link
Member

With this change and:

diff --git a/server/src/main/java/org/elasticsearch/common/settings/AbstractScopedSettings.java b/server/src/main/java/org/elasticsearch/common/settings/AbstractScopedSettings.java
index e25d954aa4f..aee85b6ff57 100644
--- a/server/src/main/java/org/elasticsearch/common/settings/AbstractScopedSettings.java
+++ b/server/src/main/java/org/elasticsearch/common/settings/AbstractScopedSettings.java
@@ -54,7 +54,7 @@ public abstract class AbstractScopedSettings extends AbstractComponent {
     private final List<SettingUpdater<?>> settingUpdaters = new CopyOnWriteArrayList<>();
     private final Map<String, Setting<?>> complexMatchers;
     private final Map<String, Setting<?>> keySettings;
-    private final Map<Setting<?>, Function<Map.Entry<String, String>, Map.Entry<String, String>>> settingUpgraders;
+    private final Map<Setting<?>, SettingUpgrader<?>> settingUpgraders;
     private final Setting.Property scope;
     private static final Pattern KEY_PATTERN = Pattern.compile("^(?:[-\\w]+[.])*[-\\w]+$");
     private static final Pattern GROUP_KEY_PATTERN = Pattern.compile("^(?:[-\\w]+[.])+$");
@@ -70,12 +70,8 @@ public abstract class AbstractScopedSettings extends AbstractComponent {
 
         this.settingUpgraders =
                 Collections.unmodifiableMap(
-                        settingUpgraders
-                                .stream()
-                                .collect(
-                                        Collectors.toMap(
-                                                SettingUpgrader::getSetting,
-                                                u -> e -> new AbstractMap.SimpleEntry<>(u.getKey(e.getKey()), u.getValue(e.getValue())))));
+                        settingUpgraders.stream().collect(Collectors.toMap(SettingUpgrader::getSetting, Function.identity())));
+
 
         this.scope = scope;
         Map<String, Setting<?>> complexMatchers = new HashMap<>();
@@ -786,15 +782,24 @@ public abstract class AbstractScopedSettings extends AbstractComponent {
         boolean changed = false; // track if any settings were upgraded
         for (final String key : settings.keySet()) {
             final Setting<?> setting = getRaw(key);
-            final Function<Map.Entry<String, String>, Map.Entry<String, String>> upgrader = settingUpgraders.get(setting);
+            final SettingUpgrader<?> upgrader = settingUpgraders.get(setting);
             if (upgrader == null) {
                 // the setting does not have an upgrader, copy the setting
                 builder.copy(key, settings);
             } else {
                 // the setting has an upgrader, so mark that we have changed a setting and apply the upgrade logic
                 changed = true;
-                final Map.Entry<String, String> upgrade = upgrader.apply(new Entry(key, settings));
-                builder.put(upgrade.getKey(), upgrade.getValue());
+                if (setting.isListSetting()) {
+                    final List<String> value = settings.getAsList(key);
+                    final String upgradedKey = upgrader.getKey(key);
+                    final List<String> upgradedValue = upgrader.getListValue(value);
+                    builder.putList(upgradedKey, upgradedValue);
+                } else {
+                    final String value = settings.get(key);
+                    final String upgradedKey = upgrader.getKey(key);
+                    final String upgradedValue = upgrader.getValue(value);
+                    builder.put(upgradedKey, upgradedValue);
+                }
             }
         }
         // we only return a new instance if there was an upgrade
diff --git a/server/src/main/java/org/elasticsearch/common/settings/SettingUpgrader.java b/server/src/main/java/org/elasticsearch/common/settings/SettingUpgrader.java
index 91f2bead300..bc41b554905 100644
--- a/server/src/main/java/org/elasticsearch/common/settings/SettingUpgrader.java
+++ b/server/src/main/java/org/elasticsearch/common/settings/SettingUpgrader.java
@@ -19,6 +19,8 @@
 
 package org.elasticsearch.common.settings;
 
+import java.util.List;
+
 /**
  * Represents the logic to upgrade a setting.
  *
@@ -51,4 +53,8 @@ public interface SettingUpgrader<T> {
         return value;
     }
 
+    default List<String> getListValue(final List<String> value) {
+        return value;
+    }
+
 }
diff --git a/server/src/test/java/org/elasticsearch/common/settings/ScopedSettingsTests.java b/server/src/test/java/org/elasticsearch/common/settings/ScopedSettingsTests.java
index 0ee1d2e9c4a..6766316fafd 100644
--- a/server/src/test/java/org/elasticsearch/common/settings/ScopedSettingsTests.java
+++ b/server/src/test/java/org/elasticsearch/common/settings/ScopedSettingsTests.java
@@ -47,6 +47,7 @@ import java.util.concurrent.atomic.AtomicReference;
 import java.util.function.BiConsumer;
 import java.util.function.Consumer;
 import java.util.function.Function;
+import java.util.stream.Collectors;
 
 import static org.hamcrest.CoreMatchers.containsString;
 import static org.hamcrest.CoreMatchers.equalTo;
@@ -1171,4 +1172,47 @@ public class ScopedSettingsTests extends ESTestCase {
         }
     }
 
+    public void testUpgradeListSetting() {
+        final Setting<List<String>> oldSetting =
+                Setting.listSetting("foo.old", Collections.emptyList(), Function.identity(), Property.NodeScope);
+        final Setting<List<String>> newSetting =
+                Setting.listSetting("foo.new", Collections.emptyList(), Function.identity(), Property.NodeScope);
+
+        final AbstractScopedSettings service =
+                new ClusterSettings(
+                        Settings.EMPTY,
+                        new HashSet<>(Arrays.asList(oldSetting, newSetting)),
+                        Collections.singleton(new SettingUpgrader<List<String>>() {
+
+                            @Override
+                            public Setting<List<String>> getSetting() {
+                                return oldSetting;
+                            }
+
+                            @Override
+                            public String getKey(final String key) {
+                                return "foo.new";
+                            }
+
+                            @Override
+                            public List<String> getListValue(final List<String> value) {
+                                return value.stream().map(s -> "new." + s).collect(Collectors.toList());
+                            }
+                        }));
+
+        final int length = randomIntBetween(0, 16);
+        final List<String> values = length == 0 ? Collections.emptyList() : new ArrayList<>(length);
+        for (int i = 0; i < length; i++) {
+            values.add(randomAlphaOfLength(8));
+        }
+
+        final Settings settings = Settings.builder().putList("foo.old", values).build();
+        final Settings upgradedSettings = service.upgradeSettings(settings);
+        assertFalse(oldSetting.exists(upgradedSettings));
+        assertTrue(newSetting.exists(upgradedSettings));
+        assertThat(
+                newSetting.get(upgradedSettings),
+                equalTo(oldSetting.get(settings).stream().map(s -> "new." + s).collect(Collectors.toList())));
+    }
+
 }

I can also get the upgrade settings case to work too.

Copy link
Member

@jasontedor jasontedor left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I am good with this change, but let us see what @rjernst thinks too.

Copy link
Member

@rjernst rjernst left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks, this looks much better. Two minor suggestions.

* Returns the setting value associated with the setting key. If it does not exists,
* returns the default value provided.
*/
public String get(String setting, String defaultValue, boolean isList) {
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

can this be package private?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

yea I think so will adjust in a bit :)

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

+1

@@ -453,7 +458,7 @@ public final String getRaw(final Settings settings) {
* @return the raw string representation of the setting value
*/
String innerGetRaw(final Settings settings) {
return settings.get(getKey(), defaultValue.apply(settings));
return settings.get(getKey(), defaultValue.apply(settings), isListSetting());
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Instead of having an isListSetting, since this is already package private anyways, can this just check instanceof ListSetting?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hmm, yea it could :) -> will change to that :)

Copy link
Member

@jasontedor jasontedor Sep 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@original-brownbear @rjernst See my diff, we need to expose Setting#isListSetting in AbstractScopedSettings.

Copy link
Member

@jasontedor jasontedor Sep 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Its implementation though can be final and be an instance of check against ListSetting.

@original-brownbear
Copy link
Member Author

Adjusted as discussed, will merge once green :)

@jasontedor
Copy link
Member

Thanks @original-brownbear.

@jasontedor jasontedor merged commit 6075e15 into elastic:master Sep 10, 2018
@jasontedor
Copy link
Member

@original-brownbear I merged this so I can open a PR on top of it. Can you handle backporting this to 6.x?

jasontedor pushed a commit that referenced this pull request Sep 11, 2018
When we see a settings value, it could be a list. Yet this should only
happen if the underlying setting type is a list setting type. This
commit adds validation that when we get a setting value that is a list,
that the setting that we are getting is a list setting. And similarly,
if we get a value for a list setting, the underlying value should be a
list.
@jasontedor
Copy link
Member

@original-brownbear I ended up backporting this in e084a0a.

jasontedor added a commit to jasontedor/elasticsearch that referenced this pull request Sep 11, 2018
* master:
  Add full cluster restart base class (elastic#33577)
  Validate list values for settings (elastic#33503)
  Copy and validatie soft-deletes setting on resize (elastic#33517)
  Test: Fix package name
  SQL: Fix result column names for arithmetic functions (elastic#33500)
  Upgrade to latest Lucene snapshot (elastic#33505)
  Enable not wiping cluster settings after REST test (elastic#33575)
  MINOR: Remove Dead Code in SearchScript (elastic#33569)
  [Test] Remove duplicate method in TestShardRouting (elastic#32815)
  mute test on windows
  Update beats template to include apm-server metrics (elastic#33286)
  Fix typos (elastic#33499)
  [CCR] Delay auto follow license check (elastic#33557)
  [CCR] Add create_follow_index privilege (elastic#33559)
  Strengthen FilterRoutingTests (elastic#33149)
  Correctly handle PKCS#11 tokens for system keystore (elastic#33460)
  Remove some duplicate request conversion methods. (elastic#33538)
@original-brownbear
Copy link
Member Author

@jasontedor thanks!

jasontedor added a commit to martijnvg/elasticsearch that referenced this pull request Sep 11, 2018
* master: (91 commits)
  Preserve cluster settings on full restart tests (elastic#33590)
  Use IndexWriter.getFlushingBytes() rather than tracking it ourselves (elastic#33582)
  Fix upgrading of list settings (elastic#33589)
  Add read-only Engine (elastic#33563)
  HLRC: Add ML get categories API (elastic#33465)
  SQL: Adds MONTHNAME, DAYNAME and QUARTER functions (elastic#33411)
  Add predicate_token_filter (elastic#33431)
  Fix Replace function. Adds more tests to all string functions. (elastic#33478)
  [ML] Rename input_fields to column_names in file structure (elastic#33568)
  Add full cluster restart base class (elastic#33577)
  Validate list values for settings (elastic#33503)
  Copy and validatie soft-deletes setting on resize (elastic#33517)
  Test: Fix package name
  SQL: Fix result column names for arithmetic functions (elastic#33500)
  Upgrade to latest Lucene snapshot (elastic#33505)
  Enable not wiping cluster settings after REST test (elastic#33575)
  MINOR: Remove Dead Code in SearchScript (elastic#33569)
  [Test] Remove duplicate method in TestShardRouting (elastic#32815)
  mute test on windows
  Update beats template to include apm-server metrics (elastic#33286)
  ...
original-brownbear added a commit to original-brownbear/elasticsearch that referenced this pull request Sep 13, 2018
* Reverts setting type validation introduced in elastic#33503
original-brownbear added a commit that referenced this pull request Sep 13, 2018
* Reverts setting type validation introduced in #33503
original-brownbear added a commit to original-brownbear/elasticsearch that referenced this pull request Sep 13, 2018
* Reverts setting type validation introduced in elastic#33503
original-brownbear added a commit that referenced this pull request Sep 13, 2018
* Reverts setting type validation introduced in #33503
@jimczi jimczi added v7.0.0-beta1 and removed v7.0.0 labels Feb 7, 2019
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants