feat: add support for self-joins with sub-query by GurtejSohi · Pull Request #220 · hypertrace/document-store

GurtejSohi · 2025-04-01T12:23:47Z

Description

This PR has changes to add support for self-joins with sub-query (represented by SubQueryJoinExpression) in MongoDB.
Note that postgres impl of FromTypeExpressionVisitor for visiting SubQueryJoinExpression throws an UnsupportedOperationException. Currently, only the mongo implementation supports this operation.

Testing

Added an integration test for testing the mongo impl of self-join with sub-query. It also demonstrates its sample usage

suresh-prakash · 2025-04-10T07:39:20Z

+  {"date":"2014-03-01T09:00:00Z","item":"Mirror","quantity":1},
+  {"date":"2014-04-04T11:21:39.736Z","item":"Shampoo","quantity":20},
+  {"date":"2016-02-06T20:20:13Z","item":"Soap","quantity":5}
+]


Nit: Empty line at the end of this file.

suresh-prakash · 2025-04-10T07:39:32Z

+[
+  {"date":"2015-09-10T08:43:00Z","item":"Comb","quantity":10},
+  {"date":"2014-03-01T09:00:00Z","item":"Mirror","quantity":1},
+  {"date":"2014-04-04T11:21:39.736Z","item":"Shampoo","quantity":20},


Nit: Can we format this file as JSON?

suresh-prakash · 2025-04-10T07:40:04Z

+ */
+@Value
+public class AliasedIdentifierExpression extends IdentifierExpression {
+  String contextAlias;


Why not just alias?

Alias sounds like we are giving the field name as alias (like in SelectionSpec. If you look at the example above, this is not the alias for the field, but the alias for the context in which we are referring the field.

Makes sense.

suresh-prakash · 2025-04-10T07:41:35Z

+
+  @Override
+  public String toString() {
+    return String.format("JOIN (%s) AS %s ON %s", subQuery, subQueryAlias, joinCondition);


Nit:

Suggested change

return String.format("JOIN (%s) AS %s ON %s", subQuery, subQueryAlias, joinCondition);

return String.format("JOIN (%s) AS %s ON (%s)", subQuery, subQueryAlias, joinCondition);

suresh-prakash · 2025-04-10T07:44:39Z

          List.of(
              query -> singleton(getFilterClause(query, Query::getFilter)),
-              MongoFromTypeExpressionParser::getFromClauses,
+              query -> new MongoFromTypeExpressionParser(this).getFromClauses(query),


Any way to avoid passing down the this pointer? It seems to be going away from the standard way.

suresh-prakash · 2025-04-10T08:01:36Z

+  public BasicDBObject visit(LogicalExpression expression) {
+    BasicDBObject letClause = new BasicDBObject();
+    for (FilterTypeExpression operand : expression.getOperands()) {
+      letClause.putAll((Map<String, Object>) operand.accept(this));


We seem to be casting BasicDBObject into a Map. This could be confusing (probably compiler might be generating a warning).

Can we either convert everything to BasicDBObject or everything to Map?

changed the return type to map

suresh-prakash · 2025-04-10T08:03:01Z

+
+    @Override
+    public BasicDBObject visit(ConstantExpression expression) {
+      return new BasicDBObject();


Is this deliberate?
If this is not expected here, why not extend the MongoSelectTypeExpressionParser?

yes, this is deliberate as this visitor is expected to go through SelectTypeExpressions and only return the fields that have to be declared in the let clause. So any other field other than AliasedIdentifierExpression has to be ignored but it should not throw an error in that case and for the case of AliasedIdentifierExpression it should return the field name that we need to add to the let clause.

Got it. This may be a common use-case. So, can we create a common base class that returns default (empty map/BasicDBObject) for every implementation and then override it?

That way

The default returning shared base-class can be re-used

Implementation here would look concise

suresh-prakash · 2025-04-10T08:04:29Z

+      AliasedIdentifierExpression aliasedExpression, String subQueryAlias) {
+    BasicDBObject letClause = new BasicDBObject();
+    if (aliasedExpression.getContextAlias().equals(subQueryAlias)) {
+      letClause.put(aliasedExpression.getName(), "$" + aliasedExpression.getName());


What happens if the AliasedExpression contains a nested field like "a.b".
Does it need escaping/encoding?

It would be beneficial to add an integration for the same.

The field name of the aliasedExpression is given by the user and if we can make an assumption that the user won't give a dot in the field name then we should be fine as we are only using the field name here and not the context alias.

I'm a bit confused. Does it allow me to pick the value of "a.b" and alias it as "c"?

In other words, are you saying that the alias can't contain a dot, but the identifier name can?

Sorry, my bad, I misunderstood your previous comment. I went through it again and it made sense this time. I've updated the implementation to encode the variable name so that it can handle nested fields also. I've also added an integration test for it.

suresh-prakash · 2025-04-10T08:06:51Z

+   * Subquery join expression is not supported by default. Override this method to support it.
+   */
+  default <T> T visit(SubQueryJoinExpression subQueryJoinExpression) {
+    throw new UnsupportedOperationException("This operation is not supported");


Typically, we should force all the implementors of visitor interface to implement all the methods.

Whether to throw exception or not should be the implementor's responsibility.

We probably created the base class
MongoSelectTypeExpressionParser for a similar purpose (without having a default implementation in the interface).

The reason being, it may make sense to implement this in other implementations (other than MongoDB).

suresh-prakash · 2025-04-10T08:07:08Z

  <T> T visit(final IdentifierExpression expression);
+
+  default <T> T visit(final AliasedIdentifierExpression expression) {
+    throw new UnsupportedOperationException();


Same here. Not a typical visitor flow.

…or nested fields as well

suresh-prakash · 2025-04-11T05:58:21Z

+ */
+@Value
+public class AliasedIdentifierExpression extends IdentifierExpression {
+  String contextAlias;


Makes sense.

suresh-prakash · 2025-04-11T05:58:56Z

    return key.replace("\\u002e", FIELD_SEPARATOR).replace("\\u0024", PREFIX).replace("\\\\", "\\");
  }

+  public static String encodeVariableName(final String variableName) {


Why not use the existing "encode" method in MongoUtils?

Encode method in MongoUtils also adds backslashes. Backslashes are allowed in keys but not in variable names, as mentioned here - https://www.mongodb.com/docs/manual/reference/aggregation-variables/.
So I had to introduce a new method to encode variable names.

suresh-prakash · 2025-04-11T06:01:11Z

+  public Map<String, Object> visit(KeyExpression expression) {
+    return Collections.emptyMap();
+  }
+
+  @Override
+  public Map<String, Object> visit(ArrayRelationalFilterExpression expression) {
+    return Collections.emptyMap();
+  }
+
+  @Override
+  public Map<String, Object> visit(DocumentArrayFilterExpression expression) {


We don't need to override these, right?

I just realized that it might be confusing because I put the inner class in between the outer class methods.
Did you think that these are methods of the inner class? If so, please note that these are outer class methods and since it implements filter type expression it has to implement these methods as well.

To avoid confusion, I've moved the inner class below the outer class methods

Oh, yeah. That was the confusion. Makes sense to keep the inner classes towards the end.

…lass methods

codecov · 2025-04-11T07:05:51Z

Codecov Report

Attention: Patch coverage is 79.41176% with 42 lines in your changes missing coverage. Please review.

Project coverage is 79.61%. Comparing base (577aaac) to head (f7a0118).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
...mongo/query/MongoAggregationPipelineConverter.java	87.87%	1 Missing and 7 partials ⚠️
...e/expression/impl/AliasedIdentifierExpression.java	66.66%	2 Missing and 4 partials ⚠️
...go/query/parser/MongoEmptySelectionTypeParser.java	28.57%	5 Missing ⚠️
...tore/mongo/query/parser/MongoLetClauseBuilder.java	81.81%	3 Missing and 1 partial ⚠️
...tstore/expression/impl/SubQueryJoinExpression.java	75.00%	2 Missing ⚠️
...ypertrace/core/documentstore/mongo/MongoUtils.java	33.33%	1 Missing and 1 partial ⚠️
...transformer/PostgresSelectionQueryTransformer.java	0.00%	2 Missing ⚠️
...v1/transformer/PostgresUnnestQueryTransformer.java	0.00%	2 Missing ⚠️
.../documentstore/mongo/query/MongoQueryExecutor.java	88.88%	1 Missing ⚠️
.../documentstore/mongo/query/parser/AliasParser.java	0.00%	1 Missing ⚠️
... and 9 more

Additional details and impacted files

@@             Coverage Diff              @@
##               main     #220      +/-   ##
============================================
- Coverage     79.73%   79.61%   -0.12%     
- Complexity     1043     1048       +5     
============================================
  Files           199      204       +5     
  Lines          4945     5083     +138     
  Branches        410      416       +6     
============================================
+ Hits           3943     4047     +104     
- Misses          711      740      +29     
- Partials        291      296       +5

Flag	Coverage Δ
integration	`79.61% <79.41%> (-0.12%)`	⬇️
unit	`56.32% <33.33%> (-1.35%)`	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

github-actions · 2025-04-11T07:06:44Z

Test Results

40 files ±0 40 suites ±0 31s ⏱️ -1s
246 tests ±0 246 ✅ ±0 0 💤 ±0 0 ❌ ±0
502 runs +2 502 ✅ +2 0 💤 ±0 0 ❌ ±0

Results for commit f7a0118. ± Comparison against base commit 577aaac.

feat: add support for self-joins

c029bdb

suresh-prakash reviewed Apr 2, 2025

View reviewed changes

GurtejSohi added 2 commits April 2, 2025 14:15

chore: address review comments

26932df

refactor: api changes for join expression

0378c9c

GurtejSohi requested a review from suresh-prakash April 2, 2025 14:43

style: spotless formatting

fb4d86d

suresh-prakash reviewed Apr 3, 2025

View reviewed changes

GurtejSohi added 3 commits April 8, 2025 15:14

feat: implementation for sub query joins within the same collection

72cbcff

chore: complete the impl for sub query self join

ca4d5f8

test: add integration test for self join with sub-query

986c619

GurtejSohi marked this pull request as ready for review April 9, 2025 21:15

GurtejSohi requested review from avinashkolluru, kotharironak and skjindal93 as code owners April 9, 2025 21:15

GurtejSohi requested a review from suresh-prakash April 9, 2025 21:15

GurtejSohi changed the title ~~feat: add support for self-joins~~ feat: add support for self-joins with sub-query Apr 9, 2025

chore: cleaned up code and added docs

a774807

suresh-prakash reviewed Apr 10, 2025

View reviewed changes

GurtejSohi added 3 commits April 10, 2025 17:54

chore: address review comments

a72ddf6

chore: address review comments p2

9ee78fb

test: add integration test to verify self join with sub-query works f…

11f5d28

…or nested fields as well

GurtejSohi requested a review from suresh-prakash April 10, 2025 18:48

chore: update collection data for integration test

ab72570

suresh-prakash reviewed Apr 11, 2025

View reviewed changes

refactor: move inner class in MongoLetClauseBuilder below the outer c…

b572d5f

…lass methods

GurtejSohi requested a review from suresh-prakash April 11, 2025 06:41

suresh-prakash approved these changes Apr 11, 2025

View reviewed changes

Merge branch 'main' into add-support-for-self-joins

f7a0118

GurtejSohi merged commit 015ee5c into main Apr 11, 2025
5 of 7 checks passed

GurtejSohi deleted the add-support-for-self-joins branch April 11, 2025 07:18

	return String.format("JOIN (%s) AS %s ON %s", subQuery, subQueryAlias, joinCondition);
	return String.format("JOIN (%s) AS %s ON (%s)", subQuery, subQueryAlias, joinCondition);

Conversation

GurtejSohi commented Apr 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Testing

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

GurtejSohi Apr 11, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

codecov Bot commented Apr 11, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

GurtejSohi commented Apr 1, 2025 •

edited

Loading

GurtejSohi Apr 11, 2025 •

edited

Loading

codecov Bot commented Apr 11, 2025 •

edited

Loading