From 0b74159526c9d2a7adbc8624e90e534ba12ce275 Mon Sep 17 00:00:00 2001
From: ylamgarchal <yassine.lamgarchal@redhat.com>
Date: Thu, 7 Nov 2024 12:04:45 +0100
Subject: [PATCH] analytics: fix regexp generation on nested fields

nrt added

Change-Id: I1bbcf3561ad3f312e24bae911cd1344e8a5d5afa
---
 dci/analytics/query_es_dsl.py        |  9 ++++--
 tests/analytics/test_query_es_dsl.py | 48 ++++++++++++++++++++++++++++
 2 files changed, 55 insertions(+), 2 deletions(-)

diff --git a/dci/analytics/query_es_dsl.py b/dci/analytics/query_es_dsl.py
index 92e8f4092..547a0c49f 100644
--- a/dci/analytics/query_es_dsl.py
+++ b/dci/analytics/query_es_dsl.py
@@ -17,7 +17,9 @@
 import pyparsing as pp
 
 _field = pp.Word(pp.alphanums + "_" + ".")
-_value = pp.Word(pp.alphanums + "_" + "-" + "%" + "." + ":" + "\\" + "*" + "?")
+_value = pp.Word(
+    pp.alphanums + "_" + "-" + "%" + "." + ":" + "\\" + "*" + "?" + "+" + "{" + "}"
+)
 _word = pp.Word(pp.alphanums + "_" + "-" + "." + " " + ":")
 _comma = pp.Suppress(pp.Literal(","))
 _lp = pp.Suppress(pp.Literal("("))
@@ -91,7 +93,7 @@ def _generate_from_operators(parsed_query, handle_nested=False):
             handle_nested, operator, operand_1, operand_2
         )
     elif operator == "=~":
-        return {
+        _regexp = {
             "regexp": {
                 operand_1: {
                     "value": operand_2,
@@ -100,6 +102,9 @@ def _generate_from_operators(parsed_query, handle_nested=False):
                 }
             }
         }
+        if handle_nested and "." in operand_1:
+            return {"nested": {"path": operand_1.split(".")[0], "query": _regexp}}
+        return _regexp
     elif operator == "not_in":
         if handle_nested and "." in operand_1:
             return {
diff --git a/tests/analytics/test_query_es_dsl.py b/tests/analytics/test_query_es_dsl.py
index 8da95478a..d5d210ae2 100644
--- a/tests/analytics/test_query_es_dsl.py
+++ b/tests/analytics/test_query_es_dsl.py
@@ -383,3 +383,51 @@ def test_query_build_comparison_operator():
             ]
         }
     }
+
+
+def test_nrt_query_build_nested_regexp():
+    ret = qed.build(
+        "((components.type=ocp) and (name=~.*upgrade.*) and (team.name=~Intel.+) and (topic.name=OCP-4.16)  and (tags in [daily]))"
+    )
+    assert ret == {
+        "bool": {
+            "filter": [
+                {
+                    "nested": {
+                        "path": "components",
+                        "query": {"term": {"components.type": "ocp"}},
+                    }
+                },
+                {
+                    "regexp": {
+                        "name": {
+                            "value": ".*upgrade.*",
+                            "flags": "ALL",
+                            "case_insensitive": True,
+                        }
+                    }
+                },
+                {
+                    "nested": {
+                        "path": "team",
+                        "query": {
+                            "regexp": {
+                                "team.name": {
+                                    "value": "Intel.+",
+                                    "flags": "ALL",
+                                    "case_insensitive": True,
+                                }
+                            }
+                        },
+                    }
+                },
+                {
+                    "nested": {
+                        "path": "topic",
+                        "query": {"term": {"topic.name": "OCP-4.16"}},
+                    }
+                },
+                {"terms": {"tags": ["daily"]}},
+            ]
+        }
+    }