From 2eb4e3606ad8cfa35681603141294fd9e519742d Mon Sep 17 00:00:00 2001
From: Saurabh Misra <misra.saurabh1@gmail.com>
Date: Wed, 15 Oct 2025 11:59:10 -0700
Subject: [PATCH] find common tags

Signed-off-by: Saurabh Misra <misra.saurabh1@gmail.com>
---
 codeflash/result/common_tags.py | 11 +++++++++++
 tests/test_common_tags.py       | 22 ++++++++++++++++++++++
 2 files changed, 33 insertions(+)
 create mode 100644 codeflash/result/common_tags.py
 create mode 100644 tests/test_common_tags.py

diff --git a/codeflash/result/common_tags.py b/codeflash/result/common_tags.py
new file mode 100644
index 000000000..504cb5cd6
--- /dev/null
+++ b/codeflash/result/common_tags.py
@@ -0,0 +1,11 @@
+from __future__ import annotations
+
+
+def find_common_tags(articles: list[dict[str, list[str]]]) -> set[str]:
+    if not articles:
+        return set()
+
+    common_tags = articles[0].get("tags", [])
+    for article in articles[1:]:
+        common_tags = [tag for tag in common_tags if tag in article.get("tags", [])]
+    return set(common_tags)
diff --git a/tests/test_common_tags.py b/tests/test_common_tags.py
new file mode 100644
index 000000000..754261b55
--- /dev/null
+++ b/tests/test_common_tags.py
@@ -0,0 +1,22 @@
+from codeflash.result.common_tags import find_common_tags
+
+
+def test_common_tags_1() -> None:
+    articles_1 = [
+        {"title": "Article 1", "tags": ["Python", "AI", "ML"]},
+        {"title": "Article 2", "tags": ["Python", "Data Science", "AI"]},
+        {"title": "Article 3", "tags": ["Python", "AI", "Big Data"]},
+    ]
+
+    expected = {"Python", "AI"}
+
+    assert find_common_tags(articles_1) == expected
+
+    articles_2 = [
+        {"title": "Article 1", "tags": ["Python", "AI", "ML"]},
+        {"title": "Article 2", "tags": ["Python", "Data Science", "AI"]},
+        {"title": "Article 3", "tags": ["Python", "AI", "Big Data"]},
+        {"title": "Article 4", "tags": ["Python", "AI", "ML"]},
+    ]
+
+    assert find_common_tags(articles_2) == expected