From 4605b0d5b8708429c2a91a355c9c25d9ddc63519 Mon Sep 17 00:00:00 2001
From: Jose Rego <jose@lightdash.com>
Date: Wed, 12 Nov 2025 11:12:14 +0000
Subject: [PATCH] feat: pr files

---
 properties.sample.json           | 56 ++++++++++++++++++++++++++++++++
 tap_github/schemas/pr_files.json | 21 ++++++++++++
 tap_github/streams.py            | 27 ++++++++++++++-
 3 files changed, 103 insertions(+), 1 deletion(-)
 create mode 100644 properties.sample.json
 create mode 100644 tap_github/schemas/pr_files.json

diff --git a/properties.sample.json b/properties.sample.json
new file mode 100644
index 00000000..dc7bae04
--- /dev/null
+++ b/properties.sample.json
@@ -0,0 +1,56 @@
+{
+  "streams": [
+    {
+      "stream": "pull_requests",
+      "tap_stream_id": "pull_requests",
+      "key_properties": ["id"],
+      "schema": {
+        "properties": {
+          "id": {"type": ["null", "integer"]},
+          "number": {"type": ["null", "integer"]},
+          "updated_at": {"type": ["null", "string"], "format": "date-time"}
+        }
+      },
+      "metadata": [
+        {
+          "breadcrumb": [],
+          "metadata": {
+            "selected": true,
+            "tap-stream-id": "pull_requests",
+            "table-key-properties": ["id"],
+            "valid-replication-keys": ["updated_at"],
+            "replication-method": "INCREMENTAL"
+          }
+        },
+        {"breadcrumb": ["properties", "id"], "metadata": {"inclusion": "automatic"}},
+        {"breadcrumb": ["properties", "number"], "metadata": {"inclusion": "automatic"}},
+        {"breadcrumb": ["properties", "updated_at"], "metadata": {"inclusion": "automatic"}}
+      ]
+    },
+    {
+      "stream": "pr_files",
+      "tap_stream_id": "pr_files",
+      "key_properties": ["id"],
+      "schema": {
+        "properties": {
+          "id": {"type": ["null", "string"]},
+          "updated_at": {"type": ["null", "string"], "format": "date-time"},
+          "filename": {"type": ["null", "string"]}
+        }
+      },
+      "metadata": [
+        {
+          "breadcrumb": [],
+          "metadata": {
+            "selected": true,
+            "tap-stream-id": "pr_files",
+            "table-key-properties": ["id"],
+            "valid-replication-keys": ["updated_at"],
+            "replication-method": "INCREMENTAL",
+            "parent-tap-stream-id": "pull_requests"
+          }
+        }
+      ]
+    }
+  ]
+}
\ No newline at end of file
diff --git a/tap_github/schemas/pr_files.json b/tap_github/schemas/pr_files.json
new file mode 100644
index 00000000..629a4e27
--- /dev/null
+++ b/tap_github/schemas/pr_files.json
@@ -0,0 +1,21 @@
+{
+  "type": ["null", "object"],
+  "properties": {
+    "_sdc_repository": {"type": ["string"]},
+    "sha": {"type": ["null", "string"]},
+    "filename": {"type": ["null", "string"]},
+    "status": {"type": ["null", "string"]},
+    "additions": {"type": ["null", "integer"]},
+    "deletions": {"type": ["null", "integer"]},
+    "changes": {"type": ["null", "integer"]},
+    "blob_url": {"type": ["null", "string"]},
+    "raw_url": {"type": ["null", "string"]},
+    "contents_url": {"type": ["null", "string"]},
+    "patch": {"type": ["null", "string"]},
+    "previous_filename": {"type": ["null", "string"]},
+    "id": {"type": ["null", "string"]},
+    "updated_at": {"type": ["null", "string"], "format": "date-time"},
+    "pr_number": {"type": ["null", "integer"]},
+    "pr_id": {"type": ["null", "integer"]}
+  }
+}
diff --git a/tap_github/streams.py b/tap_github/streams.py
index 707898ef..15f539a6 100644
--- a/tap_github/streams.py
+++ b/tap_github/streams.py
@@ -504,6 +504,30 @@ def add_fields_at_1st_level(self, record, parent_record = None):
         record['pr_id'] = parent_record.get('id')
         record['id'] = '{}-{}'.format(parent_record.get('id'), record.get('sha'))
 
+class PRFiles(IncrementalStream):
+    '''
+    https://docs.github.com/en/rest/pulls/pulls#list-pull-requests-files
+    '''
+    tap_stream_id = "pr_files"
+    replication_method = "INCREMENTAL"
+    replication_keys = "updated_at"
+    key_properties = ["id"]
+    path = "pulls/{}/files"
+    use_repository = True
+    id_keys = ['number']
+    parent = 'pull_requests'
+
+    def add_fields_at_1st_level(self, record, parent_record = None):
+        """
+        Add fields in the record explicitly at the 1st level of JSON.
+        """
+        # Use the PR's updated_at for incremental replication because file objects lack timestamps
+        record['updated_at'] = parent_record.get('updated_at')
+        record['pr_number'] = parent_record.get('number')
+        record['pr_id'] = parent_record.get('id')
+        # Build a stable id based on PR id and filename
+        record['id'] = '{}-{}'.format(parent_record.get('id'), record.get('filename'))
+
 class PullRequests(IncrementalOrderedStream):
     '''
     https://developer.github.com/v3/pulls/#list-pull-requests
@@ -513,7 +537,7 @@ class PullRequests(IncrementalOrderedStream):
     replication_keys = "updated_at"
     key_properties = ["id"]
     path = "pulls?state=all&sort=updated&direction=desc"
-    children = ['reviews', 'review_comments', 'pr_commits']
+    children = ['reviews', 'review_comments', 'pr_commits', 'pr_files']
     pk_child_fields = ["number"]
 
 class TeamMemberships(FullTableStream):
@@ -716,6 +740,7 @@ def add_fields_at_1st_level(self, record, parent_record = None):
     "reviews": Reviews,
     "review_comments": ReviewComments,
     "pr_commits": PRCommits,
+    "pr_files": PRFiles,
     "teams": Teams,
     "team_members": TeamMembers,
     "team_memberships": TeamMemberships,