From f33d3a79efbb8dc8b3fb6ed6e6588a2dd85f50f8 Mon Sep 17 00:00:00 2001
From: Raymond Kim <109366641+tt-rkim@users.noreply.github.com>
Date: Thu, 26 Sep 2024 18:06:58 -0400
Subject: [PATCH] #12878: Add links to job and pipeline for CI/CD analytics
 (#13183)

* #12878: Fix some hardcodes in data collection and add pipeline + job links

* #12878: Add links to pipeline + job in pydantic model so we can actually start getting it into the JSON
---
 infra/data_collection/github/utils.py    | 12 ++++++++----
 infra/data_collection/pydantic_models.py | 10 ++++++++++
 2 files changed, 18 insertions(+), 4 deletions(-)

diff --git a/infra/data_collection/github/utils.py b/infra/data_collection/github/utils.py
index 84ef0d97a70..e192f87a075 100644
--- a/infra/data_collection/github/utils.py
+++ b/infra/data_collection/github/utils.py
@@ -54,8 +54,7 @@ def get_pipeline_row_from_github_info(github_runner_environment, github_pipeline
     github_pipeline_id = github_pipeline_json["id"]
     pipeline_submission_ts = github_pipeline_json["created_at"]
 
-    logger.warning("Using hardcoded value for repository_url")
-    repository_url = "https://github.com/tenstorrent/tt-metal"
+    repository_url = github_pipeline_json["repository"]["html_url"]
 
     jobs = github_jobs_json["jobs"]
     jobs_start_times = list(map(lambda job_: get_datetime_from_github_datetime(job_["started_at"]), jobs))
@@ -75,8 +74,7 @@ def get_pipeline_row_from_github_info(github_runner_environment, github_pipeline
     pipeline_end_ts = github_pipeline_json["updated_at"]
     name = github_pipeline_json["name"]
 
-    logger.warning("Using hardcoded value tt-metal for project value")
-    project = "tt-metal"
+    project = github_pipeline_json["repository"]["name"]
 
     trigger = github_runner_environment["github_event_name"]
 
@@ -92,6 +90,8 @@ def get_pipeline_row_from_github_info(github_runner_environment, github_pipeline
     logger.warning("Using hardcoded value github_actions for orchestrator value")
     orchestrator = "github_actions"
 
+    github_pipeline_link = github_pipeline_json["html_url"]
+
     return {
         "github_pipeline_id": github_pipeline_id,
         "repository_url": repository_url,
@@ -106,6 +106,7 @@ def get_pipeline_row_from_github_info(github_runner_environment, github_pipeline
         "git_commit_hash": git_commit_hash,
         "git_author": git_author,
         "orchestrator": orchestrator,
+        "github_pipeline_link": github_pipeline_link,
     }
 
 
@@ -217,6 +218,8 @@ def get_job_row_from_github_job(github_job):
     logger.warning("docker_image erroneously used in pipeline data model, but should be moved. Returning null")
     docker_image = None
 
+    github_job_link = github_job["html_url"]
+
     return {
         "github_job_id": github_job_id,
         "host_name": host_name,
@@ -231,6 +234,7 @@ def get_job_row_from_github_job(github_job):
         "is_build_job": is_build_job,
         "job_matrix_config": job_matrix_config,
         "docker_image": docker_image,
+        "github_job_link": github_job_link,
     }
 
 
diff --git a/infra/data_collection/pydantic_models.py b/infra/data_collection/pydantic_models.py
index 16ec7c682d8..9aa5a3905a5 100644
--- a/infra/data_collection/pydantic_models.py
+++ b/infra/data_collection/pydantic_models.py
@@ -48,6 +48,10 @@ class Job(BaseModel):
         None,
         description="Identifier for the Github Actions CI job, for pipelines " "orchestrated and executed by Github.",
     )
+    github_job_link: Optional[str] = Field(
+        None,
+        description="Link to the Github Actions CI job, for pipelines orchestrated and " "executed by Github.",
+    )
     name: str = Field(description="Name of the job.")
     job_submission_ts: datetime = Field(description="Timestamp with timezone when the job was submitted.")
     job_start_ts: datetime = Field(description="Timestamp with timezone when the job execution started.")
@@ -66,6 +70,8 @@ class Job(BaseModel):
     card_type: Optional[str] = Field(description="Card type and version.")
     os: Optional[str] = Field(description="Operating system of the host.")
     location: Optional[str] = Field(description="Where the host is located.")
+    failure_signature: Optional[str] = Field(None, description="Failure signature.")
+    failure_description: Optional[str] = Field(None, description="Failure description.")
     tests: List[Test] = []
 
 
@@ -82,6 +88,10 @@ class Pipeline(BaseModel):
         description="Identifier for the Github Actions CI pipeline, for pipelines "
         "orchestrated and executed by Github.",
     )
+    github_pipeline_link: Optional[str] = Field(
+        None,
+        description="Link to the Github Actions CI pipeline, for pipelines " "orchestrated and executed by Github.",
+    )
     pipeline_submission_ts: datetime = Field(
         description="Timestamp with timezone when the pipeline was submitted for " "execution.",
     )