From dce383911dfb94e9d0093ea45b6b393e9b7e2c39 Mon Sep 17 00:00:00 2001
From: h1alexbel <aliaksei.bialiauski@hey.com>
Date: Thu, 6 Jun 2024 10:44:17 +0300
Subject: [PATCH] feat(#7): metrics.sh, metrics.py, install reqs

---
 steps/install.sh                   |  1 +
 steps/{structure.sh => metrics.py} | 30 ++++++++++++++++++++++++++----
 steps/metrics.sh                   |  3 ++-
 steps/requirements.txt             |  1 +
 4 files changed, 30 insertions(+), 5 deletions(-)
 rename steps/{structure.sh => metrics.py} (60%)
 create mode 100644 steps/requirements.txt

diff --git a/steps/install.sh b/steps/install.sh
index 3e31ec1..cded1b0 100644
--- a/steps/install.sh
+++ b/steps/install.sh
@@ -24,4 +24,5 @@
 set -e
 
 npm install -g ghminer@0.0.3
+pip3 install -r requirements.txt
 echo "All dependencies are installed and up to date! Now you can run 'make' and build the dataset."
diff --git a/steps/structure.sh b/steps/metrics.py
similarity index 60%
rename from steps/structure.sh
rename to steps/metrics.py
index 30c8a45..57f436c 100644
--- a/steps/structure.sh
+++ b/steps/metrics.py
@@ -1,4 +1,3 @@
-#!/usr/bin/env bash
 # The MIT License (MIT)
 #
 # Copyright (c) 2024 Aliaksei Bialiauski
@@ -20,7 +19,30 @@
 # LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 # SOFTWARE.
+import pandas as pd
 
-set -e
-set -o pipefail
-
+frame = pd.read_csv('results.csv')
+frame["ic"] = frame["issues"] / frame["commits"]
+# @todo #7:30min Compute CPD and RC metrics too.
+#  We need to compute CPD and RC metrics too, however for now we don't have
+#  the proper dataset after ghminer execution. Need to collect required data
+#  first, and then compute CPDs and RCs. Don't forget to remove this puzzle.
+frame.drop(
+    columns=[
+        "branch",
+        "createdAt",
+        "lastCommitDate",
+        "lastReleaseDate",
+        "contributors",
+        "pulls",
+        "commits",
+        "issues",
+        "forks",
+        "stars",
+        "license",
+        "language",
+        "diskUsage"
+    ],
+    inplace=True
+)
+frame.to_csv("repos.csv", index=False)
diff --git a/steps/metrics.sh b/steps/metrics.sh
index d086c0e..87bced3 100644
--- a/steps/metrics.sh
+++ b/steps/metrics.sh
@@ -24,4 +24,5 @@
 set -e
 set -o pipefail
 
-# CPD, RC, IC
+# Compute CPD, RC, IC metrics.
+python3 metrics.py
diff --git a/steps/requirements.txt b/steps/requirements.txt
new file mode 100644
index 0000000..4b2e332
--- /dev/null
+++ b/steps/requirements.txt
@@ -0,0 +1 @@
+pandas==2.2.2