From 9ac1ed967cd0042f0761f6da85bfa480542d797a Mon Sep 17 00:00:00 2001
From: Kara Moraw <k.moraw@epcc.ed.ac.uk>
Date: Fri, 5 Jul 2024 16:33:23 +0100
Subject: [PATCH] update GitHub actions

---
 .github/workflows/test.yml | 36 ++++++++++++++++++------------------
 .gitignore                 |  1 +
 data/debug/test_set.csv    | 11 +++++++++++
 3 files changed, 30 insertions(+), 18 deletions(-)
 create mode 100644 data/debug/test_set.csv

diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
index 80b28c1..0216475 100644
--- a/.github/workflows/test.yml
+++ b/.github/workflows/test.yml
@@ -21,8 +21,8 @@ jobs:
           cp config_example.cfg config.cfg
           sed -i "s/<your-access-token>/${{ secrets.TOKEN_TEST }}/" config.cfg
       - name: Test crawling contents script
-        working-directory: ./github
-        run: python3 crawl_contents.py -f ../data/test_set.csv -n github_id -v
+        working-directory: ./src/github
+        run: python3 crawl_contents.py -f ../../data/debug/test_set.csv -n github_id -v
       - run: ls -l data/
   github-crawl-contributions:
     runs-on: ubuntu-latest
@@ -43,9 +43,9 @@ jobs:
           cp config_example.cfg config.cfg
           sed -i "s/<your-access-token>/${{ secrets.TOKEN_TEST }}/" config.cfg
       - name: Test crawling contributions script
-        working-directory: ./github
-        run: python3 crawl_contributions.py -f ../data/test_set.csv -n github_id -v
-      - run: ls -l data/
+        working-directory: ./src/github
+        run: python3 crawl_contributions.py -f ../../data/debug/test_set.csv -n github_id -v
+      - run: ls -l data/raw/github
   github-crawl-engagement:
     runs-on: ubuntu-latest
     defaults:
@@ -65,9 +65,9 @@ jobs:
           cp config_example.cfg config.cfg
           sed -i "s/<your-access-token>/${{ secrets.TOKEN_TEST }}/" config.cfg
       - name: Test crawling engagement script
-        working-directory: ./github
-        run: python3 crawl_engagement.py -f ../data/test_set.csv -n github_id -v
-      - run: ls -l data/        
+        working-directory: ./src/github
+        run: python3 crawl_engagement.py -f ../../data/debug/test_set.csv -n github_id -v
+      - run: ls -l data/raw/github       
   github-crawl-issues:
     runs-on: ubuntu-latest
     defaults:
@@ -87,9 +87,9 @@ jobs:
           cp config_example.cfg config.cfg
           sed -i "s/<your-access-token>/${{ secrets.TOKEN_TEST }}/" config.cfg
       - name: Test crawling issues script
-        working-directory: ./github
-        run: python3 crawl_issues.py -f ../data/test_set.csv -n github_id -v
-      - run: ls -l data/        
+        working-directory: ./src/github
+        run: python3 crawl_issues.py -f ../../data/debug/test_set.csv -n github_id -v
+      - run: ls -l data/raw/github    
   github-crawl-metadata:
     runs-on: ubuntu-latest
     defaults:
@@ -109,9 +109,9 @@ jobs:
           cp config_example.cfg config.cfg
           sed -i "s/<your-access-token>/${{ secrets.TOKEN_TEST }}/" config.cfg
       - name: Test crawling issues script
-        working-directory: ./github
-        run: python3 crawl_metadata.py -f ../data/test_set.csv -n github_id -v
-      - run: ls -l data/             
+        working-directory: ./src/github
+        run: python3 crawl_metadata.py -f ../../data/debug/test_set.csv -n github_id -v
+      - run: ls -l data/raw/github         
   eprints-parse:
     runs-on: ubuntu-latest
     defaults:
@@ -131,12 +131,12 @@ jobs:
           cp config_example.cfg config.cfg
           sed -i "s/<your-access-token>/${{ secrets.TOKEN_TEST }}/" config.cfg
       - name: Test parsing remote eprints
-        working-directory: ./eprints
+        working-directory: ./src/eprints
         run: python3 parse_eprints.py --repo eprints.soton.ac.uk --date 2022-2023 -v
       - name: Test parsing PDFs
-        working-directory: ./eprints
+        working-directory: ./src/eprints
         run: python3 parse_pdfs.py --repo eprints.soton.ac.uk --date 2022-2023 --domain github.com -v
       - name: Test cleaning links
-        working-directory: ./eprints
+        working-directory: ./src/eprints
         run: python3 clean_eprints_links.py --repo eprints.soton.ac.uk --date 2022-2023 --domain github.com -v
-      - run: ls -l data/
\ No newline at end of file
+      - run: ls -l data/raw/eprints
\ No newline at end of file
diff --git a/.gitignore b/.gitignore
index 9d90c08..1ebc5ba 100644
--- a/.gitignore
+++ b/.gitignore
@@ -141,6 +141,7 @@ dmypy.json
 
 data/debug/*
 !data/debug/representative_set.csv
+!data/debug/test_set.csv
 
 # tex
 *.aux
diff --git a/data/debug/test_set.csv b/data/debug/test_set.csv
new file mode 100644
index 0000000..2e00bf2
--- /dev/null
+++ b/data/debug/test_set.csv
@@ -0,0 +1,11 @@
+user_name,repo_name,stars,watchers,forks,commits_no,contributors_no,size_kb,github_id
+ObaidaNa,albaath-marks,0,1,0,5,1,630,ObaidaNa/albaath-marks
+ja153903,jabbariao-dotcom-api,0,1,0,16,1,36,ja153903/jabbariao-dotcom-api
+drone,awesome-drone,100,10,17,22,7,17,drone/awesome-drone
+syun0216,goforeat,100,6,31,1011,2,240094,syun0216/goforeat
+ng-matero,ng-matero,1000,36,293,1391,10,88339,ng-matero/ng-matero
+cov-lineages,pango-designation,1000,91,85,2683,15,1681848,cov-lineages/pango-designation
+ellie,atuin,9985,44,268,527,109,42795,ellie/atuin
+tiangolo,sqlmodel,9978,132,419,236,53,4835,tiangolo/sqlmodel
+996icu,996.ICU,40000,4259,21210,3189,405,188163,996icu/996.ICU
+codecrafters-io,build-your-own-x,40000,4510,19327,523,104,971,codecrafters-io/build-your-own-x