feat(#204): experiment with QUERY

.github/workflows/datasets.yml

@@ -60,6 +60,7 @@ jobs:
       - name: Create
         run: |
           docker run  --rm -v "$(pwd)/output:/collection" \
+            -e QUERY="stars:>10 size:>=20 mirror:false template:false NOT android" \
             -e START="${{ inputs.start }}" -e END="${{ inputs.end }}" \
             -e COLLECT_TOKEN="${{ secrets.COLLECT_TOKEN_1 }}" \
             -e GH_TOKEN="${{ secrets.GITHUB_TOKEN }}" \

Dockerfile

@@ -55,6 +55,7 @@ RUN curl https://sh.rustup.rs -sSf | bash -s -- -y \
 
 # Run variables.
 ENV RUN="/root/.cargo/bin" \
+    QUERY="stars:>10 language:java size:>=20 mirror:false template:false NOT android" \
     PATS="pats.txt" \
     OUT="data"
 

data.sh

@@ -24,7 +24,7 @@ echo "$COLLECT_TOKEN" >> "$PATS"
 {
   "$RUN"/just install
   poetry install
-  "$RUN"/just collect "collection/$OUT" "$START" "$END" "repos"
+  "$RUN"/just collect "collection/$OUT" "$QUERY" "$START" "$END" "repos"
   "$RUN"/just pulls "../repos.csv" "$GH_TOKEN" "../repos-with-pulls.csv"
   "$RUN"/just filter "../repos-with-pulls.csv" "../after-filter.csv"
   "$RUN"/just extract "../after-filter.csv" "../after-extract.csv"

justfile

@@ -62,10 +62,10 @@ clean:
   rm sr-data/experiment/* && rmdir sr-data/experiment
 
 # Collect repositories.
-collect dir start end out:
+collect dir query start end out:
   mkdir -p {{dir}}
-  ghminer --query "stars:>10 language:java size:>=20 mirror:false template:false NOT android" \
-    --start "{{start}}" --end "{{end}}" --tokens "$PATS" --filename "{{out}}"
+  ghminer --query "{{query}}" --start "{{start}}" --end "{{end}}" \
+   --tokens "$PATS" --filename "{{out}}"
 
 # Fetch pulls count for collected repos.
 pulls repos token out="experiment/with-pulls.csv":