study-config.json

{
  "$schema": "schemas/0.1.2/study-config.schema.json",
  "datasets": {
    "kotlin": "2021 Aug/Kotlin",
    "python": "2021 Aug/Python",
    "python-ds": "2020 August/Python-DS",
    "java": "2019 October/GitHub",
    "java-sf": "2013 September/SF",
    "original": "2012 July/SF"
  },
  "queries": {
    "kotlin/hashes.txt": {
      "query": "queries/hashes.boa",
      "dataset": "kotlin",
      "processors": {
        "gendupes.py": {
          "output": "data/txt/kotlin/dupes.txt",
          "csv": "kotlin/dupes.csv",
          "cacheclean": [
            "kotlin/*-deduped.parquet"
          ]
        }
      }
    },
    "kotlin/rq1.txt": {
      "query": "queries/rq1.boa",
      "dataset": "kotlin",
      "csv": {
        "output": "kotlin/rq1.csv",
        "test": [
          "3,\\.kts?$"
        ]
      }
    },
    "kotlin/project-count.txt": {
      "query": "queries/project-count.boa",
      "dataset": "kotlin",
      "csv": {
        "output": "kotlin/project-count.csv",
        "drop": [
          1
        ]
      }
    },
    "python/project-count.txt": {
      "query": "queries/project-count.boa",
      "dataset": "python",
      "csv": {
        "output": "python/project-count.csv",
        "drop": [
          1
        ]
      }
    }
  },
  "substitutions": [
    {
      "target": "{@escape@}",
      "file": "escape.boa"
    },
    {
      "target": "{@project-filter@}",
      "replacement": "if (input.stars >= 5)"
    }
  ],
  "analyses": {
    "rq1.py": {
      "input": [
        "kotlin/rq1.csv",
        "kotlin/dupes.csv"
      ]
    }
  }
}