Move datasets to assets (#44)

LSYS · Apr 1, 2024 · 4fd3a01 · 4fd3a01
1 parent 9b08ef6
commit 4fd3a01
Show file tree

Hide file tree

Showing 59 changed files with 16 additions and 14 deletions.
diff --git a/Makefile b/Makefile
@@ -13,7 +13,7 @@ test: ## Run tests with pytest and coverage
 	@rm -rf temp
 	make clean-dta
 	# Git checkout census.dta
-	git checkout datasets/census.dta
+	git checkout assets/datasets/census.dta
 
 .PHONY: lint
 MYPY_OPTS := --ignore-missing-imports

diff --git a/datasets/Makefile → assets/datasets/Makefile b/datasets/Makefile → assets/datasets/Makefile
diff --git a/datasets/README.md → assets/datasets/README.md b/datasets/README.md → assets/datasets/README.md
diff --git a/datasets/auto.dta → assets/datasets/auto.dta b/datasets/auto.dta → assets/datasets/auto.dta
diff --git a/datasets/auto2.dta → assets/datasets/auto2.dta b/datasets/auto2.dta → assets/datasets/auto2.dta
diff --git a/datasets/autornd.dta → assets/datasets/autornd.dta b/datasets/autornd.dta → assets/datasets/autornd.dta
diff --git a/datasets/bplong.dta → assets/datasets/bplong.dta b/datasets/bplong.dta → assets/datasets/bplong.dta
diff --git a/datasets/cancer.dta → assets/datasets/cancer.dta b/datasets/cancer.dta → assets/datasets/cancer.dta
diff --git a/assets/datasets/census.dta b/assets/datasets/census.dta
diff --git a/datasets/citytemp.dta → assets/datasets/citytemp.dta b/datasets/citytemp.dta → assets/datasets/citytemp.dta
diff --git a/datasets/educ99gdp.dta → assets/datasets/educ99gdp.dta b/datasets/educ99gdp.dta → assets/datasets/educ99gdp.dta
diff --git a/datasets/getdata.do → assets/datasets/getdata.do b/datasets/getdata.do → assets/datasets/getdata.do
diff --git a/datasets/gnp96.dta → assets/datasets/gnp96.dta b/datasets/gnp96.dta → assets/datasets/gnp96.dta
diff --git a/datasets/lifeexp.dta → assets/datasets/lifeexp.dta b/datasets/lifeexp.dta → assets/datasets/lifeexp.dta
diff --git a/datasets/manifest.txt → assets/datasets/manifest.txt b/datasets/manifest.txt → assets/datasets/manifest.txt
diff --git a/datasets/nickchk-causaldata/Mroz.dta → assets/datasets/nickchk-causaldata/Mroz.dta b/datasets/nickchk-causaldata/Mroz.dta → assets/datasets/nickchk-causaldata/Mroz.dta
diff --git a/datasets/nickchk-causaldata/abortion.dta → .../datasets/nickchk-causaldata/abortion.dta b/datasets/nickchk-causaldata/abortion.dta → .../datasets/nickchk-causaldata/abortion.dta
diff --git a/...ets/nickchk-causaldata/adult_services.dta → ...ets/nickchk-causaldata/adult_services.dta b/...ets/nickchk-causaldata/adult_services.dta → ...ets/nickchk-causaldata/adult_services.dta
diff --git a/datasets/nickchk-causaldata/avocado.dta → ...s/datasets/nickchk-causaldata/avocado.dta b/datasets/nickchk-causaldata/avocado.dta → ...s/datasets/nickchk-causaldata/avocado.dta
diff --git a/.../nickchk-causaldata/black_politicians.dta → .../nickchk-causaldata/black_politicians.dta b/.../nickchk-causaldata/black_politicians.dta → .../nickchk-causaldata/black_politicians.dta
diff --git a/datasets/nickchk-causaldata/castle.dta → ...ts/datasets/nickchk-causaldata/castle.dta b/datasets/nickchk-causaldata/castle.dta → ...ts/datasets/nickchk-causaldata/castle.dta
diff --git a/...sets/nickchk-causaldata/close_college.dta → ...sets/nickchk-causaldata/close_college.dta b/...sets/nickchk-causaldata/close_college.dta → ...sets/nickchk-causaldata/close_college.dta
diff --git a/...ickchk-causaldata/close_elections_lmb.dta → ...ickchk-causaldata/close_elections_lmb.dta b/...ickchk-causaldata/close_elections_lmb.dta → ...ickchk-causaldata/close_elections_lmb.dta
diff --git a/datasets/nickchk-causaldata/cps_mixtape.dta → ...tasets/nickchk-causaldata/cps_mixtape.dta b/datasets/nickchk-causaldata/cps_mixtape.dta → ...tasets/nickchk-causaldata/cps_mixtape.dta
diff --git a/datasets/nickchk-causaldata/credit_cards.dta → ...asets/nickchk-causaldata/credit_cards.dta b/datasets/nickchk-causaldata/credit_cards.dta → ...asets/nickchk-causaldata/credit_cards.dta
diff --git a/datasets/nickchk-causaldata/gapminder.dta → ...datasets/nickchk-causaldata/gapminder.dta b/datasets/nickchk-causaldata/gapminder.dta → ...datasets/nickchk-causaldata/gapminder.dta
diff --git a/datasets/nickchk-causaldata/google_stock.dta → ...asets/nickchk-causaldata/google_stock.dta b/datasets/nickchk-causaldata/google_stock.dta → ...asets/nickchk-causaldata/google_stock.dta
diff --git a/...sets/nickchk-causaldata/gov_transfers.dta → ...sets/nickchk-causaldata/gov_transfers.dta b/...sets/nickchk-causaldata/gov_transfers.dta → ...sets/nickchk-causaldata/gov_transfers.dta
diff --git a/...kchk-causaldata/gov_transfers_density.dta → ...kchk-causaldata/gov_transfers_density.dta b/...kchk-causaldata/gov_transfers_density.dta → ...kchk-causaldata/gov_transfers_density.dta
diff --git a/datasets/nickchk-causaldata/greek_data.dta → ...atasets/nickchk-causaldata/greek_data.dta b/datasets/nickchk-causaldata/greek_data.dta → ...atasets/nickchk-causaldata/greek_data.dta
diff --git a/datasets/nickchk-causaldata/mortgages.dta → ...datasets/nickchk-causaldata/mortgages.dta b/datasets/nickchk-causaldata/mortgages.dta → ...datasets/nickchk-causaldata/mortgages.dta
diff --git a/datasets/nickchk-causaldata/nhefs.dta → assets/datasets/nickchk-causaldata/nhefs.dta b/datasets/nickchk-causaldata/nhefs.dta → assets/datasets/nickchk-causaldata/nhefs.dta
diff --git a/...ets/nickchk-causaldata/nhefs_complete.dta → ...ets/nickchk-causaldata/nhefs_complete.dta b/...ets/nickchk-causaldata/nhefs_complete.dta → ...ets/nickchk-causaldata/nhefs_complete.dta
diff --git a/datasets/nickchk-causaldata/nsw_mixtape.dta → ...tasets/nickchk-causaldata/nsw_mixtape.dta b/datasets/nickchk-causaldata/nsw_mixtape.dta → ...tasets/nickchk-causaldata/nsw_mixtape.dta
diff --git a/...ts/nickchk-causaldata/organ_donations.dta → ...ts/nickchk-causaldata/organ_donations.dta b/...ts/nickchk-causaldata/organ_donations.dta → ...ts/nickchk-causaldata/organ_donations.dta
diff --git a/...chk-causaldata/restaurant_inspections.dta → ...chk-causaldata/restaurant_inspections.dta b/...chk-causaldata/restaurant_inspections.dta → ...chk-causaldata/restaurant_inspections.dta
diff --git a/datasets/nickchk-causaldata/ri.dta → assets/datasets/nickchk-causaldata/ri.dta b/datasets/nickchk-causaldata/ri.dta → assets/datasets/nickchk-causaldata/ri.dta
diff --git a/datasets/nickchk-causaldata/scorecard.dta → ...datasets/nickchk-causaldata/scorecard.dta b/datasets/nickchk-causaldata/scorecard.dta → ...datasets/nickchk-causaldata/scorecard.dta
diff --git a/datasets/nickchk-causaldata/snow.dta → assets/datasets/nickchk-causaldata/snow.dta b/datasets/nickchk-causaldata/snow.dta → assets/datasets/nickchk-causaldata/snow.dta
diff --git a/...sets/nickchk-causaldata/social_insure.dta → ...sets/nickchk-causaldata/social_insure.dta b/...sets/nickchk-causaldata/social_insure.dta → ...sets/nickchk-causaldata/social_insure.dta
diff --git a/datasets/nickchk-causaldata/texas.dta → assets/datasets/nickchk-causaldata/texas.dta b/datasets/nickchk-causaldata/texas.dta → assets/datasets/nickchk-causaldata/texas.dta
diff --git a/datasets/nickchk-causaldata/thornton_hiv.dta → ...asets/nickchk-causaldata/thornton_hiv.dta b/datasets/nickchk-causaldata/thornton_hiv.dta → ...asets/nickchk-causaldata/thornton_hiv.dta
diff --git a/datasets/nickchk-causaldata/titanic.dta → ...s/datasets/nickchk-causaldata/titanic.dta b/datasets/nickchk-causaldata/titanic.dta → ...s/datasets/nickchk-causaldata/titanic.dta
diff --git a/...hk-causaldata/training_bias_reduction.dta → ...hk-causaldata/training_bias_reduction.dta b/...hk-causaldata/training_bias_reduction.dta → ...hk-causaldata/training_bias_reduction.dta
diff --git a/...s/nickchk-causaldata/training_example.dta → ...s/nickchk-causaldata/training_example.dta b/...s/nickchk-causaldata/training_example.dta → ...s/nickchk-causaldata/training_example.dta
diff --git a/datasets/nickchk-causaldata/yule.dta → assets/datasets/nickchk-causaldata/yule.dta b/datasets/nickchk-causaldata/yule.dta → assets/datasets/nickchk-causaldata/yule.dta
diff --git a/datasets/nlsw88.dta → assets/datasets/nlsw88.dta b/datasets/nlsw88.dta → assets/datasets/nlsw88.dta
diff --git a/datasets/pop2000.dta → assets/datasets/pop2000.dta b/datasets/pop2000.dta → assets/datasets/pop2000.dta
diff --git a/datasets/sp500.dta → assets/datasets/sp500.dta b/datasets/sp500.dta → assets/datasets/sp500.dta
diff --git a/datasets/testopen-converted-log.txt → assets/datasets/testopen-converted-log.txt b/datasets/testopen-converted-log.txt → assets/datasets/testopen-converted-log.txt
diff --git a/datasets/testopen-converted.do → assets/datasets/testopen-converted.do b/datasets/testopen-converted.do → assets/datasets/testopen-converted.do
diff --git a/datasets/testopen-log.txt → assets/datasets/testopen-log.txt b/datasets/testopen-log.txt → assets/datasets/testopen-log.txt
diff --git a/datasets/testopen.do → assets/datasets/testopen.do b/datasets/testopen.do → assets/datasets/testopen.do
diff --git a/datasets/txt2macro.ado → assets/datasets/txt2macro.ado b/datasets/txt2macro.ado → assets/datasets/txt2macro.ado
diff --git a/datasets/uslifeexp.dta → assets/datasets/uslifeexp.dta b/datasets/uslifeexp.dta → assets/datasets/uslifeexp.dta
diff --git a/datasets/uslifeexp2.dta → assets/datasets/uslifeexp2.dta b/datasets/uslifeexp2.dta → assets/datasets/uslifeexp2.dta
diff --git a/datasets/voter.dta → assets/datasets/voter.dta b/datasets/voter.dta → assets/datasets/voter.dta
diff --git a/datasets/census.dta b/datasets/census.dta
diff --git a/tests/test_wbstata.py b/tests/test_wbstata.py
@@ -14,6 +14,8 @@
 )
 
 
+DATAPATH = "assets/datasets"
+
 def test_normalize_dta_filename():
     expected = "census.dta"
 
@@ -25,13 +27,13 @@ def test_normalize_dta_filename():
     result = normalize_dta_filename(filename)
     assert result == expected
 
-    expected = "datasets/census.dta"
+    expected = f"{DATAPATH}/census.dta"
 
-    filename = "datasets/census"
+    filename = f"{DATAPATH}/census"
     result = normalize_dta_filename(filename)
     assert result == expected
 
-    filename = "datasets/census.dta"
+    filename = f"{DATAPATH}/census.dta"
     result = normalize_dta_filename(filename)
     assert result == expected
 
@@ -56,7 +58,7 @@ def test_add_suffix():
 
 
 def test_get_output_name():
-    file = "census.dta"
+    file = f"{DATAPATH}/census.dta"
 
     # Overwriting
     expected = file
@@ -83,7 +85,7 @@ def test_get_output_name():
 
 
 def test_is_dta_file():
-    valid_file = "datasets/census.dta"
+    valid_file = f"{DATAPATH}/census.dta"
     is_dta_file(valid_file)
 
     invalid_file = "wrongfile.dta"
@@ -108,8 +110,8 @@ def test_glob_dta_files():
 def test_convert_dta():
     for version in range(10, 17 + 1):
         convert_dta(
-            "datasets/census.dta",
-            "datasets/test-output.dta",
+            f"{DATAPATH}/census.dta",
+            f"{DATAPATH}/test-output.dta",
             target_version=version,
         )
 
@@ -135,13 +137,13 @@ def test_wbstata():
     )
 
     # Check that prompt works with just file (prompt for version)
-    result = runner.invoke(wbstata, ["datasets/census.dta"])
+    result = runner.invoke(wbstata, [f"{DATAPATH}/census.dta"])
     assert result.exit_code == 0
     assert "> Target version" in result.output
 
     # Check minimal command
-    dta = "datasets/census.dta"
-    expected_output = f"{dta} to datasets/census-wbstata.dta in version 13.\n"
+    dta = f"{DATAPATH}/census.dta"
+    expected_output = f"{dta} to {DATAPATH}/census-wbstata.dta in version 13.\n"
     result = runner.invoke(
         wbstata, [f"{dta}", "--target-version", "13", "--verbose"]
     )
@@ -150,9 +152,9 @@ def test_wbstata():
     assert COMPLETION_MSG in result.output
 
     # Check multiple
-    dta1 = "datasets/census.dta"
-    dta2 = "datasets/auto.dta"
-    dta3 = "datasets/lifeexp.dta"
+    dta1 = f"{DATAPATH}/census.dta"
+    dta2 = f"{DATAPATH}/auto.dta"
+    dta3 = f"{DATAPATH}/lifeexp.dta"
     result = runner.invoke(
         wbstata,
         [