diff --git a/.github/workflows/lint_sqlfluff.yml b/.github/workflows/lint_sqlfluff.yml index 0986785b..5b603162 100644 --- a/.github/workflows/lint_sqlfluff.yml +++ b/.github/workflows/lint_sqlfluff.yml @@ -6,6 +6,9 @@ on: jobs: lint-mimic-iv: runs-on: ubuntu-latest + permissions: + contents: read + checks: write steps: - name: checkout uses: actions/checkout@v7 @@ -35,8 +38,10 @@ jobs: shell: bash run: sqlfluff lint --format github-annotation --annotation-level failure --nofail ${{ steps.get_files_to_lint.outputs.lintees }} > annotations.json - name: Annotate + if: steps.get_files_to_lint.outputs.lintees != '' uses: yuzutech/annotations-action@v0.6.0 with: repo-token: "${{ secrets.GITHUB_TOKEN }}" title: "SQLFluff Lint" - input: "./annotations.json" \ No newline at end of file + input: "./annotations.json" + ignore-unauthorized-error: true diff --git a/mimic-iii/buildmimic/duckdb/import_duckdb.sh b/mimic-iii/buildmimic/duckdb/import_duckdb.sh index 3eb1e129..1069f790 100755 --- a/mimic-iii/buildmimic/duckdb/import_duckdb.sh +++ b/mimic-iii/buildmimic/duckdb/import_duckdb.sh @@ -85,9 +85,11 @@ make_table_name () { # load data into database find "$MIMIC_DIR" -type f -regex '.*\.csv\(.gz\)*' | while IFS= read -r FILE; do make_table_name "$FILE" - echo "Loading $FILE .. \c" + # Escape single quotes for SQL string literal + FILE_SQL=$(printf '%s' "$FILE" | sed "s/'/''/g") + printf "Loading %s .. " "$FILE" try duckdb "$OUTFILE" <<-EOSQL - COPY $TABLE_NAME FROM '$FILE' (HEADER, DELIM ',', QUOTE '"', ESCAPE '"'); + COPY $TABLE_NAME FROM '$FILE_SQL' (HEADER, DELIM ',', QUOTE '"', ESCAPE '"'); EOSQL echo "done!" done && echo "Successfully finished loading data into $OUTFILE." diff --git a/mimic-iii/buildmimic/postgres/Makefile b/mimic-iii/buildmimic/postgres/Makefile index 15b4c521..83a61ee5 100644 --- a/mimic-iii/buildmimic/postgres/Makefile +++ b/mimic-iii/buildmimic/postgres/Makefile @@ -104,7 +104,7 @@ mimic-build-gz: @echo '------------------' @echo '' @sleep 2 - psql "$(DBSTRING)" -v ON_ERROR_STOP=1 -f postgres_load_data_gz.sql -v mimic_data_dir=${datadir} + psql "$(DBSTRING)" -v ON_ERROR_STOP=1 -f postgres_load_data_gz.sql -v "mimic_data_dir=$(DATADIR)" @echo '' @echo '--------------------' @echo '-- Adding indexes --' @@ -151,7 +151,7 @@ mimic-build: @echo '------------------' @echo '' @sleep 2 - psql "$(DBSTRING)" -v ON_ERROR_STOP=1 -f postgres_load_data.sql -v mimic_data_dir=${DATADIR} + psql "$(DBSTRING)" -v ON_ERROR_STOP=1 -f postgres_load_data.sql -v "mimic_data_dir=$(DATADIR)" @echo '' @echo '--------------------' @echo '-- Adding indexes --' @@ -184,7 +184,7 @@ ifeq ("$(physionetuser)","") @echo 'Call the makefile again with physionetuser=' @echo ' e.g. make eicu-download datadir=/path/to/data physionetuser=hello@physionet.org' else - wget --user $(physionetuser) --ask-password -P $(DATADIR) -A csv.gz -m -p -E -k -K -np -nd "$(PHYSIONETURL)" + wget --user $(physionetuser) --ask-password -P "$(DATADIR)" -A csv.gz -m -p -E -k -K -np -nd "$(PHYSIONETURL)" endif mimic-demo-download: @@ -192,7 +192,7 @@ mimic-demo-download: @echo '-- Downloading MIMIC-III from PhysioNet --' @echo '------------------------------------------' @echo '' - wget --user $(physionetuser) --ask-password -P $(DATADIR) -A csv.gz -m -p -E -k -K -np -nd "$(PHYSIONETDEMOURL)" + wget --user $(physionetuser) --ask-password -P "$(DATADIR)" -A csv.gz -m -p -E -k -K -np -nd "$(PHYSIONETDEMOURL)" #This is fairly inelegant and could be tidied with a for loop and an if to check for gzip, #but need to maintain compatibility with Windows, which baffling lacks these things diff --git a/mimic-iii/buildmimic/postgres/create_mimic_user.sh b/mimic-iii/buildmimic/postgres/create_mimic_user.sh index 9b333dcf..838ddc9e 100755 --- a/mimic-iii/buildmimic/postgres/create_mimic_user.sh +++ b/mimic-iii/buildmimic/postgres/create_mimic_user.sh @@ -20,6 +20,13 @@ else echo "User is set to '$MIMIC_USER'"; fi +# escape ' in password for SQL string literal +MIMIC_PASSWORD_SQL=$(printf '%s' "$MIMIC_PASSWORD" | sed "s/'/''/g") +# quote SQL identifiers (double any embedded ") +sql_ident () { printf '%s' "$1" | sed 's/"/""/g; s/^/"/; s/$/"/'; } +MIMIC_USER_SQL=$(sql_ident "$MIMIC_USER") +MIMIC_DB_SQL=$(sql_ident "$MIMIC_DB") + PSQL='psql' # add in the host/port, if they were specified (not null, -n) @@ -58,11 +65,11 @@ fi if [ "$MIMIC_USER" != "postgres" ]; then # we need to create this user via postgres # use SUDO to login as postgres - $PSQL -U postgres -d postgres -c "DROP USER IF EXISTS $MIMIC_USER; CREATE USER $MIMIC_USER WITH PASSWORD '$MIMIC_PASSWORD';" + $PSQL -U postgres -d postgres -c "DROP USER IF EXISTS $MIMIC_USER_SQL; CREATE USER $MIMIC_USER_SQL WITH PASSWORD '$MIMIC_PASSWORD_SQL';" fi if [ "$MIMIC_DB" != "postgres" ]; then # drop and recreate the database - $PSQL -U postgres -d postgres -c "DROP DATABASE IF EXISTS $MIMIC_DB;" - $PSQL -U postgres -d postgres -c "CREATE DATABASE $MIMIC_DB OWNER $MIMIC_USER;" -fi \ No newline at end of file + $PSQL -U postgres -d postgres -c "DROP DATABASE IF EXISTS $MIMIC_DB_SQL;" + $PSQL -U postgres -d postgres -c "CREATE DATABASE $MIMIC_DB_SQL OWNER $MIMIC_USER_SQL;" +fi diff --git a/mimic-iii/buildmimic/sqlite/import.py b/mimic-iii/buildmimic/sqlite/import.py index 801651c7..66052b2a 100644 --- a/mimic-iii/buildmimic/sqlite/import.py +++ b/mimic-iii/buildmimic/sqlite/import.py @@ -26,9 +26,16 @@ def _table_name_from_csv(filename: str) -> str: print(msg) sys.exit() +# Prefer .csv.gz when both exist for the same table; also load plain .csv +# (import.sh already accepts both). +files_by_table = {} +for f in glob("*.csv"): + files_by_table[_table_name_from_csv(f)] = f for f in glob("*.csv.gz"): + files_by_table[_table_name_from_csv(f)] = f + +for table, f in sorted(files_by_table.items()): print("Starting processing {}".format(f)) - table = _table_name_from_csv(f) if os.path.getsize(f) < THRESHOLD_SIZE: df = pd.read_csv(f, index_col="ROW_ID") df.to_sql(table, CONNECTION_STRING) diff --git a/mimic-iii/buildmimic/sqlite/import.sh b/mimic-iii/buildmimic/sqlite/import.sh index dbf6b9ce..58b85abb 100755 --- a/mimic-iii/buildmimic/sqlite/import.sh +++ b/mimic-iii/buildmimic/sqlite/import.sh @@ -28,11 +28,11 @@ for FILE in *; do TABLE_NAME=$(echo "${FILE%%.*}" | tr "[:upper:]" "[:lower:]") case "$FILE" in *csv) - IMPORT_CMD=".import $FILE $TABLE_NAME" + IMPORT_CMD=".import \"$FILE\" $TABLE_NAME" ;; # need to decompress csv before load *csv.gz) - IMPORT_CMD=".import \"|gzip -dc $FILE\" $TABLE_NAME" + IMPORT_CMD=".import \"|gzip -dc \\\"$FILE\\\"\" $TABLE_NAME" ;; # not a data file so skip *) @@ -40,7 +40,7 @@ for FILE in *; do ;; esac echo "Loading $FILE." - sqlite3 $OUTFILE < "$LOAD_COUNT_FILE" +find "$MIMIC_DIR" -type f \( -name '*.csv' -o -name '*.csv.gz' \) | sort | while IFS= read -r FILE; do make_table_name "$FILE" # skip directories which we do not expect in mimic-iv-ed # avoids syntax errors if mimic-iv in the same dir - case $DIRNAME in + case "$DIRNAME" in (ed) ;; # OK (*) continue; esac + FILE_SQL=$(printf '%s' "$FILE" | sed "s/'/''/g") echo "Loading $FILE .. " try duckdb "$OUTFILE" <<-EOSQL - COPY $TABLE_NAME FROM '$FILE' (HEADER, DELIM ',', QUOTE '"', ESCAPE '"'); + COPY $TABLE_NAME FROM '$FILE_SQL' (HEADER, DELIM ',', QUOTE '"', ESCAPE '"'); EOSQL + echo x >> "$LOAD_COUNT_FILE" echo "done!" -done && echo "Successfully finished loading data into $OUTFILE." +done || exit $? + +if [ ! -s "$LOAD_COUNT_FILE" ]; then + die "No .csv / .csv.gz files loaded from $MIMIC_DIR (expected ed/)." +fi +echo "Successfully finished loading data into $OUTFILE." diff --git a/mimic-iv-note/buildmimic/duckdb/import_duckdb.sh b/mimic-iv-note/buildmimic/duckdb/import_duckdb.sh index 20e28313..4db098c1 100644 --- a/mimic-iv-note/buildmimic/duckdb/import_duckdb.sh +++ b/mimic-iv-note/buildmimic/duckdb/import_duckdb.sh @@ -29,7 +29,7 @@ usage () { die " USAGE: ./import_duckdb.sh mimic_data_dir [output_db] WHERE: - mimic_data_dir directory that contains csv.gz or csv files + mimic_data_dir directory that contains csv.tar.gz or csv files output_db: optional filename for duckdb file (default: mimic4_note.db)\ " } @@ -97,18 +97,23 @@ make_table_name () { # load data into database -find "$MIMIC_DIR" -type f -name '*.csv???' | sort | while IFS= read -r FILE; do +# Match both .csv and .csv.gz ( '*.csv???' only matched .csv.gz ). +LOAD_COUNT_FILE=$(mktemp) || die "mktemp failed" +trap 'rm -f "$LOAD_COUNT_FILE"' EXIT +: > "$LOAD_COUNT_FILE" +find "$MIMIC_DIR" -type f \( -name '*.csv' -o -name '*.csv.gz' \) | sort | while IFS= read -r FILE; do make_table_name "$FILE" # skip directories which we do not expect in mimic-iv-note # avoids syntax errors if mimic-iv in the same dir - case $DIRNAME in + case "$DIRNAME" in (note) ;; # OK (*) continue; esac + FILE_SQL=$(printf '%s' "$FILE" | sed "s/'/''/g") echo "Loading $FILE .." OUTPUT=$(duckdb "$OUTFILE" 2>&1 <<-EOSQL - COPY $TABLE_NAME FROM '$FILE' (HEADER, DELIM ',', QUOTE '"', ESCAPE '"'); + COPY $TABLE_NAME FROM '$FILE_SQL' (HEADER, DELIM ',', QUOTE '"', ESCAPE '"'); EOSQL ) # If the table is missing in the DB, we emit a warning and continue. @@ -123,5 +128,11 @@ EOSQL yell "$OUTPUT" die "Exiting due to load error." fi + echo x >> "$LOAD_COUNT_FILE" echo "done!" -done && echo "Successfully finished loading data into $OUTFILE." +done || exit $? + +if [ ! -s "$LOAD_COUNT_FILE" ]; then + die "No .csv / .csv.gz files loaded from $MIMIC_DIR (expected note/)." +fi +echo "Successfully finished loading data into $OUTFILE." diff --git a/mimic-iv/buildmimic/duckdb/import_duckdb.sh b/mimic-iv/buildmimic/duckdb/import_duckdb.sh index 4ff9fc21..3213a8fd 100755 --- a/mimic-iv/buildmimic/duckdb/import_duckdb.sh +++ b/mimic-iv/buildmimic/duckdb/import_duckdb.sh @@ -29,7 +29,7 @@ usage () { die " USAGE: ./import_duckdb.sh mimic_data_dir [output_db] WHERE: - mimic_data_dir directory that contains csv.gz or csv files + mimic_data_dir directory that contains csv.tar.gz or csv files output_db: optional filename for duckdb file (default: mimic4.db)\ " } @@ -102,18 +102,26 @@ make_table_name () { # load data into database -find "$MIMIC_DIR" -type f -name '*.csv???' | sort | while IFS= read -r FILE; do +# Match both uncompressed (.csv) and gzip (.csv.gz). The old '*.csv???' +# pattern only matched names with exactly three chars after ".csv" (i.e. .gz), +# so plain .csv files were silently skipped while the script still reported success. +LOAD_COUNT_FILE=$(mktemp) || die "mktemp failed" +trap 'rm -f "$LOAD_COUNT_FILE"' EXIT +: > "$LOAD_COUNT_FILE" +find "$MIMIC_DIR" -type f \( -name '*.csv' -o -name '*.csv.gz' \) | sort | while IFS= read -r FILE; do make_table_name "$FILE" # skip directories which we do not expect in mimic-iv # avoids syntax errors if mimic-iv-ed in the same dir - case $DIRNAME in + case "$DIRNAME" in (hosp|icu) ;; # OK (*) continue; esac - echo "Loading $FILE .. \c" + # Escape single quotes for SQL string literal + FILE_SQL=$(printf '%s' "$FILE" | sed "s/'/''/g") + printf "Loading %s .. " "$FILE" OUTPUT=$(duckdb "$OUTFILE" 2>&1 <<-EOSQL - COPY $TABLE_NAME FROM '$FILE' (HEADER, DELIM ',', QUOTE '"', ESCAPE '"'); + COPY $TABLE_NAME FROM '$FILE_SQL' (HEADER, DELIM ',', QUOTE '"', ESCAPE '"'); EOSQL ) # If the table is missing in the DB, we emit a warning and continue. @@ -128,5 +136,11 @@ EOSQL yell "$OUTPUT" die "Exiting due to load error." fi + echo x >> "$LOAD_COUNT_FILE" echo "done!" -done && echo "Successfully finished loading data into $OUTFILE." +done || exit $? + +if [ ! -s "$LOAD_COUNT_FILE" ]; then + die "No .csv / .csv.gz files loaded from $MIMIC_DIR (expected hosp/ and icu/)." +fi +echo "Successfully finished loading data into $OUTFILE." diff --git a/mimic-iv/buildmimic/mysql/validate_demo.sql b/mimic-iv/buildmimic/mysql/validate_demo.sql index 575e8413..ebf4f02d 100644 --- a/mimic-iv/buildmimic/mysql/validate_demo.sql +++ b/mimic-iv/buildmimic/mysql/validate_demo.sql @@ -31,14 +31,17 @@ FROM ( SELECT 'poe_detail' AS tbl, 3795 AS row_count UNION ALL SELECT 'prescriptions' AS tbl, 18087 AS row_count UNION ALL SELECT 'procedures_icd' AS tbl, 722 AS row_count UNION ALL + SELECT 'provider' AS tbl, 40508 AS row_count UNION ALL SELECT 'services' AS tbl, 319 AS row_count UNION ALL SELECT 'transfers' AS tbl, 1190 AS row_count UNION ALL -- icu data SELECT 'icustays' AS tbl, 140 AS row_count UNION ALL + SELECT 'caregiver' AS tbl, 15468 AS row_count UNION ALL SELECT 'd_items' AS tbl, 4014 AS row_count UNION ALL SELECT 'chartevents' AS tbl, 668862 AS row_count UNION ALL SELECT 'datetimeevents' AS tbl, 15280 AS row_count UNION ALL SELECT 'inputevents' AS tbl, 20404 AS row_count UNION ALL + SELECT 'ingredientevents' AS tbl, 25728 AS row_count UNION ALL SELECT 'outputevents' AS tbl, 9362 AS row_count UNION ALL SELECT 'procedureevents' AS tbl, 1468 AS row_count ) exp @@ -64,14 +67,17 @@ INNER JOIN SELECT 'poe_detail' AS tbl, count(*) AS row_count FROM poe_detail UNION ALL SELECT 'prescriptions' AS tbl, count(*) AS row_count FROM prescriptions UNION ALL SELECT 'procedures_icd' AS tbl, count(*) AS row_count FROM procedures_icd UNION ALL + SELECT 'provider' AS tbl, count(*) AS row_count FROM provider UNION ALL SELECT 'services' AS tbl, count(*) AS row_count FROM services UNION ALL SELECT 'transfers' AS tbl, count(*) AS row_count FROM transfers UNION ALL -- icu data SELECT 'icustays' AS tbl, count(*) AS row_count FROM icustays UNION ALL + SELECT 'caregiver' AS tbl, count(*) AS row_count FROM caregiver UNION ALL SELECT 'chartevents' AS tbl, count(*) AS row_count FROM chartevents UNION ALL SELECT 'd_items' AS tbl, count(*) AS row_count FROM d_items UNION ALL SELECT 'datetimeevents' AS tbl, count(*) AS row_count FROM datetimeevents UNION ALL SELECT 'inputevents' AS tbl, count(*) AS row_count FROM inputevents UNION ALL + SELECT 'ingredientevents' AS tbl, count(*) AS row_count FROM ingredientevents UNION ALL SELECT 'outputevents' AS tbl, count(*) AS row_count FROM outputevents UNION ALL SELECT 'procedureevents' AS tbl, count(*) AS row_count FROM procedureevents ) obs diff --git a/mimic-iv/buildmimic/postgres/validate_demo.sql b/mimic-iv/buildmimic/postgres/validate_demo.sql index 303dd98e..049497e5 100644 --- a/mimic-iv/buildmimic/postgres/validate_demo.sql +++ b/mimic-iv/buildmimic/postgres/validate_demo.sql @@ -22,14 +22,17 @@ WITH expected AS SELECT 'poe_detail' AS tbl, 3795 AS row_count UNION ALL SELECT 'prescriptions' AS tbl, 18087 AS row_count UNION ALL SELECT 'procedures_icd' AS tbl, 722 AS row_count UNION ALL + SELECT 'provider' AS tbl, 40508 AS row_count UNION ALL SELECT 'services' AS tbl, 319 AS row_count UNION ALL SELECT 'transfers' AS tbl, 1190 AS row_count UNION ALL -- icu data SELECT 'icustays' AS tbl, 140 AS row_count UNION ALL + SELECT 'caregiver' AS tbl, 15468 AS row_count UNION ALL SELECT 'd_items' AS tbl, 4014 AS row_count UNION ALL SELECT 'chartevents' AS tbl, 668862 AS row_count UNION ALL SELECT 'datetimeevents' AS tbl, 15280 AS row_count UNION ALL SELECT 'inputevents' AS tbl, 20404 AS row_count UNION ALL + SELECT 'ingredientevents' AS tbl, 25728 AS row_count UNION ALL SELECT 'outputevents' AS tbl, 9362 AS row_count UNION ALL SELECT 'procedureevents' AS tbl, 1468 AS row_count ) @@ -54,14 +57,17 @@ WITH expected AS SELECT 'poe_detail' AS tbl, count(*) AS row_count FROM mimiciv_hosp.poe_detail UNION ALL SELECT 'prescriptions' AS tbl, count(*) AS row_count FROM mimiciv_hosp.prescriptions UNION ALL SELECT 'procedures_icd' AS tbl, count(*) AS row_count FROM mimiciv_hosp.procedures_icd UNION ALL + SELECT 'provider' AS tbl, count(*) AS row_count FROM mimiciv_hosp.provider UNION ALL SELECT 'services' AS tbl, count(*) AS row_count FROM mimiciv_hosp.services UNION ALL SELECT 'transfers' AS tbl, count(*) AS row_count FROM mimiciv_hosp.transfers UNION ALL -- icu data SELECT 'icustays' AS tbl, count(*) AS row_count FROM mimiciv_icu.icustays UNION ALL + SELECT 'caregiver' AS tbl, count(*) AS row_count FROM mimiciv_icu.caregiver UNION ALL SELECT 'chartevents' AS tbl, count(*) AS row_count FROM mimiciv_icu.chartevents UNION ALL SELECT 'd_items' AS tbl, count(*) AS row_count FROM mimiciv_icu.d_items UNION ALL SELECT 'datetimeevents' AS tbl, count(*) AS row_count FROM mimiciv_icu.datetimeevents UNION ALL SELECT 'inputevents' AS tbl, count(*) AS row_count FROM mimiciv_icu.inputevents UNION ALL + SELECT 'ingredientevents' AS tbl, count(*) AS row_count FROM mimiciv_icu.ingredientevents UNION ALL SELECT 'outputevents' AS tbl, count(*) AS row_count FROM mimiciv_icu.outputevents UNION ALL SELECT 'procedureevents' AS tbl, count(*) AS row_count FROM mimiciv_icu.procedureevents ) diff --git a/mimic-iv/buildmimic/sqlite/import.py b/mimic-iv/buildmimic/sqlite/import.py index 68d43647..31b880c4 100644 --- a/mimic-iv/buildmimic/sqlite/import.py +++ b/mimic-iv/buildmimic/sqlite/import.py @@ -110,7 +110,7 @@ def main(): if args.limit > 0: for f in data_files: if 'patients' in f.name: - pt = pd.read_csv(f) + pt = pd.read_csv(f, low_memory=False) break if pt is None: raise FileNotFoundError('Unable to find a patients file in current folder.') @@ -158,7 +158,7 @@ def main(): tablename = tablenames[i] print("Starting processing {}".format(tablename), end='.. ') if os.path.getsize(f) < THRESHOLD_SIZE: - df = pd.read_csv(f, dtype=mimic_dtypes) + df = pd.read_csv(f, dtype=mimic_dtypes, low_memory=False) df = process_dataframe(df, subjects=subjects) df.to_sql(tablename, connection, index=False) row_counts[tablename] += len(df) diff --git a/mimic-iv/buildmimic/sqlite/import.sh b/mimic-iv/buildmimic/sqlite/import.sh index 08ded288..fb642d1c 100755 --- a/mimic-iv/buildmimic/sqlite/import.sh +++ b/mimic-iv/buildmimic/sqlite/import.sh @@ -29,11 +29,11 @@ for FILE in */**.csv*; do TABLE_NAME=$(echo "${BASENAME%%.*}" | tr "[:upper:]" "[:lower:]") case "$FILE" in *csv) - IMPORT_CMD=".import $FILE $TABLE_NAME" + IMPORT_CMD=".import \"$FILE\" $TABLE_NAME" ;; # need to decompress csv before load *csv.gz) - IMPORT_CMD=".import \"|gzip -dc $FILE\" $TABLE_NAME" + IMPORT_CMD=".import \"|gzip -dc \\\"$FILE\\\"\" $TABLE_NAME" ;; # not a data file so skip *) @@ -41,7 +41,7 @@ for FILE in */**.csv*; do ;; esac echo "Loading $FILE." - sqlite3 $OUTFILE < str: # The logic is as follows: # * DAY -> difference of the two calendar dates (date subtraction = whole days) # * YEAR -> difference of the two calendar years +# * MONTH -> year*12 + month difference (calendar month boundaries) # * sub-day units -> truncate both operands to the unit (which makes the # elapsed seconds an exact multiple of the unit) then divide. +# WEEK is intentionally unsupported: BigQuery week starts (SUNDAY/ISO) do not +# match PostgreSQL DATE_TRUNC('week') Monday semantics. # https://cloud.google.com/bigquery/docs/reference/standard-sql/datetime_functions#datetime_diff _SECONDS_PER_UNIT = {"SECOND": 1, "MINUTE": 60, "HOUR": 3600} @@ -39,6 +42,19 @@ def _datetime_diff_sql(self: Postgres.Generator, expression: exp.Expression) -> return f"(CAST({end} AS DATE) - CAST({start} AS DATE))" if unit == "YEAR": return f"CAST(EXTRACT(YEAR FROM {end}) - EXTRACT(YEAR FROM {start}) AS BIGINT)" + if unit == "MONTH": + # Calendar month boundaries (matches BigQuery DATETIME_DIFF MONTH). + return ( + "CAST((EXTRACT(YEAR FROM {end}) - EXTRACT(YEAR FROM {start})) * 12 " + "+ (EXTRACT(MONTH FROM {end}) - EXTRACT(MONTH FROM {start})) AS BIGINT)" + ).format(end=end, start=start) + if unit not in _SECONDS_PER_UNIT: + # WEEK (and WEEK(SUNDAY)/ISO) need BigQuery's week-start semantics; + # refuse rather than emit a silently wrong days/7 division. + raise ValueError( + f"Unsupported DATETIME_DIFF unit {unit!r}; " + "expected SECOND, MINUTE, HOUR, DAY, MONTH, or YEAR" + ) lo = unit.lower() factor = _SECONDS_PER_UNIT[unit] diff --git a/src/mimic_utils/transpile.py b/src/mimic_utils/transpile.py index d4a65541..29924c18 100644 --- a/src/mimic_utils/transpile.py +++ b/src/mimic_utils/transpile.py @@ -113,7 +113,7 @@ def transpile_file( f"CREATE TABLE {derived_schema}{Path(source_file).stem} AS\n" ) + transpiled_query - with open(destination_file, "w") as write_file: + with open(destination_file, "w", encoding="utf-8") as write_file: write_file.write(transpiled_query) diff --git a/tests/test_transpile.py b/tests/test_transpile.py index 952240bc..8907c7ed 100644 --- a/tests/test_transpile.py +++ b/tests/test_transpile.py @@ -52,6 +52,9 @@ def t(bq: str, dialect: str) -> str: "SELECT (CAST(a.dischtime AS DATE) - CAST(a.admittime AS DATE)) FROM t AS a"), ("diff_year_pg", "SELECT DATETIME_DIFF(a.b, a.c, YEAR) FROM t a", "postgres", "SELECT CAST(EXTRACT(YEAR FROM a.b) - EXTRACT(YEAR FROM a.c) AS BIGINT) FROM t AS a"), + ("diff_month_pg", "SELECT DATETIME_DIFF(a.b, a.c, MONTH) FROM t a", "postgres", + "SELECT CAST((EXTRACT(YEAR FROM a.b) - EXTRACT(YEAR FROM a.c)) * 12 " + "+ (EXTRACT(MONTH FROM a.b) - EXTRACT(MONTH FROM a.c)) AS BIGINT) FROM t AS a"), # DuckDB handles DATETIME_DIFF natively (boundary count), operands swapped ("diff_hour_duckdb", "SELECT DATETIME_DIFF(a.outtime, a.intime, HOUR) FROM t a", "duckdb", "SELECT DATE_DIFF('HOUR', a.intime, a.outtime) FROM t AS a"), @@ -216,6 +219,7 @@ def test_mimic_iii_concept_transpiles_and_reparses(sql_file, dialect): ("2150-01-02 01:00:00", "2150-01-01 23:00:00", "DAY", 1), # crosses midnight ("2150-01-01 23:00:00", "2150-01-01 01:00:00", "DAY", 0), # same day ("2155-06-15 00:00:00", "2150-01-01 00:00:00", "YEAR", 5), + ("2150-04-01 00:00:00", "2150-01-15 00:00:00", "MONTH", 3), ("2150-01-01 00:01:00", "2150-01-01 00:00:59", "MINUTE", 1), ("2150-01-01 00:00:30", "2150-01-01 00:00:10", "SECOND", 20), ("2150-01-01 01:00:00", "2150-01-01 03:00:00", "HOUR", -2), # negative direction