Merge pull request #255 from jim-dotcom/new-commands

SQL-332: Run SQL query on CSV file contents from command-line

Author: ykxvqze

sql-queries-6/run-sql-queries-on-csv-file/data.csv

@@ -0,0 +1,9 @@
+name,sex,age,weight,height 
+Alex,M,41,74,170 
+Bert,M,42,66,166 
+Dave,M,39,72,167 
+Elly,F,30,70,124 
+Luke,M,34,72,163 
+Omar,M,38,69,145 
+Page,F,31,67,135 
+Ruth,F,28,65,131

sql-queries-6/run-sql-queries-on-csv-file/example2_csv.py

@@ -0,0 +1,11 @@
+import duckdb 
+
+# Run a query directly on the CSV file 
+query = "SELECT AVG(weight) AS avg_weight FROM read_csv_auto('data.csv')" 
+avg_weight = duckdb.query(query).to_df() 
+print(avg_weight) 
+
+# Another query to count individuals by gender 
+count_gender_query = "SELECT sex, COUNT(*) AS count FROM read_csv_auto('data.csv') GROUP BY sex" 
+count_gender = duckdb.query(count_gender_query).to_df() 
+print(count_gender)

sql-queries-6/run-sql-queries-on-csv-file/example_csv.py

@@ -0,0 +1,19 @@
+import pandas as pd
+from pandasql import sqldf 
+
+# Load the CSV file into a DataFrame 
+df = pd.read_csv('data.csv') 
+
+# Define the query function 
+pysqldf = lambda q: sqldf(q, globals()) 
+
+# Example 1: Getting average weight 
+avg_weight_query = "SELECT AVG(weight) AS avg_weight FROM df" 
+avg_weight = pysqldf(avg_weight_query) 
+print(avg_weight) 
+
+
+# Example 2: Count individuals by gender 
+count_gender_query = "SELECT sex, COUNT(*) AS count FROM df GROUP BY sex" 
+count_gender = pysqldf(count_gender_query) 
+print(count_gender)

sql-queries-6/run-sql-queries-on-csv-file/import-data-to-sqlite.sh

@@ -0,0 +1 @@
+sqlite3 :memory: -cmd '.mode csv' -cmd '.import data.csv my_table'

sql-queries-6/run-sql-queries-on-csv-file/install-duckdb.sh

@@ -0,0 +1 @@
+pip install duckdb

sql-queries-6/run-sql-queries-on-csv-file/install-pandas-and-pandasql.sh

@@ -0,0 +1,2 @@
+pip install pandas pandasql sqlalchemy==1.4.46
+

sql-queries-6/run-sql-queries-on-csv-file/install-sqlite.sh

@@ -0,0 +1,5 @@
+# Intall SQLite
+sudo apt-get install sqlite3
+
+# Check version of SQLite
+sqlite3 --version

sql-queries-6/run-sql-queries-on-csv-file/run-sql-queries-noninteractively.sh

@@ -0,0 +1 @@
+sqlite3 :memory: -cmd '.mode csv' -cmd '.import data.csv my_table' 'SELECT sex, COUNT(*) FROM my_table GROUP BY sex;'

sql-queries-6/run-sql-queries-on-csv-file/running-example2_csv.py.sh

@@ -0,0 +1 @@
+python example2_csv.py

sql-queries-6/run-sql-queries-on-csv-file/running-example_csv.py.sh

@@ -0,0 +1 @@
+python example_csv.py