How to programatically get table structure with pyscopg2

Question

when I execute the program application.py ,I get this error :

psycopg2.ProgrammingError
ProgrammingError: ERREUR:  erreur de syntaxe sur ou près de « "/d" »
LINE 1: "/d" carto."BADGES_SFR"

(in english "ProgrammingError: ERREUR: syntax error at or near « "/d" » ") My objective for this program is to get the table structure

this is the following code application.py :

#!/usr/bin/python 2.7.6
# -*- coding:utf-8 -*-
import os
import sys
from flask import Flask,render_template
import psycopg2
reload(sys)  
sys.setdefaultencoding('utf8')
app = Flask(__name__)

@app.route('/')
def fihoum():
    conn = psycopg2.connect(database="carto", user="postgres", password="daed5Aemo", host="192.168.12.54")
    cur = conn.cursor()
    #cur.execute("SELECT * FROM carto.\"BADGES_SFR\"")
    cur.execute("/d carto.\"BADGES_SFR\"")
    rows = cur.fetchall()
    return render_template('hello.html', titre="Données du client BADGES_SFR !",mots=rows)

if __name__=="__main__":
    app.run(host=os.getenv('IP', '0.0.0.0'), 
            port=int(os.getenv('PORT',5000)),
            debug=True)

Answer 1

First of all it's \d <table_name> (note the backslash \d not /d), but that's only available in the psql interactive terminal

My [objective] with this program is to get the table structure

You can use the SQL table information_schema.columns which carries the information you want

SELECT column_name, data_type, is_nullable
FROM information_schema.columns
WHERE table_name = '<table_name>';

    column_name     |     data_type     | is_nullable 
--------------------+-------------------+-------------
 column_a           | integer           | YES
 column_b           | boolean           | NO

Here's a list of the columns available for use: https://www.postgresql.org/docs/9.4/static/infoschema-columns.html

Snippet

import psycopg2
conn = psycopg2.connect(database='carto', user=..., password=...)

q = """                              
SELECT column_name, data_type, is_nullable
FROM information_schema.columns
WHERE table_name = %s;
"""

cur = conn.cursor()
cur.execute(q, ('BADGES_SFR',))  # (table_name,) passed as tuple
cur.fetchall()

# Example Output 
[('column_a', 'integer', 'YES'),
 ('column_b', 'boolean', 'NO'),
 ...,
]

Tip for Jinja2/Flask integration:

Consider using psycopg2.extras.DictCursor, it'll make it easier for you to pull the information based on the column name (dict key) in your Flask template, because your'll a have a dict that you can access e.g. using row['data_type'], row['column_name'] etc.

Answer 2

If you only need the columns description:

>>> cur.execute('select * from t')
>>> description = cur.description
>>> print description
(Column(name='record_timestamptz', type_code=1184, display_size=None, internal_size=8, precision=None, scale=None, null_ok=None),)
>>> columns = description[0]
>>> print columns.name, columns.type_code
record_timestamptz 1184

For the data type name cache the pg_type relation at the application start and make it a dictionary:

cursor = conn.cursor(cursor_factory=psycopg2.extras.DictCursor)

cursor.execute ('select oid, * from pg_type')

pg_type = dict([(t['oid'], t['typname']) for t in cursor.fetchall()])

cursor.execute ('select * from t')

for t in cursor.description:
    print t.name, t.type_code, pg_type[t.type_code]