Accessing Data Programmatically

Accessing Postgresql from .NET using the Npgsql provider

One of the more popular .NET providers for Postgresql is Npgsql, which is ADO.NET compatible and is used nearly identically as other .NET database providers.

A typical query is performed by creating a command, binding parameters, and then executing the command. In C#:

var connString = "Host=myserv;Username=myuser;Password=mypass;Database=mydb";
using (var conn = new NpgsqlConnection(connString))
{
    var querystring = "INSERT INTO data (some_field) VALUES (@content)";
    
    conn.Open();
    // Create a new command with CommandText and Connection constructor
    using (var cmd = new NpgsqlCommand(querystring, conn))
    {
        // Add a parameter and set its type with the NpgsqlDbType enum
        var contentString = "Hello World!";
        cmd.Parameters.Add("@content", NpgsqlDbType.Text).Value = contentString;

        // Execute a query that returns no results
        cmd.ExecuteNonQuery();


        /* It is possible to reuse a command object and open connection instead of creating new ones */

        // Create a new query and set its parameters
        int keyId = 101;
        cmd.CommandText = "SELECT primary_key, some_field FROM data WHERE primary_key = @keyId";
        cmd.Parameters.Clear();
        cmd.Parameters.Add("@keyId", NpgsqlDbType.Integer).Value = keyId;

        // Execute the command and read through the rows one by one
        using (NpgsqlDataReader reader = cmd.ExecuteReader())
        {
            while (reader.Read())   // Returns false for 0 rows, or after reading the last row of the results
            {
                // read an integer value
                int primaryKey = reader.GetInt32(0);
                // or 
                primaryKey = Convert.ToInt32(reader["primary_key"]);

                // read a text value
                string someFieldText = reader["some_field"].ToString();
            }
        }
    }
}    // the C# 'using' directive calls conn.Close() and conn.Dispose() for us

Accessing PostgreSQL from PHP using Pomm2

On the shoulders of the low level drivers, there is pomm. It proposes a modular approach, data converters, listen/notify support, database inspector and much more.

Assuming, Pomm has been installed using composer, here is a complete example:

<?php
use PommProject\Foundation\Pomm;
$loader = require __DIR__ . '/vendor/autoload.php';
$pomm = new Pomm(['my_db' => ['dsn' => 'pgsql://user:pass@host:5432/db_name']]);

// TABLE comment (
// comment_id uuid PK, created_at timestamptz NN,
// is_moderated bool NN default false,
// content text NN CHECK (content !~ '^\s+$'), author_email text NN)
$sql = <<<SQL
SELECT
  comment_id,
  created_at,
  is_moderated,
  content,
  author_email
FROM comment
  INNER JOIN author USING (author_email)
WHERE
  age(now(), created_at) < $*::interval
ORDER BY created_at ASC
SQL;

// the argument will be converted as it is cast in the query above
$comments = $pomm['my_db']
    ->getQueryManager()
    ->query($sql, [DateInterval::createFromDateString('1 day')]);

if ($comments->isEmpty()) {
    printf("There are no new comments since yesterday.");
} else {
    foreach ($comments as $comment) {
        printf(
                "%s has posted at %s. %s\n",
                $comment['author_email'],
                $comment['created_at']->format("Y-m-d H:i:s"),
                $comment['is_moderated'] ? '[OK]' : '');
    }
}

Pomm’s query manager module escapes query arguments to prevent SQL injection. When the arguments are cast, it also converts them from a PHP representation to valid Postgres values. The result is an iterator, it uses a cursor internally. Every row is converted on the fly, booleans to booleans, timestamps to \DateTime etc.

Accessing PostgreSQL from python using psycopg2

You can find description of the driver here.

The quick example is:

import psycopg2

db_host = 'postgres.server.com'
db_port = '5432'
db_un = 'user'
db_pw = 'password'
db_name = 'testdb'

conn = psycopg2.connect("dbname={} host={} user={} password={}".format(
                         db_name,  db_host, db_un, db_pw),
                         cursor_factory=RealDictCursor)
cur = conn.cursor()
sql = 'select * from testtable where id > %s and id < %s'
args = (1, 4)
cur.execute(sql, args)

print(cur.fetchall())

Will result:

[{'id': 2, 'fruit': 'apple'}, {'id': 3, 'fruit': 'orange'}]

Accessing PostgreSQL with the C-API

The C-API is the most powerful way to access PostgreSQL and it is surprisingly comfortable.

Compilation and linking

During compilation, you have to add the PostgreSQL include directory, which can be found with pg_config --includedir, to the include path.
You must link with the PostgreSQL client shared library (libpq.so on UNIX, libpq.dll on Windows). This library is in the PostgreSQL library directory, which can be found with pg_config --libdir.

Note: For historical reason, the library is called libpq.soand not libpg.so, which is a popular trap for beginners.

Given that the below code sample is in file coltype.c, compilation and linking would be done with

gcc -Wall -I "$(pg_config --includedir)" -L "$(pg_config --libdir)" -o coltype coltype.c -lpq

with the GNU C compiler (consider adding -Wl,-rpath,"$(pg_config --libdir)" to add the library search path) or with

cl /MT /W4 /I <include directory> coltype.c <path to libpq.lib>

on Windows with Microsoft Visual C.

Sample program

/* necessary for all PostgreSQL client programs, should be first */
#include <libpq-fe.h>

#include <stdio.h>
#include <string.h>

#ifdef TRACE
#define TRACEFILE "trace.out"
#endif

int main(int argc, char **argv) {
#ifdef TRACE
    FILE *trc;
#endif
    PGconn *conn;
    PGresult *res;
    int rowcount, colcount, i, j, firstcol;
    /* parameter type should be guessed by PostgreSQL */
    const Oid paramTypes[1] = { 0 };
    /* parameter value */
    const char * const paramValues[1] = { "pg_database" };

    /*
     * Using an empty connectstring will use default values for everything.
     * If set, the environment variables PGHOST, PGDATABASE, PGPORT and
     * PGUSER will be used.
     */
    conn = PQconnectdb("");

    /*
     * This can only happen if there is not enough memory
     * to allocate the PGconn structure.
     */
    if (conn == NULL)
    {
        fprintf(stderr, "Out of memory connecting to PostgreSQL.\n");
        return 1;
    }

    /* check if the connection attempt worked */
    if (PQstatus(conn) != CONNECTION_OK)
    {
        fprintf(stderr, "%s\n", PQerrorMessage(conn));
        /*
         * Even if the connection failed, the PGconn structure has been
         * allocated and must be freed.
         */
        PQfinish(conn);
        return 1;
    }

#ifdef TRACE
    if (NULL == (trc = fopen(TRACEFILE, "w")))
    {
        fprintf(stderr, "Error opening trace file \"%s\"!\n", TRACEFILE);
        PQfinish(conn);
        return 1;
    }

    /* tracing for client-server communication */
    PQtrace(conn, trc);
#endif

    /* this program expects the database to return data in UTF-8 */
    PQsetClientEncoding(conn, "UTF8");

    /* perform a query with parameters */
    res = PQexecParams(
        conn,
        "SELECT column_name, data_type "
            "FROM information_schema.columns "
            "WHERE table_name = $1",
        1,            /* one parameter */
        paramTypes,        
        paramValues,
        NULL,        /* parameter lengths are not required for strings */
        NULL,        /* all parameters are in text format */
        0            /* result shall be in text format */
    );

    /* out of memory or sever communication broken */
    if (NULL == res)
    {
        fprintf(stderr, "%s\n", PQerrorMessage(conn));
        PQfinish(conn);
#ifdef TRACE
        fclose(trc);
#endif
        return 1;
    }

    /* SQL statement should return results */
    if (PGRES_TUPLES_OK != PQresultStatus(res))
    {
        fprintf(stderr, "%s\n", PQerrorMessage(conn));
        PQfinish(conn);
#ifdef TRACE
        fclose(trc);
#endif
        return 1;
    }

    /* get count of result rows and columns */
    rowcount = PQntuples(res);
    colcount = PQnfields(res);

    /* print column headings */
    firstcol = 1;

    printf("Description of the table \"pg_database\"\n");

    for (j=0; j<colcount; ++j)
    {
        if (firstcol)
            firstcol = 0;
        else
            printf(": ");

        printf(PQfname(res, j));
    }

    printf("\n\n");

    /* loop through rosult rows */
    for (i=0; i<rowcount; ++i)
    {
        /* print all column data */
        firstcol = 1;

        for (j=0; j<colcount; ++j)
        {
            if (firstcol)
                firstcol = 0;
            else
                printf(": ");

            printf(PQgetvalue(res, i, j));
        }

        printf("\n");
    }

    /* this must be done after every statement to avoid memory leaks */
    PQclear(res);
    /* close the database connection and release memory */
    PQfinish(conn);
#ifdef TRACE
    fclose(trc);
#endif
    return 0;
}