pg_query()

(PHP 4 >= 4.2.0, PHP 5, PHP 7)

执行查询

说明

pg_query(resource $connection,string $query): resource

pg_query()在查询可以执行时返回查询结果资源号。如果查询失败或者提供的连接号无效则返回FALSE。如果连接号有效，则可以用pg_last_error()函数来提取详细的错误信息。pg_query()发送一条 SQL 语句到$connection资源指定的 PostgreSQL 数据库。$connection必须是由pg_connect()或pg_pconnect()返回的合法连接号。本函数返回值是一个其它 PostgreSQL 函数例如pg_fetch_array()可以用来访问查询结果的查询结果资源号。

Note:$connection是pg_query()中的可选参数。如果没有指定$connection，则使用默认连接。默认连接是pg_connect()或pg_pconnect()所打开的最后一个连接。尽管$connection参数可以省略，但不推荐这样做。因为这样可能会导致很难发现脚本中的错误。

Note:

本函数以前的名字为pg_exec()。pg_exec()函数为了兼容性的原因仍然可以使用，但是鼓励用户使用新的名字。

参见pg_connect()，pg_pconnect()，pg_fetch_array()，pg_fetch_object()，pg_num_rows()和pg_affected_rows()。

Regarding david.bouriaud@ac-rouen.fr:
You misunderstand SQL. When a query is issued, results applicable at the time of the query are returned to the application (i.e. PHP). There is no further reference to the database required. Thus, all of the pg_fetch_* functions are acting on an internal data storage, NOT the database itself. This is because SQL does not have a concept of sets, or of state (except in limited circumstances provided by transactions). However, if you use a cursor instead, fetching only one record at a time, you may get an error if you delete the table. If you don't, it is an issue with Postgres, not PHP.

expanding on the note left by "cmoore" -
To check to see if the recordset returned no records,
<?php
 $result=pg_query($conn, "SELECT * FROM x WHERE a=b;");
 if (!$result) {
  echo "query did not execute";
 }
 $rs = pg_fetch_assoc($result);
 if (!$rs) {
  echo "0 records"
 }
?>
-jack

A quick note for novice users: when gathering input from fields on a web form that maintains a database connection, *never* use pg_query to do queries from the field. Always sanitize input using pg_prepare and pg_execute.

There was a typo in the code that I posted:
<?php
 $result=pg_query($conn, "SELECT * FROM x WHERE a=b;");
 if (!$result) {
  echo "query did not execute";
 }
 if (pg_num_rows($result) == 0) {
  echo "0 records"
 }
 else {
  while ($row = pg_fetch_array($result)) {
   //do stuff with $row
  }
 }
?>

It would be better this way:
<?php
 $result=pg_query($conn, "SELECT COUNT(*) AS rows FROM x WHERE a=b;");
 if (!$result) {
  echo "query did not execute";
 }
 if ($line = pg_fetch_assoc($result)) {
  if ($line['rows'] == 0) {
   echo "0 records"
  }
 }
 else {
  while ($row = pg_fetch_array($result)) {
   //do stuff with $row
  }
 }
?> 
This solution doesn't raise the load of the system with the move of matching rows (perhaps 0,1, perhaps 100, 1000, ... rows)

In reply to david dot bouriaud at ac-rouen dot fr:
All it is doing is internal caching. How can that be dangerous. If you are going to be deleting records after you have closed the connection it is your problem to make sure you have the latest and greatest records, and not some cached ones. Considering you are writing the script I don't see why it is a problem, you know what you are doing in the script, so it is quite useless for PHP to invalidate the cache, when that could be done upon exiting the script, which would mean there was less time spent cleaning out the cache when it counts most (when returning data to the user).

One thing to note that wasn't obvious to me at first. If your query returns zero rows, that is not a "failed" query. So the following is wrong:
 $result=pg_query($conn, "SELECT * FROM x WHERE a=b;");
 if (!$result) {
  echo "No a=b in x\n";
 }
pg_query returns FALSE if the query can not be executed for some reason. If the query is executed but returns zero rows then you get back a resul with no rows.

Use pg_query to call your stored procedures, and use pg_fetch_result when getting a value (like a smallint as in this example) returned by your stored procedure.
<?php
$pgConnection = pg_connect("dbname=users user=me");
$userNameToCheckFor = "metal";
$result = pg_query($pgConnection, "SELECT howManyUsersHaveThisName('$userNameToCheckFor')");
$count = pg_fetch_result($result, 0, 'howManyUsersHaveThisName');
?>

$GLOBALS["PG_CONNECT"]=pg_connect(...);
....
function query ($sqlQuery,$var=0) {
  if (!$GLOBALS["PG_CONNECT"]) return 0;
  $lev=error_reporting (8); //NO WARRING!!
  $result=pg_query ($sqlQuery);
  error_reporting ($lev); //DEFAULT!!
  if (strlen ($r=pg_last_error ($GLOBALS["PG_CONNECT"]))) {
   if ($var) {
    echo "<p color=\"red\">ERROR:<pre>";
    echo $sqlQuery;
    echo "</pre>";
    echo $r;
    echo "&lt/p>";
   }
   close_db ();
   return 0;
  }
  return $result;
}

Improving upon what jsuzuki said:
It's probably better to use pg_num_rows() to see if no rows were returned, as that leaves the resultset cursor pointed to the first row so you can use it in a loop.
Example:
<?php
 $result=pg_query($conn, "SELECT * FROM x WHERE a=b;");
 if (!$result) {
  echo "query did not execute";
 }
 if (pg_num_rows($result) == 0) {
  echo "0 records"
 }
 else {
  while ($row = pg_fetch_array($result) {
   //do stuff with $row
  }
 }
?>
I, personally, also find it more readable.

Here is my small function to make it easier for me to use data from select queries (attention, it is sensitive to sql injection)
<?php
function requestToDB($connection,$request){ 
  if(!$result=pg_query($connection,$request)){
    return False;
  }
  $combined=array();
  while ($row = pg_fetch_assoc($result)) {
    $combined[]=$row;
  }
  return $combined;
}
?>
Example:
<?php
$conn = pg_pconnect("dbname=mydatabase");
$results=requestToDB($connect,"select * from mytable");
//You can now access a "cell" of your table like this:
$rownumber=0;
$columname="mycolumn";
$mycell=$results[$rownumber][$columname];
var_dump($mycell);

Hi to all !
It seems that the old pg_exec function does not do what it is expected to.
In the doc, it is said that it returns a resource identifier on the successful querry that was send to the backend.
It seems to me that it is more than a resource identifier.
Follow the example :
<?php
$ConnId = pg_connect ("blablabla");
$ResId = pg_exec ("select * from table", $ConnId);
pg_close ($ConnId);
$row = pg_fetch_array ($ResId, 4);
?>
I closed the connection voluntarily before the pg_fetch_array. It WORKS !
Now, imagine the following script :
<?php
$ConnId = pg_connect ("blablabla");
$ResId = pg_exec ("select * from table", $ConnId);
pg_close ($ConnId);
system ("psql base -c delete from table");
$row = pg_fetch_array ($ResId, 4);
?>
See how it could be harmful !!!! I think that the coders have done this for performances reasons, but it is not the right way do do so !!!

create table from pg_query results.. hope it helps newbies...
function table_create($result)
{
  $numrows = pg_num_rows($result);
  $fnum = pg_num_fields($result);
  echo "<table border width='100%'>";
  echo "<tr>";
  for ($x = 0; $x < $fnum; $x++) {
    echo "<td><b>";
    echo strtoupper(pg_field_name($result, $x));
    echo "</b></td>";
  }
  echo "</tr>";
  for ($i = 0; $i < $numrows; $i++) {
    $row = pg_fetch_object($result, $i);
    echo "<tr align='center'>";
    for ($x = 0; $x < $fnum; $x++) {
  $fieldname = pg_field_name($result, $x);
  echo "<td>";
  echo $row->$fieldname;
  echo "</td>";
    }
    echo"</tr>";
  }
  echo "</table>";
  
  return 0;
}

That's why your code should never assume it has the very latest data unless it locks it.

Took me a while to track this down so I thought it might be useful for others:
If you use stored procedures and need to get result sets back from them:
function dbquery($link,$query){
 pg_query($link,"BEGIN;");
 $tr=pg_query($link,$query);
 $r=pg_fetch_row($tr);
 $name=$r[0];
 $rs=pg_query($link,"FETCH ALL IN \"" . $name . "\";");
 pg_query($link,"END;");
 return $rs;
}
(Error checking removed for clarity)