function main()
{
global $argv;
global $datasets;
// from require_once "../analysis/common/functions.php"
global $api_timezone;
// from reqire_once "./lib/common.php"
//----------------
// Obtain parameters
if (PHP_SAPI != 'cli') {
// Invoked by Web server
expected_query_parameters(['action', 'startdate', 'enddate', 'format']);
// Determine resource ($querybin_name will be set or not)
$components = explode('/', $_SERVER['PATH_INFO']);
assert($components[0] === '');
// since PATH_INFO started with a "/"
array_shift($components);
// remove empty component from first "/"
switch (count($components)) {
case 0:
// No query bin specified: list all (e.g. /api/querybin.php)
$resource = 'all';
$querybin_name = NULL;
break;
case 1:
// Query bin (e.g. /api/querybin.php/foobar)
$resource = 'querybin';
$querybin_name = $components[0];
break;
case 2:
// Query bin's tweets (e.g. /api/querybin.php/foobar/tweets)
if ($components[1] != 'tweets') {
abort_with_error(404, "Not found: " . $_SERVER['PATH_INFO']);
}
$resource = 'querybin/tweets';
$querybin_name = $components[0];
break;
default:
$resource = NULL;
abort_with_error(404, "Not found: " . $_SERVER['PATH_INFO']);
}
// Determine action
if ($_SERVER['REQUEST_METHOD'] === 'GET' && array_key_exists('action', $_GET)) {
$action = $_GET['action'];
// explicit action
} else {
if ($_SERVER['REQUEST_METHOD'] === 'POST' && array_key_exists('action', $_POST)) {
$action = $_POST['action'];
// explicit action
} else {
// Use default action for resource
switch ($resource) {
case 'all':
$action = 'list';
break;
case 'querybin':
$action = 'bin-info';
break;
case 'querybin/tweets':
$action = 'tweet-info';
break;
default:
$action = NULL;
abort_with_error(500, "Internal error: bad rsrc: {$resource}");
}
}
}
// Check combination of resource, action and method
$method = $_SERVER['REQUEST_METHOD'];
if ($method != 'GET' && $method != 'POST' && $method != 'DELETE') {
abort_with_error(405, "Method not allowed: {$method}");
}
$bad_combination = false;
switch ($resource) {
case 'all':
if (!($action === 'list' && $method === 'GET')) {
$bad_combination = true;
}
break;
case 'querybin':
if (!($action === 'bin-info' && $method === 'GET')) {
$bad_combination = true;
}
break;
case 'querybin/tweets':
if (!($action === 'tweet-info' && $method === 'GET' || $action === 'export-tweets' && $method === 'GET' || $action === 'purge-tweets' && $method === 'DELETE' || $action === 'purge-tweets' && $method === 'POST')) {
$bad_combination = true;
}
break;
default:
abort_with_error(500, "Internal error: bad rsrc: {$resource}");
}
if ($bad_combination) {
abort_with_error(400, "Invalid: {$resource}, {$action}, {$method}");
}
// Get parameters
$str_start = $_REQUEST['startdate'];
$str_end = $_REQUEST['enddate'];
if (isset($_REQUEST['format']) && $_REQUEST['format'] !== '') {
$format = $_REQUEST['format'];
if ($format != 'csv' && $format != 'tsv') {
abort_with_error(400, "Invalid format: {$format}");
}
} else {
$format = 'csv';
// default
}
} else {
// Invoked from command line
// PHP's getopt is terrible, but it is always available.
$skip_num = 1;
$options = getopt("hlbtps:e:", ['help', 'list', 'bin-info', 'tweet-info', 'purge-tweets', 'start:', 'end:']);
if ($options != false) {
foreach ($options as $opt => $optarg) {
$skip_num++;
if ($optarg != NULL) {
$skip_num++;
}
switch ($opt) {
case 'h':
case 'help':
$script_name = basename($argv[0]);
$dtz = $api_timezone != NULL ? $api_timezone : "UTC";
echo <<<END
Usage: php {$script_name} [options] [queryBinName]
Options:
-l | --list list names of all query bins (default without name)
-b | --bin-info show information about query bin (default with name)
-t | --tweet-info show information about tweets in named query bin
-p | --purge-tweets purge tweets in named query bin
-s | --start tm start time for tweet viewing/purging
-e | --end tm end time for tweet viewing/purging
-h | --help show this help message
Format for tm: yyyy-mm-ddThh:mm:ss[tz]
tz: 'Z', 'UTC' or [+|-]HH:MM (default: {$dtz})
END;
exit(0);
break;
case 'l':
case 'list':
if (isset($action)) {
fwrite(STDERR, "Usage error: multiple actions\n");
exit(2);
}
$action = 'list';
break;
case 'b':
case 'bin-info':
if (isset($action)) {
fwrite(STDERR, "Usage error: multiple actions\n");
exit(2);
}
$action = 'bin-info';
break;
case 't':
case 'tweet-info':
if (isset($action)) {
fwrite(STDERR, "Usage error: multiple actions\n");
exit(2);
}
$action = 'tweet-info';
break;
case 'p':
case 'purge-tweets':
if (isset($action)) {
fwrite(STDERR, "Usage error: multiple actions\n");
exit(2);
}
$action = 'purge-tweets';
break;
case 's':
case 'start':
$str_start = $optarg;
break;
case 'e':
case 'end':
$str_end = $optarg;
break;
}
}
}
$args = array_slice($argv, $skip_num);
if (0 < count($args)) {
// Compensate for PHP's getopt's incomplete processing
// Check for options after getopt stops processing
if ($args[0] === '--') {
array_shift($args);
}
foreach ($args as $arg) {
if ($arg[0] === '-') {
fwrite(STDERR, "Usage error: options not allowed after arguments: {$arg}\n");
exit(2);
}
}
}
// Arguments
if (count($args) == 0) {
$querybin_name = NULL;
} else {
if (count($args) == 1) {
$querybin_name = $args[0];
} else {
fwrite(STDERR, "Usage error: too many arguments\n");
exit(2);
}
}
// Derive implied action if no explicit action was provided
if (!isset($action)) {
if (isset($querybin_name)) {
$action = 'bin-info';
} else {
$action = 'list';
}
}
// Check for invalid option/argument combinations
if ($action !== 'list') {
// All these actions require a query bin to be specified
if (!isset($querybin_name)) {
fwrite(STDERR, "Usage error: missing query bin name\n");
exit(2);
}
} else {
if (isset($querybin_name)) {
fwrite(STDERR, "Usage error: query bin name not required\n");
exit(2);
}
}
if ($action === 'list' || $action === 'bin-info') {
// These actions do not use start/end time
if (isset($str_start) || isset($str_end)) {
fwrite(STDERR, "Usage error: start/end time not required\n");
exit(2);
}
}
}
//----------------
// Check/process parameters
// Check and set $querybin (if needed)
if (isset($querybin_name)) {
if (!isset($datasets[$querybin_name])) {
abort_with_error(404, "Unknown query bin: {$querybin_name}");
}
$querybin = $datasets[$querybin_name];
} else {
$querybin = NULL;
}
// Check start and end time
//
// Code convention:
// $str_start and $str_end are string values (or null).
// $dt_start and $dt_end are DateTime objects (or null).
$dt_start = NULL;
if (isset($str_start) && $str_start !== '') {
try {
$dt_start = dt_parse($str_start, false, $api_timezone);
} catch (DtException $e) {
$m = $e->getMessage();
abort_with_error(400, "Bad start time ({$m}): {$e->dt_str}");
}
}
$dt_end = NULL;
if (isset($str_end) && $str_end !== '') {
try {
$dt_end = dt_parse($str_end, true, $api_timezone);
} catch (DtException $e) {
$m = $e->getMessage();
abort_with_error(400, "Bad end time ({$m}): {$e->dt_str}");
}
}
if (isset($dt_start) && isset($dt_end)) {
if ($dt_end < $dt_start) {
abort_with_error(400, "End time is before start time.");
}
}
//----------------
// Perform action and produce response
switch ($action) {
case 'list':
do_list_bins();
break;
case 'bin-info':
do_bin_info($querybin);
break;
case 'tweet-info':
do_view_or_export_tweets($querybin, $dt_start, $dt_end, NULL);
break;
case 'export-tweets':
assert(isset($format));
do_view_or_export_tweets($querybin, $dt_start, $dt_end, $format);
break;
case 'purge-tweets':
do_purge_tweets($querybin, $dt_start, $dt_end);
break;
default:
abort_with_error(500, "Internal error: unexpected action: {$action}");
}
exit(0);
}
function dt_from_utc($str)
{
$r = array();
if (preg_match('/^\\d{4}-\\d\\d-\\d\\d \\d\\d:\\d\\d:\\d\\d$/', $str, $r)) {
return dt_parse($str . 'Z', false, NULL);
} else {
throw new DtException($str, DtException::SYNTAX_ERROR);
}
}
