nur-sery/nur_src/mapper/csv/CsvReader.php

213 lines
6.4 KiB
PHP

<?php
namespace nur\mapper\csv;
use nur\A;
use nur\b\io\EOFException;
use nur\b\io\IReader;
use nur\b\io\Tfilter;
use nur\b\params\parametrable_utils;
use nur\b\params\Tparametrable;
use nur\data\types\Metadata;
use nur\mapper\base\encoding_utils;
use nur\mapper\base\Producer;
use nur\mapper\base\Tencoding;
use nur\mapper\line\IconvMapper;
use nur\reader;
/**
* Class CsvReader: produit un flux de données à partir d'une source au format
* CSV.
*
* NB: cette classe supporte les fichiers CSV multi-lignes, contrairement à
* {@link Csv2AssocMapper}
*
* --autogen-properties-and-methods--
* @method string|null getInputEncoding()
* @method string|null getOutputEncoding()
* @method setInput($value)
* @method string|null setInputEncoding(?string $value)
* @method string|null setOutputEncoding(?string $value)
* @method setCsvFlavour($value)
* @method bool setMultiSchema(bool $value)
* @method bool|null setParseHeaders(?bool $value)
* @method array|null setHeaders(?array $value)
* @method int setSkipLines(int $value)
* @method string|null setMapEmpty(?string $value)
*/
class CsvReader extends Producer {
use Tparametrable, Tencoding, Tfilter;
const SEPARATOR = csv_defaults::OO_SEPARATOR;
const ENCLOSURE = csv_defaults::OO_ENCLOSURE;
const ESCAPE = csv_defaults::OO_ESCAPE;
function __construct($input=null) {
$this->csv2assoc = new Csv2AssocMapper();
$this->csv2assoc->setCsvFlavour([static::SEPARATOR, static::ENCLOSURE, static::ESCAPE]);
$params = null;
if ($input !== null) $params["input"] = $input;
parent::__construct($params);
}
function setEncodingFilter(string $from, string $to="utf-8"): void {
$this->_setEncodingFilter($from, $to);
}
const PARAMETRABLE_PARAMS_SCHEMA = [
"input" => [null, null, "fichier en entrée"],
];
static function _get_parametrable_params_schema(): array {
return array_merge(self::PARAMETRABLE_PARAMS_SCHEMA
, encoding_utils::PARAMETRABLE_PARAMS_SCHEMA
, AbstractCsvMapper::PARAMETRABLE_PARAMS_SCHEMA
, Csv2AssocMapper::PARAMETRABLE_PARAMS_SCHEMA);
}
protected function afterSetParametrableParams(array $modifiedKeys, ?Metadata $md=null): void {
if (!in_array("input_encoding", $modifiedKeys)
&& in_array("csv_flavour", $modifiedKeys)) {
$flavourName = $this->csvFlavourName;
if ($flavourName !== null) {
$flavourName = csv_defaults::verifix_name($flavourName);
$this->ppInputEncoding = csv_defaults::get_encoding($flavourName);
$modifiedKeys[] = "input_encoding";
}
}
$this->encodingInput__afterSetParametrableParams($modifiedKeys);
}
protected $ppInput;
function pp_setInput($input): void {
if ($input instanceof IReader) $this->reader = $input;
else $this->ppInput = $input;
}
/** @var Csv2AssocMapper */
protected $csv2assoc;
/** @var string */
private $csvFlavourName;
function pp_setCsvFlavour($csvFlavour): void {
if (is_string($csvFlavour)) $this->csvFlavourName = $csvFlavour;
$this->csv2assoc->setCsvFlavour($csvFlavour);
}
function pp_setMultiSchema(bool $multiSchema): void {
$this->csv2assoc->setMultiSchema($multiSchema);
}
function pp_setParseHeaders(?bool $parseHeaders): void {
$this->csv2assoc->setParseHeaders($parseHeaders);
}
function pp_setHeaders(?array $headers): void {
$this->csv2assoc->setHeaders($headers);
}
function pp_setSkipLines(int $skipLines): void {
$this->csv2assoc->setSkipLines($skipLines);
}
function pp_setMapEmpty($mapEmpty=null): void {
$this->csv2assoc->setMapEmpty($mapEmpty);
}
/** @var IReader */
protected $reader;
protected function setup(): void {
if ($this->reader === null) {
$this->reader = reader::with($this->ppInput);
$this->_rwAppendFilters($this->reader);
}
$this->csv2assoc->ensureSetup();
}
protected function teardown(): void {
$this->csv2assoc->close();
if ($this->reader !== null) {
$this->reader->close();
$this->reader = null;
}
}
function producer() {
$parser = $this->csv2assoc;
[$separator, $enclosure, $escape] = $parser->getCsvFlavour();
$reader = $this->reader;
$resource = $reader->getResource();
if ($resource !== null) {
while (!$parser->_parseLine()) {
if (fgets($resource) === false) break;
}
while (($values = fgetcsv($resource, 0, $separator, $enclosure, $escape)) !== false) {
if (!$parser->_checkHeader($values)) {
yield $parser->_mapRow($values);
}
}
} else {
while (true) {
try {
$line = $reader->readLine();
} catch (EOFException $e) {
break;
}
if (!$parser->_parseLine()) continue;
$values = str_getcsv($line, $separator, $enclosure, $escape);
if (!$parser->_checkHeader($values)) {
yield $parser->_mapRow($values);
}
}
}
}
#############################################################################
const _AUTOGEN_CONSTS = [
"" => [self::class, "_autogen_consts", true],
];
const _AUTOGEN_LITERALS = /*autogen*/[
[
\nur\b\params\parametrable_utils::class,
'\\nur\\b\\params\\parametrable_utils::class',
],
[self::class, 'self::class'],
[
self::PARAMETRABLE_PARAMS_SCHEMA,
'self::PARAMETRABLE_PARAMS_SCHEMA',
],
];
const _AUTOGEN_METHODS = /*autogen*/[
[
\nur\b\params\parametrable_utils::class,
'_autogen_methods_getters',
self::PARAMETRABLE_PARAMS_SCHEMA,
self::class,
],
[
\nur\b\params\parametrable_utils::class,
'_autogen_methods_setters',
self::PARAMETRABLE_PARAMS_SCHEMA,
self::class,
],
];
const _AUTO_GETTERS = /*autogen*/[
'getInput' => 'input',
'getInputEncoding' => 'input_encoding',
'getOutputEncoding' => 'output_encoding',
'getCsvFlavour' => 'csv_flavour',
'isMultiSchema' => 'multi_schema',
'getParseHeaders' => 'parse_headers',
'getHeaders' => 'headers',
'getSkipLines' => 'skip_lines',
'getMapEmpty' => 'map_empty',
];
const _AUTO_SETTERS = /*autogen*/[
'setInput' => 'input',
'setInputEncoding' => 'input_encoding',
'setOutputEncoding' => 'output_encoding',
'setCsvFlavour' => 'csv_flavour',
'setMultiSchema' => 'multi_schema',
'setParseHeaders' => 'parse_headers',
'setHeaders' => 'headers',
'setSkipLines' => 'skip_lines',
'setMapEmpty' => 'map_empty',
];
#--autogen-dynamic--
}