File Coverage

blib/lib/API/INSEE/Sirene.pm

Criterion	Covered	Total	%
statement	136	259	52.5
branch	26	70	37.1
condition	5	12	41.6
subroutine	29	38	76.3
pod	17	18	94.4
total	213	397	53.6

package API::INSEE::Sirene;

use strict;
use warnings;

use Carp 'croak';
use JSON;
use HTTP::Request::Common qw/ GET POST /;
use HTTP::Status ':constants';
use List::Util 'any';
use LWP::UserAgent;
use POSIX 'strftime';
use Switch;

our $VERSION = 4.04;

use constant {
    API_AUTH_URL        => 'https://api.insee.fr/token',
    API_BASE_URL        => 'https://api.insee.fr/entreprises/sirene/V3',
    DEFAULT_MAX_RESULTS => 20, # from documentation
    DEFAULT_TIMEOUT     => 20,
    HARD_MAX_RESULTS    => 1_000, # from documentation
    MAX_SIREN_LENGTH    => 9,
    MAX_SIRET_LENGTH    => 14,
    MIN_LENGTH          => 3,
};

my $EMPTY = q//;

my $historized_fields = {
    siren => [ qw/
        dateFin dateDebut
        etatAdministratifUniteLegale changementEtatAdministratifUniteLegale
        nomUniteLegale changementNomUniteLegale nomUsageUniteLegale changementNomUsageUniteLegale
        denominationUniteLegale changementDenominationUniteLegale denominationUsuelle1UniteLegale
        denominationUsuelle2UniteLegale denominationUsuelle3UniteLegale changementDenominationUsuelleUniteLegale
        categorieJuridiqueUniteLegale changementCategorieJuridiqueUniteLegale activitePrincipaleUniteLegale
        nomenclatureActivitePrincipaleUniteLegale changementActivitePrincipaleUniteLegale nicSiegeUniteLegale
        changementNicSiegeUniteLegale economieSocialeSolidaireUniteLegale changementEconomieSocialeSolidaireUniteLegale
        caractereEmployeurUniteLegale changementCaractereEmployeurUniteLegale
    / ],
    siret => [ qw/
        etatAdministratifEtablissement changementEtatAdministratifEtablissement
        enseigne1Etablissement enseigne2Etablissement enseigne3Etablissement changementEnseigneEtablissement
        denominationUsuelleEtablissement changementDenominationUsuelleEtablissement
        activitePrincipaleEtablissement nomenclatureActivitePrincipaleEtablissement changementActivitePrincipaleEtablissement
        caractereEmployeurEtablissement changementCaractereEmployeurEtablissement
    / ],
};

my $useful_fields_legal_unit = [
    qw/
        siren
        dateCreationUniteLegale
        sigleUniteLegale
        categorieEntreprise
        denominationUniteLegale denominationUsuelle1UniteLegale nomUniteLegale
        categorieJuridiqueUniteLegale
        activitePrincipaleUniteLegale nomenclatureActivitePrincipaleUniteLegale
        nicSiegeUniteLegale
    /
];

my $useful_fields_establishment = [
    qw/
        siren siret
        denominationUsuelleEtablissement denominationUniteLegale denominationUsuelle1UniteLegale nomUniteLegale
        activitePrincipaleUniteLegale
        numeroVoieEtablissement typeVoieEtablissement libelleVoieEtablissement
        codePostalEtablissement libelleCommuneEtablissement
    /
];

my $useful_fields_aliases = {
    nicSiege                        => 'nicSiegeUniteLegale',
    nom                             => [ 'denominationUniteLegale', 'nomUniteLegale' ],
    dateCreation                    => 'dateCreationUniteLegale',
    sigle                           => 'sigleUniteLegale',
    categorieJuridique              => 'categorieJuridiqueUniteLegale',
    nomenclatureActivitePrincipale  => 'nomenclatureActivitePrincipaleUniteLegale',
    activitePrincipale              => 'activitePrincipaleUniteLegale',
    numvoie                         => 'numeroVoieEtablissement',
    typevoie                        => 'typeVoieEtablissement',
    nomvoie                         => 'libelleVoieEtablissement',
    codePostal                      => 'codePostalEtablissement',
    nomCommune                      => 'libelleCommuneEtablissement',
    adresseComplete                 => [
                                        'numeroVoieEtablissement',
                                        'typeVoieEtablissement', 'libelleVoieEtablissement',
                                        'codePostalEtablissement', 'libelleCommuneEtablissement'
                                    ],
};

sub new {
    my $class = shift;
    my ($credentials, $timeout, $max_results, $proxy) = @_;

    my $self = bless {
        credentials      => $credentials,
        user_agent       => undef,
        token_expiration => undef,
        max_results      => undef,
        debug_mode       => 0,
        current_endpoint => undef,
    }, $class;

    $self->_initUserAgent();
    $self->setProxy($proxy);
    $self->setMaxResults($max_results);
    $self->setTimeout($timeout);

    return $self;
}

sub setCredentials {
    my ($self, $credentials) = @_;

    $self->{'credentials'} = $credentials;
}

sub setMaxResults {
    my ($self, $max_results) = @_;

    $max_results //= DEFAULT_MAX_RESULTS;
    $self->{'max_results'} = $max_results > HARD_MAX_RESULTS ? HARD_MAX_RESULTS : $max_results;
}

sub setDebugMode {
    my ($self, $debug_value) = @_;

    $self->{'debug_mode'} = $debug_value;
}

sub setProxy {
    my ($self, $proxy) = @_;

    defined $proxy ? $self->{'user_agent'}->proxy([ 'http', 'https' ], $proxy) : $self->{'user_agent'}->env_proxy;
}

sub setTimeout {
    my ($self, $timeout) = @_;

    $timeout //= DEFAULT_TIMEOUT;
    $self->{'user_agent'}->timeout($timeout);
}

sub setCurrentEndpoint {
    my ($self, $endpoint) = @_;

    $self->{'current_endpoint'} = $endpoint;
}

sub _dumpRequest {
    my ($self, $request, $response) = @_;

    my $dump = sprintf "Sent request:\n%s\n", $request->as_string;
    $dump .= sprintf "Received response:\n%s\n", $response->as_string if defined $response;

    return $dump;
}

sub _initUserAgent {
    my $self = shift;

    $self->{'user_agent'} = LWP::UserAgent->new(protocols_allowed => [ 'http', 'https' ]);

    $self->{'user_agent'}->agent("Perl API::INSEE::Sirene V$VERSION");
    $self->{'user_agent'}->default_header('Accept' => 'application/json');
}

sub _getToken {
    my $self = shift;

    croak 'Please provide your credentials.' if !defined $self->{'credentials'};

    my $request = POST API_AUTH_URL,
        Authorization => "Basic $self->{'credentials'}",
        Content       => [ grant_type => 'client_credentials' ];

    my $response = $self->{'user_agent'}->request($request);
    my $json_obj;

    if ($response->content_type =~ m/^application\/json/) {
        $json_obj = decode_json($response->content);
    }
    else {
        return 1, $self->_dumpRequest($request, $response); # the API may return xml intead of json...
    }

    switch ($response->code) {
        case HTTP_OK {
            $self->{'token_expiration'} = time + $json_obj->{'expires_in'};
            $self->{'user_agent'}->default_header( Authorization => "Bearer $json_obj->{'access_token'}" );
            return 0;
        }
        case HTTP_UNAUTHORIZED { # wrong credentials
            return 1 , $json_obj->{'error_description'};
        }
        else { # oh dear we are in trouble
            return 1, $self->_dumpRequest($request, $response);
        }
    }
}

sub _sendRequest {
    my ($self, $parameters) = @_;

    my $request;
    if (!exists $parameters->{'q'}) {
        my @url_parameters;

        foreach my $key (keys %{ $parameters }) {
            push @url_parameters, join '=', $key, $parameters->{$key};
        }

        my $endpoint = join '?', $self->{'current_endpoint'}, join '&', @url_parameters;
        $request = GET join '/', API_BASE_URL, $endpoint;
    }
    else {
        $request = POST join('/', API_BASE_URL, $self->{'current_endpoint'}),
            Content => [ %{ $parameters } ];
    }

    if ($self->{'debug_mode'}) { # Requests will not be sent in debug mode
        return 0, $self->_dumpRequest($request);
    }

    if (!defined $self->{'token_expiration'} || $self->{'token_expiration'} < time) {
        my ($err, $msg) = $self->_getToken();
        croak $msg if $err;
    }

    my $response = $self->{'user_agent'}->request($request);

    switch ($response->code) {
        case [ HTTP_OK, HTTP_NOT_FOUND ] {
            return 0, $response->content;
        }
        case HTTP_MOVED_PERMANENTLY { # duplicated legal unit/ establishment
            return 1, sprintf "%s\n%s", $response->message, $response->header('Location');
        }
        case [
            HTTP_REQUEST_URI_TOO_LARGE, HTTP_TOO_MANY_REQUESTS,
            HTTP_UNAUTHORIZED, HTTP_FORBIDDEN,
            HTTP_SERVICE_UNAVAILABLE
        ] {
# There is no syntax error in request, the http message should be sufficient to understand the problem
            return 1, $response->message;
        }
        else { # case HTTP_BAD_REQUEST || HTTP_INTERNAL_SERVER_ERROR
            return 1, $self->_dumpRequest($request, $response);
        }
    }
}

sub _buildParameters {
    my ($self, $usefull_fields, $desired_fields, $criteria) = @_;

# Parameters names come from the documentation
    my $parameters = {
        date   => strftime('%Y-%m-%d', localtime),
        nombre => $self->{'max_results'},
    };
    $parameters->{'champs'} = $self->_buildFields($usefull_fields, $desired_fields) if (defined $desired_fields && $desired_fields ne 'all');
    $parameters->{'q'}      = sprintf('(%s)', $criteria) if defined $criteria;

    return $parameters;
}

sub _buildFields {
    my ($self, $usefull_fields, $desired_fields) = @_;

    if (defined $desired_fields) {
        return $self->_mapAliases($desired_fields);
    }
    else {
        return join ',', @{ $usefull_fields };
    }
}

sub _mapAliases {
    my ($self, $desired_fields) = @_;

    my @desired_fields = ref $desired_fields eq 'ARRAY' ? @{ $desired_fields } : $desired_fields;

    foreach my $desired_field (@desired_fields) {
        if (exists $useful_fields_aliases->{$desired_field}) {
            if (ref $useful_fields_aliases->{$desired_field} eq 'ARRAY') {
                $desired_field = join ',', @{ $useful_fields_aliases->{$desired_field} };
            }
            else {
                $desired_field = $useful_fields_aliases->{$desired_field};
            }
        }
    }

    return join ',', @desired_fields;
}

sub getCustomCriteria {
    my ($self, $field_name, $value, $search_mode) = @_;

    croak 'No endpoint specified.' if !defined $self->{'current_endpoint'};

    $search_mode //= 'aproximate';
    if (exists $useful_fields_aliases->{$field_name}) {
        if (ref $useful_fields_aliases->{$field_name} eq 'ARRAY') {
            croak "Can't use the alias $field_name in custom criteria";
        }
        $field_name = $useful_fields_aliases->{$field_name};
    }

    if ($search_mode eq 'aproximate') {
        my @criteria;
        my @words = split /[ \/-]/, $value;

        foreach my $word (@words) {
            $word =~ s/&/%26/ig;
            $word = sprintf '(%s:"%s"~ OR %s:*%s*)', $field_name, $word, $field_name, $word;
            $word = "periode$word" if any { $_ eq $field_name } @{ $historized_fields->{$self->{'current_endpoint'}} };


            push @criteria, $word;
        }

        return join ' AND ', @criteria;
    }

    my $criteria;
    $value =~ s/&/%26/ig;

    if ($search_mode eq 'exact') {
        $criteria = sprintf '%s:%s', $field_name, $value;
    }
    elsif ($search_mode eq 'begin') {
        $criteria = sprintf '%s:%s*', $field_name, $value;
    }

    $criteria = "periode($criteria)" if any { $_ eq $field_name } @{ $historized_fields->{$self->{'current_endpoint'}} };

    return $criteria;
}

sub searchByCustomCriteria {
    my ($self, $criteria, $desired_fields) = @_;

    my $parameters;
    switch ($self->{'current_endpoint'}) {
        case 'siren' { $parameters = $self->_buildParameters($useful_fields_legal_unit, $desired_fields, $criteria) }
        case 'siret' { $parameters = $self->_buildParameters($useful_fields_establishment, $desired_fields, $criteria) }
        else { croak 'Bad endpoint specified.' }
    }

    return $self->_sendRequest($parameters);
}

sub getLegalUnitBySIREN {
    my ($self, $siren_number, $desired_fields) = @_;

    return 1, "Invalid SIREN $siren_number -> Must be a ${ \MAX_SIREN_LENGTH } digits number."
        if $siren_number !~ m/^\d{${ \MAX_SIREN_LENGTH }}$/;

    $self->setCurrentEndpoint("siren/$siren_number");
    my $parameters = $self->_buildParameters($useful_fields_legal_unit, $desired_fields);

    return $self->_sendRequest($parameters);
}

sub searchLegalUnitBySIREN {
    my ($self, $siren_number, $desired_fields) = @_;

    return 1, "Invalid SIREN $siren_number -> Must be a ${ \MIN_LENGTH } digits min and ${ \MAX_SIREN_LENGTH } digits number max."
        if $siren_number !~ m/^\d{${ \MIN_LENGTH },${ \MAX_SIREN_LENGTH }}$/;

    $self->setCurrentEndpoint('siren');
    my $criteria = $self->getCustomCriteria('siren', $siren_number, 'begin');

    return $self->searchByCustomCriteria($criteria, $desired_fields);
}

sub getEstablishmentBySIRET {
    my ($self, $siret_number, $desired_fields) = @_;

    return 1, "Invalid SIRET $siret_number -> Must be a ${ \MAX_SIRET_LENGTH } digits number."
        if $siret_number !~ m/^\d{${ \MAX_SIRET_LENGTH }}$/;

    $self->setCurrentEndpoint("siret/$siret_number");
    my $parameters = $self->_buildParameters($useful_fields_establishment, $desired_fields);

    return $self->_sendRequest($parameters);
}

sub getEstablishmentsBySIREN {
    my ($self, $siren_number, $desired_fields) = @_;

    return (1, "Invalid SIREN $siren_number -> Must be a ${ \MAX_SIREN_LENGTH } digits number.")
        if $siren_number !~ m/^\d{${ \MAX_SIREN_LENGTH }}$/;

    $self->setCurrentEndpoint('siret');
    my $criteria = $self->getCustomCriteria('siren', $siren_number);

    return $self->searchByCustomCriteria($criteria, $desired_fields);
}

sub searchEstablishmentBySIRET {
    my ($self, $siret_number, $desired_fields) = @_;

    return 1, "Invalid SIRET $siret_number -> Must be a ${ \MIN_LENGTH } digits min and a ${ \MAX_SIRET_LENGTH } digits number max."
        if $siret_number !~ m/^\d{${ \MIN_LENGTH },${ \MAX_SIRET_LENGTH }}$/;

    $self->setCurrentEndpoint('siret');
    my $criteria = $self->getCustomCriteria('siret', $siret_number);

    return $self->searchByCustomCriteria($criteria, $desired_fields);
}

sub getLegalUnitsByName {
    my ($self, $name, $desired_fields) = @_;

    $self->setCurrentEndpoint('siren');
    my $criteria = $self->getCustomCriteria('denominationUniteLegale', $name);

    return $self->searchByCustomCriteria($criteria, $desired_fields);
}

sub getEstablishmentsByName {
    my ($self, $name, $desired_fields) = @_;

    $self->setCurrentEndpoint('siret');
    my $criteria = $self->getCustomCriteria('denominationUniteLegale', $name);

    return $self->searchByCustomCriteria($criteria, $desired_fields);
}

sub getLegalUnitsByUsualName {
    my ($self, $name, $desired_fields) = @_;


    $self->setCurrentEndpoint('siren');
    my $criteria = $self->getCustomCriteria('denominationUsuelle1UniteLegale', $name);

    return $self->searchByCustomCriteria($criteria, $desired_fields);
}

sub getEstablishmentsByUsualName {
    my ($self, $name, $desired_fields) = @_;

    $self->setCurrentEndpoint('siret');
    my $criteria = $self->getCustomCriteria('denominationUsuelle1UniteLegale', $name);

    return $self->searchByCustomCriteria($criteria, $desired_fields);
}

1;

__END__