File Coverage

blib/lib/Finance/YahooJPN/Quote.pm

Criterion	Covered	Total	%
statement	199	261	76.2
branch	48	92	52.1
condition	26	35	74.2
subroutine	15	18	83.3
pod	5	6	83.3
total	293	412	71.1

package Finance::YahooJPN::Quote;

use 5.16.3;
use warnings;
use utf8;

our $VERSION = '1.08'; # 2014-07-18 (since 2001-05-30)

use Carp;
use IO::Socket;
use Encode;
use Module::Load;

=head1 NAME

Finance::YahooJPN::Quote -- Fetch historical Japanese stock quotes on Yahoo! Japan Finance.

=head1 SYNOPSIS

  use Finance::YahooJPN::Quote;

  # Get the quote of Sony Corp. at the Tokyo stock exchange.
  my @quote = Finance::YahooJPN::Quote->historical('6758.t');

  print join("\n", @quote);

=head1 DESCRIPTION

Historical quote data is basis for analyzing stock market. Here in Japan, standard quote data is indicated as a set of data: the four prices (open, high, low, close) and the volume of each day. This module provides its user some functions to get historical quote of a company.

=cut

# initialize global variables of this package
my $Japan_Standard_Time = time + 32400; # 9h * 60m * 60s = 32400s
my $Today = join '-', (
                    ( gmtime($Japan_Standard_Time) )[5] + 1900 ,
    sprintf('%02d', ( gmtime($Japan_Standard_Time) )[4] + 1   ),
    sprintf('%02d', ( gmtime($Japan_Standard_Time) )[3]       ),
    );
undef $Japan_Standard_Time;

my $Server = 'info.finance.yahoo.co.jp';

my $Debug = 0;

=head1 METHODS

=over

=item historical($symbol [, 'start' => $start] [, 'noadjust' => 1])

This class method automatically C and C then C a historical series of quote of the stock which specified with C<$symbol> argument.

See the descriptions about the following methods for the argument and attributes: C<$symbol>, C and C.

=cut

sub debug {
    my($class, $level, $symbol, $start, $last) = @_;
    if ($level !~ m/^[1-6]$/) {
        croak 'You must specify debug level [1-6]';
    }
    $Debug = $level;

    my $self = $class->new($symbol);
    $self->scan('start' => $start, 'last' => $last);
    say join("\n", $self->output('noadjust' => 1)) if $Debug == 5;
    say join("\n", $self->output(               )) if $Debug == 6;
}

sub historical {
    my($class, $symbol, %option) = @_;
    my $self = $class->new($symbol);

    foreach my $key (keys %option) {
        my $lowercase = $key;
        $lowercase =~ tr/A-Z/a-z/;
        unless ($lowercase eq 'start' or $lowercase eq 'last' or $lowercase eq 'noadjust') {
            carp "unknown attribute name: $key (will be ignored)";
        }
        $option{$lowercase} = $option{$key};
    }

    if ($option{'start'}) {
        if ($option{'last'}) {
            $self->scan('start' => $option{'start'}, 'last' => $option{'last'});
        } else {
            $self->scan('start' => $option{'start'});
        }
    } elsif ($option{'last'}) {
            $self->scan('last' => $option{'last'});
    } else {
        $self->scan();
    }

    if ($option{'noadjust'} and $option{'noadjust'} == 1) {
        $self->output('noadjust' => 1);
    } else {
        $self->output();
    }
}

=item new($symbol)

Constructor class method. A stock C<$symbol> should be given with 4-digit code number and optionaly followed by a letter extension (dot `.' and an alphabet). (i.e. `6758' or `6758.t')

Japanese stock markets use 4-digit code numbers for stock symbols. In addtion to that, an alphabetical letter extention is used for indicating its exchanging place. For example, the stock symbol code of Sony Corp. is '6758' and the letter extention of the Tokyo Stock Exchange is '.t'. Hence, the stock quote of Sony Corp. at Tokyo Stock Exchange is specified as '6758.t'.

According to the Yahoo-Japan-Finance's description L the letter extentions of exchanging place are:

 .t: Tokyo   Stock Exchange
 .q: JASDAQ
 .n: Nagoya  Stock Exchange
 .s: Sapporo Stock Exchange
 .f: Fukuoka Stock Exchange

Letter extention is omittable. When it is omit, the default exchange market is chosen by the Yahoo-Japan-Finance's server. It is not certain but I guess that the default one should be the main exchange market of the stock. Note: since almost symbols should work without letter extention, I have experienced certain problems with a few symbols those which have originally `.j' letter extention. This is of course not for the module but owe to the Yahoo-Japan-Finance server's behavior.

There is an exception for above. A few symbols of index are indicated in 5 to 7 digit code numbers. They are '998405' (TOPIX), '998407' (NIKKEI) and '23337' (JASDAQ). L

=cut

sub new {
    my($class, $symbol) = @_;
    my $self = {};
    bless $self, $class;

    unless ($symbol) {
        croak "'symbol' argument can't be omittable";
    }
    if (
           $symbol eq '998405'
        or $symbol eq '998407'
        or $symbol eq '23337'
        or $symbol =~ /^\d{4}(\.[a-zA-Z]){0,1}$/
       ) {
        $self->{'symbol'} = $symbol;
    }
    else {
        croak "Stock symbol must be given in a 4-digit number and optionally which can be\nfollowed by a letter extension (a dot `.' and an alphabet).\n\tFor example: `6758' or `6758.t'\nExcept for these special index codes: 998405, 998407 and 23337.\n\n";	}

    return $self;
}

my $MAX_RETRY = 10;

# _fetch($url)
# This private method is for fetching a web page.
sub _fetch {
    my($self, $abs_path) = @_;

    my @html;
    for (my $i = 0; $i < $MAX_RETRY; $i++) {
        my $sock;
        unless ($sock = IO::Socket::INET->new(
            PeerAddr => $Server,
            PeerPort => 'http(80)',
            Proto    => 'tcp',
            Timeout  => 10,
            )) {
            # retry upto $MAX_RETRY times
            if ($i < $MAX_RETRY - 1) {
                sleep 20;
                next;
            } else {
                die "Network connection error to $Server; reached max retry: $MAX_RETRY";
            }
        }

        print $sock <<"EOF";
GET $abs_path HTTP/1.1
Host: $Server
Connection: close

EOF

        @html = <$sock>;
        close $sock;

        if ($html[0] =~ /^HTTP\/1\.1 200 OK/) {
            last;
        } else {
            # retry upto $MAX_RETRY times
            if ($i < $MAX_RETRY - 1) {
                sleep 10;
                next;
            } else {
                die "Network communication error with $Server; reached max retry: $MAX_RETRY";
            }
        }
    }

    my $html = join '', @html;
    # newline character unification
    $html =~ s/\x0D\x0A|\x0D|\x0A/\n/g;

    return $html;
}

=item set_proxy($proxy [, 'socks_version' => $version]) *EXPERIMENTAL*

Set proxy. C<$proxy> must be given in format like '192.168.0.1:8080' (port is needed).

If the proxy server is other than SOCKS5 (inevitably it is SOCKS4), you must give c<$socks_version> (i.e. 'socks_version' => 4).

Unfortunately, I can offer very limited support for this feature since I don't have proper testing environment for proxy. So this feature is just experimental and I have no plan for upgrading it to non-experimental state in the future. And also, the install package's test suite is only for non-proxy.

=cut

sub set_proxy {
    my $self = shift;
    $self->{'proxy'} = shift;
    my %option = @_;

    $self->{'socks_version'} = $option{'version'} || 5;

    return $self;
}

# _fetch_proxy($url)
# Another method for fetching a web page with proxy.
sub _fetch_proxy {
    load 'IO::Socket::Socks';
    my($self, $abs_path) = @_;

    my($proxy_addr, $proxy_port) = split ':', $self->{'proxy'};

    my @html;
    for (my $i = 0; $i < $MAX_RETRY; $i++) {
        my $socks = IO::Socket::Socks->new(
            ConnectAddr  => $Server,
            ConnectPort  => 80,
            Proto        => 'tcp',
            ProxyAddr    => $proxy_addr,
            ProxyPort    => $proxy_port,
            SocksVersion => $self->{'socks_version'}, # default 5
            # SocksDebug   => 1,
            # Timeout  => 10,
            ) or die "Couldn't connect to $Server via " . $self->{'proxy'};

        print $socks <<"EOF";
GET $abs_path HTTP/1.1
Host: $Server
Connection: close

EOF

        @html = <$socks>;
        close $socks;

        if ($html[0] =~ /^HTTP\/1\.1 200 OK/) {
            last;
        } else {
            # retry upto $MAX_RETRY times
            if ($i >= $MAX_RETRY - 2) {
                die 'Network connection error: reached $MAX_RETRY';
            }
        }
    }

    my $html = join '', @html;
    # newline character unification
    $html =~ s/\x0D\x0A|\x0D|\x0A/\n/g;

    return $html;
}

=item scan(['start' => $start])

This object method is for scanning the stock's historical quote pages of Yahoo-Japan-Finance from the C<$start> date to the current date. And for picking up quote data of each day on those pages.

Date of C<$start> must be given in the format `YYYY-MM-DD' (ex. `2003-08-14'). Be careful, don't forget to quote the word, because bare word 2000-01-01 will be comprehend by Perl as '2000 - 1 - 1 = 1998'. This attribute is omittable. The default value of C<$start> is '1990-01-01'.

You cannot specify a date of last day. Because, to find the splits you must scan the quote during whole of the period from the C<$start> day. Without split data, estimation of value adjustment for split cannot be done exactly.

Note that datetime of this module is based on JST (Japan Standard Time: GMT +09:00).

=cut

sub scan {
    my($self, %term) = @_;

    $self->{'start'} = '1983-01-01';
    $self->{'last' } = $Today;

    foreach my $key (keys %term) {
        my $lowercase = $key;
        $lowercase =~ tr/A-Z/a-z/;
        unless ($lowercase eq 'start' or $lowercase eq 'last') {
            carp "unknown attribute name: $key (will be ignored)";
        }
        unless ($term{$key} =~ /^\d{4}-\d{2}-\d{2}$/) {
            croak "A date should be given in the format `YYYY-MM-DD'. (ex. `2003-08-14')";
        }
        $self->{$lowercase} = $term{$key};
    }

    # estimate term to fetch
    my($yearStart, $monthStart, $dayStart) = split /-/, $self->{'start'};
    my($yearEnd, $monthEnd, $dayEnd) = split /-/, $self->{'last' };

    # multi page fetching
    while (1) {
        # 50rows/1page is max at Yahoo-Japan-Finance
        # So the poor Yahoo-Japan-Finance's paging algorithm has a certain serious bug that I can't utilize paging method.

        my $abs_path;
        if ($yearEnd == $yearStart && $monthEnd == $monthStart) {
            $abs_path = "/history/?code=$self->{'symbol'}&sy=$yearEnd&sm=$monthEnd&sd=$dayStart&ey=$yearEnd&em=$monthEnd&ed=$dayEnd&tm=d";
        } else {
            $abs_path = "/history/?code=$self->{'symbol'}&sy=$yearEnd&sm=$monthEnd&sd=01&ey=$yearEnd&em=$monthEnd&ed=$dayEnd&tm=d";
        }
        my $remotedoc;
        if ($self->{'proxy'}) {
            $remotedoc = decode('utf-8', $self->_fetch_proxy($abs_path));
        } else {
            $remotedoc = decode('utf-8', $self->_fetch($abs_path));
        }

        # testing whether it is valid symbol or not.
        if ($remotedoc =~ m/該当する銘柄はありません。/) {
            say '該当する銘柄はありません。' if $Debug == 1;
            last;
        }
        # testing whether it is valid term or not.
        if ($remotedoc =~ m/該当する期間のデータはありません。/) {
            say '該当する期間のデータはありません。' if $Debug == 1;
            last;
        }
        # testing whether it is the overrun page (with bulk quote table) or not (currently it is no use withou paging method)
        #if ($remotedoc =~ m|