File Coverage

blib/lib/Text/Fragment.pm

Criterion	Covered	Total	%
statement	182	187	97.3
branch	85	98	86.7
condition	16	22	72.7
subroutine	21	21	100.0
pod	5	5	100.0
total	309	333	92.7

package Text::Fragment;

our $DATE = '2017-07-10'; # DATE
our $VERSION = '0.10'; # VERSION

use 5.010001;
use strict;
use warnings;
use Log::ger;

use Data::Clone;

require Exporter;
our @ISA       = qw(Exporter);
our @EXPORT_OK = qw(
                       list_fragments
                       get_fragment
                       set_fragment_attrs
                       insert_fragment
                       delete_fragment
               );

our $re_id = qr/\A[A-Za-z0-9_.,:-]+\z/;

our %SPEC;

sub _format_quoted {
    my $unquoted = shift;
    my $res = "";
    my $i = -1;
    while (++$i < length($unquoted)) {
        my $c = substr($unquoted, $i, 1);
        if ($c eq '\\' or $c eq '"') {
            $res .= "\\$c";
        } elsif ($c !~ /[\x20-\x7F]/) {
            # strip non-printables
        } else {
            $res .= $c;
        }
    }
    qq("$res");
}

sub _parse_quoted {
    my $quoted = shift;
    $quoted =~ s/\A"//; $quoted =~ s/"\z//;
    my $res = "";
    my $i = -1;
    while (++$i < length($quoted)) {
        my $c = substr($quoted, $i, 1);
        if ($c eq '\\') {
            $res .= substr($quoted, ++$i, 1);
        } else {
            $res .= $c;
        }
    }
    $res;
}

sub _format_attr_value {
    my $val = shift;
    $val =~ /\s|"|[^\x20-\x7f]/ ? _format_quoted($val) : $val;
}

sub _label {
    my %args  = @_;
    my $id            = $args{id} // "";
    my $label         = $args{label}; # str
    my $comment_style = $args{comment_style};
    my $attrs         = $args{attrs};

    my $quoted_re = qr/"(?:[^\n\r"\\]|\\[^\n\r])*"/;

    my $a_re;  # regex to match attributes
    my $ai_re; # also match attributes, but attribute id must be present
    if (length $id) {
        $ai_re = qr/(?:\w+=\S*[ \t]+)*id=(?\Q$id\E)(?:[ \t]+\w+=(?:$quoted_re|\S+))*/;
    } else {
        $ai_re = qr/(?:\w+=\S*[ \t]+)*id=(?\S*)(?:[ \t]+\w+=(?:$quoted_re|\S+))*/;
    }
    $a_re  = qr/(?:\w+=\S*)?(?:[ \t]+\w+=(?:$quoted_re|\S+))*/;

    my ($ts, $te); # tag start and end
    if ($comment_style eq 'shell') {
        $ts = "#";
        $te = "";
    } elsif ($comment_style eq 'c') {
        $ts = "/*";
        $te = "*/";
    } elsif ($comment_style eq 'cpp') {
        $ts = "//";
        $te = "";
    } elsif ($comment_style eq 'html') {
        $ts = "";
    } elsif ($comment_style eq 'ini') {
        $ts = ";";
        $te = "";
    }
    # regex to detect fragment
    my $ore = qr!^(?.*?)[ \t]*\Q$ts\E[ \t]*
                 \Q$label\E[ \t]+
                 (?$ai_re)[ \t]*
                 \Q$te\E[ \t]*(?\R|\z)!mx;

    my $mre = qr!^\Q$ts\E[ \t]*
                 BEGIN[ \t]+\Q$label\E[ \t]+
                 (?$ai_re)[ \t]*
                 \Q$te\E[ \t]*(?\R)
                 (?:
                     (?.*)
                     ^\Q$ts\E[ \t]*END[ \t]+\Q$label\E[ \t]+
                       (?:\w+=\S*[ \t]+)*id=\g{id}(?:[ \t]+\w+=\S+)*
                       [ \t]*\Q$te\E |
                     (?.*?) # without any ID at the ending comment
                     ^\Q$ts\E[ \t]*END[ \t]+\Q$label\E(?:[ \t]+$a_re)?[ \t]*
                     \Q$te\E
                 )
                 [ \t]*(?\R|\z)!msx;

    my $parse_attrs = sub {
        my $s = shift // "";
        my %a;
        while ($s =~ /(\w+)=(?:($quoted_re)|(\S+))(?:\s+|\z)/g) {
            $a{$1} = $2 ? _parse_quoted($2) : $3;
        }
        \%a;
    };

    return {
        one_line_pattern   => $ore,
        multi_line_pattern => $mre,
        parse_attrs        => $parse_attrs,
        format_fragment    => sub {
            my %f = @_;

            # formatted attrs as string
            my $as = "";
            if (ref($f{attrs})) {
                for (sort keys %{ $f{attrs} }) {
                    $as .= " " . "$_="._format_attr_value($f{attrs}{$_});
                }
            } else {
                my $a = $parse_attrs->($f{attrs});
                $as = join("", map {" $_="._format_attr_value($a->{$_})}
                               grep {$_ ne 'id'}
                               sort keys %$a);
            }

            my $pl = $f{payload};

            # to keep things simple here, regardless of whether the replaced
            # pattern contains ending newline (enl), we still format with ending
            # newline. then we'll just need to strip ending newline if it's not
            # needed.

            if ($f{is_multi} || $pl =~ /\R/) {
                $pl .= "\n" unless $pl =~ /\R\z/;
                "$ts BEGIN $label id=$id$as" . ($te ? " $te":"") . "\n" .
                $pl .
                "$ts END $label id=$id" . ($te ? " $te":"") . "\n";
            } else {
                "$pl $ts $label id=$id$as" . ($te ? " $te":"") . "\n";
            }
        },
    };
}

sub _doit {
    my ($which, %args) = @_;

    die "BUG: invalid which"
        unless $which =~ /\A(?:list|get|insert|delete|set_attrs)\z/;
    my ($label_str, $label_sub);
    if (ref($args{label}) eq 'CODE') {
        $label_str = "FRAGMENT";
        $label_sub = $args{label};
    } else {
        $label_str = $args{label} || "FRAGMENT";
        $label_sub = \&_label;
    }

    my $text               = $args{text};
    defined($text) or return [400, "Please specify text"];
    my $id                 = $args{id};
    if ($which =~ /\A(?:get|insert|set_attrs|delete)\z/) {
        defined($id) or return [400, "Please specify id"];
    }
    if (defined $id) {
        $id =~ $re_id or return [400, "Invalid ID syntax '$id', please use ".
                                     "letters/numbers/dots/dashes only"];
    }
    my $attrs              = $args{attrs} // {};
    for (keys %$attrs) {
        /\A\w+\z/ or return [400, "Invalid attribute name '$_', please use ".
                                 "letters/numbers only"];
        if (!defined($attrs->{$_})) {
            if ($which eq 'set_attrs') {
                next;
            } else {
                return [400, "Undefined value for attribute name '$_'"];
            }
        }
    }

    my $good_pattern       = $args{good_pattern};
    my $replace_pattern    = $args{replace_pattern};
    my $top_style          = $args{top_style};
    my $comment_style      = $args{comment_style} // "shell";
    $comment_style =~ /\A(cpp|c|shell|html|ini)\z/ or return [
        400, "Invalid comment_style '$comment_style', ".
            "please use cpp/c/shell/html/ini"];
    my $res                = $label_sub->(id=>$id, label=>$label_str,
                                          comment_style=>$comment_style);
    my $one_line_pattern   = $res->{one_line_pattern};
    my $multi_line_pattern = $res->{multi_line_pattern};
    my $parse_attrs        = $res->{parse_attrs};
    my $format_fragment    = $res->{format_fragment};
    my $payload            = $args{payload};
    if ($which eq 'insert') {
        defined($payload) or return [400, "Please specify payload"];
    }

    if ($which eq 'list') {

        my @ff;
        while ($text =~ /($one_line_pattern|$multi_line_pattern)/xg) {
            push @ff, {
                raw     => $1,
                id      => $+{id},
                payload => $+{payload},
                attrs   => $parse_attrs->($+{attrs}),
            };
        }
        return [200, "OK", \@ff];

    } elsif ($which eq 'get') {

        if ($text =~ /($one_line_pattern|$multi_line_pattern)/x) {
            return [200, "OK", {
                raw     => $1,
                id      => $+{id},
                payload => $+{payload},
                attrs   => $parse_attrs->($+{attrs}),
            }];
        } else {
            return [404, "Fragment with ID '$id' not found"];
        }

    } elsif ($which eq 'set_attrs') {

        my $orig_attrs;
        my $sub = sub {
            my %f = @_;
            $orig_attrs = $parse_attrs->($f{attrs});
            my %a = %$orig_attrs; delete $a{id};
            for my $k (keys %$attrs) {
                my $v = $attrs->{$k};
                if (defined $v) {
                    $a{$k} = $v;
                } else {
                    delete $a{$k};
                }
            }
            $f{attrs} = \%a;
            $format_fragment->(%f);
        };
        if ($text =~ s{$one_line_pattern | $multi_line_pattern}
                      {$sub->(%+)}egx) {
            return [200, "OK", {text=>$text, orig_attrs=>$orig_attrs}];
        } else {
            return [404, "Fragment with ID '$id' not found"];
        }

    } elsif ($which eq 'delete') {

        my %f;
        my $sub = sub {
            %f = @_;
            $f{enl} ? $f{bnl} : "";
        };
        if ($text =~ s{(?\R?)
                       (?$one_line_pattern | $multi_line_pattern)}
                      {$sub->(%+)}egx) {
            return [200, "OK", {text=>$text,
                                orig_fragment=>$f{fragment},
                                orig_payload=>$f{payload}}];
        } else {
            return [304, "Fragment with ID '$id' already does not exist"];
        }

    } else { # insert

        my $replaced;
        my %f;
        my $sub = sub {
            %f = @_;
            return $f{fragment} if $payload eq $f{payload};
            $replaced++;
            $f{orig_fragment} = $f{fragment};
            $f{orig_payload} = $f{payload};
            $f{payload} = $payload;
            $format_fragment->(%f);
        };
        if ($good_pattern && $text =~ /$good_pattern/) {
            return [304, "Text contains good pattern"];
        }

        if ($text =~ s{(?(?:$one_line_pattern | $multi_line_pattern))}
                      {$sub->(%+)}ex) {
            if ($replaced) {
                return [200, "Payload replaced", {
                    text=>$text, orig_fragment=>$f{orig_fragment},
                    orig_payload=>$f{orig_payload}}];
            } else {
                return [304, "Fragment with ID '$id' already exist with ".
                            "same content"];
            }
        }

        my $fragment = $format_fragment->(payload=>$payload, attrs=>$attrs);
        if ($replace_pattern && $text =~ /($replace_pattern)/) {
            my $orig_fragment = $1;
            $text =~ s/$replace_pattern/$fragment/;
            return [200, "Replace pattern replaced", {
                text=>$text, orig_fragment=>$orig_fragment}];
        }

        if ($top_style) {
            $text = $fragment . $text;
        } elsif (length($text)) {
            my $enl = $text =~ /\R\z/; # text ends with newline
            $fragment =~ s/\R\z// unless $enl;
            $text .= ($enl ? "" : "\n") . $fragment;
        } else {
            # insert at bottom of empty string
            $text = $fragment;
        }
        return [200, "Fragment inserted at the ".
                    ($top_style ? "top" : "bottom"), {text=>$text}];
    }

}

$SPEC{':package'} = {
    v => 1.1,
    summary     => 'Manipulate fragments in text',
    description => <<'_',

A fragment is a single line or a group of lines (called payload) with a metadata
encoded in the comment that is put adjacent to it (for a single line fragment)
or enclosing it (for a multiline fragment). Fragments are usually used in
configuration files or code. Here is the structure of a single-line fragment:

     #

Here is the structure of a multi-line fragment:

    # BEGIN

    # END  []

Label is by default `FRAGMENT` but can be other string. Attributes are a
sequence of `name=val` separated by whitespace, where name must be alphanums
only and val is zero or more non-whitespace characters. There must at least be
an attribute with name `id`, it is used to identify fragment and allow the
fragment to be easily replaced/modified/deleted from text. Attributes are
optional in the ending comment.

Comment character used is by default `#` (`shell`-style comment), but other
comment styles are supported (see below).

Examples of single-line fragments (the second example uses `c`-style comment and
the third uses `cpp`-style comment):

    RSYNC_ENABLE=1 # FRAGMENT id=enable
    some text /* FRAGMENT id=id2 */
    some text // FRAGMENT id=id3 foo=1 bar=2

An example of multi-line fragment (using `html`-style comment instead of
`shell`):


    some
    lines
    of
    text


Another example (using `ini`-style comment):

    ; BEGIN FRAGMENT id=default-settings
    register_globals=On
    extension=mysql.so
    extension=gd.so
    memory_limit=256M
    post_max_size=64M
    upload_max_filesize=64M
    browscap=/c/share/php/browscap.ini
    allow_url_fopen=0
    ; END FRAGMENT

_
};

my $arg_comment_style = {
    summary => 'Comment style',
    schema  => ['str' => {
        default => 'shell',
        in      => [qw/c cpp html shell ini/],
    }],
};
my $arg_label = {
    schema  => [str => {default=>'FRAGMENT'}],
    summary => 'Comment label',
};

my $arg_id = {
    summary => 'Fragment ID',
    schema  => ['str*' => { match => qr/\A[\w-]+\z/ }],
    req     => 1,
};

my $arg_payload = {
    summary => 'Fragment content',
    schema  => 'str*',
    req     => 1,
};

$SPEC{list_fragments} = {
    v => 1.1,
    summary => 'List fragments in text',
    args => {
        text          => {
            summary => 'The text which contain fragments',
            schema  => 'str*',
            req     => 1,
            pos     => 0,
        },
        comment_style => $arg_comment_style,
        label         => $arg_label,
    },
    result => {
        summary => 'List of fragments',
        schema  => 'array*',
        description => <<'_',

Will return status 200 if operation is successful. Result will be an array of
fragments, where each fragment is a hash containing these keys: `raw` (string),
`payload` (string), `attrs` (hash), `id` (string, can also be found in
attributes).

_
    },
};
sub list_fragments {
    _doit('list', @_);
}

$SPEC{get_fragment} = {
    v => 1.1,
    summary => 'Get fragment with a certain ID in text',
    description => <<'_',

If there are multiple occurences of the fragment with the same ID ,

_
    args => {
        text          => {
            summary => 'The text which contain fragments',
            schema  => 'str*',
            req     => 1,
            pos     => 0,
        },
        comment_style => $arg_comment_style,
        label         => $arg_label,
        id            => $arg_id,
    },
    result => {
        summary => 'Fragment',
        schema  => 'array*',
        description => <<'_',

Will return status 200 if fragment is found. Result will be a hash with the
following keys: `raw` (string), `payload` (string), `attrs` (hash), `id`
(string, can also be found in attributes).

Return 404 if fragment is not found.

_
    },
};
sub get_fragment {
    _doit('get', @_);
}

$SPEC{set_fragment_attrs} = {
    v => 1.1,
    summary => 'Set/unset attributes of a fragment',
    description => <<'_',

If there are multiple occurences of the fragment with the same ID ,

_
    args => {
        text          => {
            summary => 'The text which contain fragments',
            schema  => 'str*',
            req     => 1,
            pos     => 0,
        },
        comment_style => $arg_comment_style,
        label         => $arg_label,
        id            => $arg_id,
        attrs         => {
            schema => 'hash*',
            description => <<'_',

To delete an attribute in the fragment, you can set the value to undef.

_
            req    => 1,
        },
    },
    result => {
        summary => 'New text and other data',
        schema  => 'array*',
        description => <<'_',

Will return status 200 if fragment is found. Result will be a hash containing
these keys: `text` (string, the modified text), `orig_attrs` (hash, the old
attributes before being modified).

Return 404 if fragment is not found.

_
    },
};
sub set_fragment_attrs {
    _doit('set_attrs', @_);
}

$SPEC{insert_fragment} = {
    v => 1.1,
    summary => 'Insert or replace a fragment in text',
    description => <<'_',

Newline insertion behaviour: if fragment is inserted at the bottom and text does
not end with newline (which is considered bad style), the inserted fragment will
also not end with newline. Except when original text is an empty string, in
which case an ending newline will still be added.

_
    args => {
        text      => {
            summary => 'The text to insert fragment into',
            schema  => 'str*',
            req     => 1,
            pos     => 0,
        },
        id        => $arg_id,
        payload   => $arg_payload,
        top_style => {
            summary => 'Whether to append fragment at beginning of file '.
                'instead of at the end',
            schema  => [bool => { default=>0 }],
            description => <<'_',

Default is false, which means to append at the end of file.

Note that this only has effect if `replace_pattern` is not defined or replace
pattern is not found in file. Otherwise, fragment will be inserted to replace
the pattern.

_
        },
        replace_pattern => {
            summary => 'Regex pattern which if found will be used for '.
                'placement of fragment',
            schema  => 'str',
            description => <<'_',

If fragment needs to be inserted into file, then if `replace_pattern` is defined
then it will be searched. If found, fragment will be placed to replace the
pattern. Otherwise, fragment will be inserted at the end (or beginning, see
`top_style`) of file.

_
        },
        good_pattern => {
            summary => 'Regex pattern which if found means fragment '.
                'need not be inserted',
            schema  => 'str',
        },
        comment_style => $arg_comment_style,
        label         => $arg_label,
        attrs         => {
            schema => [hash => {default=>{}}],
        },
    },
    result => {
        summary => 'A hash of result',
        schema  => 'hash*',
        description => <<'_',

Will return status 200 if operation is successful and text is changed. The
result is a hash with the following keys: `text` will contain the new text,
`orig_payload` will contain the original payload before being removed/replaced,
`orig_fragment` will contain the original fragment (or the text that matches
`replace_pattern`).


Will return status 304 if nothing is changed (i.e. if fragment with the
same payload that needs to be inserted already exists in the text).

_
    },
};
sub insert_fragment {
    _doit('insert', @_);
}

$SPEC{delete_fragment} = {
    v => 1.1,
    summary => 'Delete fragment in text',
    description => <<'_',

If there are multiple occurences of fragment (which is considered an abnormal
condition), all occurences will be deleted.

Newline deletion behaviour: if fragment at the bottom of text does not end with
newline (which is considered bad style), the text after the fragment is deleted
will also not end with newline.

_
    args => {
        text => {
            summary => 'The text to delete fragment from',
            schema  => 'str*',
            req     => 1,
            pos     => 0,
        },
        id => {
            summary => 'Fragment ID',
            schema  => ['str*' => { match => qr/\A[\w-]+\z/ }],
            req     => 1,
            pos     => 1,
        },
        comment_style => $arg_comment_style,
        label => {
            schema  => ['any' => {
                of => ['str*', 'code*'],
                default => 'FRAGMENT',
            }],
            summary => 'Comment label',
        },
    },
    result => {
        summary => 'A hash of result',
        schema  => 'hash*',
        description => <<'_',

Will return status 200 if operation is successful and text is deleted. The
result is a hash with the following keys: `text` will contain the new text,
`orig_payload` will contain the original fragment payload before being deleted,
`orig_fragment` will contain the original fragment. If there are multiple
occurences (which is considered an abnormal condition), only the last deleted
fragment will be returned in `orig_payload` and `orig_fragment`.

Will return status 304 if nothing is changed (i.e. when the fragment that needs
to be deleted already does not exist in the text).

_
    },
};
sub delete_fragment {
    _doit('delete', @_);
}

1;
# ABSTRACT: Manipulate fragments in text

__END__