comparison blast2html.py @ 116:f5066973029a

refactor

115:0c2a03f9740b

    colors = ('black', 'blue', 'green', 'magenta', 'red')

    max_scale_labels = 10

    def __init__(self, input, templatedir, templatename, dbname, genelinks={}):
        self.input = input
        self.templatename = templatename
        self.dbname = dbname
        self.genelinks = genelinks

        self.blast = objectify.parse(self.input).getroot()
        self.loader = jinja2.FileSystemLoader(searchpath=templatedir)
        self.environment = jinja2.Environment(loader=self.loader,

        display_nolink: boolean, if false don't display anything if no link can be created. Default True.
        """
        
        db = hit.getroottree().getroot().BlastOutput_db

        if isinstance(self.genelinks, six.string_types):
            template = self.genelinks
        else:
            template = self.genelinks[db].template

        if text is None:
            if text_from == 'hitid':
                text = hitid(hit)
            elif text_from == 'dbname':
                text = self.dbname or self.genelinks[db].dbname or 'Gene Bank'
            else:
                raise ValueError("Unknown value for text_from: '{0}'. Use 'hitid' or 'dbname'.".format(text_from))

        if template is None:
            return text if display_nolink else ''

        classattr = 'class="{0}" '.format(jinja2.escape(cssclass)) if cssclass is not None else ''
        return jinja2.Markup("<a {0}href=\"{1}\">{2}</a>".format(classattr, jinja2.escape(link), jinja2.escape(text)))


genelinks_entry = namedtuple('genelinks_entry', 'dbname template')
def read_genelinks(dir):
    links = defaultdict(lambda: genelinks_entry(None, None))
    # blastdb.loc, blastdb_p.loc, blastdb_d.loc, etc.
    files = sorted(glob.glob(path.join(dir, 'blastdb*.loc')))
    # reversed, so blastdb.loc will take precedence
    for f in reversed(files):
        try:

            for l in f.readlines():
                if l.strip().startswith('#'):
                    continue
                line = l.rstrip('\n').split('\t')
                try:
                    links[line[2]] = genelinks_entry(dbname=line[3], template=line[4])
                except IndexError:
                    continue
            f.close()
        except OSError:
            continue

def main():
    default_template = path.join(path.dirname(__file__), 'blast2html.html.jinja')

    parser = argparse.ArgumentParser(description="Convert a BLAST XML result into a nicely readable html page",
                                     usage="{0} [-i] INPUT [-o OUTPUT] [--genelink-template URL_TEMPLATE]".format(sys.argv[0]))
    input_group = parser.add_mutually_exclusive_group(required=True)
    input_group.add_argument('positional_arg', metavar='INPUT', nargs='?', type=argparse.FileType(mode='r'),
                             help='The input Blast XML file, same as -i/--input')
    input_group.add_argument('-i', '--input', type=argparse.FileType(mode='r'), 
                             help='The input Blast XML file')

    # jinja later tries to re-open the template file, but we don't
    # care too much.
    parser.add_argument('--template', type=argparse.FileType(mode='r'), default=default_template,
                        help='The template file to use. Defaults to blast_html.html.jinja')

    parser.add_argument('--dbname', type=str, default=None,
                        help="The link text to use for external links to a gene bank database. Defaults to 'Gene Bank'")
    dblink_group = parser.add_mutually_exclusive_group()
    dblink_group.add_argument('--genelink-template', metavar='URL_TEMPLATE',
                              default='http://www.ncbi.nlm.nih.gov/nucleotide/{accession}?report=genbank&log$=nuclalign',
                              help="""A link template to link hits to a gene bank webpage. The template string is a 
                              Python format string. It can contain the following replacement elements: {id[N]}, {fullid}, 
                              {defline[N]}, {fulldefline}, {accession}, where N is a number. id[N] and defline[N] will be 
                              replaced by the Nth element of the id or defline, where '|' is the field separator. 

                              The default is 'http://www.ncbi.nlm.nih.gov/nucleotide/{accession}?report=genbank&log$=nuclalign',
                              which is a link to the NCBI nucleotide database.""")

    dblink_group.add_argument('--db-config-dir',
                              help="""The directory where databases are configured in blastdb*.loc files. These files
                              are consulted for creating a gene bank link. The files should be tab-separated tables (with lines
                              starting with '#' ignored), where the third field of a line should be a database path and the fourth
                              a genebank link template conforming to the --genelink-template option syntax.

                              This option is incompatible with --genelink-template.""")
    
    args = parser.parse_args()
    if args.input == None:
        args.input = args.positional_arg
    if args.input == None:

    templatedir, templatename = path.split(args.template.name)
    args.template.close()
    if not templatedir:
        templatedir = '.'

    if args.db_config_dir is None:
        genelinks = defaultdict(lambda: genelinks_entry(template=args.genelink_template, dbname=None))
    elif not path.isdir(args.db_config_dir):
        parser.error('db-config-dir does not exist or is not a directory')
    else:
        genelinks = read_genelinks(args.db_config_dir)

    b = BlastVisualize(args.input, templatedir, templatename, dbname=args.dbname, genelinks=genelinks)
    b.render(args.output)
    args.output.close()


if __name__ == '__main__':

116:f5066973029a

    colors = ('black', 'blue', 'green', 'magenta', 'red')

    max_scale_labels = 10

    def __init__(self, input, templatedir, templatename, genelinks):
        self.input = input
        self.templatename = templatename
        self.genelinks = genelinks

        self.blast = objectify.parse(self.input).getroot()
        self.loader = jinja2.FileSystemLoader(searchpath=templatedir)
        self.environment = jinja2.Environment(loader=self.loader,

        display_nolink: boolean, if false don't display anything if no link can be created. Default True.
        """
        
        db = hit.getroottree().getroot().BlastOutput_db

        template = self.genelinks[db].template

        if text is None:
            if text_from == 'hitid':
                text = hitid(hit)
            elif text_from == 'dbname':
                text = self.genelinks[db].dbname
            else:
                raise ValueError("Unknown value for text_from: '{0}'. Use 'hitid' or 'dbname'.".format(text_from))

        if template is None:
            return text if display_nolink else ''

        classattr = 'class="{0}" '.format(jinja2.escape(cssclass)) if cssclass is not None else ''
        return jinja2.Markup("<a {0}href=\"{1}\">{2}</a>".format(classattr, jinja2.escape(link), jinja2.escape(text)))


genelinks_entry = namedtuple('genelinks_entry', 'dbname template')
def read_blastdb(dir, default):
    links = defaultdict(lambda: default)
    # blastdb.loc, blastdb_p.loc, blastdb_d.loc, etc.
    files = sorted(glob.glob(path.join(dir, 'blastdb*.loc')))
    # reversed, so blastdb.loc will take precedence
    for f in reversed(files):
        try:

            for l in f.readlines():
                if l.strip().startswith('#'):
                    continue
                line = l.rstrip('\n').split('\t')
                try:
                    links[line[2]] = genelinks_entry(dbname=line[3] or default.dbname, template=line[4])
                except IndexError:
                    continue
            f.close()
        except OSError:
            continue

def main():
    default_template = path.join(path.dirname(__file__), 'blast2html.html.jinja')

    parser = argparse.ArgumentParser(description="Convert a BLAST XML result into a nicely readable html page",
                                     usage="{0} [-i] INPUT [-o OUTPUT] [--genelink-template URL_TEMPLATE] [--dbname DBNAME]".format(sys.argv[0]))
    input_group = parser.add_mutually_exclusive_group(required=True)
    input_group.add_argument('positional_arg', metavar='INPUT', nargs='?', type=argparse.FileType(mode='r'),
                             help='The input Blast XML file, same as -i/--input')
    input_group.add_argument('-i', '--input', type=argparse.FileType(mode='r'), 
                             help='The input Blast XML file')

    # jinja later tries to re-open the template file, but we don't
    # care too much.
    parser.add_argument('--template', type=argparse.FileType(mode='r'), default=default_template,
                        help='The template file to use. Defaults to blast_html.html.jinja')

    parser.add_argument('--dbname', type=str, default='Gene Bank',
                        help="The link text to use for external links to a gene bank database. Defaults to 'Gene Bank'")
    parser.add_argument('--genelink-template', metavar='URL_TEMPLATE',
                        default='http://www.ncbi.nlm.nih.gov/nucleotide/{accession}?report=genbank&log$=nuclalign',
                        help="""A link template to link hits to a gene bank webpage. The template string is a 
                        Python format string. It can contain the following replacement elements: {id[N]}, {fullid}, 
                        {defline[N]}, {fulldefline}, {accession}, where N is a number. id[N] and defline[N] will be 
                        replaced by the Nth element of the id or defline, where '|' is the field separator. 
                        
                        The default is 'http://www.ncbi.nlm.nih.gov/nucleotide/{accession}?report=genbank&log$=nuclalign',
                        which is a link to the NCBI nucleotide database.""")

    parser.add_argument('--db-config-dir',
                        help="""The directory where databases are configured in blastdb*.loc files. These files
                        are consulted for creating a gene bank link. The files should conform to the format that
                        Galaxy's BLAST expect, i.e. tab-separated tables (with lines starting with '#' ignored),
                        with two extra fields. The third field of a line should be a database path and the fourth
                        a genebank link template conforming to the --genelink-template option syntax. Entries in
                        these config files override links specified using --genelink-template and --dbname.""")
    
    args = parser.parse_args()
    if args.input == None:
        args.input = args.positional_arg
    if args.input == None:

    templatedir, templatename = path.split(args.template.name)
    args.template.close()
    if not templatedir:
        templatedir = '.'

    defaultentry = genelinks_entry(args.dbname, args.genelink_template)
    if args.db_config_dir is None:
        genelinks = defaultdict(lambda: defaultentry)
    elif not path.isdir(args.db_config_dir):
        parser.error('db-config-dir does not exist or is not a directory')
    else:
        genelinks = read_blastdb(args.db_config_dir, default=defaultentry)

    b = BlastVisualize(args.input, templatedir, templatename, genelinks=genelinks)
    b.render(args.output)
    args.output.close()


if __name__ == '__main__':