comparison blast2html.py @ 115:0c2a03f9740b

make external gene bank name configurable

114:4f0ed3b5ae46

import warnings
import six, codecs, io
from six.moves import builtins
from os import path
from itertools import repeat
from collections import defaultdict
import glob
import argparse
from lxml import objectify
import jinja2

    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,

                       # FIXME: is this the correct formula vv?
                       # float(...) because non-flooring division doesn't work with lxml elements in python 2.6
                       ident = "{0:.0%}".format(float(min(float(hsp.Hsp_identity) / blastxml_len(hsp) for hsp in hsps))))

    @filter
    def genelink(self, hit, text=None, clas=None, display_nolink=True):
        """Create a html link from a hit node to a configured gene bank webpage.
        text: The text of the link, defaults to the hit_id
        clas: extra css classes that will be added to the <a> element
        display_nolink: boolean, if false don't display anything if no link can be created. Default True.
        """
        
        if text is None:
            text = hitid(hit)

        db = hit.getroottree().getroot().BlastOutput_db

        if isinstance(self.genelinks, six.string_types):
            template = self.genelinks
        else:
            template = self.genelinks.get(db)
        if template is None:
            return text if display_nolink else ''
        args = dict(id=hitid(hit).split('|'),
                    fullid=hitid(hit),
                    defline=str(hit.Hit_def).split(' ', 1)[0].split('|'),
                    fulldefline=str(hit.Hit_def).split(' ', 1)[0],
                    accession=str(hit.Hit_accession))

            link = template.format(**args)
        except Exception as e:
            warnings.warn('Error in formatting gene bank link {} with {}: {}'.format(template, args, e))
            return text if display_nolink else ''

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


def read_genelinks(dir):
    links = {}
    # 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]] = line[3]
                except IndexError:
                    continue
            f.close()
        except OSError:
            continue

    # handle the errors. This introduces a small race condition when
    # 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')
    
    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.

        #         self.write(i)
        # args.output.writelines = fixed_writelines

        args.output.close()
        args.output = io.open(args.output.name, 'w', encoding='utf-8')
    
    templatedir, templatename = path.split(args.template.name)
    args.template.close()
    if not templatedir:
        templatedir = '.'

    if args.db_config_dir is None:
        genelinks = args.genelink_template
    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, genelinks)
    b.render(args.output)
    args.output.close()


if __name__ == '__main__':

115:0c2a03f9740b

import warnings
import six, codecs, io
from six.moves import builtins
from os import path
from itertools import repeat
from collections import defaultdict, namedtuple
import glob
import argparse
from lxml import objectify
import jinja2

    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,

                       # FIXME: is this the correct formula vv?
                       # float(...) because non-flooring division doesn't work with lxml elements in python 2.6
                       ident = "{0:.0%}".format(float(min(float(hsp.Hsp_identity) / blastxml_len(hsp) for hsp in hsps))))

    @filter
    def genelink(self, hit, text=None, text_from='hitid', cssclass=None, display_nolink=True):
        """Create a html link from a hit node to a configured gene bank webpage.
        text: The text of the link. If not set applies text_from.
        text_from: string, if text is not specified, take it from specified source. Either 'hitid' (default) or 'dbname'.
        cssclass: extra css classes that will be added to the <a> element
        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 ''

        args = dict(id=hitid(hit).split('|'),
                    fullid=hitid(hit),
                    defline=str(hit.Hit_def).split(' ', 1)[0].split('|'),
                    fulldefline=str(hit.Hit_def).split(' ', 1)[0],
                    accession=str(hit.Hit_accession))

            link = template.format(**args)
        except Exception as e:
            warnings.warn('Error in formatting gene bank link {} with {}: {}'.format(template, args, e))
            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

    # handle the errors. This introduces a small race condition when
    # 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.

        #         self.write(i)
        # args.output.writelines = fixed_writelines

        args.output.close()
        args.output = io.open(args.output.name, 'w', encoding='utf-8')

    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__':