comparison blast2html.py @ 83:40ce2043006a draft

merge py2.6 branch

74:6ecbfebb9dd9

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

# Copyright The Hyve B.V. 2014
# License: GPL version 3 or (at your option) any higher version

from __future__ import unicode_literals

import sys
import math
import warnings
from os import path
from itertools import repeat
import argparse
from lxml import objectify
import jinja2



_filters = {}
def filter(func_or_name):
    "Decorator to register a function as filter in the current jinja environment"
    if isinstance(func_or_name, str):
        def inner(func):
            _filters[func_or_name] = func.__name__
            return func
        return inner
    else:

@filter
def fmt(val, fmt):
    return format(float(val), fmt)

@filter
def firsttitle(hit):
    return hit.Hit_def.text.split('>')[0]

@filter
def othertitles(hit):

    qseq = hsp.Hsp_qseq.text
    midline = hsp.Hsp_midline.text
    hseq = hsp.Hsp_hseq.text

    if not qframe in (1, -1):
        warnings.warn("Error in BlastXML input: Hsp node {} has a Hsp_query-frame of {}. (should be 1 or -1)".format(nodeid(hsp), qframe))
        qframe = -1 if qframe < 0 else 1
    if not hframe in (1, -1):
        warnings.warn("Error in BlastXML input: Hsp node {} has a Hsp_hit-frame of {}. (should be 1 or -1)".format(nodeid(hsp), hframe))
        hframe = -1 if hframe < 0 else 1
    
    def split(txt):
        return [txt[i:i+linewidth] for i in range(0, len(txt), linewidth)]

    for qs, mid, hs, offset in zip(split(qseq), split(midline), split(hseq), range(0, len(qseq), linewidth)):
        yield (
            "Query  {:>7}  {}  {}\n".format(qfrom+offset*qframe, qs, qfrom+(offset+len(qs)-1)*qframe) +
            "       {:7}  {}\n".format('', mid) +
            "Subject{:>7}  {}  {}".format(hfrom+offset*hframe, hs, hfrom+(offset+len(hs)-1)*hframe)
        )
        
    if qfrom+(len(qseq)-1)*qframe != qto:
        warnings.warn("Error in BlastXML input: Hsp node {} qseq length mismatch: from {} to {} length {}".format(
            nodeid(hsp), qfrom, qto, len(qseq)))
    if hfrom+(len(hseq)-1)*hframe != hto:
        warnings.warn("Error in BlastXML input: Hsp node {} hseq length mismatch: from {} to {} length {}".format(
            nodeid(hsp), hfrom, hto, len(hseq)))

    

@filter('len')

    elif frame == -1:
        return 'Minus'
    raise Exception("frame should be either +1 or -1")

def genelink(hit, type='genbank', hsp=None):
    if not isinstance(hit, str):
        hit = hitid(hit)
    link = "http://www.ncbi.nlm.nih.gov/nucleotide/{}?report={}&log$=nuclalign".format(hit, type)
    if hsp != None:
        link += "&from={}&to={}".format(hsp['Hsp_hit-from'], hsp['Hsp_hit-to'])
    return link


# javascript escape filter based on Django's, from https://github.com/dsissitka/khan-website/blob/master/templatefilters.py#L112-139
# I've removed the html escapes, since html escaping is already being performed by the template engine.

_base_js_escapes = (
    ('\\', r'\u005C'),
    ('\'', r'\u0027'),
    ('"', r'\u0022'),
    # ('>', r'\u003E'),
    # ('<', r'\u003C'),
    # ('&', r'\u0026'),
    # ('=', r'\u003D'),
    # ('-', r'\u002D'),
    # (';', r'\u003B'),
    # (u'\u2028', r'\u2028'),
    # (u'\u2029', r'\u2029')
)

# Escape every ASCII character with a value less than 32. This is
# needed a.o. to prevent html parsers from jumping out of javascript
# parsing mode.

    Javascript string literal escape. Note that this only escapes data
    for embedding within javascript string literals, not in general
    javascript snippets.
    """

    value = str(value)

    for bad, good in _js_escapes:
        value = value.replace(bad, good)

    return value

    def _addfilters(self, environment):
        for filtername, funcname in _filters.items():
            try:
                environment.filters[filtername] = getattr(self, funcname)
            except AttributeError:
                environment.filters[filtername] = globals()[funcname]

    def render(self, output):
        template = self.environment.get_template(self.templatename)

        params = (('Query ID', self.blast["BlastOutput_query-ID"]),

                  ('Query length', self.blast["BlastOutput_query-len"]),
                  ('Program', self.blast.BlastOutput_version),
                  ('Database', self.blast.BlastOutput_db),
        )

        output.write(template.render(blast=self.blast,
                                     iterations=self.blast.BlastOutput_iterations.Iteration,
                                     colors=self.colors,
                                     # match_colors=self.match_colors(),
                                     # hit_info=self.hit_info(),
                                     genelink=genelink,
                                     params=params))

    @filter
    def match_colors(self, result):
        """
        An iterator that yields lists of length-color pairs. 

            def hsp_val(path):
                return (float(hsp[path]) for hsp in hsps)

            yield dict(hit = hit,
                      title = firsttitle(hit),
                       maxscore = "{:.1f}".format(max(hsp_val('Hsp_bit-score'))),
                       totalscore = "{:.1f}".format(sum(hsp_val('Hsp_bit-score'))),
                       cover = "{:.0%}".format(cover_count / query_length),
                       e_value = "{:.4g}".format(min(hsp_val('Hsp_evalue'))),
                       # FIXME: is this the correct formula vv?
                       ident = "{:.0%}".format(float(min(hsp.Hsp_identity / blastxml_len(hsp) for hsp in hsps))),
                       accession = hit.Hit_accession)


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="{} [-i] INPUT [-o OUTPUT]".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')

83:40ce2043006a

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

# Actually this program works with both python 2 and 3, tested against python 2.6

# Copyright The Hyve B.V. 2014
# License: GPL version 3 or (at your option) any higher version

from __future__ import unicode_literals, division

import sys
import math
import warnings
import six, codecs
from six.moves import builtins
from os import path
from itertools import repeat
import argparse
from lxml import objectify
import jinja2



_filters = dict(float='float')
def filter(func_or_name):
    "Decorator to register a function as filter in the current jinja environment"
    if isinstance(func_or_name, six.string_types):
        def inner(func):
            _filters[func_or_name] = func.__name__
            return func
        return inner
    else:

@filter
def fmt(val, fmt):
    return format(float(val), fmt)

@filter
def numfmt(val):
    """Format numbers in decimal notation, but without excessive trailing 0's.
    Default python float formatting will use scientific notation for some values,
    or append trailing zeros with the 'f' format type, and the number of digits differs
    between python 2 and 3."""
    fpart, ipart = math.modf(val)
    if fpart == 0:
        return str(int(val))
    # round to 10 to get identical representations in python 2 and 3
    s = format(round(val, 10), '.10f').rstrip('0')
    if s[-1] == '.':
        s += '0'
    return s

@filter
def firsttitle(hit):
    return hit.Hit_def.text.split('>')[0]

@filter
def othertitles(hit):

    qseq = hsp.Hsp_qseq.text
    midline = hsp.Hsp_midline.text
    hseq = hsp.Hsp_hseq.text

    if not qframe in (1, -1):
        warnings.warn("Error in BlastXML input: Hsp node {0} has a Hsp_query-frame of {1}. (should be 1 or -1)".format(nodeid(hsp), qframe))
        qframe = -1 if qframe < 0 else 1
    if not hframe in (1, -1):
        warnings.warn("Error in BlastXML input: Hsp node {0} has a Hsp_hit-frame of {1}. (should be 1 or -1)".format(nodeid(hsp), hframe))
        hframe = -1 if hframe < 0 else 1
    
    def split(txt):
        return [txt[i:i+linewidth] for i in range(0, len(txt), linewidth)]

    for qs, mid, hs, offset in zip(split(qseq), split(midline), split(hseq), range(0, len(qseq), linewidth)):
        yield (
            "Query  {0:>7}  {1}  {2}\n".format(qfrom+offset*qframe, qs, qfrom+(offset+len(qs)-1)*qframe) +
            "       {0:7}  {1}\n".format('', mid) +
            "Subject{0:>7}  {1}  {2}".format(hfrom+offset*hframe, hs, hfrom+(offset+len(hs)-1)*hframe)
        )
        
    if qfrom+(len(qseq)-1)*qframe != qto:
        warnings.warn("Error in BlastXML input: Hsp node {0} qseq length mismatch: from {1} to {2} length {3}".format(
            nodeid(hsp), qfrom, qto, len(qseq)))
    if hfrom+(len(hseq)-1)*hframe != hto:
        warnings.warn("Error in BlastXML input: Hsp node {0} hseq length mismatch: from {1} to {2} length {3}".format(
            nodeid(hsp), hfrom, hto, len(hseq)))

    

@filter('len')

    elif frame == -1:
        return 'Minus'
    raise Exception("frame should be either +1 or -1")

def genelink(hit, type='genbank', hsp=None):
    if not isinstance(hit, six.string_types):
        hit = hitid(hit)
    link = "http://www.ncbi.nlm.nih.gov/nucleotide/{0}?report={1}&log$=nuclalign".format(hit, type)
    if hsp != None:
        link += "&from={0}&to={1}".format(hsp['Hsp_hit-from'], hsp['Hsp_hit-to'])
    return link


# javascript escape filter based on Django's, from https://github.com/dsissitka/khan-website/blob/master/templatefilters.py#L112-139
# I've removed the html escapes, since html escaping is already being performed by the template engine.

# The r'\u0027' syntax doesn't work the way we need to in python 2.6 with unicode_literals
_base_js_escapes = (
    ('\\', '\\u005C'),
    ('\'', '\\u0027'),
    ('"', '\\u0022'),
    # ('>', '\\u003E'),
    # ('<', '\\u003C'),
    # ('&', '\\u0026'),
    # ('=', '\\u003D'),
    # ('-', '\\u002D'),
    # (';', '\\u003B'),
    (u'\u2028', '\\u2028'),
    (u'\u2029', '\\u2029')
)

# Escape every ASCII character with a value less than 32. This is
# needed a.o. to prevent html parsers from jumping out of javascript
# parsing mode.

    Javascript string literal escape. Note that this only escapes data
    for embedding within javascript string literals, not in general
    javascript snippets.
    """

    value = six.text_type(value)

    for bad, good in _js_escapes:
        value = value.replace(bad, good)

    return value

    def _addfilters(self, environment):
        for filtername, funcname in _filters.items():
            try:
                environment.filters[filtername] = getattr(self, funcname)
            except AttributeError:
                try:
                    environment.filters[filtername] = globals()[funcname]
                except KeyError:
                    environment.filters[filtername] = getattr(builtins, funcname)

    def render(self, output):
        template = self.environment.get_template(self.templatename)

        params = (('Query ID', self.blast["BlastOutput_query-ID"]),

                  ('Query length', self.blast["BlastOutput_query-len"]),
                  ('Program', self.blast.BlastOutput_version),
                  ('Database', self.blast.BlastOutput_db),
        )

        result = template.render(blast=self.blast,
                                 iterations=self.blast.BlastOutput_iterations.Iteration,
                                 colors=self.colors,
                                 genelink=genelink,
                                 params=params)
        if six.PY2:
            result = result.encode('utf-8')
        output.write(result)

    @filter
    def match_colors(self, result):
        """
        An iterator that yields lists of length-color pairs. 

            def hsp_val(path):
                return (float(hsp[path]) for hsp in hsps)

            yield dict(hit = hit,
                       title = firsttitle(hit),
                       maxscore = "{0:.1f}".format(max(hsp_val('Hsp_bit-score'))),
                       totalscore = "{0:.1f}".format(sum(hsp_val('Hsp_bit-score'))),
                       cover = "{0:.0%}".format(cover_count / query_length),
                       e_value = "{0:.4g}".format(min(hsp_val('Hsp_evalue'))),
                       # 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))),
                       accession = hit.Hit_accession)


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]".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')