Homec4science
Diffusion invenio-infoscience (maint-1.1) Browse

R3600/modules/bibformat/lib/bibformat_utils.py
4ba7d6c55f7bmaint-1.1

Last Change
Raw File

bibformat_utils.py
View Options

# -*- coding: utf-8 -*-
##
## This file is part of CDS Invenio.
## Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007, 2008 CERN.
##
## CDS Invenio is free software; you can redistribute it and/or
## modify it under the terms of the GNU General Public License as
## published by the Free Software Foundation; either version 2 of the
## License, or (at your option) any later version.
##
## CDS Invenio is distributed in the hope that it will be useful, but
## WITHOUT ANY WARRANTY; without even the implied warranty of
## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
## General Public License for more details.
##
## You should have received a copy of the GNU General Public License
## along with CDS Invenio; if not, write to the Free Software Foundation, Inc.,
## 59 Temple Place, Suite 330, Boston, MA 02111-1307, USA.
"""
Utilities for special formatting of records.
API functions: highlight, get_contextual_content, encode_for_xml
Used mainly by BibFormat elements.
"""
__revision__ = "$Id$"
import re
import zlib
import os
from invenio.config import \
CFG_OAI_ID_FIELD, \
CFG_WEBSEARCH_FULLTEXT_SNIPPETS, \
CFG_WEBSEARCH_FULLTEXT_SNIPPETS_WORDS, \
CFG_PATH_PDFTOTEXT
from invenio.dbquery import run_sql
from invenio.urlutils import string_to_numeric_char_reference
from invenio.textutils import encode_for_xml
from invenio.shellutils import run_shell_command
def highlight(text, keywords=None, prefix_tag='<strong>', suffix_tag="</strong>"):
"""
Returns text with all words highlighted with given tags (this
function places 'prefix_tag' and 'suffix_tag' before and after
words from 'keywords' in 'text').
for example set prefix_tag='<b style="color: black; background-color: rgb(255, 255, 102);">' and suffix_tag="</b>"
@param text: the text to modify
@param keywords: a list of string
@return: highlighted text
"""
if not keywords:
return text
#FIXME decide if non english accentuated char should be desaccentuaded
def replace_highlight(match):
""" replace match.group() by prefix_tag + match.group() + suffix_tag"""
return prefix_tag + match.group() + suffix_tag
#Build a pattern of the kind keyword1 | keyword2 | keyword3
pattern = '|'.join(keywords)
compiled_pattern = re.compile(pattern, re.IGNORECASE)
#Replace and return keywords with prefix+keyword+suffix
return compiled_pattern.sub(replace_highlight, text)
def get_contextual_content(text, keywords, max_lines=2):
"""
Returns some lines from a text contextually to the keywords in
'keywords_string'
@param text: the text from which we want to get contextual content
@param keywords: a list of keyword strings ("the context")
@param max_lines: the maximum number of line to return from the record
@return: a string
"""
def grade_line(text_line, keywords):
"""
Grades a line according to keywords.
grade = number of keywords in the line
"""
grade = 0
for keyword in keywords:
grade += text_line.upper().count(keyword.upper())
return grade
#Grade each line according to the keywords
lines = text.split('.')
#print 'lines: ',lines
weights = [grade_line(line, keywords) for line in lines]
#print 'line weights: ', weights
def grade_region(lines_weight):
"""
Grades a region. A region is a set of consecutive lines.
grade = sum of weights of the line composing the region
"""
grade = 0
for weight in lines_weight:
grade += weight
return grade
if max_lines > 1:
region_weights = []
for index_weight in range(len(weights)- max_lines + 1):
region_weights.append(grade_region(weights[index_weight:(index_weight+max_lines)]))
weights = region_weights
#print 'region weights: ',weights
#Returns line with maximal weight, and (max_lines - 1) following lines.
index_with_highest_weight = 0
highest_weight = 0
i = 0
for weight in weights:
if weight > highest_weight:
index_with_highest_weight = i
highest_weight = weight
i += 1
#print 'highest weight', highest_weight
if index_with_highest_weight+max_lines > len(lines):
return lines[index_with_highest_weight:]
else:
return lines[index_with_highest_weight:index_with_highest_weight+max_lines]
def record_get_xml(recID, format='xm', decompress=zlib.decompress,
on_the_fly=False):
"""
Returns an XML string of the record given by recID.
The function builds the XML directly from the database,
without using the standard formatting process.
'format' allows to define the flavour of XML:
- 'xm' for standard XML
- 'marcxml' for MARC XML
- 'oai_dc' for OAI Dublin Core
- 'xd' for XML Dublin Core
If record does not exist, returns empty string.
If the record is deleted, returns an empty MARCXML (with recid
controlfield, OAI ID fields and 980__c=DELETED)
@param recID: the id of the record to retrieve
@param on_the_fly: if False, try to fetch precreated one in database
@return: the xml string of the record
"""
from invenio.search_engine import record_exists
def get_fieldvalues(recID, tag):
"""Return list of field values for field TAG inside record RECID."""
out = []
if tag == "001___":
# we have asked for recID that is not stored in bibXXx tables
out.append(str(recID))
else:
# we are going to look inside bibXXx tables
digit = tag[0:2]
bx = "bib%sx" % digit
bibx = "bibrec_bib%sx" % digit
query = "SELECT bx.value FROM %s AS bx, %s AS bibx WHERE bibx.id_bibrec='%s' AND bx.id=bibx.id_bibxxx AND bx.tag LIKE '%s'" \
"ORDER BY bibx.field_number, bx.tag ASC" % (bx, bibx, recID, tag)
res = run_sql(query)
for row in res:
out.append(row[0])
return out
def get_creation_date(recID, fmt="%Y-%m-%d"):
"Returns the creation date of the record 'recID'."
out = ""
res = run_sql("SELECT DATE_FORMAT(creation_date,%s) FROM bibrec WHERE id=%s", (fmt, recID), 1)
if res:
out = res[0][0]
return out
def get_modification_date(recID, fmt="%Y-%m-%d"):
"Returns the date of last modification for the record 'recID'."
out = ""
res = run_sql("SELECT DATE_FORMAT(modification_date,%s) FROM bibrec WHERE id=%s", (fmt, recID), 1)
if res:
out = res[0][0]
return out
#_ = gettext_set_language(ln)
out = ""
# sanity check:
record_exist_p = record_exists(recID)
if record_exist_p == 0: # doesn't exist
return out
# print record opening tags, if needed:
if format == "marcxml" or format == "oai_dc":
out += " <record>\n"
out += " <header>\n"
for identifier in get_fieldvalues(recID, CFG_OAI_ID_FIELD):
out += " <identifier>%s</identifier>\n" % identifier
out += " <datestamp>%s</datestamp>\n" % get_modification_date(recID)
out += " </header>\n"
out += " <metadata>\n"
if format.startswith("xm") or format == "marcxml":
res = None
if on_the_fly == False:
# look for cached format existence:
query = """SELECT value FROM bibfmt WHERE
id_bibrec='%s' AND format='%s'""" % (recID, format)
res = run_sql(query, None, 1)
if res and record_exist_p == 1:
# record 'recID' is formatted in 'format', so print it
out += "%s" % decompress(res[0][0])
else:
# record 'recID' is not formatted in 'format' -- they are
# not in "bibfmt" table; so fetch all the data from
# "bibXXx" tables:
if format == "marcxml":
out += """ <record xmlns="http://www.loc.gov/MARC21/slim">\n"""
out += " <controlfield tag=\"001\">%d</controlfield>\n" % int(recID)
elif format.startswith("xm"):
out += """ <record>\n"""
out += " <controlfield tag=\"001\">%d</controlfield>\n" % int(recID)
if record_exist_p == -1:
# deleted record, so display only OAI ID and 980:
oai_ids = get_fieldvalues(recID, CFG_OAI_ID_FIELD)
if oai_ids:
out += "<datafield tag=\"%s\" ind1=\"%s\" ind2=\"%s\"><subfield code=\"%s\">%s</subfield></datafield>\n" % \
(CFG_OAI_ID_FIELD[0:3],
CFG_OAI_ID_FIELD[3:4],
CFG_OAI_ID_FIELD[4:5],
CFG_OAI_ID_FIELD[5:6],
oai_ids[0])
out += "<datafield tag=\"980\" ind1=\" \" ind2=\" \"><subfield code=\"c\">DELETED</subfield></datafield>\n"
else:
# controlfields
query = "SELECT b.tag,b.value,bb.field_number FROM bib00x AS b, bibrec_bib00x AS bb "\
"WHERE bb.id_bibrec='%s' AND b.id=bb.id_bibxxx AND b.tag LIKE '00%%' "\
"ORDER BY bb.field_number, b.tag ASC" % recID
res = run_sql(query)
for row in res:
field, value = row[0], row[1]
value = encode_for_xml(value)
out += """ <controlfield tag="%s">%s</controlfield>\n""" % \
(encode_for_xml(field[0:3]), value)
# datafields
i = 1 # Do not process bib00x and bibrec_bib00x, as
# they are controlfields. So start at bib01x and
# bibrec_bib00x (and set i = 0 at the end of
# first loop)
for digit1 in range(0, 10):
for digit2 in range(i, 10):
bx = "bib%d%dx" % (digit1, digit2)
bibx = "bibrec_bib%d%dx" % (digit1, digit2)
query = "SELECT b.tag,b.value,bb.field_number FROM %s AS b, %s AS bb "\
"WHERE bb.id_bibrec='%s' AND b.id=bb.id_bibxxx AND b.tag LIKE '%s%%' "\
"ORDER BY bb.field_number, b.tag ASC" % (bx,
bibx,
recID,
str(digit1)+str(digit2))
res = run_sql(query)
field_number_old = -999
field_old = ""
for row in res:
field, value, field_number = row[0], row[1], row[2]
ind1, ind2 = field[3], field[4]
if ind1 == "_" or ind1 == "":
ind1 = " "
if ind2 == "_" or ind2 == "":
ind2 = " "
# print field tag
if field_number != field_number_old or \
field[:-1] != field_old[:-1]:
if field_number_old != -999:
out += """ </datafield>\n"""
out += """ <datafield tag="%s" ind1="%s" ind2="%s">\n""" % \
(encode_for_xml(field[0:3]),
encode_for_xml(ind1),
encode_for_xml(ind2))
field_number_old = field_number
field_old = field
# print subfield value
value = encode_for_xml(value)
out += """ <subfield code="%s">%s</subfield>\n""" % \
(encode_for_xml(field[-1:]), value)
# all fields/subfields printed in this run, so close the tag:
if field_number_old != -999:
out += """ </datafield>\n"""
i = 0 # Next loop should start looking at bib%0 and bibrec_bib00x
# we are at the end of printing the record:
out += " </record>\n"
elif format == "xd" or format == "oai_dc":
# XML Dublin Core format, possibly OAI -- select only some bibXXx fields:
out += """ <dc xmlns="http://purl.org/dc/elements/1.1/"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://purl.org/dc/elements/1.1/
http://www.openarchives.org/OAI/1.1/dc.xsd">\n"""
if record_exist_p == -1:
out += ""
else:
for f in get_fieldvalues(recID, "041__a"):
out += " <language>%s</language>\n" % f
for f in get_fieldvalues(recID, "100__a"):
out += " <creator>%s</creator>\n" % encode_for_xml(f)
for f in get_fieldvalues(recID, "700__a"):
out += " <creator>%s</creator>\n" % encode_for_xml(f)
for f in get_fieldvalues(recID, "245__a"):
out += " <title>%s</title>\n" % encode_for_xml(f)
for f in get_fieldvalues(recID, "65017a"):
out += " <subject>%s</subject>\n" % encode_for_xml(f)
for f in get_fieldvalues(recID, "8564_u"):
out += " <identifier>%s</identifier>\n" % encode_for_xml(f)
for f in get_fieldvalues(recID, "520__a"):
out += " <description>%s</description>\n" % encode_for_xml(f)
out += " <date>%s</date>\n" % get_creation_date(recID)
out += " </dc>\n"
# print record closing tags, if needed:
if format == "marcxml" or format == "oai_dc":
out += " </metadata>\n"
out += " </record>\n"
return out
def parse_tag(tag):
"""
Parse a marc code and decompose it in a table with:
0-tag 1-indicator1 2-indicator2 3-subfield
The first 3 chars always correspond to tag.
The indicators are optional. However they must both be indicated, or both ommitted.
If indicators are ommitted or indicated with underscore '_', they mean "No indicator".
"No indicator" is also equivalent indicator marked as whitespace.
The subfield is optional. It can optionally be preceded by a dot '.' or '$$' or '$'
Any of the chars can be replaced by wildcard %
THE FUNCTION DOES NOT CHECK WELLFORMNESS OF 'tag'
Any empty chars is not considered
For example:
>> parse_tag('245COc') = ['245', 'C', 'O', 'c']
>> parse_tag('245C_c') = ['245', 'C', '', 'c']
>> parse_tag('245__c') = ['245', '', '', 'c']
>> parse_tag('245__$$c') = ['245', '', '', 'c']
>> parse_tag('245__$c') = ['245', '', '', 'c']
>> parse_tag('245 $c') = ['245', '', '', 'c']
>> parse_tag('245 $$c') = ['245', '', '', 'c']
>> parse_tag('245__.c') = ['245', '', '', 'c']
>> parse_tag('245 .c') = ['245', '', '', 'c']
>> parse_tag('245C_$c') = ['245', 'C', '', 'c']
>> parse_tag('245CO$$c') = ['245', 'C', 'O', 'c']
>> parse_tag('245C_.c') = ['245', 'C', '', 'c']
>> parse_tag('245$c') = ['245', '', '', 'c']
>> parse_tag('245.c') = ['245', '', '', 'c']
>> parse_tag('245$$c') = ['245', '', '', 'c']
>> parse_tag('245__%') = ['245', '', '', '']
>> parse_tag('245__$$%') = ['245', '', '', '']
>> parse_tag('245__$%') = ['245', '', '', '']
>> parse_tag('245 $%') = ['245', '', '', '']
>> parse_tag('245 $$%') = ['245', '', '', '']
>> parse_tag('245$%') = ['245', '', '', '']
>> parse_tag('245.%') = ['245', '', '', '']
>> parse_tag('245$$%') = ['245', '', '', '']
>> parse_tag('2%5$$a') = ['2%5', '', '', 'a']
"""
p_tag = ['', '', '', ''] # tag, ind1, ind2, code
tag = tag.replace(" ", "") # Remove empty characters
tag = tag.replace("$", "") # Remove $ characters
tag = tag.replace(".", "") # Remove . characters
#tag = tag.replace("_", "") # Remove _ characters
p_tag[0] = tag[0:3] # tag
if len(tag) == 4:
p_tag[3] = tag[3] # subfield
elif len(tag) == 5:
ind1 = tag[3] # indicator 1
if ind1 != "_":
p_tag[1] = ind1
ind2 = tag[4] # indicator 2
if ind2 != "_":
p_tag[2] = ind2
elif len(tag) == 6:
p_tag[3] = tag[5] # subfield
ind1 = tag[3] # indicator 1
if ind1 != "_":
p_tag[1] = ind1
ind2 = tag[4] # indicator 2
if ind2 != "_":
p_tag[2] = ind2
return p_tag
def get_all_fieldvalues(recID, tags_in):
"""
Returns list of values that belong to fields in tags_in for record
with given recID.
Note that when a partial 'tags_in' is specified (eg. '100__'), the
subfields of all corresponding datafields are returned all 'mixed'
together. Eg. with:
123 100__ $a Ellis, J $u CERN
123 100__ $a Smith, K
>> get_all_fieldvalues(123, '100__')
['Ellis, J', 'CERN', 'Smith, K']
"""
out = []
if type(tags_in) is not list:
tags_in = [tags_in, ]
dict_of_tags_out = {}
if not tags_in:
for i in range(0, 10):
for j in range(0, 10):
dict_of_tags_out["%d%d%%" % (i, j)] = '%'
else:
for tag in tags_in:
if len(tag) == 0:
for i in range(0, 10):
for j in range(0, 10):
dict_of_tags_out["%d%d%%" % (i, j)] = '%'
elif len(tag) == 1:
for j in range(0, 10):
dict_of_tags_out["%s%d%%" % (tag, j)] = '%'
elif len(tag) <= 5:
dict_of_tags_out["%s%%" % tag] = '%'
else:
dict_of_tags_out[tag[0:5]] = tag[5:6]
tags_out = dict_of_tags_out.keys()
tags_out.sort()
# search all bibXXx tables as needed:
for tag in tags_out:
digits = tag[0:2]
try:
intdigits = int(digits)
if intdigits < 0 or intdigits > 99:
raise ValueError
except ValueError:
# invalid tag value asked for
continue
bx = "bib%sx" % digits
bibx = "bibrec_bib%sx" % digits
query = "SELECT b.tag,b.value,bb.field_number FROM %s AS b, %s AS bb "\
"WHERE bb.id_bibrec=%%s AND b.id=bb.id_bibxxx AND b.tag LIKE %%s"\
"ORDER BY bb.field_number, b.tag ASC" % (bx, bibx)
res = run_sql(query, (recID, str(tag)+dict_of_tags_out[tag]))
# go through fields:
for row in res:
field, value, field_number = row[0], row[1], row[2]
out.append(value)
return out
re_bold_latex = re.compile('\$?\\\\textbf\{(?P<content>.*?)\}\$?')
re_emph_latex = re.compile('\$?\\\\emph\{(?P<content>.*?)\}\$?')
re_generic_start_latex = re.compile('\$?\\\\begin\{(?P<content>.*?)\}\$?')
re_generic_end_latex = re.compile('\$?\\\\end\{(?P<content>.*?)\}\$?')
re_verbatim_env_latex = re.compile('\\\\begin\{verbatim.*?\}(?P<content>.*?)\\\\end\{verbatim.*?\}')
def latex_to_html(text):
"""
Do some basic interpretation of LaTeX input. Gives some nice
results when used in combination with JSMath.
"""
# Process verbatim environment first
def make_verbatim(match_obj):
"""Replace all possible special chars by HTML character
entities, so that they are not interpreted by further commands"""
return '<br/><pre class="tex2math_ignore">' + \
string_to_numeric_char_reference(match_obj.group('content')) + \
'</pre><br/>'
text = re_verbatim_env_latex.sub(make_verbatim, text)
# Remove trailing "line breaks"
text = text.strip('\\\\')
# Process special characters
text = text.replace("\\%", "%")
text = text.replace("\\#", "#")
text = text.replace("\\$", "$")
text = text.replace("\\&", "&amp;")
text = text.replace("\\{", "{")
text = text.replace("\\}", "}")
text = text.replace("\\_", "_")
text = text.replace("\\^{} ", "^")
text = text.replace("\\~{} ", "~")
text = text.replace("\\textregistered", "&#0174;")
text = text.replace("\\copyright", "&#0169;")
text = text.replace("\\texttrademark", "&#0153; ")
# Remove commented lines and join lines
text = '\\\\'.join([line for line in text.split('\\\\') \
if not line.lstrip().startswith('%')])
# Line breaks
text = text.replace('\\\\', '<br/>')
# Non-breakable spaces
text = text.replace('~', '&nbsp;')
# Styled text
def make_bold(match_obj):
"Make the found pattern bold"
# FIXME: check if it is valid to have this inside a formula
return '<b>' + match_obj.group('content') + '</b>'
text = re_bold_latex.sub(make_bold, text)
def make_emph(match_obj):
"Make the found pattern emphasized"
# FIXME: for the moment, remove as it could cause problem in
# the case it is used in a formula. To be check if it is valid.
return ' ' + match_obj.group('content') + ''
text = re_emph_latex.sub(make_emph, text)
# Lists
text = text.replace('\\begin{enumerate}', '<ol>')
text = text.replace('\\end{enumerate}', '</ol>')
text = text.replace('\\begin{itemize}', '<ul>')
text = text.replace('\\end{itemize}', '</ul>')
text = text.replace('\\item', '<li>')
# Remove remaining non-processed tags
text = re_generic_start_latex.sub('', text)
text = re_generic_end_latex.sub('', text)
return text
def get_pdf_snippets(recID, patterns,
nb_words_around=CFG_WEBSEARCH_FULLTEXT_SNIPPETS_WORDS,
max_snippets=CFG_WEBSEARCH_FULLTEXT_SNIPPETS):
"""
Extract text snippets around 'patterns' from the newest PDF file of 'recID'
The search is case-insensitive.
The snippets are meant to look like in the results of the popular search
engine: using " ... " between snippets.
For empty patterns it returns ""
"""
from invenio.bibdocfile import BibRecDocs
text_path = ""
text_path_courtesy = ""
for bd in BibRecDocs(recID).list_bibdocs():
if bd.get_text():
text_path = bd.get_text_path()
text_path_courtesy = bd.get_status()
break # stop at the first good PDF textable file
if text_path:
out = get_text_snippets(text_path, patterns, nb_words_around, max_snippets)
if not out:
# no hit, so check stemmed versions:
from invenio.bibindex_engine_stemmer import stem
stemmed_patterns = [stem(p, 'en') for p in patterns]
print stemmed_patterns
out = get_text_snippets(text_path, stemmed_patterns, nb_words_around, max_snippets)
if out:
out_courtesy = ""
if text_path_courtesy:
out_courtesy = '<strong>Snippets courtesy of ' + text_path_courtesy + '</strong><br>'
return """<div class="snippetbox">%s%s</div>""" % (out_courtesy, out)
else:
return ""
else:
return ""
def get_text_snippets(textfile_path, patterns, nb_words_around, max_snippets):
"""
Extract text snippets around 'patterns' from file found at 'textfile_path'
The snippets are meant to look like in the results of the popular search
engine: using " ... " between snippets.
For empty patterns it returns ""
The idea is to first produce big snippets with grep and narrow them
TODO: - distinguish the beginning of sentences and try to make the snippets
start there
"""
if len(patterns) == 0:
return ""
# the max number of words that can still be added to the snippet
words_left = max_snippets * (nb_words_around * 2 + 1)
# Assuming that there will be at least one word per line we can produce the
# big snippets like this
cmd = "grep -i -A%s -B%s -m%s"
cmdargs = [str(nb_words_around), str(nb_words_around), str(max_snippets)]
for p in patterns:
cmd += " -e %s"
cmdargs.append(p)
cmd += " %s"
cmdargs.append(textfile_path)
(dummy1, output, dummy2) = run_shell_command(cmd, cmdargs)
result = []
big_snippets = output.split("--")
# cut the snippets to match the nb_words_around parameter precisely:
for s in big_snippets:
small_snippet = cut_out_snippet(s, patterns, nb_words_around, words_left)
#count words
words_left -= len(small_snippet.split())
#if words_left <= 0:
#print "Error: snippet too long"
result.append(small_snippet)
# combine snippets
out = ""
for snippet in result:
if snippet:
if out:
out += "<br>"
out += "..." + highlight(snippet, patterns) + "..."
return out
def cut_out_snippet(text, patterns, nb_words_around, max_words):
# the snippet can include many occurances of the patterns if they are not
# further appart than 2 * nb_words_around
def starts_with_any(word, patterns):
# Check whether the word's beginning matches any of the patterns.
# The second argument is an array of patterns to match.
ret = False
lower_case = word.lower()
for p in patterns:
if lower_case.startswith(str(p).lower()):
ret = True
break
return ret
# make the nb_words_around smaller if required by max_words
# to make sure that at least one pattern is included
while nb_words_around * 2 + 1 > max_words:
nb_words_around -= 1
if nb_words_around < 1:
return ""
snippet = ""
words = text.split()
last_written_word = -1
i = 0
while i < len(words):
if starts_with_any(words[i], patterns):
# add part before first or following occurance of a word
j = max(last_written_word + 1, i - nb_words_around)
while j < i:
snippet += (" " + words[j])
j += 1
# write the pattern
snippet += (" " + words[i])
last_written_word = i
# write the suffix. If pattern found, break
j = 1
while j <= nb_words_around and i + j < len(words):
if starts_with_any(words[i+j], patterns):
break
else:
snippet += (" " + words[i+j])
last_written_word = i + j
j += 1
i += j
else:
i += 1
# apply max_words param if needed
snippet_words = snippet.split()
length = len(snippet_words)
if (length > max_words):
j = 0
shorter_snippet = ""
while j < max_words:
shorter_snippet += " " + snippet_words[j]
j += 1
return shorter_snippet
else:
return snippet
c4science · Help