patacrep/patacrep/index.py

"""Manage indexes.

Generate indexes files for the songbook compilation. This is a replacement for
the original makeindex program written in C that produces an index file (.sbx)
from a file generated by the latex compilation of the songbook (.sxd).
"""

import locale
import unidecode
import re

from patacrep import authors
from patacrep import encoding
from patacrep.latex import tex2plain

EOL = "\n"

# Pattern set to ignore latex command in title prefix
KEYWORD_PATTERN = re.compile(r"^%(\w+)\s?(.*)$", re.LOCALE)
FIRST_LETTER_PATTERN = re.compile(r"^(?:\{?\\\w+\}?)*[^\w]*(\w)", re.LOCALE)

def process_sxd(filename):
    """Parse sxd file.

    Return an Index object.
    """
    data = []
    index_file = None
    with encoding.open_read(filename) as index_file:
        for line in index_file:
            data.append(line.strip())

    i = 1
    idx = Index(data[0])

    while len(data) > i and data[i].startswith('%'):
        keywords = KEYWORD_PATTERN.match(data[i]).groups()
        idx.add_keyword(keywords[0], keywords[1])
        i += 1

    idx.compile_keywords()
    for i in range(i, len(data), 3):
        entry = data[i:i + 3]
        idx.add(entry[0], entry[1], entry[2])

    return idx


class Index(object):
    """Title, author or scripture Index representation."""

    def __init__(self, indextype):
        self.data = dict()
        self.keywords = dict()
        self.authwords = dict()
        self.prefix_patterns = []
        if indextype == "TITLE INDEX DATA FILE":
            self.indextype = "TITLE"
        elif indextype == "SCRIPTURE INDEX DATA FILE":
            self.indextype = "SCRIPTURE"
        elif indextype == "AUTHOR INDEX DATA FILE":
            self.indextype = "AUTHOR"
        else:
            self.indextype = ""

    @staticmethod
    def get_first_letter(key):
        """Return the uppercase first letter of key."""
        try:
            letter = FIRST_LETTER_PATTERN.match(key).group(1)
        except AttributeError:
            # classify as number all the non letter characters
            letter = "0"
        if re.match(r'\d', letter):
            letter = '0-9'
        return letter.upper()

    def add_keyword(self, key, word):
        """Add 'word' to self.keywords[key]."""
        if not key in self.keywords:
            self.keywords[key] = []
        self.keywords[key].append(word)

    def compile_keywords(self):
        """Turn keywords (self.keywords) into regular expressions."""
        if self.indextype == "TITLE":
            if 'prefix' in self.keywords:
                for prefix in self.keywords['prefix']:
                    self.prefix_patterns.append(re.compile(
                        r"^({prefix})(\b|\\)(\s*.*)$".format(prefix=prefix),
                        re.LOCALE
                        ))

        if self.indextype == "AUTHOR":
            self.authwords = authors.compile_authwords(self.keywords)

    def _raw_add(self, key, number, link):
        """Add a song to the list.

        No processing is done on data. It is added raw. See add() for a
        similar method with processing.
        """
        first = self.get_first_letter(key[0])
        if not first in self.data:
            self.data[first] = dict()
        if not key in self.data[first]:
            self.data[first][key] = {
                    'sortingkey': [
                        unidecode.unidecode(tex2plain(item)).lower()
                        for item in key
                        ],
                    'entries': [],
                    }
        self.data[first][key]['entries'].append({'num': number, 'link': link})

    def add(self, key, number, link):
        """Add a song to the list.

        Process data before adding it.
        """
        if self.indextype == "TITLE":
            # Removing prefixes before titles
            for pattern in self.prefix_patterns:
                match = pattern.match(key)
                if match:
                    self._raw_add(
                            (
                                (match.group(2) + match.group(3)).strip(),
                                match.group(1).strip(),
                            ),
                            number,
                            link
                            )
                    return
            self._raw_add((key, ""), number, link)

        if self.indextype == "AUTHOR":
            # Processing authors
            for author in authors.processauthors(
                    key,
                    **self.authwords):
                self._raw_add(author, number, link)

    @staticmethod
    def ref_to_str(ref):
        """Return the LaTeX code corresponding to the reference."""
        return r'\hyperlink{{{0[link]}}}{{{0[num]}}}'.format(ref)

    def key_to_str(self, key):
        """Convert the key (title or author) to the LaTeX command rendering it.

        """
        if self.indextype == "AUTHOR":
            return r"\indexauthor{{{first}}}{{{last}}}".format(
                first=key[1],
                last=key[0],
                )

        if self.indextype == "TITLE":
            return r"\indextitle{{{0[1]}}}{{{0[0]}}}".format(key)

    def entry_to_str(self, key, entry):
        """Return the LaTeX code corresponding to the entry."""
        return r"""\idxentry{{
                {0}
            }}{{
                {1}
            }}""".format(
                self.key_to_str(key),
                r'\\'.join([self.ref_to_str(ref) for ref in entry]),
                )

    def idxblock_to_str(self, letter, entries):
        """Return the LaTeX code corresponding to an index block.

        Here, an index block is a letter, and all data beginning with this
        letter.
        """
        def sortkey(key):
            """Return something sortable for `entries[key]`."""
            return [
                    locale.strxfrm(item)
                    for item
                    in entries[key]['sortingkey']
                    ]
        string = r'\begin{idxblock}{' + letter + '}' + EOL
        for key in sorted(entries, key=sortkey):
            string += "  " + self.entry_to_str(key, entries[key]['entries'])
        string += EOL + r'\end{idxblock}'
        return string

    def entries_to_str(self):
        """Return the LaTeX code corresponding to the index."""
        string = ""
        for letter in sorted(self.data.keys()):
            string += self.idxblock_to_str(letter, self.data[letter]) + EOL
        return string