Dynastie/search.py

# -*- coding: utf-8 -*-
import re
import unicodedata
import os
import operator
import pickle
from django.db import models
from dynastie.models import Post

class Search:
    MINIMUM_LETTERS = 3

    def __init__(self):
        self.report = ''

        self.tagreg = re.compile('<[^>]+>')
        self.htmlreg = re.compile('&[^;]+;')
        self.numreg = re.compile('[0-9]+')
        self.pat = re.compile(r'\s+')

        self.replace_by_space = ('(', ')', '#', '\'', '{', '}', '[', ']',
                                 '-', '|', '\t', '\\', '_', '^' '=', '+', '$',
                                 '£', '%', 'µ', '*', ',', '?', ';', '.', '/',
                                 ':', '!', '§', '€', '²')

    def _addReport(self, string, color=''):
        if color != '':
            self.report = self.report + '<span style="color:' + color + '">'
        self.report = self.report + '<b>' + self.__class__.__name__ + '</b> : '
        self.report = self.report + string
        if color != '':
            self.report = self.report + '</span>'
        self.report = self.report + '<br/>\n'

    def _addWarning(self, string):
        self.addReport(string, 'yellow')

    def _addError(self, string):
        self.addReport(string, 'red')


    def _saveDatabase(self, blog, hashtable):
        d = pickle.dumps(hashtable)

        f = open(blog.src_path + '/_search.db', 'w')
        f.write(d)
        f.close()

    def _loadDatabase(self, blog):
        filename = blog.src_path + '/_search.db'

        if not os.path.exists(filename):
            print 'No search index !'
            return None

        f = open(filename, 'rb')
        hashtable = pickle.load(f)
        f.close()

        return hashtable

    def _remove_reg(self, content, reg):
        found = re.search(reg, content)
        while found != None:
        #print str(found.start()) + ' ' + str(found.end())
            # print content[found.start(0):found.end(0)]
            # print "============================================="
            content = content[:found.start(0)].lstrip() + ' ' + content[found.end(0):]

            found = re.search(reg, content)

        return content

    def _strip_accents(self, s):
        return ''.join((c for c in unicodedata.normalize('NFD', s) if unicodedata.category(c) != 'Mn'))

    def _remove_tag(self, content):
        content = self._remove_reg(content, self.htmlreg)
        content = self._remove_reg(content, self.numreg)

        content = content.replace('\n', '')
        content = content.replace('\r', '')
        content = content.replace('"', '')

        for c in self.replace_by_space:
            content = content.replace(c, ' ')
        
        content = self._remove_reg(content, self.tagreg)

        content = self.pat.sub(' ', content)

        return content

    def _prepare_string(self, content):
        content = self._remove_tag(content)
        content = self._strip_accents(unicode(content, 'utf8'))

        return content

    def _indexContent(self, hashtable, index, content, word_weight):
        content = self._prepare_string(content)

        wordlist = content.split(' ')

        for word in wordlist:
            if len(word) < self.MINIMUM_LETTERS:
                continue
            word = word.lower()
            if not word in hashtable:
                hashtable[word] = []
            if not index in hashtable[word]:
                hashtable[word].insert(0, [index, word_weight])
            else:
                weight = hashtable[word][1]
                hashtable[word][1] = weight + word_weight

    def _index_file(self, hashtable, filename, index):
        f = open(filename, 'r')
        content = f.read()
        f.close()

        try:
            post = Post.objects.get(pk=index)
            if post.published == False: return
        except:
            return

        self._indexContent(hashtable, index, content, 1)
        self._indexContent(hashtable, index, post.title.encode('utf-8'), 5)
        
    def create_index(self, blog):
        hashtable = {}
        
        root = blog.src_path + '/_post' 

        if os.path.exists(root):
            for post in os.listdir(root):
                # Not a post number
                if not re.search(self.numreg, post): continue
                self._index_file(hashtable, root + '/' + post, int(post))

        self._saveDatabase(blog, hashtable)

        self._addReport('Search generated @ ' + blog.src_path + '/_search.db')

        return self.report

    def _index_post(self, blog, post, saveDatabase=True):
        hashtable = self._loadDatabase(blog)

        filename = blog.src_path + '/_post/' + post

        if hashtable is None:
            return self.create_index(blog)

        self._index_file(hashtable, filename, int(post))

        if saveDatabase:
            self._saveDatabase(blog, hashtable)

    def _remove_post(self, blog, post, saveDatabase=True):
        hashtable = self._loadDatabase(blog)

        if hashtable is None: return

        for k, v in hashtable.items():
            if post in v:
                v.remove(post)

        if saveDatabase:
            self._saveDatabase(blog, hashtable)

    def index_post(self, blog, post):
        return self._index_post(blog, post, True)

    def delete_post(self, blog, post):
        return self._remove_post(blog, post, True)

    def edit_post(self, blog, post, saveDatabase=True):
        self._remove_post(blog, post, False)
        self._index_post(blog, post, True)

    def search(self, blog, string):
        hashtable = self._loadDatabase(blog)
        
        string = self._prepare_string(string.encode('utf-8'))

        wordlist = string.split(' ')

        res = {}
        for word in wordlist:
            if len(word) < Search.MINIMUM_LETTERS:
                continue
            word = word.lower()
            reg = re.compile('.*' + word + '.*')
            for key in hashtable.keys():
                if reg.match(key):
                    for post in hashtable[key]:
                        if not post[0] in res:
                            res[post[0]] = post[1]
                        else:
                            res[post[0]] += post[1]

        sorted_res = sorted(res.iteritems(), key=operator.itemgetter(1))
        sorted_res.reverse()

        res = []
        for i in range(len(sorted_res)):
            res .append(sorted_res[i][0])
        return res