Добавлена возможность обновления (создания) конкретной версии

README.md

@@ -1,11 +1,28 @@
 # py-fias
 WSGI application they can serve FIAS (Russian Address Object DB)
+
 Простое приложение для работы с БД ФИАС, написано для Python 2.7
 
 ## Установка
 Протестирована работа на следующих ОС: Windows (8.1) и Debian Jessie
 
-Предполагается, что у Вас уже установлена БД PostgreSql версии 9.5, интерпретатор Python 2.7
+### Зависимости
 
-1. Windows
-    1. Установим *уйню...
+Для работы приложения необходимо достаточное кол-во RAM (1Gb+) и 4.5Gb места на диске 
+(3-3.5Gb для скачивания архива с базой и  300-400Mb для индексов Sphinx). Также необходимы root права 
+(или Администратора, для OS Windows), для работы демона Sphinx и предварительной установки. 
+
+Предварительно нужно установить и настроить:
+
+1. Python 2.7 [Windows](https://www.python.org/downloads/windows/), [Debian](https://www.python.org/downloads/source/) 
+(`sudo apt-get install python2.7 python2.7-dev`), pip
+
+2. PostgreSql 9.5 и выше (из-за _ON CONFLICT_)
+
+3. Sphinx 2.2.3 и новее (из-за синтаксиса _MAYBE_)
+
+### Windows
+1. Установить sphinxapi последней версии: 
+
+`python -m pip install https://github.com/Romamo/sphinxapi/zipball/master`
+    

aore/aoutils/importer.py

@@ -1,39 +0,0 @@
-# -*- coding: utf-8 -*-
-
-from pysimplesoap.client import SoapClient
-
-
-class Importer:
-    def __init__(self):
-        self.client = SoapClient(
-            location="http://fias.nalog.ru/WebServices/Public/DownloadService.asmx",
-            action='http://fias.nalog.ru/WebServices/Public/DownloadService.asmx/',
-            namespace="http://fias.nalog.ru/WebServices/Public/DownloadService.asmx",
-            soap_ns='soap', trace=False, ns=False)
-
-    def get_current_fias_version(self):
-        return 224  # TODO FIXIT
-
-    def get_full(self):
-        response = self.client.GetLastDownloadFileInfo()
-
-        assert response, "Response is null"
-        downloadfileinfo = response.GetLastDownloadFileInfoResponse.GetLastDownloadFileInfoResult
-
-        assert downloadfileinfo.VersionId < self.get_current_fias_version(), "DB is already up-to-date"
-
-        yield dict(intver=int(downloadfileinfo.VersionId), strver=str(downloadfileinfo.TextVersion),
-                   url=str(downloadfileinfo.FiasCompleteXmlUrl))
-
-    # return (intver, strver, url)
-    def get_updates(self):
-        response = self.client.GetAllDownloadFileInfo()
-
-        assert response, "Response is null"
-
-        current_fias_version = self.get_current_fias_version()
-
-        for DownloadFileInfo in response.GetAllDownloadFileInfoResponse.GetAllDownloadFileInfoResult.DownloadFileInfo:
-            if int(DownloadFileInfo.VersionId) > current_fias_version:
-                yield dict(intver=int(DownloadFileInfo.VersionId), strver=str(DownloadFileInfo.TextVersion),
-                           url=str(DownloadFileInfo.FiasDeltaXmlUrl))

aore/fias/fiasfactory.py

@@ -0,0 +1,17 @@
+# -*- coding: utf-8 -*-
+from aore.fias.search import SphinxSearch
+
+
+class FiasFactory:
+    def __init__(self):
+        self.searcher = SphinxSearch()
+
+    # text - строка поиска
+    # strong - строгий поиск или "мягкий" (с допущением ошибок, опечаток)
+    # out_format - "full" or "simple" - полный (подробно для каждого подпункта) или простой (только строка и AOID)
+    def find(self, text, strong=False, out_format="simple"):
+        try:
+            results = self.searcher.find(text, strong)
+
+        except:
+            return []

aore/fias/search.py

@@ -4,7 +4,7 @@ import re
 
 import Levenshtein
 import psycopg2
-import aore.sphinxapi as sphinxapi
+import sphinxapi
 
 from aore.config import db as dbparams, sphinx_index_sugg, sphinx_index_addjobj
 from aore.dbutils.dbimpl import DBImpl
@@ -80,12 +80,12 @@ class SphinxSearch:
         phrase = unicode(phrase).replace('-', '').replace('@', '').lower()
         return re.split(r"[ ,:.#$]+", phrase)
 
-    def __add_word_variations(self, word_entry):
-        if word_entry.MT_MANY_SUGG:
+    def __add_word_variations(self, word_entry, strong):
+        if word_entry.MT_MANY_SUGG and not strong:
             suggs = self.__get_suggest(word_entry.word, self.rating_limit_soft, 6)
             for suggestion in suggs:
                 word_entry.add_variation(suggestion[0])
-        if word_entry.MT_SOME_SUGG:
+        if word_entry.MT_SOME_SUGG and not strong:
             suggs = self.__get_suggest(word_entry.word, self.rating_limit_hard, 3)
             for suggestion in suggs:
                 word_entry.add_variation(suggestion[0])
@@ -96,16 +96,18 @@ class SphinxSearch:
         if word_entry.MT_ADD_SOCR:
             word_entry.add_variation_socr()
 
-    def __get_word_entries(self, words):
+    def __get_word_entries(self, words, strong):
         for word in words:
             if word != '':
                 we = WordEntry(self.db, word)
-                self.__add_word_variations(we)
+                self.__add_word_variations(we, strong)
+                if we.get_variations() == "()":
+                    raise BaseException("Cannot process sentence.")
                 yield we
 
-    def find(self, text):
+    def find(self, text, strong):
         words = self.__split_phrase(text)
-        word_entries = self.__get_word_entries(words)
+        word_entries = self.__get_word_entries(words, strong)
         sentence = "{}".format(" MAYBE ".join(x.get_variations() for x in word_entries))
 
         self.__configure(sphinx_index_addjobj)
@@ -114,4 +116,4 @@ class SphinxSearch:
         results = []
         for ma in rs['matches']:
             results.append([ma['attrs']['aoid'], ma['attrs']['fullname'], ma['weight']])
-        print results
+        return results

aore/miscutils/sphinx.py

@@ -5,9 +5,9 @@ import os
 
 from bottle import template
 
-from aore.aoutils.aoxmltableentry import AoXmlTableEntry
+from aore.updater.aoxmltableentry import AoXmlTableEntry
+from aore.updater.dbhandler import DbHandler
 from aore.config import db as dbconfig, sphinx_index_addjobj, sphinx_var_dir, trashfolder, sphinx_index_sugg
-from aore.dbutils.dbhandler import DbHandler
 from trigram import trigram
 
 

aore/updater/aodataparser.py

@@ -1,7 +1,7 @@
 # -*- coding: utf-8 -*-
 import os
 
-from aore.aoutils.aoxmltableentry import AoXmlTableEntry
+from aore.updater.aoxmltableentry import AoXmlTableEntry
 from aore.config import trashfolder
 from aore.dbutils.dbschemas import db_shemas
 from xmlparser import XMLParser
@@ -18,18 +18,12 @@ class AoDataParser:
         self.pagesize = pagesize
         self.currentpage = 0
         self.counter = 0
-        self.addrobj_filter = self.datasource.table_name == 'ADDROBJ' and self.datasource.operation_type == AoXmlTableEntry.OperationType.create
 
         self.base_filename = ""
         self.csv_file = None
         self.data_bereit_callback = None
 
     def import_update(self, attr):
-        # Addrobj anvanced filter
-        if self.addrobj_filter:
-            if attr['ACTSTATUS'] == '0' or 'NEXTID' in attr:
-                return
-
         if self.counter > self.pagesize:
             # Send old file to DB engine
             if self.csv_file:

aore/updater/dbhandler.py

@@ -5,7 +5,7 @@ import logging
 import psycopg2
 from bottle import template
 
-from aore.aoutils.aoxmltableentry import AoXmlTableEntry
+from aore.updater.aoxmltableentry import AoXmlTableEntry
 from aore.config import db as dbparams
 from aore.dbutils.dbimpl import DBImpl
 from aore.dbutils.dbschemas import db_shemas

aore/updater/soapreceiver.py

@@ -0,0 +1,26 @@
+# -*- coding: utf-8 -*-
+
+from pysimplesoap.client import SoapClient
+
+
+class SoapReceiver:
+    def __init__(self):
+        self.client = SoapClient(
+            location="http://fias.nalog.ru/WebServices/Public/DownloadService.asmx",
+            action='http://fias.nalog.ru/WebServices/Public/DownloadService.asmx/',
+            namespace="http://fias.nalog.ru/WebServices/Public/DownloadService.asmx",
+            soap_ns='soap', trace=False, ns=False)
+
+    def get_current_fias_version(self):
+        return 224  # TODO FIXIT
+
+    # return (intver, strver, url)
+    def get_update_list(self):
+        response = self.client.GetAllDownloadFileInfo()
+
+        assert response, "Response is null"
+
+        for DownloadFileInfo in response.GetAllDownloadFileInfoResponse.GetAllDownloadFileInfoResult.DownloadFileInfo:
+            yield dict(intver=int(DownloadFileInfo.VersionId), strver=str(DownloadFileInfo.TextVersion),
+                       delta_url=str(DownloadFileInfo.FiasDeltaXmlUrl),
+                       complete_url=str(DownloadFileInfo.FiasCompleteXmlUrl))

aore/updater/updater.py

@@ -3,15 +3,15 @@
 import logging
 from os import walk, path
 
-from aore.aoutils.aodataparser import AoDataParser
-from aore.aoutils.aorar import AoRar
-from aore.aoutils.aoxmltableentry import AoXmlTableEntry
-from aore.aoutils.importer import Importer
-from aore.dbutils.dbhandler import DbHandler
+from aore.updater.aodataparser import AoDataParser
+from aore.updater.aorar import AoRar
+from aore.updater.aoxmltableentry import AoXmlTableEntry
+from aore.updater.dbhandler import DbHandler
+from aore.updater.soapreceiver import SoapReceiver
 from aore.dbutils.dbschemas import allowed_tables
 
 
-class AoUpdater:
+class Updater:
     # Source: "http", directory (as a full path to unpacked xmls)
     def __init__(self, source="http"):
         self.db_handler = DbHandler()
@@ -31,7 +31,7 @@ class AoUpdater:
 
     def __get_updates_from_folder(self, foldername):
         # TODO: Вычислять версию, если берем данные из каталога
-        yield dict(intver=0, textver="Unknown", url=foldername)
+        yield dict(intver=0, textver="Unknown", delta_url=foldername, complete_url=foldername)
 
     def __get_updates_from_rar(self, url):
         aorar = AoRar()
@@ -39,14 +39,11 @@ class AoUpdater:
         for table_entry in aorar.get_table_entries(fname, allowed_tables):
             yield table_entry
 
-    def __init_update_entries(self, full_base):
+    def __init_update_entries(self, updates_generator):
         if self.mode == "http":
+            assert updates_generator
             self.tablelist_generator = self.__get_updates_from_rar
-            imp = Importer()
-            if full_base:
-                self.updalist_generator = imp.get_full()
-            else:
-                self.updalist_generator = imp.get_updates()
+            self.updalist_generator = updates_generator
         else:
             assert path.isdir(self.mode), "Invalid directory {}".format(self.mode)
             self.updalist_generator = self.__get_updates_from_folder(self.mode)
@@ -56,12 +53,13 @@ class AoUpdater:
         aoparser = AoDataParser(table_xmlentry, chunck_size)
         aoparser.parse(lambda x, y: self.db_handler.bulk_csv(operation_type, table_xmlentry.table_name, x, y))
 
-    def create(self):
-        self.__init_update_entries(True)
+    def create(self, updates_generator):
+        self.__init_update_entries(updates_generator)
         self.db_handler.pre_create()
 
         for update_entry in self.updalist_generator:
-            for table_entry in self.tablelist_generator(update_entry['url']):
+            logging.info("Processing update #{}".format(update_entry['intver']))
+            for table_entry in self.tablelist_generator(update_entry['complete_url']):
                 if table_entry.operation_type == AoXmlTableEntry.OperationType.update:
                     table_entry.operation_type = AoXmlTableEntry.OperationType.create
                 self.process_single_entry(table_entry.operation_type, table_entry)
@@ -70,18 +68,13 @@ class AoUpdater:
 
         logging.info("Create success")
 
-    def update(self, count=1):
-        self.__init_update_entries(False)
+    def update(self, updates_generator):
+        self.__init_update_entries(updates_generator)
         self.db_handler.pre_update()
 
-        counter = 0
-        for update_entry in self.updalist_generator:
-            counter += 1
-            if counter > count:
-                logging.warning("Maximum count of updates ({}) are processed - exit".format(count))
-                break
-
-            for table_entry in self.tablelist_generator(update_entry['url']):
+        for update_entry in self.updates_generator:
+            logging.info("Processing update #{}".format(update_entry['intver']))
+            for table_entry in self.tablelist_generator(update_entry['delta_url']):
                 self.process_single_entry(table_entry.operation_type, table_entry)
 
         logging.info("Update success")

manage.py

@@ -2,30 +2,82 @@
 
 import optparse
 
-from aore.aoutils.aoupdater import AoUpdater
-from aore.miscutils.sphinx import SphinxHelper
 from aore.fias.search import SphinxSearch
+from aore.miscutils.sphinx import SphinxHelper
+from aore.updater.updater import Updater
+from aore.updater.soapreceiver import SoapReceiver
 
 
-def update_base(xml_source, updates_count):
-    aoupdater = AoUpdater(xml_source)
-    aoupdater.update(updates_count)
+def print_fias_versions():
+    imp = SoapReceiver()
+    current_version = imp.get_current_fias_version()
+    all_versions = imp.get_update_list()
+
+    print("Installed version: {}".format(current_version))
+    print("Avaliable updates:")
+    print("Number\t\tDate")
+    for upd in all_versions:
+        mark_current = (' ', '*')[int(upd['intver']) == current_version]
+        print "{}{}\t\t{}".format(mark_current, upd['intver'], upd['strver'])
 
 
-def create_base(xml_source):
-    aoupdater = AoUpdater(xml_source)
-    aoupdater.create()
+def parse_update_str(updates_str):
+    if updates_str == "all":
+        return None
+
+    upd_list = updates_str.lower().replace(' ','').split(',')
+    out_list = []
+
+    for u_entry in upd_list:
+        if '-' in u_entry:
+            u_range = u_entry.split('-')
+            out_list += range(int(u_range[0]), int(u_range[1]))
+        else:
+            out_list.append(int(u_entry))
+
+    return out_list
+
+
+def get_allowed_updates(updates_str, mode = "create"):
+    imp = SoapReceiver()
+    current_version = imp.get_current_fias_version()
+    all_versions = [x for x in imp.get_update_list()]
+
+    user_defined_list = parse_update_str(updates_str)
+    out_list = []
+
+    if mode == "create" and not user_defined_list:
+        yield all_versions[-1]
+
+    assert (mode == "create" and len(user_defined_list) == 1)
+
+    for uv in all_versions:
+        uv_ver = uv['intver']
+        if uv_ver > current_version and (not user_defined_list or uv_ver in user_defined_list):
+            out_list.append(uv)
+
+    out_list.sort(key=lambda x: x['intver'])
+    for ol_entry in out_list:
+        yield ol_entry
 
 
 def main():
     # Parse options
     p = optparse.OptionParser()
     p.add_option('--database', '-b', action="store", type="string",
-                 help="Manage database. Value: create - create new DB, update - update existing DB without loose the data")
-    p.add_option('--update-count', '-u', default=1, type="int",
-                 help="Count of updates to process, only for '--database update' option")
+                 help="Manage database. Values: "
+                      "create - create new DB, "
+                      "update - update existing DB without loose the data")
+    p.add_option('--update-version', '-u', default="all", type="string",
+                 help="Valid for updating via HTTP. "
+                      "Versions of updates to process. Can be 111 or 111-222 or 111,222,333."
+                      "For '--database-create' only one value is necessary. If not specified, "
+                      "all updates will be processed (for '--database update') or last DB snapshot "
+                      "(for '--database create')")
+    p.add_option('--show-versions', '-v', action="store_true", dest="show_versions", default=False,
+                 help="Show allowed fias versions")
     p.add_option('--source', '-s', default="http",
-                 help="Create/update DB from source. Value: \"http\" or absolute path to folder")
+                 help="Create/update DB from source. Value: 'http' or absolute path to folder containing XMLs")
     p.add_option('--sphinx-configure', '-c', action="store_true", dest="sphinx", default="False",
                  help="Configure sphinx. Creates sphinx.conf specified in '--output-conf'")
     p.add_option('--indexer-path', '-i',
@@ -37,14 +89,25 @@ def main():
 
     options, arguments = p.parse_args()
 
+    # Show FIAS updates
+    if options.show_versions:
+        print_fias_versions()
+        return
+
     # Manage DB
     if options.database:
         # create new database
+        aoupdater = Updater(options.source)
+        allowed_updates = None
+        if options.source == "http":
+            allowed_updates = get_allowed_updates(options.update_version)
+
         if options.database == "create":
-            create_base(options.source)
+            aoupdater.create(allowed_updates)
+
         # update database
         if options.database == "update":
-            update_base(options.source, int(options.update_count))
+            aoupdater.update(allowed_updates)
 
     # Manage Sphinx
     if options.sphinx and options.indexer_path and options.output_conf:
@@ -54,7 +117,8 @@ def main():
     # 4 Debug purposes..
     if options.test:
         sph = SphinxSearch()
-        sph.find('кридовая паскаул')
+        sph.find('кедровая пасраул')
+
 
 if __name__ == '__main__':
     main()