[python] Added staistics.

8 files changed, 255 insertions, 15 deletions

diff --git a/tools/python/maps_generator/__main__.py b/tools/python/maps_generator/__main__.py
index ead17107d2..97a55b9371 100644
--- a/tools/python/maps_generator/__main__.py
+++ b/tools/python/maps_generator/__main__.py
@@ -8,7 +8,8 @@ from .generator.exceptions import ContinueError, SkipError, ValidationError
 from .maps_generator import (generate_maps, generate_coasts, reset_to_stage,
                              ALL_STAGES, stage_download_production_external,
                              stage_descriptions, stage_ugc, stage_popularity,
-                             stage_localads, stages_as_string)
+                             stage_localads, stage_statistics,
+                             stages_as_string)
 from .utils.collections import unique
 
 logger = logging.getLogger("maps_generator")
@@ -25,16 +26,16 @@ def parse_options():
         nargs="?",
         type=str,
         help="Continue the last build or specified in CONTINUE from the "
-        "last stopped stage.")
+             "last stopped stage.")
     parser.add_argument(
         "--countries",
         type=str,
         default="",
         help="List of regions, separated by a comma or a semicolon, or path to "
-        "file with regions, separated by a line break, for which maps"
-        " will be built. The names of the regions can be seen "
-        "in omim/data/borders. It is necessary to set names without "
-        "any extension.")
+             "file with regions, separated by a line break, for which maps"
+             " will be built. The names of the regions can be seen "
+             "in omim/data/borders. It is necessary to set names without "
+             "any extension.")
     parser.add_argument(
         "--skip",
         type=str,
@@ -58,7 +59,7 @@ def parse_options():
         default=False,
         action="store_true",
         help="Build production maps. In another case, 'osm only maps' are built"
-        " - maps without additional data and advertising.")
+             " - maps without additional data and advertising.")
     return vars(parser.parse_args())
 
 
@@ -131,10 +132,13 @@ def main():
         ]
     options["skip"] = options_skip
     if not options["production"]:
-        options["skip"] += stages_as_string(stage_download_production_external,
-                                            stage_ugc, stage_popularity,
-                                            stage_descriptions,
-                                            stage_localads)
+        options["skip"] += stages_as_string(
+            stage_download_production_external,
+            stage_ugc, stage_popularity,
+            stage_descriptions,
+            stage_localads,
+            stage_statistics
+        )
     if not all(s in ALL_STAGES for s in options["skip"]):
         raise SkipError(f"Stages {set(options['skip']) - set(ALL_STAGES)} "
                         f"not found.")
diff --git a/tools/python/maps_generator/generator/env.py b/tools/python/maps_generator/generator/env.py
index 7d0211993b..ad8dca2851 100644
--- a/tools/python/maps_generator/generator/env.py
+++ b/tools/python/maps_generator/generator/env.py
@@ -153,6 +153,12 @@ class Env:
         return path
 
     @property
+    def stats_path(self):
+        path = os.path.join(self.out_path, "stats")
+        self._create_if_not_exist(path)
+        return path
+
+    @property
     def types_path(self):
         return os.path.join(self.user_resource_path, "types.txt")
 
diff --git a/tools/python/maps_generator/generator/gen_tool.py b/tools/python/maps_generator/generator/gen_tool.py
index ca8a05d4e5..1899a12291 100644
--- a/tools/python/maps_generator/generator/gen_tool.py
+++ b/tools/python/maps_generator/generator/gen_tool.py
@@ -37,6 +37,7 @@ class GenTool:
         "no_ads": bool,
         "preprocess": bool,
         "split_by_polygons": bool,
+        "type_statistics": bool,
         "planet_version": int,
         "booking_data": str,
         "brands_data": str,
diff --git a/tools/python/maps_generator/generator/settings.py b/tools/python/maps_generator/generator/settings.py
index b0adebe3cd..8df50f2a38 100644
--- a/tools/python/maps_generator/generator/settings.py
+++ b/tools/python/maps_generator/generator/settings.py
@@ -50,6 +50,8 @@ SUBWAY_URL = ""
 FOOD_URL = ""
 FOOD_TRANSLATIONS_URL = ""
 
+STATS_TYPES_CONFIG = ""
+
 PLANET = "planet"
 
 GEN_TOOL = "generator_tool"
@@ -120,6 +122,9 @@ FOOD_URL = _get_opt(config, "External", "FOOD_URL", FOOD_URL)
 FOOD_TRANSLATIONS_URL = _get_opt(config, "External", "FOOD_TRANSLATIONS_URL", 
                                  FOOD_TRANSLATIONS_URL)
 
+STATS_TYPES_CONFIG = _get_opt_path(config, "Stats", "STATS_TYPES_CONFIG",
+                                   STATS_TYPES_CONFIG)
+
 PLANET_O5M = os.path.join(MAIN_OUT_PATH, PLANET + ".o5m")
 PLANET_PBF = os.path.join(MAIN_OUT_PATH, PLANET + ".osm.pbf")
 PLANET_COASTS_GEOM_URL = os.path.join(PLANET_COASTS_URL, "latest_coasts.geom")
diff --git a/tools/python/maps_generator/generator/statistics.py b/tools/python/maps_generator/generator/statistics.py
new file mode 100644
index 0000000000..6051e97c51
--- /dev/null
+++ b/tools/python/maps_generator/generator/statistics.py
@@ -0,0 +1,116 @@
+import re
+import os
+import datetime
+from collections import defaultdict
+
+
+RE_STAT = re.compile(r"(?:\d+\. )?([\w:|-]+?)\|: "
+                     r"size = \d+; "
+                     r"count = (\d+); "
+                     r"length = ([0-9.e+-]+) m; "
+                     r"area = ([0-9.e+-]+) m²; "
+                     r"names = (\d+)\s*")
+
+RE_TIME_DELTA = re.compile(r'^(?:(?P<days>-?\d+) (days?, )?)?'
+                           r'((?:(?P<hours>-?\d+):)(?=\d+:\d+))?'
+                           r'(?:(?P<minutes>-?\d+):)?'
+                           r'(?P<seconds>-?\d+)'
+                           r'(?:\.(?P<microseconds>\d{1,6})\d{0,6})?$')
+
+RE_FINISH_STAGE = re.compile(r"(.*)Stage (\w+): finished in (.+)$")
+
+
+def read_stat(f):
+    stats = []
+    for line in f:
+        m = RE_STAT.match(line)
+        stats.append({
+            "name": m.group(1).replace("|", "-"),
+            "cnt": int(m.group(2)),
+            "len": float(m.group(3)),
+            "area": float(m.group(4)),
+            "names": int(m.group(5))
+        })
+    return stats
+
+
+def read_config(f):
+    config = []
+    for line in f:
+        columns = [c.strip() for c in line.split(";", 2)]
+        columns[0] = re.compile(columns[0])
+        columns[1] = columns[1].lower()
+        config.append(columns)
+    return config
+
+
+def process_stat(config, stats):
+    result = {}
+    for param in config:
+        res = 0
+        for typ in stats:
+            if param[0].match(typ["name"]):
+                if param[1] == "len":
+                    res += typ["len"]
+                elif param[1] == "area":
+                    res += typ["area"]
+                elif param[1] == "cnt_names":
+                    res += typ["names"]
+                else:
+                    res += typ["cnt"]
+        result[str(param[0]) + param[1]] = res
+    return result
+
+
+def format_res(res, typ):
+    if typ == "len":
+        unit = "м"
+    elif typ == "area":
+        unit = "м²"
+    else:
+        unit = "шт."
+    return res, unit
+
+
+def make_stats(config_path, stats_path):
+    with open(config_path) as f:
+        config = read_config(f)
+    with open(stats_path) as f:
+        stats = process_stat(config, read_stat(f))
+    lines = []
+    for param in config:
+        k = str(param[0]) + param[1]
+        st = format_res(stats[k], param[1])
+        lines.append({"type": param[2], "quantity": st[0], "unit": st[1]})
+    return lines
+
+
+def parse_time(time_str):
+    parts = RE_TIME_DELTA.match(time_str)
+    if not parts:
+        return
+    parts = parts.groupdict()
+    time_params = {}
+    for name, param in parts.items():
+        if param:
+            time_params[name] = int(param)
+    return datetime.timedelta(**time_params)
+
+
+def get_stages_info(log_path):
+    result = defaultdict(lambda: defaultdict(dict))
+    for file in os.listdir(log_path):
+        path = os.path.join(log_path, file)
+        with open(path) as f:
+            for line in f:
+                m = RE_FINISH_STAGE.match(line)
+                if not m:
+                    continue
+                stage_name = m.group(2)
+                dt = parse_time(m.group(3))
+                if file.startswith("stage_"):
+                    result["stages"][stage_name] = dt
+                else:
+                    country = file.split(".")[0]
+                    result["countries"][country][stage_name] = dt
+    return result
diff --git a/tools/python/maps_generator/maps_generator.py b/tools/python/maps_generator/maps_generator.py
index 4018a56863..78e15d7868 100644
--- a/tools/python/maps_generator/maps_generator.py
+++ b/tools/python/maps_generator/maps_generator.py
@@ -3,7 +3,10 @@ import os
 import shutil
 from functools import partial
 from multiprocessing.pool import ThreadPool
+from collections import defaultdict
 import multiprocessing
+import json
+import datetime
 
 from descriptions.descriptions_downloader import (check_and_get_checker,
                                                   download_from_wikipedia_tags,
@@ -22,6 +25,7 @@ from .generator.env import (planet_lock_file, build_lock_file,
 from .generator.exceptions import (ContinueError, BadExitStatusError,
                                    wait_and_raise_if_fail)
 from .generator.gen_tool import run_gen_tool
+from .generator.statistics import make_stats, get_stages_info
 from .utils.file import is_verified, download_file, make_tarfile
 
 logger = logging.getLogger("maps_generator")
@@ -238,8 +242,9 @@ def stage_external_resources(env):
     for ttf_file in resources:
         shutil.copy2(ttf_file, env.intermediate_path)
 
-    shutil.copy2(os.path.join(env.user_resource_path, "WorldCoasts_obsolete.mwm"),
-                 env.mwm_path)
+    shutil.copy2(
+        os.path.join(env.user_resource_path, "WorldCoasts_obsolete.mwm"),
+        env.mwm_path)
 
     for file in os.listdir(env.mwm_path):
         if file.startswith(WORLD_NAME) and file.endswith(".mwm"):
@@ -260,6 +265,46 @@ def stage_localads(env):
 
 
 @stage
+def stage_statistics(env):
+    result = defaultdict(lambda: defaultdict(dict))
+
+    @country_stage_log
+    def stage_mwm_statistics(env, country, **kwargs):
+        stats_tmp = os.path.join(env.draft_path, f"{country}.stat")
+        with open(stats_tmp, "w") as f:
+            maps_stages.run_gen_tool_with_recovery_country(
+                env,
+                env.gen_tool,
+                out=f,
+                err=env.get_subprocess_out(country),
+                data_path=env.mwm_path,
+                user_resource_path=env.user_resource_path,
+                type_statistics=True,
+                output=country,
+                **kwargs
+            )
+        result["countries"][country]["types"] = \
+            make_stats(settings.STATS_TYPES_CONFIG, stats_tmp)
+
+    mwms = env.get_mwm_names()
+    countries = filter(lambda x: x not in WORLDS_NAMES, mwms)
+    with ThreadPool() as pool:
+        pool.map(partial(stage_mwm_statistics, env), countries)
+    stages_info = get_stages_info(env.log_path)
+    result["stages"] = stages_info["stages"]
+    for c in stages_info["countries"]:
+        result["countries"][c]["stages"] = stages_info["countries"][c]
+
+    def default(o):
+        if isinstance(o, datetime.timedelta):
+            return str(o)
+
+    with open(os.path.join(env.stats_path, "stats.json"), "w") as f:
+        json.dump(result, f, ensure_ascii=False, sort_keys=True,
+                  indent=2, default=default)
+
+
+@stage
 def stage_cleanup(env):
     osm2ft_path = os.path.join(env.out_path, "osm2ft")
     os.makedirs(osm2ft_path, exist_ok=True)
@@ -283,7 +328,7 @@ STAGES = [s.__name__ for s in
            stage_download_and_convert_planet, stage_update_planet,
            stage_coastline, stage_preprocess, stage_features, stage_mwm,
            stage_descriptions, stage_countries_txt, stage_external_resources,
-           stage_localads, stage_cleanup)]
+           stage_localads, stage_statistics, stage_cleanup)]
 
 ALL_STAGES = STAGES + COUNTRIES_STAGES
 
@@ -343,6 +388,7 @@ def generate_maps(env):
             stage_countries_txt(env)
             stage_external_resources(env)
             stage_localads(env)
+            stage_statistics(env)
             stage_cleanup(env)
 
 
diff --git a/tools/python/maps_generator/var/etc/map_generator.ini.default b/tools/python/maps_generator/var/etc/map_generator.ini.default
index 2fc56d6361..593781d5a0 100644
--- a/tools/python/maps_generator/var/etc/map_generator.ini.default
+++ b/tools/python/maps_generator/var/etc/map_generator.ini.default
@@ -32,5 +32,8 @@ OSM_TOOLS_PATH: ~/osmctools
 # POPULARITY_URL:
 # SUBWAY_URL:
 # FOOD_URL:
-# FOOD_TRANSLATIONS_URL: 
+# FOOD_TRANSLATIONS_URL:
+
+[Stats]
+STATS_TYPES_CONFIG: ${Developer:OMIM_PATH}/tools/python/maps_generator/var/etc/stats_types_config.txt
 
diff --git a/tools/python/maps_generator/var/etc/stats_types_config.txt b/tools/python/maps_generator/var/etc/stats_types_config.txt
new file mode 100644
index 0000000000..3026e9f86e
--- /dev/null
+++ b/tools/python/maps_generator/var/etc/stats_types_config.txt
@@ -0,0 +1,59 @@
+barrier-(fence|gate);len;Заборы
+building;cnt;Здания
+(amenity|shop|historic)-.*;cnt;POI
+(amenity|shop|historic)-.*;cnt_names;POI c именами
+amenity-(cafe|restaurant|fast_food).*;cnt;Кафе и рестораны
+amenity-(pub|bar);cnt;Бары и пабы
+amenity-kindergarten;cnt;Детские сады
+amenity-(school|university|college);cnt;Школы и университеты
+amenity-parking.*;cnt;Автостоянки
+amenity-parking.*;area;Автостоянки
+amenity-pharmacy;cnt;Аптеки
+amenity-place_of_worship.*;cnt;Храмы
+amenity-(hospital|doctors);cnt;Больницы и поликлиники
+amenity-toilets;cnt;Туалеты
+amenity-(waste_disposal|recycling);cnt;Мусорные баки
+highway-(motorway|trunk|primary|secondary|tertiary|residential|unclassified|service|track|living_street)(_link)?(-.*)?;len;Автодорожная сеть
+highway-(footway|path|pedestrian|steps).*;len;Пешеходные дорожки
+highway-.*-bridge;len;Мосты
+highway-.*-tunnel;len;Туннели
+highway-(footway|path|steps)-bridge;len;Пешеходные мосты
+highway-(footway|path|steps)-tunnel;len;Пешеходные туннели
+highway-steps.*;len;Лестницы
+highway-speed_camera;cnt;Камеры контроля скорости
+internet_access-wlan;cnt;Точки доступа Wi-Fi
+leisure-(pitch|stadium|playing_fields|track|sports_centre).*;cnt;Спортплощадки и комплексы
+leisure-playground;cnt;Детские площадки
+man_made-lighthouse;cnt;Маяки
+man_made-windmill;cnt;Ветряные мельницы
+man_made-pipeline.*;len;Трубопроводы
+natural-beach;cnt;Пляжи
+natural-tree;cnt;Отдельностоящие деревья
+natural-waterfall;cnt;Водопады
+piste:type.*;len;Лыжни
+place-(city.*|town|village|hamlet);cnt;Населённые пункты
+place-island;cnt;Острова
+power-(minor_)?line.*;len;Линии электропередачи
+power-(pole|tower);cnt;Опоры ЛЭП
+railway-(rail|monorail|light_rail|narrow_gauge|preserved|siding|spur|yard|disused|incline).*;len;Железные дороги
+railway-.*-(bridge|tunnel);len;Железнодорожные мосты и туннели
+railway-(razed|abandoned).*;len;Снятые ветки ж/д
+railway-narrow_gauge.*;len;Узкоколейные ж/д
+railway-tram(-.*)?;len;Трамвайные пути
+railway-(halt|station);cnt;Станции железной дороги
+railway-subway.*;len;Линии метро
+highway-bus_stop|railway-tram_stop;cnt;Остановки наземного транспорта
+shop-bakery;cnt;Пекарни
+shop-books;cnt;Книжные магазины
+shop-clothes;cnt;Магазины одежды
+shop-shoes;cnt;Магазины обуви
+shop-(convenience|supermarket);cnt;Продуктовые магазины
+shop-florist;cnt;Цветочные салоны
+shop-(hairdresser|beauty);cnt;Парикмахерские и салоны красоты
+tourism-(guest_house|hos?tel|motel);cnt;Гостиницы и хостелы
+tourism-(attraction|viewpoint);cnt;Достопримечательности и точки обзора
+waterway-(canal|river|stream)(-.*)?;len;Реки, каналы и ручьи
+landuse-cemetery.*;area;Кладбища
+leisure-park.*;area;Парки
+natural-beach;area;Пляжи
+sponsored-booking;cnt;Booking отели