refactoring of picons downloading

This commit is contained in:
DYefremov
2021-02-10 23:21:30 +03:00
parent d87c878d4d
commit 7922f368b5
3 changed files with 177 additions and 135 deletions

View File

@@ -5,8 +5,11 @@ import shutil
from collections import namedtuple
from html.parser import HTMLParser
import requests
from app.commons import run_task, log
from app.settings import SettingsType
from .satellites import _HEADERS
_ENIGMA2_PICON_KEY = "{:X}:{:X}:{}"
_NEUTRINO_PICON_KEY = "{:x}{:04x}{:04x}.png"
@@ -17,6 +20,7 @@ Picon = namedtuple("Picon", ["ref", "ssid", "v_pid"])
class PiconsParser(HTMLParser):
""" Parser for package html page. (https://www.lyngsat.com/packages/*provider-name*.html) """
_BASE_URL = "https://www.lyngsat.com"
def __init__(self, entities=False, separator=' ', single=None):
@@ -58,14 +62,14 @@ class PiconsParser(HTMLParser):
row = self._current_row
ln = len(row)
if self._single and ln == 4 and row[0].startswith("../../logo/"):
self.picons.append(Picon(row[0].strip("../"), "0", "0"))
if self._single and ln == 4 and row[0].startswith("/logo/"):
self.picons.append(Picon(row[0].strip(), "0", "0"))
else:
if 9 < ln < 13:
url = None
if row[0].startswith("../logo/"):
if row[0].startswith("/logo/"):
url = row[0]
elif row[1].startswith("../logo/"):
elif row[1].startswith("/logo/"):
url = row[1]
ssid = row[-4]
@@ -78,37 +82,44 @@ class PiconsParser(HTMLParser):
pass
@staticmethod
def parse(open_path, picons_path, tmp_path, provider, picon_ids, s_type=SettingsType.ENIGMA_2):
if not os.path.isfile(open_path):
log("PiconsParser error [parse]. No such file or directory: {}".format(open_path))
def parse(provider, picons_path, picon_ids, s_type=SettingsType.ENIGMA_2):
""" Returns tuple(url, picon file name) list. """
req = requests.get(provider.url, timeout=5)
if req.status_code == 200:
logo_data = req.text
else:
log("Provider picons downloading error: {} {}".format(provider.url, req.reason))
return
with open(open_path, encoding="utf-8", errors="replace") as f:
on_id, pos, ssid, single = provider.on_id, provider.pos, provider.ssid, provider.single
neg_pos = pos.endswith("W")
pos = int("".join(c for c in pos if c.isdigit()))
# For negative (West) positions 3600 - numeric position value!!!
if neg_pos:
pos = 3600 - pos
parser = PiconsParser(single=single)
parser.reset()
parser.feed(f.read())
picons = parser.picons
if picons:
os.makedirs(picons_path, exist_ok=True)
for p in picons:
try:
if single:
on_id, freq = on_id.strip().split("::")
namespace = "{:X}{:X}".format(int(pos), int(freq))
else:
namespace = "{:X}0000".format(int(pos))
name = PiconsParser.format(ssid if single else p.ssid, on_id, namespace, picon_ids, s_type)
p_name = picons_path + (name if name else os.path.basename(p.ref))
shutil.copyfile(tmp_path + "www.lyngsat.com/" + p.ref.lstrip("."), p_name)
except (TypeError, ValueError) as e:
msg = "Picons format parse error: {}".format(p) + "\n" + str(e)
log(msg)
on_id, pos, ssid, single = provider.on_id, provider.pos, provider.ssid, provider.single
neg_pos = pos.endswith("W")
pos = int("".join(c for c in pos if c.isdigit()))
# For negative (West) positions 3600 - numeric position value!!!
if neg_pos:
pos = 3600 - pos
parser = PiconsParser(single=provider.single)
parser.reset()
parser.feed(logo_data)
picons = parser.picons
picons_data = []
if picons:
for p in picons:
try:
if single:
on_id, freq = on_id.strip().split("::")
namespace = "{:X}{:X}".format(int(pos), int(freq))
else:
namespace = "{:X}0000".format(int(pos))
name = PiconsParser.format(ssid if single else p.ssid, on_id, namespace, picon_ids, s_type)
p_name = picons_path + (name if name else os.path.basename(p.ref))
picons_data.append(("{}{}".format(PiconsParser._BASE_URL, p.ref), p_name))
except (TypeError, ValueError) as e:
msg = "Picons format parse error: {}".format(p) + "\n" + str(e)
log(msg)
return picons_data
@staticmethod
def format(ssid, on_id, namespace, picon_ids, s_type):
@@ -127,7 +138,8 @@ class ProviderParser(HTMLParser):
_POSITION_PATTERN = re.compile("at\s\d+\..*(?:E|W)']")
_ONID_TID_PATTERN = re.compile("^\d+-\d+.*")
_TRANSPONDER_FREQUENCY_PATTERN = re.compile("^\d+ [HVLR]+")
_DOMAINS = {"/tvchannels/", "/radiochannels/", "/packages/"}
_DOMAINS = {"/tvchannels/", "/radiochannels/", "/packages/", "/logo/"}
_BASE_URL = "https://www.lyngsat.com"
def __init__(self, entities=False, separator=' '):
@@ -155,7 +167,7 @@ class ProviderParser(HTMLParser):
if tag == 'tr':
self._is_th = True
if tag == "img":
if attrs[0][1].startswith("logo/"):
if attrs[0][1].startswith("/logo/"):
self._current_row.append(attrs[0][1])
if tag == "a":
url = attrs[0][1]
@@ -187,41 +199,47 @@ class ProviderParser(HTMLParser):
self._current_row.append(final_cell)
self._current_cell = []
elif tag == 'tr':
r = self._current_row
row = self._current_row
# Satellite position
if not self._positon:
pos = re.findall(self._POSITION_PATTERN, str(r))
pos = re.findall(self._POSITION_PATTERN, str(row))
if pos:
self._positon = "".join(c for c in str(pos) if c.isdigit() or c in ".EW")
len_row = len(r)
len_row = len(row)
if len_row > 2:
m = self._TRANSPONDER_FREQUENCY_PATTERN.match(r[1])
m = self._TRANSPONDER_FREQUENCY_PATTERN.match(row[1])
if m:
self._freq = m.group().split()[0]
if len_row == 12:
if len_row == 14:
# Providers
name = r[5]
name = row[6]
self._prv_names.add(name)
m = self._ONID_TID_PATTERN.match(str(r[-2]))
m = self._ONID_TID_PATTERN.match(str(row[9]))
if m:
on_id, tid = m.group().split("-")
if on_id not in self._ids:
r[-2] = on_id
row[-2] = on_id
self._ids.add(on_id)
r[0] = self._positon
row[0] = self._positon
if name + on_id not in self._prv_names:
self._prv_names.add(name + on_id)
self.rows.append(Provider(logo=r[2], name=name, pos=self._positon, url=r[6], on_id=on_id,
logo_data = None
req = requests.get(self._BASE_URL + row[3], timeout=5)
if req.status_code == 200:
logo_data = req.content
else:
log("Downloading provider logo error: {}".format(req.reason))
self.rows.append(Provider(logo=logo_data, name=name, pos=self._positon, url=row[5], on_id=on_id,
ssid=None, single=False, selected=True))
elif 6 < len_row < 10:
elif 6 < len_row < 14:
# Single services
name, url, ssid = None, None, None
if r[0].startswith("http"):
name, url, ssid = r[1], r[0], r[4]
elif r[1].startswith("http"):
name, url, ssid = r[2], r[1], r[5]
if row[0].startswith("http"):
name, url, ssid = row[1], row[0], row[0]
elif row[1].startswith("http"):
name, url, ssid = row[2], row[1], row[0]
if name and url:
on_id = "{}::{}".format(self._on_id if self._on_id else "1", self._freq)
@@ -237,14 +255,51 @@ class ProviderParser(HTMLParser):
super().reset()
def parse_providers(open_path):
def parse_providers(url):
""" Returns a list of providers sorted by logo [single channels after providers]. """
parser = ProviderParser()
parser.reset()
with open(open_path, encoding="utf-8", errors="replace") as f:
parser.feed(f.read())
request = requests.get(url=url, headers=_HEADERS)
if request.status_code == 200:
parser.feed(request.text)
else:
log("Parse providers error [{}]: {}".format(url, request.reason))
return parser.rows
def srt(p):
if p.logo is None:
return 1
return 0
providers = parser.rows
providers.sort(key=srt)
return providers
def download_picon(src_url, dest_path, callback):
""" Downloads and saves the picon to file. """
err_msg = "Picon download error: {} [{}]"
timeout = (3, 5) # connect and read timeouts
if callback:
callback("Downloading: {}.\n".format(os.path.basename(dest_path)))
req = requests.get(src_url, timeout=timeout, stream=True)
if req.status_code != 200:
err_msg = err_msg.format(src_url, req.reason)
log(err_msg)
if callback:
callback(err_msg + "\n")
else:
try:
with open(dest_path, "wb") as f:
for chunk in req:
f.write(chunk)
except OSError as e:
err_msg = "Saving picon [{}] error: {}".format(dest_path, e)
log(err_msg)
if callback:
callback(err_msg + "\n")
@run_task

View File

@@ -1704,6 +1704,8 @@ Author: Dmitriy Yefremov
<property name="can_focus">True</property>
<property name="editable">False</property>
<property name="wrap_mode">word-char</property>
<property name="left_margin">5</property>
<property name="right_margin">5</property>
<property name="overwrite">True</property>
</object>
</child>

View File

@@ -1,17 +1,16 @@
import os
import re
import shutil
import subprocess
import tempfile
from pathlib import Path
from urllib.parse import urlparse, unquote
from gi.repository import GLib, GdkPixbuf
from gi.repository import GLib, GdkPixbuf, Gio
from app.commons import run_idle, run_task, run_with_delay
from app.connections import upload_data, DownloadType, download_data, remove_picons
from app.settings import SettingsType, Settings
from app.tools.picons import PiconsParser, parse_providers, Provider, convert_to
from app.tools.picons import PiconsParser, parse_providers, Provider, convert_to, download_picon
from app.tools.satellites import SatellitesParser, SatelliteSource
from .dialogs import show_dialog, DialogType, get_message
from .main_helper import (update_entry_data, append_text_to_tview, scroll_to, on_popup_menu, get_base_model, set_picon,
@@ -30,6 +29,7 @@ class PiconsDialog:
self._POS_PATTERN = re.compile(r"^\d+\.\d+[EW]?$")
self._current_process = None
self._terminate = False
self._is_downloading = False
self._filter_binding = None
self._services = None
self._current_picon_info = None
@@ -480,7 +480,7 @@ class PiconsDialog:
try:
for sat in sats:
pos = sat[1]
name, pos = "{} ({})".format(sat[0], pos), "{}{}".format("-" if pos[-1] == "W" else "", pos[:-1])
name = "{} ({})".format(sat[0], pos)
if not self._terminate and model:
if pos in self._sat_positions:
@@ -492,49 +492,28 @@ class PiconsDialog:
model = view.get_model()
self._url_entry.set_text(model.get(model.get_iter(path), 1)[0])
@run_idle
def on_load_providers(self, item):
self._expander.set_expanded(True)
self.on_info_bar_close()
self._cancel_button.show()
url = self._url_entry.get_text()
try:
self._current_process = subprocess.Popen(["wget", "-pkP", self._TMP_DIR, url],
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
universal_newlines=True)
except FileNotFoundError as e:
self._cancel_button.hide()
self.show_info_message(str(e), Gtk.MessageType.ERROR)
else:
GLib.io_add_watch(self._current_process.stderr, GLib.IO_IN, self.write_to_buffer)
model = self._providers_view.get_model()
model.clear()
self.append_providers(url, model)
model = self._providers_view.get_model()
model.clear()
self.get_providers(model)
@run_task
def append_providers(self, url, model):
self._current_process.wait()
try:
self._terminate = False
providers = parse_providers(self._TMP_DIR + url[url.find("w"):])
except FileNotFoundError:
pass # NOP
else:
if providers:
for p in providers:
if self._terminate:
return
model.append((self.get_pixbuf(p[0]) if p[0] else TV_ICON, *p[1:]))
self.update_receive_button_state()
finally:
GLib.idle_add(self._cancel_button.hide)
self._terminate = False
def get_providers(self, model):
providers = parse_providers(self._url_entry.get_text())
if providers:
self.append_providers(providers, model)
def get_pixbuf(self, img_url):
return GdkPixbuf.Pixbuf.new_from_file_at_scale(filename=self._TMP_DIR + "www.lyngsat.com/" + img_url,
width=48, height=48, preserve_aspect_ratio=True)
@run_idle
def append_providers(self, providers, model):
for p in providers:
model.append((self.get_pixbuf(p[0]) if p[0] else TV_ICON, *p[1:]))
self.update_receive_button_state()
def get_pixbuf(self, img_data):
if img_data:
f = Gio.MemoryInputStream.new_from_data(img_data)
return GdkPixbuf.Pixbuf.new_from_stream_at_scale(f, 48, 32, True, None)
def on_receive(self, item):
self._cancel_button.show()
@@ -542,12 +521,12 @@ class PiconsDialog:
@run_task
def start_download(self):
if self._current_process.poll() is None:
if self._is_downloading:
self.show_dialog("The task is already running!", DialogType.ERROR)
return
self._terminate = False
self._expander.set_expanded(True)
self._is_downloading = True
GLib.idle_add(self._expander.set_expanded, True)
providers = self.get_selected_providers()
for prv in providers:
@@ -558,38 +537,47 @@ class PiconsDialog:
return
try:
for prv in providers:
if self._terminate:
return
self.process_provider(Provider(*prv))
picons_path = self._picons_dir_entry.get_text()
os.makedirs(os.path.dirname(picons_path), exist_ok=True)
picons = []
self.show_info_message(get_message("Please, wait..."), Gtk.MessageType.INFO)
import concurrent.futures
with concurrent.futures.ThreadPoolExecutor(max_workers=10) as executor:
# Getting links to picons.
futures = {executor.submit(self.process_provider, Provider(*p), picons_path): p for p in providers}
for future in concurrent.futures.as_completed(futures):
if not self._is_downloading:
executor.shutdown()
return
picons.extend(future.result())
# Getting picon images.
futures = {executor.submit(download_picon, *pic, self.append_output): pic for pic in picons}
done, not_done = concurrent.futures.wait(futures, timeout=0)
while self._is_downloading and not_done:
done, not_done = concurrent.futures.wait(not_done, timeout=5)
for future in not_done:
future.cancel()
concurrent.futures.wait(not_done)
if not self._is_downloading:
return
if not self._resize_no_radio_button.get_active():
self.resize(self._picons_dir_entry.get_text())
self.resize(picons_path)
else:
self.show_info_message(get_message("Done!"), Gtk.MessageType.INFO)
finally:
GLib.idle_add(self._cancel_button.hide)
self._terminate = False
self._is_downloading = False
def process_provider(self, prv):
url = prv.url
self.show_info_message(get_message("Please, wait..."), Gtk.MessageType.INFO)
self._current_process = subprocess.Popen(["wget", "-pkP", self._TMP_DIR, url],
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
universal_newlines=True)
GLib.io_add_watch(self._current_process.stderr, GLib.IO_IN, self.write_to_buffer)
self._current_process.wait()
path = self._TMP_DIR + (url[url.find("//") + 2:] if prv.single else self._BASE_URL + url[url.rfind("/") + 1:])
PiconsParser.parse(path, self._picons_dir_entry.get_text(),
self._TMP_DIR, prv, self._picon_ids, self.get_picons_format())
def write_to_buffer(self, fd, condition):
if condition == GLib.IO_IN:
char = fd.read(1)
self.append_output(char)
return True
return False
def process_provider(self, prv, picons_path):
self.append_output("Getting links to picons for: {}.\n".format(prv.name))
return PiconsParser.parse(prv, picons_path, self._picon_ids, self.get_picons_format())
@run_idle
def append_output(self, char):
@@ -615,7 +603,7 @@ class PiconsDialog:
self.show_info_message(get_message("Done!"), Gtk.MessageType.INFO)
def on_cancel(self, item=None):
if self.is_task_running() and show_dialog(DialogType.QUESTION, self._dialog) == Gtk.ResponseType.CANCEL:
if self._is_downloading and show_dialog(DialogType.QUESTION, self._dialog) == Gtk.ResponseType.CANCEL:
return True
self.terminate_task()
@@ -623,16 +611,15 @@ class PiconsDialog:
@run_task
def terminate_task(self):
self._terminate = True
if self._current_process:
self._current_process.terminate()
self.show_info_message(get_message("The task is canceled!"), Gtk.MessageType.WARNING)
self._is_downloading = False
self.show_info_message(get_message("The task is canceled!"), Gtk.MessageType.WARNING)
def on_close(self, window, event):
if self.on_cancel():
return True
self._terminate = True
self._is_downloading = False
self.save_window_size(window)
self.clean_data()
self._app.update_picons()
@@ -667,9 +654,10 @@ class PiconsDialog:
@run_idle
def show_info_message(self, text, message_type):
self._info_bar.set_visible(True)
self._info_bar.set_message_type(message_type)
self._info_bar.set_visible(False)
self._message_label.set_text(get_message(text))
self._info_bar.set_message_type(message_type)
self._info_bar.set_visible(True)
def on_picons_dir_open(self, entry, icon, event_button):
update_entry_data(entry, self._dialog, settings=self._settings)
@@ -834,9 +822,6 @@ class PiconsDialog:
return picon_format
def is_task_running(self):
return self._current_process and self._current_process.poll() is None
if __name__ == "__main__":
pass