Merge branch 'broken-jtv-header' into 'master'

Alternative JTV header and encoding support

See merge request andrey/jtv2xmltv!5

jtv2xmltv/convert.py

@@ -1,10 +1,13 @@
+from __future__ import print_function
 from xml.etree import ElementTree
 import zipfile
+import sys
 from jtv2xmltv.parser import parse_schedule, parse_titles
 from jtv2xmltv.utils import fix_zip_filename
 
 
-def convert_jtv_to_xmltv(jtv_filename, epg_encoding="utf8", epg_timezone="UTC", epg_lang=None):
+def convert_jtv_to_xmltv(jtv_filename, jtv_encoding="cp1251", xmltv_encoding="utf8", xmltv_timezone="UTC",
+                         xmltv_lang=None):
     archive = zipfile.ZipFile(jtv_filename, 'r')
     xmltv_tv = ElementTree.Element('tv')
     channel_id = 0
@@ -14,23 +17,34 @@ def convert_jtv_to_xmltv(jtv_filename, epg_encoding="utf8", epg_timezone="UTC",
             channel_name = unicode_name[0:-4]
             channel_id = channel_id + 1
             xmltv_channel = ElementTree.SubElement(xmltv_tv, 'channel', id=str(channel_id))
-            if epg_lang is not None:
-                ElementTree.SubElement(xmltv_channel, 'display-name', lang=str(epg_lang)).text = channel_name
+            if xmltv_lang is not None:
+                ElementTree.SubElement(xmltv_channel, 'display-name', lang=str(xmltv_lang)).text = channel_name
             else:
                 ElementTree.SubElement(xmltv_channel, 'display-name').text = channel_name
     channel_id = 0
     for filename in archive.namelist():
         if filename.endswith('.pdt'):
             channel_id = channel_id + 1
-            titles = archive.read(filename)
-            channel_titles = parse_titles(titles)
-            schedules = archive.read(filename[0:-4] + ".ndx")
-            channel_schedules = parse_schedule(schedules)
+            try:
+                titles_filename = filename
+                titles = archive.read(titles_filename)
+                channel_titles = parse_titles(titles, encoding=jtv_encoding)
+            except Exception as e:
+                print("Failed to process titles in {}: {}".format(titles_filename, str(e)), file=sys.stderr)
+                continue
+
+            try:
+                schedules_filename = filename[0:-4] + ".ndx"
+                schedules = archive.read(schedules_filename)
+                channel_schedules = parse_schedule(schedules)
+            except Exception as e:
+                print("Failed to process schedule in {}: {}".format(schedules_filename, str(e)), file=sys.stderr)
+                continue
 
             i = 0
             for curr_title in channel_titles:
-                if epg_timezone != "UTC":
-                    time_format = '%Y%m%d%H%M%S ' + epg_timezone
+                if xmltv_timezone != "UTC":
+                    time_format = '%Y%m%d%H%M%S ' + xmltv_timezone
                 else:
                     time_format = '%Y%m%d%H%M%S'
                 if i < len(channel_schedules) - 1:
@@ -42,10 +56,10 @@ def convert_jtv_to_xmltv(jtv_filename, epg_encoding="utf8", epg_timezone="UTC",
                     xmltv_programme = ElementTree.SubElement(xmltv_tv, 'programme',
                                                              start=channel_schedules[i].strftime(time_format),
                                                              channel=str(channel_id))
-                if epg_lang is not None:
-                    ElementTree.SubElement(xmltv_programme, 'title', lang=str(epg_lang)).text = curr_title
+                if xmltv_lang is not None:
+                    ElementTree.SubElement(xmltv_programme, 'title', lang=str(xmltv_lang)).text = curr_title
                 else:
                     ElementTree.SubElement(xmltv_programme, 'title').text = curr_title
                 i = i + 1
     archive.close()
-    return ElementTree.tostring(xmltv_tv, encoding=epg_encoding, method="xml").decode(encoding=epg_encoding)
+    return ElementTree.tostring(xmltv_tv, encoding=xmltv_encoding, method="xml").decode(encoding=xmltv_encoding)

jtv2xmltv/main.py

@@ -16,16 +16,18 @@ def main():
     parser.add_argument('-i', '--inputfile', required=True)
     parser.add_argument('-o', '--outputfile', default='-')
     parser.add_argument('-t', '--timezone')
+    parser.add_argument('-e', '--encoding', default='cp1251')
     args = parser.parse_args()
     jtv_filename = args.inputfile
     xmltv_filename = args.outputfile
+    jtv_encoding = args.encoding
     if args.timezone is None:
         tz_format = 'UTC'
     elif args.timezone[0] == '-' or args.timezone[0] == '+':
         tz_format = str(args.timezone)
     else:
         tz_format = '+' + str(args.timezone)
-    xmltv_content = convert_jtv_to_xmltv(jtv_filename, epg_timezone=tz_format)
+    xmltv_content = convert_jtv_to_xmltv(jtv_filename, jtv_encoding=jtv_encoding, xmltv_timezone=tz_format)
     if xmltv_filename is None or xmltv_filename == "-":
         print(xmltv_content)
     else:

jtv2xmltv/parser.py

@@ -2,17 +2,16 @@
 from jtv2xmltv.utils import filetime_to_datetime
 
 
-def parse_titles(data):
-    jtv_header = b"JTV 3.x TV Program Data\x0a\x0a\x0a"
-    if data[0:26] != jtv_header:
+def parse_titles(data, encoding="cp1251"):
+    jtv_headers = [b"JTV 3.x TV Program Data\x0a\x0a\x0a", b"JTV 3.x TV Program Data\xa0\xa0\xa0"]
+    if data[0:26] not in jtv_headers:
         raise Exception('Invalid JTV format')
     data = data[26:]
     titles = []
     while data:
         title_length = int(struct.unpack('<H', data[0:2])[0])
         data = data[2:]
-        # TODO: epg titles encoding could be not cp1251
-        title = data[0:title_length].decode('cp1251')
+        title = data[0:title_length].decode(encoding)
         data = data[title_length:]
         titles.append(title)
     return titles

setup.py

@@ -3,7 +3,7 @@
 
 setup(
     name="jtv2xmltv",
-    version="0.1.1",
+    version="0.2.0",
     packages=find_packages(),
     entry_points={
         'console_scripts': [
@@ -14,7 +14,7 @@
     author_email="tataranovich@gmail.com",
     description="JTV to XMLTV converter",
     long_description="JTV to XMLTV converter",
-    license="GPL-3",
+    license="MIT",
     keywords="epg xmltv jtv",
     url="https://github.com/tataranovich/jtv2xmltv/",
 )

tests/test_parser.py

@@ -8,8 +8,13 @@ def test_correct_jtv_header():
     assert parse_titles(data) == []
 
 
-def test_incorrect_jtv_header():
+def test_alternative_jtv_header():
     data = b'JTV 3.x TV Program Data\xA0\xA0\xA0'
+    assert parse_titles(data) == []
+
+
+def test_incorrect_jtv_header():
+    data = b'JTV 3.x TV Program Data\xFF\xFF\xFF'
     with pytest.raises(Exception):
         parse_titles(data)