Merge branch 'release-1.3' back into develop

For some asian countries there is the original name following after a
linebreak. This is not needed in overview, so cat it of.

Related Task: [https://fs.golderweb.de/index.php?do=details&task_id=118 FS#118]

.gitignore

@@ -62,3 +62,5 @@ target/
 
 # Test
 test.py
+
+disabled

README.md

@@ -0,0 +1,43 @@
+# wiki-jogobot-charts
+This is a [Pywikibot](https://www.mediawiki.org/wiki/Manual:Pywikibot) based [Wikipedia Bot](https://de.wikipedia.org/wiki/Wikipedia:Bots)
+of [User:JogoBot](https://de.wikipedia.org/wiki/Benutzer:JogoBot) on the
+[German Wikipedia](https://de.wikipedia.org/wiki/Wikipedia:Hauptseite).
+
+On [JogoBots wikipedia user page](https://de.wikipedia.org/wiki/Benutzer:JogoBot/Charts) a more detailed description can be found.
+
+## Requirements
+* python3.4+
+
+
+* pywikibot-core
+
+
+The libraries above need to be installed and configured manualy considering [documentation of pywikibot-core](https://www.mediawiki.org/wiki/Manual:Pywikibot).
+
+* mwparserfromhell
+* isoweek
+* [jogobot-core module](https://git.golderweb.de/wiki/jogobot)
+
+Those can be installed using pip and the _requirements.txt_ file provided with this packet
+
+    pip install -r requirements.txt
+
+## Versions
+
+* v1.3
+  - jogobot module not longer included
+* v1.2
+  - improved repo structure
+* v1.1
+  - Cut Titel and Interpret on first linebreak
+* v1.0
+  - first stable release
+
+## Bugs
+[jogobot-charts Issues](https://git.golderweb.de/wiki/jogobot-charts/issues)
+
+## License
+GPLv3+
+
+## Author Information
+Copyright 2018 Jonathan Golder <jonathan@golderweb.de>

__init__.py

@@ -3,7 +3,7 @@
 #
 #  __init__.py
 #
-#  Copyright 2015 GOLDERWEB – Jonathan Golder <jonathan@golderweb.de>
+#  Copyright 2016 Jonathan Golder <jonathan@golderweb.de>
 #
 #  This program is free software; you can redistribute it and/or modify
 #  it under the terms of the GNU General Public License as published by

charts.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8  -*-
 #
-#  chartsbot.py
+#  charts.py
 #
 #  original version by:
 #
@@ -11,7 +11,7 @@
 #
 #  modified by:
 #
-#  Copyright 2015 GOLDERWEB – Jonathan Golder <jonathan@golderweb.de>
+#  Copyright 2017 Jonathan Golder <jonathan@golderweb.de>
 #
 #  This program is free software; you can redistribute it and/or modify
 #  it under the terms of the GNU General Public License as published by
@@ -46,10 +46,14 @@ The following parameters are supported:
 
 
 import locale
+import os
+import sys
 
 import pywikibot
 from pywikibot import pagegenerators
 
+import jogobot
+
 from summarypage import SummaryPage
 
 # This is required for the text that is shown when you run this script
@@ -87,23 +91,41 @@ class ChartsBot( ):
         # Force parsing of countrylist
         self.force_reload = force_reload
 
-        # Set the edit summary message
+        # Output Information
+        jogobot.output( "Chartsbot invoked" )
+
+        # Save pywikibot site object
         self.site = pywikibot.Site()
-        self.summary = "Bot: Aktualisiere Übersichtsseite Nummer-eins-Hits"
+
+        # Define edit summary
+        self.summary = jogobot.config["charts"]["edit_summary"].strip()
+
+        # Make sure summary starts with "Bot:"
+        if not self.summary[:len("Bot:")] == "Bot:":
+            self.summary = "Bot: " + self.summary.strip()
 
         # Set locale to 'de_DE.UTF-8'
         locale.setlocale(locale.LC_ALL, 'de_DE.UTF-8')
 
     def run(self):
         """Process each page from the generator."""
+        # Count skipped pages (redirect or missing)
+        skipped = 0
         for page in self.generator:
-            self.treat(page)
+            if not self.treat(page):
+                skipped += 1
+
+        if skipped:
+            jogobot.output( "Chartsbot finished, {skipped} page(s) skipped"
+                            .format( skipped=skipped ) )
+        else:
+            jogobot.output( "Chartsbot finished successfully" )
 
     def treat(self, page):
         """Load the given page, does some changes, and saves it."""
         text = self.load(page)
         if not text:
-            return
+            return False
 
         ################################################################
         # NOTE: Here you can modify the text in whatever way you want. #
@@ -118,7 +140,9 @@ class ChartsBot( ):
             text = sumpage.get_new_text()
 
         if not self.save(text, page, self.summary, False):
-            pywikibot.output(u'Page %s not saved.' % page.title(asLink=True))
+            jogobot.output(u'Page %s not saved.' % page.title(asLink=True))
+
+        return True
 
     def load(self, page):
         """Load the text of the given page."""
@@ -126,27 +150,31 @@ class ChartsBot( ):
             # Load the page
             text = page.get()
         except pywikibot.NoPage:
-            pywikibot.output(u"Page %s does not exist; skipping."
-                             % page.title(asLink=True))
+            jogobot.output( u"Page %s does not exist; skipping."
+                            % page.title(asLink=True), "ERROR" )
         except pywikibot.IsRedirectPage:
-            pywikibot.output(u"Page %s is a redirect; skipping."
-                             % page.title(asLink=True))
+            jogobot.output( u"Page %s is a redirect; skipping."
+                            % page.title(asLink=True), "ERROR" )
         else:
             return text
-        return None
+        return False
 
     def save(self, text, page, comment=None, minorEdit=True,
              botflag=True):
         """Update the given page with new text."""
         # only save if something was changed (and not just revision)
         if text != page.get():
-            # Show the title of the page we're working on.
-            # Highlight the title in purple.
-            pywikibot.output(u"\n\n>>> \03{lightpurple}%s\03{default} <<<"
-                             % page.title())
-            # show what was changed
-            pywikibot.showDiff(page.get(), text)
-            pywikibot.output(u'Comment: %s' % comment)
+
+            # Show diff only in interactive mode or in verbose mode
+            if not self.always or pywikibot.config.verbose_output:
+
+                # Show the title of the page we're working on.
+                # Highlight the title in purple.
+                jogobot.output( u">>> \03{lightpurple}%s\03{default} <<<"
+                                % page.title())
+                # show what was changed
+                pywikibot.showDiff(page.get(), text)
+                jogobot.output(u'Comment: %s' % comment)
 
             if self.always or pywikibot.input_yn(
                     u'Do you want to accept these changes?',
@@ -157,17 +185,17 @@ class ChartsBot( ):
                     page.save(summary=comment or self.comment,
                               minor=minorEdit, botflag=botflag)
                 except pywikibot.LockedPage:
-                    pywikibot.output(u"Page %s is locked; skipping."
-                                     % page.title(asLink=True))
+                    jogobot.output( u"Page %s is locked; skipping."
+                                    % page.title(asLink=True), "ERROR" )
                 except pywikibot.EditConflict:
-                    pywikibot.output(
+                    jogobot.output(
                         u'Skipping %s because of edit conflict'
-                        % (page.title()))
+                        % (page.title()), "ERROR")
                 except pywikibot.SpamfilterError as error:
-                    pywikibot.output(
+                    jogobot.output(
                         u'Cannot change %s because of spam blacklist \
 entry %s'
-                        % (page.title(), error.url))
+                        % (page.title(), error.url), "ERROR")
                 else:
                     return True
         return False
@@ -182,42 +210,65 @@ def main(*args):
     @param args: command line arguments
     @type args: list of unicode
     """
+
     # Process global arguments to determine desired site
     local_args = pywikibot.handle_args(args)
 
-    # This factory is responsible for processing command line arguments
-    # that are also used by other scripts and that determine on which pages
-    # to work on.
-    genFactory = pagegenerators.GeneratorFactory()
-    # The generator gives the pages that should be worked upon.
-    gen = None
+    # Get the jogobot-task_slug (basename of current file without ending)
+    task_slug = os.path.basename(__file__)[:-len(".py")]
 
-    # If always is True, bot won't ask for confirmation of edit (automode)
-    always = False
+    # Before run, we need to check wether we are currently active or not
+    try:
+        # Will throw Exception if disabled/blocked
+        jogobot.is_active( task_slug )
 
-    # If force_reload is True, bot will always parse Countrylist regardless of
-    # parsing is needed or not
-    force_reload = False
+    except jogobot.jogobot.Blocked:
+        (type, value, traceback) = sys.exc_info()
+        jogobot.output( "\03{lightpurple} %s (%s)" % (value, type ),
+                        "CRITICAL" )
 
-    # Parse command line arguments
-    for arg in local_args:
-        if arg.startswith("-always"):
-            always = True
-        elif arg.startswith("-force-reload"):
-            force_reload = True
-        else:
-            genFactory.handleArg(arg)
+    except jogobot.jogobot.Disabled:
+        (type, value, traceback) = sys.exc_info()
+        jogobot.output( "\03{red} %s (%s)" % (value, type ),
+                        "ERROR" )
 
-    if not gen:
-        gen = genFactory.getCombinedGenerator()
-    if gen:
-        # The preloading generator is responsible for downloading multiple
-        # pages from the wiki simultaneously.
-        gen = pagegenerators.PreloadingGenerator(gen)
-        bot = ChartsBot(gen, always, force_reload)
-        bot.run()
+    # Bot/Task is active
     else:
-        pywikibot.showHelp()
+        # This factory is responsible for processing command line arguments
+        # that are also used by other scripts and that determine on which pages
+        # to work on.
+        genFactory = pagegenerators.GeneratorFactory()
+        # The generator gives the pages that should be worked upon.
+        gen = None
+
+        # If always is True, bot won't ask for confirmation of edit (automode)
+        always = False
+
+        # If force_reload is True, bot will always parse Countrylist regardless
+        # if parsing is needed or not
+        force_reload = False
+
+        # Parse command line arguments
+        for arg in local_args:
+            if arg.startswith("-always"):
+                always = True
+            elif arg.startswith("-force-reload"):
+                force_reload = True
+            else:
+                pass
+                genFactory.handleArg(arg)
+
+        if not gen:
+            gen = genFactory.getCombinedGenerator()
+        if gen:
+            # The preloading generator is responsible for downloading multiple
+            # pages from the wiki simultaneously.
+            gen = pagegenerators.PreloadingGenerator(gen)
+            bot = ChartsBot(gen, always, force_reload)
+            if bot:
+                bot.run()
+        else:
+            pywikibot.showHelp()
 
 if( __name__ == "__main__" ):
     main()

countrylist.py

@@ -3,7 +3,7 @@
 #
 #  countrylist.py
 #
-#  Copyright 2015 GOLDERWEB – Jonathan Golder <jonathan@golderweb.de>
+#  Copyright 2017 Jonathan Golder <jonathan@golderweb.de>
 #
 #  This program is free software; you can redistribute it and/or modify
 #  it under the terms of the GNU General Public License as published by
@@ -25,6 +25,7 @@
 Provides a class for handling charts list per country and year
 """
 
+import re
 import locale
 from datetime import datetime
 
@@ -33,6 +34,8 @@ from isoweek import Week
 import pywikibot
 import mwparserfromhell as mwparser
 
+import jogobot
+
 
 class CountryList():
     """
@@ -66,7 +69,8 @@ class CountryList():
 
         # Check if page exits
         if not self.page.exists():
-            return False
+            raise CountryListError( "CountryList " +
+                                    str(wikilink.title) + " does not exists!" )
 
         # Initialise attributes
         __attr = (  "wikicode", "entry", "chartein", "_chartein_raw",
@@ -96,15 +100,15 @@ class CountryList():
 
     def find_year( self ):
         """
-        Try to find the year related to CountryList
+        Try to find the year related to CountryList using regex
         """
-        self.year = datetime.now().year
+        match = re.search( r"^.+\((\d{4})\)", self.page.title() )
 
-        # Check if year is in page.title, if not try last year
-        if str( self.year ) not in self.page.title():
-            self.year -= 1
-        # If last year does not match, raise YearError
-        if str( self.year ) not in self.page.title():
+        # We matched something
+        if match:
+            self.year = int(match.group(1))
+
+        else:
             raise CountryListError( "CountryList year is errorneous!" )
 
     def parse( self ):
@@ -112,6 +116,9 @@ class CountryList():
         Handles the parsing process
         """
 
+        # Set revid
+        self.revid = self.page.latest_revision_id
+
         # Parse page with mwparser
         self.generate_wikicode()
 
@@ -126,6 +133,10 @@ class CountryList():
         # For easy detecting wether we have parsed self
         self.parsed = True
 
+        # Log parsed page
+        jogobot.output( "Parsed revision {revid} of page [[{title}]]".format(
+            revid=self.revid, title=self.page.title() ) )
+
     def detect_belgian( self ):
         """
         Detect wether current entry is on of the belgian (Belgien/Wallonien)
@@ -156,11 +167,18 @@ class CountryList():
         # For belgian list we need to select subsection of country
         belgian = self.detect_belgian()
 
-        if belgian:
-            singles_section = self.wikicode.get_sections(
-                matches=belgian )[0].get_sections( matches="Singles" )[0]
-        else:
-            singles_section = self.wikicode.get_sections(matches="Singles")[0]
+        # Select Singles-Section
+        # Catch Error if we have none
+        try:
+            if belgian:
+                singles_section = self.wikicode.get_sections(
+                    matches=belgian )[0].get_sections( matches="Singles" )[0]
+            else:
+                singles_section = self.wikicode.get_sections(
+                    matches="Singles" )[0]
+
+        except IndexError:
+            raise CountryListError( "No Singles-Section found!")
 
         # Since we have multiple categories in some countrys we need
         # to select the first wrapping template
@@ -230,7 +248,15 @@ class CountryList():
         If param is not present raise Error
         """
         if self.entry.has( "Chartein" ):
-            self._chartein_raw = self.entry.get("Chartein").value.strip()
+            self._chartein_raw = self.entry.get("Chartein").value
+
+            # Remove possible ref-tags
+            for ref in self._chartein_raw.ifilter_tags(matches="ref"):
+                self._chartein_raw.remove( ref )
+
+            # Remove whitespace
+            self._chartein_raw = str(self._chartein_raw).strip()
+
         else:
             raise CountryListEntryError( "Template Parameter 'Chartein' is \
 missing!" )
@@ -256,7 +282,17 @@ missing!" )
         If param is not present raise Error
         """
         if self.entry.has( "Titel" ):
-            self._titel_raw = self.entry.get("Titel").value.strip()
+            self._titel_raw = self.entry.get("Titel").value
+
+            # Only use part before possible "<br"
+            self.remove_lines(self._titel_raw)
+
+            # Remove possible ref-tags
+            for ref in self._titel_raw.ifilter_tags(matches="ref"):
+                self._titel_raw.remove( ref )
+
+            # Remove whitespace
+            self._titel_raw = str(self._titel_raw).strip()
         else:
             raise CountryListEntryError( "Template Parameter 'Titel' is \
 missing!" )
@@ -321,7 +357,56 @@ missing!" )
         If param is not present raise Error
         """
         if self.entry.has( "Interpret" ):
-            self._interpret_raw = self.entry.get("Interpret").value.strip()
+            self._interpret_raw = self.entry.get("Interpret").value
+
+            # Only use part before possible "<br"
+            self.remove_lines(self._interpret_raw)
+
+            # Remove possible ref-tags
+            for ref in self._interpret_raw.ifilter_tags(matches="ref"):
+                self._interpret_raw.remove( ref )
+
+            # Handle SortKeyName and SortKey
+            for template in self._interpret_raw.ifilter_templates(
+                    matches="SortKey" ):
+
+                if template.name == "SortKeyName":
+                    # Differing Link-Destination is provided as param 3
+                    if template.has(3):
+                        # Construct link out of Template, Params:
+                        # 1 = Surname
+                        # 2 = Name
+                        # 3 = Link-Dest
+                        interpret_link = mwparser.nodes.wikilink.Wikilink(
+                            str(template.get(3).value),
+                            str(template.get(1).value) + " " +
+                            str(template.get(2).value) )
+
+                    # Default Link-Dest [[Surname Name]]
+                    else:
+                        interpret_link = mwparser.nodes.wikilink.Wikilink(
+                            str(template.get(1).value) + " " +
+                            str(template.get(2).value) )
+
+                    # Replace Template with link
+                    self._interpret_raw.replace( template, interpret_link )
+
+                # SortKey
+                else:
+                    # Replace SortKey with text from param 2 if present
+                    if template.has(2):
+                        self._interpret_raw.replace( template,
+                                                     template.get(2).value)
+                    # Else Remove SortKey (text should follow behind SortKey)
+                    else:
+                        self._interpret_raw.replace( template, None)
+
+                # Normally won't be needed as there should be only one
+                # SortKey-Temlate but ... its a wiki
+                break
+
+            # Remove whitespace
+            self._interpret_raw = str(self._interpret_raw).strip()
         else:
             raise CountryListEntryError( "Template Parameter 'Interpret' is \
 missing!" )
@@ -379,6 +464,42 @@ missing!" )
         else:
             return str(keywords[0])
 
+    def remove_lines(self, wikicode):
+        """
+        Removes linebreaks (<br>) and everything after them in given wikicode
+        """
+        # Catch wrong typed param
+        if not isinstance(wikicode, mwparser.wikicode.Wikicode):
+            raise TypeError(str(type(self)) + "._remove_lines() expects " +
+                            "parameter 'wikicode' of type " +
+                            "'mwparserfromhell.wikicode.Wikicode', " +
+                            str(type(wikicode)) + " was given!")
+
+        # Find first linebreak
+        br = next(wikicode.ifilter_tags(matches="br"), None)
+
+        # If there is one, get its position and slice nodes-list
+        if br:
+            brpos = wikicode.nodes.index(br)
+            wikicode.nodes = wikicode.nodes[0:brpos]
+
+    def __str__( self ):
+        """
+        Returns str repression for Object
+        """
+        if self.parsed:
+            return ("CountryList( Link = \"{link}\", Revid = \"{revid}\", " +
+                    "Interpret = \"{interpret}\", Titel = \"{titel}\", " +
+                    "Chartein = \"{chartein}\" )").format(
+                        link=repr(self.wikilink),
+                        revid=self.revid,
+                        interpret=self.interpret,
+                        titel=self.titel,
+                        chartein=repr(self.chartein))
+        else:
+            return "CountryList( Link = \"{link}\" )".format(
+                link=repr(self.wikilink))
+
 
 class CountryListError( Exception ):
     """
@@ -392,3 +513,98 @@ class CountryListEntryError( CountryListError ):
     Handles errors occuring in class CountryList related to entrys
     """
     pass
+
+
+class CountryListUnitTest():
+    """
+    Defines Test-Functions for CountryList-Module
+    """
+
+    testcases = ( { "Link": mwparser.nodes.Wikilink( "Benutzer:JogoBot/Charts/Tests/Liste der Nummer-eins-Hits in Frankreich (2015)" ),  # noqa
+                    "revid": 148453827,
+                    "interpret": "[[Adele (Sängerin)|Adele]]",
+                    "titel": "[[Hello (Adele-Lied)|Hello]]",
+                    "chartein": datetime( 2015, 10, 23 ) },
+                  { "Link": mwparser.nodes.Wikilink( "Benutzer:JogoBot/Charts/Tests/Liste der Nummer-eins-Hits in Belgien (2015)", "Wallonien"),  # noqa
+                    "revid": 148455281,
+                    "interpret": "[[Nicky Jam]] & [[Enrique Iglesias (Sänger)|Enrique Iglesias]]",  # noqa
+                    "titel": "El perdón",
+                    "chartein": datetime( 2015, 9, 12 ) } )
+
+    def __init__( self, page=None ):
+        """
+        Constructor
+        Set attribute page
+        """
+        if page:
+            self.page_link = mwparser.nodes.Wikilink( page  )
+        else:
+            self.page_link = None
+
+    def treat( self ):
+        """
+        Start testing either manually with page provided by cmd-arg page or
+        automatically with predefined test case
+        """
+        if self.page_link:
+            self.man_test()
+        else:
+            self.auto_test()
+
+    def auto_test( self ):
+        """
+        Run automatic tests with predefined test data from wiki
+        """
+
+        for case in type(self).testcases:
+
+            self.countrylist = CountryList( case["Link"] )
+
+            if( self.countrylist.is_parsing_needed( case["revid"] ) or not
+                self.countrylist.is_parsing_needed( case["revid"] + 1 ) ):
+                    raise Exception(
+                        "CountryList.is_parsing_needed() does not work!" )
+
+            self.countrylist.parse()
+
+            for key in case:
+
+                if key == "Link":
+                    continue
+
+                if not case[key] == getattr(self.countrylist, key ):
+                    raise Exception( key + " – " + str(
+                                     getattr(self.countrylist, key ) ))
+
+    def man_test( self ):
+        """
+        Run manual test with page given in parameter
+        """
+        self.countrylist = CountryList( self.page_link )
+
+        self.countrylist.parse()
+
+        print( self.countrylist )
+        print( "Since we have no data to compare, you need to manually " +
+               "check data above against given page to ensure correct " +
+               "working of module!" )
+
+
+def main(*args):
+    """
+    Handling direct calls --> unittest
+    """
+    # Process global arguments to determine desired site
+    local_args = pywikibot.handle_args(args)
+
+    # Parse command line arguments
+    for arg in local_args:
+        if arg.startswith("-page:"):
+            page = arg[ len("-page:"): ]
+
+    # Call unittest-class
+    test = CountryListUnitTest( page )
+    test.treat()
+
+if __name__ == "__main__":
+    main()

requirements.txt

@@ -0,0 +1,8 @@
+# mwparserfromhell
+mwparserfromhell
+
+# isoweek
+isoweek
+
+# jogobot
+git+https://git.golderweb.de/wiki/jogobot.git#egg=jogobot

summarypage.py

@@ -3,7 +3,7 @@
 #
 #  summarypage.py
 #
-#  Copyright 2015 GOLDERWEB – Jonathan Golder <jonathan@golderweb.de>
+#  Copyright 2017 Jonathan Golder <jonathan@golderweb.de>
 #
 #  This program is free software; you can redistribute it and/or modify
 #  it under the terms of the GNU General Public License as published by
@@ -30,6 +30,8 @@ from datetime import datetime, timedelta
 # import pywikibot
 import mwparserfromhell as mwparser
 
+import jogobot
+
 from countrylist import CountryList, CountryListError
 
 
@@ -143,11 +145,11 @@ class SummaryPageEntry():
         # Get current year
         current_year = datetime.now().year
 
-        # Store old link.title
-        link_title = self.countrylist_wikilink.title
-
         # If list is from last year, replace year
-        if (current_year - 1) in link_title:
+        if (current_year - 1) in self.countrylist_wikilink.title:
+            jogobot.output( "Trying to use new years list for [[{page}]]"
+                            .format( page=self.countrylist_wikilink.title ) )
+
             self.countrylist_wikilink.title.replace( (current_year - 1),
                                                      current_year )
 
@@ -160,7 +162,15 @@ class SummaryPageEntry():
         # Maybe fallback to last years list
         except CountryListError:
 
-            self.countrylist_wikilink.title = link_title
+            # If list is from last year, replace year
+            if (current_year ) in self.countrylist_wikilink.title:
+                jogobot.output( "New years list for [[{page}]] does not " +
+                                "exist, fall back to old list!".format(
+                                    page=self.countrylist_wikilink.title ) )
+
+                self.countrylist_wikilink.title.replace( current_year,
+                                                         (current_year - 1) )
+
             self.countrylist = CountryList( self.countrylist_wikilink )
 
             self.maybe_parse_countrylist()