Merge pull request #381 from PyAr/transmit-redirection-score

Transmit redirection scores to destination articles

Author: facundobatista

src/preprocessing/preprocess.py

@@ -1,7 +1,7 @@
 #!/usr/bin/python
 # -*- coding: utf-8 -*-
 
-# Copyright 2020 CDPedistas (see AUTHORS.txt)
+# Copyright 2020-2021 CDPedistas (see AUTHORS.txt)
 #
 # This program is free software: you can redistribute it and/or modify it
 # under the terms of the GNU General Public License version 3, as published
@@ -177,9 +177,10 @@ def process(self):
                     scores_log.write("{}|R|{:d}\n".format(
                         to3dirs.to_pagina(page_path), this_total_score))
 
-                    # save the extra pages score (that may exist or not in the dump)
-                    for extra_page, extra_score in other_pages_scores:
-                        scores_log.write("{}|E|{:d}\n".format(extra_page, extra_score))
+                # save the extra pages score (that may exist or not in the dump) even if page
+                # is discarded (e.g. for transfering score from redirect pages to its targets)
+                for extra_page, extra_score in other_pages_scores:
+                    scores_log.write("{}|E|{:d}\n".format(extra_page, extra_score))
 
                 # with score or discarded, log it as processed
                 processed_before_log.write(page_path + "\n")

tests/test_preprocess.py

@@ -1,4 +1,4 @@
-# Copyright 2013-2020 CDPedistas (see AUTHORS.txt)
+# Copyright 2013-2021 CDPedistas (see AUTHORS.txt)
 #
 # This program is free software: you can redistribute it and/or modify it
 # under the terms of the GNU General Public License version 3, as published
@@ -199,6 +199,28 @@ def test_empty_dir(self, tmp_path, wikisite):
         ws.commit()
         assert os.path.getsize(config.LOG_PREPROCESADO) == 0
 
+    def test_transmit_redirection_score_to_destination(self, mocker, tmp_path, wikisite):
+        """Test that extra scores produced while processing a redirection are not discarded."""
+        ws = wikisite(str(tmp_path))
+
+        # mock preprocessor that discards the redirection and transmits the score to destination
+        omit_redirects = mocker.Mock(return_value=(None, [('destination', 1234)]))
+        mocker.patch.object(ws, 'preprocessors', [omit_redirects])
+
+        # dummy redirection article that will be discarded
+        article = tmp_path.joinpath('r', 'e', 'd', 'redirection')
+        article.parent.mkdir(parents=True)
+        article.touch()
+
+        ws.process()
+        ws.commit()
+
+        with open(preprocess.LOG_SCORES_ACCUM, 'rt', encoding='utf-8') as fh:
+            scores = fh.read()
+
+        # real score of redirection is discarded, extra score of destination is saved
+        assert scores == 'destination|E|1234\n'
+
 
 class TestPagesSelector(object):
     """Tests for the PagesSelector"""