Write output files to output folders

Author: lopuhin

run_boilerpipe.py

@@ -16,7 +16,8 @@ def main():
         item_id = os.path.basename(path).split('.')[0]
         extractor = Extractor(extractor='ArticleExtractor', html=html)
         output[item_id] = {'articleBody': extractor.getText()}
-    with codecs.open('output-boilerpipe.json', 'wt', encoding='utf8') as f:
+    with codecs.open(os.path.join('output', 'boilerpipe.json'),
+                     'wt', encoding='utf8') as f:
         json.dump(output, f, sort_keys=True, ensure_ascii=False, indent=4)
 
 

run_dragnet.py

@@ -14,7 +14,7 @@ def main():
         item_id = path.stem.split('.')[0]
         content = extract_content(html, encoding='utf8')
         output[item_id] = {'articleBody': content}
-    Path('output-dragnet.json').write_text(
+    (Path('output') / 'dragnet.json').write_text(
         json.dumps(output, sort_keys=True, ensure_ascii=False, indent=4),
         encoding='utf8')
 

run_html_text.py

@@ -13,7 +13,7 @@ def main():
             html = f.read()
         item_id = path.stem.split('.')[0]
         output[item_id] = {'articleBody': html_text.extract_text(html)}
-    Path('output-html-text.json').write_text(
+    (Path('output') / 'html-text.json').write_text(
         json.dumps(output, sort_keys=True, ensure_ascii=False, indent=4),
         encoding='utf8')
 

run_newspaper.py

@@ -18,7 +18,7 @@ def main():
         article.set_html(html)
         article.parse()
         output[item_id] = {'articleBody': article.text}
-    Path('output-newspaper.json').write_text(
+    (Path('output') / 'newspaper.json').write_text(
         json.dumps(output, sort_keys=True, ensure_ascii=False, indent=4),
         encoding='utf8')
 

run_readability.py

@@ -16,7 +16,7 @@ def main():
         doc = Document(html)
         text = html_text.extract_text(doc.summary(html_partial=True))
         output[item_id] = {'articleBody': text}
-    Path('output-readability.json').write_text(
+    (Path('output') / 'readability.json').write_text(
         json.dumps(output, sort_keys=True, ensure_ascii=False, indent=4),
         encoding='utf8')