| Index: trunk/tools/editor_trends/etl/enricher.py |
| — | — | @@ -257,8 +257,8 @@ |
| 258 | 258 | 'User Talk': 3, |
| 259 | 259 | } |
| 260 | 260 | ns = {} |
| 261 | | - if title.text != None: |
| 262 | | - title = title.text |
| | 261 | + if title != None: |
| | 262 | + #title = title.text |
| 263 | 263 | for namespace in namespaces: |
| 264 | 264 | if title.startswith(namespace): |
| 265 | 265 | ns['namespace'] = namespaces[namespace] |
| — | — | @@ -324,8 +324,8 @@ |
| 325 | 325 | |
| 326 | 326 | |
| 327 | 327 | def create_variables(article, cache, bots): |
| 328 | | - title = article['title'].text |
| 329 | | - namespace = determine_namespace(article['title']) |
| | 328 | + title = article['title'] |
| | 329 | + namespace = determine_namespace(title) |
| 330 | 330 | |
| 331 | 331 | if namespace != False: |
| 332 | 332 | cache.stats.count_articles += 1 |
| — | — | @@ -379,7 +379,7 @@ |
| 380 | 380 | |
| 381 | 381 | article = {} |
| 382 | 382 | id = False |
| 383 | | - article[root.tag] = root |
| | 383 | + article[root.tag] = root.text |
| 384 | 384 | article['revisions'] = [] |
| 385 | 385 | for event, elem in context: |
| 386 | 386 | if event == 'end' and elem.tag == 'revision': |
| — | — | @@ -388,6 +388,7 @@ |
| 389 | 389 | article[elem.tag] = elem |
| 390 | 390 | id = True |
| 391 | 391 | |
| | 392 | + root.clear() |
| 392 | 393 | return article |
| 393 | 394 | |
| 394 | 395 | |