| Index: trunk/tools/editor_trends/etl/enricher.py |
| — | — | @@ -382,7 +382,11 @@ |
| 383 | 383 | def parse_xml(buffer): |
| 384 | 384 | context = iterparse(buffer, events=('end',)) |
| 385 | 385 | context = iter(context) |
| 386 | | - event, root = context.next() |
| | 386 | + try: |
| | 387 | + event, root = context.next() |
| | 388 | + except SyntaxError, e: |
| | 389 | + print e |
| | 390 | + print buffer.getvalue() |
| 387 | 391 | |
| 388 | 392 | article = {} |
| 389 | 393 | id = False |
| — | — | @@ -436,7 +440,8 @@ |
| 437 | 441 | if dataset == 'training': |
| 438 | 442 | function(article, cache, bots) |
| 439 | 443 | else: |
| 440 | | - counts = function(article, counts, bots) |
| | 444 | + #counts = function(article, counts, bots) |
| | 445 | + pass |
| 441 | 446 | buffer = cStringIO.StringIO() |
| 442 | 447 | |
| 443 | 448 | if i % 10000 == 0: |