From eacebcd6dd2ae1a7ff545ea1375c61760a0b829e Mon Sep 17 00:00:00 2001 From: Kovid Goyal Date: Tue, 29 Jan 2008 03:57:08 +0000 Subject: [PATCH] Warn when server returns a zero-length document --- src/libprs500/web/fetch/simple.py | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/src/libprs500/web/fetch/simple.py b/src/libprs500/web/fetch/simple.py index 6d8ee191b5..2d5c85ab45 100644 --- a/src/libprs500/web/fetch/simple.py +++ b/src/libprs500/web/fetch/simple.py @@ -259,6 +259,9 @@ class RecursiveFetcher(object): try: self.current_dir = linkdiskpath f = self.fetch_url(iurl) + dsrc = f.read() + if len(dsrc) == 0: + raise Exception('No content') soup = self.get_soup(f.read()) self.logger.debug('Processing images...') self.process_images(soup, f.geturl()) @@ -272,7 +275,7 @@ class RecursiveFetcher(object): self.process_links(soup, iurl, recursion_level+1) else: self.process_return_links(soup, iurl) - self.logger.debug('Recursion limit reached. Skipping %s', iurl) + self.logger.debug('Recursion limit reached. Skipping links in %s', iurl) save_soup(soup, res) self.localize_link(tag, 'href', res)