Adding JMCR-Stable version

[Benchmarks_CSolver.git] / JMCR-Stable / real-world application / weblech / Spider.properties

diff --git a/JMCR-Stable/real-world application/weblech/Spider.properties b/JMCR-Stable/real-world application/weblech/Spider.properties
new file mode 100644 (file)
index 0000000..9c8b41a
--- /dev/null
+++ b/JMCR-Stable/real-world application/weblech/Spider.properties
@@ -0,0 +1,61 @@
+# Spider configuration file\r
+#\r
+# All of these settings default to sensible values if not specified.\r
+\r
+# Directory in which to save downloaded files, defaults to "."#c:/weblech/sites\r
+saveRootDirectory = download\r
+\r
+# Filename in which to save mailto links\r
+mailtoLogFile = mailto.txt\r
+\r
+# Tell the spider to reload HTML pages each time, but not images\r
+# or other files\r
+refreshHTMLs = true\r
+refreshImages = false\r
+refreshOthers = false\r
+\r
+# Set the extensions the Spider should use to determine which\r
+# pages are of MIME type text/html. The Spider also learns new\r
+# types as it downloads them.\r
+htmlExtensions = htm,html,shtm,shtml\r
+\r
+# Similarly for MIME type image/*\r
+imageExtensions = gif,jpg,jpeg,png,bmp\r
+\r
+# URL at which we should start the spider\r
+startLocation = http://www.google.com/\r
+\r
+# Whether to do depth first search, or the default breadth\r
+# first search when finding URLs to download\r
+depthFirst = false\r
+\r
+# Maximum depth of pages to retrieve (the first page is depth\r
+# 0, links from there depth 1, etc). Setting to 0 is "unlimited"\r
+maxDepth = 2\r
+\r
+# Basic URL filtering. URLs must contain this string in order\r
+# to be downloaded by WebLech\r
+urlMatch = slashdot.org\r
+\r
+# Basic URL prioritisation. URLs which are "interesting" are\r
+# downloaded first, URLs which are "boring" last.\r
+interestingURLs=pollBooth.pl,faq\r
+boringURLs=article.pl\r
+\r
+# User Agent header\r
+userAgent = Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 5.0)\r
+\r
+# Username and password for basic HTTP authentication, if required.\r
+# The same username and password will be used for all authentication\r
+# challenges during a download session.\r
+basicAuthUser = myUser\r
+basicAuthPassword = 1234\r
+\r
+# Number of download threads to start\r
+spiderThreads = 2\r
+\r
+# How often to checkpoint the Spider. A checkpoint file is named\r
+# "spider.checkpoint" and can be used to start the spider in the\r
+# middle of a run. Setting this value to 0 disables checkpoints.\r
+# Here we checkpoint every 30 seconds\r
+checkpointInterval = 30000
\ No newline at end of file