summaryrefslogtreecommitdiff
path: root/sa-extract/lcp_ops.py
diff options
context:
space:
mode:
authorChris Dyer <prguest11@taipan.cs>2012-02-02 06:29:50 +0000
committerChris Dyer <prguest11@taipan.cs>2012-02-02 06:29:50 +0000
commit8e5fad9bcbadf36bbab3c1c5b053e3c8f7dddbce (patch)
tree9c812b3f267aa1975cdf8b7af928c4b20eb36f93 /sa-extract/lcp_ops.py
parentff496d3089e84846c8562c574155d8df1e4d911c (diff)
lopez suffix array extractor with copyrighted david chiang code excised
Diffstat (limited to 'sa-extract/lcp_ops.py')
-rwxr-xr-xsa-extract/lcp_ops.py52
1 files changed, 52 insertions, 0 deletions
diff --git a/sa-extract/lcp_ops.py b/sa-extract/lcp_ops.py
new file mode 100755
index 00000000..9df6e82a
--- /dev/null
+++ b/sa-extract/lcp_ops.py
@@ -0,0 +1,52 @@
+#!/usr/bin/env python
+
+import lcp
+import sys
+import optparse
+import csuf
+
+'''Run various computations using the LCP array'''
+def main():
+
+ optparser = optparse.OptionParser()
+# optparser.add_option("-c", "--config", dest="config", help="configuration module")
+ optparser.add_option("-s", "--sa-check", dest="sa_check", default=False, action="store_true")
+ optparser.add_option("-l", "--lcp-check", dest="lcp_check", default=False, action="store_true")
+ optparser.add_option("-t", "--stats", dest="stats", default=0, type="int", action="store")
+ optparser.add_option("-u", "--unigram", dest="uni_check", default=False, action="store_true")
+ optparser.add_option("-r", "--report-long-lcps", dest="long_lcp", type="int", default=0, action="store")
+ (opts,args) = optparser.parse_args()
+
+ if len(args) < 1:
+ print >>sys.stderr, "Usage: lcp.py [opts] <sa file>"
+ sys.exit(1)
+
+ safile = args[0]
+ sa = csuf.SuffixArray(safile, from_binary=True)
+
+# if opts.sa_check:
+# check_sufarray(sa)
+
+ l = lcp.LCP(sa)
+
+ if opts.lcp_check:
+ print >>sys.stderr, "Checking LCP Array..."
+ l.check()
+ print >>sys.stderr, "Check finished"
+
+ if opts.stats > 0:
+ l.compute_stats(opts.stats)
+
+# if opts.uni_check:
+# if lcp is None:
+# lcp = LCP(sa)
+# unigram_stats(sa, lcp)
+#
+# if opts.long_lcp:
+# if lcp is None:
+# lcp = LCP(sa, opts.long_lcp)
+
+if __name__ == "__main__":
+ sys.exit(main())
+
+