7 files changed, 85 insertions, 4 deletions
diff --git a/pig/tests/files/example.sha1b32 b/pig/tests/files/example.sha1b32
new file mode 100644
index 0000000..20a1357
--- /dev/null
+++ b/pig/tests/files/example.sha1b32
@@ -0,0 +1,4 @@
+EJWYVOPONJRARK7SGG6COFRN7CSTHROY
+V32E3CCO7NMI2M4OHLKG73DXD72LR4B2
+3I42H3S6NNFQ2MSVX7XZKYAYSCX5QBYJ
+E3WSNQ7JAFOW7N3ZJ6GLV27T52T25JDK
diff --git a/pig/tests/files/sourcecode.cdx b/pig/tests/files/sourcecode.cdx
new file mode 100644
index 0000000..eeb397c
--- /dev/null
+++ b/pig/tests/files/sourcecode.cdx
@@ -0,0 +1,6 @@
+# match
+edu,cmu,cs,adm,reports-archive)/anon/usr0/ftp/usr0/anon/2002/cmu-cs-02-119.java 20170706005950 http://reports-archive.adm.cs.cmu.edu/anon/usr0/ftp/usr0/anon/2002/CMU-CS-02-119.java text/plain 200 MQHD36X5MNZPWFNMD5LFOYZSFGCHUN3V - - 361006 17120058 CITESEERX-CRAWL-2017-06-20-20170706004100259-00924-00932-wbgrp-svc284/CITESEERX-CRAWL-2017-06-20-20170706005946792-00926-31209~wbgrp-svc284.us.archive.org~8443.warc.gz
+# no
+fi,tkk,lib)/diss/2001/isbn951225459x/isbn951225459x.pyc 20170705074926 http://lib.tkk.fi/Diss/2001/isbn951225459X/isbn951225459X.pyc text/plain 200 KJBCOT7LGBNIAVGEGPUELK5OK6RTFORR - - 344175 255650124 CITESEERX-CRAWL-2017-06-20-20170705074433815-00129-00138-wbgrp-svc284/CITESEERX-CRAWL-2017-06-20-20170705074843696-00134-31209~wbgrp-svc284.us.archive.org~8443.warc.gz
+# no
+org,oxfordjournals,nar)/cgi/reprint/gkl1060v1.pdf 20170706035441 http://nar.oxfordjournals.org/cgi/reprint/gkl1060v1.pdf text/html 301 OX6MLVDFURLT2KSYCXUYW2PZNOVFSEVF - - 697 49346051 CITESEERX-CRAWL-2017-06-20-20170706034741172-00140-00149-wbgrp-svc285/CITESEERX-CRAWL-2017-06-20-20170706035435634-00148-3671~wbgrp-svc285.us.archive.org~8443.warc.gz
diff --git a/pig/tests/files/tarballs.cdx b/pig/tests/files/tarballs.cdx
new file mode 100644
index 0000000..7a81b79
--- /dev/null
+++ b/pig/tests/files/tarballs.cdx
@@ -0,0 +1,10 @@
+#http://research.fit.edu/sealevelriselibrary/documents/doc_mgr/448/Florida_Keys_Low_Island_Biodiversity_&_SLR_-_Ross_et_al_2009.pdf
+#http://ijs.sgmjournals.org:80/cgi/reprint/54/6/2217.pdf
+#http://eprints.ecs.soton.ac.uk/12020/1/mind-the-semantic-gap.pdf
+#http://eprint.uq.edu.au/archive/00004120/01/R103_Forrester_pp.pdf
+
+# should match 2:
+
+edu,fit,research)/sealevelriselibrary/documents/doc_mgr/448/Florida_Keys_Low_Island_Biodiversity_&_SLR_-_Ross_et_al_2009.pdf 20170706005950 http://mit.edu/file.pdf application/pdf 200 MQHD36X5MNZPWFNMD5LFOYZSFGCHUN3I - - 123 456 CRAWL/CRAWL.warc.gz
+edu,fit,research)/sealevelriselibrary/documents/doc_mgr/448/Florida_Keys_Low_Island_Biodiversity_&_SLR_-_Ross_et_al_2009.tar.gz 20170706005950 http://mit.edu/file.tar.gz application/octet-stream 200 NQHD36X5MNZPWFNMD5LFOYZSFGCHUN3I - - 123 456 CRAWL/CRAWL.warc.gz
+org,sgmjournals,ijs)//cgi/reprint/54/6/2217.tar.gz 20170706005950 http://mit.edu/file.tar.gz application/gzip 200 TQHD36X5MNZPWFNMD5LFOYZSFGCHUN3V - - 123 456 CRAWL/CRAWL.warc.gz
diff --git a/pig/tests/pighelper.py b/pig/tests/pighelper.py
index 4aa4259..95e0426 100644
--- a/pig/tests/pighelper.py
+++ b/pig/tests/pighelper.py
@@ -17,6 +17,9 @@ import unittest
 import subprocess
 
 
+def count_lines(s):
+    return len([l for l in s.strip().split('\n') if len(l) > 0])
+
 class PigTestHelper(unittest.TestCase):
 
     @classmethod
@@ -50,7 +53,7 @@ class PigTestHelper(unittest.TestCase):
         return retval
 
     def run_pig(self, script_path, in_file, **kwargs):
-        """Convenience helper around run_pig().
+        """Convenience helper around run_pig_raw().
         
         INPUT parameter is set to in_file.
         OUTPUT parameter is set to a random file.
diff --git a/pig/tests/test_filter_cdx_paper_pdfs.py b/pig/tests/test_filter_cdx_paper_pdfs.py
index a8ebd9f..c2d2e6b 100644
--- a/pig/tests/test_filter_cdx_paper_pdfs.py
+++ b/pig/tests/test_filter_cdx_paper_pdfs.py
@@ -1,10 +1,8 @@
 
 import os
 import unittest
-from pighelper import PigTestHelper
+from pighelper import PigTestHelper, count_lines
 
-def count_lines(s):
-    return len([l for l in s.strip().split('\n') if len(l) > 0])
 
 class TestFilterCDXPaperPdfs(PigTestHelper):
 
diff --git a/pig/tests/test_filter_software.py b/pig/tests/test_filter_software.py
new file mode 100644
index 0000000..cce90b4
--- /dev/null
+++ b/pig/tests/test_filter_software.py
@@ -0,0 +1,16 @@
+
+import os
+import unittest
+from pighelper import PigTestHelper, count_lines
+
+
+class TestFilterCDXSoftware(PigTestHelper):
+
+    def test_tarballs(self):
+        r = self.run_pig("filter-cdx-tarball.pig", "tests/files/tarballs.cdx")
+        assert count_lines(r) == 2
+
+    def test_source_code(self):
+        r = self.run_pig("filter-cdx-source-code-crude.pig", "tests/files/sourcecode.cdx")
+        assert count_lines(r) == 1
+
diff --git a/pig/tests/test_join_cdx.py b/pig/tests/test_join_cdx.py
new file mode 100644
index 0000000..e6eca6a
--- /dev/null
+++ b/pig/tests/test_join_cdx.py
@@ -0,0 +1,44 @@
+
+import os
+import unittest
+import tempfile
+import subprocess
+from pighelper import PigTestHelper, count_lines
+
+class TestJoinCDXSha1(PigTestHelper):
+
+    def run_pig_join(self, script_path, cdx_file, digest_file, **kwargs):
+        """Convenience helper around run_pig().
+        
+        INPUT parameter is set to in_file.
+        OUTPUT parameter is set to a random file.
+        Any keyword args are passed as parameters.
+        """
+
+        pargs = []
+        for key, value in kwargs.items():
+            pargs.append('-p')
+            pargs.append('{}={}'.format(key, value))
+
+        out_file = tempfile.mktemp(dir=self._tmpdir)
+        params = [
+            '-f', script_path,
+            '-p', 'INPUT_CDX={}'.format(cdx_file),
+            '-p', 'INPUT_DIGEST={}'.format(digest_file),
+            '-p', 'OUTPUT={}'.format(out_file),
+            ] + pargs
+        status = self.run_pig_raw(params)
+        assert status.returncode == 0
+        # Capture all the part-r-* files
+        print("out_file: {}".format(out_file))
+        subprocess.run("/bin/ls -la {}/part-*".format(out_file), shell=True)
+        sub = subprocess.run("/bin/cat {}/part-*".format(out_file), stdout=subprocess.PIPE, shell=True)
+        out = sub.stdout.decode('utf-8')
+        print(out)
+        return out
+
+    # TODO: helper to verify that output matches an expected file
+
+    def test_thing(self):
+        r = self.run_pig_join("join-cdx-sha1.pig", "tests/files/example.cdx", "tests/files/example.sha1b32")
+        assert count_lines(r) == 4