[open-ils-commits] r620 - servres/trunk/conifer/libsystems/z3950 (artunit)

[svn at svn.open-ils.org]

Author: artunit
Date: 2009-08-17 01:05:04 -0400 (Mon, 17 Aug 2009)
New Revision: 620

Added:
   servres/trunk/conifer/libsystems/z3950/pyz3950_search.py
Log:
whoops, add missing z3950 setup

Added: servres/trunk/conifer/libsystems/z3950/pyz3950_search.py
===================================================================
--- servres/trunk/conifer/libsystems/z3950/pyz3950_search.py	                        (rev 0)
+++ servres/trunk/conifer/libsystems/z3950/pyz3950_search.py	2009-08-17 05:05:04 UTC (rev 620)
@@ -0,0 +1,101 @@
+# z39.50 search using yaz-client. 
+# dependencies: yaz-client, pexpect
+
+# I found that pyz3950.zoom seemed wonky when testing against conifer
+# z3950, so I whipped up this expect-based version instead.
+
+import warnings
+import re
+import sys
+from marcxml import marcxml_to_dictionary
+
+try:
+
+    import profile
+    import lex
+    import yacc
+except ImportError:
+
+    sys.modules['profile'] = sys # just get something called 'profile';
+                                     # it's not actually used.
+    import ply.lex
+    import ply.yacc             # pyz3950 thinks these are toplevel modules.
+    sys.modules['lex'] = ply.lex
+    sys.modules['yacc'] = ply.yacc
+
+# for Z39.50 support, not sure whether this is the way to go yet but
+# as generic as it gets
+from PyZ3950 import zoom, zmarc
+
+
+LOG = None              #  for pexpect debugging, try LOG = sys.stderr
+GENERAL_TIMEOUT = 40
+PRESENT_TIMEOUT = 60
+
+def search(host, port, database, query, start=1, limit=10):
+
+
+    query = query.encode('utf-8') # is this okay? Is it enough??
+
+    conn = zoom.Connection(host, port)
+    conn.databaseName = database
+    conn.preferredRecordSyntax = 'XML'
+    
+    query = zoom.Query ('CCL', str(query))
+    res = conn.search (query)
+    collector = []
+    #if we were dealing with marc8 results, would probably need this
+    #m = zmarc.MARC8_to_Unicode ()
+
+    # how many to present? At most 10 for now.
+    to_show = min(len(res)-(start - 1), limit)
+    if limit:
+        to_show = min(to_show, limit)
+
+
+    #this seems to an efficient way of snagging the records
+    #would be good to cache the result set for iterative display
+    for r in range(start - 1,(start-1) + to_show):
+        #would need to translate marc8 records, evergreen doesn't need this
+        #collector.append(m.translate(r.data))
+        collector.append(str(res.__getitem__(r)).replace('\n',''))
+    conn.close ()
+
+
+    raw = "" . join(collector)
+
+    raw_records = []
+    err = None
+
+    pat = re.compile('<record .*?</record>', re.M)
+    raw_records = pat.findall(raw)
+
+    parsed = []
+    for rec in raw_records:
+        try:
+            rec = _marc_utf8_pattern.sub(_decode_marc_utf8, rec)
+            dct = marcxml_to_dictionary(rec)
+        except 'x':
+            raise rec
+        parsed.append(dct)
+    return parsed, len(res)
+
+
+# decoding MARC \X.. UTF-8 patterns.
+
+_marc_utf8_pattern = re.compile(r'\\X([0-9A-F]{2})')
+
+def _decode_marc_utf8(regex_match):
+    return chr(int(regex_match.group(1), 16))
+
+
+#------------------------------------------------------------
+# some tests
+
+if __name__ == '__main__':
+    tests = [
+        ('zed.concat.ca:210', 'OSUL', 'chanson'),
+        ]
+    for host, db, query in tests:
+        print (host, db, query)
+        print len(search(host, db, query, limit=33))