University of Calgary
Library - Information Technology Services
Exporting UCALGARY MARC records for National Library and OCLC...L.Pearce's instructions.
Approximately 3 times a year we are supposed to report all our new cataloging
records to both the National Library and OCLC. For the National Library we must report
medical and non-medical records separately, and once this is done the two files
are concatenated together for reporting as one file to OCLC.
The files are sent via ftp to both sites.
A paper record is kept on orange paper on Linda Pearce's bulletin board of the
numbers of records sent and the dates they were ftp'd across.
Extracting MARC records
In the shell, in ...Unicorn/Xfer, rename dumpkeys and dumpdate if they exist.
This is because they will be overlaid, and you may have needed the old files for something.
In Unicorn Reports, in the Bib group, Create a Prepmarcexp report.
Choose itemcat2 / Do NOT contain / MEDICAL
Choose #Copies > 0
Enter date cataloged range (refer to orange chart for dates)
Schedule and run the report ASAP. When this runs, it creates new dumpdate and
dumpkeys files in ...Unicorn/Xfer.
When the job completes, view it and write down on the orange chart the number of
records extracted.
Record KEYS have been extracted but MARC records must still be created.
check to see that the records in MARA.Dyymmdd look OK
To see records as flat ascii, cat filename | marcprint | more
WARNING: NEVER do a tail on the MARC file
Sending the records to the National Library
login to the production server as the sirsi user
cd Unicorn/Xfer
ftp mara-ftp.nlc-bnc.ca
username: acu_ft
password: acuftp
binary
put MARA.Dyymmdd
Finish procedures for non-medical records
Fill in the orange chart
Procedures to send Medical records
The medical records must be done the NEXT day because the filename does not
distinguish between medical and non-medical records, so they must be distinguished by date.
Do all the procedures above EXCEPT in itemcat2 choose CONTAIN MEDICAL
To send the medical records to National Library, login as acum_ft and use password
acumftp
Sending files to OCLC
After producing the output files for both medical and nonmedical records for N.L.,
make a backup copy of both files in the shell.
cat MARAMED.mmddbak >> MARANONMED.yymmddbak
This concatenates the medical records to the end of the nonmedical records
change the filename to DATA.Dyymmdd
check the file sizes. The new DATA file should be equal to the sum of the old
nonmed and medical files combined.
DO NOT DO A 'tail' on the MARC file! Bad things happen in unix....
Creating the LABEL file for OCLC
OCLC demands both the data file itself and also a LABEL file that describes the
contents of the DATA file.
Copy the previous LABEL file with a new date on the end. Change the number of
records sent (the second line of the file after RBF) and the date on the end of the
UAUD.D line
eg of LABEL file: LABEL.D020326
DAT 20001103123456.0
RBF 18426
DSN UAUD.D020326
ORS UAU
FOR M
FDI P123456
Sending records to OCLC
Send the DATA file first and the LABEL file second.
login to the Sirsi production server as the sirsi user
ftp to EDX.OCLC.ORG
login: TUAU1
password: this expires EVERY time so you always have to type the oldpw/newpw/newpw
unless you just used it in the past few days. Password is attached to the orange chart.
cd 'edx.ebsb.uau.ftp'
binary
put filename
check OCLC's website for confirmation that they received the files
an email should also be received from them saying they got the file. This doesn't
always happen!
Other info
After all records have been sent perfectly, erase the files sent 3 months previously,
and erase the prepmarcexp reports from Unicorn as well.
Contact person at National Library is Michael Andrassy: michael.andrassy@nlc-bnc.ca
Contact person at OCLC is Manon Barbeau: barbeaum@oclc.org
Exporting MARC records for AMBROSE to OCLC
This process was begun for the first time March 26, 2010 and extracted all of the AMBROSE non-shadow batch created/never catalogued records and AMBROSE non-shadow date created before Jan 1, 2010 records.
Extract the Marc records via report template Ambrose_extract1 (just be careful to adjust the date catalogued to reflect the records you want to extract).
After it runs, go to the shell and in ...Unicorn/Xfer, rename the dumpkeys and dumpdate to something like ambrosemar262010a.keys and ambrosemar262010.dumpdate
Now extract the next batch of marc record via report template Ambrose_extract_forNEVERBATCH.
After it runs, again go into the shell and rename the dumpkeys and dumpdate since these two files get overlaid each time prepmarcexp report is run.
The reason to always run the batch loaded/never catalogued extraction is because Ambrose can add copies at anytime to our old loaded records so there is no way to easily detect those cases, it is just easier to do a full Ambrose database extract each time for OCLC.
Now create the MARC records from the keys extracted via:
In the shell, in Unicorn/Xfer take the file of renamed keys and do:
See that the records look okay by doing: cat DATA.D100326.FILE1 | marcprint | more use CNTL-c to escape the view whenever you want. Or use: cat DATA.D100326.FILE1 | flatskip -aMARC -im -of -w5 (use 5 or whatever number of records you want to view).
OCLC stated that each DATA file should not exceed 90,000 records and you should contact them if you plan on sending more than 90,000 records per day. So, I split the records into 2 files of 80,329 records and 15,471 records for this first March 26, 2010 extract.
create a label file to go with each of these files for OCLC
copy the previous LABEL file with a new date on the end.
Change the number of records sent (the second line of the file after RBF prefix).
Change the date on the DAT (line one)
Change the date in the DSN line (filename)
Example of an Ambrose label file:
$ more LABEL.D100326.FILE1
DAT 20100326000000.0
RBF 15471
DSN DATA.D100326.FILE1
ORS CNTAB
FDI P011108
Send the records to OCLC via ftp in the shell in the Unicorn/Xfer directory
ftp edx.oclc.org
Type logon/username of: TCNTAB1
Type password your password* oldpasswd/newpasswd/newpasswd (where your password should be 8 characters long with one internal number but no punctuation)
the password is in my password word document
cd 'EDX.EBSB.CNTAB.FTP'
passive
bin
put DATA.D100326.FILE1
put LABEL.D100326.FILE1
Customer support at OCLC is: 1-800-848-5800
Exporting MARC records for AMBROSE to National Library
Using the same records extracted for OCLC, merge the files if more than one and then split as necessary. NLC wants files to contain 40,000-60,000 records only. For this first data set submitted for Ambrose, there were 95,800 records extracted March 26, 2010 so I split it into two files of 47,900 records.
I have these files saved in Unicorn/Xfer/Ambrose
I named the files MARA_AMB.100531a and MARA_AMB.100531b in a similar naming fashion to that outlined above.
The records were ftp'd to NLC on Jun 1, 2010 and named for the day I created the files although the records themselves were extracted March 26.
Details on filenaming etc can be found on the NLC site
login to the production server as the sirsi user
cd Unicorn/Xfer/Ambrose
ftp mara-ftp.nlc-bnc.ca
username: acau1
password: acau8
binary
put MARA_AMB.100531a
put MARA_AMB.100531b
ls (and ensure files are there)
bye
NOTE: do NOT use PASSIVE ftp mode -- it does not work for this site.