Skip to content


Subversion checkout URL

You can clone with
Download ZIP


refextract: improve argument check #487

tiborsimko opened this Issue · 4 comments

2 participants


Originally on 2011-02-10

When you run refextract on a non-existent file, it tracebacks with:

UnboundLocalError: local variable 'count_auth_group' referenced before assignment

It should check the parameters properly and report the concrete problem, e.g. file non-existent, permission denied, etc.


Originally by chayward on 2011-02-14

on branch: refextract-countauth-fix

paired with support for arxiv file names


Originally on 2011-02-15

Running refextract on a non-existent file does not report an error for
me; it produces empty output:

/opt/invenio/bin/refextract 1:/tmp/nonexistentfile.pdf
<?xml version="1.0" encoding="UTF-8"?>
<collection xmlns="">
   <controlfield tag="001">1</controlfield>
   <datafield tag="999" ind1="C" ind2="6">
      <subfield code="a">Invenio/1.0.0-rc0.40-106ba refextract/1.0.0-rc0.40-106ba-1297762113-1-0-0-0-0-0-0</subfield>

It should print an error message and exit with some error code
so that clients can know something went wrong. (e.g. workflow can
halt etc)


Originally by chayward on 2011-02-15

on branch: refextract-countauth-fix


Originally by Christopher Hayward on 2011-02-16

In [3680959]:

#CommitTicketReference repository="" revision="3680959b3640bbe54a013e9c3bf50af0ff7a29bf"
refextract: fix unassigned count_auth_group

- Initialised count_auth_group in main.
  (closes #487)

- Updated usage information with inspire journal title output option.

- When obtaining the recid and the file location for each extraction job,
  split a maximum of one time, since file names with arvix notation
  regularly include a colon e.g. 'arxiv:9834.pdf'.

- Display an error message, regardless of the output mode, and halt
  extraction when either a non-existent or an unreadable pdf/text input
  file is provided as input.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Something went wrong with that request. Please try again.