RESUB(1l) MISC. REFERENCE MANUAL PAGES RESUB(1l) NAME resub - Prepare database for resubmission of text by data capture agency SYNOPSIS resub [-y] [-D database] -f file_or_key_list | file_or_key... DESCRIPTION Resub prepares the BNC database for resubmission of a text or texts by a data capture agency. Typically, it is neces- sary to do this when a text which has been bounced back to the data capture agency by OUCS is returned to OUCS after rework. Unless resub is run very soon after the replacement files are loaded, the database update procedures will view the replacement texts as duplicates of existing texts, and will reject them. Resub arranges that existing data per- taining to the files related to a text is archived, and that the record of a text's unique name is deleted. This done, the resubmitted text will be accepted by the database update procedures the next time they are run. (See the description of overnight in TGCW50.) A text may be specified in two ways: - Its 5- or 6-character mixed-case unique key may be given - for example, Kingdm ; or - The name of a file related to the text may be given. For example .Kingdm, books/A_Kingdm and $oup930421/.Kingdm are all acceptable. The two types of reference may be mixed, either on the com- mand line, or in a file named by the -f option. (But not both.) Because deleting database records is a dangerous operation, the user is asked to confirm the superseding of each text. Allowable responses are: y Confirm that text has been resubmitted n Do not mark text as resubmitted q Do not mark text as resubmitted; exit ! Confirm that text has been resubmitted; suppress further prompting. (Use with caution.) All prompting can be suppressed with the -y flag, which should be used with the greatest caution. Sun Release 4.1 Last change: TGCW49: 12 May, 1993 1 RESUB(1l) MISC. REFERENCE MANUAL PAGES RESUB(1l) If a text should be marked as resubmitted by mistake, it is possible to restore the deleted information to the database - see Dominic. It is allowable for a given text to be resubmitted more than once. The user is warned that this is the case on the second and subsequent resubmissions. TGCW48 describes file loading procedures. The procedure for loading resubmitted files is identical. In particular, resubmitted files, like original files, should be placed in new /corpus/Incoming/... and /corpus/Work/... directories named for the day on which they are received; they should not replace the original (and subsequently bounced) versions of the texts. Care is required to ensure that the BNC names of the resubmitted texts are exactly the same as those that they supersede. The clash utility (see TGCW48, step 13) may be useful in ensuring that names match. This done, resub may be run, for example as resub .?????{,?} in the newly-created /corpus/Work/... directory containing the resubmitted versions of the files. After resub has been run, all the superseded files related to a text may be deleted to save disk space if necessary: the database does not track further changes in their status. Alternatively, the files may be left on-line as they are, or in compressed (zipped) form. In any event, the original files in the corresponding As_received directory should be left in place just in case there is a need to refer back to them. OPTIONS -Ddatabase Use database instead of the default database, bnc. -ffile Names a file containing mixed case unique keys or filenames, one per line, corresponding to texts that have been resubmitted. -y Mark texts as resubmitted unconditionally: do not prompt the user. DIAGNOSTICS Warns on encountering keys which do not correspond to texts known to the database. Sun Release 4.1 Last change: TGCW49: 12 May, 1993 2 RESUB(1l) MISC. REFERENCE MANUAL PAGES RESUB(1l) Warns if a given text has already been superseded. Gives SQL errors if problems arise with data access. Aborts if unable to open an input channel for user response unless -y option in force. ENVIRONMENT II_SYSTEM Location of Ingres files. Defaults to /usr/local. FILES ~natcorp/bin/resub The program itself. The man page is embedded: hand the program to nroff -man. AUTHOR Dominic Dunlop SEE ALSO gzip(1), perl(1), Ingres/SQL Reference Manual, TGCW40, TGCW48, TGCW50 NOTES Resub should not be run if a text has simply been bounced: this will hide information about the files related to the text, and, in particular, to its bounced status. Run resub only when a resubmitted version of a text has been received. Resub does not touch the bibliographic information related to a text. If it should be necessary to supersede or amend this information, the changes must be made by hand. The word-counts for old versions of resubmitted texts con- tribute to the totals used for bounced texts and to gross totals of words submitted in the output of weeklyReport. (See TGCW40.) The status of a text which has been superseded by a resub- mitted version should not change. If it does - for example, if a B_ file appears in the directory of the a superseded A_ file - the overnight program notices and complains. BUGS When a resubmitted text is reinserted into the database by the overnight program, ``duplicate key on insert'' warnings are given when an attempt is made to rewrite bncSelClas and bncWork records which have not been deleted by resub because they contain useful bibliographic information about the text. These warnings can be ignored. Sun Release 4.1 Last change: TGCW49: 12 May, 1993 3