HowTo: Create a simple Document Management Application Part 2
Novell Cool Solutions: Feature
By Stomfi
Reader Rating 
|
Digg This -
Slashdot This
Posted: 28 Jul 2005 |
Learning to use Linux at Home and WorkThis HowTo shows you how to create a simple Document Management System using the Shell and Runtime Revolution.
I have used some new shell tools, and some new RunRev functions in this project, so that you can increase your knowledge.
 |
|
|
Special Offer for Cool Solutions Readers: Free Copy of Runtime Revolution The good folks at Runtime Revolution have extended a special offer to Cool Solutions readers to make it easier for you to implement the great ideas in Stomfi's articles. They are offering it for free to Novell customers who know the secret code. See this page for details. |
|
|
The new shell tools are some of the bibliographic database tools, namely "indxbib" and "lkbib". These are part of the "refer" suite. Refer was designed to add troff/nroff/groff formatted references to a document. We shall be using them for our own purposes, using a subset of the possible database entries. You can look up the manual pages for "refer" to see all the possible entries. The reason we are using these tools is that they are suitable, and who wants to waste time reinventing them.
I've used a new RunRev function in this project for making and selecting items from a menu, and I've used one of the windows for two activities, both adding and editing, by changing the property of the display field, and adding a flag to tell which mode we are using. I've also set up a new construct that passes a flag to the program about which calling screen to return to. This is really simple RunRev scripting as I do all the information processing with the Linux shell tools, which were designed for that purpose.
The RunRev windows and the first of the scripts were developed in Part one of this HowTo.
This is a picture of the ADDDOC window where you can add details for a new document or edit those of an existing one.
This picture shows that the Title of the record for the `SCOOL document details has been changed.
These are the rest of the scripts that allow this application to perform its activities:
ADDDOC card script
on openCard
global TITL
global AUTH
global NDATE
global PUBL
global PUBLC
global EZINE
global EZINES
global DBFILE
global EDITME
if EDITME <> empty
then
set the listBehavior of field "DOCUMENT" to false
put ($HOME & "/docmanage/bin/looked.sh" && quote & EDITME & quote) into ERECORD
replace return with empty in ERECORD
put the shell of ERECORD into EDOC
#put EDOC into field "DEBUG"
put line 7 of EDOC into TITL
put line 1 of EDOC into AUTH
put line 4 of EDOC into NDATE
put line 2 of EDOC into PUBL
put line 3 of EDOC into PUBLC
put line 6 of EDOC into EZINE
put line 5 of EDOC into EZINES
put line 9 of EDOC into DBFILE
put line 4 of EDOC & return into field "DOCUMENT"
put line 1 of EDOC & return after field "DOCUMENT"
put line 7 of EDOC & return after field "DOCUMENT"
put line 2 of EDOC & return after field "DOCUMENT"
put line 3 of EDOC & return after field "DOCUMENT"
put line 6 of EDOC & return after field "DOCUMENT"
put line 5 of EDOC & return after field "DOCUMENT"
put line 9 of EDOC into field "BIBFILE"
else
put empty into TITL
put empty into AUTH
put empty into NDATE
put empty into PUBL
put empty into PUBLC
put empty into EZINE
put empty into EZINES
put empty into field "BIBFILE"
put empty into field "DOCUMENT"
put ("First Choose File from this list or" & return & "click empty line for New Database") into field "DOCUMENT"
put (return & "***********************" & return) after field "DOCUMENT"
put ( $HOME & "/docmanage/dict/papers") into DOCDBDIR
put ("ls" && DOCDBDIR & "| grep -v" && quote & "\.i" & quote ) into FLIST
replace return with empty in FLIST
put the shell of FLIST after field "DOCUMENT"
set the listBehavior of field "DOCUMENT" to true
end if
end openCard
This is our double purpose window. The if condition looks at EDITME and if it finds something in there, sets the list behaviour and uses it to find the record using the following shell script.
#!/bin/bash
#
#looked.sh SearchTerms
#Returns fields for editing function
#Set Database Folder
DBFLDR="$HOME/docmanage/dict/papers/"
#Change folder to make command line less wordy
cd $DBFLDR
#Make a list of Databases
DBLIST=`ls | grep -v "\.i"`
#
#Strip the leading Title: string from $1
LKEY=`echo $1 | cut -d":" -f2-`
#
#Create command line DB option list
DBOPTIONS=""
BIBDBF=""
for DBF in $DBLIST
do
#First make sure that the indexes are up to date
indxbib $DBF
#Construct the options part of the line
DBOPTIONS="$DBOPTIONS -p $DBF"
#Find out if this is the BIBDBF file
if [ "${#BIBDBF}" -lt 1 ]
then
BIBDBF=`grep -iF "$LKEY" $DBF`
if [ "${#BIBDBF}" -gt 1 ]
then
DBFILE=$DBF
fi
fi
done
#
#list matching record
lkbib $DBOPTIONS "$LKEY" |\
sed -f $HOME/docmanage/bin/noheadings.sed > $HOME/docmanage/erec.txt
#echo the BIBDBF file
echo $DBFILE >> $HOME/docmanage/erec.txt
cat $HOME/docmanage/erec.txt
Much of this script is similar to the lookup.sh script. We find out which database file contains the Title string and use that for the lkbib option. We list the record, use the sed script noheadings.sed to strip off the % references and send the result to a temporary file. We echo the database name to the end of that file and send the result back to RunRev. This is the sed file.
s/%A// s/%B// s/%C// s/%D// s/%K// s/%S// s/%T//
You can see it substitutes the first pattern with nothing.
The else statement for the ADDDOC card script is for adding a new record when the ADD menu is clicked from the START button. It sets all the name holders to empty, puts a message into the DOCUMENT display field, calls a shell function to put the names of the existing databases after the message, and sets the list behaviour to "true" so you can click in it to perform the required action.
DOCUMENT field script.
on mouseDown
global DBFILE
put the selectedText of me into DBFILE
if DBFILE <> empty
then
put DBFILE into field "BIBFILE"
else
ask "Please Enter a New or Old File Name to save your record"
put it into DBFILE
put DBFILE into field "BIBFILE"
end if
put empty into field "DOCUMENT"
set the listBehavior of field "DOCUMENT" to false
end mouseDown
This script only works when the list behaviour is set to "true" helping the ADDDOC window perform its double function.
Year button
on mouseUp ask "Year" global NDATE put it into NDATE put "DATE:" && NDATE into line 1 of field "DOCUMENT" end mouseUp
All the document details buttons have similar scripts. You can work out the name holders from the next script and the line number to put it into from the order of the buttons on the window.
EXIT button script
on mouseUp
global TITL
global AUTH
global NDATE
global PUBL
global PUBLC
global EZINE
global EZINES
global DBFILE
global EDITME
put ($HOME & "/docmanage/dict/papers/" & DBFILE) into BIBDBFILE
replace return with empty in BIBDBFILE
if TITL <> empty and AUTH <> empty and NDATE <> empty
then
put(quote & NDATE & "#" & TITL & "#" & AUTH & "#" & PUBL & "#" & PUBLC & "#" & EZINE & "#" & EZINES & quote) into NEWBIB
if EDITME <> empty
then
put ($HOME & "/docmanage/bin/editbib.sh" && BIBDBFILE && NEWBIB) into DUREC
else
put ($HOME & "/docmanage/bin/writebib.sh" && BIBDBFILE && NEWBIB) into DUREC
end if
replace return with empty in DUREC
put the shell of DUREC into DUNREC
end if
put empty into EDITME
global WITCH
put the name of this card into CNAME
if CNAME contains WITCH
then
put "DocManage" into WITCH
end if
go card WITCH
end mouseUp
This script first creates the database path name. It checks to see if enough fields to create a record are filled. If there is, it creates a pattern containing all the field names separated by the # symbol. It checks to see if this is an edit of an existing record or a new record, and sets the relevant shell script. It clears the edit flag, and calls the shell script. Lastly, whether it edited, created or did nothing - it returns to the calling window.
These are the two shell scripts:
#!/bin/bash
#
#editbib.sh BIB Record
#Add records to specified BIBDB
BIBDB="$1"
#These are the record entries in $2
#PDATE=%D $1 AUTH=%A $2 TITL=%T $3 PUBL=%B $4 PUBLC=%C $5 EZINE=%S $6 ESECT=%K $7
#
#Create BIB entry using sed to get rid of the initial space. The back slash escapes the literal meaning of the space.
echo "$2" |\
awk -F"#" '{print "\n","%A "$2"\n","%B "$4"\n","%C "$5"\n","%D "$1"\n","%K "$7"\n","%S "$6"\n","%T "$3"\n","\n"}'|\
sed -e s/^\ // > $HOME/docmanage/newent.txt
#
#Find out which record number contains %T $3
TITL=`echo $2 | cut -d"#" -f3`
RKEY="%T $TITL"
NUMR=`awk -v KEYR="$RKEY" 'BEGIN{RS = ""};{if($0 ~ KEYR) print NR }' $BIBDB`
#
#remove temporary file
rm -f $HOME/docmanage/newfile.txt
#
#Use the awk getline function to replace the record
awk -v NEWREC="$HOME/docmanage/newent.txt" -v RNUM=$NUMR '\
BEGIN{RS = "";ORS = "\n\n";OFS= "\n"};\
{if(NR != RNUM) {print $0}\
else {getline < NEWREC; print $0}}' $BIBDB > $HOME/docmanage/newfile.txt
#
#If success replace original database
if [ ! -z $HOME/docmanage/newfile.txt ]
then
rm -f $BIBDB
mv $HOME/docmanage/newfile.txt $BIBDB
echo "DUN"
#Update BIB index
indxbib $BIBDB
fi
The editbib.sh script uses the awk getline function to replace the record equal to the record number of the identified record, printing out all other lines. We have used the get line function before in the Home Cookbook articles, so you should be familiar with its use. To make sure that the look up functions work in other parts of the application, the "indxbib" shell tool is used after any changes or additions to the database.
#!/bin/bash
#
#addbib.sh BIB Record
#Add records to specified BIBDB
BIBDB="$1"
#Create BIBDB if not exist
if [ ! -e "$BIBDB" ]
then
touch $BIBDB
fi
#These are the record entries in $2
#PDATE=%D $1 AUTH=%A $2 TITL=%T $3 PUBL=%B $4 PUBLC=%C $5 EZINE=%S $6 ESECT=%K $7
#
#Create BIB entry
echo "$2" |\
awk -F"#" '{print " \n","%A "$2"\n","%B "$4"\n","%C "$5"\n","%D "$1"\n","%K "$7"\n","%S "$6"\n","%T "$3"\n"}' >> "$1"
echo "DUN"
#Update BIB index
indxbib $BIBDB
This script is pretty straight forward as its function craete a new database if it doesn't exist and to write a new record on the end of the database file, making sure that there is at least one newline separating each record. This is done by the "\n" at each end of the awk print function.
Card Help Button exit
on mouseUp
global WITCH
put the name of this card into CNAME
if CNAME contains WITCH
then
put "DocManage" into WITCH
end if
go card WITCH
end mouseUp
This is the help button exit script. I haven't been window specific with the help, but if you want to do this then you can create a text file for each window, copy this same button to any that are missing it, and change the HELP button script adding these extra lines to each window. This example is for the START window.
on mouseUp global WITCH put "DocManage" into WITCH #Extra lines for specific window help put ($HOME &"/docmanage/starthelp.txt") into HELPDOC put empty into field "HELPTXT" of card "HELP" put HELPDOC into field "HELPTXT" of card "HELP" #End of extra lines go card "HELP" end mouseUp
You can copy this script from the "DocManage" card, paste it into all the other Help button scripts, and change "starthelp.txt" to the name of the relevant help file for that window, and change "DocManage" into the name of the card the HELP button is on. Or you can copy and past just the extra lines changing the name of the help file.
This is the end of this series. It is nice and simple allowing you to add extra features. You can use the other fields described in the refer manual page for what ever extra details you want. For example, you could add a page reference for magazine or journal articles, or you could add a web page URL for web articles, or products and services your business may be interested in, adding extra buttons on the ADDDOC window, and modifying the scripts accordingly.
You can separate your records into many databases, and if you want to store them in a logical place in your file system, put symbolic links to them into the $HOME/docmanage/dict/papers folder.
In a networked environment, you could distribute the Stand Alone to other people and set the database links folder in the application to a readable folder on the server. This way you can access similar databases with different detail fields across a network. Your network administrator may have to set the links up for you and your fellow workers.
There are plenty of professional tools that do similar things, but this works, you made it yourself, you or anyone with basic Linux skills can modify it to suit specific needs, it's not bloated with features you will never need, and it didn't cost any money. What could be better than all of that. Go Linux Newbie!
Novell Cool Solutions (corporate web communities) are produced by WebWise Solutions. www.webwiseone.com