"nb" -- New Books Watch 

requirements: unix shell, cron, standard text tools 

--------------------------------------------

'nb' is a Unix script that polls LibraryThing author pages and sends an 
email alert whenever a new work shows up that previously was not on 
LibraryThing. It is useful for keeping track of favorite authors who 
publish new books. The script uses LibraryThing as a backend database 
but you don't need an account there, it is accessed anonymously.

INSTALL and RUN

1. Copy nb.sh to its own directory and edit: change the parameters at 
   the top (self explanatory).

    NOTE: Be careful of line breaks (use an editor without word-wrap 
    such as vi or 'pico -w'). Check for paths on your system with "which 
    program"

   Set the script executable:

	   chmod 755 nb.sh

2. Create a list of authors in a file called nb.cfg (or any 
   name - the name is a command line argument). The file takes the 
   form:

      lastnamefirstname
      lastnamefirstname
      etc..

      Here's a sample cfg file with two authors:

        deli notify/nb> cat nb.cfg
        winchestersimon
        zolaemile
        deli notify/nb>

   These are the LibraryThing ID's as seen in the author page URL. 
   Generally it is simply lastnamefirstname (lowercase no-space) but 
   best to check the author's page on LibraryThing and look at the URL 
   for the author's ID. For example, the Author Page for Emile Zola has 
   a URL of:

     http://www.librarything.com/author/zolaemile&norefer=1

   thus the ID is "zolaemile" (the part after "/author/" and before "&")

3. Run it manually for a test. Example:

     [/home/bob/nb] ./nb nb.cfg

   If no email arrives, check for the file "emailbody.txt" - if it's 
   there, the problem is with the email system somewhere.

3. Add to cron. Here is a sample cron entry to run at 11:45pm

45 23 * * * /home/steve/nb/nb.sh /home/steve/nb/nb.cfg >> /dev/null 2>&1

   I run mine once a week, it keeps the noise down, new books don't come 
   out that often to need to run it that often.

FILES

   If you're curious how it works..

   For each author nb creates three file:

    a. authorname.new - the latest list of books 
    b. authorname.old - the previous list of books 
    c. authorname.htm - the raw HTML data from LibraryThing

   nb then compares the difference between .old and .new and emails 
   any works that are in .new but not .old

NOTES/BUGS

   a. Known bug: any book title with quotes (""), the part in quotes 
      won't show up. Thus if the full title is in quotes (rare) it will 
      be invisable. If only part of the title is in quotes, only that 
      part won't show up.
   b. It only shows additional works, not subtractions (works removed by 
      combination, deletion or canonical title change (see next note)).
   c. Work titles are determined by the "canonical title" under Common 
      Knowledge. If there is no canonical title, LT uses the most common 
      title. Thus, old works can show up as new if someone adds/changes 
      the canonical title in common knowledge.
   d. The script logs into LibraryThing anonymous. You don't need a LT 
      account.

WARNINGS/ERRORS

   Warnings and Errors can be turned on/off as one of the customize 
   variables at the top of the script (email_warnings, email_errors)

   Errors:

   a. "Bad HTML" - This could be caused by an inaccurate author ID in 
    nb.cfg, or if LibraryThing is down or network unreachable. Recommend 
    to always leave this on.

   Warnings:

   b. "Highload" - LibraryThing occasionally has high server load and 
    blocks anonymous users from doing certain resource intensive 
    actions. The script detects this and just skips doing anything. 

   c. "Tempdown" - LibraryThing is reporting it is temporarily down. nb 
    skips doing anything for that run.

   If there are too many warnings, change the time of day when the 
   script runs to a slow period when LT is not as busy, like early 
   morning.  

---
"New Books Watch" 
Stephen Balbach
stephen@balbach.net
http://bachlab.balbach.net/nb.html