NAME¶
AutoSearch -- a web-search tracking application
SYNOPSIS¶
AutoSearch [--stats] [--verbose] -n "Query Name" -s "query
string" --engine engine [--mail you@where.com] [--options
"opt=val"]... [--filter "filter"] [--host host] [--port
port] [--userid bbunny --password c4rr0t5] [--ignore_channels KABC,KCBS,KNBC]
qid
AutoSearch --VERSION AutoSearch --help AutoSearch --man
DESCRIPTION¶
AutoSearch performs a web-based search and puts the results set in
qid/index.html. Subsequent searches (i.e., the second form above)
AutoSearch determine what changes (if any) occured to the results sent
since the last run. These incremental changes are recorded in
qid/YYYYMMDD.html.
AutoSearch is amenable to be run as a
cron job because all the
input parameters are saved in the web pages.
AutoSearch can act as a
automated query agent for a particular search. The output files are designed
to be a set of web pages to easily display the results set with a web browser.
Example:
AutoSearch -n 'LSAM Replication'
-s '"lsam replication"'
-e AltaVista
replication_query
This query (which should be all on one line) creates a directory
replication_query and fills it with the fascinating output of the AltaVista
query on "lsam replication", with pages titled ``LSAM Replication''.
(Note the quoting: the single quotes in '"lsam replication"' are for
the shell, the double quotes are for AltaVista to search for the phrase rather
than the separate words.)
A more complicated example:
AutoSearch -n 'External Links to LSAM'
-s '(link:www.isi.edu/lsam or link:www.isi.edu/~lsam) -url:isi.edu'
-e AltaVista::AdvancedWeb
-o coolness=hot
This query does an advanced AltaVista search and specifies the (hypothetical)
``coolness'' option to the search engine.
OPTIONS¶
- "qid"
- The query identifer specifies the directory in which all the files
that relate to this query and search results will live. It can be an
absolute path, or a relative path from cwd. If the directory does not
exist, it will be created and a new search started.
- "--stats"
- Show search statistics: the query string, number of hits, number of
filtered hits, filter string, number of suspended (deleted) hits, previous
set size, current set size, etc.
- "-v" or "--verbose"
- Verbose: output additional messages and warnings.
- "-n" or "--qn" or "--queryname"
- Specify the query name. The query name is used as a heading in the web
pages, therefore it should be a 'nice' looking version of the query
string.
- "-s" or "--qs" or "--querystring"
- Specify the query string. The query string is the character string which
will be submitted to the search engine. You may include special characters
to group or to qualify the search.
- "-e" or "--engine"
- Specify the search engine. The query string will be submitted to the user
specified search engine.
In many cases there are specialized versions of search engines. For example,
AltaVista::AdvancedWeb and AltaVista::News allow more
powerful and Usenet searches. See AltaVista or the man page for your
search engine for details about specialized variations.
- "--listnewurls"
- In addition to all the normal file maintenance, print all new URLs to
STDOUT, one per line.
- "-o" or "--options"
- Specify the query options. The query options will be submitted to the user
search engine with the query string. This feature permits modification of
the query string for a specific search engine or option. More than one
query option may be specified.
Example: "-o what=news" causes AltaVista to search Usenet.
Although this works, the preferred mechanism in this case would be
"-e AltaVista::News" or "-e AltaVista::AdvancedNews".
Options are intended for internal or expert use.
- "-f" or "--uf" or "--urlfilter"
- This option specifies a regular expression which will be compared against
the URLs of any results; if they match the case-insensitive regular
expression, they will be removed from the hit set.
Example: "-f '.*\.isi\.edu'" avoids all of ISI's web pages.
- "--cleanup i"
- Delete all traces of query results from more than i days ago. If --cleanup
is given, all other options other than the qid will be ignored.
- "--cmdline"
- Reconstruct the complete command line (AutoSearch and all its arguments)
that was used to create the query results. Command line will be shown on
STDERR. If --cmdline is given, all other options other than the qid will
be ignored.
- "--mail user@address" or "-m user@address"
- After search is complete, send email to that user, listing the NEW
results. Email is HTML format. Requires the Email::Send and related
modules. If you send email through an SMTP server, you must set
environment variable SMTPSERVER to your server name or IP address. If your
SMTP server requires password, you must set environment variables
SMTPUSERNAME and SMTPPASSWORD. If you send email via sendmail, you should
set environment variable SENDMAIL if the sendmail executable is not in the
path.
- "--emailfrom user@address"
- If your outgoing mail server rejects email from certain users, you can use
this argument to set the From: header.
- "--userid bbunny"
- If the search engine requires a login/password (e.g. Ebay::Completed), use
this.
- "--password Carr0t5"
- If the search engine requires a login/password (e.g. Ebay::Mature), use
this.
DESCRIPTION¶
AutoSearch submits a query to a search engine, produces HTML pages that
reflect the set of 'hits' (filtered search results) returned by the search
engine, and tracks these results over time. The URL and title are displayed in
the
qid/index.html, the URL, the title, and description are displayed
in the 'weekly' files.
To organize these results, each search result is placed in a query information
directory (qid). The directory becomes the search results 'handle', an easy
way to track a set of results. Thus a qid of
"/usr/local/htdocs/lsam/autosearch/load_balancing" might locate the
results on your web server at
"
http://www.isi.edu/lsam/autosearch/load_balancing".
Inside the qid directory you will find files relating to this query. The primary
file is
index.html, which reflects the latest search results. Every
not-filtered hit for every search is stored in
index.html. When a hit
is no longer found by the search engine it a removed from
index.html.
As new results for a search are returned from the search engine they are
placed in
index.html.
At the bottom of
index.html, there is a heading "Weekly Search
Results", which is updated each time the search is submitted (see
"AUTOMATED SEARCHING"). The list of search runs is stored in reverse
chronological order. Runs which provide no new information are identified with
No Unique Results found for search on <date>
Runs which contain changes are identified by
Web search results for search on <date>
which will be linked a page detailing the changes from that run.
Detailed search results are noted in weekly files. These files are named
YYYYMMDD.html and are stored in the qid directory. The weekly files
include THE URL, title, and a the description (if available). The title is a
link to the original web page.
AUTOMATED SEARCHING¶
On UNIX-like systems,
cron(1) may be used to establish periodic searches
and the web pages will be maintained by
AutoSearch. To establish the
first search, use the first example under SYNOPSIS. You must specify the qid,
query name and query string. If any of the items are missing, you will be
interactively prompted for the missing item(s).
Once the first search is complete you can re-run the search with the second form
under SYNOPSIS.
A cron entry like:
0 3 * * 1 /nfs/u1/wls/AutoSearch.pl /www/div7/lsam/autosearch/caching
might be used to run the search each Monday at 3:00 AM. The query name and query
string may be repeated; but they will not be used. This means that with a cron
line like:
0 3 * * 1 /nfs/u1/wls/AutoSearch.pl /www/div7/lsam/autosearch/caching -n caching -s caching
a whole new search series can be originated by
rm -r /www/div7/lsam/autosearch/caching
However, the only reason to start a new search series would be to throw away the
old weekly files.
We don't recommend running searches more than once per day, but if so the
per-run files will be updated in-place. Any changes are added to the page with
a comment that "Recently Added:"; and deletions are indicated with
"Recently Suspended:."
CHANGING THE LOOK OF THE PAGES¶
The basic format of these two pages is simple and customizable. One requirement
is that the basic structure remain unchanged. HTML comments are used to
identify sections of the document. Almost everything can be changed except for
the strings which identify the section starts and ends.
Noteworthy tags and their meaning:
- <!--Top-->.*<!--/Top-->
- The text contained within this tag is placed at the top of the output
page. If the text contains AutoSearch WEB Searching, then the query
name will replace it. If the text does not contain this magic string and
it is the first ever search, the user will be asked for a query name.
- <!--Query{.*}/Query-->
- The text contained between the braces is the query string. This is how
AutoSearch maintains the query string. You may edit this string to
change the query string; but only in qid/index.html. The text
ask user is special and will force AutoSearch to request the
search string from the user.
- <!--SearchEngine{.*}/SearchEngine-->
- The text contained between the braces is the search engine. Other engines
supported are HotBot and Lycos. You may edit this string to change the
engine used; but only in qid/index.html. The text ask
user is special and will force AutoSearch to to request the
search string from the user.
- <!--QueryOptions{.*}/QueryOptions-->
- The text contained between the braces specifies a query options. Multiple
occurrencs of this command are allowed to specify multiple options.
- <!--URLFilter{.*}/URLFilter-->
- The text contained between the braces is the URL filter. This is how
AutoSearch maintains the filter. Again you may edit this string to
change the query string; but only in qid/index.html. The text
ask user is special and will force AutoSearch to ask the
user (STDIN) for the query string. When setting up the first search, you
must edit first_index.html, not qid/index.html. The URL
filter is a standard perl5 regular expression. URLs which do not match
will be kept.
- <!--Bottom-->.*<!--/Bottom-->
- The text contained within this tag is placed at the bottom of the output
page. This is a good place to put navigation, page owner information,
etc.
The remainder of the tags fall into a triplet of
~Heading,
~Template, and
~, where ~ is Summary, Weekly, Appended, and
Suspended. The sub-sections appear in the order given. To produce a section
AutoSearch outputs the heading, the template, the section, n copies of
the formatted data, and an /section. The tags and their function are:
- ~Heading
- The heading tag identifies the heading for a section of the output file.
The SummaryHeading is for the summary portion, etc. The section may be
empty (e.g., Suspended) and thus no heading is output.
- ~Template
- The template tag identifies how each item is to be formatted. Simple text
replacement is used to change the template into the actual output text.
The text to be replaced is noted in ALLCAPS.
- ~
- This tag is used to locate the section (Summary, Weekly, etc.). This
section represents the actual n-items of data.
You can edit these values in the
qid/index.html page of an existing
search. The file
first_index.html (in the directory above
qid)
will be used as a default template for new queries.
Examples of these files can be seen in the pages under
"
http://www.isi.edu/lsam/tools/autosearch/", or in the output
generated by a new AutoSearch.
FILES¶
- first_index.html
- optional file to determine the default format of the index.html
file of a new query.
- first_date.html
- optional file to determine the default format of the YYYYMMDD.html
file for a new query.
- qid/index.html
- (automatically created) latest search results, and reverse chronological
list of periodic searches.
- qid/date.html
- file used as a template for the YYYYMMDD.html files.
- qid/YYYYMMDD.html
- (automatically created) summary of changes for a particular date (AKA
'Weekly' file).
Optional files
first_index.html and
first_date.html are used for
the initial search as a template for
qid/index.html and
date.html, respectively. If either of these files does not exist; a
default-default template is stored within the
AutoSearch source. The
intention of these two files is to permit a user to establish a framework for
a group of search sets which have a common format. By leaving the default
query name and query string alone, they will be overridden by command line
inputs.
SEE ALSO¶
For the library, see WWW::Search, for the perl regular expressions, see perlre.
AUTHORS¶
Wm. L. Scheding
AutoSearch is a re-implementation of an earlier version written by Kedar
Jog.
COPYRIGHT¶
Copyright (C) 1996-1997 University of Southern California. All rights reserved.
Redistribution and use in source and binary forms are permitted provided that
the above copyright notice and this paragraph are duplicated in all such forms
and that any documentation, advertising materials, and other materials related
to such distribution and use acknowledge that the software was developed by
the University of Southern California, Information Sciences Institute. The
name of the University may not be used to endorse or promote products derived
from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED "AS IS" AND WITHOUT ANY EXPRESS OR IMPLIED
WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
DESIRED FEATURES¶
These are good ideas that people have suggested.
- URL validation.
- Validate the status of each URL (with HTTP HEAD requests) and indicate
this status in the output.
- Multi-search.
- It should be possible to merge the results of searches from two
search-engines. If this merger were done as a new search engine, this
operation would be transparent to AutoSearch.
BUGS¶
None known at this time; please inform the maintainer mthurn@cpan.org if any
crop up.