I implemented my own search engine back before Google became popular.
It's usually more complete than external search engines (for searching this site), but isn't as friendly.
I've included Google search too.
Notes:
My search tool returns a list of documents on the server which contain one or more of the specified keywords.
It searches every HTML file in the specified path(s), and does not use indexes.
This makes it a little slow, but pretty thorough.
-
Use the + prefix to denote required keywords.
For example, bill +gates will return documents with at least one occurance of "gates" (but maybe not "bill").
Among the results, documents with lots of "gates" will rank higher than those with lots of "bill".
-
Use double quotes to enclose phrases.
So, "bill gates" will return documents which contain the term "bill" followed by white space and then "gates".
Independent occurances of "bill" and "gates" are ignored.
- Searches are case-insensitive.
-
Keywords can currently match substrings in the source documents.
This means the query "q" will return a list of all the documents which contain the letter q.
If you want to search for words starting with q, you can try " q",
and for words that consist solely of the letter q (surrounded by white space), you can try " q ".
Right now I don't handle punctuation specially, so " q. " won't result in a match if the query is " q ".
- This specifies which directories to search for files containing the keywords.
- For root directory (the entire server), type /
- Leading or trailing / symbols are ignored.
- Separate multiple search directories with spaces. (e.g., cope/rail-on tripecac/albums)
- You can use wildcards (e.g., cope/*) to search all the direct subdirectories in a tree.
- Currently there is no recursive directory search, and we only search .htm and .html files.
- This specifies the maximum number of documents to return to the user.
- Lowering this number does not speed up searches; it simply reduces the size of the results page.
- If this option is Yes, you get additional output.
- It's mostly useful if you have multiple search terms.
If you can see this message, then you're viewing this page w/o CSS because either:
- Your browser doesn't fully support CSS (e.g., NS4, Lynx, etc.).
- You disabled CSS support in your browser or used my "nocss" link.
- You're viewing this offline, w/o having the expected CSS file.
- Some other reason which I haven't anticipated.
The page should still be usable without CSS. It's just ugly.
back to top