guile-llama-cpp archive search

Query

Long messages and words are ignored

Messages longer than 100,000 letters or 500,000 bytes are ignored. Words longer than 40 characters are ignored. Attachments are ignored.

Single term query

The query specifies only one term for retrieving all documents which contain the term. e.g.,

namazu

AND query

The query specifies two or more terms for retrieving all documents which contain both terms. You can insert the and operator between the terms. e.g.,

Linux and Netscape

You can ommit the and operator. Terms which is separated by one ore more spaces is assumed to be AND query.

OR query

The query specifies two or more terms for retrieving all documents which contain either term. You can insert the or operator between the terms. e.g.,

Linux or FreeBSD

NOT query

The query specifies two or more terms for retrieving all documents which contain a first term but does't contain the following terms. You can insert the not operator between the terms to do NOT query. e.g.,

Linux not UNIX

Grouping

You can group queries by surrounding them by parentheses. The parentheses should be separated by one or more spaces. e.g.,

( Linux or FreeBSD ) and Netscape not Windows

Phrase searching

You can search for a phrase which consists of two or more terms by surrounding them with double quotes like "..." or with braces like {...}. In Namazu, precision of phrase searching is not 100 %, so it causes wrong results occasionally. e.g.,

{GNU Emacs}

Substring matching

The are three types of substring matching searching.

Prefix matching

inter* (terms which begin with inter)

Inside matching

*text* (terms which contain text)

Suffix matching

*net (terms which terminated with net)

Regular expressions

You can use regular expressions for pattern matching. The regular expressions must be surrounded by slashes like /.../. Namazu uses Ruby's regular regular expressions engine. It offers generally Perl compatible flavor. e.g.,

/pro(gram|blem)s?/

Field-specified searching

You can limit your search to specific fields such as Subject:, From:, Message-Id:. It's especially convenient for Mail/News documents. e.g.,

• +subject:Linux
  (Retrieving all documents which contain Linux in a Subject: field)
• +subject:"GNU Emacs"
  (Retrieving all documents which contain GNU Emacs in a Subject: field)
• +from:foo@bar.jp
  (Retrieving all documents which contain foo@bar.jp in a From: field)
• +message-id:<199801240555.OAA18737@foo.bar.jp>
  (Retrieving a certain document which contains specified Message-Id:)

Notes

• In any queries, Namazu ignores case distinctions of alphabet characters. In other words, Namazu does case-insensitive pattern matching in any time.
• Japanese phrases are forced to be segmented into morphemes automatically and are handled them as phrase searching. This processing causes invalid segmentation occasionally.
• Alphabet, numbers or a part of symbols (duplicated in ASCII) characters which defined in JIS X 0208 (Japanese Industrial Standards) are handled as ASCII characters.
• Namazu can handle a term which contains symbols like TCP/IP. Since this handling isn't complete, you can describe TCP and IP instead of TCP/IP, but it may cause noisy results.
• Substring matching and field-specified searching takes more time than other methods.
• If you want to use and, or or not simply as terms, you can surround them respectively with double quotes like "..." or braces like {...}.