sed & awk

sed & awkSearch this book
Previous: 1.1 May You Solve Interesting ProblemsChapter 1
Power Tools for Editing
Next: 1.3 A Pattern-Matching Programming Language
 

1.2 A Stream Editor

Sed is a "non-interactive" stream-oriented editor. It is stream-oriented because, like many UNIX programs, input flows through the program and is directed to standard output. (vi, for instance, is not stream-oriented. Nor are most DOS applications.) Input typically comes from a file but can be directed from the keyboard.[2] Output goes to the terminal screen by default but can be captured in a file instead. Sed works by interpreting a script specifying the actions to be performed.

[2] Doing so, however, is not particularly useful.

Sed offers capabilities that seem a natural extension of interactive text editing. For instance, it offers a search-and-replace facility that can be applied globally to a single file or a group of files. While you would not typically use sed to change a term that appears once in a particular file, you will find it very useful to make a series of changes across a number of files. Think about making 20 different edits in over 100 files in a matter of minutes, and you get an idea of how powerful sed can be.

Using sed is similar to writing simple shell scripts (or batch files in DOS). You specify a series of actions to be performed in sequence. Most of these actions could be done manually from within vi: replacing text, deleting lines, inserting new text, etc. The advantage of sed is that you can specify all editing instructions in one place and then execute them on a single pass through the file. You don't have to go into each file to make each change. Sed can also be used effectively to edit very large files that would be slow to edit interactively.

There are many opportunities to use sed in the course of creating and maintaining a document, especially when the document consists of individual chapters, each placed in a separate file. Typically, after a draft of a document has returned from review, there are a number of changes that can be applied to all files. For instance, during the course of a software documentation project, the name of the software or its components might change, and you have to track down and make these changes. With sed, this is a simple process.

Sed can be used to achieve consistency throughout a document. You can search for all the different ways a particular term might be used and make them all the same. You can use sed to insert special typesetting codes or symbols prior to formatting by troff. For instance, it can be used to replace quotation marks with the ASCII character codes for forward and back double quotes ("curly quotes" instead of "straight" quotes).

Sed also has the ability to be used as an editing filter. In other words, you could process an input file and send the output to another program. For instance, you could use sed to analyze a plain text file and insert troff macros before directing the output to troff for formatting. It allows you to make edits on the fly, perhaps ones that are temporary.

An author or publisher can use sed to write numerous conversion programs - translating formatting codes in Scribe or TeX files into troff, for example, or converting PC word processing files, such as WordStar. Later on, we will look at a sed script that converts troff macros into stylesheet tags for use in Ventura Publisher. (Perhaps sed could be used to translate a program written in the syntax of one language to the syntax of another language.) When Sun Microsystems first produced Xview, they released a conversion program for converting SunView programs to XView, and the program largely consisted of sed scripts, converting the names of various functions.

Sed has a few rudimentary programming constructs that can be used to build more complicated scripts. It also has a limited ability to work on more than one line at a time.

All but the simplest sed scripts are usually invoked from a "shell wrapper," a shell script that invokes sed and also contains the commands that sed executes. A shell wrapper is an easy way to name and execute a single-word command. Users of the command don't even need to know that sed is being used. One example of such a shell wrapper is the script phrase, which we'll look at later in this book. It allows you to match a pattern that might fall over two lines, addressing a specific limitation of grep.

In summary, use sed:

  1. To automate editing actions to be performed on one or more files.

  2. To simplify the task of performing the same edits on multiple files.

  3. To write conversion programs.


Previous: 1.1 May You Solve Interesting Problemssed & awkNext: 1.3 A Pattern-Matching Programming Language
1.1 May You Solve Interesting ProblemsBook Index1.3 A Pattern-Matching Programming Language

The UNIX CD Bookshelf NavigationThe UNIX CD BookshelfUNIX Power ToolsUNIX in a NutshellLearning the vi Editorsed & awkLearning the Korn ShellLearning the UNIX Operating System


Banner.Novgorod.Ru