sdiff and More

More on GNU diffutils. A look into sdiff, which shows file comparisons side-by-side and lets you merge two files interactively.

As was said in last month’s column (the first of this series), there are good graphical file-comparison programs available, and many of those take advantage of color and of a scrollbar to make file comparisons easy to see. However, the GUI tools aren’t always the best way to compare files.

The topic last month was diff, which shows differences “vertically” — showing first the lines from one file, then lines from the other file. This month, let’s look into sdiff, which shows differences “horizontally” or side-by-side. It also lets you merge two files interactively. sdiff has more than its share of oddities, too– let’s explore those in detail, too, so you’ll be ready to use this handy tool.

(The GNU diffutils documentation says that the name sdiff is obsolete for non-interactive use, without -o. It recommends using diff –y or diff ––side-by-side instead. For consistency, though, this column uses sdiff everywhere.)

Why Use sdiff?

There are a number of good reasons to use the command-line sdiff instead of graphical difference programs:

*You have extra control over the output. For instance, sdiff –I ignores changed lines that match one or more regular expressions you specify. It can also ignore changes in tabs, whitespace, carriage returns, and more. (sdiff understands many of the diff options covered in last month’s column.)

*The default (non-interactive) output is 130 columns wide, designed for printing on line printers– those clattering machines that use stacks of folded paper with holes on the edges. However, it’s also good for printing on sheet-fed printers. Choose landscape mode, and use a constant-width font that maintains the position of the center “gutter” between the files. (There are examples of the gutter later in this column.)

*Although the non-interactive sdiff is designed for printing, it’s also quite useful in a terminal window. Stretch a window wide enough to view the output, or use the handy pager program less to pan side-to-side through the output.

*sdiff can also merge two files interactively, letting you choose the lines from each file that make up the merged output file.

Let’s dig in!

Showing Differences On-Screen

By default (without its –o option), sdiff compares two files and shows the two in a side-by-side format 130-characters (columns) wide. Since typical terminal windows are 80-characters wide, you’ll need to make some adjustments to view the output sensibly. We’ll cover some of those in the next section.

Let’s compare two short files, file1a and file1b, shown in Table One.

Table One: Two short files, file1a and file1b, left and right, respectively
File file1a File file1b
line 1 line one
line 2 line 2
line 2a line 3
line 3 line 3b

To save printing space, let’s use the option –w 30 to make the output lines 30-characters wide, which is enough to show the short lines from both sample files. Here’s the sdiff output:

$ sdiff –w 30 file1a file1b
line 1 | line one
line 2 line 2
line 2a <
line 3 line 3
> line 3b

The default sdiff output shows the first 61 characters from each line of both files. The –w 30 forces the output to be much narrower. If lines in the first file are too short, sdiff outputs enough tabs and spaces at the end of the first file’s line to make the complete 61 columns.

The two files’ contents are separated by a gutter. If the two lines differ, the gutter has a space, a character that denotes the type of difference, and a tab. If the two lines are the same, the gutter is just a tab.

*A | (vertical bar) means that the lines are different. In the example, the first line is different.

*A < (left angle bracket) denotes that the line is only in the first file (in the left column). Here, line 2a only occurs in the first file (file1a).

*A > (right angle bracket) marks a line that’s only in the second file (right column). Here, that’s line 3b.

Cutting Clutter

Seeing common lines repeated can clutter the output. The –l (lowercase “L”) or ––left-column option outputs only the left column of common lines. Let’s try it with the same two files, and also use the shell wildcard [ab] to avoid typing both filenames separately:

$ sdiff –w 30 –l file1[ab]
line 1 | line one
line 2 (
line 2a <
line 3 (
> line 3b

The ( (left parenthesis) character in the gutter indicates that line 2 and line 3 are in both files, but it’s only shown in the left column.

If you don’t need to see the common lines at all, use –s or ––suppress-common-lines, as in:

$ sdiff –w 30 –s file1[ab]
line 1 | line one
line 2a <
> line 3b

Incomplete Lines

You may see the marker \ or / in the gutter. The reason for it is worth understanding– especially if you use the Emacs editor, which can cause the problem if you aren’t careful.

Each line in a Linux text file ends with a newline character. In Emacs, you can create a file whose last line doesn’t end with a newline if you don’t press Enter at the end of the last line. When plain diff compares a file like that, it warns “No newline at end of file”. sdiff works differently.

If both files have a line with the same text, but one of those lines is incomplete (doesn’t have a newline), sdiff puts the marker \ or / in the gutter. However, if only one of the files has a line, and that line is also incomplete, sdiff uses the marker < or >– and also doesn’t output a final newline. (This is the behavior of GNU sdiff version 2.8.1.)

Figure One shows an example you can try. It uses echo –n and the shell’s “append” operator >> to create files without a final newline. The file file2a has two newline-terminated lines; file2b has two lines, the last incomplete; and file2c has three lines, the last incomplete.

Figure One: Files with incomplete lines

$ echo” line 1 nl” > file2a

$ cp file2a file2b

$ echo” line 2 nl” >> file2a

$ echo –n” line 2″ >> file2b

$ diff file2[ ab]


< line 2 nl

> line 2

\ No newline at end of file

$ sdiff –w 30 file2[ ab]

line 1 nl line 1 nl

line 2 nl / line 2

$ sdiff –w 30 file2b file2a

line 1 nl line 1 nl

line 2 \ line 2 nl

$ cp file2a file2c

$ echo –n” line 3″ >> file2c

$ diff file2[ ac]


> line 3

\ No newline at end of file

$ sdiff –w 30 file2[ ac]

line 1 nl line 1 nl

line 2 nl line 2 nl

> line 3$

Notice the $ prompt at the end of the last line. It’s there because sdiff output line 3 from file2c as-is– without a newline — and the shell did what it always does: outputs its prompt string after the program exits.

Alignment Problems

If there’s misalignment in the characters in the gutter, or at the start of each line in the second file, your terminal window may not be using a constant-width font. Be sure you’re using a typewriter-like font, such as Courier, where, for instance, a period (.) takes the same horizontal space as an uppercase W.

sdiff tries to handle tab characters correctly on both the input file and the output lines to your terminal. If it doesn’t seem to work, though, try the sdiff option –t or ––expand-tabs. It expands tabs in the input files to spaces on output.

If problems persist, your terminal might not be handling tabs correctly. Try piping the output of sdiff to the expand utility, which converts tabs into the proper number of space characters to keep the default eight-character tabstop settings. (However, don’t use expand with the –o option, which you’ll see momentarily. Pipe buffering can keep you from seeing all of the text.) If the output of expand looks right, then your terminal’s tabstop settings may be wrong. In that case, try using the reset utility or opening another terminal window.

Line-Width Problems

If a line from one of the input files is wider than the space available in the sdiff output, the line is truncated, even if the difference is in that truncated line part!

One useful adjustment is to make the output wider: 160 columns or more. This lets you compare two files with 80-character lines and see all of each file. The option –w nnn sets the total output line width, so –w 168 makes lines wide enough for two 80-character-wide files and the 8-character gutter. Of course, if your terminal isn’t 168 characters wide (and you can’t stretch it that far or use a smaller font), the output lines will either wrap or be truncated.

A handy way to show sdiff output that’s wider than your terminal is by piping the sdiff output to less –S. (That’s an uppercase “S”.) This truncates each line at the screen width; you can scroll right and left across the full line width with the right-arrow and left-arrow keys. (This won’t work with interactive sdiff, though.)

Merging Files With sdiff

To merge two files interactively, use sdiff –o outfile or sdiff ––output= outfile, where outfile is the destination of the merged text. sdiff starts from the beginning of the files, outputting the first set of common lines to outfile. Then it shows the first hunk of differing lines and prompts with %. At the prompt, you can type one of these commands and press Enter:

*l copies the left version to outfile, while r copies the right version.

*el invokes an editor on the left version, then copies the edited output to outfile. er does the same for the right version.

*eb concatenates both versions into a temporary file, lets you edit the temporary file, then copies the result to outfile.

*ed is like eb, but it puts a header with filename and line numbers before each version. (Remove the header when you edit the temporary file.)

*e discards both versions and opens an editor on an empty file. The result is copied to outfile.

*q quits early.

*Enter gives a brief summary of the commands.

Figure Two shows an example editing session. The editor used is ed, the simple line-oriented editor, so you can see what’s happening, but you’ll probably prefer your usual editor (which may already be set as EDITOR by default).

Figure Two: Interactive file merging with sdiff –o
$ export EDITOR=ed
$ sdiff –o file1.merge –w 30 file1[ab]
line 1 | line one
line 2 line 2
line 2a <
line 3 line 3
> line 3b
That’s all.
$ cat file1.merge
line 1
line 2
line 3
That’s all.

When sdiff shows line 2a, which is only in the left file, press r to ignore the line (output nothing to file1.merge). When sdiff shows line 3b, press e to open an empty file with ed. In the editor, add the single line That’s all. to the file, and save it.

Quick-and-Dirty Merging

Interactive merging with sdiff can be tedious. Here’s a kludge that’s often easier. Make a diff –u (unified diff format) listing of both files, with enough context to include all lines. Then edit the merged listing, removing the lines you don’t want. Finally, remove the first character (+, -, and space) from the remaining lines.

For example, to merge file1a and file1b with a maximum 4 lines of context (which is plenty for 4-line files):

$ diff –u4 file1[ab] > file1.merge
$ cat file1.merge
--- file1a ...
+++ file1b ...
@@ -1,4 +1,4 @@
-line 1
+line one
line 2
-line 2a
line 3
+line 3b
$ vi file1.merge

(To count the number of lines in a file, use wc –l.) Remove the three header lines and any other lines you don’t want to keep. Then remove the leading space, + and - characters (in vi, for example, use the command :%s/^.//).

To Be Continued…

Next time, we’ll look into the little-known diff3, including using it to merge files automatically. We’ll also cover hints about using patch to merge.

Comments on "sdiff and More"

Wonderful story, reckoned we could combine several unrelated data, nevertheless definitely worth taking a appear, whoa did a single find out about Mid East has got more problerms too.

It¡¯s really a nice and useful piece of info. I am glad that you shared this useful information with us. Please keep us up to date like this. Thanks for sharing.

Here are some hyperlinks to sites that we link to because we think they’re really worth visiting.

Always a major fan of linking to bloggers that I appreciate but don?t get quite a bit of link adore from.

Just beneath, are many absolutely not associated internet sites to ours, nonetheless, they may be surely worth going over.

Just beneath, are many absolutely not connected sites to ours, having said that, they’re certainly worth going over.

Here is a good Weblog You might Find Exciting that we encourage you to visit.

Wonderful story, reckoned we could combine a few unrelated data, nonetheless definitely really worth taking a appear, whoa did one find out about Mid East has got more problerms as well.

I used to be recommended this web site by way of my cousin. I am not positive whether or not this post is written by way of him as nobody else realize such particular approximately my trouble. You’re incredible! Thank you!

Here are some of the web-sites we suggest for our visitors.

Although websites we backlink to below are considerably not related to ours, we feel they are truly worth a go by way of, so possess a look.

The time to study or pay a visit to the material or web sites we’ve linked to beneath.

Just beneath, are quite a few entirely not related web-sites to ours, nevertheless, they may be surely worth going over.

Always a massive fan of linking to bloggers that I appreciate but really don’t get a lot of link adore from.

The time to study or stop by the material or websites we’ve linked to below.

The information mentioned within the report are a number of the very best offered.

I just could not depart your web site before suggesting that I extremely enjoyed the standard info a person provide for your visitors? Is going to be back often in order to check up on new posts

Its like you read my thoughts! You appear to know so much about this, such as you wrote the e-book in it or something. I feel that you simply could do with a few percent to power the message house a little bit, but instead of that, this is great blog. A great read. I will definitely be back.

That could be the end of this write-up. Here you?ll come across some sites that we believe you?ll appreciate, just click the links.

Always a huge fan of linking to bloggers that I really like but do not get a good deal of link adore from.

That would be the end of this article. Right here you will come across some web-sites that we assume you?ll value, just click the hyperlinks.

Usually posts some extremely fascinating stuff like this. If you are new to this site.

We like to honor quite a few other world wide web internet sites around the net, even though they aren?t linked to us, by linking to them. Beneath are some webpages worth checking out.

The heart of your writing while appearing agreeable at first, did not sit properly with me personally after some time. Someplace throughout the sentences you actually were able to make me a believer unfortunately just for a short while. I still have got a problem with your leaps in logic and you would do nicely to fill in those gaps. If you actually can accomplish that, I could surely end up being fascinated.

We came across a cool site that you simply could appreciate. Take a look in case you want.

Very few web sites that take place to become in depth below, from our point of view are undoubtedly well really worth checking out.

We prefer to honor several other world-wide-web internet sites on the web, even when they aren?t linked to us, by linking to them. Below are some webpages really worth checking out.

Here are some hyperlinks to web sites that we link to due to the fact we assume they are really worth visiting.

Very couple of internet websites that come about to become detailed below, from our point of view are undoubtedly well worth checking out.

Usually posts some really intriguing stuff like this. If you?re new to this site.

Every when inside a although we choose blogs that we study. Listed beneath are the most up-to-date websites that we select.

The time to study or check out the material or internet sites we have linked to below.

The time to read or stop by the content material or web-sites we have linked to beneath.

Although web-sites we backlink to beneath are considerably not related to ours, we really feel they are truly worth a go via, so have a look.

Check beneath, are some absolutely unrelated web sites to ours, even so, they are most trustworthy sources that we use.

Wonderful story, reckoned we could combine a number of unrelated information, nonetheless genuinely really worth taking a search, whoa did 1 understand about Mid East has got additional problerms also.

Here is an excellent Blog You might Discover Fascinating that we encourage you to visit.

Please check out the internet sites we adhere to, such as this one, because it represents our picks from the web.

Below you will locate the link to some web-sites that we believe you ought to visit.

Below you?ll obtain the link to some web pages that we assume you’ll want to visit.

Here is an excellent Blog You might Come across Fascinating that we encourage you to visit.

Always a large fan of linking to bloggers that I adore but don?t get a whole lot of link appreciate from.

The data talked about in the report are a few of the best offered.

Check below, are some completely unrelated websites to ours, nonetheless, they’re most trustworthy sources that we use.

875969 295812I?d forever want to be update on new posts on this site , bookmarked ! . 897595

Every after in a although we select blogs that we read. Listed beneath are the most up-to-date internet sites that we choose.

One of our guests a short while ago advised the following website.

Here is an excellent Weblog You may Locate Interesting that we encourage you to visit.

One of our guests not long ago encouraged the following website.

One of our visitors recently advised the following website.

Leave a Reply