Minimum unique abbreviation of option is acceptable. You may use double hyphens instead of single hyphen to denote options. You may use white space in place of the equals sign to separate an option name from its value.
This program is part of Netpbm.
pbmtext takes the specified text, either a single line from the command line or multiple lines from standard input, and renders it into a PBM graphical image.
The text rendered is all the non-option command line arguments, separated by spaces, except that if there are no non-option command line arguments, it is Standard Input.
In the image, each line of input is a line of output. Lines are delimited by newline characters.
The program renders any character in an input line that is not in the font as a space. Note that control characters usually aren't in the font, but some fonts have glyphs for them. The newline characters that delimit lines in of Standard Input are not in any line.
Tab characters are rendered as a number of spaces; any entry in the font for the tab code point is irrelevant. The number of spaces is such as to create tab stops every 8 characters. Note that this is not useful if the font is proportional.
The image is just wide enough for the longest line of text, plus margins, and just high enough to contain the lines of text, plus margins.
The left and right margins are twice the width of the widest character in the font; the top and bottom margins are the height of the tallest character in the font. But if the text is only one line, all the margins are half of this. You can use the -nomargins option to eliminate the margins.
pbmtext renders left to right. It cannot render vertically or right to left.
pbmtextps does the same thing as pbmtext, but uses Ghostscript to generate the characters, which means you can use Postscript fonts. But it also means you have to have Ghostscript installed and it isn't as fast. Also, pbmtextps generates only one line of text, whereas pbmtext can create multiple lines.
pbmtext is meant for simple text. If you're working with a document, you would be better off using a document formatting program to "print" to a Postscript file, then feeding that Postscript to pstopnm.
In addition to the options common to all programs based on libnetpbm (most notably -quiet, see Common Options), pbmtext recognizes the following command line options:
By default, pbmtext takes a single-byte character stream as input. When you specify -wchar, it treats input text as a multibyte character stream encoded according to the current locale. Normally, the user would supply a BDF font file encoded in ISO-10646-1 with a -font option.
With -wchar, you cannot supply the text on the command line; it must be fed from standard input.
This option was new in Netpbm 10.82 (March 2018).
The default is the built in font "bdf."
"bdf" is Times-Roman 15 pixels high. (That's about 14 point type printed at 75 dpi).
"fixed" is a built in fixed width font.
For information about other fonts, and how to make one of your own, see Fonts below.
pixels may be fractional, in which case the number of pixels added varies so as to achieve the specified average. For example -space=1.5 causes half the spaces to be 1 pixel and half to be 2 pixels.
pixels may be negative to crowd text together, but the author has not put much thought or testing into how this works in every possible case, so it might cause disastrous results.
pixels must be a whole number.
pixels may be negative to crowd lines together, but the author has not put much thought or testing into how this works in every possible case, so it might cause disastrous results.
Note that there may still be space beyond the edges of the type because a character itself may include space at its edges. To eliminate all surrounding background, so the type touches all four edges of the image, use pnmcrop.
If the input is one line, pbmtext breaks it into multiple lines as needed to fit the specified width. It breaks it between characters, but does not pay attention to white space; it may break in the middle of a word and a line may begin or end with white space.
If the input is multiple lines, pbmtext assumes you already have line breaks where they make sense, and pbmtext simply truncates each line as needed to fit the specified width.
This option was new in Netpbm 10.91 (June 2020). Before that, pbmtext always reads the entire font.
With this option, instead of outputting an image of the text, pbmtext just writes to Standard Output a message telling the dimensions of the image it would have produced.
You can specify only one of -dry-run and -text-dump.
This option was new in Netpbm 10.75 (June 2016).
When -wchar is in effect, the output text will be in the encoding specified by the current locale.
You can specify only one of -dry-run and -text-dump.
This option was new in Netpbm 10.82 (March 2018).
Often, you want to place text over another image. One way to do this is with ppmlabel. For more flexible (but complex) drawing of text on an image, there is ppmdraw. These do not give you the font options that pbmtext does, though.
Another way is to use pbmtext to create an image containing the text, then use pamcomp to overlay the text image onto your base image. To make only the text (and not the entire rectangle containing it) cover the base image, you will need to give pamcomp a mask, via its -alpha option. You can just use the text image itself as the mask, as long as you also specify the -invert option to pamcomp.
If you want to overlay colored text instead of black, just use ppmchange to change all black pixels to the color of your choice before overlaying the text image. But still use the original black and white image for the transparency mask.
If you want the text at an angle, use pnmrotate on the text image (and transparency mask) before overlaying.
There are three kinds of fonts you an use with pbmtext:
There are two built in fonts: bdf and fixed. You select these fonts with a -builtin option.
bdf is the default when you specify no font information on the command line. The naming reflects the fact that it shares many characteristics of BDF style fonts. When this font was implemented, pbmtext did not have the ability to read arbitrary BDF fonts specified by the -font option. There is no external font file involved.
bdf is encoded in ISO 8859-1 (Latin 1, 8-bit). In addition to English it can handle most West European languages (Spanish, French, German, Swedish ...) This set lacks the Euro currency sign.
fixed is ASCII (7-bit) only.
While it is not an error to do so, you should not use the above built-in fonts with -wchar.
BDF is an ancient font format that at one time was standard for the X Window System. Now, you don't see it very often. Until about 2024, you could find some BDF fonts on the Xfree86 web site. Today, you can get the full package of the BDF fonts from XFree86 from the Netpbm web site.
To create a font as a PBM file (to use with the -font option), you just create a PBM image of the text matrix below.
The first step is to display text matrix below on the screen, e.g. in an X11 window.
M ",/^_[`jpqy| M
/ !"#$%&'()*+ /
< ,-./01234567 <
> 89:;<=>?@ABC >
@ DEFGHIJKLMNO @
_ PQRSTUVWXYZ[ _
{ \]^_`abcdefg {
} hijklmnopqrs }
~ tuvwxyz{|}~ ~
M ",/^_[`jpqy| M
Make sure it's a fixed width font -- This should display as a perfect rectangle.
Also, try to use a simple display program. Pbmtext divides this into a matrix of cells, all the same size, each containing one character, so it is important that whatever you use to display it display with uniform horizontal and vertical spacing. Fancy word processing programs sometimes stretch characters in both directions to fit certain dimensions, and that won't work. Sometimes a display program scales a font to show a character larger or smaller than its natural size. That too won't often work because the rounding involved in such scaling causes non-uniform distances between characters.
If you display the text matrix improperly, the usual symptom is that when you try to use the font, pbmtext fails with an error message telling you that the number of lines in the font isn't divisible by 11, or it can't find the blank band around the inner rectangle. Sometimes the symptom is that one of the characters displays with a piece of the character that is next to it in the matrix. For example, "l" might display with a little piece of the "m" attached on its right.
Do a screen grab or window dump of that text, using for instance xwd, xgrabsc, or screendump. Convert the result into a pbm file. If necessary, use pamcut to remove anything you grabbed in addition to the text pictured above (or be a wimp and use a graphical editor such as Gimp). Finally, run it through pnmcrop to make sure the edges are right up against the text. pbmtext can figure out the sizes and spacings from that.
There are some historical computer fonts, such as that used by the original IBM PC, in the form that you can screen-grab and turn into a PBM font file available from Stewart C Russell. There are fonts with various duodecimal digit glyphs at treisara.deviantart.com.
PBM fonts cannot be used with -wchar.
In the past, English text was encoded in 7-bit ASCII. 8-bit and multibyte encodings were needed only for non-English languages. This is not the case today. As of this writing, 90% of all web pages are encoded in UTF-8. While many of them are actually restricted to 7-bit ASCII, which is a subset of UTF-8, English text encoded in UTF-8 commonly employs "66 99" style quotation marks, which do not appear in ASCII.
If your input text is UTF-8, you should use -wchar. You may have to tweak the locale setting. pbmtext recognizes code points up to 65535. This is sufficient for the vast majority of text written in modern languages.
In the default single-byte (or "narrow") character mode, pbmtext can handle 7-bit and 8-bit character sets. Examples are ASCII, ISO 8859 family, koi8-r/u and VISCII. It is up to the user to supply a BDF file covering the necessary glyphs with the "-font" option. The font file must be in the right encoding.
pbmtext does not inspect the encoding of the font file.
If the text is from Standard Input, no line may be longer than 4999 characters. If one is, the program aborts with an appropriate error message.
If the text is from Standard Input and contains a null character, the results are abnormal. Lines may be truncated, and a single line may be considered multiple lines. Normal text does not contain null characters, so this isn't a big problem.
If you get garbled output, check the input text encoding and font file encoding. When using -wchar, also check the current locale.
To convert the encoding of a text file, use iconv or luit.
To check the encoding of a BDF file, examine the CHARSET_REGISTRY line and the next line, which should be CHARSET_ENCODING:
$ grep -A1 CHARSET_REGISTRY font-a.bdf
CHARSET_REGISTRY "ISO8859"
CHARSET_ENCODING "1"
$ grep -A1 CHARSET_REGISTRY font-b.bdf
CHARSET_REGISTRY "ISO10646"
CHARSET_ENCODING "1"
The latter is Unicode. BDF files coded in ISO 16046-1 usually work for Western European languages, because ISO 16046-1 expands ISO 8859-1 (also called "Latin-1") while maintaining the first 256 code points. ISO 8859-1 itself is a superset of ASCII. Run the above command and verify the necessary glyphs are present.
IMPORTANT: For input text, a different rule applies. If you feed ISO 8859-1 text to pbmtext -wchar set up for UTF-8, the output will be garbled. Unicode provides several encoding schemes and different ones are in effect for input text and font. The difference between Unicode codepoint and the various encodings is a formidable stumbling block; beware of web pages that get the concept wrong.
75% of the BDF files in the font collection available from the Netpbm website are in ISO 10646-1. Many have the Euro sign, Greek letters, etc., but they are placed at code points available to pbmtext only with -wchar.
Before pbmtext had the -wchar option, one often had to produce a BDF file in an 8-bit encoding from a master BDF file encoded in ISO 10646-1.
There are several programs that perform BDF encoding conversion. If you have the X Window System installed, first look for ucs2any. If you don't, you can download ucs2any.pl from Unicode fonts and tools for X11. This website has much useful information on fonts.
Another converter is trbdf, included in the "trscripts" package, available in some GNU/Linux distributions.
BDF files encoded in ISO 8859-2, ISO 8859-7, koi8-r, etc. are available from ISO 8859 Alphabet Soup and its sister page The Cyrillic Charset Soup. Though the information is dated, these pages give a good overview of 8-bit character sets.
To convert OTF or TTF font files to BDF, use otf2bdf by Mike Leisher or otftobdf (part of the libotf package, which is part of the m17n project).