.TH prettypr 3erl "syntax_tools 3.0.1" "" "Erlang Module Definition"
.SH NAME
prettypr \- A generic pretty printer library.
.SH DESCRIPTION
.LP
A generic pretty printer library\&. This module uses a strict-style context passing implementation of John Hughes algorithm, described in "The design of a Pretty-printing Library"\&. The paragraph-style formatting, empty documents, floating documents, and null strings are my own additions to the algorithm\&.
.LP
To get started, you should read about the document() data type; the main constructor functions: text/1, above/2, beside/2, nest/2, sep/1, and par/2; and the main layout function format/3\&.
.LP
If you simply want to format a paragraph of plain text, you probably want to use the text_par/2 function, as in the following example:
.LP
.nf
prettypr:format(prettypr:text_par("Lorem ipsum dolor sit amet"), 20)
.fi
.SH "DATA TYPES"
.RS 2
.TP 2
.B
deep_string() = [char() | deep_string()]:
.TP 2
.B
document() = null | #text{s=deep_string()} | #nest{n=integer(), d=document()} | #beside{d1=document(), d2=document()} | #above{d1=document(), d2=document()} | #sep{ds=[document()], i=integer(), p=boolean()} | #float{d=document(), h=integer(), v=integer()} | #union{d1=document(), d2=document()} | #fit{d=document()}:
.RE
.SH EXPORTS
.LP
.B
above(D1::document(), D2::document()) -> #above{d1=document(), d2=document()}
.br
.RS
.LP
Concatenates documents vertically\&. Returns a document representing the concatenation of the documents \fID1\fR\& and \fID2\fR\& such that the first line of \fID2\fR\& follows directly below the last line of \fID1\fR\&, and the first character of \fID2\fR\& is in the same horizontal column as the first character of \fID1\fR\&, in all possible layouts\&.
.LP
Examples:
.LP
.nf
ab cd => ab
cd
abc
abc fgh => de
de ij fgh
ij
.fi
.RE
.LP
.B
beside(D1::document(), D2::document()) -> #beside{d1=document(), d2=document()}
.br
.RS
.LP
Concatenates documents horizontally\&. Returns a document representing the concatenation of the documents \fID1\fR\& and \fID2\fR\& such that the last character of \fID1\fR\& is horizontally adjacent to the first character of \fID2\fR\&, in all possible layouts\&. (Note: any indentation of \fID2\fR\& is lost\&.)
.LP
Examples:
.LP
.nf
ab cd => abcd
ab ef ab
cd gh => cdef
gh
.fi
.RE
.LP
.B
best(D::document(), W::integer(), R::integer()) -> empty | document()
.br
.RS
.LP
Selects a "best" layout for a document, creating a corresponding fixed-layout document\&. If no layout could be produced, the atom \fIempty\fR\& is returned instead\&. For details about \fIPaperWidth\fR\& and \fILineWidth\fR\&, see format/3\&. The function is idempotent\&.
.LP
One possible use of this function is to compute a fixed layout for a document, which can then be included as part of a larger document\&. For example:
.LP
.nf
above(text("Example:"), nest(8, best(D, W - 12, L - 6)))
.fi
.LP
will format \fID\fR\& as a displayed-text example indented by 8, whose right margin is indented by 4 relative to the paper width \fIW\fR\& of the surrounding document, and whose maximum individual line length is shorter by 6 than the line length \fIL\fR\& of the surrounding document\&.
.LP
This function is used by the format/3 function to prepare a document before being laid out as text\&.
.RE
.LP
.B
break(D::document()) -> #above{d1=document(), d2=document()}
.br
.RS
.LP
Forces a line break at the end of the given document\&. This is a utility function; see empty/0 for details\&.
.RE
.LP
.B
empty() -> null
.br
.RS
.LP
Yields the empty document, which has neither height nor width\&. (\fIempty\fR\& is thus different from an empty text string, which has zero width but height 1\&.)
.LP
Empty documents are occasionally useful; in particular, they have the property that \fIabove(X, empty())\fR\& will force a new line after \fIX\fR\& without leaving an empty line below it; since this is a common idiom, the utility function break/1 will place a given document in such a context\&.
.LP
\fISee also:\fR\& text/1\&.
.RE
.LP
.B
floating(D::document()) -> #float{d=document(), h=integer(), v=integer()}
.br
.RS
.LP
Equivalent to floating(D, 0, 0)\&.
.RE
.LP
.B
floating(D::document(), H::integer(), V::integer()) -> #float{d=document(), h=integer(), v=integer()}
.br
.RS
.LP
Creates a "floating" document\&. The result represents the same set of layouts as \fID\fR\&; however, a floating document may be moved relative to other floating documents immediately beside or above it, according to their relative horizontal and vertical priorities\&. These priorities are set with the \fIHp\fR\& and \fIVp\fR\& parameters; if omitted, both default to zero\&.
.LP
Notes: Floating documents appear to work well, but are currently less general than you might wish, losing effect when embedded in certain contexts\&. It is possible to nest floating-operators (even with different priorities), but the effects may be difficult to predict\&. In any case, note that the way the algorithm reorders floating documents amounts to a "bubblesort", so don\&'t expect it to be able to sort large sequences of floating documents quickly\&.
.RE
.LP
.B
follow(D1::document(), D2::document()) -> #beside{d1=document(), d2=document()}
.br
.RS
.LP
Equivalent to follow(D1, D2, 0)\&.
.RE
.LP
.B
follow(D1::document(), D2::document(), N::integer()) -> #beside{d1=document(), d2=document()}
.br
.RS
.LP
Separates two documents by either a single space, or a line break and indentation\&. In other words, one of the layouts
.LP
.nf
abc def
.fi
.LP
or
.LP
.nf
abc
def
.fi
.LP
will be generated, using the optional offset in the latter case\&. This is often useful for typesetting programming language constructs\&.
.LP
This is a utility function; see par/2 for further details\&.
.LP
\fISee also:\fR\& follow/2\&.
.RE
.LP
.B
format(D::document()) -> string()
.br
.RS
.LP
Equivalent to format(D, 80)\&.
.RE
.LP
.B
format(D::document(), W::integer()) -> string()
.br
.RS
.LP
Equivalent to format(D, PaperWidth, 65)\&.
.RE
.LP
.B
format(D::document(), W::integer(), R::integer()) -> string()
.br
.RS
.LP
Computes a layout for a document and returns the corresponding text\&. See document() for further information\&. Throws \fIno_layout\fR\& if no layout could be selected\&.
.LP
\fIPaperWidth\fR\& specifies the total width (in character positions) of the field for which the text is to be laid out\&. \fILineWidth\fR\& specifies the desired maximum width (in number of characters) of the text printed on any single line, disregarding leading and trailing white space\&. These parameters need to be properly balanced in order to produce good layouts\&. By default, \fIPaperWidth\fR\& is 80 and \fILineWidth\fR\& is 65\&.
.LP
\fISee also:\fR\& best/3\&.
.RE
.LP
.B
nest(N::integer(), D::document()) -> document()
.br
.RS
.LP
Indents a document a number of character positions to the right\&. Note that \fIN\fR\& may be negative, shifting the text to the left, or zero, in which case \fID\fR\& is returned unchanged\&.
.RE
.LP
.B
null_text(S::string()) -> #text{s=deep_string()}
.br
.RS
.LP
Similar to text/1, but the result is treated as having zero width\&. This is regardless of the actual length of the string\&. Null text is typically used for markup, which is supposed to have no effect on the actual layout\&.
.LP
The standard example is when formatting source code as HTML to be placed within \fI\&.\&.\&.
\fR\& markup, and using e\&.g\&. \fI\fR\& and \fI\fR\& to make parts of the source code stand out\&. In this case, the markup does not add to the width of the text when viewed in an HTML browser, so the layout engine should simply pretend that the markup has zero width\&.
.LP
\fISee also:\fR\& empty/0, text/1\&.
.RE
.LP
.B
par(Ds::[document()]) -> #sep{ds=[document()], i=integer(), p=boolean()}
.br
.RS
.LP
Equivalent to par(Ds, 0)\&.
.RE
.LP
.B
par(Ds::[document()], N::integer()) -> #sep{ds=[document()], i=integer(), p=boolean()}
.br
.RS
.LP
Arranges documents in a paragraph-like layout\&. Returns a document representing all possible left-aligned paragraph-like layouts of the (nonempty) sequence \fIDocs\fR\& of documents\&. Elements in \fIDocs\fR\& are separated horizontally by a single space character and vertically with a single line break\&. All lines following the first (if any) are indented to the same left column, whose indentation is specified by the optional \fIOffset\fR\& parameter relative to the position of the first element in \fIDocs\fR\&\&. For example, with an offset of -4, the following layout can be produced, for a list of documents representing the numbers 0 to 15:
.LP
.nf
0 1 2 3
4 5 6 7 8 9
10 11 12 13
14 15
.fi
.LP
or with an offset of +2:
.LP
.nf
0 1 2 3 4 5 6
7 8 9 10 11
12 13 14 15
.fi
.LP
The utility function text_par/2 can be used to easily transform a string of text into a \fIpar\fR\& representation by splitting it into words\&.
.LP
Note that whenever a document in \fIDocs\fR\& contains a line break, it will be placed on a separate line\&. Thus, neither a layout such as
.LP
.nf
ab cd
ef
.fi
.LP
nor
.LP
.nf
ab
cd ef
.fi
.LP
will be generated\&. However, a useful idiom for making the former variant possible (when wanted) is \fIbeside(par([D1, text("")], N), D2)\fR\& for two documents \fID1\fR\& and \fID2\fR\&\&. This will break the line between \fID1\fR\& and \fID2\fR\& if \fID1\fR\& contains a line break (or if otherwise necessary), and optionally further indent \fID2\fR\& by \fIN\fR\& character positions\&. The utility function follow/3 creates this context for two documents \fID1\fR\& and \fID2\fR\&, and an optional integer \fIN\fR\&\&.
.LP
\fISee also:\fR\& par/1, text_par/2\&.
.RE
.LP
.B
sep(Ds::[document()]) -> #sep{ds=[document()], i=integer(), p=boolean()}
.br
.RS
.LP
Arranges documents horizontally or vertically, separated by whitespace\&. Returns a document representing two alternative layouts of the (nonempty) sequence \fIDocs\fR\& of documents, such that either all elements in \fIDocs\fR\& are concatenated horizontally, and separated by a space character, or all elements are concatenated vertically (without extra separation)\&.
.LP
Note: If some document in \fIDocs\fR\& contains a line break, the vertical layout will always be selected\&.
.LP
Examples:
.LP
.nf
ab
ab cd ef => ab cd ef | cd
ef
ab ab
cd ef => cd
ef
.fi
.LP
.LP
\fISee also:\fR\& par/2\&.
.RE
.LP
.B
text(S::string()) -> #text{s=deep_string()}
.br
.RS
.LP
Yields a document representing a fixed, unbreakable sequence of characters\&. The string should contain only \fIprintable\fR\& characters (tabs allowed but not recommended), and \fInot\fR\& newline, line feed, vertical tab, etc\&. A tab character (\fI\\t\fR\&) is interpreted as padding of 1-8 space characters to the next column of 8 characters \fIwithin the string\fR\&\&.
.LP
\fISee also:\fR\& empty/0, null_text/1, text_par/2\&.
.RE
.LP
.B
text_par(S::string()) -> document()
.br
.RS
.LP
Equivalent to text_par(Text, 0)\&.
.RE
.LP
.B
text_par(S::string(), N::integer()) -> document()
.br
.RS
.LP
Yields a document representing paragraph-formatted plain text\&. The optional \fIIndentation\fR\& parameter specifies the extra indentation of the first line of the paragraph\&. For example, \fItext_par("Lorem ipsum dolor sit amet", N)\fR\& could represent
.LP
.nf
Lorem ipsum dolor
sit amet
.fi
.LP
if \fIN\fR\& = 0, or
.LP
.nf
Lorem ipsum
dolor sit amet
.fi
.LP
if \fIN\fR\& = 2, or
.LP
.nf
Lorem ipsum dolor
sit amet
.fi
.LP
if \fIN\fR\& = -2\&.
.LP
(The sign of the indentation is thus reversed compared to the par/2 function, and the behaviour varies slightly depending on the sign in order to match the expected layout of a paragraph of text\&.)
.LP
Note that this is just a utility function, which does all the work of splitting the given string into words separated by whitespace and setting up a par with the proper indentation, containing a list of text elements\&.
.LP
\fISee also:\fR\& par/2, text/1, text_par/1\&.
.RE
.SH AUTHORS
.LP
Richard Carlsson
.I