Pretty-printing.
This module implements a pretty-printing facility to format values within 'pretty-printing boxes' combined with a set of printf-like functions. The pretty-printer splits lines at specified break hints, and indents lines according to the box structure.
This pretty-printing facility is implemented as an overlay on top of abstract formatters which provide basic output functions. Some formatters are predefined, notably:
std_formatter outputs to stdouterr_formatter outputs to stderr
Most functions in the Format module come in two variants: a short version that operates on std_formatter and the generic version prefixed by pp_ that takes a formatter as its first argument.
More formatters can be created with formatter_of_out_channel, formatter_of_buffer, formatter_of_symbolic_output_buffer or using custom formatters.
Introduction
For a gentle introduction to the basics of pretty-printing using Format, read http://caml.inria.fr/resources/doc/guides/format.en.html.
You may consider this module as providing an extension to the printf facility to provide automatic line splitting. The addition of pretty-printing annotations to your regular printf format strings gives you fancy indentation and line breaks. Pretty-printing annotations are described below in the documentation of the function Format.fprintf.
You may also use the explicit pretty-printing box management and printing functions provided by this module. This style is more basic but more verbose than the concise fprintf format strings.
For instance, the sequence open_box 0; print_string "x ="; print_space ();
print_int 1; close_box (); print_newline () that prints x = 1 within a pretty-printing box, can be abbreviated as printf "@[%s@ %i@]@." "x =" 1, or even shorter printf "@[x =@ %i@]@." 1.
Rule of thumb for casual users of this library:
- use simple pretty-printing boxes (as obtained by
open_box 0); - use simple break hints as obtained by
print_cut () that outputs a simple break hint, or by print_space () that outputs a space indicating a break hint; - once a pretty-printing box is open, display its material with basic printing functions (e. g.
print_int and print_string); - when the material for a pretty-printing box has been printed, call
close_box () to close the box; - at the end of pretty-printing, flush the pretty-printer to display all the remaining material, e.g. evaluate
print_newline ().
The behavior of pretty-printing commands is unspecified if there is no open pretty-printing box. Each box opened by one of the open_ functions below must be closed using close_box for proper formatting. Otherwise, some of the material printed in the boxes may not be output, or may be formatted incorrectly.
In case of interactive use, each phrase is executed in the initial state of the standard pretty-printer: after each phrase execution, the interactive system closes all open pretty-printing boxes, flushes all pending text, and resets the standard pretty-printer.
Warning: mixing calls to pretty-printing functions of this module with calls to Pervasives low level output functions is error prone.
The pretty-printing functions output material that is delayed in the pretty-printer queue and stacks in order to compute proper line splitting. In contrast, basic I/O output functions write directly in their output device. As a consequence, the output of a basic I/O function may appear before the output of a pretty-printing function that has been called before. For instance,
Pervasives.print_string "<";
Format.print_string "PRETTY";
Pervasives.print_string ">";
Format.print_string "TEXT";
leads to output <>PRETTYTEXT.
Abstract data corresponding to a pretty-printer (also called a formatter) and all its machinery. See also formatter.
Pretty-printing boxes
The pretty-printing engine uses the concepts of pretty-printing box and break hint to drive indentation and line splitting behavior of the pretty-printer.
Each different pretty-printing box kind introduces a specific line splitting policy:
- within an horizontal box, break hints never split the line (but the line may be split in a box nested deeper),
- within a vertical box, break hints always split the line,
- within an horizontal/vertical box, if the box fits on the current line then break hints never split the line, otherwise break hint always split the line,
- within a compacting box, a break hint never splits the line, unless there is no more room on the current line.
Note that line splitting policy is box specific: the policy of a box does not rule the policy of inner boxes. For instance, if a vertical box is nested in an horizontal box, all break hints within the vertical box will split the line.
pp_open_box ppf d opens a new compacting pretty-printing box with offset d in the formatter ppf.
Within this box, the pretty-printer prints as much as possible material on every line.
A break hint splits the line if there is no more room on the line to print the remainder of the box.
Within this box, the pretty-printer emphasizes the box structure: a break hint also splits the line if the splitting ``moves to the left'' (i.e. the new line gets an indentation smaller than the one of the current line).
This box is the general purpose pretty-printing box.
If the pretty-printer splits the line in the box, offset d is added to the current indentation.
Closes the most recently open pretty-printing box.
pp_open_hbox ppf () opens a new 'horizontal' pretty-printing box.
This box prints material on a single line.
Break hints in a horizontal box never split the line. (Line splitting may still occur inside boxes nested deeper).
pp_open_vbox ppf d opens a new 'vertical' pretty-printing box with offset d.
This box prints material on as many lines as break hints in the box.
Every break hint in a vertical box splits the line.
If the pretty-printer splits the line in the box, d is added to the current indentation.
pp_open_hvbox ppf d opens a new 'horizontal/vertical' pretty-printing box with offset d.
This box behaves as an horizontal box if it fits on a single line, otherwise it behaves as a vertical box.
If the pretty-printer splits the line in the box, d is added to the current indentation.
val pp_open_hovbox : formatter -> int -> unitpp_open_hovbox ppf d opens a new 'horizontal-or-vertical' pretty-printing box with offset d.
This box prints material as much as possible on every line.
A break hint splits the line if there is no more room on the line to print the remainder of the box.
If the pretty-printer splits the line in the box, d is added to the current indentation.
val pp_print_string : formatter -> string -> unitpp_print_string ppf s prints s in the current pretty-printing box.
val pp_print_as : formatter -> int -> string -> unitpp_print_as ppf len s prints s in the current pretty-printing box. The pretty-printer formats s as if it were of length len.
Print an integer in the current pretty-printing box.
val pp_print_float : formatter -> float -> unitPrint a floating point number in the current pretty-printing box.
val pp_print_char : formatter -> char -> unitPrint a character in the current pretty-printing box.
val pp_print_bool : formatter -> bool -> unitPrint a boolean in the current pretty-printing box.
Break hints
A 'break hint' tells the pretty-printer to output some space or split the line whichever way is more appropriate to the current pretty-printing box splitting rules.
Break hints are used to separate printing items and are mandatory to let the pretty-printer correctly split lines and indent items.
Simple break hints are:
- the 'space': output a space or split the line if appropriate,
- the 'cut': split the line if appropriate.
Note: the notions of space and line splitting are abstract for the pretty-printing engine, since those notions can be completely redefined by the programmer. However, in the pretty-printer default setting, ``output a space'' simply means printing a space character (ASCII code 32) and ``split the line'' means printing a newline character (ASCII code 10).
val pp_print_space : formatter -> unit -> unitpp_print_space ppf () emits a 'space' break hint: the pretty-printer may split the line at this point, otherwise it prints one space.
pp_print_space ppf () is equivalent to pp_print_break ppf 1 0.
pp_print_cut ppf () emits a 'cut' break hint: the pretty-printer may split the line at this point, otherwise it prints nothing.
pp_print_cut ppf () is equivalent to pp_print_break ppf 0 0.
val pp_print_break : formatter -> int -> int -> unitpp_print_break ppf nspaces offset emits a 'full' break hint: the pretty-printer may split the line at this point, otherwise it prints nspaces spaces.
If the pretty-printer splits the line, offset is added to the current indentation.
val pp_force_newline : formatter -> unit -> unitForce a new line in the current pretty-printing box.
The pretty-printer must split the line at this point,
Not the normal way of pretty-printing, since imperative line splitting may interfere with current line counters and box size calculation. Using break hints within an enclosing vertical box is a better alternative.
val pp_print_if_newline : formatter -> unit -> unitExecute the next formatting command if the preceding line has just been split. Otherwise, ignore the next formatting command.
Pretty-printing termination
val pp_print_flush : formatter -> unit -> unitEnd of pretty-printing: resets the pretty-printer to initial state.
All open pretty-printing boxes are closed, all pending text is printed. In addition, the pretty-printer low level output device is flushed to ensure that all pending text is really displayed.
Note: never use print_flush in the normal course of a pretty-printing routine, since the pretty-printer uses a complex buffering machinery to properly indent the output; manually flushing those buffers at random would conflict with the pretty-printer strategy and result to poor rendering.
Only consider using print_flush when displaying all pending material is mandatory (for instance in case of interactive use when you want the user to read some text) and when resetting the pretty-printer state will not disturb further pretty-printing.
Warning: If the output device of the pretty-printer is an output channel, repeated calls to print_flush means repeated calls to Pervasives.flush to flush the out channel; these explicit flush calls could foil the buffering strategy of output channels and could dramatically impact efficiency.
val pp_print_newline : formatter -> unit -> unitEnd of pretty-printing: resets the pretty-printer to initial state.
All open pretty-printing boxes are closed, all pending text is printed.
Equivalent to print_flush followed by a new line. See corresponding words of caution for print_flush.
Note: this is not the normal way to output a new line; the preferred method is using break hints within a vertical pretty-printing box.
Margin
pp_set_margin ppf d sets the right margin to d (in characters): the pretty-printer splits lines that overflow the right margin according to the break hints given. Nothing happens if d is smaller than 2. If d is too large, the right margin is set to the maximum admissible value (which is greater than 10 ^ 9). If d is less than the current maximum indentation limit, the maximum indentation limit is decreased while trying to preserve a minimal ratio max_indent/margin>=50% and if possible the current difference margin - max_indent.
Returns the position of the right margin.
Maximum indentation limit
val pp_set_max_indent : formatter -> int -> unitpp_set_max_indent ppf d sets the maximum indentation limit of lines to d (in characters): once this limit is reached, new pretty-printing boxes are rejected to the left, if they do not fit on the current line.
Nothing happens if d is smaller than 2. If d is too large, the limit is set to the maximum admissible value (which is greater than 10 ^ 9).
If d is greater or equal than the current margin, it is ignored, and the current maximum indentation limit is kept.
val pp_get_max_indent : formatter -> unit -> intReturn the maximum indentation limit (in characters).
The maximum formatting depth is the maximum number of pretty-printing boxes simultaneously open.
Material inside boxes nested deeper is printed as an ellipsis (more precisely as the text returned by get_ellipsis_text ()).
val pp_set_max_boxes : formatter -> int -> unitpp_set_max_boxes ppf max sets the maximum number of pretty-printing boxes simultaneously open.
Material inside boxes nested deeper is printed as an ellipsis (more precisely as the text returned by get_ellipsis_text ()).
Nothing happens if max is smaller than 2.
val pp_get_max_boxes : formatter -> unit -> intReturns the maximum number of pretty-printing boxes allowed before ellipsis.
val pp_over_max_boxes : formatter -> unit -> boolTests if the maximum number of pretty-printing boxes allowed have already been opened.
Tabulation boxes
A tabulation box prints material on lines divided into cells of fixed length. A tabulation box provides a simple way to display vertical columns of left adjusted text.
This box features command set_tab to define cell boundaries, and command print_tab to move from cell to cell and split the line when there is no more cells to print on the line.
Note: printing within tabulation box is line directed, so arbitrary line splitting inside a tabulation box leads to poor rendering. Yet, controlled use of tabulation boxes allows simple printing of columns within module Format.
open_tbox () opens a new tabulation box.
This box prints lines separated into cells of fixed width.
Inside a tabulation box, special tabulation markers defines points of interest on the line (for instance to delimit cell boundaries). Function Format.set_tab sets a tabulation marker at insertion point.
A tabulation box features specific tabulation breaks to move to next tabulation marker or split the line. Function Format.print_tbreak prints a tabulation break.
val pp_close_tbox : formatter -> unit -> unitCloses the most recently opened tabulation box.
Sets a tabulation marker at current insertion point.
print_tab () emits a 'next' tabulation break hint: if not already set on a tabulation marker, the insertion point moves to the first tabulation marker on the right, or the pretty-printer splits the line and insertion point moves to the leftmost tabulation marker.
It is equivalent to print_tbreak 0 0.
val pp_print_tbreak : formatter -> int -> int -> unitprint_tbreak nspaces offset emits a 'full' tabulation break hint.
If not already set on a tabulation marker, the insertion point moves to the first tabulation marker on the right and the pretty-printer prints nspaces spaces.
If there is no next tabulation marker on the right, the pretty-printer splits the line at this point, then insertion point moves to the leftmost tabulation marker of the box.
If the pretty-printer splits the line, offset is added to the current indentation.
Ellipsis
val pp_set_ellipsis_text : formatter -> string -> unitSet the text of the ellipsis printed when too many pretty-printing boxes are open (a single dot, ., by default).
val pp_get_ellipsis_text : formatter -> unit -> stringReturn the text of the ellipsis.
pp_print_list ?pp_sep pp_v ppf l prints items of list l, using pp_v to print each item, and calling pp_sep between items (pp_sep defaults to pp_print_cut. Does nothing on empty lists.
val pp_print_text : formatter -> string -> unitFormatted pretty-printing
Module Format provides a complete set of printf like functions for pretty-printing using format string specifications.
Specific annotations may be added in the format strings to give pretty-printing commands to the pretty-printing engine.
Those annotations are introduced in the format strings using the @ character. For instance, @ means a space break, @, means a cut, @[ opens a new box, and @] closes the last open box.
fprintf ff fmt arg1 ... argN formats the arguments arg1 to argN according to the format string fmt, and outputs the resulting string on the formatter ff.
The format string fmt is a character string which contains three types of objects: plain characters and conversion specifications as specified in the Printf module, and pretty-printing indications specific to the Format module.
The pretty-printing indication characters are introduced by a @ character, and their meanings are:
@[: open a pretty-printing box. The type and offset of the box may be optionally specified with the following syntax: the < character, followed by an optional box type indication, then an optional integer offset, and the closing > character. Pretty-printing box type is one of h, v, hv, b, or hov. 'h' stands for an 'horizontal' pretty-printing box, 'v' stands for a 'vertical' pretty-printing box, 'hv' stands for an 'horizontal/vertical' pretty-printing box, 'b' stands for an 'horizontal-or-vertical' pretty-printing box demonstrating indentation, 'hov' stands a simple 'horizontal-or-vertical' pretty-printing box. For instance, @[<hov 2> opens an 'horizontal-or-vertical' pretty-printing box with indentation 2 as obtained with open_hovbox 2. For more details about pretty-printing boxes, see the various box opening functions open_*box.@]: close the most recently opened pretty-printing box.@,: output a 'cut' break hint, as with print_cut ().@ : output a 'space' break hint, as with print_space ().@;: output a 'full' break hint as with print_break. The nspaces and offset parameters of the break hint may be optionally specified with the following syntax: the < character, followed by an integer nspaces value, then an integer offset, and a closing > character. If no parameters are provided, the good break defaults to a 'space' break hint.@.: flush the pretty-printer and split the line, as with print_newline ().@<n>: print the following item as if it were of length n. Hence, printf "@<0>%s" arg prints arg as a zero length string. If @<n> is not followed by a conversion specification, then the following character of the format is printed as if it were of length n.@?: flush the pretty-printer as with print_flush (). This is equivalent to the conversion %!.@\n: force a newline, as with force_newline (), not the normal way of pretty-printing, you should prefer using break hints inside a vertical pretty-printing box.
Note: To prevent the interpretation of a @ character as a pretty-printing indication, escape it with a % character. Old quotation mode @@ is deprecated since it is not compatible with formatted input interpretation of character '@'.
Example: printf "@[%s@ %d@]@." "x =" 1 is equivalent to open_box (); print_string "x ="; print_space ();
print_int 1; close_box (); print_newline (). It prints x = 1 within a pretty-printing 'horizontal-or-vertical' box.
Same as printf above, but instead of printing on a formatter, returns a string containing the result of formatting the arguments. Note that the pretty-printer queue is flushed at the end of each call to sprintf.
In case of multiple and related calls to sprintf to output material on a single string, you should consider using fprintf with the predefined formatter str_formatter and call flush_str_formatter () to get the final result.
Alternatively, you can use Format.fprintf with a formatter writing to a buffer of your own: flushing the formatter and the buffer at the end of pretty-printing returns the desired string.
Same as printf above, but instead of printing on a formatter, returns a string containing the result of formatting the arguments. The type of asprintf is general enough to interact nicely with %a conversions.
Same as fprintf above, but does not print anything. Useful to ignore some material when conditionally printing.
Formatted Pretty-Printing with continuations.
Same as fprintf above, but instead of returning immediately, passes the formatter to its first argument at the end of printing.
Same as kfprintf above, but does not print anything. Useful to ignore some material when conditionally printing.
Same as sprintf above, but instead of returning the string, passes it to the first argument.
Same as asprintf above, but instead of returning the string, passes it to the first argument.