The application described by this document, page, is
actually not just a parser generator, as the name implies, but a generic
tool for the execution of arbitrary transformations on texts.
Its genericity comes through the use of plugins for
reading, transforming, and writing data, and the predefined set of plugins
provided by Tcllib is for the generation of memoizing recursive descent
parsers (aka packrat parsers) from grammar specifications (Parsing
Expression Grammars).
page is written on top of the package
page::pluginmgr, wrapping its functionality into a command line based
application. All the other page::* packages are plugin and/or
supporting packages for the generation of parsers. The parsers themselves
are based on the packages grammar::peg, grammar::peg::interp,
and grammar::mengine.
- page
?options...? ?input ?output??
- This is general form for calling page. The application will read
the contents of the file input, process them under the control of
the specified options, and then write the result to the file
output.
If input is the string - the data to process
will be read from stdin instead of a file. Analogously the result
will be written to stdout instead of a file if output is
the string -. A missing output or input specification causes the
application to assume -.
The detailed specifications of the recognized options
are provided in section OPTIONS.
- path input
(in)
- This argument specifies the path to the file to be processed by the
application, or -. The last value causes the application to read
the text from stdin. Otherwise it has to exist, and be readable. If
the argument is missing - is assumed.
- path output
(in)
- This argument specifies where to write the generated text. It can be the
path to a file, or -. The last value causes the application to
write the generated documented to stdout.
If the file output does not exist then [file dirname
$output] has to exist and must be a writable directory, as the
application will create the fileto write to.
If the argument is missing - is assumed.
... reading ... transforming ... writing - plugins - pipeline
...
This section describes all the options available to the user of
the application. Options are always processed in order. I.e. of both
--help and --version are specified the option encountered
first has precedence.
Unknown options specified before any of the options -rd,
-wr, or -tr will cause processing to abort with an error.
Unknown options coming in between these options, or after the last of them
are assumed to always take a single argument and are associated with the
last plugin option coming before them. They will be checked after all the
relevant plugins, and thus the options they understand, are known. I.e. such
unknown options cause error if and only if the plugin option they are
associated with does not understand them, and was not superceded by a plugin
option coming after.
Default options are used if and only if the command line did not
contain any options at all. They will set the application up as a PEG-based
parser generator. The exact list of options is
And now the recognized options and their arguments, if they have
any:
- --help
- -h
- -?
- When one of these options is found on the command line all arguments
coming before or after are ignored. The application will print a short
description of the recognized options and exit.
- --version
- -V
- When one of these options is found on the command line all arguments
coming before or after are ignored. The application will print its own
revision and exit.
- -P
- This option signals the application to activate visual feedback while
reading the input.
- -T
- This option signals the application to collect statistics while reading
the input and to print them after reading has completed, before processing
started.
- -D
- This option signals the application to activate logging in the Safe base,
for the debugging of problems with plugins.
- -r parser
- -rd parser
- --reader
parser
- These options specify the plugin the application has to use for reading
the input. If the options are used multiple times the last one will
be used.
- -w generator
- -wr generator
- --writer
generator
- These options specify the plugin the application has to use for generating
and writing the final output. If the options are used multiple
times the last one will be used.
- -t process
- -tr process
- --transform
process
- These options specify a plugin to run on the input. In contrast to readers
and writers each use will not supersede previous uses, but add each
chosen plugin to a list of transformations, either at the front, or the
end, per the last seen use of either option -p or -a. The
initial default is to append the new transformations.
- -a
- --append
- These options signal the application that all following transformations
should be added at the end of the list of transformations.
- -p
- --prepend
- These options signal the application that all following transformations
should be added at the beginning of the list of transformations.
- --reset
- This option signals the application to clear the list of transformations.
This is necessary to wipe out the default transformations used.
- -c file
- --configuration
file
- This option causes the application to load a configuration file and/or
plugin. This is a plugin which in essence provides a pre-defined set of
commandline options. They are processed exactly as if they have been
specified in place of the option and its arguments. This means that
unknown options found at the beginning of the configuration file are
associated with the last plugin, even if that plugin was specified before
the configuration file itself. Conversely, unknown options coming after
the configuration file can be associated with a plugin specified in the
file.
If the argument is a file which cannot be loaded as a plugin
the application will assume that its contents are a list of options and
their arguments, separated by space, tabs, and newlines. Options and
argumentes containing spaces can be quoted via double-quotes (")
and quotes ('). The quote character can be specified within in a quoted
string by doubling it. Newlines in a quoted string are accepted as
is.
page makes use of four different types of plugins, namely:
readers, writers, transformations, and configurations. Here we provide only
a basic introduction on how to use them from page. The exact APIs
provided to and expected from the plugins can be found in the documentation
for page::pluginmgr, for those who wish to write their own
plugins.
Plugins are specified as arguments to the options -r,
-w, -t, -c, and their equivalent longer forms. See the
section OPTIONS for reference.
Each such argument will be first treated as the name of a file and
this file is loaded as the plugin. If however there is no file with that
name, then it will be translated into the name of a package, and this
package is then loaded. For each type of plugins the package management
searches not only the regular paths, but a set application- and
type-specific paths as well. Please see the section PLUGIN LOCATIONS
for a listing of all paths and their sources.
- -c name
- Configurations. The name of the package for the plugin name is
"page::config::name".
We have one predefined plugin:
- peg
- It sets the application up as a parser generator accepting parsing
expression grammars and writing a packrat parser in Tcl. The actual
arguments it specifies are:
--reset
--append
--reader peg
--transform reach
--transform use
--writer me
- -r name
- Readers. The name of the package for the plugin name is
"page::reader::name".
We have five predefined plugins:
- peg
- Interprets the input as a parsing expression grammar (PEG) and
generates a tree representation for it. Both the syntax of PEGs and the
structure of the tree representation are explained in their own
manpages.
- hb
- Interprets the input as Tcl code as generated by the writer plugin
hb and generates its tree representation.
- ser
- Interprets the input as the serialization of a PEG, as generated by the
writer plugin ser, using the package grammar::peg.
- lemon
- Interprets the input as a grammar specification as understood by Richard
Hipp's LEMON parser generator and generates a tree representation
for it. Both the input syntax and the structure of the tree representation
are explained in their own manpages.
- treeser
- Interprets the input as the serialization of a struct::tree. It is
validated as such, but nothing else. It is not assumed to be the
tree representation of a grammar.
- -w name
- Writers. The name of the package for the plugin name is
"page::writer::name".
We have eight predefined plugins:
- identity
- Simply writes the incoming data as it is, without making any changes. This
is good for inspecting the raw result of a reader or transformation.
- null
- Generates nothing, and ignores the incoming data structure.
- tree
- Assumes that the incoming data structure is a struct::tree and
generates an indented textual representation of all nodes, their parental
relationships, and their attribute information.
- peg
- Assumes that the incoming data structure is a tree representation of a
PEG or other other grammar and writes it out as a PEG. The result
is nicely formatted and partially simplified (strings as sequences of
characters). A pretty printer in essence, but can also be used to obtain a
canonical representation of the input grammar.
- tpc
- Assumes that the incoming data structure is a tree representation of a
PEG or other other grammar and writes out Tcl code defining a
package which defines a grammar::peg object containing the grammar
when it is loaded into an interpreter.
- hb
- This is like the writer plugin tpc, but it writes only the
statements which define stat expression and grammar rules. The code making
the result a package is left out.
- ser
- Assumes that the incoming data structure is a tree representation of a
PEG or other other grammar, transforms it internally into a
grammar::peg object and writes out its serialization.
- me
- Assumes that the incoming data structure is a tree representation of a
PEG or other other grammar and writes out Tcl code defining a
package which implements a memoizing recursive descent parser based on the
match engine (ME) provided by the package grammar::mengine.
- -t name
- Transformers. The name of the package for the plugin name is
"page::transform::name".
We have two predefined plugins:
- reach
- Assumes that the incoming data structure is a tree representation of a
PEG or other other grammar. It determines which nonterminal symbols
and rules are reachable from start-symbol/expression. All nonterminal
symbols which were not reached are removed.
- use
- Assumes that the incoming data structure is a tree representation of a
PEG or other other grammar. It determines which nonterminal symbols
and rules are able to generate a finite sequences of terminal
symbols (in the sense for a Context Free Grammar). All nonterminal symbols
which were not deemed useful in this sense are removed.
The application-specific paths searched by page either are,
or come from:
- [1]
- The directory "~/.page/plugin"
- [2]
- The environment variable PAGE_PLUGINS
- [3]
- The registry entry HKEY_LOCAL_MACHINE\SOFTWARE\PAGE\PLUGINS
- [4]
- The registry entry HKEY_CURRENT_USER\SOFTWARE\PAGE\PLUGINS
The type-specific paths searched by page either are, or
come from:
- [1]
- The directory "~/.page/plugin/<TYPE>"
- [2]
- The environment variable PAGE_<TYPE>_PLUGINS
- [3]
- The registry entry
HKEY_LOCAL_MACHINE\SOFTWARE\PAGE\<TYPE>\PLUGINS
- [4]
- The registry entry
HKEY_CURRENT_USER\SOFTWARE\PAGE\<TYPE>\PLUGINS
Where the placeholder <TYPE> is always one of the
values below, properly capitalized.
- [1]
- reader
- [2]
- writer
- [3]
- transform
- [4]
- config
The registry entries are specific to the Windows(tm) platform, all
other platforms will ignore them.
The contents of both environment variables and registry entries
are interpreted as a list of paths, with the elements separated by either
colon (Unix), or semicolon (Windows).
This document, and the package it describes, will undoubtedly
contain bugs and other problems. Please report such in the category
page of the Tcllib Trackers
[http://core.tcl.tk/tcllib/reportlist]. Please also report any ideas for
enhancements you may have for either package and/or documentation.
When proposing code changes, please provide unified diffs,
i.e the output of diff -u.
Note further that attachments are strongly preferred over
inlined patches. Attachments can be made by going to the Edit form of
the ticket immediately after its creation, and then using the left-most
button in the secondary navigation bar.