[comment {-*- tcl -*- doctools manpage}] [manpage_begin doctools::tcl::parse n 1] [copyright {2009 Andreas Kupries }] [moddesc {Documentation tools}] [titledesc {Processing text in 'subst -novariables' format}] [category {Documentation tools}] [keywords doctools parser subst command word {Tcl syntax}] [require Tcl 8.4] [require snit] [require fileutil] [require logger] [require struct::list] [require struct::stack] [require struct::set] [require treeql] [description] This package provides commands for parsing text with embedded Tcl commands as accepted by the Tcl builtin command [cmd {subst -novariables}]. The result of the parsing is an abstract syntax tree. [para] This is an internal package of doctools, for use by the higher level parsers processing the [term docidx], [term doctoc], and [term doctools] markup languages. [section API] [list_begin definitions] [call [cmd ::doctools::tcl::parse] [method text] \ [arg tree] [arg text] [opt [arg root]]] The command takes the [arg text] and parses it under the assumption that it contains a string acceptable to the Tcl builtin command [cmd {subst -novariables}]. Errors are thrown otherwise during the parsing. The format used for these errors in described in section [sectref {Error format}]. [para] The command returns the empty string as it result. The actual result of the parsing is entered into the tree structure [arg tree], under the node [arg root]. If [arg root] is not specified the root of [arg tree] is used. The [arg tree] has to exist and be the command of a tree object which supports the same methods as trees created by the package [package struct::tree]. [para] In case of errors [arg tree] will be left in an undefined state. [call [cmd ::doctools::tcl::parse] [method file] \ [arg tree] [arg path] [opt [arg root]]] The same as [method text], except that the text to parse is read from the file specified by [arg path]. [list_end] [section {Error format}] When the parser encounters a problem in the input it will throw an error using the format described here. [list_begin enumerated] [enum] The message will contain the reason for the problem (unexpected character or end of input in input), the character in question, if any, and the line and column the problem was found at, in a human readable form. This part is not documented further as its format may change as we see fit. It is intended for human consumption, not machine. [enum] The error code however will contain a machine-readable representation of the problem, in the form of a 5-element list containing, in the order listed below [list_begin enumerated] [enum] the constant string [const doctools::tcl::parse] [enum] the cause of the problem, one of [list_begin definitions] [def [const char]] Unexpected character in input [def [const eof]] Unexpected end of the input [list_end] [enum] The location of the problem as offset from the beginning of the input, counted in characters. Note: Line markers count as one character. [enum] The line the problem was found on (counted from 1 (one)), [enum] The column the problem was found at (counted from 0 (zero)) [list_end] [list_end] [section {Tree Structure}] After successfully parsing a string the generated tree will have the following structure: [list_begin enumerated] [enum] In the following items the word 'root' refers to the node which was specified as the root of the tree when invoking either [method text] or [method file]. This may be the actual root of the tree. [enum] All the following items further ignore the possibility of pre-existing attributes in the pre-existing nodes. If attributes exists with the same names as the attributes used by the parser the pre-existing values are written over. Attributes with names not clashing with the parser's attributes are not touched. [enum] The root node has no attributes. [enum] All other nodes have the attributes [list_begin definitions] [def type] The value is a string from the set { Command , Text , Word } [def range] The value is either empty or a 2-element list containing integer numbers. The numbers are the offsets of the first and last character in the input text, of the token described by the node,. [def line] The value is an integer, it describes the line in the input the token described by the node ends on. Lines are counted from 1 ([const one]). [def col] The value is an integer, it describes the column in the line in the input the token described by the node ends on. Columns are counted from 0 ([const zero]). [list_end] [enum] The children of the root, if any, are of type Command and Text, in semi-alternation. This means: After a Text node a Command node has to follow, and anything can follow a Command node, a Text or other Command node. [enum] The children of a Command node, if any, are of type Command, and Text, and Word, they describe the arguments of the command. [enum] The children of a Word node, if any, are of type Command, Text, in semi-alternation. This means: After a Text node a Command node has to follow, and anything can follow a Command node, a Text or other Command node. [enum] A Word node without children represents the empty string. [enum] All Text nodes are leaves of the tree. [enum] All leaves of the tree are either Text or Command nodes. Word nodes cannot be leaves. [list_end] [vset CATEGORY doctools] [include include/feedback.inc] [manpage_end]