wysiib / clpstr

ConString

A string constraint solver for SWI-Prolog having interfaces to CLP(FD), CLP(B) and CLP(R).

User Information

Regular Expression Dialect

The employed alphabet consists of ASCII characters and some special characters like umlauts and accented characters. Additionally, dynamic character matching is possible by specifying ranges inside square brackets, e.g., [0-9a-f], or by using the dot operator . which matches with any character. Parentheses can be used to enforce operator precedence. We represent a whitespace character in regular expressions by \s while actual whitespace characters can be used to structure regular expressions without being part of the accepted language. Further, we support the usual regular expression operators on characters, i.e., quantity operators (*, + and ?) and the alternative choice operator (|). On top of those, our regular expressions offer more strict repetition definitions for convenience, noted by {n} (exactly n times), {m,n} (m to n times) and {m,+} (at least m times).

Supported Constraints

• str_in/2: membership to a string domain, e.g., str_in(S, "[0-9]*"

• str_size/2: exact string length

• str_max_size/2: upper bound string length

• str_concatenation/3: string concatenation

• str_repeat/[2, 3, 4]: finite string repetition

• str_union/3: string union, i.e., third argument string variable accept strings that are in one of both domains

• str_intersection/3: string intersection

• str_prefix/2, str_suffix/2, str_infix/2: e.g., str_prefix(S, Prefix) where Prefix can be a constant string or a string variable itself

• str_upper_case/1, str_lower_case/1: set the string to accept upper/lower case characters only

• str_diff/2: inequality between two string variables

• str_all_diff/1: pairwise inequality for a list of string variables

• str_to_int/2: bidirectional string to integer conversion integrating SWI-Prologs CLP(FD) library

  • any constraint of library(clpfd) can be used on finite domain integer variables
  • does not accept leading zeros, e.g., str_to_int("01", 1) does not hold

• str_to_intl/2: same as str_to_int/2 but accepting leading zeros

• str_to_real/2: bidirectional string to real conversion integrating SWI-Prologs CLP(R) library

  • any constraint of library(clpr) can be used on floating point numbers
  • does not accept leading zeros

• str_to_reall/2: same as str_to_real/2 but accepting leading zeros

• str_to_bool/2: bidirectional string to boolean conversion integrating SWI-Prolog's CLP(B) library

  • any constraint of library(clpb) can be used on boolean variables
  • does not accept leading zeros

• str_to_booll/2: same as str_to_bool/2 but accepting leading zeros

• str_labeling/2: labeling with options (CLP(FD) options can be used, too)

  • supported uninformed search strategies are depth-first search (dfs), breadth-first search (bfs) and iterative deepening depth-first search (idfs)

• str_label/1: labeling with default options, i.e., dfs

Developer Information

Domain Representation

ConString works on regular expressions.

There are two types of domains implemented in ConString: string_dom(STRING) represents a constant string.

Regular expressions are represented as finite automata.

automaton_dom(STATES,DELTA,START,END) is the representation of an automaton. STATES is the list of states. cplstr states are a list of ordered integers, e.g. [1,2,3] would be a valid list of states. DELTA is the transition function of the automaton. In ConString it is implemented as a Prolog list containing a triple of the state transitions. A valid transition would be (1,epsilon,2) START is the list of valid start states and END a list of valid final states. Both are subsets of STATES.

It is not recommended to construct these domains oneself but instead use one of the constructors provided by ConString:

• constant_string_domain(STRING,RESULT) takes a string and constructs a string_dom.
• single_char_domain(CHAR,AUTOMATON) takes a single character and constructs an automata which language is only the chosen character.
• any_char_domain(DOMAIN) constructs an automata that language is any single character.
• generate_domain(REGEX,AUTOMATON) constructs an automata for a regular expression or a constant string.

Coding Guidelines

These are guidelines that shall keep the code tidy and clear, as well as ensure a certain degree of readability. These guidelines should be kept in mind by contributors and followed as much as possible.

• between every predicate 2 lines of whitespace for readability.
• indentation is made by two spaces.
• only commit documented code using Prolog Doc conventionen.
• make test cases for the desired behavior for at least every exported predicate.