scala-tptp-parser is a Scala library (for Scala 2.13.x) for parsing the input languages of the TPTP infrastructure.

The package contains a data structure for the abstract syntax tree (AST) of the parsed input as well as the parser for the different language of the TPTP, see http://tptp.org for details. In particular, the parser supports:

• THF (TH0/TH1): Monomorphic and polymorphic higher-order logic,
• TFF (TF0/TF1): Monomorphic and polymorphic typed first-order logic, including extended TFF (TXF, formerly called TFX),
• FOF: Untyped first-order logic,
• TCF: Typed clause-normal form,
• CNF: (Untyped) clause-normal form,
• TPI: TPTP Process Instruction language, and
• NXF/NHF: the new (partly experimental) non-classical TPTP languages based on TXF resp. THF.

The parser was originally developed for v7.5.0.0 of the TPTP syntax BNF (http://tptp.org/TPTP/SyntaxBNF.html); but it's constantly being updated to follow more recent developments.

scala-tptp-parser may be referenced using

The Scala TPTP parser is available on Maven Central (current version: 1.7.1). Snapshots are available on Sonatype.

In order to include scala-tptp-parser into your project via Maven, just add the following dependency:

<dependency>
  <groupId>io.github.leoprover</groupId>
  <artifactId>scala-tptp-parser_2.13</artifactId>
  <version>1.7.1</version>
</dependency>

In order to include scala-tptp-parser into your project via SBT, just add the following dependency to your build.sbt:

libraryDependencies += "io.github.leoprover" %% "scala-tptp-parser" % "1.7.1"

In order to use the library with a non-sbt project, you can simply compile the library and use the class files as an unmanaged dependency/class path. The latest release JAR can also be downloaded from the Maven Central link above.

See the brief API documentation.

The parser object TPTPParser offers several methods for parsing TPTP problems, annotated formulas or simple formulas. The input is transformed into an abstract syntax tree (AST) provided at leo.datastructures.TPTP. The ASTs are mostly case classes that can be further processed by pattern matching.

A small sample application can be seen below:

import leo.modules.input.{TPTPParser => Parser}
import Parser.TPTPParseException
import leo.datastructures.TPTP.THF

try {
 val result = Parser.problem(io.Source.fromFile("/path/to/file"))
 println(s"Parsed ${result.formulas.size} formulae and ${result.includes.size} include statements.")
 // ...
 val annotatedFormula = Parser.annotatedTHF("thf(f, axiom, ![X:$i]: (p @ X)).")
 println(s"${annotatedFormula.name} is an ${annotatedFormula.role}.")
 // ...
 val formula = Parser.thf("![X:$i]: (p @ X)")
 formula match {
   case THF.FunctionTerm(f, args) => // ...
   case THF.QuantifiedFormula(quantifier, variableList, body) => // ...
   case THF.Variable(name) => // ...
   case THF.UnaryFormula(connective, body) => // ...
   case THF.BinaryFormula(connective, left, right) => // ...
   case THF.Tuple(elements) => // ...
   case THF.ConditionalTerm(condition, thn, els) => // ...
   case THF.LetTerm(typing, binding, body) => // ...
   case THF.DefinedTH1ConstantTerm(constant) => // ...
   case THF.ConnectiveTerm(conn) => // ...
   case THF.DistinctObject(name) => // ...
   case THF.NumberTerm(value) => // ...
   case THF.NonclassicalPolyaryFormula(connective, args) => // ...
 }
 // ...
} catch {
 case e: TPTPParseException => println(s"Parse error at line ${e.line}:${e.offset}: ${e.getMessage}")
}

• 1.7.1: Allow any TFF terms on the left side of a meta equality (==).
• 1.7.0: Rework non-classical formulas in TFF and THF, so that both languages now have their dedicated case class for non-classical formulas, called NonclassicalPolyaryFormula. This breaks a few things from previous versions (new formula structure for NCL formulas). Also, the NCL short-form connectives [.] and <.> are now unary connectives (and can be used as such). The long-form connectives {name} still require an explicit application with @.
• 1.6.5: Fix usage of single-quoted identifiers that start with $ (or $$). Now the single quotes are retained in the functor name (in the AST) if the name is not a TPTP lower word. So the functor with name (as string) "$true" is not equal to "'$true'", as it should be (note the single quotes in the second variant). Single quotes are stripped automatically, if they are not necessary (i.e., parsing "'abc'" will produce functor "abc" without single quotes).
• 1.6.4: Fix TFF NCL pretty printing: Include introduced @ signs introduced by 1.6.3.
• 1.6.3: Change first-order NCL syntax according to PAAR paper (introducing @ sign in TFF). Fix pretty printing of NCL operators.
• 1.6.2: Minor update to remove parentheses around meta equalities in TFX.
• 1.6.1: Minor update to support applied form of $ite conditionals in THF (e.g., $ite @ c @ a @ b). Functional style ($ite(c,a,b)) is still supported.
• 1.6: Support for sequents in TFX and THF
• 1.5: Support for block and line (user/defined/system) comments into the AST (limited to those which start at the beginning of a line). Thanks to @XBagon for adding this feature!
• 1.5: Support for sub-roles (e.g., axiom-something or conjecture-[strength(0.85)]). In general, any <general_term> may be used as sub-role.
• 1.4: Support for non-classical TPTP (http://tptp.org/NonClassicalLogic/)