@hackage / phino

Command-Line Manipulator of 𝜑-Calculus Expressions

This is a command-line normalizer, rewriter, and dataizer of 𝜑-calculus expressions.

First, you write a simple 𝜑-calculus expression in the hello.phi file:

⟦ φ ↦ ⟦ Δ ⤍ 68-65-6C-6C-6F ⟧, t ↦ ξ.k, k ↦ ⟦⟧ ⟧

Installation

Then you can install phino in two ways:

Install Cabal first and then:

cabal update
cabal install --overwrite-policy=always phino-0.0.121
phino --version

Or download binary from the internet using curl or wget:

sudo curl -o /usr/local/bin/phino http://phino.objectionary.com/releases/macos-15/phino-latest
sudo chmod +x /usr/local/bin/phino
phino --version

Download paths are:

• Ubuntu 22.04: http://phino.objectionary.com/releases/ubuntu-22.04/phino-latest
• Ubuntu 24.04: http://phino.objectionary.com/releases/ubuntu-24.04/phino-latest
• MacOS (ARM): http://phino.objectionary.com/releases/macos-15/phino-latest
• MacOS (Intel): http://phino.objectionary.com/releases/macos-14-large/phino-latest
• Windows: http://phino.objectionary.com/releases/windows-2022/phino-latest.exe

Build

To build phino from source, clone this repository:

git clone git@github.com:objectionary/phino.git
cd phino

Then, run the following command (ensure you have Cabal installed):

cabal build all

Next, run this command to install phino system-wide:

sudo cp "$(cabal list-bin phino)" /usr/local/bin/phino

Verify that phino is installed correctly:

$ phino --version
0.0.0

You can ensure scripts are run with a specific version of phino using the --pin global option. It exits with an error when the version supplied doesn't match the installed one:

phino --pin=0.0.0.67 dataize hello.phi

Dataize

Then, you dataize the expression:

$ phino dataize hello.phi
68-65-6C-6C-6F

Atoms

Which λ functions exist is a property of the object model being dataized, not of the calculus, so phino implements none of them. They come from a JSON registry given with --atoms, keyed by regular expressions over λ names:

{
  "L_number_plus": {
    "rt": "node",
    "script": "const readline = require('readline'); ..."
  }
}

The rt field names the interpreter the script is run under. Only node is supported for now; a registry naming any other interpreter is refused when the file is read, before dataization starts.

When 𝔼 reaches a λ function the registry carries, phino writes its script to a temporary file and runs it as a POSIX process under that interpreter:

node /tmp/phino-atom-4f2a.js

An atom that is already a program needs no interpreter and no staging. Such an entry says exec and gives a path instead of a script:

{
  "L_number_plus": {
    "rt": "exec",
    "path": "/opt/eo/atoms/number-plus"
  }
}

phino spawns that file directly, as the executable binary it is, with no arguments. A path that names no file, or a file nobody may run, is refused where the registry is read, together with the unknown runtimes.

Whichever way it is run, the program is talked to over stdin and stdout, one JSON object per line, in the letters of the evaluation rule of the 𝜑-calculus paper, 𝔼(𝑏, 𝑒, 𝑠) = 𝑛, where 𝑏 is the formation, 𝑒 the universe and 𝑛 the normal form the atom answers with:

{"𝑒": "⟦ bytes ↦ ⟦ … ⟧, number ↦ ⟦ … ⟧, φ ↦ … ⟧"}
{"id": 1, "λ": "L_number_plus", "𝑏": "⟦ x ↦ Φ.number( … ), ρ ↦ ⟦ … ⟧ ⟧"}
{"id": 1, "𝑛": "11"}

The first two lines are phino's, the third is the program's. The universe Φ goes under 𝑒, in a line of its own, before the first request. Then comes the request: an id, the λ name under λ — one program may be registered under several names and branch on it — and, under 𝑏, the formation being evaluated, with its λ binding removed. Both payloads are canonical 𝜑-calculus on a single line — no syntax sugar, whatever --sweet says about the output of the run — so a program never has to know about phino's sugar in order to find a datum: every byte array is spelled out as a Δ binding.

The program answers with one line carrying the same id and, under 𝑛, the 𝜑-expression the atom answers with, in any syntax phino's parser reads — syntax sugar included, so the 11 above and the Φ.number( … ) it stands for are the same answer. phino parses it back and hands it to 𝔼 as the atom's raw result, normalizing it exactly as it normalizes anything else, so --evaluations, --partial and --max-steps keep working unchanged.

A program started for the fire is asked one request, always id 1, and its stdin is closed behind it, so it may read its input whole or line by line, as it pleases. It is waited for once it has answered, and a non-zero exit fails the run. So does a reply that is not JSON, carries no 𝑛, answers another id, or an 𝑛 that does not parse, or a program that quits without answering — always with the program's own stderr in the message.

Each key of the registry is a regular expression, and it must match the whole λ name, so a plain name such as L_number_plus means that one atom and nothing else, while L_number_.* stands for every atom of number. When 𝔼 reaches a λ function, the keys are tried top to bottom, in the order the file lists them, and the first one that matches is the entry fired, so a key placed above another hides whatever the two have in common. A key that is not a regular expression is refused where the registry is read.

A λ name no key matches has no λ function at all, so 𝔼 gets stuck on it. Without --atoms the registry is empty and every atom gets stuck.

One process per fire is where a program that is slow to start — a JVM, say — spends most of the run. An entry saying serve has phino start its program once, on the first fire, and keep it for the rest of the run, whether it is a script or a path. Together with a key that matches many names, this is how one program stands for a whole object model without being spelled once per atom:

{
  "L_bytes_eq": {
    "rt": "node",
    "script": "const readline = require('readline'); ..."
  },
  ".*": {
    "rt": "exec",
    "path": "/opt/eo/atoms/resident",
    "serve": true
  }
}

Every λ name registered on the same program, under one key or under several, is served by the same process, so there is one of it, however many atoms it stands for. The lines are the same: the program reads request after request off its stdin, each with the next id, and answers each in turn. The universe is told again only when a fire comes with a different one; the program keeps the last one it was told. When the run is over, whatever it ended with, phino closes the program's stdin, which is its cue to quit, and terminates it if it has not quit within a second.

Reducing the operands of an atom

A program gets at the parts of 𝑏 by calling phino again, so no API has to be exposed for it. The --inside option is how it asks: the expression it names is bound to a fresh synthetic attribute of the input expression, which the run takes as the universe, normalized there, and then dataized. This is the same trick phino plays internally whenever it has to reduce a sub-expression the program does not contain:

$ phino dataize --atoms=atoms.json --inside='5.plus( 6 )' universe.phi
40-26-00-00-00-00-00-00

Here universe.phi is the 𝜑-program the atom is being fired inside — the very text the program was told under 𝑒, which it feeds back on stdin.

So a L_number_plus that reduces its own operands reads like this:

const readline = require('readline');
const { execFileSync } = require('child_process');
let universe;
const dataized = (expr) => execFileSync(
  'phino',
  ['dataize', '--atoms=atoms.json', `--inside=${expr}`],
  { input: universe, encoding: 'utf8' }
).trim();
const number = (expr) => Buffer
  .from(dataized(expr).replace(/-/g, ''), 'hex')
  .readDoubleBE(0);
const hex = (value) => {
  const bytes = Buffer.alloc(8);
  bytes.writeDoubleBE(value);
  return [...bytes]
    .map((octet) => octet.toString(16).toUpperCase().padStart(2, '0'))
    .join('-');
};
readline.createInterface({ input: process.stdin }).on('line', (line) => {
  const message = JSON.parse(line);
  if ('𝑒' in message) {
    universe = message['𝑒'];
    return;
  }
  if (message['λ'] !== 'L_number_plus') {
    throw new Error(`unsupported atom ${message['λ']}`);
  }
  const b = message['𝑏'];
  const sum = hex(number(`${b}.ρ`) + number(`${b}.x`));
  process.stdout.write(`${JSON.stringify({
    id: message.id,
    '𝑛': `Φ.number( as-bytes ↦ Φ.bytes( data ↦ ⟦ Δ ⤍ ${sum} ⟧ ) )`,
  })}\n`);
});

Written this way, reading until its stdin closes, the same program runs once per fire and serves the whole run alike; only the registry entry decides.

The --inside option cannot be combined with --locator, since it aims the run at the binding it mints itself. Both dataize and morph take --atoms and --inside.

Recording what fired

Every atom fired on the way to the bytes may be recorded in a machine-readable protocol, with the --evaluations option. One firing is one line of three tab-separated fields: the name of the λ function, the formation it was applied to, and the expression it returned:

$ cat sum.phi
⟦
  bytes(data) ↦ ⟦ φ ↦ data ⟧,
  number(as-bytes) ↦ ⟦ φ ↦ as-bytes, plus(x) ↦ ⟦ λ ⤍ L_number_plus ⟧ ⟧,
  φ ↦ 5.plus( 6 )
⟧
$ phino dataize --atoms=atoms.json --evaluations=atoms.tsv --quiet \
    --sweet --hide-rho sum.phi
$ cat -T atoms.tsv
L_number_plus^I⟦ x ↦ 6 ⟧^I11

Records follow the syntax of the other options, such as --sweet and --hide-rho, but always stay on one line. The file is truncated at the beginning of every run, and --output=phi is the only output format it works with, since one record must fit into one line.

Partial evaluation

An atom that cannot fire fails the run: its λ function is not in the registry given with --atoms. This is what happens when an operation is deliberately left unimplemented — a data input replaced by a placeholder formation such as ⟦ λ ⤍ Sym_arg_0 ⟧, or an operation whose answer is not known yet. With --partial, dataization becomes partial evaluation instead: what the known inputs decide is computed, the rest survives as the residual program, which is printed in place of the bytes, and the run ends successfully:

$ cat partial.phi
⟦
  bytes(data) ↦ ⟦ φ ↦ data ⟧,
  number(as-bytes) ↦ ⟦
    φ ↦ as-bytes,
    plus(x) ↦ ⟦ λ ⤍ L_number_plus ⟧,
    times(x) ↦ ⟦ λ ⤍ L_number_times ⟧,
    as-bool ↦ ⟦ λ ⤍ L_number_as_bool ⟧
  ⟧,
  φ ↦ 2.times( 3 ).plus( 4 ).as-bool
⟧
$ phino dataize --atoms=atoms.json --partial --sweet --hide-rho partial.phi
⟦ λ ⤍ L_number_as_bool ⟧

Here 2.times( 3 ).plus( 4 ) was decided by the atoms the registry carries, so it was computed (its result, 10, sits in the hidden ρ of the residual program), while as-bool names a λ function no script answers for, so it stays in place as a normal-form subterm. Each such stuck site also lands in the --evaluations file, as a record with the first two fields only, since there is no result to report:

$ phino dataize --atoms=atoms.json --partial --evaluations=atoms.tsv --quiet \
    --sweet --hide-rho partial.phi
$ cat -T atoms.tsv
L_number_times^I⟦ x ↦ 3 ⟧^I6
L_number_plus^I⟦ x ↦ 4 ⟧^I10
L_number_as_bool^I⟦⟧

Evaluation stays demand-driven, as the calculus prescribes: an argument that nothing asked for before the run got stuck is left as it is in the residual program, for the next iteration.

The nested morphing and dataization recursion is bounded by the --max-steps option (default 1000): when the budget is exhausted, the run fails with Dataization did not finish before reaching the limit of steps. This guards against non-terminating terms, which used to loop forever before the bound was introduced:

$ phino dataize --max-steps=50 problem.phi
[ERROR]: Dataization did not finish before reaching the limit of steps: --max-steps=50

Morph

Dataization insists on bytes. Morphing 𝕄 asks a different question: evaluate as far as the object model allows, without demanding data. It resolves Φ against the universe, peels dispatches and applications through normalization, fires whichever atoms sit under a dispatch, and stops at the first formation it reaches, handing that formation back untouched. The morph command runs 𝕄 on its own:

$ cat two.phi
⟦
  bytes(data) ↦ ⟦ φ ↦ data ⟧,
  number(as-bytes) ↦ ⟦ φ ↦ as-bytes, plus(x) ↦ ⟦ λ ⤍ L_number_plus ⟧ ⟧,
  φ ↦ 5.plus( 6 ).plus( 7 )
⟧
$ phino dataize --atoms=atoms.json --sweet --hide-rho two.phi
40-32-00-00-00-00-00-00
$ phino morph --atoms=atoms.json --locator=Q.φ --sweet --hide-rho two.phi
⟦ x ↦ 7, λ ⤍ L_number_plus ⟧

The inner 5.plus( 6 ) fires, because .plus is dispatched on its result, and 11 lands in the ρ hidden by --hide-rho. The outer application is saturated but bare, so 𝕄 returns it and is finished; firing it is dataization's job and takes dataize on to 18.

The default locator Q morphs the whole top formation, which 𝕄 returns unchanged, so --locator is how one aims 𝕄 at a subterm, exactly as in dataize. Unlike 𝔻, 𝕄 is total: where no formation is reachable the answer is the terminator ⊥, printed rather than reported as a failed run:

$ phino morph --locator=Q.x <<< '⟦ x ↦ ξ ⟧'
⊥

The whole dataize option surface applies unchanged — --atoms, --inside, --sequence, --headers, --steps-dir, --evaluations, --partial, --max-steps, --shuffle/--seed, --output, --focus and the rest.

Deep morphing

𝕄 stops at the first formation it reaches and hands its bindings back as they were written, since firing a bare λ is dataization's job, and dataize follows the one path dataization demands and ends in bytes. What a program holds but nothing demands — the argument of an atom the registry does not serve, for one — is therefore reduced by neither. The --deep flag enters it:

$ cat gap.phi
⟦
  bytes(data) ↦ ⟦ φ ↦ ξ.data ⟧,
  number(as-bytes) ↦ ⟦ φ ↦ ξ.as-bytes, times(x) ↦ ⟦ λ ⤍ L_number_times ⟧ ⟧,
  bar(x) ↦ ⟦ λ ⤍ L_bar ⟧,
  demo ↦ ⟦ foo ↦ ⟦ n ↦ 3, φ ↦ Φ.bar( ξ.n.times( 5 ).times( 7 ) ) ⟧ ⟧
⟧
$ phino morph --atoms=atoms.json --inside='Q.demo.foo' \
    --sweet --hide-rho gap.phi
⟦ n ↦ 3, φ ↦ Φ.bar( n.times( 5 ).times( 7 ) ) ⟧
$ phino morph --deep --atoms=atoms.json --inside='Q.demo.foo' \
    --sweet --hide-rho gap.phi
⟦ n ↦ 3, φ ↦ Φ.bar( 105 ) ⟧

Every binding of the formation is entered, recursively. 𝕄 is asked about the term standing there and, where it lands on a saturated formation whose λ the registry serves, that λ is fired and 𝕄 is asked about the answer again. A term on whose way an atom fired is replaced by the answer of the last firing, which is the 𝜑-program the atom wrote rather than the normal form of it, so 105 stands where the arithmetic stood. A term no atom touched stays exactly as it was written and only its own parts are walked, so Φ.bar keeps its name and what comes back is still the same program, reduced as far as the registry allows. The step joins the chain under the name deep, so --sequence shows it, and --max-steps bounds the walk.

Two things are left alone. A λ the registry does not serve is not fired at all, so --deep stays as total as 𝕄 itself and needs no --partial; an atom that gets stuck deeper on a spine still fails the run, and --partial parks it, leaving that term as it was written. A formation still holding a void binding is not fired either: the void is an argument the program has not given yet, so times(x) ↦ ⟦ λ ⤍ L_number_times ⟧ is a method waiting to be applied, not an application waiting to be computed. Walking the whole program therefore folds what it can and leaves the object model as it was declared:

$ phino morph --deep --atoms=atoms.json --sweet --hide-rho gap.phi
⟦
  bytes(data) ↦ ⟦ φ ↦ data ⟧,
  number(as-bytes) ↦ ⟦ φ ↦ as-bytes, times(x) ↦ ⟦ λ ⤍ L_number_times ⟧ ⟧,
  bar(x) ↦ ⟦ λ ⤍ L_bar ⟧,
  demo ↦ ⟦ foo ↦ ⟦ n ↦ 3, φ ↦ Φ.bar( 105 ) ⟧ ⟧
⟧

Rewrite

You can rewrite this expression with the help of rules defined in the my-rule.yml YAML file (here, the !d is a capturing group, similar to regular expressions):

name: My custom rule
pattern: Δ ⤍ !d
result: Δ ⤍ 62-79-65

Then, rewrite:

$ phino rewrite --rule=my-rule.yml hello.phi
⟦ φ ↦ ⟦ Δ ⤍ 62-79-65 ⟧, t ↦ ξ.k, k ↦ ⟦⟧ ⟧

If you want to use many rules, just use --rule as many times as you need:

phino rewrite --rule=rule1.yaml --rule=rule2.yaml ...

You can also use built-in rules, which are designed to normalize expressions:

phino rewrite --normalize hello.phi

Both flags may be combined, so that your own rules are applied alongside the built-in ones, in a single rewriting session:

phino rewrite --normalize --rule=my-rule.yaml hello.phi

Some rules mint fresh synthetic names via the random-string built-in. To keep the output reproducible across runs, phino seeds the random generator deterministically with 0 by default. Use --seed to pick a different seed:

phino rewrite --seed=42 --rule=my-rule.yml hello.phi

If no input file is provided, the 𝜑-expression is taken from stdin:

$ echo '⟦ φ ↦ ⟦ Δ ⤍ 68-65-6C-6C-6F ⟧ ⟧' | phino rewrite --rule=my-rule.yml
⟦ φ ↦ ⟦ Δ ⤍ 62-79-65 ⟧ ⟧

You're able to pass XMIR as input. Use --input=xmir and phino will parse given XMIR from file or stdin and convert it to phi AST.

phino rewrite --rule=my-rule.yaml --input=xmir file.xmir

Also phino supports 𝜑-expressions in ASCII format and with syntax sugar. The rewrite command also allows you to desugar the expression and print it in canonical syntax:

$ echo '[[ @ -> Q.io.stdout("hello") ]]' | phino rewrite
⟦
  φ ↦ Φ.io.stdout(
    α0 ↦ Φ.string(
      α0 ↦ Φ.bytes(
        α0 ↦ ⟦ Δ ⤍ 68-65-6C-6C-6F ⟧
      )
    )
  )
⟧

Merge

You can merge several 𝜑-expressions into a single one by merging their top level formations:

$ cat bytes.phi
⟦ bytes(data) ↦ ⟦ φ ↦ data ⟧ ⟧
$ cat number.phi
⟦
  number(as-bytes) ↦ ⟦
    φ ↦ as-bytes,
    plus(x) ↦ ⟦ λ ⤍ L_number_plus ⟧
  ⟧
⟧
$ cat minus.phi
⟦ number ↦ ⟦ minus(x) ↦ ⟦ λ ⤍ L_number_minus ⟧ ⟧ ⟧
$ phino merge bytes.phi number.phi minus.phi --sweet
⟦
  bytes(data) ↦ ⟦ φ ↦ data ⟧,
  number(as-bytes) ↦ ⟦
    φ ↦ as-bytes,
    plus(x) ↦ ⟦ λ ⤍ L_number_plus ⟧,
    minus(x) ↦ ⟦ λ ⤍ L_number_minus ⟧
  ⟧
⟧

Match

You can test the 𝜑-expression matches against the rule pattern. The result output contains matched substitutions:

$ phino match --pattern='⟦ Δ ⤍ !d, !B ⟧' hello.phi
B >> ⟦ ρ ↦ ∅ ⟧
d >> 68-65-6C-6C-6F

Explain

You can explain the built-in rules by printing them in LaTeX format. Pass exactly one of --normalize, --morph, --dataize or --contextualize for the rewriting, morphing (𝕄), dataization (𝔻) or contextualization (𝒞) rules (or --rule for a custom rule file):

$ phino explain --normalize
\begin{tabular}{rl}
\phinoNormalizationRule{alpha}
  { [[ B_1, \tau -> ?, B_2 ]] ( \phiTerminal{\alpha_{i}} -> e ) }
  { [[ B_1, \tau -> ?, B_2 ]] ( \tau -> e ) }
  { $ i = \vert \overline{ B_1 } \vert $ }
  { }
\phinoNormalizationRule{dc}
  { T ( \tau -> e ) }
  { T }
  { }
  { }
...
\phinoNormalizationRule{stop}
  { [[ B ]] . \tau }
  { T }
  { $ \tau \notin B \;\text{and}\; @ \notin B \;\text{and}\; L \notin B $ }
  { }
\end{tabular}

The morphing and dataization rules are printed the same way:

$ phino explain --morph
\begin{tabular}{rl}
\phinoMorphingRule{mf}
  { \mathbb{M}( [[ B ]], e ) }
  { [[ B ]] }
  { }
  { }
...
\phinoMorphingRule{universe}
  { \mathbb{M}( Q, e ) }
  { \mathbb{M}( \phinoNormalize{ e }, e ) }
  { $ e \not= Q $ }
  { }
\end{tabular}

$ phino explain --dataize
\begin{tabular}{rl}
\phinoDataizationRule{delta}
  { \phinoDataize{ [[ B_1, D> δ, B_2 ]] } }
  { δ }
  { }
  { }
...
\phinoDataizationRule{norm}
  { \phinoDataize{ n } }
  { \phinoDataize{ \mathbb{M}( n, e ) } }
  { }
  { }
\end{tabular}

$ phino explain --contextualize
\begin{phinoContextualizationInference}
  \phinoName{cxi}
  \phinoConclusion{ \phinoContextualize{ \phiTerminal{\xi} }{ k }{ k } }
\end{phinoContextualizationInference}
...
\begin{phinoContextualizationInference}
  \phinoName{cd}
  \phinoPremise{ \phinoContextualize{ n }{ k }{ n_1 } }
  \phinoConclusion{ \phinoContextualize{ n . \tau }{ k }{ n_1 . \tau } }
\end{phinoContextualizationInference}

For more details, use phino [COMMAND] --help option.

Rule structure

This is BNF-like yaml rule structure. Here types ended with apostrophe, like Attribute' are built types from 𝜑-expression AST

Rule:
  name: String
  pattern: String
  result: String
  when: Condition?       # predicate, works with substitutions before extension
  where: [Extension]?    # substitution extensions
  having: Condition?     # predicate, works with substitutions after extension

Condition:
  = and: [Condition]     # logical AND
  | or:  [Condition]     # logical OR
  | not: Condition       # logical NOT
  | eq:                  # compare two comparable objects
      - Comparable
      - Comparable
  | in:                  # check if attributes exist in bindings
      - Attribute'
      - Binding'
  | nf: Expression'      # returns True if given expression in normal form
                         # which means that no more other normalization rules
                         # can be applied
  | absolute: Expression' # returns True if given expression is xi-free, i.e.
                         # there is no ξ outside of a formation: it is Φ, a
                         # formation, a dispatch with a xi-free subject, or an
                         # application with a xi-free subject and argument.
                         # Combined with a normal-form check by the '𝑘'/'!k'
                         # meta variable, which ranges over the absolute
                         # expressions 𝒦 ⊆ 𝒩, used by the Rcopy rule.
  | matches:             # returns True if given expression after dataization
      - String           # matches to given regex
      - Expression
  | part-of:             # returns True if given expression is attached to any
      - Expression'      # attribute in ginve bindings
      - BiMeta'
  | formation:           # returns True if given expression is a formation
      Expression'        # (an abstraction ⟦…⟧); used by morphing 'md'
                         # as 'not (formation 𝑛)', so a non-formation head is
                         # morphed and a formation head is left to 'ml'
  | gt:                  # returns True if the first comparable object is
      - Comparable       # greater than the second one
      - Comparable
  | disjoint:            # returns True if none of the given attributes exists
      - [Attribute']     # in the given bindings
      - Binding'

Comparable:              # comparable object that may be used in 'eq' condition
  = Attribute'
  | Number
  | Expression'

Number:                  # comparable number
  = Integer              # just regular integer
  | IndexMeta'           # 𝑖 (or !i), the index captured by an α𝑖 argument
  | length: BiMeta'      # calculate length of bindings by given meta binding
  | domain: BiMeta'      # calculate number of unique attributes in given
                         # meta binding (excluding 'assets')

Extension:               # substitutions extension used to introduce new meta variables
  meta: [ExtArgument]    # new introduced meta variable
  function: String       # name of the function
  args: [ExtArgument]    # arguments of the function

ExtArgument
  = Bytes'               # !d
  | Binding'             # !B
  | Expression'          # !e
  | Attribute'           # !t

Here's list of functions that are supported for extensions:

• contextualize - function of two arguments, that rewrites given expression depending on provided context according to the contextualization rules
• random-tau - creates attribute with random unique name. Accepts bindings, and attributes. Ensures that created attribute is not present in list of provided attributes and does not exist as attribute in provided bindings.
• dataize - dataizes given expression and returns bytes.
• concat - accepts bytes or dataizable expressions as arguments, concatenates them into single sequence and convert it to expression that can be pretty printed as human readable string: Φ.string(Φ.bytes⟦ Δ ⤍ !d ⟧).
• sed - pattern replacer, works like unix sed function. Accepts two arguments: target expression and pattern. Pattern must start with s/, consists of three parts separated by /, for example, this pattern s/\\s+//g replaces all the spaces with empty string. To escape braces and slashes in pattern and replacement parts - use them with \\, e.g. s/\\(.+\\)//g.
• random-string - accepts dataizable expression or bytes as pattern. Replaces %x and %d formatters with random hex numbers and decimals accordingly. Uniqueness is guaranteed during one execution of phino.
• size - accepts exactly one meta binding and returns size of it and Φ.number.
• tau - accepts Φ.string, dataizes it and converts it to attribute. If dataized string can't be converted to attribute - an error is thrown.
• string - accepts Φ.string or Φ.number or attribute and converts it to Φ.string.
• number - accepts Φ.string and converts it Φ.number
• sum - accepts list of Φ.number or Φ.bytes and returns sum of them as Φ.number
• join - accepts list of bindings and returns list of joined bindings. Duplicated ρ, Δ and λ attributes are ignored, all other duplicated attributes are replaced with unique attributes using random-tau function.

Meta variables

The phino supports meta variables to write 𝜑-expression patterns for capturing attributes, bindings, etc.

This is the list of supported meta variables:

• !t || 𝜏 - attribute
• !i || 𝑖 - the index of a positional (α) application argument, captured by writing α𝑖 (or ~!i)
• !e || 𝑒 - any expression
• !n || 𝑛 - any expression that is already in normal form (behaves like !e/𝑒, but only binds a sub-expression in NF, so no explicit nf: guard is needed)
• !k || 𝑘 - any expression that is absolute, i.e. xi-free and in normal form (ranges over 𝒦 ⊆ 𝒩); behaves like !e/𝑒 but only binds an absolute sub-expression, so no explicit absolute: or nf: guard is needed
• !B || 𝐵 - list of bindings
• !d || δ - bytes in meta delta binding
• !F || 𝑓 - function name in meta lambda binding

A meta variable carries a suffix, like !B1 or 𝜏0, to name what it captured, so that the result, when, where and having of a rule can read it back.

Written bare, with no suffix at all, a meta variable is anonymous: it matches whatever term stands in its place, every occurrence on its own, and binds no name. Two anonymous metas of one kind are therefore two different captures, which is what lets a pattern ask for any two attributes without inventing names for them:

name: two-attributes
pattern: '⟦ 𝜏 ↦ 𝑒, 𝜏 ↦ 𝑒 ⟧'
result: '⟦ x ↦ ⟦ Δ ⤍ 2A- ⟧ ⟧'

Spelled with suffixes, that pattern would read ⟦ 𝜏1 ↦ 𝑒1, 𝜏2 ↦ 𝑒2 ⟧ and name four captures the result never mentions, while ⟦ 𝜏1 ↦ 𝑒1, 𝜏1 ↦ 𝑒1 ⟧ would be rejected as a duplicated attribute.

Nothing can refer to an anonymous meta, since it has no name to be referred to by. Writing one outside a pattern (or the match, e-match and c-match of an inference rule) is a mistake in the rule and is reported as the rule loads.

A positional (α) application argument is written as α0, ~0 (ASCII), or α𝑖/~!i when its index is captured by an !i/𝑖 meta variable.

Incorrect usage of meta variables in 𝜑-expression patterns leads to parsing errors.

Benchmark

To run performance benchmarks, you need Java 8+ and curl. Maven is downloaded automatically on first run via benchmark/mvnw.

The benchmark uses the compiled Native class from JNA — a large real-world Java class — as its test input. On first run, make bench downloads the class, disassembles it to XMIR via jeo-maven-plugin, converts it to 𝜑 using phino rewrite, and caches the results in benchmark/tmp/. Subsequent runs skip straight to the benchmarks.

make bench

=== parse/phi ===
  warmup:     3 iterations
  batches:    10 x 1
  total:      1775922.415 μs
  avg:        177592.242 μs
  min:        161735.069 μs
  max:        206519.406 μs
  std dev:    15129.161 μs
=== parse/xmir ===
  warmup:     3 iterations
  batches:    10 x 1
  total:      7571386.164 μs
  avg:        757138.616 μs
  min:        691286.480 μs
  max:        813614.260 μs
  std dev:    32034.618 μs
=== rewrite/normalize ===
  warmup:     3 iterations
  batches:    10 x 1
  total:      579674.000 μs
  avg:        57967.400 μs
  min:        56388.139 μs
  max:        61829.190 μs
  std dev:    1703.063 μs
=== print/sweet/multiline ===
  warmup:     3 iterations
  batches:    10 x 1
  total:      3876263.188 μs
  avg:        387626.319 μs
  min:        373639.628 μs
  max:        405513.521 μs
  std dev:    10018.938 μs
=== print/sweet/flat ===
  warmup:     3 iterations
  batches:    10 x 1
  total:      3789406.984 μs
  avg:        378940.698 μs
  min:        367216.617 μs
  max:        395082.263 μs
  std dev:    8761.575 μs
=== print/salty/multiline ===
  warmup:     3 iterations
  batches:    10 x 1
  total:      13982361.998 μs
  avg:        1398236.200 μs
  min:        1366361.442 μs
  max:        1433666.397 μs
  std dev:    17717.682 μs

The results were calculated in this GHA job on 2026-09-09 at 15:41, on Linux with 4 CPUs.

How to Contribute

Fork repository, make changes, then send us a pull request. We will review your changes and apply them to the master branch shortly, provided they don't violate our quality standards. To avoid frustration, before sending us your pull request please make sure all your tests pass:

make all

To generate a local coverage report for development, run:

make coverage

To build a phino executable into the root of the repository, run:

make phino

This produces an executable phino (or phino.exe on Windows) in the project root, which you can run directly for quick local testing:

./phino --version

You will need GHC ≥ 9.6.7 and Cabal ≥ 3.0 (recommended) or Stack ≥ 3.0 installed.