thousandproblems 0.1.0__tar.gz → 0.2.0__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (25) hide show
  1. thousandproblems-0.2.0/CHANGELOG.md +12 -0
  2. {thousandproblems-0.1.0/src/thousandproblems.egg-info → thousandproblems-0.2.0}/PKG-INFO +42 -6
  3. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/README.md +41 -5
  4. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/pyproject.toml +1 -1
  5. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/src/thousandproblems/__init__.py +15 -3
  6. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/src/thousandproblems/_cparser.c +371 -24
  7. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/src/thousandproblems/_pyparser.py +98 -12
  8. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/src/thousandproblems/library.py +3 -2
  9. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/src/thousandproblems/nodes.py +45 -1
  10. {thousandproblems-0.1.0 → thousandproblems-0.2.0/src/thousandproblems.egg-info}/PKG-INFO +42 -6
  11. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/tests/test_tptp.py +198 -0
  12. thousandproblems-0.1.0/CHANGELOG.md +0 -8
  13. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/LICENSE +0 -0
  14. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/MANIFEST.in +0 -0
  15. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/benchmarks/bench.py +0 -0
  16. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/setup.cfg +0 -0
  17. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/setup.py +0 -0
  18. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/src/thousandproblems/_cparser.pyi +0 -0
  19. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/src/thousandproblems/py.typed +0 -0
  20. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/src/thousandproblems.egg-info/SOURCES.txt +0 -0
  21. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/src/thousandproblems.egg-info/dependency_links.txt +0 -0
  22. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/src/thousandproblems.egg-info/top_level.txt +0 -0
  23. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/tests/corpus.py +0 -0
  24. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/tests/test_backends.py +0 -0
  25. {thousandproblems-0.1.0 → thousandproblems-0.2.0}/tests/test_library.py +0 -0
@@ -0,0 +1,12 @@
1
+ # Changelog
2
+
3
+ ## 0.1.0
4
+
5
+ - `Variable.index` and `TypedVariable.index`: the distinct variables of a formula are numbered
6
+ 1, 2, 3 ... in order of first appearance (scoped by quantifier; 0 outside a formula).
7
+ - `TPTPFile.symbols` and `AnnotatedFormula.symbols`: the symbol table, a tuple of
8
+ `Symbol(name, arity)` in order of first appearance, without `$` and `$$` symbols.
9
+ - First release: parser and printer for every TPTP language (`thf`, `tff`, `tcf`, `fof`,
10
+ `cnf`, `tpi`), with a C extension and a pure-Python fallback that build identical trees.
11
+ - `fetch_problem` / `open_library`: read problems, and expand their includes, from a TPTP
12
+ folder, a tarball or the TPTP website.
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: thousandproblems
3
- Version: 0.1.0
3
+ Version: 0.2.0
4
4
  Summary: Fast parser and printer for the TPTP problem and solution syntax (thf, tff, tcf, fof, cnf, tpi).
5
5
  Author-email: Ontology Portal <pypi@ontologyportal.org>
6
6
  License: GNU GENERAL PUBLIC LICENSE
@@ -731,11 +731,47 @@ Import it as `thousandproblems`; the examples below alias it to `tptp`.
731
731
  >>> ax.language, ax.name, ax.role
732
732
  ('fof', 'ax1', 'axiom')
733
733
  >>> ax.formula.quantifier, ax.formula.variables
734
- ('!', (Variable(name='X'),))
734
+ ('!', (Variable(name='X', index=1),))
735
735
  >>> print(ax)
736
736
  fof(ax1,axiom,! [X] : (p(X) => q(X))).
737
737
  ```
738
738
 
739
+ ## Variable indexes and the symbol table
740
+
741
+ Each `Variable` has an `index`: the distinct variables of one formula are numbered 1, 2, 3 ... in
742
+ order of first appearance.
743
+
744
+ ```python
745
+ >>> clause = tptp.parse("cnf(c, axiom, f(X) | f(Y) | g(X, f(Y))).").formulae[0]
746
+ >>> [(v.name, v.index) for v in tptp.walk(clause.formula) if isinstance(v, tptp.Variable)]
747
+ [('X', 1), ('Y', 2), ('X', 1), ('Y', 2)]
748
+ ```
749
+
750
+ A quantifier counts as the first appearance of the variables it binds, and every binding is its
751
+ own variable, so `! [Y,X] : p(X,Y)` numbers Y 1 and X 2, and `(! [X] : p(X)) & (! [X] : q(X))`
752
+ gives its two `X` the indexes 1 and 2. Numbering restarts for each formula, and a variable that
753
+ is not part of a formula (in an annotation, or built by hand) has index 0. `index` is not
754
+ compared, so `Variable("X") == Variable("X", 1)`.
755
+
756
+ `problem.symbols` is the symbol table: a tuple of `Symbol(name, arity)`, one for each distinct
757
+ function, constant or predicate the formulae use, in order of first appearance. `f/1` and `f/2`
758
+ are different symbols, and a constant has arity 0.
759
+
760
+ ```python
761
+ >>> problem = tptp.parse("cnf(a, axiom, p(f(a), X) | ~ q(f(a,b))). cnf(b, axiom, p(c, Y)).")
762
+ >>> [str(s) for s in problem.symbols]
763
+ ['p/2', 'f/1', 'a/0', 'q/1', 'f/2', 'b/0', 'c/0']
764
+ >>> problem.formulae[1].symbols
765
+ (Symbol(name='p', arity=2), Symbol(name='c', arity=0))
766
+ ```
767
+
768
+ Variables, equality, numbers, distinct objects, `$` and `$$` symbols (`$true`, `$less` ...),
769
+ type declarations and annotations are not symbols. In THF, `f @ a` uses the symbols `f/0` and
770
+ `a/0`. Each formula has its own `symbols`, and the file's table is rebuilt when
771
+ `expand_includes` brings other files in, so it only covers formulae that are really there.
772
+
773
+ ## Functions
774
+
739
775
  | Function | |
740
776
  | --- | --- |
741
777
  | `parse(text, filename=None)` | Parse the text of a file. |
@@ -821,10 +857,10 @@ Measured on an Apple M-series laptop, Python 3.14, on generated problems of 0.4
821
857
 
822
858
  | | pure Python | C | speedup |
823
859
  | --- | --- | --- | --- |
824
- | cnf | 0.259 s | 0.024 s | 10.7x |
825
- | fof | 0.252 s | 0.024 s | 10.6x |
826
- | tff | 0.206 s | 0.018 s | 11.3x |
827
- | thf | 0.156 s | 0.012 s | 12.9x |
860
+ | cnf | 0.274 s | 0.026 s | 10.5x |
861
+ | fof | 0.269 s | 0.027 s | 10.0x |
862
+ | tff | 0.221 s | 0.020 s | 10.9x |
863
+ | thf | 0.160 s | 0.013 s | 12.4x |
828
864
 
829
865
  Run `python benchmarks/bench.py` to measure on your machine.
830
866
 
@@ -29,11 +29,47 @@ Import it as `thousandproblems`; the examples below alias it to `tptp`.
29
29
  >>> ax.language, ax.name, ax.role
30
30
  ('fof', 'ax1', 'axiom')
31
31
  >>> ax.formula.quantifier, ax.formula.variables
32
- ('!', (Variable(name='X'),))
32
+ ('!', (Variable(name='X', index=1),))
33
33
  >>> print(ax)
34
34
  fof(ax1,axiom,! [X] : (p(X) => q(X))).
35
35
  ```
36
36
 
37
+ ## Variable indexes and the symbol table
38
+
39
+ Each `Variable` has an `index`: the distinct variables of one formula are numbered 1, 2, 3 ... in
40
+ order of first appearance.
41
+
42
+ ```python
43
+ >>> clause = tptp.parse("cnf(c, axiom, f(X) | f(Y) | g(X, f(Y))).").formulae[0]
44
+ >>> [(v.name, v.index) for v in tptp.walk(clause.formula) if isinstance(v, tptp.Variable)]
45
+ [('X', 1), ('Y', 2), ('X', 1), ('Y', 2)]
46
+ ```
47
+
48
+ A quantifier counts as the first appearance of the variables it binds, and every binding is its
49
+ own variable, so `! [Y,X] : p(X,Y)` numbers Y 1 and X 2, and `(! [X] : p(X)) & (! [X] : q(X))`
50
+ gives its two `X` the indexes 1 and 2. Numbering restarts for each formula, and a variable that
51
+ is not part of a formula (in an annotation, or built by hand) has index 0. `index` is not
52
+ compared, so `Variable("X") == Variable("X", 1)`.
53
+
54
+ `problem.symbols` is the symbol table: a tuple of `Symbol(name, arity)`, one for each distinct
55
+ function, constant or predicate the formulae use, in order of first appearance. `f/1` and `f/2`
56
+ are different symbols, and a constant has arity 0.
57
+
58
+ ```python
59
+ >>> problem = tptp.parse("cnf(a, axiom, p(f(a), X) | ~ q(f(a,b))). cnf(b, axiom, p(c, Y)).")
60
+ >>> [str(s) for s in problem.symbols]
61
+ ['p/2', 'f/1', 'a/0', 'q/1', 'f/2', 'b/0', 'c/0']
62
+ >>> problem.formulae[1].symbols
63
+ (Symbol(name='p', arity=2), Symbol(name='c', arity=0))
64
+ ```
65
+
66
+ Variables, equality, numbers, distinct objects, `$` and `$$` symbols (`$true`, `$less` ...),
67
+ type declarations and annotations are not symbols. In THF, `f @ a` uses the symbols `f/0` and
68
+ `a/0`. Each formula has its own `symbols`, and the file's table is rebuilt when
69
+ `expand_includes` brings other files in, so it only covers formulae that are really there.
70
+
71
+ ## Functions
72
+
37
73
  | Function | |
38
74
  | --- | --- |
39
75
  | `parse(text, filename=None)` | Parse the text of a file. |
@@ -119,10 +155,10 @@ Measured on an Apple M-series laptop, Python 3.14, on generated problems of 0.4
119
155
 
120
156
  | | pure Python | C | speedup |
121
157
  | --- | --- | --- | --- |
122
- | cnf | 0.259 s | 0.024 s | 10.7x |
123
- | fof | 0.252 s | 0.024 s | 10.6x |
124
- | tff | 0.206 s | 0.018 s | 11.3x |
125
- | thf | 0.156 s | 0.012 s | 12.9x |
158
+ | cnf | 0.274 s | 0.026 s | 10.5x |
159
+ | fof | 0.269 s | 0.027 s | 10.0x |
160
+ | tff | 0.221 s | 0.020 s | 10.9x |
161
+ | thf | 0.160 s | 0.013 s | 12.4x |
126
162
 
127
163
  Run `python benchmarks/bench.py` to measure on your machine.
128
164
 
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
4
4
 
5
5
  [project]
6
6
  name = "thousandproblems"
7
- version = "0.1.0"
7
+ version = "0.2.0"
8
8
  description = "Fast parser and printer for the TPTP problem and solution syntax (thf, tff, tcf, fof, cnf, tpi)."
9
9
  readme = "README.md"
10
10
  requires-python = ">=3.11"
@@ -10,7 +10,7 @@ directives, and annotations (sources, inference records, useful info).
10
10
  >>> ax.language, ax.name, ax.role
11
11
  ('fof', 'ax1', 'axiom')
12
12
  >>> ax.formula.quantifier, ax.formula.variables
13
- ('!', (Variable(name='X'),))
13
+ ('!', (Variable(name='X', index=1),))
14
14
  >>> print(ax)
15
15
  fof(ax1,axiom,! [X] : (p(X) => q(X))).
16
16
 
@@ -27,6 +27,18 @@ The AST is the same for all languages:
27
27
  :class:`Binary`. ``>`` is right-associative.
28
28
  * Constants are :class:`Function` with no arguments. Defined and system symbols keep
29
29
  their prefix, as in ``Function('$true')``.
30
+ * ``Variable.index`` numbers the distinct variables of a formula 1, 2, 3 ... in order of
31
+ first appearance, so in ``f(X) | f(Y) | g(X, f(Y))`` every ``X`` has index 1 and every
32
+ ``Y`` index 2. A quantifier counts as the appearance of the variables it binds, and each
33
+ binding is its own variable: the two ``X`` of ``(! [X] : p(X)) & (! [X] : q(X))`` get
34
+ indexes 1 and 2. Numbering starts again for each formula; a variable outside a formula (in
35
+ an annotation, or built by hand) has index 0.
36
+ * ``TPTPFile.symbols`` is the symbol table: a tuple of :class:`Symbol` ``(name, arity)``,
37
+ one for each distinct function, constant or predicate the formulae use, in order of first
38
+ appearance, so ``f/1`` and ``f/2`` are different symbols. Variables, equality, numbers,
39
+ distinct objects, ``$`` and ``$$`` symbols, type declarations and annotations are not
40
+ symbols. Each ``AnnotatedFormula`` has its own ``symbols`` too, and the file's table follows
41
+ ``expand_includes``.
30
42
  * Quotes are removed from a word when the result is still a valid lower-case word,
31
43
  so ``'cat'`` and ``cat`` compare equal, as the specification requires. Any other
32
44
  quoted word keeps its quotes; use :func:`unquote` to get the raw text.
@@ -59,7 +71,7 @@ from typing import Optional, Union
59
71
 
60
72
  from . import _pyparser
61
73
  from .nodes import * # noqa: F403
62
- from .nodes import _error_at, __all__ as _node_names
74
+ from .nodes import _error_at, _symbol_table, __all__ as _node_names
63
75
 
64
76
  _LIBRARY_NAMES = ("DirectoryLibrary", "Library", "LibraryError", "OnlineLibrary", "ProblemNotFound",
65
77
  "RateLimited", "TarballLibrary", "fetch_problem", "open_library")
@@ -158,4 +170,4 @@ def _load(path: Path, root: Optional[Path], expand: bool, stack: tuple[Path, ...
158
170
  raise fail(f"{item.path!r} has no formulae named {sorted(missing)}")
159
171
  formulae = tuple(selected)
160
172
  inputs.extend(formulae)
161
- return TPTPFile(tuple(inputs), parsed.comments, parsed.filename)
173
+ return TPTPFile(tuple(inputs), parsed.comments, parsed.filename, _symbol_table(i for i in inputs if isinstance(i, AnnotatedFormula)))