alphapulldown-input-parser 0.3.0__tar.gz → 0.5.0__tar.gz

{alphapulldown_input_parser-0.3.0/src/alphapulldown_input_parser.egg-info → alphapulldown_input_parser-0.5.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alphapulldown-input-parser
-Version: 0.3.0
+Version: 0.5.0
 Summary: Fold specification parser for AlphaPulldown
 Author-email: Kosinski Lab <alphapulldown@embl-hamburg.de>
 License: MIT
@@ -30,7 +30,7 @@ Dynamic: license-file
 Reusable parser for AlphaPulldown-style fold specifications. Install it with:
 
 ```bash
-pip install alphapulldown-input-parser
+pip install "alphapulldown-input-parser>=0.4.0"
 ```
 
 or, for local development:
@@ -39,12 +39,37 @@ or, for local development:
 pip install -e /path/to/alphapulldown-input-parser
 ```
 
-The package exposes two helpers:
+The package exposes three helpers:
 
 * `parse_fold(...)` – mirrors the historical AlphaPulldown helper and performs
   feature existence checks.
 * `expand_fold_specification(...)` – expands a single fold string without
   raising if features are missing.
+* `parse_fold_chains(spec, protein_delimiter="+")` – **pure-syntactic** parse
+  of a single spec into `(chain_name, copies, regions)` triples, with no
+  filesystem access or feature lookup. Useful for resource sizing or input
+  validation before features exist on disk.
 
 The parser is dependency-free and works across AlphaPulldown, the Snakemake
 pipeline, or any other tooling that consumes the same fold syntax.
+
+As of `0.4.0`, AF3 JSON feature files support the same copy/range suffixes as
+classic AlphaPulldown feature pickles, including discontinuous regions and copy
+counts. For example:
+
+```python
+parse_fold(
+    [
+        "P01258_af3_input.json:1-100",
+        "P01258_af3_input.json:1-100:150-200",
+        "P01258_af3_input.json:2:1-100:150-200+P01579_af3_input.json",
+    ],
+    features_directory=["/path/to/features"],
+    protein_delimiter="+",
+)
+```
+
+AlphaPulldown and AlphaPulldownSnakemake can then preserve those AF3 JSON
+regions during input preparation. For the AlphaFold 3 backend, discontinuous
+regions are expanded into separate cropped chains rather than one continuous
+polymer chain.

alphapulldown_input_parser-0.5.0/README.md

@@ -0,0 +1,48 @@
+# alphapulldown-input-parser
+
+Reusable parser for AlphaPulldown-style fold specifications. Install it with:
+
+```bash
+pip install "alphapulldown-input-parser>=0.4.0"
+```
+
+or, for local development:
+
+```bash
+pip install -e /path/to/alphapulldown-input-parser
+```
+
+The package exposes three helpers:
+
+* `parse_fold(...)` – mirrors the historical AlphaPulldown helper and performs
+  feature existence checks.
+* `expand_fold_specification(...)` – expands a single fold string without
+  raising if features are missing.
+* `parse_fold_chains(spec, protein_delimiter="+")` – **pure-syntactic** parse
+  of a single spec into `(chain_name, copies, regions)` triples, with no
+  filesystem access or feature lookup. Useful for resource sizing or input
+  validation before features exist on disk.
+
+The parser is dependency-free and works across AlphaPulldown, the Snakemake
+pipeline, or any other tooling that consumes the same fold syntax.
+
+As of `0.4.0`, AF3 JSON feature files support the same copy/range suffixes as
+classic AlphaPulldown feature pickles, including discontinuous regions and copy
+counts. For example:
+
+```python
+parse_fold(
+    [
+        "P01258_af3_input.json:1-100",
+        "P01258_af3_input.json:1-100:150-200",
+        "P01258_af3_input.json:2:1-100:150-200+P01579_af3_input.json",
+    ],
+    features_directory=["/path/to/features"],
+    protein_delimiter="+",
+)
+```
+
+AlphaPulldown and AlphaPulldownSnakemake can then preserve those AF3 JSON
+regions during input preparation. For the AlphaFold 3 backend, discontinuous
+regions are expanded into separate cropped chains rather than one continuous
+polymer chain.

{alphapulldown_input_parser-0.3.0 → alphapulldown_input_parser-0.5.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "alphapulldown-input-parser"
-version = "0.3.0"
+version = "0.5.0"
 description = "Fold specification parser for AlphaPulldown"
 readme = "README.md"
 license = {text = "MIT"}

{alphapulldown_input_parser-0.3.0 → alphapulldown_input_parser-0.5.0}/src/alphapulldown_input_parser/__init__.py

@@ -8,11 +8,13 @@ from .parser import (
     expand_fold_specification,
     generate_fold_specifications,
     parse_fold,
+    parse_fold_chains,
 )
 
 __all__ = [
     "expand_fold_specification",
     "parse_fold",
+    "parse_fold_chains",
     "FormatError",
     "FeatureIndex",
     "Region",

{alphapulldown_input_parser-0.3.0 → alphapulldown_input_parser-0.5.0}/src/alphapulldown_input_parser/parser.py

@@ -72,7 +72,8 @@ class RegionSelection:
         return cls(regions=None)
 
 
-# Either {"json_input": "/path/to.json"} or {"CHAIN_A": RegionSelection(...)}
+# Either {"json_input": "/path/to.json", "regions": RegionSelection(...)?}
+# or {"CHAIN_A": RegionSelection(...)}
 FoldEntry = Dict[str, Union[str, RegionSelection]]
 
 
@@ -275,6 +276,39 @@ def _parse_regions(region_tokens: Sequence[str], spec: str) -> RegionSelection:
     return RegionSelection(regions=tuple(regions))
 
 
+def parse_fold_chains(
+    spec: str,
+    protein_delimiter: str = "+",
+) -> List[Tuple[str, int, RegionSelection]]:
+    """Parse a single fold spec into ``(chain_name, copies, regions)`` triples.
+
+    Pure-syntactic parse — no filesystem access, no feature-index lookup —
+    intended for tools that need the chain composition of a spec *before* the
+    corresponding features exist on disk (e.g. for resource sizing or input
+    validation). Follows the AlphaPulldown ``name[:copies][:region...]``
+    convention; the chain ``name`` is returned **unchanged** (no path or
+    extension stripping; the caller can normalise if needed).
+
+    Example:
+        >>> parse_fold_chains("protA:2:1-100+protB", "+")
+        [('protA', 2, RegionSelection(regions=(Region(start=1, end=100),))),
+         ('protB', 1, RegionSelection(regions=None))]
+    """
+    chains: List[Tuple[str, int, RegionSelection]] = []
+    for raw_pf in str(spec).split(protein_delimiter):
+        pf = raw_pf.strip()
+        if not pf:
+            continue
+        tokens = [token.strip() for token in pf.split(":") if token.strip()]
+        if not tokens:
+            continue
+        name = tokens[0]
+        copies, region_tokens = _extract_copy_and_regions(tokens, spec)
+        regions = _parse_regions(region_tokens, spec)
+        chains.append((name, copies, regions))
+    return chains
+
+
 # ---------------------------------------------------------------------------
 # Expansion logic
 # ---------------------------------------------------------------------------
@@ -312,39 +346,21 @@ def expand_fold_specification(
         tokens = [token.strip() for token in pf.split(":")]
         base_token = tokens[0] if tokens else ""
 
-        # JSON inputs: support optional copy number, but no ranges.
+        # JSON inputs: support optional copy number and region ranges.
         if base_token.endswith(".json"):
             path_pf = Path(base_token)
             json_path: Optional[str] = None
             for json_key in (path_pf.name, path_pf.stem):
                 json_path = index.json_path(json_key)
                 if json_path:
-                    # Handle optional copy number for JSON inputs.
-                    if len(tokens) == 1:
-                        copies = 1
-                    else:
-                        extra_tokens = tokens[1:]
-                        # Ranges (e.g. "1-10") are not supported for JSON feature files.
-                        if any("-" in tok for tok in extra_tokens):
-                            _format_error(
-                                spec,
-                                msg="Region ranges are not supported for JSON feature files.",
-                            )
-                        if len(extra_tokens) != 1:
-                            _format_error(
-                                spec,
-                                msg="JSON feature files support only an optional copy number.",
-                            )
-                        try:
-                            copies = int(extra_tokens[0])
-                        except ValueError:
-                            _format_error(
-                                spec,
-                                msg="Copy number for JSON feature file must be an integer.",
-                            )
+                    copies, region_tokens = _extract_copy_and_regions(tokens, spec)
+                    regions = _parse_regions(region_tokens, spec)
 
                     for _ in range(copies):
-                        formatted_folds.append({"json_input": json_path})
+                        json_entry: FoldEntry = {"json_input": json_path}
+                        if not regions.is_all:
+                            json_entry["regions"] = regions
+                        formatted_folds.append(json_entry)
                     break
             if json_path:
                 continue

{alphapulldown_input_parser-0.3.0 → alphapulldown_input_parser-0.5.0/src/alphapulldown_input_parser.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alphapulldown-input-parser
-Version: 0.3.0
+Version: 0.5.0
 Summary: Fold specification parser for AlphaPulldown
 Author-email: Kosinski Lab <alphapulldown@embl-hamburg.de>
 License: MIT
@@ -30,7 +30,7 @@ Dynamic: license-file
 Reusable parser for AlphaPulldown-style fold specifications. Install it with:
 
 ```bash
-pip install alphapulldown-input-parser
+pip install "alphapulldown-input-parser>=0.4.0"
 ```
 
 or, for local development:
@@ -39,12 +39,37 @@ or, for local development:
 pip install -e /path/to/alphapulldown-input-parser
 ```
 
-The package exposes two helpers:
+The package exposes three helpers:
 
 * `parse_fold(...)` – mirrors the historical AlphaPulldown helper and performs
   feature existence checks.
 * `expand_fold_specification(...)` – expands a single fold string without
   raising if features are missing.
+* `parse_fold_chains(spec, protein_delimiter="+")` – **pure-syntactic** parse
+  of a single spec into `(chain_name, copies, regions)` triples, with no
+  filesystem access or feature lookup. Useful for resource sizing or input
+  validation before features exist on disk.
 
 The parser is dependency-free and works across AlphaPulldown, the Snakemake
 pipeline, or any other tooling that consumes the same fold syntax.
+
+As of `0.4.0`, AF3 JSON feature files support the same copy/range suffixes as
+classic AlphaPulldown feature pickles, including discontinuous regions and copy
+counts. For example:
+
+```python
+parse_fold(
+    [
+        "P01258_af3_input.json:1-100",
+        "P01258_af3_input.json:1-100:150-200",
+        "P01258_af3_input.json:2:1-100:150-200+P01579_af3_input.json",
+    ],
+    features_directory=["/path/to/features"],
+    protein_delimiter="+",
+)
+```
+
+AlphaPulldown and AlphaPulldownSnakemake can then preserve those AF3 JSON
+regions during input preparation. For the AlphaFold 3 backend, discontinuous
+regions are expanded into separate cropped chains rather than one continuous
+polymer chain.

{alphapulldown_input_parser-0.3.0 → alphapulldown_input_parser-0.5.0}/test/test_parser.py

@@ -18,6 +18,7 @@ from alphapulldown_input_parser import (
     RegionSelection,
     generate_fold_specifications,
     parse_fold,
+    parse_fold_chains,
 )
 
 
@@ -232,10 +233,40 @@ def patch_feature_index(monkeypatch):
             ["dir1"],
             "+",
             {"pkl": {}, "json": {"protein1.json": "dir1/protein1.json"}},
+            [[{"json_input": "dir1/protein1.json", "regions": selection_ranges((1, 10))}]],
             None,
-            FormatError,
             None,
-            id="json_with_range_not_supported",
+            id="json_with_range",
+        ),
+        pytest.param(
+            ["protein1.json:2:1-10:20-30"],
+            ["dir1"],
+            "+",
+            {"pkl": {}, "json": {"protein1.json": "dir1/protein1.json"}},
+            [
+                [
+                    {"json_input": "dir1/protein1.json", "regions": selection_ranges((1, 10), (20, 30))},
+                    {"json_input": "dir1/protein1.json", "regions": selection_ranges((1, 10), (20, 30))},
+                ]
+            ],
+            None,
+            None,
+            id="json_with_copy_and_regions",
+        ),
+        pytest.param(
+            ["protein1.json:1-10:20-30:2"],
+            ["dir1"],
+            "+",
+            {"pkl": {}, "json": {"protein1.json": "dir1/protein1.json"}},
+            [
+                [
+                    {"json_input": "dir1/protein1.json", "regions": selection_ranges((1, 10), (20, 30))},
+                    {"json_input": "dir1/protein1.json", "regions": selection_ranges((1, 10), (20, 30))},
+                ]
+            ],
+            None,
+            None,
+            id="json_with_regions_and_copy",
         ),
     ],
 )
@@ -309,3 +340,62 @@ def test_generate_fold_specifications_writes_to_disk(tmp_path: Path) -> None:
 
     assert result == ["p1+p2"]
     assert output_path.read_text(encoding="utf-8") == "p1+p2\n"
+
+
+# ---------------------------------------------------------------------------
+# parse_fold_chains
+# ---------------------------------------------------------------------------
+
+
+def test_parse_fold_chains_basic_heteromer() -> None:
+    assert parse_fold_chains("A+B") == [
+        ("A", 1, RegionSelection.all()),
+        ("B", 1, RegionSelection.all()),
+    ]
+
+
+def test_parse_fold_chains_copies() -> None:
+    # copy number as the second token (canonical form)
+    assert parse_fold_chains("A:2") == [("A", 2, RegionSelection.all())]
+    # copy + region: name:copies:region
+    assert parse_fold_chains("A:2:1-100") == [
+        ("A", 2, RegionSelection(regions=(Region(start=1, end=100),))),
+    ]
+
+
+def test_parse_fold_chains_region_without_copies() -> None:
+    # A region alone (not a bare integer) implies a single copy
+    assert parse_fold_chains("A:1-100") == [
+        ("A", 1, RegionSelection(regions=(Region(start=1, end=100),))),
+    ]
+
+
+def test_parse_fold_chains_multiple_regions_and_copies() -> None:
+    chains = parse_fold_chains("A:2:1-100:200-300+B")
+    assert chains[0][0] == "A"
+    assert chains[0][1] == 2
+    assert chains[0][2] == RegionSelection(
+        regions=(Region(start=1, end=100), Region(start=200, end=300))
+    )
+    assert chains[1] == ("B", 1, RegionSelection.all())
+
+
+def test_parse_fold_chains_preserves_paths_and_json_names() -> None:
+    """Names are returned verbatim — no extension or path stripping."""
+    chains = parse_fold_chains("/path/to/protA_af3_input.json:2+protB.fasta")
+    assert chains[0][0] == "/path/to/protA_af3_input.json"
+    assert chains[0][1] == 2
+    assert chains[1][0] == "protB.fasta"
+    assert chains[1][1] == 1
+
+
+def test_parse_fold_chains_custom_delimiter_and_whitespace() -> None:
+    assert parse_fold_chains(" A , B ", protein_delimiter=",") == [
+        ("A", 1, RegionSelection.all()),
+        ("B", 1, RegionSelection.all()),
+    ]
+    # empty tokens are skipped
+    assert parse_fold_chains("A++B") == [
+        ("A", 1, RegionSelection.all()),
+        ("B", 1, RegionSelection.all()),
+    ]

alphapulldown_input_parser-0.3.0/README.md

@@ -1,23 +0,0 @@
-# alphapulldown-input-parser
-
-Reusable parser for AlphaPulldown-style fold specifications. Install it with:
-
-```bash
-pip install alphapulldown-input-parser
-```
-
-or, for local development:
-
-```bash
-pip install -e /path/to/alphapulldown-input-parser
-```
-
-The package exposes two helpers:
-
-* `parse_fold(...)` – mirrors the historical AlphaPulldown helper and performs
-  feature existence checks.
-* `expand_fold_specification(...)` – expands a single fold string without
-  raising if features are missing.
-
-The parser is dependency-free and works across AlphaPulldown, the Snakemake
-pipeline, or any other tooling that consumes the same fold syntax.