Move the anchor attribute in ANCH(n)/REC(n), not just its values; release 0.5.1

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "pyregtab"
-version = "0.5.0"
+version = "0.5.1"
 edition = "2021"
 description = "Native core of pyRegTab: RTL compiler, ATP matcher and table interpreter"
 license = "MIT"

README.md

@@ -16,7 +16,7 @@ and interprets the match into a relational **recordset**:
 TableSyntax → RtlCompiler/TablePattern → AtpMatcher → TableInterpreter → Recordset
 ```
 
-**pyRegTab 0.5.0 ≙ jRegTab 0.5.0** (same API, same semantics, same test
+**pyRegTab 0.5.1 ≙ jRegTab 0.5.1** (same API, same semantics, same test
 corpus), including the
 embedded RTL DSL `pyregtab.dsl` — a port of jRegTab's `ru.icc.regtab.dsl`
 (added upstream in jRegTab 0.3.0). Python-side extras on top of the Java API:
@@ -143,7 +143,7 @@ Rust (`pyregtab._core`, built with [PyO3](https://pyo3.rs) and
 
 ## Testing
 
-`pytest tests` runs (1 908 tests):
+`pytest tests` runs (1 925 tests):
 
 - the full benchmark suite — tasks 001–150 (Foofah, RegTab, Baikal),
   every fixture variant, **both** via RTL patterns and via ATP patterns

conformance/UPSTREAM

@@ -1,3 +1,3 @@
-commit: 035ff1a139e885e4cea85aa66a33e89a6b30f8c9
-tag: v0.5.0
+commit: c12633763b309fd00f65d8b236a4ab91795303b4
+tag: v0.5.1
 path: conformance/

conformance/semantic/anch_named_attrs/expected.csv

@@ -0,0 +1,3 @@
+"Dato","Lokaler","Klasse"
+"20.05","AU","0"
+"11.06","A2.1","1"

conformance/semantic/anch_named_attrs/input.csv

@@ -0,0 +1,3 @@
+"Dato","Lokaler","Klasse"
+"20.05","AU","0"
+"11.06","A2.1","1"

conformance/semantic/anch_named_attrs/options.json

@@ -0,0 +1 @@
+{ "expectedHasHeader": true }

conformance/semantic/anch_named_attrs/pattern.rtl

@@ -0,0 +1,3 @@
+<ANCH(1)>
+[ [ATTR]+ ]
+[ COL->AVP [VAL] [VAL: ROW*->REC] [VAL] ]+

conformance/semantic/anch_named_inline_delim/expected.csv

@@ -0,0 +1,4 @@
+"Dato","Lokaler","Klasse"
+"20.05","AU","0"
+"20.05"," C1.1","0"
+"11.06","A2.1","1"

conformance/semantic/anch_named_inline_delim/input.csv

@@ -0,0 +1,3 @@
+"Dato","Lokaler","Klasse"
+"20.05","AU, C1.1","0"
+"11.06","A2.1","1"

conformance/semantic/anch_named_inline_delim/options.json

@@ -0,0 +1 @@
+{ "expectedHasHeader": true }

conformance/semantic/anch_named_inline_delim/pattern.rtl

@@ -0,0 +1,2 @@
+[ [ATTR]+ ]
+[ COL->AVP [VAL] [(VAL: ROW*->REC(1)){','}] [VAL] ]+

docs/api.md

@@ -124,6 +124,9 @@ byte-identical to `RtlCompiler.compile` for lambda-free patterns. See the
 with `pattern.with_transformations(...)` or to an interpreter with
 `.with_transformations([...])`. RTL settings `<NORM, ANCH(n), SPLIT("s")>` and
 inline `REC(n)` / `REC('s')` parameters compile to these transformations.
+`AnchorAttributeAtPosition(pos)` moves the anchor attribute — its name together with
+its values — to 0-based position `pos`; it renames nothing, so every attribute-value
+binding survives and only the schema order changes.
 
 ## RTL bindings
 

docs/index.md

@@ -109,4 +109,4 @@ Requires **Python 3.10+**; binary wheels for Windows, Linux, and macOS.
 ---
 
 !!! note "Status"
-    Current release: **0.5.0** (feature parity with jRegTab 0.5.0) · License: **MIT** · [PyPI](https://pypi.org/project/pyregtab/) · [GitHub](https://github.com/regtab/pyregtab)
+    Current release: **0.5.1** (feature parity with jRegTab 0.5.1) · License: **MIT** · [PyPI](https://pypi.org/project/pyregtab/) · [GitHub](https://github.com/regtab/pyregtab)

docs/model/atp.md

@@ -331,7 +331,7 @@ chosen operation (see [ITM — Interpretation actions](itm.md#interpretation-act
     | Operation | Python factory | Notes |
     |---|---|---|
     | `REC` | `ActionSpec.rec(providers…)` | anchor → first field; providers supply remaining fields |
-    | `REC(n)` | `ActionSpec.rec(int n, providers…)` | adds `AnchorAttributeAtPosition` post-step |
+    | `REC(n)` | `ActionSpec.rec(int n, providers…)` | adds `AnchorAttributeAtPosition` post-step — moves the anchor attribute (name with its values) to position *n* |
     | `REC('s')` | `ActionSpec.rec(String delim, providers…)` | adds `DelimitedFieldSplit` post-step |
     | `AVP` | `ActionSpec.avp(provider)` | associates VAL anchor with ATTR item |
     | `AVP "name"` | `ActionSpec.avp("ATTR_NAME")` | context-derived ATTR constant |

docs/model/itm.md

@@ -317,7 +317,8 @@ satisfy the constraints of the chosen operation (Tab. I in the paper):
     | AVP with literal | `ActionSpec.avp("ATTR_NAME")` | Context-derived ATTR constant |
 
     `ActionSpec.rec(int anchorPos, providers…)` adds an `AnchorAttributeAtPosition`
-    post-processing step (RTL: `REC(n)`).
+    post-processing step (RTL: `REC(n)`) — it moves the anchor attribute, name and
+    values together, to position `anchorPos` of the schema.
     `ActionSpec.rec(String splitDelimiter, providers…)` adds a `DelimitedFieldSplit`
     step (RTL: `REC('s')`).
 
@@ -431,7 +432,7 @@ The extracted recordset may be further post-processed by optional operations:
     | Delimited field split | `DelimitedFieldSplit` | `ActionSpec.rec(String delimiter, …)` — RTL `REC('s')` |
     | Field splitting | `FieldSplitting` | explicit split spec |
     | Whitespace normalisation | `WhitespaceNormalization` | `with_transformations(…)` |
-    | Anchor attribute at position | `AnchorAttributeAtPosition` | `ActionSpec.rec(int pos, …)` — RTL `REC(n)` |
+    | Anchor attribute at position | `AnchorAttributeAtPosition` | `ActionSpec.rec(int pos, …)` — RTL `REC(n)` / `<ANCH(n)>`; moves the anchor attribute (name with its values) to position `pos`, preserving every attribute-value binding |
 
 ---
 

docs/rtl-reference.md

@@ -131,15 +131,29 @@ post-extraction `RecordsetTransformation`s on the resulting `TablePattern`.
 | Setting | Effect |
 |---|---|
 | `NORM` | Apply whitespace normalisation to all field values after extraction |
-| `ANCH(n)` | Use position *n* in the first record as the attribute name for all records |
+| `ANCH(n)` | Move the anchor attribute to 0-based position *n* in the schema |
 | `SPLIT("s")` | Split all field values by delimiter *s* after extraction |
 
 Example: `<NORM, ANCH(2)> [ … ]` — normalise and anchor at position 2.
 
+The anchor is the first attribute of the extracted schema, and `ANCH(n)` moves that
+**attribute** — its name travels together with its values, so the attribute-value binding
+of every record is untouched and only the order of the schema changes. A position of 0,
+a position beyond the schema, or a single-attribute schema leaves the recordset as is.
+
+The rule is the same for named attributes (produced by `AVP`) and for the anonymous
+`$a_i` names the interpreter invents: an anonymous name is **not** renumbered, it moves
+with its attribute. A schema `$a_1, $a_2, $a_3` under `ANCH(2)` therefore becomes
+`$a_2, $a_3, $a_1` — the values sit in the same positions as before, and the name shows
+which attribute was moved.
+
 !!! note "Inline equivalents"
     The same two transformations can be requested *inline* on a `REC` action:
     `REC(n)` is equivalent to the `ANCH(n)` setting, and `REC('s')` is equivalent to
     `SPLIT("s")`. Inline forms are by far the more common in practice (see Tasks 02, 03).
+    The position of the inline form in the pattern does not matter — `REC(n)` is picked up
+    anywhere, including inside a delimited content specification such as
+    `[(VAL: ROW*->REC(1)){','}]`, and always yields the same transformation.
     The compiler merges inline and prefix forms and raises `RtlCompileError` if they
     conflict (e.g. `ANCH(1)` together with `REC(2)`).
 
@@ -397,7 +411,7 @@ provSpecs -> op
 |---|---|---|
 | `REC` | `prov->REC` | Anchor item → record entry; provider supplies additional field values |
 | `REC` | `()->REC` | Anchor item → single-field record (no additional providers; useful after `SUFFIX`/`PREFIX`/`FILL` has enriched the anchor value) |
-| `REC(n)` | `prov->REC(n)` | Same + use attribute at position *n* as the record's attribute name |
+| `REC(n)` | `prov->REC(n)` | Same + move the anchor attribute (name with its values) to position *n* |
 | `REC('s')` | `prov->REC('s')` | Same + split field values by delimiter *s* |
 | `AVP` | `prov->AVP` | Associate anchor (VAL) with an attribute from the provider (ATTR) |
 | `JOIN` | `prov->JOIN` | Join item-based records: all items included, then dedup by named attribute (K=∅) |