feat(diagrams): add vector diagram fixtures and DOT ground truth

ATTRIBUTIONS.md

@@ -133,3 +133,10 @@ Vendored WordPerfect-family test files. Per-file provenance (upstream path, revi
 - **License:** CDLA-Permissive-1.0
 - **Used here:** 2 financial-report pages (`images/doclaynet_page_01..02.jpg`) from the `test` split (`NASDAQ_ATRI_2003.pdf` p24, `NYSE_MGM_2004.pdf` p49).
 - **Modifications:** authoritative PDF text cells linearized in reading order (top-to-bottom, then left-to-right). No OCR and no model used; text is DocLayNet's verbatim PDF-layer cell text.
+
+## Diagram fixtures (this repository)
+
+- **Source:** authored here, not derived from any third-party dataset.
+- **License:** same as this repository.
+- **Used here:** `diagrams/*.svg` plus their `diagrams/src/*.dot` sources and `ground_truth/dot/*.dot`.
+- **Modifications:** four SVGs are rendered from the committed `.dot` sources with Graphviz 15.1.1 (`dot`/`neato`), which is EPL-1.0 licensed; its output is not a derivative of the tool. `nested_transforms.svg` is hand-written. Ground truth is the source graph restated by node label, so it is independent of any recogniser's numbering.

diagrams/README.md

@@ -0,0 +1,61 @@
+# Diagram fixtures
+
+Vector diagrams for node/edge recovery (xberg-io/xberg#579). A vector diagram
+carries its own graph, so recovery from one is exact rather than probabilistic,
+and these fixtures are what say whether it actually is.
+
+## Why generated, not hand-drawn
+
+Hand-written SVG exercises almost none of what a real diagram tool emits. The
+four SVGs already in `xml/` have no transform chain, no arrowheads, no double
+borders and no curved connectors, and a recogniser can pass all four while being
+wrong about every file a user would actually bring.
+
+Rendering a known graph through Graphviz fixes that, and it also gives ground
+truth for free: the source `.dot` **is** the correct answer, so the corpus
+measures recovery rather than freezing whatever it currently produces.
+
+## Layout
+
+| file | exercises |
+|---|---|
+| `graphviz_flow.svg` | box, diamond and ellipse nodes; arrowheads; edge labels; a dashed edge; root `translate` with negative coordinates |
+| `graphviz_states.svg` | `doublecircle` (one node drawn as two concentric outlines); a pair of antiparallel edges between adjacent nodes |
+| `graphviz_network.svg` | undirected `--` edges, so no arrowhead anywhere; `neato` layout |
+| `graphviz_bidirectional.svg` | `dir=both` and `dir=back` |
+| `nested_transforms.svg` | hand-authored: nested `translate`/`scale` groups plus a viewBox that differs from the viewport, so nothing sits at the coordinate it is written at |
+| `src/*.dot` | the Graphviz sources, so every SVG above is regenerable |
+
+`ground_truth/dot/<stem>.dot` holds the graph each fixture draws, keyed by node
+label rather than by generated id so it does not depend on any one recogniser's
+numbering. Two entries are deliberately empty, `data_dashboard` and
+`simple_svg`: those files are a bar chart and a two-shape drawing, they are not
+diagrams, and recovering a graph from either would be a false positive.
+
+Ground truth also covers `xml/org_chart.svg` and `xml/flowchart.svg`, which stay
+where they are.
+
+## Regenerate
+
+```
+brew install graphviz     # or apt-get install graphviz
+cd diagrams/src
+for f in graphviz_flow graphviz_states graphviz_bidirectional; do
+  dot -Tsvg "$f.dot" -o "../$f.svg"
+done
+neato -Tsvg graphviz_network.dot -o ../graphviz_network.svg
+```
+
+Output is stable for a given Graphviz version. This was built with 15.1.1;
+a different version may lay the graphs out differently, which changes
+coordinates but not the graph, and the ground truth is written in terms of the
+graph.
+
+`nested_transforms.svg` is hand-authored and is not regenerated.
+
+## Not included
+
+A vector PDF of the same graph (`dot -Tpdf`) would exercise the PDF path, but
+`*.pdf` is excluded from git here and lives in the corpus bucket, so it has to
+go through `scripts/publish_corpus.py`. Happy to supply the file for a
+maintainer to publish.

diagrams/src/graphviz_bidirectional.dot

@@ -0,0 +1,10 @@
+digraph graphviz_bidirectional {
+  rankdir=LR;
+  node [shape=box style=filled fillcolor="#fdb462"];
+  api [label="API"];
+  db  [label="Database"];
+  cache [label="Cache"];
+  api -> db [dir=both];
+  api -> cache;
+  cache -> db [dir=back];
+}

diagrams/src/graphviz_flow.dot

@@ -0,0 +1,11 @@
+digraph graphviz_flow {
+  rankdir=TB;
+  a [label="Start" shape=box style=filled fillcolor="#a6cee3"];
+  b [label="Validate" shape=diamond style=filled fillcolor="#b2df8a"];
+  c [label="Process" shape=box style=filled fillcolor="#fb9a99"];
+  d [label="Done" shape=ellipse style=filled fillcolor="#fdbf6f"];
+  a -> b;
+  b -> c [label="ok"];
+  b -> d [label="skip" style=dashed];
+  c -> d;
+}

diagrams/src/graphviz_network.dot

@@ -0,0 +1,13 @@
+graph graphviz_network {
+  layout=neato;
+  node [shape=ellipse style=filled fillcolor="#cccccc"];
+  gw [label="Gateway" fillcolor="#8dd3c7"];
+  s1 [label="Switch A" fillcolor="#ffffb3"];
+  s2 [label="Switch B" fillcolor="#ffffb3"];
+  h1 [label="Host 1"];
+  h2 [label="Host 2"];
+  gw -- s1;
+  gw -- s2;
+  s1 -- h1;
+  s2 -- h2;
+}