mirror of
https://github.com/hanzoai/semantic.git
synced 2026-10-30 15:04:19 +00:00
zeekayandHanzo Dev
1cd0cd7644
The first commit was the shape. This is the implementation: every stage of the pipeline, the graph and its measures, the type system over it, entailment, deduplication and conflict resolution, the decision layer, the stores and the export formats. Twenty-six thousand lines of Go and sixteen thousand of tests, against no dependency but the standard library. Where the Python reaches for torch, spacy or a database driver, the Go states an interface and reports what is missing rather than pretending. A caller can tell "this build cannot serve that" from "nobody has heard of it", which is the difference between a port and a stub. The deterministic paths are complete and are what the tests exercise; the model-shaped ones are one adapter away. Porting a bug is not parity, so several places now disagree with the original on purpose. The JSON repair that closed a truncated reply counted each bracket kind separately and closed them backwards on the one input it exists for; it now works off a stack and ignores brackets inside strings. The entity patterns joined words with \s, so a heading ran into the sentence beneath it and became one name while the real relation was lost; a name is separated by horizontal space now. Eigenvector centrality ran plain power iteration, which on a bipartite graph swings between the two ends of a sign-paired eigenvalue forever and reported every node in a star as equally central; it iterates with A+I, which moves no eigenvector and settles. Similarity summed its weighted parts in map order, and float addition is not associative, so the same comparison scored differently between runs and duplicate detection returned a different answer each time. An inferred ontology named object-property endpoints by raw source type and data-property endpoints by class, so the schema referred to classes it did not contain. A property with no IRI of its own was minted as a class and came back Pascal-cased, renaming the term on the way out of the document. The command built a pipeline with every stage nil and panicked on each invocation. It is now the composition the package documents, and Run reports a missing ingester instead of dereferencing it. Co-authored-by: Hanzo Dev <[email protected]>
132 lines
3.2 KiB
Go
132 lines
3.2 KiB
Go
package store
|
|
|
|
import (
|
|
"math"
|
|
"testing"
|
|
)
|
|
|
|
const tol = 1e-9
|
|
|
|
func eq(t *testing.T, got, want float64) {
|
|
t.Helper()
|
|
if math.Abs(got-want) > tol {
|
|
t.Errorf("got %.17g, want %.17g", got, want)
|
|
}
|
|
}
|
|
|
|
func TestDot(t *testing.T) {
|
|
for _, c := range []struct {
|
|
name string
|
|
a, b []float32
|
|
want float64
|
|
}{
|
|
{"orthogonal", []float32{1, 0}, []float32{0, 1}, 0},
|
|
{"same", []float32{1, 2, 3}, []float32{4, 5, 6}, 32},
|
|
{"opposed", []float32{1, 2}, []float32{-1, -2}, -5},
|
|
{"empty", nil, nil, 0},
|
|
{"short second is zero padded", []float32{1, 2}, []float32{3}, 3},
|
|
{"short first is zero padded", []float32{3}, []float32{1, 2}, 3},
|
|
} {
|
|
t.Run(c.name, func(t *testing.T) { eq(t, Dot(c.a, c.b), c.want) })
|
|
}
|
|
}
|
|
|
|
func TestNorm(t *testing.T) {
|
|
for _, c := range []struct {
|
|
name string
|
|
v []float32
|
|
want float64
|
|
}{
|
|
{"three four five", []float32{3, 4}, 5},
|
|
{"unit", []float32{1, 0, 0}, 1},
|
|
{"zero", []float32{0, 0}, 0},
|
|
{"empty", nil, 0},
|
|
{"negative", []float32{-3, -4}, 5},
|
|
} {
|
|
t.Run(c.name, func(t *testing.T) { eq(t, Norm(c.v), c.want) })
|
|
}
|
|
}
|
|
|
|
func TestUnit(t *testing.T) {
|
|
got := Unit([]float32{3, 4})
|
|
want := []float32{0.6, 0.8}
|
|
for i := range want {
|
|
eq(t, float64(got[i]), float64(want[i]))
|
|
}
|
|
// A unit vector held in float32 is unit to float32 precision.
|
|
if n := Norm(got); math.Abs(n-1) > 1e-6 {
|
|
t.Errorf("length is %v, want 1", n)
|
|
}
|
|
|
|
zero := Unit([]float32{0, 0})
|
|
if len(zero) != 2 || zero[0] != 0 || zero[1] != 0 {
|
|
t.Errorf("zero vector became %v, want it left alone", zero)
|
|
}
|
|
|
|
v := []float32{3, 4}
|
|
Unit(v)
|
|
if v[0] != 3 || v[1] != 4 {
|
|
t.Errorf("Unit changed its argument: %v", v)
|
|
}
|
|
}
|
|
|
|
func TestCos(t *testing.T) {
|
|
// 32 / (sqrt(14) * sqrt(77)), computed by hand.
|
|
const skew = 0.9746318461970762
|
|
for _, c := range []struct {
|
|
name string
|
|
a, b []float32
|
|
want float64
|
|
}{
|
|
{"identical", []float32{1, 0}, []float32{1, 0}, 1},
|
|
{"orthogonal", []float32{1, 0}, []float32{0, 1}, 0},
|
|
{"opposed", []float32{1, 0}, []float32{-1, 0}, -1},
|
|
{"length ignored", []float32{1, 1}, []float32{5, 5}, 1},
|
|
{"skew", []float32{1, 2, 3}, []float32{4, 5, 6}, skew},
|
|
{"zero has no direction", []float32{0, 0}, []float32{1, 1}, 0},
|
|
} {
|
|
t.Run(c.name, func(t *testing.T) { eq(t, Cos(c.a, c.b), c.want) })
|
|
}
|
|
}
|
|
|
|
func TestDist(t *testing.T) {
|
|
for _, c := range []struct {
|
|
name string
|
|
a, b []float32
|
|
want float64
|
|
}{
|
|
{"same point", []float32{1, 2}, []float32{1, 2}, 0},
|
|
{"three four five", []float32{1, 2}, []float32{4, 6}, 5},
|
|
{"one axis", []float32{0}, []float32{2}, 2},
|
|
{"short is zero padded", []float32{3}, []float32{0, 4}, 5},
|
|
} {
|
|
t.Run(c.name, func(t *testing.T) { eq(t, Dist(c.a, c.b), c.want) })
|
|
}
|
|
}
|
|
|
|
func TestMetricScore(t *testing.T) {
|
|
a, b := []float32{1, 2}, []float32{4, 6}
|
|
for _, c := range []struct {
|
|
metric Metric
|
|
want float64
|
|
}{
|
|
{Cosine, Cos(a, b)},
|
|
{Inner, 16},
|
|
{Euclid, -5},
|
|
{"", Cos(a, b)},
|
|
{"nonsense", Cos(a, b)},
|
|
} {
|
|
t.Run(string(c.metric), func(t *testing.T) { eq(t, c.metric.Score(a, b), c.want) })
|
|
}
|
|
}
|
|
|
|
func TestMetricOrdersNearestFirst(t *testing.T) {
|
|
q := []float32{1, 0}
|
|
near, far := []float32{1, 0.1}, []float32{0, 1}
|
|
for _, m := range []Metric{Cosine, Inner, Euclid} {
|
|
if m.Score(q, near) <= m.Score(q, far) {
|
|
t.Errorf("%s scored the far vector at least as near", m)
|
|
}
|
|
}
|
|
}
|