hyf

commit d17f335f02e068ea59429f5de58a4ee5b1b4d96f
parent 067e7ce760f162cceecc2066a58f3496089420fe
Author: triesap <tyson@radroots.org>
Date:   Mon, 21 Sep 2026 13:03:41 +0000

core: add source-span candidate discovery

Diffstat:
A
src/hyf_core/normalization/candidates.mojo
 | 
94
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
M
tests/test_domain.mojo
 | 
32
++++++++++++++++++++++++++++++++

2 files changed, 126 insertions(+), 0 deletions(-)
diff --git a/src/hyf_core/normalization/candidates.mojo b/src/hyf_core/normalization/candidates.mojo
@@ -0,0 +1,94 @@
+from std.collections import List
+
+
+@fieldwise_init
+struct Candidate(Copyable, Movable):
+    var kind: String
+    var text: String
+    var start: Int
+    var end: Int
+
+
+def _is_digit_byte(byte: UInt8) -> Bool:
+    return byte >= UInt8(ord("0")) and byte <= UInt8(ord("9"))
+
+
+def _backward_number_start(text: String, unit_start: Int) -> Int:
+    var bytes = text.as_bytes()
+    var index = unit_start - 1
+    while index >= 0:
+        var byte = bytes[index]
+        if byte == UInt8(ord(" ")) or byte == UInt8(ord("\t")):
+            index -= 1
+            continue
+        break
+    var end = index
+    while index >= 0 and _is_digit_byte(bytes[index]):
+        index -= 1
+    if index == end:
+        return -1
+    return index + 1
+
+
+def discover_candidates(
+    text: String, known_products: List[String], known_units: List[String], known_dates: List[String]
+) -> List[Candidate]:
+    var candidates = List[Candidate]()
+    var lowered = text.lower()
+
+    for product in known_products:
+        if product.strip() == "":
+            continue
+        var index = lowered.find(product.lower())
+        if index >= 0:
+            candidates.append(
+                Candidate(
+                    kind="product",
+                    text=String(product),
+                    start=index,
+                    end=index + product.byte_length(),
+                )
+            )
+
+    for unit in known_units:
+        if unit.strip() == "":
+            continue
+        var index = lowered.find(unit.lower())
+        if index >= 0:
+            var number_start = _backward_number_start(text, index)
+            if number_start >= 0:
+                candidates.append(
+                    Candidate(
+                        kind="quantity",
+                        text=String(text[byte=number_start:index + unit.byte_length()]),
+                        start=number_start,
+                        end=index + unit.byte_length(),
+                    )
+                )
+
+    for date_word in known_dates:
+        if date_word.strip() == "":
+            continue
+        var index = lowered.find(date_word.lower())
+        if index >= 0:
+            candidates.append(
+                Candidate(
+                    kind="date",
+                    text=String(date_word),
+                    start=index,
+                    end=index + date_word.byte_length(),
+                )
+            )
+    return candidates^
+
+
+def candidate_kinds(candidates: List[Candidate]) -> List[String]:
+    var kinds = List[String]()
+    for candidate in candidates:
+        var seen = False
+        for kind in kinds:
+            if kind == candidate.kind:
+                seen = True
+        if not seen:
+            kinds.append(String(candidate.kind))
+    return kinds^
diff --git a/tests/test_domain.mojo b/tests/test_domain.mojo
@@ -559,3 +559,35 @@ def test_execution_metadata_is_separate_from_business_outcome() raises:
         _ = execution_meta("degraded", 0, None, None, None)
     with assert_raises():
         _ = execution_meta("weird", 0, None, None, None)
+
+
+from hyf_core.normalization.candidates import Candidate, candidate_kinds, discover_candidates
+
+
+def test_candidate_discovery_spans_and_unknown_products() raises:
+    var products = List[String]()
+    products.append("roma tomatoes")
+    var units = List[String]()
+    units.append("lb")
+    var dates = List[String]()
+    dates.append("friday")
+    var candidates = discover_candidates(
+        "Got about 80 lb of Roma tomatoes. Can deliver Friday.", products, units, dates
+    )
+    var kinds = candidate_kinds(candidates)
+    assert_true(len(kinds) == 3)
+    var saw_quantity = False
+    for candidate in candidates:
+        if candidate.kind == "quantity":
+            assert_equal(candidate.text, "80 lb")
+            saw_quantity = True
+    assert_true(saw_quantity)
+
+    var unknown_products = List[String]()
+    unknown_products.append("kohlrabi")
+    var empty_units = List[String]()
+    var empty_dates = List[String]()
+    assert_equal(
+        len(discover_candidates("mystery greens", unknown_products, empty_units, empty_dates)),
+        0,
+    )