@mailwoman/codex 8.2.0 → 8.4.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +21 -0
- package/package.json +2 -2
package/README.md
CHANGED
|
@@ -67,6 +67,27 @@ import { ADDRESS_SYSTEM_CONVENTIONS, conventionsForSystem } from "@mailwoman/cod
|
|
|
67
67
|
- **Single source of truth.** The resolver, the decoder's convention masks, the
|
|
68
68
|
corpus synthesis layer, and the matcher all import from `@mailwoman/codex`.
|
|
69
69
|
|
|
70
|
+
## The normative tier (codex vs the libpostal dictionaries)
|
|
71
|
+
|
|
72
|
+
Mailwoman carries two closed-class vocabularies that overlap on purpose and must not
|
|
73
|
+
be merged:
|
|
74
|
+
|
|
75
|
+
- **Codex is normative.** USPS Pub-28 (and each system's equivalent) verbatim: the
|
|
76
|
+
canonical word, every _recognized_ variant, and the one _preferred_ abbreviation.
|
|
77
|
+
Its consumers are precision-shaped — corpus synthesis recipes, the eval harness's
|
|
78
|
+
invariance transforms, and formatting (rendering `N` vs `North` requires knowing
|
|
79
|
+
which form the authority prints).
|
|
80
|
+
- **The libpostal dictionaries** (`core/data/libpostal/dictionaries/`, Pelias
|
|
81
|
+
lineage) **are descriptive**: everything people actually write, including forms no
|
|
82
|
+
authority recognizes (`en/directionals.txt` lists `lower`/`upper`/`central`).
|
|
83
|
+
Their consumers are recall-shaped — evidence-lexicon curation laws, street
|
|
84
|
+
decomposition for training gold, the street-morphology FST. See the README in
|
|
85
|
+
that directory for the full consumer map and the four-tier curated-data layering.
|
|
86
|
+
|
|
87
|
+
Broadening codex with descriptive forms would corrupt formatting; narrowing the
|
|
88
|
+
descriptive lists to normative forms would weaken the evidence guards. Different
|
|
89
|
+
questions, different tables.
|
|
90
|
+
|
|
70
91
|
## Related
|
|
71
92
|
|
|
72
93
|
- [`@mailwoman/core`](../core) — `ComponentTag` schema, pipeline infrastructure
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@mailwoman/codex",
|
|
3
|
-
"version": "8.
|
|
3
|
+
"version": "8.4.0",
|
|
4
4
|
"description": "Per-address-system postal reference data + branded types (USPS street suffixes, US ZIP codes). Pure, zero-runtime-dep — the shared canonical home for postal-system primitives the parser, resolver, and synthesis layers all reach for.",
|
|
5
5
|
"license": "AGPL-3.0-only OR LicenseRef-Commercial",
|
|
6
6
|
"repository": {
|
|
@@ -124,6 +124,6 @@
|
|
|
124
124
|
"type-fest": "^5.8.0"
|
|
125
125
|
},
|
|
126
126
|
"devDependencies": {
|
|
127
|
-
"@mailwoman/annotations": "8.
|
|
127
|
+
"@mailwoman/annotations": "8.4.0"
|
|
128
128
|
}
|
|
129
129
|
}
|