vehicles 0.9.0 → 0.9.2

data/data/PROVENANCE-vehicles.md

@@ -4,8 +4,8 @@
 (the gem projection) at a release tag:
 
 - **Source**: https://github.com/vehiclesdb/vehiclesdb `dist/vehicles.json`
-- **Tag**: `v2026.11.0` (commit `a5df1047239f1b9a017052cea34eca8a5c947dfd`), dataset version `2026.11.0`
-- **country_ranks**: 15516 model(s) carried it natively; 0 merged from
+- **Tag**: `v2026.11.6` (commit `7b2b37f41e676ad42cdc0b1dfed1460670f1595b`), dataset version `2026.11.6`
+- **country_ranks**: 16223 model(s) carried it natively; 0 merged from
   the same tag's `catalog/<kind>/models.json` `popularity.by_country[cc].rank`
 - **License**: CC-BY 4.0 (see the data repo's LICENSE + ATTRIBUTION.md)
 

data/data/plates/PROVENANCE.md

@@ -1,10 +1,10 @@
-# Bundled plates dataset (PRD-PLATES, full corpus)
+# Bundled plates dataset (full corpus)
 
 Copy of `plates/` from the public data repo (EXCLUDING `_art/`),
 refreshed on gem releases:
 
 - **Source**: https://github.com/vehiclesdb/vehiclesdb `plates/`
-- **Commit**: `c7dd65c6f4144d39c62706b7be10231a12bc27e1`
+- **Commit**: `db0ec0abf7eae90b719a0a66f40b231f903bfac6`
 - **License**: CC-BY 4.0 (see the data repo's LICENSE + ATTRIBUTION.md)
 
 Refresh: `rake plates:refresh` (this file is generated by it).

data/data/plates/_decode/at-districts.yml

@@ -8,7 +8,7 @@
 # discharges the delegation by pointing at ANLAGE 5d: "Die Bezeichnung der den
 # Zulassungsschein ausstellenden Behörde im Kennzeichen bestimmt sich nach
 # Anlage 5d." Anlage 5d is a CLOSED, PRIMARY, 98-row table printed inside the
-# regulation — so unlike the German ~400-code Bundesanzeiger list (PRD-PLATES
+# regulation — so unlike the German ~400-code Bundesanzeiger list (the plates spec
 # §2.4), it is authored here in full, verbatim, in the annex's own order
 # (by Bundesland, then as the annex prints them).
 #
@@ -23,7 +23,7 @@
 # at-historic, at-electric, at-dealer-probefahrt, at-temporary-*,
 # at-trailer-foreign, at-wunschkennzeichen) via region_encoding.
 #
-# PERIOD DISCIPLINE (PRD-PLATES §2.4): these are the codes in the CONSOLIDATED
+# PERIOD DISCIPLINE (the plates spec §2.4): these are the codes in the CONSOLIDATED
 # Anlage 5d in force on the access date (Fassung im RIS seit 26.09.2024,
 # NOR40265416). Austrian district reform has repeatedly merged Bezirke
 # (Bruck-Mürzzuschlag, Hartberg-Fürstenfeld, Murtal and Südoststeiermark are

data/data/plates/_decode/cn-issuing-offices.yml

@@ -44,7 +44,7 @@
 #            豫V (Zhengzhou), announced by 郑州市人民政府 for 2020-09-16 and
 #            cited below — it is NOT in Annex A because Annex A predates it.
 #
-# PERIOD DISCIPLINE (PRD-PLATES § 2.4). Rows are the 2007 state and are never
+# PERIOD DISCIPLINE (the plates spec § 2.4). Rows are the 2007 state and are never
 # deleted: a plate issued under an old code stays on the road, and several of
 # the place names below have since been renamed or reorganised (襄樊市 →
 # 襄阳市 2010; 巢湖市 dissolved 2011; 铜仁地区/毕节地区/海东地区/吐鲁番地区/
@@ -92,7 +92,7 @@ period_note: >-
   2007 is the year GA 36-2007 came into force (实施 2007-11-01), NOT the year
   these codes began: most of them were already allocated under GA 36-1992 and
   the 1992-format plates that carry them were issued from 1992 onward. This is
-  the instrument-in-force case in PRD-PLATES § 2.4 exactly — a consolidation
+  the instrument-in-force case in the plates spec § 2.4 exactly — a consolidation
   documenting an older practice. The 2014 and 2018 editions each reissued
   Annex A and neither text is public, so no row here can be dated later than
   2007 and none can be shown to have been withdrawn.

data/data/plates/_decode/cn-provinces.yml

@@ -68,7 +68,7 @@
 # rather than by this table will disagree with the road.
 #
 # ============================================================================
-# PERIOD DISCIPLINE (PRD-PLATES § 2.4) — ONE ROW IS YOUNGER THAN THE OTHERS
+# PERIOD DISCIPLINE (the plates spec § 2.4) — ONE ROW IS YOUNGER THAN THE OTHERS
 # ============================================================================
 # GA 36-2007's 前言 lists, among the changes that edition made to GA 36-1992,
 # verbatim: "――增加了重庆的简称；" and "――增加了"渝"、"港"、"澳"、"使"、"警"、
@@ -591,7 +591,7 @@ limits:
     dataset with the British mark, which names the six-month period of first
     registration to 2050.
   cross_file_divergences: >-
-    RECORDED, NOT AVERAGED (PRD-PLATES § 5.3). Two `row_period_start` values in
+    RECORDED, NOT AVERAGED (the plates spec § 5.3). Two `row_period_start` values in
     `classification_han` above disagree with the period on the plates/cn.yml
     series that uses the same character, and the disagreement is left standing
     for the verifier rather than reconciled by one file overwriting the other.

data/data/plates/_decode/cz-kraj.yml

@@ -21,7 +21,7 @@
 # parse API that learned one and inferred the other would be wrong on twelve of
 # the fourteen rows. Both are authored here, in the instrument's own order.
 #
-# PERIOD DISCIPLINE (PRD-PLATES §2.4). The fourteen kraje were created by
+# PERIOD DISCIPLINE (the plates spec §2.4). The fourteen kraje were created by
 # ústavní zákon č. 347/1997 Sb. and have not been re-cut since, so unlike the
 # German and Austrian tables this one has no withdrawn rows and no merger
 # history. What DID change is the meaning of the code on a plate, and it

data/data/plates/_decode/dk-serial-blocks.yml

@@ -39,7 +39,7 @@
 #   * `secondary-wikipedia` — the rest of the pre-2012 rows. Both the English
 #     and the Danish Wikipedia article carry a pre-2012 block table, each
 #     crediting the enthusiast register nrpl.dk. THE TWO TABLES DISAGREE, and
-#     PRD-PLATES §5.3 says record disagreements rather than average them, so
+#     the plates spec §5.3 says record disagreements rather than average them, so
 #     conflicting rows carry both readings and a `conflict:` note. Nothing here
 #     is presented as an instrument.
 #

data/data/plates/_decode/es-provinces.yml

@@ -10,7 +10,7 @@
 #
 # Referenced by: es-provincial (plates/es.yml) via region_encoding.
 #
-# PERIOD DISCIPLINE (PRD-PLATES §2.4 — "province codes changed over time"):
+# PERIOD DISCIPLINE (the plates spec §2.4 — "province codes changed over time"):
 # these codes are the set in force in the CONSOLIDATED Anexo XVIII, and they
 # use the CURRENT official province names (Illes Balears, Girona, Lleida,
 # Ourense, A Coruña). Historic contraseñas that predate the RGV and appear on

data/data/plates/_decode/fr-departements.yml

@@ -21,7 +21,7 @@
 #      avoir de lien avec le domicile du titulaire du certificat
 #      d'immatriculation." A SIV plate DOES NOT DECODE. Reading region off a
 #      modern French plate is the single most expensive mistake a plate-parse
-#      API can make, and PRD-PLATES §2.4 already names it ("France's SIV
+#      API can make, and the plates spec §2.4 already names it ("France's SIV
 #      département band is OWNER-CHOSEN, not geographic ... the decode table
 #      must say so or the parse API will lie"). It says so.
 #
@@ -42,7 +42,7 @@
 # 96 metropolitan départements including Corse-du-Sud (2A) and Haute-Corse (2B),
 # plus the five overseas départements 971-974 and 976. Fetched 2026-08-02.
 #
-# PERIOD DISCIPLINE (PRD-PLATES §2.4 — decode tables rot). This is the list in
+# PERIOD DISCIPLINE (the plates spec §2.4 — decode tables rot). This is the list in
 # force TODAY. Historic FNI codes that are NOT in it are NOT invented here:
 #   * 20 = Corse before the 1976 split into 2A/2B;
 #   * 75 covering the whole former Seine before the 1968 Île-de-France

data/data/plates/_decode/gb-age-identifiers.yml

@@ -5,7 +5,7 @@
 # § 23(4) does not reach the composition of a mark, so the twice-yearly code
 # is an administrative allocation published in DVLA's leaflet INF104 and, for
 # the export series, in HMRC's VAT manual. Both are authority publications and
-# therefore PRIMARY under PRD-PLATES § 4.
+# therefore PRIMARY under the plates spec § 4.
 #
 # THE RULE, and it is complete and mechanical:
 #   a vehicle first registered March-August of year YY takes the code YY

data/data/plates/_decode/gb-memory-tags.yml

@@ -9,7 +9,7 @@
 # Secretary of State under § 23(1), and DVLA publishes it in the leaflet
 # INF104. So where plates/_decode/es-provinces.yml is enacted law and the
 # German district list is a Bundesanzeiger publication, this is a leaflet —
-# and a leaflet from the issuing authority is still a PRIMARY (PRD-PLATES § 4:
+# and a leaflet from the issuing authority is still a PRIMARY (the plates spec § 4:
 # "Government/authority pages ... PRIMARY").
 #
 # Every row below is INF104 (4/22) § 05 "DVLA memory tags and age identifiers",

data/data/plates/_decode/gr-prefectures.yml

@@ -77,7 +77,7 @@
 # plate. This is the Greek analogue of the two-table trap recorded in
 # plates/_decode/it-provinces.yml.
 #
-# PERIOD DISCIPLINE (PRD-PLATES §2.4). 34 of the 195 rows are combinations that
+# PERIOD DISCIPLINE (the plates spec §2.4). 34 of the 195 rows are combinations that
 # are ALLOCATED BUT NOT YET IN ISSUE (`status: allocated-not-yet-issued`). They
 # are kept, not dropped, because they are the forward half of the allocation —
 # but a consumer must not report them as observed geography, and a validator

data/data/plates/_decode/hr-cities.yml

@@ -81,7 +81,7 @@ period_note: >-
   period, so the 1990s code set is not asserted here.
 
 # ---------------------------------------------------------------------------
-# THE DIACRITIC PROBLEM — a PRD-PLATES §2.6 finding, recorded, not folded away.
+# THE DIACRITIC PROBLEM — a the plates spec §2.6 finding, recorded, not folded away.
 # ---------------------------------------------------------------------------
 diacritics:
   affected_codes: ["ČK", "KŽ", "PŽ", "ŠI", "VŽ", "ŽU"]
@@ -91,7 +91,7 @@ diacritics:
     it as A-Z 0-9), so no `format.pattern` or `format.regex` in plates/hr.yml
     can spell them: the area-code position is written `[A-Z]{2}` there and is
     therefore NARROWER than the issuing grammar for these six areas. Per
-    PRD-PLATES §2.7 that is still `matching: strict` — narrower is strict — but
+    the plates spec §2.7 that is still `matching: strict` — narrower is strict — but
     a consumer matching a photographed Croatian plate MUST either ASCII-fold the
     input or widen the class itself. Extending the serial alphabet is a schema
     amendment (§2.6), not a data entry, and is NOT attempted here.
@@ -110,7 +110,7 @@ diacritics:
 
 # ---------------------------------------------------------------------------
 # What moved between the 2008 annex and the 2017 annex, from diffing the two
-# primaries. Recorded because decode tables rot (PRD-PLATES §9) and because it
+# primaries. Recorded because decode tables rot (the plates spec §9) and because it
 # shows the annex is maintained at the MUNICIPALITY level, not the code level.
 # ---------------------------------------------------------------------------
 changed_2008_2017:

data/data/plates/_decode/ie-counties.yml

@@ -23,7 +23,7 @@
 #     POST-MERGER identifiers T (Tipperary), L (Limerick City and County) and
 #     W (Waterford City and County).
 #
-# THE DISCREPANCY IS REAL AND IS RECORDED, NOT AVERAGED (PRD-PLATES §5.3): the
+# THE DISCREPANCY IS REAL AND IS RECORDED, NOT AVERAGED (the plates spec §5.3): the
 # VRT First Schedule table still reads TN / TS / LK / L / WD / W and has never
 # been amended to add T or to merge LK into L and WD into W, while the
 # trade-licence table WAS amended for exactly that reason — S.I. 328/2013's own
@@ -36,7 +36,7 @@
 # Referenced by: plates/ie.yml (ie-standard-1987, ie-standard-1991,
 # ie-standard-2013, ie-trade-1993) via region_encoding.
 #
-# PERIOD DISCIPLINE (PRD-PLATES §2.4): these are the marks assignable to
+# PERIOD DISCIPLINE (the plates spec §2.4): these are the marks assignable to
 # vehicles FIRST REGISTERED on or after 1 January 1987. The pre-1987 marks are a
 # different, disjoint set and are handled in `historic_pre_1987` below — they are
 # NOT in `codes:` and a pre-1987 plate must never be decoded through it. Marks
@@ -193,7 +193,7 @@ historic_pre_1987:
   exhaustive_list_note: >-
     The complete pre-1987 mark list is the 1958 Schedule as amended fourteen
     times, and merging those fourteen insertions row-by-row is gate-L4 work
-    (PRD-PLATES §7). It is deliberately NOT half-done here: a partial list read
+    (the plates spec §7). It is deliberately NOT half-done here: a partial list read
     as complete would make a parse API declare real plates invalid, which is
     worse than declaring them unresolved. The amending instruments, all read and
     all conforming to the expansion rule above, are pinned so the merge can be

data/data/plates/_decode/in-state-codes.yml

@@ -98,7 +98,7 @@ cautions:
     are the registering authority (RTO district) code, which no central
     instrument enumerates and which each State creates, merges and renumbers on
     its own. There are well over a thousand of them. They are deliberately not
-    tabled (PRD-PLATES § 2.4 mechanism-note rule) and a decode must not guess.
+    tabled (the plates spec § 2.4 mechanism-note rule) and a decode must not guess.
 
 # ---------------------------------------------------------------------------
 # CODES. `evidence: instrument` = the row is quoted verbatim in a gazette
@@ -146,7 +146,7 @@ codes:
   - { code: "WB", name: "West Bengal", type: state, evidence: corroboration-only, verify: true }
 
 # ---------------------------------------------------------------------------
-# WITHDRAWN CODES. Rows are never deleted (PRD-PLATES § 2.4 period discipline):
+# WITHDRAWN CODES. Rows are never deleted (the plates spec § 2.4 period discipline):
 # every one of these is still on live, valid plates, because an Indian
 # registration mark is not reissued when a State is renamed or merged.
 # ---------------------------------------------------------------------------

data/data/plates/_decode/it-provinces.yml

@@ -33,7 +33,7 @@
 # (c) legally not part of the plate. On a pre-1993 plate the sigla is in the mark
 # and denotes the province OF REGISTRATION (art. 320 D.P.R. 420/1959). Same two
 # letters, opposite semantics, and a parse API must not blend them. This is
-# Italy's version of the French SIV caution recorded in PRD-PLATES §2.4.
+# Italy's version of the French SIV caution recorded in the plates spec §2.4.
 jurisdiction: it
 topic: provinces
 authority:
@@ -57,7 +57,7 @@ period_note: >-
   later (Monza-Brianza MB, Fermo FM, Barletta-Andria-Trani BT in 2004-2009, Sud
   Sardegna SU in 2016) never appeared on a provincial-era plate at all.
 
-# PERIOD DISCIPLINE (PRD-PLATES §2.4). D.P.R. 20 luglio 2017, n. 140 (GU
+# PERIOD DISCIPLINE (the plates spec §2.4). D.P.R. 20 luglio 2017, n. 140 (GU
 # 22-09-2017 n. 222) withdrew four sigle and added one, and its art. 2 comma 2
 # states the rule every decode table of this kind needs, verbatim:
 #   "I veicoli dotati di targhe di immatricolazione con le sigle delle province,
@@ -82,7 +82,7 @@ withdrawn_rule_source: "https://www.normattiva.it/do/atto/caricaAKN?dataGU=19921
 #   * Three sigle carry a heraldic device the appendix describes in words and
 #     which a render pipeline must not silently drop: Aosta ("La O è sormontata
 #     dallo stemma"), Bolzano ("La Z è sormontata dallo stemma"), Trento ("La N è
-#     sormontata dallo stemma"). PRD-PLATES §3 placeholder rule applies to those
+#     sormontata dallo stemma"). The plates spec §3 placeholder rule applies to those
 #     three emblems.
 maximal_munch: false
 

data/data/plates/_decode/jp-office-marks.yml

@@ -18,7 +18,7 @@
 # 250 cm3), and § 63-4(2) to 検査対象外軽自動車 (125-250 cm3 motorcycles).
 # One decode table therefore serves every series in plates/jp.yml.
 #
-# THE SCRIPT LINE (PRD-PLATES §2.6). These marks are written in kanji and
+# THE SCRIPT LINE (the plates spec §2.6). These marks are written in kanji and
 # hiragana. They are NOT in the dataset's serial alphabet (A-Z 0-9), so they
 # can never appear in a `format.pattern` or a `format.regex`; plates/jp.yml
 # carries them as a `format.fields[]` entry with `script: kanji-hiragana` and
@@ -35,7 +35,7 @@
 # boundary change stays valid afterwards. A decoder must therefore treat a mark
 # as evidence of the base of use AT FIRST REGISTRATION (or last renumbering),
 # never as a current-location signal. This is the same caution plates/gb.yml
-# records for DVLA memory tags and PRD-PLATES §2.4 records for the French SIV.
+# records for DVLA memory tags and the plates spec §2.4 records for the French SIV.
 #
 # ROWS ARE NEVER DELETED. No mark has been withdrawn from 別表第一 in any
 # revision e-Gov publishes (2017-04-01 onward): the table has only grown,

data/data/plates/_decode/jp-use-marks.yml

@@ -37,12 +37,12 @@
 # instrument states a reason, and none is claimed here. What IS a sourced fact is
 # the omission itself, which is why every set below is written out in full.
 #
-# SCRIPT AND CODEPOINTS (PRD-PLATES §4, the identity-is-form rule). The hiragana
+# SCRIPT AND CODEPOINTS (the plates spec §4, the identity-is-form rule). The hiragana
 # are recorded in the instrument's own characters. The Roman letters are
 # recorded TWICE: `chars` carries the instrument's FULL-WIDTH forms exactly as
 # e-Gov publishes them (Ａ = U+FF21, not U+0041), and `ascii` carries the
 # half-width normalisation that plates/jp.yml's regexes are written in, because
-# PRD-PLATES §2.6 confines the dataset's serial alphabet to A-Z and 0-9. The
+# the plates spec §2.6 confines the dataset's serial alphabet to A-Z and 0-9. The
 # plate itself is embossed, not typeset; the width distinction is an artefact of
 # the statute's typography, and recording both is what keeps that honest.
 jurisdiction: jp

data/data/plates/_decode/kr-authority-marks.yml

@@ -71,7 +71,7 @@ maximal_munch_note: >-
   no mark is a prefix of another. Read exactly two syllables and stop.
 
 # ---------------------------------------------------------------------------
-# NAMING DRIFT, RECORDED (PRD-PLATES §9 "region-decode tables rot").
+# NAMING DRIFT, RECORDED (the plates spec §9 "region-decode tables rot").
 # Two of the seventeen jurisdictions have been renamed since the notice's list
 # was last set, and the notice still prints the OLD full names. The two-syllable
 # MARKS are unaffected — 강원 and 전북 are what the plate shows either way — so

data/data/plates/_decode/mx-entity-series.yml

@@ -376,5 +376,5 @@ verification:
     Apéndice C also carries per-entity allocations for autobuses, remolques,
     convertidores, motocicletas (three grammars), demostración, ecológicos,
     ambulancias, bomberos, protección civil and escoltas. They are sourced at
-    exactly the same quality and are omitted for scope (PRD-PLATES §2.5).
+    exactly the same quality and are omitted for scope (the plates spec §2.5).
     Extending this file needs no new sourcing, only transcription.

data/data/plates/_decode/my-index-marks.yml

@@ -33,7 +33,7 @@
 #          Called once per code, 2026-08-03 00:47:43 to 00:49:21 (+08).
 #          28 of 32 returned a reading.
 # The second endpoint is the closest thing Malaysia has to the registry
-# corpora PRD-PLATES §5.2 uses elsewhere: it is the authority stating, on the
+# corpora the plates spec §5.2 uses elsewhere: it is the authority stating, on the
 # record and to the minute, what it is issuing. Those readings are the evidence
 # behind BOTH standard series in plates/my.yml and are reproduced per row as
 # `observed_2026_08_03`.
@@ -52,7 +52,7 @@
 # each colliding pair — so Sibu, Sri Aman and Limbang, which are real Sarawak
 # divisions with real plates on the road, have no row in the live list at all.
 # Both captures are recorded below and neither is silently preferred
-# (PRD-PLATES §5.3: record disagreements, never average them). The practical
+# (the plates spec §5.3: record disagreements, never average them). The practical
 # reading is that the JPJ code is a QUERY KEY for a counter, not a bijection
 # with a place, and a decode consumer must treat the mapping as MANY-TO-ONE in
 # both directions.
@@ -133,7 +133,7 @@ areas:
     codes:
       - { code: SA, area: "Kota Kinabalu (West Coast)", observed_2026_08_03: "SJT 4199", form: prefix, evidence: agency-api, note: "queried as SA, issuing SJ-series — the drift is explicit here" }
       - { code: SB, area: Beaufort, observed_2026_08_03: "SB 1644 L", form: suffix, evidence: agency-api }
-      - { code: SD, area: "Lahat Datu", observed_2026_08_03: "SPA 3076", form: prefix, evidence: agency-api, note: "JPJ spells it 'Lahat Datu'; the town is conventionally Lahad Datu. The authority's spelling is kept, per PRD-PLATES §5.3." }
+      - { code: SD, area: "Lahat Datu", observed_2026_08_03: "SPA 3076", form: prefix, evidence: agency-api, note: "JPJ spells it 'Lahat Datu'; the town is conventionally Lahad Datu. The authority's spelling is kept, per the plates spec §5.3." }
       - { code: SK, area: Kudat, observed_2026_08_03: "SK 4097 H", form: suffix, evidence: agency-api }
       - { code: SM, area: Sandakan, observed_2026_08_03: null, form: unknown, evidence: agency-api, note: "one of TWO codes JPJ maps to Sandakan; see the disagreement note in the header" }
       - { code: SS, area: Sandakan, observed_2026_08_03: "SMW 5063", form: prefix, evidence: agency-api, note: "queried as SS, issuing SM-series" }
@@ -210,7 +210,7 @@ diplomatic_dc:
   gap: >-
     THE COUNTRY-CODE TABLE IS NOT ON THIS PAGE and was not located anywhere.
     The number group above is the RANK group; the sending state's code is a
-    separate printed group whose allocation is unsourced. PRD-PLATES puts
+    separate printed group whose allocation is unsourced. The plates spec puts
     diplomatic decode tables at L4 and this one is deliberately left there
     rather than guessed.
 

data/data/plates/_decode/pt-trailer-services.yml

@@ -23,7 +23,7 @@
 # gazette printed. From the 2006 instrument's entry into force the operative
 # code is VI; VS is on the road on every trailer plated between 2005 and 2006
 # and on every earlier one issued under the same code. A parse API must accept
-# BOTH — which is exactly the "decode tables rot" hazard PRD-PLATES §9 names,
+# BOTH — which is exactly the "decode tables rot" hazard the plates spec §9 names,
 # caught here inside a single fourteen-month window.
 #
 # WHAT THIS TABLE IS NOT: it is not a district-code table for cars. There has

data/data/plates/_decode/ro-counties.yml

@@ -32,7 +32,7 @@
 #    the DIGIT BLOCK ALWAYS FOLLOWS THE CODE. In "B12ABC" the character after B
 #    is a digit, so the code is one letter; in "BC12ABC" it is a letter, so the
 #    code is two. The rule is mechanical and total, and it survives separator
-#    stripping (PRD-PLATES §2.6), which is exactly when a decoder needs it most.
+#    stripping (the plates spec §2.6), which is exactly when a decoder needs it most.
 #
 # 3. THESE ARE NOT THE PRE-1992 CODES. The communist-era system used a different
 #    territorial letter set built on the administrative regions of the day — the

data/data/plates/_decode/se-diplomatic.yml

@@ -1,6 +1,6 @@
 # plates/_decode/se-diplomatic.yml — Swedish beskickningsskylt country codes.
 #
-# WHY THIS IS A _decode FILE AND NOT AN INLINE TABLE (PRD-PLATES §2.4): 128
+# WHY THIS IS A _decode FILE AND NOT AN INLINE TABLE (the plates spec §2.4): 128
 # rows is past the "small" threshold that put FZV Anlage 2 inline in de.yml,
 # and it is a CLOSED, PRIMARY list — Transportstyrelsen publishes the whole
 # thing, so it qualifies on the other limb of the rule ("a reference … if the

data/data/plates/_decode/si-regions.yml

@@ -36,7 +36,7 @@
 # NOT A COMPLETE ARMS CATALOGUE. This file names the upravne enote whose seat
 # municipality supplies each shield; it does NOT carry the 58 blazons, their
 # artwork or their licensing. Municipal arms are protected independently of
-# copyright and PRD-PLATES §3 puts them in the placeholder slot
+# copyright and the plates spec §3 puts them in the placeholder slot
 # (`design.emblem.rendered: placeholder` in plates/si.yml). Recorded gap:
 # a per-upravna-enota arms ledger.
 jurisdiction: si

data/data/plates/_decode/ua-regions.yml

@@ -27,7 +27,7 @@
 # The two never appear on the same plate.
 #
 # ---------------------------------------------------------------------------
-# THE SCRIPT PROBLEM — recorded, never folded away (PRD-PLATES §2.6).
+# THE SCRIPT PROBLEM — recorded, never folded away (the plates spec §2.6).
 # ---------------------------------------------------------------------------
 script:
   printed: Cyrillic (Ukrainian)

data/data/plates/_meta/classes.yml

@@ -1,4 +1,4 @@
-# The controlled vocabulary of plate classes (PRD-PLATES §2.3). A class is a
+# The controlled vocabulary of plate classes (the plates spec §2.3). A class is a
 # LEGAL/ADMINISTRATIVE issuance category, not a design variant. A
 # jurisdiction's variant becomes its OWN SERIES iff format OR design OR
 # period differs from the standard series; a color-only variant of the same

data/data/plates/_meta/separators.yml

@@ -1,6 +1,6 @@
 # plates/_meta/separators.yml — THE NORMATIVE SEPARATOR CHARACTER SET.
 #
-# PRD-PLATES §2.6 declares this contract in prose; this file is its
+# the plates spec §2.6 declares this contract in prose; this file is its
 # machine-readable form, and `scripts/lint_plates.rb` enforces the dataset
 # against THIS file (single source of truth — the lint hardcodes nothing).
 #