mirror of
https://github.com/penpot/penpot.git
synced 2026-08-08 22:08:39 +00:00
📚 State what the graph schema does, not what it mirrors
The graph namespaces explained themselves by citing a separate project whose Python pipeline reads the graphs this backend writes. A reader of this repository does not have that project and should not need it, and a docstring that justifies a choice by pointing elsewhere cannot be checked here. Every claim survives; only the framing changes. Column names and types are Penpot's own decision, recorded with the reason for each divergence from the snake_case default. The transform registry describes the edges it materializes. The denormalizations in `app.graph.project.document` are justified by the walk already holding both answers. Three corrections fall out of the rewrite: - `app.graph.schema.contract` claimed a test, `graph_contract_test`, that walks a checked-in schema manifest and fails on any divergence. No such test exists. The paragraph is gone. - `app.graph.project.document` pointed at `app.graph.meta/projection-transforms`, which does not exist. - `app.graph.project.transforms/registry` claimed its entries were "in application order" while `apply-transforms!` reduced over the literal vector. The three registered transforms read disjoint columns, so the order is not load-bearing. The docstring now says so, and the one real ordering constraint is stated where it applies: `link-swap-slots!` strips `swap-slot-*` entries from `touched`, so anything reading `touched` has to run before it. `contract/pending-beadpot-columns` becomes `contract/unprojected-keys`. It is referenced nowhere else. AI-assisted-by: mixed models
This commit is contained in:
Álvaro Tejero Cantero
parent
ead3c54e5b
commit
d8bb1ae39a
@ -315,8 +315,9 @@
|
||||
primitive, so the transfer goes through Ladybug's `EXPORT DATABASE` (Parquet
|
||||
per table) into a fresh on-disk database via `IMPORT DATABASE`.
|
||||
|
||||
Note the round-trip drops table comments; beadpot resolves such tables by
|
||||
name (`beadpot.graph.context/_adopt_node_table`), so this is not load-bearing.
|
||||
Note the round trip drops table comments. Nothing in the graph is addressed
|
||||
by a table comment: every table is resolved by name, so the loss costs
|
||||
nothing.
|
||||
|
||||
Returns the path of the written database, or nil when no session is loaded.
|
||||
The caller owns the file and must delete it once streamed."
|
||||
|
||||
@ -71,7 +71,7 @@
|
||||
(str "json('" (escape-cypher-string (json/encode v)) "')"))
|
||||
|
||||
(defn format-timestamp
|
||||
"Ladybug TIMESTAMP literal (beadpot: `timestamp('…')`)."
|
||||
"Ladybug TIMESTAMP literal of the form `timestamp('<ISO-8601 instant>')`."
|
||||
[v]
|
||||
(let [s (cond
|
||||
(instance? java.time.Instant v)
|
||||
|
||||
@ -10,16 +10,13 @@
|
||||
Projects Document, Page, Component, the full shape tree (skipping the root
|
||||
frame), and `IsChildOf` edges from shapes/pages/components to their parent.
|
||||
|
||||
Two of beadpot's denormalizing transforms happen here rather than in a later
|
||||
pass, because the walk already has the answer in hand:
|
||||
Two denormalizations happen here rather than in a later pass, because the
|
||||
walk already has both answers in hand and a post-ingest statement would have
|
||||
to rediscover them:
|
||||
|
||||
- `page-id` on every shape (beadpot `DenormalizePageId`), from the page the
|
||||
walk is currently in;
|
||||
- `component-id` propagated from an instance head down to its descendants
|
||||
(beadpot `DenormalizeComponentId`), from the head context the walk carries.
|
||||
|
||||
Both ids are declared in `app.graph.meta/projection-transforms`, so beadpot
|
||||
reads the graph's own provenance and does not redo them."
|
||||
- `page-id` on every shape, from the page the walk is currently in;
|
||||
- `component-id` propagated from an instance head down to its descendants,
|
||||
from the head context the walk carries."
|
||||
(:require
|
||||
[app.common.logging :as l]
|
||||
[app.common.uuid :as uuid]
|
||||
@ -73,9 +70,8 @@
|
||||
(defn descend-component-ctx
|
||||
"The component context to pass to `shape`'s children.
|
||||
|
||||
Mirrors beadpot `DenormalizeComponentId`, whose recursive match stops at the
|
||||
nearest ancestor Frame carrying a `component-id` and treats any intermediate
|
||||
node that carries one as a barrier:
|
||||
Inheritance stops at the nearest ancestor Frame carrying a `component-id`,
|
||||
and any intermediate shape that carries one is a barrier:
|
||||
|
||||
- a Frame with its own `component-id` becomes the new context (it is an
|
||||
instance head, and its descendants belong to *it*, not to an outer head);
|
||||
|
||||
@ -5,17 +5,12 @@
|
||||
;; Copyright (c) KALEIDOS INC Sucursal en España SL
|
||||
|
||||
(ns app.graph.project.transforms
|
||||
"Derived graph links, ported from beadpot's post-projection transforms.
|
||||
"Derived graph links: edges a reader could compute from the projected
|
||||
columns, materialized once at build time so a query does not have to.
|
||||
|
||||
Each entry in `registry` carries the shared transform id
|
||||
(`app.graph.meta`), so a build records exactly what it produced and beadpot
|
||||
runs only the complement in Python. Adding a transform here is therefore the
|
||||
whole port step: nothing else has to be told about it.
|
||||
|
||||
The Cypher mirrors beadpot's, which is written through its query builder but
|
||||
reduces to the same statements — the parity harness diffs the resulting
|
||||
graphs, so a semantic drift shows up as a differing edge set rather than as
|
||||
a differing query."
|
||||
Each entry in `registry` names the transform, the relationship it produces,
|
||||
and the function that produces it, so adding one is a single entry and
|
||||
nothing else has to be told about it."
|
||||
(:require
|
||||
[app.common.logging :as l]
|
||||
[app.graph.ladybug :as ladybug]
|
||||
@ -32,8 +27,7 @@
|
||||
(defn- link-component-instances!
|
||||
"`IsInstanceOf` from Frame instance heads to their Component.
|
||||
|
||||
beadpot `graph/transform/assets.py::LinkComponentInstances`. Every head is
|
||||
linked, the main instance and any copy root alike.
|
||||
Every head is linked, the main instance and any copy root alike.
|
||||
|
||||
`component-file` is what makes a head a head here, not `component-id` alone.
|
||||
`app.common.types.component/instance-of?` requires both, and the projection
|
||||
@ -54,18 +48,17 @@
|
||||
"One statement per (from, to) shape-table pair.
|
||||
|
||||
Ladybug cannot create a relationship bound by multiple node labels in a
|
||||
single `MERGE` — inherited from Kùzu, which it forks (upstream issue
|
||||
kuzudb/kuzu#5841). beadpot loops over label pairs for the same reason; the
|
||||
loop is a dialect constraint, not a modelling choice."
|
||||
single `MERGE`, a constraint inherited from Kùzu, which it forks (upstream
|
||||
issue kuzudb/kuzu#5841). The loop over label pairs is that dialect
|
||||
constraint, not a modelling choice."
|
||||
[f]
|
||||
(for [from nodes/shape-tables
|
||||
to nodes/shape-tables]
|
||||
(f from to)))
|
||||
|
||||
(defn- link-shape-refs!
|
||||
"`RefersTo` from an instance shape to its homologue in the main instance.
|
||||
|
||||
beadpot `graph/transform/assets.py::LinkShapeRefs`, driven by `shape-ref`."
|
||||
"`RefersTo` from an instance shape to its homologue in the main instance,
|
||||
driven by `shape-ref`."
|
||||
[^Connection conn]
|
||||
(reduce
|
||||
(fn [total statement] (+ total (run-scalar! conn statement)))
|
||||
@ -86,11 +79,14 @@
|
||||
(defn- link-swap-slots!
|
||||
"`FillsSwapSlot` from a swapped-in shape to the slot it replaces.
|
||||
|
||||
beadpot `graph/transform/swap_slots.py::LinkSwapSlots`. Penpot records a
|
||||
component sub-shape swap as a `swap-slot-<uuid>` entry in the *replacing*
|
||||
shape's `touched` set, where `<uuid>` is the replaced slot shape from the
|
||||
master. The entries are then stripped from `touched`, mirroring
|
||||
`app.common.types.component/normal-touched-groups`."
|
||||
Penpot records a component sub-shape swap as a `swap-slot-<uuid>` entry in
|
||||
the *replacing* shape's `touched` set, where `<uuid>` names the replaced
|
||||
slot shape in the main instance. The entries are then stripped from
|
||||
`touched`, as `app.common.types.component/normal-touched-groups` does, so a
|
||||
reader of `touched` sees design edits rather than swap bookkeeping.
|
||||
|
||||
Stripping makes this the one transform that writes a column another
|
||||
transform could read. Anything reading `touched` has to run before it."
|
||||
[^Connection conn]
|
||||
(let [linked
|
||||
(reduce
|
||||
@ -120,10 +116,12 @@
|
||||
linked))
|
||||
|
||||
(def registry
|
||||
"Every transform this backend applies, in application order.
|
||||
"Every transform this backend applies.
|
||||
|
||||
`:id` is the shared vocabulary with beadpot (`app.graph.meta`); `:rel` names
|
||||
what the transform produces, for the ingest report."
|
||||
`:id` names the transform in the ingest report and the log. `:rel` names
|
||||
the relationship it produces. The three registered here read disjoint
|
||||
columns, so the vector order is not load-bearing. The one ordering
|
||||
constraint that exists is stated on `link-swap-slots!`."
|
||||
[{:id "link-component-instances" :rel :IsInstanceOf :run link-component-instances!}
|
||||
{:id "link-shape-refs" :rel :RefersTo :run link-shape-refs!}
|
||||
{:id "link-swap-slots" :rel :FillsSwapSlot :run link-swap-slots!}])
|
||||
@ -131,9 +129,8 @@
|
||||
(defn apply-transforms!
|
||||
"Apply every registered transform to an already loaded graph.
|
||||
|
||||
Returns `{:ids [...] :counts {...} :transforms n}`; `:ids` is what the build
|
||||
records in `GraphMeta`, so beadpot subtracts exactly this set from its own
|
||||
pipeline."
|
||||
Returns `{:ids [...] :counts {...} :transforms n}`, where `:ids` names what
|
||||
ran and `:counts` gives the edges each one produced."
|
||||
[_system ^Connection conn _data _file]
|
||||
(reduce
|
||||
(fn [acc {:keys [id rel run]}]
|
||||
|
||||
@ -5,60 +5,53 @@
|
||||
;; Copyright (c) KALEIDOS INC Sucursal en España SL
|
||||
|
||||
(ns app.graph.schema.contract
|
||||
"The graph-schema contract shared with beadpot.
|
||||
"Deliberate choices in Penpot's graph schema, recorded as data.
|
||||
|
||||
Penpot is authoritative on the *design model*: what a shape is, what a
|
||||
component means, which attributes exist. beadpot is authoritative on the
|
||||
*graph schema*: the table and column names, and the Ladybug types, that
|
||||
downstream consumers (ML graph mappings, featurization) read. A graph this
|
||||
backend writes must therefore be indistinguishable, to those consumers, from
|
||||
one beadpot's Python pipeline writes.
|
||||
Penpot must pick a spelling and a type for every graph column. A Ladybug
|
||||
column gets both once, at table creation, and neither widens afterwards. The
|
||||
choices are therefore worth making deliberately and worth recording.
|
||||
|
||||
Everything that could drift between the two lives here, as data:
|
||||
Three of them live here:
|
||||
|
||||
- `column-name` maps a Penpot key to its beadpot column. The rule is
|
||||
snake_case of the key; `renames` records every exception.
|
||||
- `dropped-keys` names Penpot keys that deliberately have no column.
|
||||
- `column-name` maps a Penpot key to its column. The rule is snake_case of
|
||||
the key, and `renames` records every exception.
|
||||
- `dropped-keys` and `per-table-dropped` name Penpot keys that deliberately
|
||||
get no column.
|
||||
- `type-overrides` pins the Ladybug type where the Malli-derived one
|
||||
(`app.graph.schema.types`) differs from beadpot's and the difference is
|
||||
load-bearing for a consumer.
|
||||
(`app.graph.schema.types`) is coarser than the column deserves.
|
||||
|
||||
Each entry carries its reason. A new divergence must be added here, which is
|
||||
the point: `backend_tests.graph_contract_test` walks the checked-in beadpot
|
||||
manifest (`resources/app/graph/beadpot-schema.json`, produced by
|
||||
`bp graph schema export`) and fails on anything this namespace does not
|
||||
account for. Schema drift becomes a failing test with a precise message
|
||||
instead of a silently renamed column in a training set."
|
||||
Each entry carries its reason. A divergence from the default rule is then a
|
||||
diff to review rather than a silent rename."
|
||||
(:require
|
||||
[app.common.json :as json]
|
||||
[clojure.string :as str]))
|
||||
|
||||
(def ^:private renames
|
||||
"Penpot key → beadpot column name, where the two differ.
|
||||
"Penpot key to column name, where the column is not snake_case of the key.
|
||||
|
||||
Keyed by the Penpot key alone: no shape type gives one of these a second
|
||||
meaning, so a per-table map would only add ceremony."
|
||||
{;; beadpot names the discriminant after the table (`BooleanNode`), not
|
||||
;; after Penpot's `:bool` shape type.
|
||||
{;; `bool` collides with the Ladybug type name, so the column is named after
|
||||
;; the table (`Boolean`) rather than after Penpot's `:bool` shape type.
|
||||
:bool-type "boolean_type"
|
||||
|
||||
;; beadpot keeps the wire name `component-root` out of the graph because
|
||||
;; the column records what the *file* saved, which can lag what the shape
|
||||
;; tree implies — `saved_` marks it as the stored value, not a derivation.
|
||||
;; The column records what the file saved, which can lag what the shape
|
||||
;; tree implies. The `saved_` prefix marks it as the stored value rather
|
||||
;; than a derivation.
|
||||
:component-root "saved_component_root"
|
||||
|
||||
;; Penpot stores a list under a singular key; beadpot pluralizes it.
|
||||
;; The value is a list, so the plural is accurate.
|
||||
:shadow "shadows"
|
||||
|
||||
;; beadpot spells out the revision number.
|
||||
;; The column spells the revision number out.
|
||||
:revn "revision"})
|
||||
|
||||
(def dropped-keys
|
||||
"Penpot keys projected by the Malli registry that get no beadpot column.
|
||||
"Penpot keys projected by the Malli registry that get no column.
|
||||
|
||||
Dropping is the right call only when the column would be dead weight
|
||||
downstream; anything a consumer might learn from belongs in beadpot instead
|
||||
(see `pending-beadpot-columns`)."
|
||||
Dropping is right only when the column would be dead weight for every reader
|
||||
of the graph. A key a reader might learn from belongs in `unprojected-keys`
|
||||
instead."
|
||||
{:deleted-at
|
||||
"Only non-nil for a soft-deleted file, and a deleted file is never ingested."
|
||||
|
||||
@ -68,37 +61,35 @@
|
||||
:pixel-grid-opacity
|
||||
"Viewer chrome, as above."})
|
||||
|
||||
(def pending-beadpot-columns
|
||||
"Penpot keys that *should* become beadpot columns but do not exist there yet.
|
||||
(def unprojected-keys
|
||||
"Penpot keys that should become graph columns and do not have one yet.
|
||||
|
||||
Distinct from `dropped-keys` on purpose: these are a debt beadpot owes,
|
||||
not a decision to discard data. The contract test reports them separately so
|
||||
a new upstream attribute cannot be quietly buried in the drop list."
|
||||
Distinct from `dropped-keys` on purpose: these are a debt the projection
|
||||
owes, not a decision to discard data. Keeping the two apart means a new
|
||||
upstream attribute cannot be quietly buried in the drop list."
|
||||
{:background-blur
|
||||
"Landed upstream behind a default-on flag; beadpot has no field for it yet."})
|
||||
"Landed upstream behind a default-on flag. No column for it yet."})
|
||||
|
||||
(def ^:private per-table-dropped
|
||||
"Keys dropped only on certain tables.
|
||||
|
||||
`:grids` is the standing case: Penpot's shape schema admits it on every
|
||||
shape, but only a Frame ever carries one, and beadpot models it on Frame
|
||||
alone. Emitting an always-null column on ten other tables would widen every
|
||||
multi-table scan for nothing."
|
||||
shape, but only a Frame ever carries one. Emitting an always-null column on
|
||||
ten other tables would widen every multi-table scan for nothing."
|
||||
{:grids #{"Boolean" "Circle" "Group" "Image" "Path" "Rectangle" "SVGRaw" "Text"}})
|
||||
|
||||
(def type-overrides
|
||||
"Ladybug column type per beadpot column name, where beadpot's differs.
|
||||
"Ladybug column type per column name, where the derived type is too coarse.
|
||||
|
||||
`app.graph.schema.types` derives a type from the Malli schema, which is the
|
||||
right default but coarser than beadpot in places: a Malli `:map` becomes
|
||||
`JSON`, where beadpot may use a native Ladybug MAP or a fixed-size array
|
||||
that a consumer can read as a tensor without parsing.
|
||||
right default but coarser than the column deserves in places: a Malli `:map`
|
||||
becomes `JSON`, where a native Ladybug MAP or a fixed-size array lets a
|
||||
consumer read a tensor row without parsing.
|
||||
|
||||
Only load-bearing divergences are pinned here, in the order they became
|
||||
load-bearing; the rest are reported by the contract test and closed by
|
||||
moving the whole DDL onto the beadpot manifest."
|
||||
{;; `LinkAppliedTokens` (beadpot) reads this with `map_keys` /
|
||||
;; `map_extract`; as JSON the transform cannot run at all.
|
||||
load-bearing."
|
||||
{;; Must be a native MAP: a JSON blob cannot be indexed by key in Cypher, so
|
||||
;; `map_keys` and `map_extract` cannot reach a single token at all.
|
||||
"applied_tokens" "MAP(STRING, STRING)"
|
||||
|
||||
;; `grc/schema:rect` is an inline `:and` over a map, not the registered
|
||||
@ -109,38 +100,38 @@
|
||||
"selrect" "DOUBLE[4]"
|
||||
|
||||
;; The SVG provenance attributes are typed `:map` in the shape schema on
|
||||
;; purpose — legacy files hold them as plain maps rather than as `::grc/rect`
|
||||
;; and `::gmt/matrix` records, and a tighter *schema* would reject those files
|
||||
;; (`app.common.types.shape/schema:shape-generic-attrs`). A tighter *column*
|
||||
;; is free: `app.graph.schema.values/coerce` reads either form.
|
||||
;; purpose. Legacy files hold them as plain maps rather than as
|
||||
;; `::grc/rect` and `::gmt/matrix` records, and a tighter *schema* would
|
||||
;; reject those files
|
||||
;; (`app.common.types.shape/schema:shape-generic-attrs`). A tighter
|
||||
;; *column* is free: `app.graph.schema.values/coerce` reads either form.
|
||||
"svg_viewbox" "DOUBLE[4]"
|
||||
"svg_transform" "DOUBLE[6]"
|
||||
|
||||
;; `:fills` is an `:or` — the packed `app.common.types.fills` value or a
|
||||
;; plain vector of fill maps — so the schema alone cannot say it is a
|
||||
;; `:fills` is an `:or` over the packed `app.common.types.fills` value and
|
||||
;; a plain vector of fill maps, so the schema alone cannot say it is a
|
||||
;; collection. It always is one, and a fill has enough optional shape
|
||||
;; (solid, gradient, image) that JSON per element is the honest element type.
|
||||
;; (solid, gradient, image) that JSON per element is the honest element
|
||||
;; type.
|
||||
"fills" "JSON[]"})
|
||||
|
||||
(def ^:private map-key-fns
|
||||
"How to render the *keys* of a MAP column, per column.
|
||||
|
||||
Column names are snake_case because they are graph schema; the keys inside a
|
||||
MAP are not — they are values, and beadpot models them as whatever it parsed
|
||||
from the wire. `applied_tokens` is keyed by shape attribute in the camelCase
|
||||
form Penpot's own JSON encoder produces (`app.common.json/write-camel-key`),
|
||||
which is what beadpot's `AppliedTokenKey` holds and what
|
||||
`UsesToken.for_property` therefore carries: `strokeWidth`, not
|
||||
A column name is schema, so it is snake_case. The keys inside a MAP are
|
||||
values, so they keep the spelling their producer used. `applied_tokens` is
|
||||
keyed by shape attribute in the camelCase form
|
||||
`app.common.json/write-camel-key` produces: `strokeWidth`, not
|
||||
`stroke-width`."
|
||||
{"applied_tokens" json/write-camel-key})
|
||||
|
||||
(defn map-key-fn
|
||||
"Key renderer for a MAP column; `name` unless the column says otherwise."
|
||||
"Key renderer for a MAP column. `name` unless the column says otherwise."
|
||||
[column]
|
||||
(get map-key-fns column name))
|
||||
|
||||
(defn column-name
|
||||
"The beadpot column name for Penpot key `k`.
|
||||
"The graph column name for Penpot key `k`.
|
||||
|
||||
Default: snake_case of the key. `renames` overrides."
|
||||
[k]
|
||||
@ -154,6 +145,6 @@
|
||||
(contains? (get per-table-dropped k #{}) table)))
|
||||
|
||||
(defn ladybug-type
|
||||
"beadpot's Ladybug type for `column`, or `fallback` when it agrees."
|
||||
"The pinned Ladybug type for `column`, or `fallback` when nothing is pinned."
|
||||
[column fallback]
|
||||
(get type-overrides column fallback))
|
||||
|
||||
@ -11,10 +11,10 @@
|
||||
options (`:drop`, optional `:extra`). Derived artifacts — Ladybug
|
||||
DDL, Arrow fields, validation, type dispatch — all flow from that.
|
||||
|
||||
Column *names* and *types* are not ours to choose: they are the
|
||||
contract downstream consumers read, and beadpot owns it. Every
|
||||
divergence between a Penpot key and its column lives in
|
||||
`app.graph.schema.contract`."
|
||||
This registry is the single source of the graph schema. A Ladybug column
|
||||
gets its name and its type once, at table creation, and there is no
|
||||
widening afterwards. Every divergence between a Penpot key and its column
|
||||
is recorded in `app.graph.schema.contract`."
|
||||
(:require
|
||||
[app.common.exceptions :as ex]
|
||||
[app.common.schema :as sm]
|
||||
@ -31,7 +31,6 @@
|
||||
(def schema-version
|
||||
"penpot-graph-slice-4")
|
||||
|
||||
;; beadpot/graph/schemas.py drop_fields
|
||||
(def ^:private document-projection
|
||||
{:source ctf/schema:file
|
||||
:drop [:data]
|
||||
@ -154,7 +153,7 @@
|
||||
(nth entry 1)))
|
||||
|
||||
(defn column-name
|
||||
"beadpot column name for projected key `k` on `table`."
|
||||
"Graph column name for projected key `k` on `table`."
|
||||
[_table k]
|
||||
(contract/column-name k))
|
||||
|
||||
@ -170,8 +169,8 @@
|
||||
(defn column-keys
|
||||
"Projected column keys for `table`, in registry order.
|
||||
|
||||
Keys the beadpot contract drops on this table are omitted, so the column
|
||||
order, the Arrow batch and the DDL cannot disagree about what exists."
|
||||
Keys the contract drops on this table are omitted, so the column order, the
|
||||
Arrow batch, and the DDL cannot disagree about what exists."
|
||||
[table]
|
||||
(into []
|
||||
(comp (map first)
|
||||
@ -256,7 +255,7 @@
|
||||
table))
|
||||
|
||||
(defn cypher-property-key
|
||||
"Backtick-wrapped beadpot column name for inline Cypher literals."
|
||||
"Backtick-wrapped column name for inline Cypher literals."
|
||||
[table k]
|
||||
(str "`" (column-name table k) "`"))
|
||||
|
||||
@ -306,7 +305,7 @@
|
||||
", `position` INT64);"))
|
||||
|
||||
(defn is-instance-of-ddl
|
||||
"Frame instance heads → Component (beadpot `IsInstanceOf`)."
|
||||
"Frame instance heads → Component."
|
||||
[]
|
||||
"CREATE REL TABLE `IsInstanceOf` (FROM `Frame` TO `Component`);")
|
||||
|
||||
@ -315,8 +314,7 @@
|
||||
|
||||
Created up-front rather than on demand: the bulk loader must never race on
|
||||
lazy table creation, and a consumer can then tell \"this producer cannot
|
||||
emit that pair\" from \"this document happens to have none\" (beadpot
|
||||
`graph.manifest/REL_FAMILIES`)."
|
||||
emit that pair\" from \"this document happens to have none\"."
|
||||
[rel props]
|
||||
(str "CREATE REL TABLE `" rel "` ("
|
||||
(str/join ", " (for [from shape-tables
|
||||
@ -326,13 +324,13 @@
|
||||
");"))
|
||||
|
||||
(defn refers-to-ddl
|
||||
"Instance shape → its homologue in the component main instance
|
||||
(beadpot `RefersTo`, from `shape-ref`)."
|
||||
"Instance shape → its homologue in the component main instance, resolved
|
||||
from `shape-ref`."
|
||||
[]
|
||||
(shape-to-shape-rel-ddl "RefersTo" nil))
|
||||
|
||||
(defn fills-swap-slot-ddl
|
||||
"Swapped-in shape → the slot shape it replaces (beadpot `FillsSwapSlot`)."
|
||||
"Swapped-in shape → the slot shape it replaces."
|
||||
[]
|
||||
(shape-to-shape-rel-ddl "FillsSwapSlot" ["`slot_id` UUID"]))
|
||||
|
||||
|
||||
@ -7,8 +7,8 @@
|
||||
(ns app.graph.schema.projection
|
||||
"Derive Ladybug node column schemas from Penpot Malli sources.
|
||||
|
||||
Same model as beadpot's `drop_fields`: start from the canonical schema
|
||||
and remove keys that must not become graph columns."
|
||||
Start from the canonical schema and remove the keys that must not become
|
||||
graph columns."
|
||||
(:require
|
||||
[app.common.exceptions :as ex]
|
||||
[app.common.schema :as sm]
|
||||
@ -71,7 +71,7 @@
|
||||
"Build a graph node schema from canonical Malli `source`.
|
||||
|
||||
Options:
|
||||
- `:drop` - keys removed from the source (beadpot `drop_fields`)
|
||||
- `:drop` - keys removed from the source
|
||||
- `:extra` - optional extra `[:map ...]` merged on top"
|
||||
[source {:keys [drop extra]}]
|
||||
(let [projected (select-projected-keys source drop)]
|
||||
|
||||
@ -28,9 +28,9 @@
|
||||
more than one shape (a `:multi`, an `:or`, an optional-keyed map) becomes
|
||||
`JSON`, because a Ladybug column cannot be two types.
|
||||
|
||||
Every encoding here has a matching value formatter in `app.graph.ladybug`;
|
||||
the two must move together, and `bp graph schema diff` is what catches it
|
||||
when they do not."
|
||||
Every encoding here has a matching value formatter in `app.graph.ladybug`.
|
||||
The two must move together: a column type with no case there falls back to
|
||||
guessing the literal from the runtime value."
|
||||
(:require
|
||||
[app.common.logging :as l]
|
||||
[app.common.schema :as sm]
|
||||
|
||||
@ -120,7 +120,7 @@
|
||||
:else nil))
|
||||
|
||||
(defn- packed-color
|
||||
"`#RRGGBB` as `0xRRGGBBAA`, the form beadpot's `ColorValue` packs.
|
||||
"`#RRGGBB` as the packed integer `0xRRGGBBAA`.
|
||||
|
||||
Alpha defaults to opaque: the column holds a colour, and any opacity Penpot
|
||||
keeps alongside it is a separate attribute."
|
||||
|
||||
@ -118,9 +118,9 @@
|
||||
;; `:add-obj` children inherit — it is the shape's effective
|
||||
;; component-id, which loses the barrier case of a *non-Frame*
|
||||
;; carrying its own `component-id` (indistinguishable once
|
||||
;; denormalized). Cold projection, which beadpot diffs against,
|
||||
;; keeps the distinction; only a graph synced across such a
|
||||
;; shape can drift, and a Reload rebuilds it.
|
||||
;; denormalized). Cold projection keeps the distinction. Only a
|
||||
;; graph synced across such a shape can drift, and a Reload
|
||||
;; rebuilds it.
|
||||
:component-ctx (:component-id attrs)
|
||||
:page-id (or (:page-id attrs)
|
||||
(resolve-page-id shape-id parents pages))}]))
|
||||
@ -178,7 +178,7 @@
|
||||
"CREATE (s)-[:IsChildOf {position: " (ladybug/format-int position) "}]->(p);"))
|
||||
|
||||
(defn- create-instance-of-statement
|
||||
"Link a Frame instance head to its Component (beadpot `IsInstanceOf`).
|
||||
"Link a Frame instance head to its Component.
|
||||
|
||||
No-op when the Component is absent (e.g. library component not ingested)."
|
||||
[frame-id component-id]
|
||||
@ -247,9 +247,11 @@
|
||||
|
||||
|
||||
(defn- set-document-revision-statement
|
||||
"The column is `revision`, not `revn` — beadpot spells the revision number out
|
||||
(`schema.contract`). Named through `cypher-property-key` so the two cannot
|
||||
disagree again."
|
||||
"Set the Document's revision number.
|
||||
|
||||
`app.graph.schema.contract` names the column `revision`, not `revn`. The name
|
||||
is produced by `nodes/cypher-property-key`, so this statement and the DDL
|
||||
cannot disagree."
|
||||
[doc-id revn]
|
||||
(str "MATCH (d:Document {id: " (ladybug/format-uuid doc-id) "}) "
|
||||
"SET d." (nodes/cypher-property-key "Document" :revn) " = "
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user