feat: implement pi-base cross-reference attribute

Author: Deicyde

Mathlib/Tactic/CrossRefAttribute.lean

@@ -16,6 +16,7 @@ to entries in external mathematical databases:
 
 * `@[stacks TAG]` — [Stacks Project](https://stacks.math.columbia.edu/tags)
 * `@[kerodon TAG]` — [Kerodon](https://kerodon.net/tag/)
+* `@[pibase ID]` — [π-Base](https://topology.pi-base.org/)
 * `@[wikidata QID]` — [Wikidata](https://www.wikidata.org)
 
 Each attribute records the cross-reference in an environment extension and appends
@@ -35,6 +36,7 @@ namespace Mathlib.CrossRef
 /-- The supported external databases -/
 inductive Database where
   | kerodon
+  | pibase
   | stacks
   | wikidata
   deriving BEq, Hashable, Ord
@@ -44,18 +46,37 @@ namespace Database
 /-- The base URL for an external database's tag pages. Always ends with `/`. -/
 def url : Database → String
   | .kerodon => "https://kerodon.net/tag/"
+  | .pibase => "https://topology.pi-base.org/"
   | .stacks => "https://stacks.math.columbia.edu/tag/"
   | .wikidata => "https://www.wikidata.org/wiki/"
 
+/-- The URL path for the given database identifier. -/
+def urlPath (db : Database) (id : String) : String :=
+  match db with
+  | .pibase =>
+      let path := match id.toList with
+        | 'P' :: _ => "properties/"
+        | 'S' :: _ => "spaces/"
+        | 'T' :: _ => "theorems/"
+        | _ => ""
+      path ++ id
+  | _ => id
+
+/-- The URL for an external database entry. -/
+def entryUrl (db : Database) (id : String) : String :=
+  db.url ++ db.urlPath id
+
 /-- The display label used in docstring links and trace output. -/
 def label : Database → String
   | .kerodon => "Kerodon Tag"
+  | .pibase => "π-Base"
   | .stacks => "Stacks Tag"
   | .wikidata => "Wikidata"
 
 /-- A lowercase short name for the given database. Useful when exporting to JSON. -/
 def shortName : Database → String
   | .kerodon  => "kerodon"
+  | .pibase   => "pibase"
   | .stacks   => "stacks"
   | .wikidata => "wikidata"
 
@@ -92,7 +113,7 @@ This is the database-agnostic core of every cross-reference attribute's `add` ha
 def addCrossRefDoc (db : Database) (decl : Name) (idStr comment : String) : CoreM Unit := do
   let oldDoc := (← findDocString? (← getEnv) decl).getD ""
   let commentInDoc := if comment.isEmpty then "" else s!" ({comment})"
-  let link := s!"[{db.label} {idStr}]({db.url}{idStr}){commentInDoc}"
+  let link := s!"[{db.label} {idStr}]({db.entryUrl idStr}){commentInDoc}"
   addDocStringCore decl <| "\n\n".intercalate ([oldDoc, link].filter (· != ""))
   addTagEntry decl db idStr comment
 
@@ -169,6 +190,49 @@ def wikidataIdNoAntiquot : Parser := {
 def wikidataIdParser : Parser :=
   withAntiquot (mkAntiquot "wikidataId" wikidataIdKind) wikidataIdNoAntiquot
 
+/-! ### π-Base parser -/
+
+/-- `pibaseId` is the node kind of π-Base identifiers: one of the letters `P`, `S`, or `T`,
+followed by exactly six digits. -/
+abbrev pibaseIdKind : SyntaxNodeKind := `pibaseId
+
+/-- The main parser for π-Base identifiers: it accepts canonical property, space, and theorem
+identifiers, such as `P000001`, `S000023`, and `T000010`. -/
+def pibaseIdFn : ParserFn := fun c s =>
+  let i := s.pos
+  let s := takeWhileFn (fun c => c.isAlphanum) c s
+  if s.hasError then
+    s
+  else if s.pos == i then
+    ParserState.mkError s "π-Base id"
+  else
+    let id := c.extract i s.pos
+    match id.toList with
+    | kind :: rest =>
+        if kind != 'P' && kind != 'S' && kind != 'T' then
+          ParserState.mkUnexpectedError s
+            "π-Base ids must start with P, S, or T."
+        else if rest.length != 6 then
+          ParserState.mkUnexpectedError s "π-Base ids must have exactly six digits after P/S/T."
+        else if !rest.all Char.isDigit then
+          ParserState.mkUnexpectedError s
+            "π-Base ids must consist of P, S, or T followed by six digits."
+        else
+          mkNodeToken pibaseIdKind i true c s
+    | _ =>
+        ParserState.mkUnexpectedError s
+          "π-Base ids must consist of P, S, or T followed by six digits."
+
+@[inherit_doc pibaseIdFn]
+def pibaseIdNoAntiquot : Parser := {
+  fn   := pibaseIdFn
+  info := mkAtomicInfo "pibaseId"
+}
+
+@[inherit_doc pibaseIdFn]
+def pibaseIdParser : Parser :=
+  withAntiquot (mkAntiquot "pibaseId" pibaseIdKind) pibaseIdNoAntiquot
+
 end Mathlib.CrossRef
 
 open Mathlib.CrossRef
@@ -183,6 +247,11 @@ def Lean.TSyntax.getWikidataId (stx : TSyntax wikidataIdKind) : CoreM String :=
   let some val := Syntax.isLit? wikidataIdKind stx | throwError "Malformed Wikidata id"
   return val
 
+/-- Extract the underlying identifier as a string from a `pibaseId` node. -/
+def Lean.TSyntax.getPibaseId (stx : TSyntax pibaseIdKind) : CoreM String := do
+  let some val := Syntax.isLit? pibaseIdKind stx | throwError "Malformed π-Base id"
+  return val
+
 namespace Lean.PrettyPrinter
 
 namespace Formatter
@@ -195,6 +264,10 @@ namespace Formatter
 @[combinator_formatter wikidataIdNoAntiquot] def wikidataIdNoAntiquot.formatter :=
   visitAtom wikidataIdKind
 
+/-- The formatter for π-Base identifier syntax. -/
+@[combinator_formatter pibaseIdNoAntiquot] def pibaseIdNoAntiquot.formatter :=
+  visitAtom pibaseIdKind
+
 end Formatter
 
 namespace Parenthesizer
@@ -205,6 +278,9 @@ namespace Parenthesizer
 /-- The parenthesizer for Wikidata identifier syntax. -/
 @[combinator_parenthesizer wikidataIdNoAntiquot] def wikidataIdAntiquot.parenthesizer := visitToken
 
+/-- The parenthesizer for π-Base identifier syntax. -/
+@[combinator_parenthesizer pibaseIdNoAntiquot] def pibaseIdAntiquot.parenthesizer := visitToken
+
 end Lean.PrettyPrinter.Parenthesizer
 
 namespace Mathlib.CrossRef
@@ -265,6 +341,28 @@ initialize Lean.registerBuiltinAttribute {
   applicationTime := .beforeElaboration
 }
 
+/-! ### π-Base attribute -/
+
+/-- The `pibase` attribute.
+Use it as `@[pibase P000001 "Optional comment"]` to associate a Mathlib declaration with
+the corresponding [π-Base](https://topology.pi-base.org/) property, space, or theorem.
+
+The identifier must start with `P`, `S`, or `T`, followed by exactly six digits.
+-/
+syntax (name := pibaseTag) "pibase" pibaseIdParser (ppSpace str)? : attr
+
+initialize Lean.registerBuiltinAttribute {
+  name := `pibaseTag
+  descr := "Apply a π-Base identifier to a declaration."
+  add := fun decl stx _attrKind => do
+    let (id, comment) ← match stx with
+      | `(attr| pibase $id $[$comment]?) => pure (id, comment)
+      | _ => throwUnsupportedSyntax
+    addCrossRefDoc .pibase decl (← id.getPibaseId) ((comment.map (·.getString)).getD "")
+  -- docstrings are immutable once an asynchronous elaboration task has been started
+  applicationTime := .beforeElaboration
+}
+
 end Mathlib.CrossRef
 
 /-- Returns the array of `Tag`s in the environment, sorted alphabetically by tag. -/
@@ -291,7 +389,7 @@ def traceCrossRefs (db : Database) (verbose : Bool := false) :
     let (parL, parR) := if d.comment.isEmpty then ("", "") else (" (", ")")
     let cmt := parL ++ d.comment ++ parR
     msgs := msgs.push
-      m!"[{db.label} {d.tag}]({db.url ++ d.tag}) \
+      m!"[{db.label} {d.tag}]({db.entryUrl d.tag}) \
         corresponds to declaration '{.ofConstName d.declName}'.{cmt}"
     if verbose then
       let dType := ((env.find? d.declName).getD default).type
@@ -336,4 +434,16 @@ or declaration type (for definitions, structures, instances, etc.) after each su
 elab (name := wikidataTags) "#wikidata_tags" tk:("!")? : command =>
   traceCrossRefs .wikidata (tk.isSome)
 
+/-- The `#pibase_tags` command retrieves all declarations that have the `pibase` attribute.
+
+For each found declaration, it prints a line
+```
+'declaration_name' corresponds to tag 'declaration_tag'.
+```
+The variant `#pibase_tags!` also adds the theorem statement (for theorems)
+or declaration type (for definitions, structures, instances, etc.) after each summary line.
+-/
+elab (name := pibaseTags) "#pibase_tags" tk:("!")? : command =>
+  traceCrossRefs .pibase (tk.isSome)
+
 end Mathlib.CrossRef

MathlibTest/CrossRefAttribute.lean

@@ -97,6 +97,47 @@ info:
 /-- info: Q42 -/
 #guard_msgs in #parse Mathlib.CrossRef.wikidataIdFn => "Q42"
 
+namespace Pibase
+
+@[pibase P000001]
+theorem p : True := .intro
+
+@[pibase S000023]
+theorem s : True := .intro
+
+@[pibase T000001]
+theorem t : True := .intro
+
+/--
+info: some ([π-Base P000001](https://topology.pi-base.org/properties/P000001))
+-/
+#guard_msgs in
+run_cmd
+  Lean.logInfo m!"{← Lean.findDocString? (← Lean.getEnv) `Pibase.p}"
+
+/--
+info:
+[π-Base P000001](https://topology.pi-base.org/properties/P000001) corresponds to declaration 'p'.
+[π-Base S000023](https://topology.pi-base.org/spaces/S000023) corresponds to declaration 's'.
+[π-Base T000001](https://topology.pi-base.org/theorems/T000001) corresponds to declaration 't'.
+-/
+#guard_msgs in
+#pibase_tags
+
+/-- error: <input>:1:3: π-Base ids must have exactly six digits after P/S/T. -/
+#guard_msgs in #parse Mathlib.CrossRef.pibaseIdFn => "P42"
+
+/-- error: <input>:1:7: π-Base ids must consist of P, S, or T followed by six digits. -/
+#guard_msgs in #parse Mathlib.CrossRef.pibaseIdFn => "P00001X"
+
+/-- error: <input>:1:7: π-Base ids must start with P, S, or T. -/
+#guard_msgs in #parse Mathlib.CrossRef.pibaseIdFn => "Q000001"
+
+/-- info: S000023 -/
+#guard_msgs in #parse Mathlib.CrossRef.pibaseIdFn => "S000023"
+
+end Pibase
+
 section errors
 
 open Lean Parser Mathlib.CrossRef