Initial implementation of nimsuggest v3 (#19826)

* Initial implementation of nimsuggest v3

Rework `nimsuggest` to use caching to make usage of ide commands more efficient.
Previously, all commands no matter what the state of the process is were causing
clean build. In the context of Language Server Protocol(LSP) and lsp clients
this was causing perf issues and overall instability. Overall, the goal of v3 is
to fit to LSP Server needs

- added two new commands:
  - `recompile` to do clean compilation
  - `changed` which can be used by the IDEs to notify that a particular file has been changed.
The later can be utilized when using LSP file watches.
  - `globalSymbols` - searching global references

- added `segfaults` dependency to allow fallback to clean build when incremental
fails. I wish the error to be propagated to the client so we can work on fixing
the incremental build failures (typically hitting pointer)

- more efficient rebuild flow. ATM incremental rebuild is triggered when the
command needs that(i. e. it is global) while the commands that work on the
current source rebuild only it

Things missing in this PR:

- Documentation
- Extensive unit testing.

Although functional I still see this more as a POC that this approach can work.

Next steps:
- Implement `sug` request.
- Rework/extend the protocol to allow better client/server communication.
Ideally we will need push events, diagnostics should be restructored to allow
per file notifications, etc.
- implement v3 test suite.
- better logging

* Add tests for v3 and implement ideSug

* Remove typeInstCache/procInstCache cleanup

* Add ideChkFile command

* Avoid contains call when adding symbol info

* Remove log

* Remove segfaults

4 files changed, 91 insertions, 21 deletions

diff --git a/compiler/modulegraphs.nim b/compiler/modulegraphs.nim
index 147381910..8294d863e 100644
--- a/compiler/modulegraphs.nim
+++ b/compiler/modulegraphs.nim
@@ -11,7 +11,7 @@
 ## represents a complete Nim project. Single modules can either be kept in RAM
 ## or stored in a rod-file.
 
-import intsets, tables, hashes, md5_old
+import intsets, tables, hashes, md5_old, sequtils
 import ast, astalgo, options, lineinfos,idents, btrees, ropes, msgs, pathutils, packages
 import ic / [packed_ast, ic]
 
@@ -83,6 +83,8 @@ type
     doStopCompile*: proc(): bool {.closure.}
     usageSym*: PSym # for nimsuggest
     owners*: seq[PSym]
+    suggestSymbols*: Table[FileIndex, seq[tuple[sym: PSym, info: TLineInfo]]]
+    suggestErrors*: Table[FileIndex, seq[Suggest]]
     methods*: seq[tuple[methods: seq[PSym], dispatcher: PSym]] # needs serialization!
     systemModule*: PSym
     sysTypes*: array[TTypeKind, PType]
@@ -385,9 +387,19 @@ when defined(nimfind):
       c.graph.onDefinitionResolveForward(c.graph, s, info)
 
 else:
-  template onUse*(info: TLineInfo; s: PSym) = discard
-  template onDef*(info: TLineInfo; s: PSym) = discard
-  template onDefResolveForward*(info: TLineInfo; s: PSym) = discard
+  when defined(nimsuggest):
+    template onUse*(info: TLineInfo; s: PSym) = discard
+
+    template onDef*(info: TLineInfo; s: PSym) =
+      let c = getPContext()
+      if c.graph.config.suggestVersion == 3:
+        suggestSym(c.graph, info, s, c.graph.usageSym)
+
+    template onDefResolveForward*(info: TLineInfo; s: PSym) = discard
+  else:
+    template onUse*(info: TLineInfo; s: PSym) = discard
+    template onDef*(info: TLineInfo; s: PSym) = discard
+    template onDefResolveForward*(info: TLineInfo; s: PSym) = discard
 
 proc stopCompile*(g: ModuleGraph): bool {.inline.} =
   result = g.doStopCompile != nil and g.doStopCompile()
@@ -434,8 +446,7 @@ proc initOperators*(g: ModuleGraph): Operators =
   result.opNot = createMagic(g, "not", mNot)
   result.opContains = createMagic(g, "contains", mInSet)
 
-proc newModuleGraph*(cache: IdentCache; config: ConfigRef): ModuleGraph =
-  result = ModuleGraph()
+proc initModuleGraphFields(result: ModuleGraph) =
   # A module ID of -1 means that the symbol is not attached to a module at all,
   # but to the module graph:
   result.idgen = IdGenerator(module: -1'i32, symId: 0'i32, typeId: 0'i32)
@@ -445,9 +456,9 @@ proc newModuleGraph*(cache: IdentCache; config: ConfigRef): ModuleGraph =
   result.ifaces = @[]
   result.importStack = @[]
   result.inclToMod = initTable[FileIndex, FileIndex]()
-  result.config = config
-  result.cache = cache
   result.owners = @[]
+  result.suggestSymbols = initTable[FileIndex, seq[tuple[sym: PSym, info: TLineInfo]]]()
+  result.suggestErrors = initTable[FileIndex, seq[Suggest]]()
   result.methods = @[]
   initStrTable(result.compilerprocs)
   initStrTable(result.exposed)
@@ -461,6 +472,12 @@ proc newModuleGraph*(cache: IdentCache; config: ConfigRef): ModuleGraph =
   result.operators = initOperators(result)
   result.emittedTypeInfo = initTable[string, FileIndex]()
 
+proc newModuleGraph*(cache: IdentCache; config: ConfigRef): ModuleGraph =
+  result = ModuleGraph()
+  result.config = config
+  result.cache = cache
+  initModuleGraphFields(result)
+
 proc resetAllModules*(g: ModuleGraph) =
   initStrTable(g.packageSyms)
   g.deps = initIntSet()
@@ -472,6 +489,7 @@ proc resetAllModules*(g: ModuleGraph) =
   g.methods = @[]
   initStrTable(g.compilerprocs)
   initStrTable(g.exposed)
+  initModuleGraphFields(g)
 
 proc getModule*(g: ModuleGraph; fileIdx: FileIndex): PSym =
   if fileIdx.int32 >= 0:
@@ -550,7 +568,19 @@ proc transitiveClosure(g: var IntSet; n: int) =
 
 proc markDirty*(g: ModuleGraph; fileIdx: FileIndex) =
   let m = g.getModule fileIdx
-  if m != nil: incl m.flags, sfDirty
+  if m != nil:
+    g.suggestSymbols.del(fileIdx)
+    g.suggestErrors.del(fileIdx)
+    incl m.flags, sfDirty
+
+proc unmarkAllDirty*(g: ModuleGraph) =
+  for i in 0i32..<g.ifaces.len.int32:
+    let m = g.ifaces[i].module
+    if m != nil:
+      m.flags.excl sfDirty
+
+proc isDirty*(g: ModuleGraph; m: PSym): bool =
+  result = g.suggestMode and sfDirty in m.flags
 
 proc markClientsDirty*(g: ModuleGraph; fileIdx: FileIndex) =
   # we need to mark its dependent modules D as dirty right away because after
@@ -562,12 +592,26 @@ proc markClientsDirty*(g: ModuleGraph; fileIdx: FileIndex) =
 
   # every module that *depends* on this file is also dirty:
   for i in 0i32..<g.ifaces.len.int32:
+    if g.deps.contains(i.dependsOn(fileIdx.int)):
+      g.markDirty(FileIndex(i))
+
+proc needsCompilation*(g: ModuleGraph): bool =
+  # every module that *depends* on this file is also dirty:
+  for i in 0i32..<g.ifaces.len.int32:
     let m = g.ifaces[i].module
-    if m != nil and g.deps.contains(i.dependsOn(fileIdx.int)):
-      incl m.flags, sfDirty
+    if m != nil:
+      if sfDirty in m.flags:
+        return true
 
-proc isDirty*(g: ModuleGraph; m: PSym): bool =
-  result = g.suggestMode and sfDirty in m.flags
+proc needsCompilation*(g: ModuleGraph, fileIdx: FileIndex): bool =
+  let module = g.getModule(fileIdx)
+  if module != nil and g.isDirty(module):
+    return true
+
+  for i in 0i32..<g.ifaces.len.int32:
+    let m = g.ifaces[i].module
+    if m != nil and g.isDirty(m) and g.deps.contains(fileIdx.int32.dependsOn(i)):
+      return true
 
 proc getBody*(g: ModuleGraph; s: PSym): PNode {.inline.} =
   result = s.ast[bodyPos]
@@ -611,3 +655,13 @@ proc getPackage*(graph: ModuleGraph; fileIdx: FileIndex): PSym =
 func belongsToStdlib*(graph: ModuleGraph, sym: PSym): bool =
   ## Check if symbol belongs to the 'stdlib' package.
   sym.getPackageSymbol.getPackageId == graph.systemModule.getPackageId
+
+iterator suggestSymbolsIter*(g: ModuleGraph): tuple[sym: PSym, info: TLineInfo] =
+  for xs in g.suggestSymbols.values:
+    for x in xs.deduplicate:
+      yield x
+
+iterator suggestErrorsIter*(g: ModuleGraph): Suggest =
+  for xs in g.suggestErrors.values:
+    for x in xs:
+      yield x
diff --git a/compiler/options.nim b/compiler/options.nim
index 89fb66d5f..792f15d58 100644
--- a/compiler/options.nim
+++ b/compiler/options.nim
@@ -188,8 +188,9 @@ type
     # as far as usesWriteBarrier() is concerned
 
   IdeCmd* = enum
-    ideNone, ideSug, ideCon, ideDef, ideUse, ideDus, ideChk, ideMod,
-    ideHighlight, ideOutline, ideKnown, ideMsg, ideProject
+    ideNone, ideSug, ideCon, ideDef, ideUse, ideDus, ideChk, ideChkFile, ideMod,
+    ideHighlight, ideOutline, ideKnown, ideMsg, ideProject, ideGlobalSymbols,
+    ideRecompile, ideChanged
 
   Feature* = enum  ## experimental features; DO NOT RENAME THESE!
     implicitDeref,
@@ -993,12 +994,16 @@ proc parseIdeCmd*(s: string): IdeCmd =
   of "use": ideUse
   of "dus": ideDus
   of "chk": ideChk
+  of "chkFile": ideChkFile
   of "mod": ideMod
   of "highlight": ideHighlight
   of "outline": ideOutline
   of "known": ideKnown
   of "msg": ideMsg
   of "project": ideProject
+  of "globalSymbols": ideGlobalSymbols
+  of "recompile": ideRecompile
+  of "changed": ideChanged
   else: ideNone
 
 proc `$`*(c: IdeCmd): string =
@@ -1009,6 +1014,7 @@ proc `$`*(c: IdeCmd): string =
   of ideUse: "use"
   of ideDus: "dus"
   of ideChk: "chk"
+  of ideChkFile: "chkFile"
   of ideMod: "mod"
   of ideNone: "none"
   of ideHighlight: "highlight"
@@ -1016,6 +1022,9 @@ proc `$`*(c: IdeCmd): string =
   of ideKnown: "known"
   of ideMsg: "msg"
   of ideProject: "project"
+  of ideGlobalSymbols: "globalSymbols"
+  of ideRecompile: "recompile"
+  of ideChanged: "changed"
 
 proc floatInt64Align*(conf: ConfigRef): int16 =
   ## Returns either 4 or 8 depending on reasons.
diff --git a/compiler/passes.nim b/compiler/passes.nim
index 3de27575b..46c36f9d1 100644
--- a/compiler/passes.nim
+++ b/compiler/passes.nim
@@ -14,7 +14,7 @@ import
   options, ast, llstream, msgs,
   idents,
   syntaxes, modulegraphs, reorder,
-  lineinfos, pathutils, packages
+  lineinfos, pathutils, std/sha1, packages
 
 when defined(nimPreviewSlimSystem):
   import std/syncio
@@ -132,6 +132,11 @@ proc processModule*(graph: ModuleGraph; module: PSym; idgen: IdGenerator;
       return false
   else:
     s = stream
+
+  when defined(nimsuggest):
+    let filename = toFullPathConsiderDirty(graph.config, fileIdx).string
+    msgs.setHash(graph.config, fileIdx, $sha1.secureHashFile(filename))
+
   while true:
     openParser(p, fileIdx, s, graph.cache, graph.config)
 
diff --git a/compiler/suggest.nim b/compiler/suggest.nim
index 84c94d793..38751fcc7 100644
--- a/compiler/suggest.nim
+++ b/compiler/suggest.nim
@@ -117,7 +117,7 @@ proc getTokenLenFromSource(conf: ConfigRef; ident: string; info: TLineInfo): int
     elif sourceIdent != ident:
       result = 0
 
-proc symToSuggest(g: ModuleGraph; s: PSym, isLocal: bool, section: IdeCmd, info: TLineInfo;
+proc symToSuggest*(g: ModuleGraph; s: PSym, isLocal: bool, section: IdeCmd, info: TLineInfo;
                   quality: range[0..100]; prefix: PrefixMatch;
                   inTypeContext: bool; scope: int;
                   useSuppliedInfo = false): Suggest =
@@ -203,14 +203,14 @@ proc `$`*(suggest: Suggest): string =
     result.add(sep)
     when defined(nimsuggest) and not defined(noDocgen) and not defined(leanCompiler):
       result.add(suggest.doc.escape)
-    if suggest.version == 0:
+    if suggest.version in {0, 3}:
       result.add(sep)
       result.add($suggest.quality)
       if suggest.section == ideSug:
         result.add(sep)
         result.add($suggest.prefix)
 
-proc suggestResult(conf: ConfigRef; s: Suggest) =
+proc suggestResult*(conf: ConfigRef; s: Suggest) =
   if not isNil(conf.suggestionResultHook):
     conf.suggestionResultHook(s)
   else:
@@ -424,7 +424,7 @@ proc suggestFieldAccess(c: PContext, n, field: PNode, outputs: var Suggestions)
         t = skipTypes(t[0], skipPtrs)
     elif typ.kind == tyTuple and typ.n != nil:
       suggestSymList(c, typ.n, field, n.info, outputs)
-    
+
     suggestOperations(c, n, field, orig, outputs)
     if typ != orig:
       suggestOperations(c, n, field, typ, outputs)
@@ -482,7 +482,7 @@ proc findDefinition(g: ModuleGraph; info: TLineInfo; s: PSym; usageSym: var PSym
   if s.isNil: return
   if isTracked(info, g.config.m.trackPos, s.name.s.len) or (s == usageSym and sfForward notin s.flags):
     suggestResult(g.config, symToSuggest(g, s, isLocal=false, ideDef, info, 100, PrefixMatch.None, false, 0, useSuppliedInfo = s == usageSym))
-    if sfForward notin s.flags:
+    if sfForward notin s.flags and g.config.suggestVersion != 3:
       suggestQuit()
     else:
       usageSym = s
@@ -497,6 +497,8 @@ proc suggestSym*(g: ModuleGraph; info: TLineInfo; s: PSym; usageSym: var PSym; i
   ## misnamed: should be 'symDeclared'
   let conf = g.config
   when defined(nimsuggest):
+    g.suggestSymbols.mgetOrPut(info.fileIndex, @[]).add (s, info)
+
     if conf.suggestVersion == 0:
       if s.allUsages.len == 0:
         s.allUsages = @[info]