Merge branch '44-wrong-emission-ids' into 'main'

Resolve "Wrong emission IDs"

Closes #44

See merge request !58

lib/SpeechRecognitionFSTs/SpeechRecognitionFSTs.jl

@@ -46,6 +46,9 @@ function tokenfst(
     states = Set(Int[])
     arcs, init, final = [], [], []
     
+    initdict = Dict(initweights...)
+    finaldict = Dict(finalweights...)
+
     if isnothing(mapping)
         st = Set(Int[]) 
         for _ in tokens
@@ -57,9 +60,9 @@ function tokenfst(
         mapping = EmissionMapping(length(st), length(tokens))   
     end
 
-    # Extra state (infinite loop)
-    init_state = length(mapping) + 1
-    push!(states, init_state)
+    # Extra state
+    extra_state = length(mapping) + 1
+    push!(states, extra_state)
     
     for (j, token) in enumerate(tokens)
         offset = length(states) - 1
@@ -76,55 +79,60 @@ function tokenfst(
                 weight = S(weight)
             )
             
-            # Init arc to sources
-            if topo_arc[1] == 1
+            # Initialization arcs
+            if topo_arc[1] in keys(initdict) 
                 init_arc = Arc(
-                    src = init_state,
+                    src = extra_state,
                     isym = 0,
                     osym = token,
                     dest = src,
-                    weight = S(weight)
+                    weight = S(initdict[topo_arc[1]])
+                )
+                push!(arcs, init_arc)
+            end
+            if topo_arc[2] in keys(initdict)
+                init_arc = Arc(
+                    src = extra_state,
+                    isym = 0,
+                    osym = token,
+                    dest = dest,
+                    weight = S(initdict[topo_arc[2]])
                 )
                 push!(arcs, init_arc)
             end
 
-            # Final arcs to destinations
-            if topo_arc[2] == size(mapping, 1)
+            # Final arcs
+            if topo_arc[1] in keys(finaldict)
+                final_arc = Arc(
+                    src = src,
+                    isym = 0,
+                    osym = 0,
+                    dest = extra_state,
+                    weight = S(finaldict[topo_arc[1]])
+                )
+                push!(arcs, final_arc)
+            end
+            if topo_arc[2] in keys(finaldict)
                 final_arc = Arc(
                     src = dest,
                     isym = 0,
                     osym = 0,
-                    dest = init_state,
-                    weight = S(weight)
-                ) 
+                    dest = extra_state,
+                    weight = S(finaldict[topo_arc[2]])
+                )
                 push!(arcs, final_arc)
             end
 
             push!(states, src)
             push!(states, dest)
             push!(arcs, arc)
-        end
-
-        # for (state, weight) in initweights
-        #     print(state)
-        #     state = offset + state
-        #     push!(states, state)
-        #     push!(init, state => S(weight))
-        # end
-
-        for (state, weight) in finalweights
-            state = offset + state
-            push!(states, state)
-            push!(final, state => S(weight))
-        end
-    end
-
-    # Actually, there is just one init state
-    for (_, weight) in initweights
-        push!(init, init_state => S(weight))
+        end    
     end
+    
+    push!(init, extra_state => one(S))
+    push!(final, extra_state => one(S))
 
-    TensorFST(arcs, init, final)
+    TensorFST(collect(Set(arcs)), init, final)
 end
 
 include("trie.jl")

src/symboltable.jl

@@ -44,7 +44,7 @@ function SymbolTable(syms)
 end
 
 """
-    syms(t [, exclude_ϵ])
+    syms(t [, exclude_ϵ=tru])
 
 Return all symbols from symbol table `t`.
 """
@@ -57,7 +57,7 @@ function syms(t::SymbolTable; exclude_ϵ=true)
 end
 
 """
-    ids(t [, exclude_ϵ])
+    ids(t [, exclude_ϵ=true])
 
 Return all IDs from symbol table `t`.
 """

test/speechrecfsts.jl

@@ -31,11 +31,11 @@ S = BoolSemiring
             # # Extra node
             # Source
             (src = (7,), isym = 0, osym = 4, weight = one(S), dest = (5,)),
-            (src = (7,), isym = 0, osym = 3, weight = one(S), dest = (3, )),
+            (src = (7,), isym = 0, osym = 3, weight = one(S), dest = (3,)),
             (src = (7,), isym = 0, osym = 2, weight = one(S), dest = (1,)),
             # Destination
             (src = (6,), isym = 0, osym = 0, weight = one(S), dest = (7,)),
-            (src = (4,), isym = 0, osym = 0, weight = one(S), dest = (7, )),
+            (src = (4,), isym = 0, osym = 0, weight = one(S), dest = (7,)),
             (src = (2,), isym = 0, osym = 0, weight = one(S), dest = (7,)),
 
 
@@ -49,7 +49,7 @@ S = BoolSemiring
 			
         ],
         [(7,) => one(S)],
-        [(4,) => one(S), (6,) => one(S), (2,) => one(S)],
+        [(7,) => one(S)]
     )
     @test issame(U, U2)
 end