refactor some error handling in peg, more unit tests

basis/peg/ebnf/ebnf-tests.factor

@@ -3,7 +3,7 @@
 !
 USING: kernel tools.test peg peg.ebnf words math math.parser 
        sequences accessors peg.parsers parser namespaces arrays 
-       strings eval ;
+       strings eval unicode.data multiline ;
 IN: peg.ebnf.tests
 
 { T{ ebnf-non-terminal f "abc" } } [
@@ -520,3 +520,13 @@ Tok                = Spaces (Number | Special )
 { "\\" } [
   "\\" [EBNF foo="\\" EBNF]
 ] unit-test
+
+[ "USE: peg.ebnf [EBNF EBNF]" eval ] must-fail
+
+[ <" USE: peg.ebnf [EBNF
+    lol = a
+    lol = b
+  EBNF] "> eval
+] [
+    error>> [ redefined-rule? ] [ name>> "lol" = ] bi and
+] must-fail-with

basis/peg/ebnf/ebnf.factor

@@ -5,13 +5,18 @@ sequences quotations vectors namespaces make math assocs
 continuations peg peg.parsers unicode.categories multiline
 splitting accessors effects sequences.deep peg.search
 combinators.short-circuit lexer io.streams.string stack-checker
-io combinators parser ;
+io combinators parser summary ;
 IN: peg.ebnf
 
 : rule ( name word -- parser )
   #! Given an EBNF word produced from EBNF: return the EBNF rule
   "ebnf-parser" word-prop at ;
 
+ERROR: no-rule rule parser ;
+
+: lookup-rule ( rule parser -- rule' )
+    2dup rule [ 2nip ] [ no-rule ] if* ; 
+
 TUPLE: tokenizer any one many ;
 
 : default-tokenizer ( -- tokenizer )
@@ -34,8 +39,13 @@ TUPLE: tokenizer any one many ;
 : reset-tokenizer ( -- )
   default-tokenizer \ tokenizer set-global ;
 
+ERROR: no-tokenizer name ;
+
+M: no-tokenizer summary
+    drop "Tokenizer not found" ;
+
 SYNTAX: TOKENIZER: 
-  scan search [ "Tokenizer not found" throw ] unless*
+  scan dup search [ nip ] [ no-tokenizer ] if*
   execute( -- tokenizer ) \ tokenizer set-global ;
 
 TUPLE: ebnf-non-terminal symbol ;
@@ -258,7 +268,7 @@ DEFER: 'choice'
     "]]" token ensure-not ,
     "]?" token ensure-not ,
     [ drop t ] satisfy ,
-  ] seq* [ first ] action repeat0 [ >string ] action ;
+  ] seq* repeat0 [ concat >string ] action ;
 
 : 'ensure-not' ( -- parser )
   #! Parses the '!' syntax to ensure that 
@@ -368,14 +378,15 @@ M: ebnf-tokenizer (transform) ( ast -- parser )
   dup parser-tokenizer \ tokenizer set-global
   ] if ;
 
+ERROR: redefined-rule name ;
+
+M: redefined-rule summary
+  name>> "Rule '" "' defined more than once" surround ;
+  
 M: ebnf-rule (transform) ( ast -- parser )
   dup elements>> 
   (transform) [
-    swap symbol>> dup get parser? [ 
+    swap symbol>> dup get parser? [ redefined-rule ] [ set ] if
-      "Rule '" over append "' defined more than once" append throw 
-    ] [ 
-      set 
-    ] if
   ] keep ;
 
 M: ebnf-sequence (transform) ( ast -- parser )
@@ -467,13 +478,17 @@ ERROR: bad-effect quot effect ;
     [ bad-effect ]
   } cond ;
 
+: ebnf-transform ( ast -- parser quot )
+  [ parser>> (transform) ]
+  [ code>> insert-escapes ]
+  [ parser>> ] tri build-locals  
+  [ string-lines parse-lines ] call( string -- quot ) ;
+ 
 M: ebnf-action (transform) ( ast -- parser )
-  [ parser>> (transform) ] [ code>> insert-escapes ] [ parser>> ] tri build-locals  
+  ebnf-transform check-action-effect action ;
-  [ string-lines parse-lines ] call( string -- quot ) check-action-effect action ;
 
 M: ebnf-semantic (transform) ( ast -- parser )
-  [ parser>> (transform) ] [ code>> insert-escapes ] [ parser>> ] tri build-locals 
+  ebnf-transform semantic ;
-  [ string-lines parse-lines ] call( string -- quot ) semantic ;
 
 M: ebnf-var (transform) ( ast -- parser )
   parser>> (transform) ;
@@ -481,19 +496,20 @@ M: ebnf-var (transform) ( ast -- parser )
 M: ebnf-terminal (transform) ( ast -- parser )
   symbol>> tokenizer one>> call( symbol -- parser ) ;
 
+ERROR: ebnf-foreign-not-found name ;
+
+M: ebnf-foreign-not-found summary
+  name>> "Foreign word '" "' not found" surround ;
+
 M: ebnf-foreign (transform) ( ast -- parser )
-  dup word>> search
+  dup word>> search [ word>> ebnf-foreign-not-found ] unless*
-  [ "Foreign word '" swap word>> append "' not found" append throw ] unless*
   swap rule>> [ main ] unless* over rule [
     nip
   ] [
     execute( -- parser )
   ] if* ;
 
-: parser-not-found ( name -- * )
+ERROR: parser-not-found name ;
-  [
-    "Parser '" % % "' not found." %
-  ] "" make throw ;
 
 M: ebnf-non-terminal (transform) ( ast -- parser )
   symbol>>  [
@@ -504,16 +520,16 @@ M: ebnf-non-terminal (transform) ( ast -- parser )
   'ebnf' parse transform ;
 
 : check-parse-result ( result -- result )
-  dup [
+  [
-    dup remaining>> [ blank? ] trim empty? [
+    dup remaining>> [ blank? ] trim [
       [ 
         "Unable to fully parse EBNF. Left to parse was: " %
         remaining>> % 
       ] "" make throw
-    ] unless
+    ] unless-empty
   ] [
     "Could not parse EBNF" throw
-  ] if ;
+  ] if* ;
 
 : parse-ebnf ( string -- hashtable )
   'ebnf' (parse) check-parse-result ast>> transform ;
@@ -522,14 +538,18 @@ M: ebnf-non-terminal (transform) ( ast -- parser )
   parse-ebnf dup dup parser [ main swap at compile ] with-variable
   [ compiled-parse ] curry [ with-scope ast>> ] curry ;
 
-SYNTAX: <EBNF "EBNF>" reset-tokenizer parse-multiline-string parse-ebnf main swap at  
+SYNTAX: <EBNF
+  "EBNF>"
+  reset-tokenizer parse-multiline-string parse-ebnf main swap at  
   parsed reset-tokenizer ;
 
-SYNTAX: [EBNF "EBNF]" reset-tokenizer parse-multiline-string ebnf>quot nip 
+SYNTAX: [EBNF
+  "EBNF]"
+  reset-tokenizer parse-multiline-string ebnf>quot nip 
   parsed \ call parsed reset-tokenizer ;
 
 SYNTAX: EBNF: 
   reset-tokenizer CREATE-WORD dup ";EBNF" parse-multiline-string  
-  ebnf>quot swapd (( input -- ast )) define-declared "ebnf-parser" set-word-prop 
+  ebnf>quot swapd
+  (( input -- ast )) define-declared "ebnf-parser" set-word-prop 
   reset-tokenizer ;
-