Salta al contingut
Ridgun
Català

15 El Llibre de Ridgun

La lib codec — Hex, Base64, Base64url, URL Percent-Encoding i JSON Lleuger

15.1. Per què existeix la lib codec? (El problema real i la filosofia Ridgun)

«Per què necessito una lib codec escrita des de zero en Ridgun pur en lloc d'enllaçar una llibreria de C de tota la vida?»

Aquesta és la primera pregunta que es fa qualsevol enginyer de sistemes en arribar a codec. I la resposta resumeix tota la filosofia de ridgun-libs:

Siguem sincers: res no trenca més ràpid una integració web que confondre Base64 estàndard (+/=) amb Base64url (-_ sense farciment) en un token JWT, o construir una query string concatenant cadenes sense escapar.

La lib codec implementa els estàndards RFC 4648 (Hex, Base64 i Base64url), RFC 3986 (codificació percentual d'URLs i diccionaris query encode / query decode) i constructors/extractors ràpids de JSON lleuger (RFC 8259).

En un ecosistema tradicional, resoldre les tasques del domini CODEC implica arrossegar dependències externes en C, lluitar amb punters crus, conversions opaques de tipus, excepcions no verificades o pauses impredictibles del recol·lector de brossa. En Ridgun, la lib codec (src/main/ridgun/lib/codec.rg) aborda el problema amb quatre principis innegociables:

  1. Declara import codec (o import codec as alias) a l'inici del fitxer: cada lib defineix el seu espai de noms module :: codec i en mode --strict les funcions multi-paraula s'invoquen sempre amb espais naturals.
  2. Encadena les teves canonades d'esquerra a dreta amb ->: escriu entrada pas u -> pas dos -> resultat, fes servir màscares (+ _ _ +) per desestructurar seqüències i omet _ = en sentències seqüencials (o descarta amb -> ()).
  3. Recursió de cua amb TCO (Tail-Call Optimization): tots els recorreguts interns de codec es compilen a bucles natius en espai de pila $O(1)$.
  4. Memòria determinista (RMM1 / RMM2): cada estructura intermèdia s'allibera en l'instant exacte en què acaba la seva última lectura, sense Garbage Collector.

15.2. Artefactes natius, importació i catàleg complet de l'API de codec

A ridgun-libs, codec és una lib de primera classe situada a src/main/ridgun/lib/codec.rg (domini CODEC). En compilar amb stable bundle, s'empaqueta en tres formats natius:

ArtefacteRutaPropòsit
Bundle Ridguntarget/bundles/codec.rglibDistribució i importació directa amb import codec
Fitxer compilattarget/codec.rgcArtefacte intermedi verificat per ridgunc
Objecte compartit ELF64target/libcodec.soBiblioteca dinàmica nativa sense dependències de C ni libc

Cada lib declara el seu propi espai de noms amb module :: codec: n'hi ha prou d'escriure import codec (o import codec as alias) a la capçalera i cridar les seves funcions amb espais naturals (la meva funcio x) i canonades -> (x la meva funcio -> resultat).

15.2.1. Resum de famílies funcionals de codec

Signatura / Funció / TipusQuè fa i quan fer-la servir
hex encode / hex decode / hex is validCodificació hexadecimal RFC 4648 en minúscules i descodificació validada
base64 encode / base64 decodeBase64 estàndard RFC 4648 §4 amb farciment =
base64url encode / base64url decodeBase64url segur per a URLs RFC 4648 §5 (-, _, sense farciment)
url encode / url decode / query encode / query decodeCodificació percentual RFC 3986 i diccionaris de query string k=v&...
json quote / json object / json get str / json get intSerialització JSON ràpida sense AST i extracció de claus de primer nivell

15.2.2. Catàleg exhaustiu de tipus i signatures a src/main/ridgun/lib/codec.rg

A continuació tens l'inventari complet de tipus algebraics (type), operadors (op ::) i signatures de funcions (fn ::) definits a src/main/ridgun/lib/codec.rg (0 tipus, 0 operadors i 121 funcions):

#Funció / Signatura (fn ::)Tipus de Retorn
1fn :: cd raw byte ub IntString
2fn :: cd raw byte IntString
3fn :: cd sb append3 String StringString String String String
4fn :: cd sb append2 String String StringString String String String
5fn :: cd sb append1 String String String StringString String String String
6fn :: cd sb append String String String String StringString String String String
7fn :: cd sb finish String String String StringString
8fn :: cd lb append2 [String] [String] Int[String] [String] [String] Int
9fn :: cd lb append1 [String] [String] [String] Int[String] [String] [String] Int
10fn :: cd lb append String [String] [String] [String] Int[String] [String] [String] Int
11fn :: cd lb finish [String] [String] [String][String]
12fn :: cd hex lut IntString
13fn :: cd hex up lut IntString
14fn :: cd b64 lut IntString
15fn :: cd b64url lut IntString
16fn :: cd match sub at String String Int Int IntBool
17fn :: cd index of loop String String Int Int Int IntInt
18fn :: cd index of String StringInt
19fn :: cd scan b end String IntInt
20fn :: cd scan b prev String IntInt
21fn :: cd scan ascii end String Int IntInt
22fn :: cd byte slice step String Int Int Int String String String StringString
23fn :: cd byte slice loop String Int Int String String String StringString
24fn :: cd byte slice check String Int Int Int IntString
25fn :: cd byte slice String Int IntString
26fn :: cd split char loop String String Int Int Int Int [String] [String] [String] Int[String]
27fn :: cd hex val IntInt
28fn :: cd valid hex loop String Int IntBool
29fn :: cd is valid hex len String IntBool
30fn :: is valid hex StringBool
31fn :: cd hex enc byte IntString
32fn :: cd hex enc loop String Int Int String String String StringString
33fn :: hex encode StringString
34fn :: cd hex dec loop String Int Int String String String StringString
35fn :: hex decode StringString
36fn :: cd b64 char String IntString
37fn :: cd b64 pad suffix String BoolString
38fn :: cd b64 enc step String Int String Bool Int String String String StringString
39fn :: cd b64 enc rem String Int String BoolString
40fn :: base64 encode StringString
41fn :: base64url encode StringString
42fn :: cd b64 val IntInt
43fn :: cd strip one b64 pad String IntString
44fn :: cd strip b64 pad String IntString
45fn :: cd valid b64 body String Int IntBool
46fn :: cd valid b64 raw String IntBool
47fn :: is valid base64 StringBool
48fn :: cd b64 dec step String Int Int Int String String String StringString
49fn :: cd b64 dec loop String Int IntString
50fn :: base64 decode StringString
51fn :: base64url decode StringString
52fn :: cd is unreserved IntBool
53fn :: cd scan unres String String Int IntInt
54fn :: cd pct byte lut String IntString
55fn :: cd url enc step String String String Int Int Int String String String StringString
56fn :: cd url enc loop String String String Int Int String String String StringString
57fn :: url encode StringString
58fn :: cd scan url plain String Int IntInt
59fn :: cd url dec pct String Int Int Int Int String String String String StringString
60fn :: cd url dec step String Int Int Int String String String StringString
61fn :: cd url dec loop String Int Int String String String StringString
62fn :: url decode StringString
63fn :: cd pair key (String String)String
64fn :: cd pair val (String String)String
65fn :: cd qs enc loop [String String] String String String StringString
66fn :: query string encode [String String]String
67fn :: cd parse kv at String IntString String
68fn :: cd parse kv item StringString String
69fn :: cd kv lb append2 [String String] [String String] Int[String String] [String String] [String String] Int
70fn :: cd kv lb append1 [String String] [String String] [String String] Int[String String] [String String] [String String] Int
71fn :: cd kv lb append (String String) [String String] [String String] [String String] Int[String String] [String String] [String String] Int
72fn :: cd parse qs list [String] [String String] [String String] [String String] Int[String String]
73fn :: cd strip leading q StringString
74fn :: cd decode clean qs String[String String]
75fn :: query string decode String[String String]
76fn :: cd is json esc IntBool
77fn :: cd scan json plain String Int IntInt
78fn :: cd json esc hex2 IntString
79fn :: cd json esc code ub IntString
80fn :: cd json esc code IntString
81fn :: cd json esc step String Int Int Int String String String StringString
82fn :: cd json esc loop String Int Int String String String StringString
83fn :: json escape StringString
84fn :: cd json unesc char String Int IntString
85fn :: cd scan no backslash String Int IntInt
86fn :: cd json unesc cp Int IntString Int
87fn :: cd json unesc u4 Int Int Int Int IntString Int
88fn :: cd json unesc step String Int Int Int String String String StringString
89fn :: cd json unesc loop String Int Int String String String StringString
90fn :: json unescape StringString
91fn :: json quote StringString
92fn :: json int IntString
93fn :: json bool BoolString
94fn :: json nullString
95fn :: cd join json loop [String] String String String StringString
96fn :: cd join json items [String]String
97fn :: json array [String]String
98fn :: cd join json pairs loop [String String] String String String StringString
99fn :: cd join json pairs [String String]String
100fn :: json object [String String]String
101fn :: cd quote pair vals [String String] [String String] [String String] [String String] Int[String String]
102fn :: json str object [String String]String
103fn :: cd is json ws IntBool
104fn :: cd skip key sep String Int IntInt
105fn :: cd scan json str end String Int IntInt
106fn :: cd is tok delim IntBool
107fn :: cd scan json tok end String Int IntInt
108fn :: cd extract json str vend String Int IntString
109fn :: cd extract json str vstart String IntString
110fn :: cd extract json str kpos String String IntString
111fn :: codec json get str String StringString
112fn :: cd is signed int loop String Int IntBool
113fn :: cd is valid int len String IntBool
114fn :: cd is valid int str StringBool
115fn :: cd raw tok vstart String IntString
116fn :: cd raw tok kpos String String IntString
117fn :: cd raw json token String StringString
118fn :: cd parse tok int String IntInt
119fn :: codec json get int String String IntInt
120fn :: cd parse tok bool String BoolBool
121fn :: codec json get bool String String BoolBool

15.3. Explorant codec a fons: Exemples pas a pas de cada funcionalitat

No n'hi ha prou de veure la taula de signatures: explorarem cadascuna de les funcionalitats de la lib codec mitjançant exemples progressius i executables en Ridgun. Començarem per les dues receptes tabulars d'ús quotidià i després recorrerem en detall tots els escenaris funcionals, transformacions i casos límit verificats a src/test/ridgun/test codec.rg.

15.3.1. Exemple 1: Flux principal en estil tabular amb codec

Observa com en Ridgun els signes = i els passos de transformació amb codec queden alineats com una taula neta, sense variables mutables innecessàries ni efectes laterals ocults:

import codec

fn :: empaquetar parametros String String => String
      empaquetar parametros usuario redir =
        usuario base64url encode -> token_b64
        query encode [("token", token_b64), ("next", redir)]

15.3.2. Exemple 2: Composició segura i gestió de casos de vora

Tota funció de la lib codec està dissenyada per compondre's amb guardes booleanes | i amb la resta de les 32 libs de ridgun-libs:

import codec

fn :: construir respuesta rapida Int String => String Int
      construir respuesta rapida id nombre =
        json object [("id", json int id), ("name", json quote nombre), ("ok", json bool True)] -> doc
        json get int doc "id" 0 -> leido
        (doc, leido)

15.3.3. Exemple 3: Codificació i descodificació RFC 4648 Hexadecimal, Base64 estàndard i Base64url segur per a URLs

En aquest exemple pràctic explorem de manera exhaustiva com es comporta codec en aquest bloc funcional, incloent-hi entrades estàndard, combinacions de paràmetres i valors frontera:

fn :: test rfc4648 hex and base64 => ()
      test rfc4648 hex and base64 =
        -- 1. Hexadecimal RFC 4648 test vectors & negative validation
        assert str eq (hex encode "") "" "hex_encode empty"
        assert str eq (hex encode "foobar") "666f6f626172" "hex_encode foobar"
        assert str eq (hex decode "666f6f626172") "foobar" "hex_decode lowercase"
        assert str eq (hex decode "666F6F626172") "foobar" "hex_decode uppercase"
        assert true (is valid hex "deadBEEF09") "is_valid_hex valid"
        assert false (is valid hex "abc") "is_valid_hex odd length"
        assert false (is valid hex "deadZZ") "is_valid_hex invalid chars"
        assert str eq (hex decode "abc") "" "hex_decode odd -> empty"

        -- 2. Base64 & Base64url RFC 4648 test vectors
        assert str eq (base64 encode "") "" "b64 empty"
        assert str eq (base64 encode "f") "Zg==" "b64 f"
        assert str eq (base64 encode "fo") "Zm8=" "b64 fo"
        assert str eq (base64 encode "foo") "Zm9v" "b64 foo"
        assert str eq (base64 encode "foob") "Zm9vYg==" "b64 foob"
        assert str eq (base64 encode "fooba") "Zm9vYmE=" "b64 fooba"
        assert str eq (base64 encode "foobar") "Zm9vYmFy" "b64 foobar"
        assert str eq (base64 decode "Zg==") "f" "b64 dec f"
        assert str eq (base64 decode "Zm8=") "fo" "b64 dec fo"
        assert str eq (base64 decode "Zm9v") "foo" "b64 dec foo"
        assert str eq (base64 decode "Zm9vYg==") "foob" "b64 dec foob"
        assert str eq (base64 decode "Zm9vYmE=") "fooba" "b64 dec fooba"
        assert str eq (base64 decode "Zm9vYmFy") "foobar" "b64 dec foobar"
        assert str eq (base64url encode "foob") "Zm9vYg" "b64url nopad"
        assert str eq (base64url decode "Zm9vYg") "foob" "b64url dec"
        assert false (is valid base64 "A") "invalid base64 len%4==1"
        assert false (is valid base64 "Zm$v") "invalid base64 char"
        assert str eq (base64 decode "Zm$v") "" "invalid base64 -> empty"

        -- 3. LIB-8: Binary-safe decoding for raw octets >= 0x80 and 0x00
        raw4 = hex decode "c800ff41"
        assert int equals (str len raw4) 4 "str_len (hex_decode c800ff41) == 4"
        assert int equals (str char at raw4 0) 200 "raw4[0] == 0xc8"
        assert int equals (str char at raw4 1) 0 "raw4[1] == 0x00"
        assert int equals (str char at raw4 2) 255 "raw4[2] == 0xff"
        assert int equals (str char at raw4 3) 65 "raw4[3] == 0x41"
        assert str eq (hex encode raw4) "c800ff41" "hex_encode raw4 == c800ff41"
        check binary roundtrip loop 10 -> ()
        ()

15.3.4. Exemple 4: Codificació percentual RFC 3986 (url encode, url decode) i cadenes de consulta (query encode, query get)

En aquest exemple pràctic explorem de manera exhaustiva com es comporta codec en aquest bloc funcional, incloent-hi entrades estàndard, combinacions de paràmetres i valors frontera:

fn :: test rfc3986 url and query strings => ()
      test rfc3986 url and query strings =
        raw = "hello world&x=1+2%"
        enc = url encode raw
        assert str eq enc "hello%20world%26x%3D1%2B2%25" "url_encode"
        assert str eq (url decode enc) raw "url_decode roundtrip"
        assert str eq (url decode "a+b%20c") "a b c" "url_decode plus as space"
        qs  = query string encode [("q", "ridgun libs"), ("page", "2")]
        assert str eq qs "q=ridgun%20libs&page=2" "query_string_encode"
        dec = query string decode ("?" + qs)
        assert eq dec [("q", "ridgun libs"), ("page", "2")]
        ()

15.3.5. Exemple 5: Escapament i formatació JSON lleugera RFC 8259 (json escape, json unescape, json encode kv, json get str/int/bool)

En aquest exemple pràctic explorem de manera exhaustiva com es comporta codec en aquest bloc funcional, incloent-hi entrades estàndard, combinacions de paràmetres i valors frontera:

fn :: test rfc8259 lightweight json => ()
      test rfc8259 lightweight json =
        esc = json escape "line1\n\"quoted\"\\tab\t"
        assert str eq esc "line1\\n\\\"quoted\\\"\\\\tab\\t" "json_escape"
        assert str eq (json unescape esc) "line1\n\"quoted\"\\tab\t" "json_unescape"
        doc = json object [("service", json quote "solstice"), ("port", json int 8443), ("tls", json bool True), ("extra", json null)]
        assert str eq doc "{\"service\":\"solstice\",\"port\":8443,\"tls\":true,\"extra\":null}" "json_object"
        assert str eq (json get str doc "service") "solstice" "json_get_str"
        assert str eq (json get str doc "missing") "" "json_get_str missing"
        assert int equals (json get int doc "port" 80) 8443 "json_get_int"
        assert int equals (json get int doc "missing" 80) 80 "json_get_int fallback"
        assert int equals (json get int doc "service" 80) 80 "json_get_int non-int fallback"
        assert true (json get bool doc "tls" False) "json_get_bool"
        assert false (json get bool doc "missing" False) "json_get_bool fallback"
        arr = json array [json int 1, json int 2, json quote "three"]
        assert str eq arr "[1,2,\"three\"]" "json_array"
        s_o = json str object [("k1", "v1"), ("k2", "v\"2")]
        assert str eq (json get str s_o "k2") "v\"2" "json_str_object escaped roundtrip"
        ()

15.4. Sota el capó: com està construïda codec en Ridgun pur (src/main/ridgun/lib/codec.rg)

Un dels majors avantatges de ridgun-libs és que no hi ha caixes negres en C. Pots obrir src/main/ridgun/lib/codec.rg i llegir exactament com està implementada cada operació fent servir pattern matching, arbres de guardes | / |. i recursió de cua (TCO). Vegem les peces clau de la seva implementació real:

15.4.1. Implementació interna: RFC 4648 Hexadecimal Encoding & Decoding

Fixa't en com codec.rg estructura les signatures fn :: ... => ... primer i resol cada cas mitjançant equacions alineades i crides de cua:

module :: codec

fn :: cd hex val Int => Int
      cd hex val c   =
        | c >= 48 and c <= 57  = c - 48
        | c >= 65 and c <= 70  = (c - 65) + 10
        | c >= 97 and c <= 102 = (c - 97) + 10
        | otherwise            = 0 - 1

fn :: cd valid hex loop String Int Int => Bool
      cd valid hex loop s idx len =
        | idx >= len                           = True
        | (cd hex val (str char at s idx)) < 0 = False
        | otherwise                            = cd valid hex loop s (idx + 1) len

fn :: cd is valid hex len String Int => Bool
      cd is valid hex len s n =
        | (n % 2) != 0 = False
        | otherwise    = cd valid hex loop s 0 n

fn :: is valid hex String => Bool
      is valid hex s      =  cd is valid hex len s (str len s)

fn :: cd hex enc byte Int => String
      cd hex enc byte b   =
        ub  = ((b % 256) + 256) % 256
        hi  = ub / 16
        lo  = ub % 16
        tbl = "0123456789abcdef"
        (str slice tbl hi (hi + 1)) + (str slice tbl lo (lo + 1))

fn :: cd hex enc loop String Int Int String String String String => String
      cd hex enc loop s idx len c0 c1 c2 c3 =
        | idx >= len = cd sb finish c0 c1 c2 c3
        | otherwise  =
          n0, n1, n2, n3 = cd sb append (cd hex enc byte (str char at s idx)) c0 c1 c2 c3
          cd hex enc loop s (idx + 1) len n0 n1 n2 n3

fn :: hex encode String => String
      hex encode s      =  cd hex enc loop s 0 (str len s) "" "" "" ""

fn :: cd hex dec loop String Int Int String String String String => String
      cd hex dec loop s idx len c0 c1 c2 c3 =
        | idx >= len = cd sb finish c0 c1 c2 c3
        | otherwise =
          hi = cd hex val (str char at s idx)
          lo = cd hex val (str char at s (idx + 1))
          b  = cd raw byte ub ((hi * 16) + lo)
          n0, n1, n2, n3 = cd sb append b c0 c1 c2 c3
          cd hex dec loop s (idx + 2) len n0 n1 n2 n3

fn :: hex decode String => String
      hex decode s      =
        | not (is valid hex s) = ""
        | otherwise            = cd hex dec loop s 0 (str len s) "" "" "" ""

15.4.2. Implementació interna: RFC 4648 Standard Base64 & URL-Safe Base64url

Fixa't en com codec.rg estructura les signatures fn :: ... => ... primer i resol cada cas mitjançant equacions alineades i crides de cua:

fn :: cd b64 char String Int => String
      cd b64 char tbl    idx =
        i = idx % 64
        str slice tbl i (i + 1)

fn :: cd b64 pad suffix String Bool => String
      cd b64 pad suffix suffix True = suffix
      cd b64 pad suffix _ False = ""

fn :: cd b64 enc step String Int String Bool Int String String String String => String
      cd b64 enc step s idx tbl pad rem c0 c1 c2 c3 =
        | rem <= 0 = cd sb finish c0 c1 c2 c3
        | rem == 1 =
          b0   = str char at s idx
          out2 = (cd b64 char tbl (b0 / 4)) + (cd b64 char tbl ((b0 % 4) * 16)) + (cd b64 pad suffix "==" pad)
          cd sb finish (c0 + out2) c1 c2 c3
        | rem == 2 =
          b0   = str char at s idx
          b1   = str char at s (idx + 1)
          n    = (b0 * 256) + b1
          out3 = (cd b64 char tbl (n / 1024)) + (cd b64 char tbl ((n / 16) % 64)) + (cd b64 char tbl ((n % 16) * 4)) + (cd b64 pad suffix "=" pad)
          cd sb finish (c0 + out3) c1 c2 c3
        | otherwise =
          b0   = str char at s idx
          b1   = str char at s (idx + 1)
          b2   = str char at s (idx + 2)
          n    = (b0 * 65536) + (b1 * 256) + b2
          out4 = (cd b64 char tbl (n / 262144)) + (cd b64 char tbl ((n / 4096) % 64)) + (cd b64 char tbl ((n / 64) % 64)) + (cd b64 char tbl (n % 64))
          n0, n1, n2, n3 = cd sb append out4 c0 c1 c2 c3
          cd b64 enc step s (idx + 3) tbl pad (rem - 3) n0 n1 n2 n3

fn :: cd b64 enc rem String Int String Bool => String
      cd b64 enc rem s idx tbl pad = cd b64 enc step s idx tbl pad ((str len s) - idx) "" "" "" ""

fn :: base64 encode String => String
      base64 encode s      =  cd b64 enc rem s 0 (cd b64 lut 0) True

fn :: base64url encode String => String
      base64url encode s      =  cd b64 enc rem s 0 (cd b64url lut 0) False

fn :: cd b64 val Int => Int
      cd b64 val c   =
        | c >= 65 and c <= 90  = c - 65
        | c >= 97 and c <= 122 = (c - 97) + 26
        | c >= 48 and c <= 57  = (c - 48) + 52
        | c == 43 or c == 45   = 62
        | c == 47 or c == 95   = 63
        | otherwise            = 0 - 1

fn :: cd strip one b64 pad String Int => String
      cd strip one b64 pad s n =
        | n <= 0                        = ""
        | (str char at s (n - 1)) == 61 = str slice s 0 (n - 1)
        | otherwise                     = s

fn :: cd strip b64 pad String Int => String
      cd strip b64 pad s n =
        | (n % 4) != 0 = s
        | otherwise =
          s1 = cd strip one b64 pad s n
          cd strip one b64 pad s1 (str len s1)

fn :: cd valid b64 body String Int Int => Bool
      cd valid b64 body s idx len =
        | idx >= len                           = True
        | (cd b64 val (str char at s idx)) < 0 = False
        | otherwise                            = cd valid b64 body s (idx + 1) len

fn :: cd valid b64 raw String Int => Bool
      cd valid b64 raw raw n =
        | (n % 4) == 1 = False
        | otherwise    = cd valid b64 body raw 0 n

fn :: is valid base64 String => Bool
      is valid base64 s      =
        cd strip b64 pad s (str len s) -> raw
        cd valid b64 raw raw (str len raw)

fn :: cd b64 dec step String Int Int Int String String String String => String
      cd b64 dec step s idx len rem c0 c1 c2 c3 =
        | rem <= 1 = cd sb finish c0 c1 c2 c3
        | rem == 2 =
          v0 = cd b64 val (str char at s idx)
          v1 = cd b64 val (str char at s (idx + 1))
          b0 = cd raw byte ub (((v0 * 4) + (v1 / 16)) % 256)
          cd sb finish (c0 + b0) c1 c2 c3
        | rem == 3 =
          v0 = cd b64 val (str char at s idx)
          v1 = cd b64 val (str char at s (idx + 1))
          v2 = cd b64 val (str char at s (idx + 2))
          n  = (v0 * 1024) + (v1 * 16) + (v2 / 4)
          b01 = (cd raw byte ub ((n / 256) % 256)) + (cd raw byte ub (n % 256))
          cd sb finish (c0 + b01) c1 c2 c3
        | otherwise =
          v0 = cd b64 val (str char at s idx)
          v1 = cd b64 val (str char at s (idx + 1))
          v2 = cd b64 val (str char at s (idx + 2))
          v3 = cd b64 val (str char at s (idx + 3))
          n  = (v0 * 262144) + (v1 * 4096) + (v2 * 64) + v3
          b0 = (n / 65536) % 256
          b1 = (n / 256) % 256
          b2 = n % 256
          chunk = (cd raw byte ub b0) + (cd raw byte ub b1) + (cd raw byte ub b2)
          n0, n1, n2, n3 = cd sb append chunk c0 c1 c2 c3
          cd b64 dec step s (idx + 4) len (rem - 4) n0 n1 n2 n3

fn :: cd b64 dec loop String Int Int => String
      cd b64 dec loop s idx len = cd b64 dec step s idx len (len - idx) "" "" "" ""

fn :: base64 decode String => String
      base64 decode s      =
        | not (is valid base64 s) = ""
        | otherwise =
          raw = cd strip b64 pad s (str len s)
          cd b64 dec loop raw 0 (str len raw)

fn :: base64url decode String => String
      base64url decode s      =  base64 decode s

15.4.3. Implementació interna: RFC 3986 URL Percent-Encoding & Query Strings

Fixa't en com codec.rg estructura les signatures fn :: ... => ... primer i resol cada cas mitjançant equacions alineades i crides de cua:

fn :: cd is unreserved Int => Bool
      cd is unreserved c   =
        | c >= 65 and c <= 90                        = True
        | c >= 97 and c <= 122                       = True
        | c >= 48 and c <= 57                        = True
        | c == 45 or c == 95 or c == 46 or c == 126  = True
        | otherwise                                  = False

fn :: cd scan unres String String Int Int => Int
      cd scan unres s umap idx len =
        | idx >= len = len
        | (str char at umap ((str char at s idx) % 256)) == 49 =
          cd scan unres s umap (idx + 1) len
        | otherwise = idx

fn :: cd pct byte lut String Int => String
      cd pct byte lut lut b =
        ub = ((b % 256) + 256) % 256
        hi = ub / 16
        lo = ub % 16
        "%" + (str slice lut hi (hi + 1)) + (str slice lut lo (lo + 1))

fn :: cd url enc step String String String Int Int Int String String String String => String
      cd url enc step s umap lut run_start next_idx len c0 c1 c2 c3 =
        | next_idx >= len =
          | run_start == 0 = s
          | otherwise      = cd sb finish (c0 + (str slice s run_start len)) c1 c2 c3 |.
        | otherwise =
          piece = (str slice s run_start next_idx) + (cd pct byte lut lut (str char at s next_idx))
          n0, n1, n2, n3 = cd sb append piece c0 c1 c2 c3
          cd url enc loop s umap lut (next_idx + 1) len n0 n1 n2 n3

fn :: cd url enc loop String String String Int Int String String String String => String
      cd url enc loop s umap lut run_start len c0 c1 c2 c3 =
        | run_start >= len = cd sb finish c0 c1 c2 c3
        | otherwise        = cd url enc step s umap lut run_start (cd scan unres s umap run_start len) len c0 c1 c2 c3

fn :: url encode String => String
      url encode s      =
        umap = "0000000000000000000000000000000000000000000001101111111111000000011111111111111111111111111000010111111111111111111111111110001000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000"
        cd url enc loop s umap "0123456789ABCDEF" 0 (str len s) "" "" "" ""

fn :: cd scan url plain String Int Int => Int
      cd scan url plain s idx len =
        | idx >= len = len
        | (str char at s idx) == 37 or (str char at s idx) == 43 = idx
        | otherwise = cd scan url plain s (idx + 1) len

fn :: cd url dec pct String Int Int Int Int String String String String String => String
      cd url dec pct s next_idx len hi lo prefix c0 c1 c2 c3 =
        | hi >= 0 and lo >= 0 =
          piece = prefix + (cd raw byte ub ((hi * 16) + lo))
          n0, n1, n2, n3 = cd sb append piece c0 c1 c2 c3
          cd url dec loop s (next_idx + 3) len n0 n1 n2 n3
        | otherwise =
          piece = prefix + "%"
          n0, n1, n2, n3 = cd sb append piece c0 c1 c2 c3
          cd url dec loop s (next_idx + 1) len n0 n1 n2 n3

fn :: cd url dec step String Int Int Int String String String String => String
      cd url dec step s run_start next_idx len c0 c1 c2 c3 =
        | next_idx >= len =
          | run_start == 0 = s
          | otherwise      = cd sb finish (c0 + (cd byte slice s run_start len)) c1 c2 c3 |.
        | (str char at s next_idx) == 43 =
          piece = (cd byte slice s run_start next_idx) + " "
          n0, n1, n2, n3 = cd sb append piece c0 c1 c2 c3
          cd url dec loop s (next_idx + 1) len n0 n1 n2 n3
        | (next_idx + 2) < len =
          prefix = cd byte slice s run_start next_idx
          hi = cd hex val (str char at s (next_idx + 1))
          lo = cd hex val (str char at s (next_idx + 2))
          cd url dec pct s next_idx len hi lo prefix c0 c1 c2 c3
        | otherwise =
          piece = (cd byte slice s run_start next_idx) + "%"
          n0, n1, n2, n3 = cd sb append piece c0 c1 c2 c3
          cd url dec loop s (next_idx + 1) len n0 n1 n2 n3

fn :: cd url dec loop String Int Int String String String String => String
      cd url dec loop s run_start len c0 c1 c2 c3 =
        | run_start >= len = cd sb finish c0 c1 c2 c3
        | otherwise        = cd url dec step s run_start (cd scan url plain s run_start len) len c0 c1 c2 c3

fn :: url decode String => String
      url decode s      =  cd url dec loop s 0 (str len s) "" "" "" ""

fn :: cd pair key (String String) => String
      cd pair key (k, _)          =  k

fn :: cd pair val (String String) => String
      cd pair val (_, v)          =  v

fn :: cd qs enc loop [String String] String String String String => String
      cd qs enc loop []          c0 c1 c2 c3 =  cd sb finish c0 c1 c2 c3
      cd qs enc loop (pair:rest) c0 c1 c2 c3 =
        piece = "&" + (url encode (cd pair key pair)) + "=" + (url encode (cd pair val pair))
        n0, n1, n2, n3 = cd sb append piece c0 c1 c2 c3
        cd qs enc loop rest n0 n1 n2 n3

fn :: query string encode [String String] => String
      query string encode []              =  ""
      query string encode (pair:rest)     =
        first = (url encode (cd pair key pair)) + "=" + (url encode (cd pair val pair))
        cd qs enc loop rest first "" "" ""

fn :: cd parse kv at String Int => String String
      cd parse kv at item eq_idx =
        | eq_idx < 0 = (url decode item, "")
        | otherwise  = (url decode (cd byte slice item 0 eq_idx), url decode (cd byte slice item (eq_idx + 1) (str len item)))

fn :: cd parse kv item String => String String
      cd parse kv item item   =  cd parse kv at item (cd index of item "=")

fn :: cd kv lb append2 [String String] [String String] Int => [String String] [String String] [String String] Int
      cd kv lb append2 n1 l2 cnt =
        | (cnt % 1024) == 0 = ([], [], l2 :: n1, cnt)
        | otherwise         = ([], n1, l2, cnt)

fn :: cd kv lb append1 [String String] [String String] [String String] Int => [String String] [String String] [String String] Int
      cd kv lb append1 n0 l1 l2 cnt =
        | (cnt % 32) == 0 = cd kv lb append2 (l1 :: n0) l2 cnt
        | otherwise       = (n0, l1, l2, cnt)

fn :: cd kv lb append (String String) [String String] [String String] [String String] Int => [String String] [String String] [String String] Int
      cd kv lb append item l0 l1 l2 cnt =
        cd kv lb append1 (l0 :: [item]) l1 l2 (cnt + 1)

fn :: cd parse qs list [String] [String String] [String String] [String String] Int => [String String]
      cd parse qs list []       l0 l1 l2 _   =  l2 :: (l1 :: l0)
      cd parse qs list (x:rest) l0 l1 l2 cnt =
        | (str len x) == 0 = cd parse qs list rest l0 l1 l2 cnt
        | otherwise        =
          n0, n1, n2, ncnt = cd kv lb append (cd parse kv item x) l0 l1 l2 cnt
          cd parse qs list rest n0 n1 n2 ncnt

fn :: cd strip leading q String => String
      cd strip leading q qs     =
        | str starts with qs "?" = str slice qs 1 (str len qs)
        | otherwise              = qs

fn :: cd decode clean qs String => [String String]
      cd decode clean qs clean  =
        | (str len clean) == 0 = []
        | otherwise            = cd parse qs list (cd split char loop clean "&" 38 0 0 (str len clean) [] [] [] 0) [] [] [] 0

fn :: query string decode String => [String String]
      query string decode qs     =  cd decode clean qs (cd strip leading q qs)

15.4.4. Implementació interna: RFC 8259 Lightweight JSON Formatting & Field Extraction

Fixa't en com codec.rg estructura les signatures fn :: ... => ... primer i resol cada cas mitjançant equacions alineades i crides de cua:

fn :: cd is json esc Int => Bool
      cd is json esc c =
        ub = ((c % 256) + 256) % 256
        ub < 32 or ub == 34 or ub == 92

fn :: cd scan json plain String Int Int => Int
      cd scan json plain s idx len =
        | idx >= len = len
        | cd is json esc (str char at s idx) = idx
        | otherwise = cd scan json plain s (idx + 1) len

fn :: cd json esc hex2 Int => String
      cd json esc hex2 ub =
        hi  = (ub / 16) % 16
        lo  = ub % 16
        tbl = "0123456789abcdef"
        "\\u00" + (str slice tbl hi (hi + 1)) + (str slice tbl lo (lo + 1))

fn :: cd json esc code ub Int => String
      cd json esc code ub ub =
        | ub == 34  = "\\\""
        | ub == 92  = "\\\\"
        | ub == 8   = "\\b"
        | ub == 12  = "\\f"
        | ub == 10  = "\\n"
        | ub == 13  = "\\r"
        | ub == 9   = "\\t"
        | otherwise = cd json esc hex2 ub

fn :: cd json esc code Int => String
      cd json esc code c =
        cd json esc code ub (((c % 256) + 256) % 256)

fn :: cd json esc step String Int Int Int String String String String => String
      cd json esc step s idx next_idx len c0 c1 c2 c3 =
        | next_idx >= len =
          | idx == 0  = s
          | otherwise = cd sb finish (c0 + (cd byte slice s idx len)) c1 c2 c3 |.
        | otherwise =
          piece = (cd byte slice s idx next_idx) + (cd json esc code (str char at s next_idx))
          n0, n1, n2, n3 = cd sb append piece c0 c1 c2 c3
          cd json esc loop s (next_idx + 1) len n0 n1 n2 n3

fn :: cd json esc loop String Int Int String String String String => String
      cd json esc loop s idx len c0 c1 c2 c3 =
        | idx >= len = cd sb finish c0 c1 c2 c3
        | otherwise  = cd json esc step s idx (cd scan json plain s idx len) len c0 c1 c2 c3

fn :: json escape String => String
      json escape s      =  cd json esc loop s 0 (str len s) "" "" "" ""

fn :: cd json unesc char String Int Int => String
      cd json unesc char s idx nc =
        | nc == 34  = "\""
        | nc == 92  = "\\"
        | nc == 47  = "/"
        | nc == 98  = char from code 8
        | nc == 102 = char from code 12
        | nc == 110 = "\n"
        | nc == 114 = "\r"
        | nc == 116 = "\t"
        | otherwise = str slice s (idx + 1) (idx + 2)

fn :: cd scan no backslash String Int Int => Int
      cd scan no backslash s idx len =
        | idx >= len = len
        | (str char at s idx) == 92 = idx
        | otherwise = cd scan no backslash s (idx + 1) len

fn :: cd json unesc cp Int Int => String Int
      cd json unesc cp next_idx cp =
        | cp <= 255 = (cd raw byte ub cp, next_idx + 6)
        | otherwise = (char from code cp, next_idx + 6)

fn :: cd json unesc u4 Int Int Int Int Int => String Int
      cd json unesc u4 next_idx d0 d1 d2 d3 =
        | d0 < 0 or d1 < 0 or d2 < 0 or d3 < 0 = ("\\u", next_idx + 2)
        | otherwise = cd json unesc cp next_idx ((((d0 * 16) + d1) * 256) + (d2 * 16) + d3)

fn :: cd json unesc step String Int Int Int String String String String => String
      cd json unesc step s idx next_idx len c0 c1 c2 c3 =
        | next_idx >= len =
          | idx == 0  = s
          | otherwise = cd sb finish (c0 + (cd byte slice s idx len)) c1 c2 c3 |.
        | (next_idx + 5) < len and (str char at s (next_idx + 1)) == 117 =
          d0 = cd hex val (str char at s (next_idx + 2))
          d1 = cd hex val (str char at s (next_idx + 3))
          d2 = cd hex val (str char at s (next_idx + 4))
          d3 = cd hex val (str char at s (next_idx + 5))
          rep, adv = cd json unesc u4 next_idx d0 d1 d2 d3
          piece = (cd byte slice s idx next_idx) + rep
          n0, n1, n2, n3 = cd sb append piece c0 c1 c2 c3
          cd json unesc loop s adv len n0 n1 n2 n3
        | (next_idx + 1) < len =
          piece = (cd byte slice s idx next_idx) + (cd json unesc char s next_idx (str char at s (next_idx + 1)))
          n0, n1, n2, n3 = cd sb append piece c0 c1 c2 c3
          cd json unesc loop s (next_idx + 2) len n0 n1 n2 n3
        | otherwise =
          cd sb finish (c0 + (cd byte slice s idx len)) c1 c2 c3

fn :: cd json unesc loop String Int Int String String String String => String
      cd json unesc loop s idx len c0 c1 c2 c3 =
        | idx >= len = cd sb finish c0 c1 c2 c3
        | otherwise  = cd json unesc step s idx (cd scan no backslash s idx len) len c0 c1 c2 c3

fn :: json unescape String => String
      json unescape s      =
        | not (str contains s "\\") = s
        | otherwise                 = cd json unesc loop s 0 (str len s) "" "" "" ""

fn :: json quote String => String
      json quote s      =  "\"" + (json escape s) + "\""

fn :: json int Int => String
      json int n   =  int to str n

fn :: json bool Bool  => String
      json bool True  =  "true"
      json bool False =  "false"

fn :: json null => String
      json null =  "null"

fn :: cd join json loop [String] String String String String => String
      cd join json loop []       c0 c1 c2 c3 =  cd sb finish c0 c1 c2 c3
      cd join json loop (x:rest) c0 c1 c2 c3 =
        n0, n1, n2, n3 = cd sb append ("," + x) c0 c1 c2 c3
        cd join json loop rest n0 n1 n2 n3

fn :: cd join json items [String] => String
      cd join json items []       =  ""
      cd join json items (x:rest) =  cd join json loop rest x "" "" ""

fn :: json array [String] => String
      json array items    =  "[" + (cd join json items items) + "]"

fn :: cd join json pairs loop [String String] String String String String => String
      cd join json pairs loop []          c0 c1 c2 c3 =  cd sb finish c0 c1 c2 c3
      cd join json pairs loop (pair:rest) c0 c1 c2 c3 =
        piece = "," + (json quote (cd pair key pair)) + ":" + (cd pair val pair)
        n0, n1, n2, n3 = cd sb append piece c0 c1 c2 c3
        cd join json pairs loop rest n0 n1 n2 n3

fn :: cd join json pairs [String String] => String
      cd join json pairs []              =  ""
      cd join json pairs (pair:rest)     =
        first = (json quote (cd pair key pair)) + ":" + (cd pair val pair)
        cd join json pairs loop rest first "" "" ""

fn :: json object [String String] => String
      json object pairs           =  "{" + (cd join json pairs pairs) + "}"

fn :: cd quote pair vals [String String] [String String] [String String] [String String] Int => [String String]
      cd quote pair vals []          l0 l1 l2 _   =  l2 :: (l1 :: l0)
      cd quote pair vals (pair:rest) l0 l1 l2 cnt =
        n0, n1, n2, ncnt = cd kv lb append (cd pair key pair, json quote (cd pair val pair)) l0 l1 l2 cnt
        cd quote pair vals rest n0 n1 n2 ncnt

fn :: json str object [String String] => String
      json str object pairs           =  json object (cd quote pair vals pairs [] [] [] 0)

fn :: cd is json ws Int => Bool
      cd is json ws c   =  c == 32 or c == 9 or c == 10 or c == 13 or c == 58

fn :: cd skip key sep String Int Int => Int
      cd skip key sep s idx len =
        | idx >= len                        = idx
        | cd is json ws (str char at s idx) = cd skip key sep s (idx + 1) len
        | otherwise                         = idx

fn :: cd scan json str end String Int Int => Int
      cd scan json str end s idx len =
        | idx >= len                = 0 - 1
        | (str char at s idx) == 92 = cd scan json str end s (idx + 2) len
        | (str char at s idx) == 34 = idx
        | otherwise                 = cd scan json str end s (idx + 1) len

fn :: cd is tok delim Int => Bool
      cd is tok delim c   =  c == 44 or c == 125 or c == 93 or c == 32 or c == 9 or c == 10 or c == 13

fn :: cd scan json tok end String Int Int => Int
      cd scan json tok end s idx len =
        | idx >= len                           = len
        | cd is tok delim (str char at s idx)  = idx
        | otherwise                            = cd scan json tok end s (idx + 1) len

fn :: cd extract json str vend String Int Int => String
      cd extract json str vend doc vstart vend =
        | vend < 0  = ""
        | otherwise = json unescape (str slice doc (vstart + 1) vend)

fn :: cd extract json str vstart String Int => String
      cd extract json str vstart doc vstart =
        | vstart >= (str len doc)        = ""
        | (str char at doc vstart) != 34 = ""
        | otherwise                      = cd extract json str vend doc vstart (cd scan json str end doc (vstart + 1) (str len doc))

fn :: cd extract json str kpos String String Int => String
      cd extract json str kpos doc pat kpos =
        | kpos < 0  = ""
        | otherwise = cd extract json str vstart doc (cd skip key sep doc (kpos + (str len pat)) (str len doc))

fn :: codec json get str String String => String
      codec json get str doc    key    =
        pat = "\"" + (json escape key) + "\""
        cd extract json str kpos doc pat (cd index of doc pat)

fn :: cd is signed int loop String Int Int => Bool
      cd is signed int loop s idx len =
        | idx >= len                                           = True
        | (str char at s idx) < 48 or (str char at s idx) > 57 = False
        | otherwise                                            = cd is signed int loop s (idx + 1) len

fn :: cd is valid int len String Int => Bool
      cd is valid int len s n =
        | n == 0                  = False
        | (str char at s 0) == 45 =
          | n <= 1    = False
          | otherwise = cd is signed int loop s 1 n |.
        | otherwise               = cd is signed int loop s 0 n

fn :: cd is valid int str String => Bool
      cd is valid int str s = cd is valid int len s (str len s)

fn :: cd raw tok vstart String Int => String
      cd raw tok vstart doc vstart =
        | vstart >= (str len doc) = ""
        | otherwise =
          vend = cd scan json tok end doc vstart (str len doc)
          str slice doc vstart vend

fn :: cd raw tok kpos String String Int => String
      cd raw tok kpos doc pat kpos =
        | kpos < 0  = ""
        | otherwise = cd raw tok vstart doc (cd skip key sep doc (kpos + (str len pat)) (str len doc))

fn :: cd raw json token String String => String
      cd raw json token doc key =
        pat = "\"" + (json escape key) + "\""
        cd raw tok kpos doc pat (cd index of doc pat)

fn :: cd parse tok int String Int => Int
      cd parse tok int tok fallback =
        | not (cd is valid int str tok) = fallback
        | otherwise                     = str to int tok

fn :: codec json get int String String Int => Int
      codec json get int doc key fallback = cd parse tok int (cd raw json token doc key) fallback

fn :: cd parse tok bool String Bool => Bool
      cd parse tok bool tok fallback =
        | tok == "true"  = True
        | tok == "false" = False
        | otherwise      = fallback

fn :: codec json get bool String String Bool => Bool
      codec json get bool doc key fallback = cd parse tok bool (cd raw json token doc key) fallback

15.5. Garanties algorísmiques, memòria RMM1/RMM2 i les 5 Regles d'Or de codec

  • Garantia algorísmica i de seguretat: Rebuig determinista d'entrades hexadecimals o Base64 mal formades (retornant "" sense lectures fora de límits) i reversibilitat exacta $decode(encode(x)) == x$.
  • Complexitat de pila $O(1)$ amb TCO: Tots els bucles recursius a src/main/ridgun/lib/codec.rg estan en posició de cua, de manera que ridgunc els compila a salts natius sense creixement de pila.
  • Memòria determinista (RMM1/RMM2): Zero pauses de recol·lector de brossa; totes les estructures intermèdies s'alliberen de forma determinista en sortir de l'àmbit.
  • Verificació contínua: Provada exhaustivament amb stable check i la suite automatitzada src/test/ridgun/test codec.rg.

Les 5 Regles d'Or en programar amb codec

  1. Declara import codec a l'inici del fitxer i aprofita l'equivalència entre espais i guions baixos perquè el teu codi es llegeixi com a prosa.
  2. Alinea les teves canonades de dades amb =: encadena les crides a codec en passos amb nom propi i deixa que stable format alineï les columnes.
  3. Aprofita les funcions totals de codec a les teves guardes |: comprova els valors de retorn o desembolica els ADTs amb pattern matching en lloc d'assumir que l'entrada externa sempre és vàlida.
  4. Evita copiar grans estructures si només necessites consultar un camp: recorda que en RMM2 els arguments d'entrada no incrementen ni decrementen comptadors (+0 / -0).
  5. Combina codec amb test (import test): escriu sempre assercions amb assert equals o assert true per fixar els invariants del teu domini.

Al Capítol 16 passarem a explorar en profunditat la lib json (La lib json — Arbre AST Recursiu RFC 8259, Pretty-Printer i Consultes per Ruta).