selector.c

HTML-XML-utils由一套小型C程序（过滤器）组成

	s++;
	state = ATTR;
      } else errexit("Expected letter instead of \"%c\" after \"[\"",*s);
      break;
    case ATTR:					/* Inside attrib name */
      if (isnmchar(*s)) {strappc(&sel->attribs->name, *s); s++;}
      else state = AFTER_ATTR;
      break;
    case AFTER_ATTR:
      if (isspace(*s)) s++;
      else if (*s == '/') {s++; saved_state = AFTER_ATTR; state = SLASH;}
      else if (*s == ']') {s++; sel->attribs->op = Exists; state = AFTER_TYPE;}
      else if (*s == '~') {s++; sel->attribs->op = Includes; state = EQ;}
      else if (*s == '|') {s++; sel->attribs->op = LangMatch; state=EQ;}
      else if (*s == '^') {s++; sel->attribs->op = StartsWith; state=EQ;}
      else if (*s == '$') {s++; sel->attribs->op = EndsWidth; state=EQ;}
      else if (*s == '*') {s++; sel->attribs->op = Contains; state = EQ;}
      else {sel->attribs->op = Equals; state = EQ;}
      break;
    case EQ:					/* Expect '=' */
      if (*s != '=') errexit("Expected '=' instead of \"%c\"", *s);
      else {s++; sel->attribs->value = newstring(""); state=START_VALUE;}
      break;
    case START_VALUE:				/* After '=' */
      if (isspace(*s)) s++;
      else if (*s == '/') {s++; saved_state = START_VALUE; state=SLASH;}
      else if (*s == '"') {s++; state = DSTRING;}
      else if (*s == '\'') {s++; state = SSTRING;}
      else if (!isnmstart(*s)) errexit("Syntax error at \"%c\"\n", *s);
      else {strappc(&sel->attribs->value, *s); s++; state = VALUE;}
      break;
    case DSTRING:				/* Inside "..." */
      if (*s == '"') {s++; state = AFTER_VALUE;}
      else {strappc(&sel->attribs->value, *s); s++;}
      break;
    case SSTRING:				/* Inside "..." */
      if (*s == '\'') {s++; state = AFTER_VALUE;}
      else {strappc(&sel->attribs->value, *s); s++;}
      break;
    case VALUE:					/* Inside keyword */
      if (isnmchar(*s)) {strappc(&sel->attribs->value, *s); s++;}
      else state = AFTER_VALUE;
      break;
    case AFTER_VALUE:				/* Expect ']' */
      if (isspace(*s)) s++;
      else if (*s == '/') {s++; saved_state = AFTER_VALUE; state = HASH;}
      else if (*s == ']') {s++; state = AFTER_TYPE;}
      else errexit("Expected ']' instead of \"%c\"\n", *s);
      break;
    case START_PSEUDO:				/* After ':' */
      new(pseudosel);
      pseudosel->next = sel->pseudos;
      sel->pseudos = pseudosel;
      if (*s == 'r' || *s == 'R') {s++; state = PSEUDO_R;}
      else if (*s == 'n' || *s == 'N') {s++; state = PSEUDO_N;}
      else if (*s == 'f' || *s == 'F') {s++; state = PSEUDO_F;}
      else if (*s == 'l' || *s == 'L') {s++; state = PSEUDO_L;}
      else errexit("Unknown pseudo-class \":%c...\"\n", *s);
      break;
    case PSEUDO_R:				/* After ':r' */
      if (*s == 'o' || *s == 'O') {s++; state = PSEUDO_RO;}
      else errexit("Unknown pseudo-class \":r%c...\"\n", *s);
      break;
    case PSEUDO_RO:				/* After ':ro' */
      if (*s == 'o' || *s == 'O') {s++; state = PSEUDO_ROO;}
      else errexit("Unknown pseudo-class \":ro%c...\"\n", *s);
      break;
    case PSEUDO_ROO:				/* After ':roo' */
      if (*s == 't' || *s == 'T') {sel->pseudos->type = Root; s++; state=PSEUDO_ROOT;}
      else errexit("Unknown pseudo-class \":roo%c...\"\n", *s);
      break;
    case PSEUDO_ROOT:				/* After ':root' */
      if (!isnmchar(*s)) state = AFTER_TYPE;
      else errexit("Unknown pseudo-class \":root%c...\"\n", *s);
      break;
    case PSEUDO_N:				/* After ':n' */
      if (*s == 't' || *s == 'T') {s++; state = PSEUDO_NT;}
      else errexit("Unknown pseudo-class \":n%c...\"\n", *s);
      break;
    case PSEUDO_NT:				/* After ':nt' */
      if (*s == 'h' || *s == 'H') {s++; state = PSEUDO_NTH;}
      else errexit("Unknown pseudo-class \":nt%c...\"\n", *s);
      break;
    case PSEUDO_NTH:				/* After ':nth' */
      if (*s == '-') {s++; state = PSEUDO_NTH_;}
      else errexit("Unknown pseudo-class \":nth%c...\"\n", *s);
      break;
    case PSEUDO_NTH_:				/* After ':nth_' */
      if (*s == 'c' || *s == 'C') {s++; state = PSEUDO_NTH_C;}
      else if (*s == 'o' || *s == 'O') {s++; state = PSEUDO_NTH_O;}
      else errexit("Unknown pseudo-class \":nth_%c...\"\n", *s);
      break;
    case PSEUDO_NTH_C:				/* After ':nth_c' */
      if (*s == 'h' || *s == 'H') {s++; state = PSEUDO_NTH_CH;}
      else errexit("Unknown pseudo-class \":nth_c%c...\"\n", *s);
      break;
    case PSEUDO_NTH_CH:				/* After ':nth_ch' */
      if (*s == 'i' || *s == 'I') {s++; state = PSEUDO_NTH_CHI;}
      else errexit("Unknown pseudo-class \":nth_ch%c...\"\n", *s);
      break;
    case PSEUDO_NTH_CHI:			/* After ':nth_chi' */
      if (*s == 'l' || *s == 'L') {s++; state = PSEUDO_NTH_CHIL;}
      else errexit("Unknown pseudo-class \":nth_chi%c...\"\n", *s);
      break;
    case PSEUDO_NTH_CHIL:			/* After ':nth_chil' */
      if (*s == 'd' || *s == 'D') {s++; state = PSEUDO_NTH_CHILD;}
      else errexit("Unknown pseudo-class \":nth_chil%c...\"\n", *s);
      break;
    case PSEUDO_NTH_CHILD:			/* After ':nth_child' */
      if (*s == '(') {s++; state = PSEUDO_NTH_CHILD_;}
      else errexit("Unknown pseudo-class \":nth_child%c...\"\n", *s);
      break;
    case PSEUDO_NTH_CHILD_:			/* After ':nth_child(' */
      sel->pseudos->type = NthChild; sel->pseudos->a = 0;
      if (isspace(*s)) s++;
      else if (*s == '/') {s++; saved_state = state; state = SLASH;}
      else if (isdigit(*s)) {saved_state = AFTER_MUL; state = START_INT;}
      else if (*s == '-') {s++; state = PSEUDO__MINUS;}
      else if (*s == 'o' || *s == 'O') {s++; state = PSEUDO__O;}
      else if (*s == 'e' || *s == 'E') {s++; state = PSEUDO__E;}
      else if (*s == 'n' || *s == 'N') {s++; sel->pseudos->a = 1;
        state=AFTER_MUL_N;}
      else errexit("Expected digit after \"nth-child(\" at \"%c\"\n",*s);
      break;
    case PSEUDO__MINUS:				/* After :nth-...(- */
      if (isspace(*s)) s++;
      else if (*s == '/') {s++; saved_state = state; state = SLASH;}
      else if (isdigit(*s)) {saved_state = PSEUDO__NEG; state=START_INT;}
      else if (*s == 'n' || *s == 'N') {s++; sel->pseudos->a = -1;
        state=AFTER_MUL_N;}
      else errexit("Expected digit after \":nth-...(\" at \"%c\"\n",*s);
      break;
    case PSEUDO__NEG:				/* After -<num> */
      sel->pseudos->b = -sel->pseudos->b; state = AFTER_MUL;
      break;
    case PSEUDO__O:				/* After :nth...(o */
      if (*s == 'd' || *s == 'D') {s++; state = PSEUDO__OD;}
      else errexit("Illegal character \"%c\" in \":nth-...(\"\n", *s);
      break;
    case PSEUDO__OD:				/* After :nth...(od */
      if (*s == 'd' || *s == 'D') {s++; state = PSEUDO__ODD;}
      else errexit("Illegal character \"%c\" in \":nth-...(\"\n", *s);
      break;
    case PSEUDO__ODD:				/* After :nth-...(odd */
      if (!isnmchar(*s)) {state = END_PSEUDO;
        sel->pseudos->a = 2; sel->pseudos->b = 1;}
      else errexit("Illegal character \"%c\" in \":nth-...(\"\n", *s);
      break;
    case PSEUDO__E:				/* After :nth-...(e */
      if (*s == 'v' || *s == 'V') {s++; state = PSEUDO__EV;}
      else errexit("Illegal character \"%c\" in \":nth-...(\"\n", *s);
      break;
    case PSEUDO__EV:				/* After :nth-...(ev */
      if (*s == 'e' || *s == 'E') {s++; state = PSEUDO__EVE;}
      else errexit("Illegal character \"%c\" in \":nth-...(\"\n", *s);
      break;
    case PSEUDO__EVE:				/* After :nth-...(eve */
      if (*s == 'n' || *s == 'N') {s++; state = PSEUDO__EVEN;}
      else errexit("Illegal character \"%c\" in \":nth-...(\"\n", *s);
      break;
    case PSEUDO__EVEN:				/* Afte :nth-...(even */
      if (!isnmchar(*s)) {state = END_PSEUDO;
        sel->pseudos->a = 2; sel->pseudos->b = 0;}
      else errexit("Illegal character \"%c\" in \":nth-...(\"\n", *s);
      break;
    case START_INT:				/* At a digit */
      n = *s - '0'; s++; state = INT;
      break;
    case INT:					/* Inside number */
      if (isdigit(*s)) {n = 10 * n + *s - '0'; s++;}
      else {sel->pseudos->b = n; state = saved_state;}
      break;
    case AFTER_MUL:				/* After "...(<num>" */
      if (isspace(*s)) s++;
      else if (*s == '/') {s++; saved_state = state; state = SLASH;}
      else if (*s == ')') {s++; state = AFTER_TYPE;}
      else if (*s == 'n' || *s == 'N') {s++; state = AFTER_MUL_N;
        sel->pseudos->a = sel->pseudos->b; sel->pseudos->b = 0;}
      else errexit("Illegal character \"%c\"\n", *s);
      break;
    case AFTER_MUL_N:				/* After "...(<num>N" */
      if (isspace(*s)) s++;
      else if (*s == '/') {s++; saved_state = state; state = SLASH;}
      else if (*s == '+') {s++; state = AFTER_MUL_NPLUS;}
      else if (*s == ')') {s++; state = AFTER_TYPE;}
      else errexit("Illegal character \"%c\"\n", *s);
      break;
    case AFTER_MUL_NPLUS:			/* After "...(<num>N+" */
      if (isspace(*s)) s++;
      else if (*s == '/') {s++; saved_state = state; state = SLASH;}
      else if (isdigit(*s)) {saved_state = END_PSEUDO; state=START_INT;}
      else errexit("Expected a digit after \"+\" at \"%c\"\n", *s);
      break;
    case END_PSEUDO:				/* Expect ')' */
      if (isspace(*s)) s++;
      else if (*s == '/') {s++; saved_state = END_PSEUDO; state = SLASH;}
      else if (*s == ')') {s++; state = AFTER_TYPE;}
      else errexit("Illegal character \"%c\"\n", *s);
      break;
    case PSEUDO_NTH_O:				/* After ':nth_o' */
      if (*s == 'f' || *s == 'F') {s++; state = PSEUDO_NTH_OF;}
      else errexit("Unknown pseudo-class \":nth_o%c...\"", *s);
      break;
    case PSEUDO_NTH_OF:				/* After ':nth_of' */
      if (*s == '-') {s++; state = PSEUDO_NTH_OF_;}
      else errexit("Unknown pseudo-class \":nth_of%c...\"", *s);
      break;
    case PSEUDO_NTH_OF_:			/* After ':nth_of_' */
      if (*s == 't' || *s == 'T') {s++; state = PSEUDO_NTH_OF_T;}
      else errexit("Unknown pseudo-class \":nth_of_%c...\"", *s);
      break;
    case PSEUDO_NTH_OF_T:			/* After ':nth_of_t' */
      if (*s == 'y' || *s == 'Y') {s++; state = PSEUDO_NTH_OF_TY;}
      else errexit("Unknown pseudo-class \":nth_of_t%c...\"", *s);
      break;
    case PSEUDO_NTH_OF_TY:			/* After ':nth_of_ty' */
      if (*s == 'p' || *s == 'P') {s++; state = PSEUDO_NTH_OF_TYP;}
      else errexit("Unknown pseudo-class \":nth_of_ty%c...\"", *s);
      break;
    case PSEUDO_NTH_OF_TYP:			/* After ':nth_of_typ' */
      if (*s == 'e' || *s == 'E') {s++; state = PSEUDO_NTH_OF_TYPE;}
      else errexit("Unknown pseudo-class \":nth_of_typ%c...\"", *s);
      break;
    case PSEUDO_NTH_OF_TYPE:			/* After :nth_of_type */
      if (*s == '(') {s++; state = PSEUDO_NTH_OF_TYPE_;}
      else errexit("Unknown pseudo-class \":nth_of_type%c...\"", *s);
      break;
    case PSEUDO_NTH_OF_TYPE_:			/* After :nth_of_type( */
      sel->pseudos->type = NthChild; sel->pseudos->a = 0;
      if (isspace(*s)) s++;
      else if (*s == '/') {s++; saved_state = state; state = SLASH;}
      else if (isdigit(*s)) {saved_state = AFTER_MUL; state = START_INT;}
      else if (*s == '-') {s++; state = PSEUDO__MINUS;}
      else if (*s == 'o' || *s=='O') {s++; state = PSEUDO__O;}
      else if (*s == 'e' || *s=='E') {s++; state = PSEUDO__E;}
      else if (*s == 'n' || *s=='N') {s++; sel->pseudos->a=1;state=AFTER_MUL_N;}
      else errexit("Expected digit after \":nth-of_type(\" at \"%c\"\n",*s);
      break;
    case PSEUDO_F:				/* After ':f' */
      if (*s == 'i' || *s == 'I') {s++; state = PSEUDO_FI;}
      else errexit("Unknown pseudo-class \":f%c...\"", *s);
      break;
    case PSEUDO_FI:				/* After ':fi' */
      if (*s == 'r' || *s == 'R') {s++; state = PSEUDO_FIR;}
      else errexit("Unknown pseudo-class \":fi%c...\"", *s);
      break;
    case PSEUDO_FIR:				/* After ':fir' */
      if (*s == 's' || *s == 'S') {s++; state = PSEUDO_FIRS;}
      else errexit("Unknown pseudo-class \":fir%c...\"", *s);
      break;
    case PSEUDO_FIRS:				/* After ':firs' */
      if (*s == 't' || *s == 'T') {s++; state = PSEUDO_FIRST;}
      else errexit("Unknown pseudo-class \":firs%c...\"", *s);
      break;
    case PSEUDO_FIRST:				/* After ':first' */
      if (*s == '-') {s++; state = PSEUDO_FIRST_;}
      else errexit("Unknown pseudo-class \":first%c...\"", *s);
      break;
    case PSEUDO_FIRST_:				/* After ':first_' */
      if (*s == 'c' || *s == 'C') {s++; state = PSEUDO_FIRST_C;}
      else if (*s == 'o') {s++; state = PSEUDO_FIRST_O;}
      else errexit("Unknown pseudo-class \":first_%c...\"", *s);
      break;
    case PSEUDO_FIRST_C:			/* After ':first_c' */
      if (*s == 'h' || *s == 'H') {s++; state = PSEUDO_FIRST_CH;}
      else errexit("Unknown pseudo-class \":first_c%c...\"", *s);
      break;
    case PSEUDO_FIRST_CH:			/* After ':first_ch' */
      if (*s == 'i' || *s == 'I') {s++; state = PSEUDO_FIRST_CHI;}
      else errexit("Unknown pseudo-class \":first_ch%c...\"", *s);
      break;
    case PSEUDO_FIRST_CHI:			/* After ':first_chi' */
      if (*s == 'l' || *s == 'L') {s++; state = PSEUDO_FIRST_CHIL;}
      else errexit("Unknown pseudo-class \":first_chi%c...\"", *s);
      break;
    case PSEUDO_FIRST_CHIL:			/* After ':first_chil' */
      if (*s != 'd' && *s != 'D') errexit("Unknown pseudo-class \":first_chil%c...\"", *s);
      else {sel->pseudos->type = FirstChild; s++; state = PSEUDO_FIRST_CHILD;}
      break;
    case PSEUDO_FIRST_CHILD:			/* After ':first_child' */
      if (!isnmchar(*s)) {state = AFTER_TYPE;}
      else errexit("Unknown pseudo-class \":first_child%c...\"", *s);
      break;
    case PSEUDO_FIRST_O:			/* After ':first_o' */
      if (*s == 'f' || *s == 'F') {s++; state = PSEUDO_FIRST_OF;}
      else errexit("Unknown pseudo-class \":first_o%c...\"", *s);
    case PSEUDO_FIRST_OF:			/* After ':first_of' */
      if (*s == '-') {s++; state = PSEUDO_FIRST_OF_;}
      else errexit("Unknown pseudo-class \":first_of%c...\"", *s);
    case PSEUDO_FIRST_OF_:			/* After ':first_of_' */
      if (*s == 't' || *s == 'T') {s++; state = PSEUDO_FIRST_OF_T;}
      else errexit("Unknown pseudo-class \":first_of_%c...\"", *s);
    case PSEUDO_FIRST_OF_T:			/* After ':first_of_t' */
      if (*s == 'y' || *s == 'Y') {s++; state = PSEUDO_FIRST_OF_TY;}
      else errexit("Unknown pseudo-class \":first_of_t%c...\"", *s);
    case PSEUDO_FIRST_OF_TY:			/* After ':first_of_ty' */
      if (*s == 'p' || *s == 'P') {s++; state = PSEUDO_FIRST_OF_TYP;}
      else errexit("Unknown pseudo-class \":first_of_ty%c...\"", *s);
    case PSEUDO_FIRST_OF_TYP:			/* After ':first_of_typ' */
      if (*s == 'e' || *s == 'E') {s++; state = PSEUDO_FIRST_OF_TYPE;}
      else errexit("Unknown pseudo-class \":first_of_typ%c...\"", *s);
    case PSEUDO_FIRST_OF_TYPE:			/* After ':first_of_type' */
      if (!isnmchar(*s)) {sel->pseudos->type = FirstOfType; state =AFTER_TYPE;}
      else errexit("Unknown pseudo-class \":first_of_type%c...\"", *s);
    case PSEUDO_L:				/* After ':l' */
      if (*s == 'a' || *s == 'A') {s++; state = PSEUDO_LA;}
      else errexit("Unknown pseudo-class \":l%c...\"", *s);
      break;      
    case PSEUDO_LA:				/* After ':la' */
      if (*s == 'n' || *s == 'N') {s++; state = PSEUDO_LAN;}
      else errexit("Unknown pseudo-class \":la%c...\"", *s);
      break;      
    case PSEUDO_LAN:				/* After ':lan' */
      if (*s == 'g' || *s == 'G') {s++; state = PSEUDO_LANG;}
      else errexit("Unknown pseudo-class \":lan%c...\"", *s);
      break;      
    case PSEUDO_LANG:				/* After ':lang' */
      if (*s == '(') {s++; state = PSEUDO_LANG_;}
      else errexit("Unknown pseudo-class \":lang%c...\"", *s);
      break;      
    case PSEUDO_LANG_:				/* After ':lang(' */
      if (isspace(*s)) s++;
      else if (*s == '/') {s++; saved_state = PSEUDO_LANG_; state = SLASH;}
      else if (!isnmstart(*s)) errexit("Incorrect \":lang(\" at \"%c\"\n", *s);
      else {
	sel->pseudos->type = Lang;
	sel->pseudos->s = newstring("");
	strappc(&sel->pseudos->s, *s);
	s++;
	state = LANG;
      }
      break;
    case LANG:					/* Inside a language code */
      if (isnmchar(*s)) {strappc(&sel->pseudos->s, *s); s++;}
      else state = END_PSEUDO;
      break;
    default:
      assert(!"Cannot happen");
    }
  }
  return sel;
}