pcbrf.c

harvest是一个下载html网页得机器人

         mdo  msc  mso  pero rni  tagc tago litc */
da11 []={DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,ER0 ,
         DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 },/*da1*/
da11a[]={MLA_,MLA_,MLA_,MLA_,MLA_,NON_,EE_ ,GET_,RS_ ,FUN_,FUN_,MLA_,NSC_,NOP_,
         MLA_,MLA_,MLA_,MLA_,MLA_,MLA_,MLA_,TER_},

er11 []={DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,ER0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,
         DA0 ,DA0 ,DA0 ,DA0 ,CR0 ,DA0 ,DA0 ,DA0 },/*er1*/
er11a[]={LPR_,LPR_,LPR_,ERX_,LPR_,LPR_,LPR_,GET_,LPR_,LPR_,LPR_,LPR_,LPR_,LPR_,
         LPR_,LPR_,LPR_,LPR_,NOP_,LPR_,LPR_,LPR_},

cr11 []={DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,CR0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,
         DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 },/*cr1*/
cr11a[]={LP2_,CRN_,LP2_,CRA_,LP2_,LP2_,LP2_,GET_,LP2_,LP2_,LP2_,LP2_,LP2_,LP2_,
         LP2_,LP2_,LP2_,LP2_,LP2_,LP2_,LP2_,LP2_},

*litrtab[] = {da11, da11a, er11, er11a, cr11, cr11a};
struct parse pcblitr = {"LITR", lexlms, litrtab, 0, 0, 0, 0};
#undef DA0
#undef ER0
#undef CR0
/* PCBLITV: Literal parse; no references; RS ignored; RE/SPACE sequences
            become single SPACE.  Only minimum data characters allowed.
*/
/* Symbols for state names (end with a number). */
#define LS0     0   /* Leading SPACE or RE found. */
#define VA0     2   /* Valid character found. */
#define SP0     4   /* SPACE/RE sequence begun. */

static UNCH
/*       free num  min  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  
         litc */			  
ls10 []={VA0 ,VA0 ,VA0 ,VA0 ,LS0 ,VA0 ,LS0 ,LS0 ,LS0 ,LS0 ,LS0 ,VA0 ,VA0 ,
         LS0 },/*ls0*/		  
ls10a[]={MLE_,MLA_,MLA_,MLA_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,MLE_,SYS_,SYS_,
         TER_},			  
va10 []={VA0 ,VA0 ,VA0 ,VA0 ,SP0 ,VA0 ,VA0 ,VA0 ,VA0 ,SP0 ,SP0 ,VA0 ,VA0 ,
         LS0 },/*va0*/		  
da10a[]={MLE_,MLA_,MLA_,MLA_,MLA_,SYS_,EOF_,GET_,RS_ ,FUN_,MLE_,SYS_,SYS_,
         TER_},			  
sp10 []={VA0 ,VA0 ,VA0 ,VA0 ,SP0 ,VA0 ,VA0 ,SP0 ,SP0 ,SP0 ,SP0 ,VA0 ,VA0 ,
         LS0 },/*sp0*/		  
sp10a[]={MLE_,MLA_,MLA_,MLA_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,MLE_,SYS_,SYS_,
         RPR_},

*litvtab[] = {ls10, ls10a, va10, da10a, sp10, sp10a};
struct parse pcblitv = {"LITV", lexmin, litvtab, 0, 0, 0, 0};
#undef LS0
#undef VA0
#undef SP0
/* PCBLITT: Tokenized attribute value parse.
*/

/* PCBLITT: Attribute value parse; general and character references;
            function chars are translated.
*/
/* Symbols for state names (end with a number). */
#define SP0	0   /* Ignore spaces */
#define DA0     2   /* Data character */
#define ER0     4   /* ERO found; ignore space */
#define ER1     6   /* ERO found; don't ignore space */
#define CR0     8   /* CRO found (ER0, RNI); ignore space */
#define CR1     10  /* CR0 found; don't ignore space */

int pcblittda = DA0;

static UNCH
/*       free num  min  nms  spc  non  ee   eob  rs   re   sep  cde  nsc  ero
         mdo  msc  mso  pero rni  tagc tago litc */

sp14 []={DA0 ,DA0 ,DA0 ,DA0 ,SP0 ,DA0 ,DA0 ,SP0 ,SP0 ,SP0 ,SP0 ,DA0 ,DA0 ,ER0 ,
         DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 },/*sp0*/
sp14a[]={MLA_,MLA_,MLA_,MLA_,NOP_,NON_,EE_ ,GET_,RS_ ,NOP_,NOP_,MLA_,NSC_,NOP_,
         MLA_,MLA_,MLA_,MLA_,MLA_,MLA_,MLA_,TER_},

da14 []={DA0 ,DA0 ,DA0 ,DA0 ,SP0 ,DA0 ,DA0 ,DA0 ,DA0 ,SP0 ,SP0 ,DA0 ,DA0 ,ER1 ,
         DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,SP0 },/*da0*/
da14a[]={MLA_,MLA_,MLA_,MLA_,MLA_,NON_,EE_ ,GET_,RS_ ,FUN_,FUN_,MLA_,NSC_,NOP_,
         MLA_,MLA_,MLA_,MLA_,MLA_,MLA_,MLA_,TER_},

er14 []={DA0 ,DA0 ,DA0 ,SP0 ,DA0 ,DA0 ,DA0 ,ER0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,
         DA0 ,DA0 ,DA0 ,DA0 ,CR0 ,DA0 ,DA0 ,DA0 },/*er0*/
er14a[]={LPR_,LPR_,LPR_,ERX_,LPR_,LPR_,LPR_,GET_,LPR_,LPR_,LPR_,LPR_,LPR_,LPR_,
         LPR_,LPR_,LPR_,LPR_,NOP_,LPR_,LPR_,LPR_},

er15 []={DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,ER1 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,
         DA0 ,DA0 ,DA0 ,DA0 ,CR1 ,DA0 ,DA0 ,DA0 },/*er1*/
er15a[]={LPR_,LPR_,LPR_,ERX_,LPR_,LPR_,LPR_,GET_,LPR_,LPR_,LPR_,LPR_,LPR_,LPR_,
         LPR_,LPR_,LPR_,LPR_,NOP_,LPR_,LPR_,LPR_},

cr14 []={DA0 ,DA0 ,DA0 ,SP0 ,DA0 ,DA0 ,DA0 ,CR0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,
         DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 },/*cr0*/
cr14a[]={LP2_,CRN_,LP2_,CRA_,LP2_,LP2_,LP2_,GET_,LP2_,LP2_,LP2_,LP2_,LP2_,LP2_,
         LP2_,LP2_,LP2_,LP2_,LP2_,LP2_,LP2_,LP2_},

cr15 []={DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,CR1 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,
         DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 ,DA0 },/*cr1*/
cr15a[]={LP2_,CRN_,LP2_,CRA_,LP2_,LP2_,LP2_,GET_,LP2_,LP2_,LP2_,LP2_,LP2_,LP2_,
         LP2_,LP2_,LP2_,LP2_,LP2_,LP2_,LP2_,LP2_},

*litttab[] = {sp14, sp14a, da14, da14a, er14, er14a, er15, er15a, cr14, cr14a,
	      cr15, cr15a};
struct parse pcblitt = {"LITT", lexlms, litttab, 0, 0, 0, 0};
#undef SP0
#undef DA0
#undef ER0
#undef ER1
#undef CR0
#undef CR1
/* PCBMD: State and action table for markup declaration tokenization.
          Columns are based on LEXMARK.C.
*/
/* Symbols for state names (end with a number). */
#define SP1     0   /* Separator before token expected (but not -). */
#define SP2     2   /* Separator before token expected. */
#define TK1     4   /* Token expected. */
#define CM1     6   /* COM[1] found: possible comment, MGRP, or minus.*/
#define CM2     8   /* COM[2] found; in comment. */
#define CM3    10   /* Ending COM[1] found; end comment or continue it. */
#define PR1    12   /* PERO found when token was expected. */
#define PX1    14   /* PLUS found: PGRP or error. */
#define RN1    16   /* RNI found; possible reserved name start. */

int pcbmdtk = TK1;            /* PCBMD: token expected. */

static UNCH
/*       bit  nmc  num  nms  spc  non  ee   eob  rs   com  eti  grpo lit  lita
         dso  dsc  pero plus refc rni  tagc tago vi    */
sp21 []={SP1 ,SP1 ,SP1 ,SP1 ,TK1 ,SP1 ,TK1 ,SP1 ,TK1 ,SP1 ,SP1 ,TK1 ,TK1 ,TK1 ,
         TK1 ,SP1 ,PR1 ,PX1 ,SP1 ,RN1 ,SP1 ,SP1 ,SP1 },
sp21a[]={INV_,LEN_,LEN_,LEN_,NOP_,SYS_,EE_ ,GET_,RS_ ,LEN_,INV_,GRPS,LIT ,LITE,
         MDS ,INV_,NOP_,NOP_,INV_,NOP_,EMD ,INV_,INV_},

sp22 []={SP2 ,SP2 ,SP2 ,SP2 ,TK1 ,SP2 ,TK1 ,SP2 ,TK1 ,CM1 ,SP2 ,TK1 ,TK1 ,TK1 ,
         TK1 ,SP2 ,PR1 ,PX1 ,SP2 ,RN1 ,SP2 ,SP2 ,SP2 },
sp22a[]={INV_,LEN_,LEN_,LEN_,NOP_,SYS_,EE_ ,GET_,RS_ ,NOP_,INV_,GRPS,LIT ,LITE,
         MDS ,INV_,NOP_,NOP_,INV_,NOP_,EMD ,INV_,INV_},

tk21 []={SP1 ,SP1 ,SP2 ,SP1 ,TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,CM1 ,SP1 ,TK1 ,TK1 ,TK1 ,
         TK1 ,SP1 ,PR1 ,PX1 ,SP1 ,RN1 ,SP1 ,SP1 ,SP1 },
tk21a[]={INV_,NMT ,NUM ,NAS ,NOP_,SYS_,EE_ ,GET_,RS_ ,NOP_,INV_,GRPS,LIT ,LITE,
         MDS ,INV_,NOP_,NOP_,INV_,NOP_,EMD ,INV_,INV_},

/*       bit  nmc  num  nms  spc  non  ee   eob  rs   com  eti  grpo lit  lita
         dso  dsc  pero plus refc rni  tagc tago vi    */

cm21 []={TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,CM1 ,TK1 ,CM1 ,TK1 ,CM2 ,TK1 ,TK1 ,TK1 ,TK1 ,
         TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,TK1 },
cm21a[]={CDR ,CDR ,CDR ,CDR ,CDR ,SYS_,CDR ,GET_,CDR ,NOP_,CDR ,MGRP,CDR ,CDR ,
         CDR ,CDR ,CDR ,CDR ,CDR ,CDR ,CDR ,CDR ,CDR },

cm22 []={CM2 ,CM2 ,CM2 ,CM2 ,CM2 ,CM2 ,TK1 ,CM2 ,CM2 ,CM3 ,CM2 ,CM2 ,CM2 ,CM2 ,
         CM2 ,CM2 ,CM2 ,CM2 ,CM2 ,CM2 ,CM2 ,CM2 ,CM2 },
cm22a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,NOP_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

/*       bit  nmc  num  nms  spc  non  ee   eob  rs   com  eti  grpo lit  lita
         dso  dsc  pero plus refc rni  tagc tago vi    */
cm23 []={CM2 ,CM2 ,CM2 ,CM2 ,CM2 ,CM3 ,TK1 ,CM3 ,CM2 ,TK1 ,CM2 ,CM2 ,CM2 ,CM2 ,
         CM2 ,CM2 ,CM2 ,CM2 ,CM2 ,CM2 ,CM2 ,CM2 ,CM2 },
cm23a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,NOP_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

pr21 []={SP1 ,SP1 ,SP1 ,TK1 ,TK1 ,PR1 ,SP2 ,PR1 ,TK1 ,SP2 ,SP1 ,SP1 ,SP1 ,SP1 ,
         SP1 ,SP1 ,SP2 ,SP1 ,SP1 ,TK1 ,SP1 ,SP1 ,SP1 },
pr21a[]={PCI_,PCI_,PCI_,PER_,PEN ,SYS_,PENR,GET_,PEN ,PENR,PCI_,PCI_,PCI_,PCI_,
         PCI_,PCI_,PENR,PCI_,PCI_,PCI_,PCI_,PCI_,PCI_},

px21 []={SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,PX1 ,SP1 ,PX1 ,SP1 ,SP1 ,SP1 ,TK1 ,SP1 ,SP1 ,
         SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 ,SP1 },
px21a[]={PCI_,PCI_,PCI_,PCI_,PCI_,SYS_,PCI_,GET_,PCI_,PCI_,PCI_,PGRP,PCI_,PCI_,
         PCI_,PCI_,PCI_,PCI_,PCI_,PCI_,PCI_,PCI_,PCI_},

rn21 []={TK1 ,TK1 ,TK1 ,SP1 ,TK1 ,RN1 ,TK1 ,RN1 ,TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,
         TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,TK1 ,TK1 },
rn21a[]={PCI_,PCI_,PCI_,RNS ,PCI_,SYS_,PCI_,GET_,PCI_,PCI_,PCI_,PCI_,PCI_,PCI_,
         PCI_,PCI_,PCI_,PCI_,PCI_,PCI_,PCI_,PCI_,PCI_},

*mdtab[] = {sp21, sp21a, sp22, sp22a, tk21, tk21a, cm21, cm21a, cm22, cm22a,
            cm23, cm23a, pr21, pr21a, px21, px21a, rn21, rn21a};
struct parse pcbmd = {"MD", lexmark, mdtab, 0, 0, 0, 0};
#undef SP1
#undef SP2
#undef TK1
#undef CM1
#undef CM2
#undef CM3
#undef PR1
#undef PX1
#undef RN1
/* PCBMDC: State and action table for comment declaration.
*/
/* Symbols for state names (end with a number). */
#define CD2     0   /* COM[2] found; in comment. */
#define CD3     2   /* Ending COM[1] found; end comment or continue it. */
#define EM1     4   /* Ending COM[2] found; start new comment or end. */
#define CD1     6   /* COM[1] found; new comment or error. */

static UNCH
/*      bit  nmc  num  nms  spc  non  ee   eob  rs   com  eti  grpo lit  lita
        dso  dsc  pero plus refc rni  tagc tago vi    */
cd22 []={CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD3 ,CD2 ,CD2 ,CD2 ,CD2 ,
         CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 },
cd22a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,NOP_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

cd23 []={CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD3 ,CD2 ,CD3 ,CD2 ,EM1 ,CD2 ,CD2 ,CD2 ,CD2 ,
         CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 },
cd23a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,NOP_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},

em21 []={CD2 ,CD2 ,CD2 ,CD2 ,EM1 ,EM1 ,CD2 ,EM1 ,EM1 ,CD1 ,CD2 ,CD2 ,CD2 ,CD2 ,
         CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 },
em21a[]={INV_,INV_,INV_,INV_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,INV_,INV_,INV_,INV_,
         INV_,INV_,INV_,INV_,INV_,INV_,EMD ,INV_,INV_},

cd21 []={CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD1 ,CD2 ,CD1 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,
         CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 ,CD2 },
cd21a[]={PCI_,PCI_,PCI_,PCI_,PCI_,SYS_,EOF_,GET_,PCI_,NOP_,PCI_,PCI_,PCI_,PCI_,
         PCI_,PCI_,PCI_,PCI_,PCI_,PCI_,PCI_,PCI_,PCI_},

*mdctab[] = {cd22, cd22a, cd23, cd23a, em21, em21a, cd21, cd21a};
struct parse pcbmdc = {"MDC", lexmark, mdctab, 0, 0, 0, 0};
#undef CD2
#undef CD3
#undef EM1
#undef CD1
/* PCBMDI: State and action table for ignoring markup declarations.
           Literals are handled properly so a TAGC won't end the declaration.
           An error is noted if the entity ends within a declaration that
           is being ignored.
           TO DO: Handle nested declaration sets.
*/
/* Symbols for state names (end with a number). */
#define NC1     0   /* Not in a comment; TAGC ends declaration. */
#define IC1     2   /* COM[1] found; possible comment. */
#define IC2     4   /* COM[2] found; in comment. */
#define IC3     6   /* Ending COM[1] found; end comment or continue it. */
#define LI1     8   /* Literal parameter; search for LIT. */
#define LA1    10   /* Literal parameter; search for LITA. */

static UNCH
/*       bit  nmc  num  nms  spc  non  ee   eob  rs   com  eti  grpo lit  lita
         dso  dsc  pero plus refc rni  tagc tago vi    */
nc21 []={NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,IC1 ,NC1 ,NC1 ,LI1 ,LA1 ,
         NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,NC1 },
nc21a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,NOP_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,EMD ,NOP_,NOP_},

ic21 []={NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,IC1 ,NC1 ,IC1 ,NC1 ,IC2 ,NC1 ,NC1 ,LI1 ,LA1 ,
         NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,NC1 ,NC1 },
ic21a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,NOP_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,EMD ,NOP_,NOP_},

ic22 []={IC2 ,IC2 ,IC2 ,IC2 ,IC2 ,IC2 ,NC1 ,IC2 ,IC2 ,IC3 ,IC2 ,IC2 ,IC2 ,IC2 ,
         IC2 ,IC2 ,IC2 ,IC2 ,IC2 ,IC2 ,IC2 ,IC2 ,IC2 },
ic22a[]={NOP_,NOP_,NOP_,NOP_,NOP_,SYS_,EOF_,GET_,RS_ ,NOP_,NOP_,NOP_,NOP_,NOP_,
         NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_,NOP_},