From adec059e95b8e5728d8fb2d8420c5e928812c282 Mon Sep 17 00:00:00 2001 From: Takuji Tanaka Date: Tue, 9 Feb 2016 11:02:48 +0000 Subject: texk/upmendex: upmendex 0.50 git-svn-id: svn://tug.org/texlive/trunk@39638 c570f23f-e606-0410-a88d-b1316a301751 --- Build/source/texk/upmendex/fread.c | 554 +++++++++++++++++++++++++++++++++++++ 1 file changed, 554 insertions(+) create mode 100644 Build/source/texk/upmendex/fread.c (limited to 'Build/source/texk/upmendex/fread.c') diff --git a/Build/source/texk/upmendex/fread.c b/Build/source/texk/upmendex/fread.c new file mode 100644 index 00000000000..22f50ef9eb8 --- /dev/null +++ b/Build/source/texk/upmendex/fread.c @@ -0,0 +1,554 @@ +#include "mendex.h" + +#include + +#include "exkana.h" +#include "exvar.h" + +#define BUFSIZE 65535 + +static int getestr(char *buff, char *estr); +static void chkpageattr(struct page *p); +static void copy_multibyte_char(char *buff1, char *buff2, int *i, int *j); + +/* read idx file */ +int idxread(char *filename, int start) +{ + int i,j,k,l,m,n,cc,indent,wflg,flg=0,bflg=0,nest,esc,quo,eflg=0,pacc,preject; + char buff[BUFSIZE],wbuff[BUFSIZE],estr[256],table[BUFSIZE]; + UChar utable[BUFSIZE],ubuff[BUFSIZE],*tmp1,*tmp2; + FILE *fp; + + pacc=acc; + preject=reject; + + if (filename==NULL) { + fp=stdin; +#ifdef WIN32 + setmode(fileno(fp), _O_BINARY); +#endif + verb_printf(efp, "Scanning input file stdin."); + } + else { + if(kpse_in_name_ok(filename)) + fp=fopen(filename,"rb"); + else + fp=NULL; + if (fp==NULL) { + sprintf(buff,"%s.idx",filename); + if(kpse_in_name_ok(buff)) + fp=fopen(buff,"rb"); + else + fp=NULL; + if (fp==NULL) { + warn_printf(efp,"Warning: Couldn't find input file %s.\n",filename); + return 1; + } + else strcpy(filename,buff); + } + verb_printf(efp,"Scanning input file %s.",filename); + } + + for (i=start,n=1;;i++,n++) { + if (!(i%100)) + ind=(struct index *)realloc(ind,sizeof(struct index)*(i+100)); +LOOP: + ind[i].lnum=n; + if (fgets(buff,sizeof(buff)-1,fp)==NULL) break; + for (j=bflg=cc=0;j0) { + nest--; + wbuff[k]=buff[j]; + continue; + } + if (nest==0) { + if (buff[j]==level) { + esc=0; + if (indent>=2) { + fprintf(efp,"\nError: Extra `%c\' in %s, line %d.",level,filename,ind[i].lnum); + if (efp!=stderr) fprintf(stderr,"\nError: Extra `%c\' in %s, line %d.",level,filename,ind[i].lnum); + eflg++; + reject++; + n++; + goto LOOP; + } + wbuff[k]='\0'; + multibyte_to_widechar(utable,BUFSIZE,wbuff); + ind[i].idx[indent]=u_xstrdup(utable); + if (!wflg) ind[i].org[indent]=NULL; + indent++; + wflg=0; + k= -1; + continue; + } + else if (buff[j]==actual) { + esc=0; + if (wflg) { + fprintf(efp,"\nError: Extra `%c\' in %s, line %d.",actual,filename,ind[i].lnum); + if (efp!=stderr) fprintf(stderr,"\nError: Extra `%c\' in %s, line %d.",actual,filename,ind[i].lnum); + eflg++; + reject++; + n++; + goto LOOP; + } + wbuff[k]='\0'; + multibyte_to_widechar(utable,BUFSIZE,wbuff); + ind[i].org[indent]=u_xstrdup(utable); + wflg=1; + k= -1; + continue; + } + else if ((esc==0 && buff[j]==arg_close) || buff[j]==encap) { + esc=0; + if (buff[j]==encap) { + j++; + cc=getestr(&buff[j],estr); + if (cc<0) { + fprintf(efp,"\nBad encap string in %s, line %d.",filename,ind[i].lnum); + if (efp!=stderr) fprintf(stderr,"\nBad encap string in %s, line %d.",filename,ind[i].lnum); + eflg++; + reject++; + n++; + goto LOOP; + } + j+=cc; + } + else estr[0]='\0'; + + wbuff[k]='\0'; + multibyte_to_widechar(utable,BUFSIZE,wbuff); + ind[i].idx[indent]=u_xstrdup(utable); + if (u_strlen(ind[i].idx[indent])==0) { + if (wflg) { + u_strcpy(ind[i].idx[indent],ind[i].org[indent]); + } + else if (indent>0) { + indent--; + } + else { + verb_printf(efp,"\nWarning: Illegal null field in %s, line %d.",filename,ind[i].lnum); + warn++; + n++; + goto LOOP; + } + } + if (!wflg) { + ind[i].org[indent]=NULL; + } + break; + } + } + if (bcomp==1) { + if (buff[j]==' ' || buff[j]=='\t') { + esc=0; + if (k==0) { + k--; + continue; + } + else if (buff[j+1]==' ' || buff[j+1]=='\t' || buff[j+1]==encap || buff[j+1]==arg_close || buff[j+1]==actual || buff[j+1]==level) { + k--; + continue; + } + else if (buff[j]=='\t') { + wbuff[k]=' '; + continue; + } + } + } + } + else quo=0; + + if (buff[j]!=escape) esc=0; + copy_multibyte_char(buff, wbuff, &j, &k); + } + ind[i].words=indent+1; + +/* kana-convert */ + + for (k=0;kind[l].num) { + ind[l].num++; + if (!((ind[l].num)%16)) ind[l].p=(struct page *)realloc(ind[l].p,sizeof(struct page)*((int)((ind[l].num)/16)+1)*16); + + ind[l].p[ind[l].num].page=xstrdup(table); + + ind[l].p[ind[l].num].enc=xstrdup(estr); + chkpageattr(&ind[l].p[ind[l].num]); + } + } + else { + ind[i].num=0; + ind[i].p=xmalloc(sizeof(struct page)*16); + for (;buff[j]!=arg_open && buff[j]!='\n' && buff[j]!='\0';j++); + if (buff[j]=='\n' || buff[j]=='\0') { + verb_printf(efp,"\nWarning: Missing second argument in %s, line %d.",filename,ind[i].lnum); + acc--; + reject++; + warn++; + n++; + goto LOOP; + } + j++; + for (k=nest=0;;j++,k++) { + if (buff[j]=='\n' || buff[j]=='\0') { + verb_printf(efp,"\nWarning: Incomplete second argument in %s, line %d.",filename,ind[i].lnum); + acc--; + reject++; + warn++; + n++; + goto LOOP; + } + if (buff[j]==arg_open) + nest++; + if (buff[j]==arg_close) { + if (nest==0) { + table[k]='\0'; + ind[i].p[0].page=xstrdup(table); + break; + } + else nest--; + } + copy_multibyte_char(buff, table, &j, &k); + } + ind[l].p[0].enc=xstrdup(estr); + chkpageattr(&ind[i].p[0]); + } + } + } + lines=i; + if (filename != NULL) fclose(fp); + + verb_printf(efp,"...done (%d entries accepted, %d rejected).\n",acc-pacc, reject-preject); + return eflg; +} + +/* pic up encap string */ +static int getestr(char *buff, char *estr) +{ + int i,nest=0; + + for (i=0;i0) { + if ((unsigned char)buff[i-1]<0x80) { + estr[i]=buff[i]; + i++; + } + } + else { + estr[i]=buff[i]; + i++; + } + } + if (nest==0 && buff[i]==arg_close) { + estr[i]='\0'; + return i; + } + if (buff[i]==arg_open) nest++; + else if (buff[i]==arg_close) nest--; + copy_multibyte_char(buff, estr, &i, NULL); + } + + return -1; +} + +static void copy_multibyte_char(char *buff1, char *buff2, int *i, int *j) +{ + int len; + + if ((unsigned char)buff1[*i]<0x80) { + buff2[j ? *j : *i] = buff1[*i]; + return; + } + len = multibytelen((unsigned char)buff1[*i]); + if (len<0) { + verb_printf(efp,"\nWarning: Illegal input of lead byte 0x%x in UTF-8.",(unsigned char)buff1[*i]); + len=1; /* copy one byte when illegal lead byte */ + } + while(len--) { + buff2[j ? (*j)++ : *i] = buff1[*i]; + (*i)++; + } + if (j!=NULL) (*j)--; + (*i)--; +} + +int multibyte_to_widechar(UChar *wcstr, int32_t size, char *mbstr) +{ + int32_t len, olen, nsub; + UErrorCode err; + + len=strlen(mbstr)+1; + do { + err = U_ZERO_ERROR; + len--; + u_strFromUTF8WithSub(NULL, 0, &olen, mbstr, len, 0xFFFD, &nsub, &err); + } while (olen>=size); + err = U_ZERO_ERROR; + u_strFromUTF8WithSub(wcstr, size, &olen, mbstr, len, 0xFFFD, &nsub, &err); + return olen; +} + +int widechar_to_multibyte(char *mbstr, int32_t size, UChar *wcstr) +{ + int32_t len, olen, nsub; + UErrorCode err; + + len=u_strlen(wcstr)+1; + do { + err = U_ZERO_ERROR; + len--; + u_strToUTF8WithSub(NULL, 0, &olen, wcstr, len, 0xFFFD, &nsub, &err); + } while (olen>=size); + err = U_ZERO_ERROR; + u_strToUTF8WithSub(mbstr, size, &olen, wcstr, len, 0xFFFD, &nsub, &err); + return olen; +} + +static void chkpageattr(struct page *p) +{ + int i,j,cc=0; + + for (i=0;ipage);i++) { + if (strncmp(page_compositor,&p->page[i],strlen(page_compositor))==0) { + p->attr[cc]=pattr[cc]; + cc++; + i+=strlen(page_compositor)-1; + } + else { +ATTRLOOP: + if (!((p->page[i]>='0' && p->page[i]<='9') || (p->page[i]>='A' && p->page[i]<='Z') || (p->page[i]>='a' && p->page[i]<='z'))) { + p->attr[cc]= -1; + if (cc<2) p->attr[++cc]= -1; + return; + } + switch(page_precedence[pattr[cc]]) { + case 'r': + if (strchr("ivxlcdm",p->page[i])==NULL) { + pattr[cc]++; + for (j=cc+1;j<3;j++) pattr[j]=0; + goto ATTRLOOP; + } + break; + case 'R': + if (strchr("IVXLCDM",p->page[i])==NULL) { + pattr[cc]++; + for (j=cc+1;j<3;j++) pattr[j]=0; + goto ATTRLOOP; + } + break; + case 'n': + if (p->page[i]<'0' || p->page[i]>'9') { + pattr[cc]++; + for (j=cc+1;j<3;j++) pattr[j]=0; + goto ATTRLOOP; + } + break; + case 'a': + if (p->page[i]<'a' || p->page[i]>'z') { + pattr[cc]++; + for (j=cc+1;j<3;j++) pattr[j]=0; + goto ATTRLOOP; + } + break; + case 'A': + if (p->page[i]<'A' || p->page[i]>'Z') { + pattr[cc]++; + for (j=cc+1;j<3;j++) pattr[j]=0; + goto ATTRLOOP; + } + break; + default: + break; + } + } + } + p->attr[cc]=pattr[cc]; + if (cc<2) p->attr[++cc]= -1; +} -- cgit v1.2.3