From de2796508fa69aab0c649b1773871282bc24acbe Mon Sep 17 00:00:00 2001 From: Arkshine Date: Tue, 30 Dec 2014 00:41:29 +0100 Subject: [PATCH] Add string literal concatenation and stringizing operator --- compiler/libpc300/sc2.c | 181 +++++++++++++++++++++++++++++++------ compiler/libpc300/sci18n.c | 3 +- 2 files changed, 155 insertions(+), 29 deletions(-) diff --git a/compiler/libpc300/sc2.c b/compiler/libpc300/sc2.c index 1fcd5aac..77145464 100755 --- a/compiler/libpc300/sc2.c +++ b/compiler/libpc300/sc2.c @@ -36,8 +36,9 @@ #endif /* flags for litchar() */ -#define RAWMODE 1 -#define UTF8MODE 2 +#define RAWMODE 0x1 +#define UTF8MODE 0x2 +#define ISPACKED 0x4 static cell litchar(const unsigned char **lptr,int flags); static void substallpatterns(unsigned char *line,int buffersize); @@ -1493,6 +1494,7 @@ static int substpattern(unsigned char *line,size_t buffersize,char *pattern,char const unsigned char *p,*s,*e; unsigned char *args[10]; int match,arg,len,argsnum=0; + int stringize; memset(args,0,sizeof args); @@ -1590,11 +1592,18 @@ static int substpattern(unsigned char *line,size_t buffersize,char *pattern,char if (match) { /* calculate the length of the substituted string */ for (e=(unsigned char*)substitution,len=0; *e!='\0'; e++) { + if (*e=='#' && *(e+1)=='%' && isdigit(*(e+2))) { + stringize=1; + e++; /* skip '#' */ + } else { + stringize=0; + } /* if */ if (*e=='%' && isdigit(*(e+1)) && argsnum) { arg=*(e+1)-'0'; assert(arg>=0 && arg<=9); + assert(stringize == 0 || stringize == 1); if (args[arg]!=NULL) { - len+=strlen((char*)args[arg]); + len+=strlen((char*)args[arg])+2*stringize; e++; /* skip %, digit is skipped later */ } else { len++; @@ -1610,12 +1619,22 @@ static int substpattern(unsigned char *line,size_t buffersize,char *pattern,char /* substitute pattern */ strdel((char*)line,(int)(s-line)); for (e=(unsigned char*)substitution,s=line; *e!='\0'; e++) { + if (*e=='#' && *(e+1)=='%' && isdigit(*(e+2))) { + stringize=1; + e++; /* skip '#' */ + } else { + stringize=0; + } /* if */ if (*e=='%' && isdigit(*(e+1))) { arg=*(e+1)-'0'; assert(arg>=0 && arg<=9); if (args[arg]!=NULL) { + if (stringize) + strins((char*)s++,"\"",1); strins((char*)s,(char*)args[arg],strlen((char*)args[arg])); s+=strlen((char*)args[arg]); + if (stringize) + strins((char*)s++,"\"",1); e++; /* skip %, digit is skipped later */ } else { strins((char*)s,(char*)e,1); @@ -1703,6 +1722,61 @@ static void substallpatterns(unsigned char *line,int buffersize) } #endif +/* scanplus + * Look for + in the string and (if not there) in the remainder of the file, + * but restore (or keep intact): + * - the current position in the file + * - the comment parsing state + * - the line buffer used by the lexical analyser + * - the active line number and the active file + * + * The function returns 1 if a plus was found and 0 if not + */ +static int scanplus(const unsigned char *lptr) +{ + static void *inpfmark=NULL; + unsigned char *localbuf; + short localcomment,found; + + /* first look for the plus in the remainder of the string */ + while (*lptr<=' ' && *lptr!='\0') + lptr++; + if (lptr[0]=='+') + return 1; + if (*lptr!='\0') + return 0; /* stumbled on something that is not a plus and not white-space */ + + /* the plus was not on the active line, read more lines from the current + * file (but save its position first) + */ + if (inpf==NULL || pc_eofsrc(inpf)) + return 0; /* quick exit: cannot read after EOF */ + if ((localbuf=(unsigned char*)malloc((sLINEMAX+1)*sizeof(unsigned char)))==NULL) + return 0; + inpfmark=pc_getpossrc(inpf); + localcomment=icomment; + + found=0; + /* read from the file, skip preprocessing, but strip off comments */ + while (!found && pc_readsrc(inpf,localbuf,sLINEMAX)!=NULL) { + stripcom(localbuf); + lptr=localbuf; + /* skip white space */ + while (*lptr<=' ' && *lptr!='\0') + lptr++; + if (lptr[0]=='+') + found=1; + else if (*lptr!='\0') + break; /* stumbled on something that is not a plus and not white-space */ + } /* while */ + + /* clean up & reset */ + free(localbuf); + pc_resetsrc(inpf,inpfmark); + icomment=localcomment; + return found; +} + /* preprocess * * Reads a line by readline() into "pline" and performs basic preprocessing: @@ -1974,38 +2048,89 @@ SC_FUNC int lex(cell *lexvalue,char **lexsym) error(220); } /* if */ } /* if */ - } else if (*lptr=='\"' || (*lptr==sc_ctrlchar && *(lptr+1)=='\"')) - { /* unpacked string literal */ + } else if (*lptr=='\"' /* unpacked string literal */ + || (*lptr==sc_ctrlchar && *(lptr+1)=='\"') /* unpacked raw string */ + || (*lptr=='!' && *(lptr+1)=='\"') /* packed string */ + || (*lptr=='!' && *(lptr+1)==sc_ctrlchar && *(lptr+2)=='\"') /* packed raw string */ + || (*lptr==sc_ctrlchar && *(lptr+1)=='!' && *(lptr+2)=='\"')) /* packed raw string */ + { + int stringflags,segmentflags; + unsigned char *cat; if (sLiteralQueueDisabled) { _lextok=tPENDING_STRING; return _lextok; } _lextok=tSTRING; - stringflags= (*lptr==sc_ctrlchar) ? RAWMODE : 0; *lexvalue=_lexval=litidx; - lptr+=1; /* skip double quote */ - if ((stringflags & RAWMODE)!=0) - lptr+=1; /* skip "escape" character too */ - lptr=sc_packstr ? packedstring(lptr,stringflags) : unpackedstring(lptr,stringflags); - if (*lptr=='\"') - lptr+=1; /* skip final quote */ + _lexstr[0]='\0'; + stringflags=-1; /* to mark the first segment */ + for ( ;; ) { + if (*lptr=='!') + segmentflags= (*(lptr+1)==sc_ctrlchar) ? RAWMODE | ISPACKED: ISPACKED; + else if (*lptr==sc_ctrlchar) + segmentflags= (*(lptr+1)=='!') ? RAWMODE | ISPACKED : RAWMODE; + else + segmentflags=0; + if ((segmentflags & ISPACKED)!=0) + lptr+=1; /* skip '!' character */ + if ((segmentflags & RAWMODE)!=0) + lptr+=1; /* skip "escape" character too */ + assert(*lptr=='\"'); + lptr+=1; /* skip double quote too */ + if (stringflags==-1) + stringflags=segmentflags; + else if (stringflags!=segmentflags) + error(238); /* mixing packed/unpacked/raw strings in concatenation */ + cat=strchr(_lexstr,'\0'); + assert(cat!=NULL); + while ((*lptr!='\"' || *(lptr-1)==sc_ctrlchar) && *lptr!='\0' && (cat-_lexstr)