test (943fa751) · Commits · jan.koester / fresh

commands/shell/for.cpp

+13 −3

Changes for commands/shell/for.cpp: 13 added lines, 3 removed lines.
Original line number	Diff line number	Diff line
		@@ -46,14 +46,19 @@ namespace fresh { namespace cmd {
		std::string remainder=(argc>1 && argv[1]) ? argv[1] : "";
		std::vector<ShellWord> words=tokenizeShellWords(remainder);

		if(words.size()<2 \|\| words[1].text!="in"){
		// "for VAR; do" / "for VAR do" (no "in" list): the loop runs
		// over the positional parameters, "$@" (the kernel's
		// scripts/setlocalversion uses "for file; do")
		bool overArgs=words.size()>=2 && words[1].text!="in" &&
		(words[1].text==";" \|\| words[1].text=="do");
		if(words.size()<2 \|\| (words[1].text!="in" && !overArgs)){
		std::cerr << "for: expected 'VAR in ...'" << std::endl;
		return 2;
		}

		std::string varname=words[0].text;

		size_t doIdx=findShellKeyword(words,2,words.size(),"do");
		size_t doIdx=findShellKeyword(words,overArgs ? 1 : 2,words.size(),"do");
		if(doIdx==words.size()){
		std::cerr << "for: expected 'do'" << std::endl;
		return 2;
		@@ -65,11 +70,16 @@ namespace fresh { namespace cmd {
		return 2;
		}

		std::vector<std::string> items;
		if(overArgs){
		items=Environment::instance().positionalArgs();
		}else{
		size_t itemsEnd=doIdx;
		if(itemsEnd>2 && words[itemsEnd-1].text==";")
		--itemsEnd;
		std::string itemtext=shellWordsRawText(remainder,words,2,itemsEnd);
		std::vector<std::string> items=tokenizeShellArgs(itemtext);
		items=tokenizeShellArgs(itemtext);
		}

		std::string body=shellWordsRawText(remainder,words,doIdx+1,doneIdx);
		std::string trailing=shellWordsRawText(remainder,words,doneIdx+1,words.size());

src/shell.cpp

+80 −121

Changes for src/shell.cpp: 80 added lines, 121 removed lines.
Original line number	Diff line number	Diff line
		@@ -80,6 +80,70 @@ static bool isValidShellName(const std::string &s){
		return true;
		}


		// The ')' closing a "$(" whose inner text starts at `from` (npos-safe:
		// s.size() when unterminated). Quoted text, backslash escapes, backticks
		// and nested "$(" are skipped, and a case pattern's unbalanced ')'
		// ("case x in a) ...;; esac") does not close anything -- counting bare
		// parens instead ended the kernel's
		// $(ld -v \| sed -e 's/(compatible with [^)]*)//')
		// at the ')' inside the sed program.
		static size_t cmdSubstClose(const std::string &s,size_t from){
		size_t n=s.size(),i=from;
		int depth=1;
		std::vector<int> caseAt; // paren depth at each open "case"
		auto wordAt=[&](size_t k,const char *w){
		size_t l=strlen(w);
		if(s.compare(k,l,w)!=0) return false;
		bool startOk=(k==0) \|\| strchr(" \t\n;&\|(",s[k-1]);
		bool endOk=(k+l>=n) \|\| strchr(" \t\n;&\|)",s[k+l]);
		return startOk && endOk;
		};
		while(i<n){
		char c=s[i];
		if(c=='\\'){ i+=2; continue; }
		if(c=='\''){
		size_t e=s.find('\'',i+1);
		i=(e==std::string::npos) ? n : e+1;
		continue;
		}
		if(c=='"'){
		size_t k=i+1;
		while(k<n && s[k]!='"'){
		if(s[k]=='\\') { k+=2; continue; }
		if(s[k]=='$' && k+1<n && s[k+1]=='('){ k=cmdSubstClose(s,k+2)+1; continue; }
		if(s[k]=='`'){ size_t e=s.find('`',k+1); k=(e==std::string::npos) ? n : e+1; continue; }
		k++;
		}
		i=k+1;
		continue;
		}
		if(c=='`'){
		size_t k=i+1;
		while(k<n && s[k]!='`'){ if(s[k]=='\\') k++; k++; }
		i=k+1;
		continue;
		}
		if(c=='#' && (i==from \|\| strchr(" \t\n;",s[i-1]))){
		size_t e=s.find('\n',i);
		i=(e==std::string::npos) ? n : e;
		continue;
		}
		if(c=='$' && i+1<n && s[i+1]=='('){ i=cmdSubstClose(s,i+2)+1; continue; }
		if(wordAt(i,"case")){ caseAt.push_back(depth); i+=4; continue; }
		if(wordAt(i,"esac") && !caseAt.empty()){ caseAt.pop_back(); i+=4; continue; }
		if(c=='('){ depth++; i++; continue; }
		if(c==')'){
		if(!caseAt.empty() && caseAt.back()==depth){ i++; continue; } // a case pattern
		if(--depth==0) return i;
		i++;
		continue;
		}
		i++;
		}
		return n;
		}

		static void pushGlobExpanded(std::vector<std::string> &tokens,const std::string &word){
		if(word.find_first_of("*?[")==std::string::npos){
		tokens.push_back(word);
		@@ -526,13 +590,7 @@ std::vector<std::string> fresh::tokenizeShellArgs(const std::string &segment){
		++j;
		}else if(cj=='$' && j+1<n && segment[j+1]=='(' &&
		!(j+2<n && segment[j+2]=='(')){
		size_t k=j+2;
		int pdepth=1;
		while(k<n && pdepth>0){
		if(segment[k]=='(') ++pdepth;
		else if(segment[k]==')') --pdepth;
		if(pdepth>0) ++k;
		}
		size_t k=cmdSubstClose(segment,j+2);
		raw+=Shell::instance().runCaptured(segment.substr(j+2,k-(j+2)));
		j=(k<n) ? k+1 : k;
		}else if(cj=='`'){
		@@ -610,13 +668,7 @@ std::vector<std::string> fresh::tokenizeShellArgs(const std::string &segment){
		// as the rest of this tokenizer; a "$(...)" written inside "..."
		// is not expanded, just copied like any other quoted text.
		if(c=='$' && i+1<n && segment[i+1]=='(' && !(i+2<n && segment[i+2]=='(')){
		size_t j=i+2;
		int depth=1;
		while(j<n && depth>0){
		if(segment[j]=='(') ++depth;
		else if(segment[j]==')') --depth;
		if(depth>0) ++j;
		}
		size_t j=cmdSubstClose(segment,i+2);
		std::string inner=segment.substr(i+2,j-(i+2));
		flushPending();
		current+=Shell::instance().runCaptured(inner);
		@@ -767,13 +819,7 @@ std::vector<std::string> fresh::tokenizeShellArgs(const std::string &segment){
		++j;
		}else if(cj=='$' && j+1<n && segment[j+1]=='(' &&
		!(j+2<n && segment[j+2]=='(')){
		size_t k=j+2;
		int pdepth=1;
		while(k<n && pdepth>0){
		if(segment[k]=='(') ++pdepth;
		else if(segment[k]==')') --pdepth;
		if(pdepth>0) ++k;
		}
		size_t k=cmdSubstClose(segment,j+2);
		rawExpr+=Shell::instance().runCaptured(segment.substr(j+2,k-(j+2)));
		j=(k<n) ? k+1 : k;
		}else if(cj=='\\' && j+1<n && segment[j+1]=='`'){
		@@ -827,13 +873,7 @@ std::vector<std::string> fresh::tokenizeShellArgs(const std::string &segment){
		// limit the top-level "$(...)" handling already has.
		if(segment[i]=='$' && i+1<n && segment[i+1]=='(' &&
		!(i+2<n && segment[i+2]=='(')){
		size_t j=i+2;
		int pdepth=1;
		while(j<n && pdepth>0){
		if(segment[j]=='(') ++pdepth;
		else if(segment[j]==')') --pdepth;
		if(pdepth>0) ++j;
		}
		size_t j=cmdSubstClose(segment,i+2);
		std::string inner=segment.substr(i+2,j-(i+2));
		flushRaw();
		dqResult+=Shell::instance().runCaptured(inner);
		@@ -1084,46 +1124,11 @@ std::vector<fresh::ShellWord> fresh::tokenizeShellWords(const std::string &input
		// real ';'-respecting parse of its contents once this
		// whole span reaches it as one piece). Same depth/quote/
		// backtick-aware scan as extractSubshellBody() itself.
		int depth=1;
		++i;
		while(i<n && depth>0){
		char c=input[i];
		if(c=='\'' \|\| c=='"'){
		char q=c;
		++i;
		// Double-quoted: a backslash-escaped quote char
		// (`\"`) is NOT this span's own closing quote --
		// same recognized escape set as needsMoreInput()/
		// tokenizeShellArgs() elsewhere in this file. Single
		// quotes stay fully literal (no escapes at all,
		// matching POSIX): the `else` scan below handles
		// that case unchanged.
		if(q=='"'){
		while(i<n && input[i]!=q){
		if(input[i]=='\\' && i+1<n &&
		(input[i+1]=='"' \|\| input[i+1]=='\\' \|\|
		input[i+1]=='$' \|\| input[i+1]=='`'))
		i+=2;
		else
		++i;
		}
		}else{
		while(i<n && input[i]!=q)
		++i;
		}
		}else if(c=='`'){
		++i;
		while(i<n && input[i]!='`'){
		if(input[i]=='\\' && i+1<n && input[i+1]=='`')
		++i;
		++i;
		}
		}else if(c=='('){
		++depth;
		}else if(c==')'){
		--depth;
		}
		++i;
		// (cmdSubstClose(): quotes, escapes, backticks, nested
		// "$(" and case patterns' lone ')' are all skipped)
		{
		size_t close=cmdSubstClose(input,i+1);
		i=(close<n) ? close+1 : n;
		}
		continue;
		}
		@@ -2003,13 +2008,8 @@ std::vector<std::string> fresh::splitPipelineStages(const std::string &segment,
		// that one specifically needs its own branch) silently split into
		// two broken pipeline stages instead of one command.
		if(c=='$' && i+1<n && segment[i+1]=='('){
		size_t j=i+2;
		int depth=1;
		while(j<n && depth>0){
		if(segment[j]=='(') ++depth;
		else if(segment[j]==')') --depth;
		++j;
		}
		size_t j=cmdSubstClose(segment,i+2);
		if(j<n) ++j;
		current+=segment.substr(i,j-i);
		i=j;
		continue;
		@@ -2169,53 +2169,12 @@ bool fresh::extractSubshellBody(const std::string &stage,std::string &body,std::
		if(i>=n \|\| stage[i]!='(')
		return false;

		// Depth-tracked scan for the matching ')', skipping over quoted and
		// backtick-quoted spans so a '(' or ')' inside one of those (e.g. an
		// embedded "echo ')'") doesn't desync the count -- same approach as
		// splitPipelineStages()'s "$(...)" handling just above.
		size_t j=i+1;
		int depth=1;
		while(j<n && depth>0){
		char c=stage[j];
		if(c=='\'' \|\| c=='"'){
		char q=c;
		++j;
		// Double-quoted: a backslash-escaped quote (`\"`) is NOT this
		// span's own closing quote -- same escape-aware fix already
		// applied to tokenizeShellWords()'s own "(...)" absorption scan
		// (see its comment); without it, a body like
		// `( sed "s/\"//g" )` desyncs this depth count on the escaped
		// quote and returns a truncated/wrong body. Single quotes stay
		// fully literal (POSIX has no escapes inside them at all).
		if(q=='"'){
		while(j<n && stage[j]!=q){
		if(stage[j]=='\\' && j+1<n &&
		(stage[j+1]=='"' \|\| stage[j+1]=='\\' \|\|
		stage[j+1]=='$' \|\| stage[j+1]=='`'))
		j+=2;
		else
		++j;
		}
		}else{
		while(j<n && stage[j]!=q)
		++j;
		}
		}else if(c=='`'){
		++j;
		while(j<n && stage[j]!='`'){
		if(stage[j]=='\\' && j+1<n && stage[j+1]=='`')
		++j;
		++j;
		}
		}else if(c=='('){
		++depth;
		}else if(c==')'){
		--depth;
		}
		++j;
		}
		if(depth>0)
		// the matching ')' (quotes, escapes, backticks, nested "$(" and case
		// patterns skipped -- see cmdSubstClose())
		size_t close=cmdSubstClose(stage,i+1);
		if(close>=n)
		return false; // unterminated "(" -- treat as not a subshell at all
		size_t j=close+1;

		body=stage.substr(i+1,(j-1)-(i+1));
		suffix=stage.substr(j);