Improve parsing errors

Makefile

@@ -9,7 +9,7 @@ BINDIR=.
 NAME = $(shell readlink -f . | xargs basename)
 
 # global links
-LDFLAGS = -lpthread
+LDFLAGS = -Wl,--no-as-needed -lpthread
 
 # compiler
 CC=g++

include/util.hpp

@@ -7,6 +7,8 @@
 #include <exception>
 #include <stdexcept>
 
+#include <ztd/filedat.hpp>
+
 #define INDENT indent(ind)
 
 extern std::string indenting_string;
@@ -36,4 +38,7 @@ void _exec(std::string const& bin, std::vector<std::string> const& args);
 
 std::string stringReplace(std::string subject, const std::string& search, const std::string& replace);
 
+void printFormatError(ztd::format_error const& e, bool print_line=true);
+void printErrorIndex(const char* in, const int index, const std::string& message, const std::string& origin, bool print_line=true);
+
 #endif //UTIL_HPP

src/generate.cpp

@@ -44,6 +44,9 @@ std::string pipeline::generate(int ind)
 {
   std::string ret;
 
+  if(cmds.size()<=0)
+    return "";
+
   ret += cmds[0].generate(ind);
   for(uint32_t i=1 ; i<cmds.size() ; i++)
   {
@@ -57,7 +60,10 @@ std::string pipeline::generate(int ind)
 std::string condlist::generate(int ind)
 {
   std::string ret;
-  if(!opt_minimize) ret += INDENT;
+  if(pls.size() <= 0)
+    return "";
+  if(!opt_minimize)
+    ret += INDENT;
   ret += pls[0].generate(ind);
   for(uint32_t i=0 ; i<pls.size()-1 ; i++)
   {
@@ -114,7 +120,8 @@ std::string generate_resolve(std::vector<std::string> args, int ind)
     dir=pwd();
     std::string cddir=ztd::exec("dirname", g_origin).first;
     cddir.pop_back();
-    chdir(cddir.c_str());
+    if(chdir(cddir.c_str()) != 0)
+      throw std::runtime_error("Cannot cd to '"+cddir+"'");
   }
 
 
@@ -150,7 +157,8 @@ std::string generate_resolve(std::vector<std::string> args, int ind)
   }
 
   if(!opts['C'] && !piped)
-    chdir(dir.c_str());
+    if(chdir(dir.c_str()) != 0)
+      throw std::runtime_error("Cannot cd to '"+dir+"'");
 
   return ret;
 }
@@ -176,7 +184,8 @@ std::string generate_include(std::vector<std::string> args, int ind)
     dir=pwd();
     std::string cddir=ztd::exec("dirname", curfile).first;
     cddir.pop_back();
-    chdir(cddir.c_str());
+    if(chdir(cddir.c_str()) != 0)
+      throw std::runtime_error("Cannot cd to '"+cddir+"'");
   }
 
   // do shell resolution
@@ -199,7 +208,7 @@ std::string generate_include(std::vector<std::string> args, int ind)
         add_include(it) ) // not already included
     {
       file=import_file(it);
-      if(opts['e'])
+      if(opts['d'])
         file = stringReplace(file, "\"", "\\\"");
       if(opts['r'])
         ret += file;
@@ -226,7 +235,8 @@ std::string generate_include(std::vector<std::string> args, int ind)
     }
   }
   if(!opts['C'] && !piped)
-    chdir(dir.c_str());
+    if(chdir(dir.c_str()) != 0)
+      throw std::runtime_error("Cannot cd to '"+dir+"'");
   g_origin=curfile;
 
   if(!opts['r'])
@@ -321,7 +331,7 @@ std::string block::generate(int ind, bool print_shebang)
       // commands
       for(auto it: cls)
         ret += it.generate(ind+1);
-      if(opt_minimize)
+      if(opt_minimize && ret.size()>1)
         ret.pop_back(); // ) can be right after command
       else
         ret += INDENT;

src/main.cpp

@@ -65,7 +65,7 @@ int main(int argc, char* argv[])
   try
   {
     block sh(parse(import_file(file)));
-    if(options['E'])
+    if(options['e'])
     {
       std::string data=sh.generate();
       // generate path
@@ -94,11 +94,13 @@ int main(int argc, char* argv[])
   }
   catch(ztd::format_error& e)
   {
-    printFormatException(e);
+    printFormatError(e);
+    return 100;
   }
   catch(std::exception& e)
   {
     std::cerr << e.what() << std::endl;
+    return 2;
   }
 
 

src/options.cpp

@@ -4,13 +4,12 @@ ztd::option_set options = gen_options();
 bool opt_minimize;
 bool piped=false;
 
-
 ztd::option_set gen_options()
 {
   ztd::option_set ret;
   ret.add(ztd::option('h', "help",     false, "Display this help message"));
   ret.add(ztd::option('m', "minimize", false, "Minimize code"));
-  ret.add(ztd::option('E', "exec",     false, "Directly exec instead of outputting"));
+  ret.add(ztd::option('e', "exec",     false, "Directly exec instead of outputting"));
   ret.add(ztd::option("help-commands", false, "Print help for linker commands"));
   return ret;
 }

src/parse.cpp

@@ -10,7 +10,7 @@ std::string g_origin;
 
 inline bool is_in(char c, const char* set)
 {
-  return index(set, c) != NULL;
+  return strchr(set, c) != NULL;
 }
 
 inline bool is_alphanum(char c)
@@ -38,6 +38,15 @@ bool word_eq(const char* word, const char* in, uint32_t size, uint32_t start, co
   return false;
 }
 
+std::string get_word(const char* in, uint32_t size, uint32_t start, const char* end_set)
+{
+  uint32_t i=start;
+  while(i<size && !is_in(in[i], end_set))
+    i++;
+
+  return std::string(in+start, i-start);
+}
+
 uint32_t skip_chars(const char* in, uint32_t size, uint32_t start, const char* set)
 {
   for(uint32_t i=start; i<size ; i++)
@@ -77,7 +86,11 @@ std::pair<arg, uint32_t> parse_arg(const char* in, uint32_t size, uint32_t start
 {
   arg ret;
   // j : start of subarg
-  uint32_t i=start,j=start;
+  uint32_t i=start,j=start,q=start;
+
+  if(is_in(in[i], "&|;\n#()"))
+    throw ztd::format_error( strf("Unexpected token '%c'", in[i]) , g_origin, in, i);
+
   while(i<size && !is_in(in[i], " \t|&;\n()"))
   {
     if(i+1<size && is_in(in[i], "<>") && in[i+1]=='&') // special case for <& and >&
@@ -93,15 +106,13 @@ std::pair<arg, uint32_t> parse_arg(const char* in, uint32_t size, uint32_t start
     }
     else if(in[i] == '"') // start double quote
     {
+      q=i;
       i++;
-      while(i<size && in[i] != '"') // while inside quoted string
+      while(in[i] != '"') // while inside quoted string
       {
         if(in[i] == '\\') // backslash: don't check next char
         {
-          i++;
+          i+=2;
-          if(i>=size)
-            break;
-          i++;
         }
         else if( word_eq("$(", in, size, i) ) // substitution
         {
@@ -115,18 +126,20 @@ std::pair<arg, uint32_t> parse_arg(const char* in, uint32_t size, uint32_t start
         }
         else
           i++;
-      }
+
         if(i>=size)
-        break;
+          throw ztd::format_error("Unterminated double quote", g_origin, in, q);
+      }
       i++;
     }
     else if(in[i] == '\'') // start single quote
     {
+      q=i;
       i++;
       while(i<size && in[i]!='\'')
         i++;
       if(i>=size)
-        break;
+        throw ztd::format_error("Unterminated single quote", g_origin, in, q);
       i++;
     }
     else if( word_eq("$(", in, size, i) ) // substitution
@@ -155,10 +168,17 @@ std::pair<arg, uint32_t> parse_arg(const char* in, uint32_t size, uint32_t start
 // must start at a read char
 // first char has to be read
 // ends at either &|;\n#()
-std::pair<arglist, uint32_t> parse_arglist(const char* in, uint32_t size, uint32_t start)
+std::pair<arglist, uint32_t> parse_arglist(const char* in, uint32_t size, uint32_t start, bool hard_error=false)
 {
   uint32_t i=start;
   arglist ret;
+  if(is_in(in[i], "&|;\n#(){}"))
+  {
+    if(hard_error)
+      throw ztd::format_error( strf("Unexpected token '%c'", in[i]) , g_origin, in, i);
+    else
+      return std::make_pair(ret, i);
+  }
   while(i<size)
   {
     auto pp=parse_arg(in, size, i);
@@ -178,7 +198,7 @@ std::pair<block, uint32_t> parse_block(const char* in, uint32_t size, uint32_t s
 // ends at either &;\n#)
 std::pair<pipeline, uint32_t> parse_pipeline(const char* in, uint32_t size, uint32_t start)
 {
-  uint32_t i = skip_unread(in, size, start);
+  uint32_t i=start;
   pipeline ret;
   while(i<size)
   {
@@ -237,6 +257,9 @@ std::pair<condlist, uint32_t> parse_condlist(const char* in, uint32_t size, uint
       throw ztd::format_error( strf("Unexpected token: '%c%c'", in[i], in[i+1]), g_origin, in, i);
     else // unknown
       throw ztd::format_error("Unknown error", g_origin, in, i);
+    i = skip_unread(in, size, i);
+    if(i>=size)
+      throw ztd::format_error( "Unexpected end of file", g_origin, in, i );
   }
   return std::make_pair(ret, i);
 }
@@ -248,14 +271,16 @@ std::pair<block, uint32_t> parse_subshell(const char* in, uint32_t size, uint32_
 {
   uint32_t i = skip_unread(in, size, start);
   block ret(block::subshell);
-  while(i<size && in[i] != ')')
+  while(in[i] != ')')
   {
     auto pp=parse_condlist(in, size, i);
     ret.cls.push_back(pp.first);
     i = skip_unread(in, size, pp.second);
-  }
     if(i>=size)
       throw ztd::format_error("Expecting )", g_origin, in, start-1);
+  }
+  if(ret.cls.size()<=0)
+    throw ztd::format_error("Subshell is empty", g_origin, in, start-1);
   i++;
   return std::make_pair(ret,i);
 }
@@ -267,17 +292,18 @@ std::pair<block, uint32_t> parse_brace(const char* in, uint32_t size, uint32_t s
 {
   uint32_t i = skip_unread(in, size, start);
   block ret(block::brace);
-  while(i<size && in[i] != '}')
+  while(in[i] != '}')
   {
     auto pp=parse_condlist(in, size, i);
     ret.cls.push_back(pp.first);
     i = skip_unread(in, size, pp.second);
+    if(i>=size)
+      throw ztd::format_error("Expecting }", g_origin, in, start-1);
     if(is_in(in[i], ")"))
       throw ztd::format_error( strf("Unexpected token: '%c'", in[i]) , g_origin, in, i );
   }
-  if(i>=size)
+  if(ret.cls.size()<=0)
-    throw ztd::format_error("Expecting }", g_origin, in, start-1);
+    throw ztd::format_error("Brace block is empty", g_origin, in, start-1);
-
   i++;
 
   return std::make_pair(ret,i);
@@ -321,7 +347,7 @@ std::pair<block, uint32_t> parse_cmd(const char* in, uint32_t size, uint32_t sta
   }
   else // is a command
   {
-    auto pp=parse_arglist(in, size, start);
+    auto pp=parse_arglist(in, size, start, true);
     ret.args = pp.first;
     i = pp.second;
   }
@@ -335,7 +361,7 @@ std::pair<block, uint32_t> parse_cmd(const char* in, uint32_t size, uint32_t sta
 std::pair<block, uint32_t> parse_case(const char* in, uint32_t size, uint32_t start)
 {
   block ret(block::case_block);
-  uint32_t i=skip_unread(in, size, start);;
+  uint32_t i=skip_chars(in, size, start, " \t");;
 
   // get the treated argument
   auto pa = parse_arg(in, size, i);
@@ -343,10 +369,10 @@ std::pair<block, uint32_t> parse_case(const char* in, uint32_t size, uint32_t st
   i=skip_unread(in, size, pa.second);
 
   // must be an 'in'
-  if(!word_eq("in", in, size, i))
+  if(!word_eq("in", in, size, i, " \t\n"))
   {
-    auto pp=parse_arg(in, size, i);
+    std::string pp=get_word(in, size, i, " \t\n");
-    throw ztd::format_error("Unexpected word: '"+pp.first.raw+"', expecting 'in' after case", g_origin, in, i);
+    throw ztd::format_error("Unexpected word: '"+pp+"', expecting 'in' after case", g_origin, in, i);
   }
 
   i=skip_unread(in, size, i+2);
@@ -357,6 +383,8 @@ std::pair<block, uint32_t> parse_case(const char* in, uint32_t size, uint32_t st
     // toto)
     std::pair<arg, list_t> cc;
     pa = parse_arg(in, size, i);
+    if(pa.first.raw == "")
+      throw ztd::format_error("Empty case value", g_origin, in, i);
     cc.first = pa.first;
     i=skip_unread(in, size, pa.second);
 
@@ -410,6 +438,8 @@ std::pair<block, uint32_t> parse_case(const char* in, uint32_t size, uint32_t st
 std::pair<block, uint32_t> parse_block(const char* in, uint32_t size, uint32_t start)
 {
   uint32_t i = skip_chars(in, size, start, " \n\t");
+  if(i>=size)
+    throw ztd::format_error("Unexpected end of file", g_origin, in, i);
   std::pair<block, uint32_t> ret;
   if(in[i] == '{') // brace block
   {
@@ -431,7 +461,7 @@ std::pair<block, uint32_t> parse_block(const char* in, uint32_t size, uint32_t s
   }
   if(ret.first.args.args.size()<=0)
   {
-    auto pp=parse_arglist(in, size, ret.second); // in case of redirects
+    auto pp=parse_arglist(in, size, ret.second, false); // in case of redirects
     ret.second=pp.second;
     ret.first.args=pp.first;
   }

src/util.cpp

@@ -89,3 +89,49 @@ std::string stringReplace(std::string subject, const std::string& search, const
   }
   return subject;
 }
+
+void printFormatError(ztd::format_error const& e, bool print_line)
+{
+  printErrorIndex(e.data(), e.where(), e.what(), e.origin(), print_line);
+}
+
+std::string repeatString(std::string const& str, uint32_t n)
+{
+  std::string ret;
+  for(uint32_t i=0; i<n; i++)
+    ret += str;
+  return ret;
+}
+
+void printErrorIndex(const char* in, const int index, const std::string& message, const std::string& origin, bool print_line)
+{
+  int i=0, j=0; // j: last newline
+  int line=1; //n: line #
+  int in_size=strlen(in);
+  if(index >= 0)
+  {
+    while(i < in_size && i < index)
+    {
+      if(in[i] == '\n')
+      {
+        line++;
+        j=i+1;
+      }
+      i++;
+    }
+    while(i < in_size && in[i]!='\n')
+    {
+      i++;
+    }
+  }
+  if(origin != "")
+  {
+    fprintf(stderr, "%s:%u:%u: %s\n", origin.c_str(), line, index-j+1, message.c_str());
+    if(print_line)
+    {
+      std::cerr << std::string(in+j, i-j) << std::endl;
+      std::cerr << repeatString(" ", index-j) << '^' << std::endl;
+    }
+  }
+}
+