Use Byte for data from input source.

7 files changed, 78 insertions, 71 deletions

diff --git a/src/include/storage.H b/src/include/storage.H
index b75ced9d..cd3a723a 100644
--- a/src/include/storage.H
+++ b/src/include/storage.H
@@ -46,6 +46,9 @@
 #include <string>
 
 namespace OpenAxiom {
+   // Datatype for the unit of storage.
+   using Byte = unsigned char;
+
    // -----------------
    // -- SystemError --
    // -----------------
@@ -64,9 +67,6 @@ namespace OpenAxiom {
    void filesystem_error(const std::string&);
    
    namespace Memory {
-      // Datatype for the unit of storage.
-      using Byte = unsigned char;
-
       // Datatype for pointers to data.
       using Pointer = void*;
 
@@ -302,11 +302,11 @@ namespace OpenAxiom {
          explicit FileMapping(std::string);
          FileMapping(FileMapping&&);
          ~FileMapping();
-         const char* begin() const { return static_cast<const char*>(start); }
-         const char* end() const { return begin() + extent; }
+         const Byte* begin() const { return start; }
+         const Byte* end() const { return begin() + extent; }
          std::size_t size() const { return extent; }
       protected:
-         Pointer start;         // address at the mapped storage
+         Byte* start;           // address at the mapped storage
          size_t extent;         // length (in bytes) of the storage
       private:
          FileMapping(const FileMapping&) = delete;
diff --git a/src/utils/hammer.cc b/src/utils/hammer.cc
index 3ab56c72..eb07b378 100644
--- a/src/utils/hammer.cc
+++ b/src/utils/hammer.cc
@@ -1,4 +1,4 @@
-// Copyright (C) 2010, Gabriel Dos Reis.
+// Copyright (C) 2013, Gabriel Dos Reis.
 // All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
@@ -65,13 +65,13 @@ namespace OpenAxiom {
       // ---------------
       // Plain text, with no reference to any chunk.  
       struct BasicText : Element {
-         BasicText(const char* f, const char* l) : span(f, l) { }
+         BasicText(const Byte* f, const Byte* l) : span(f, l) { }
          // Pointer to the start of this basic text element
-         const char* begin() const { return span.first; }
+         const Byte* begin() const { return span.first; }
          // Oone-past-the-end of the this basic text element.
-         const char* end() const { return span.second; }
+         const Byte* end() const { return span.second; }
       private:
-         std::pair<const char*, const char*> span;
+         std::pair<const Byte*, const Byte*> span;
       };
 
       // ---------------
@@ -100,7 +100,7 @@ namespace OpenAxiom {
 
          // Augment this chunk with a basic text in the open interval
          // [f,l).
-         CompositeText& add_text(const char* f, const char* l) {
+         CompositeText& add_text(const Byte* f, const Byte* l) {
             texts.push_back(BasicText(f, l));
             push_back(&texts.back());
             return *this;
@@ -109,7 +109,7 @@ namespace OpenAxiom {
          // Augment this chunk with a reference to another chunk
          // named `n'.  Note that we don't attempt to check for
          // possible circularities.
-         CompositeText reference_chunk(const char* f, const char* l) {
+         CompositeText reference_chunk(const Byte* f, const Byte* l) {
             refs.push_back(Reference(std::string(f, l)));
             push_back(&refs.back());
             return *this;
@@ -142,11 +142,11 @@ namespace OpenAxiom {
          CompositeText prose;         // the prose around the chunks.
          ChunkTable defs;             // chunk definition table.
          CompositeText* active_chunk; // chunk under construction.
-         const char* text_start;      // begining of current basic text.
+         const Byte* text_start;      // begining of current basic text.
 
          // Append basic text in the range `[text_start,last)'
          // to the current chunk.
-         void finish_chunk(const char* last) {
+         void finish_chunk(const Byte* last) {
             if (text_start != last)
                active_chunk->add_text(text_start, last);
             active_chunk = &prose;
@@ -154,7 +154,7 @@ namespace OpenAxiom {
          }
 
          // Start a new chunk or extend an existing chunk.
-         void begin_chunk(const std::string& name, const char* start) {
+         void begin_chunk(const std::string& name, const Byte* start) {
             if (CompositeText* chunk = lookup_chunk(name))
                active_chunk = chunk;
             else {
@@ -177,7 +177,7 @@ namespace OpenAxiom {
       // Attempt to advance the cursor past newline marker.
       // Return true on sucess.
       static bool
-      saw_newline(const char*& cur, const char* end) {
+      saw_newline(const Byte*& cur, const Byte* end) {
          if (*cur == '\n') {
             ++cur;
             return true;
@@ -193,7 +193,7 @@ namespace OpenAxiom {
       // Move `cur' to end of line or `end', whichever comes first.
       // Return true if the area swept consisted only of blank characters.
       static inline bool
-      trailing_blank(const char*& cur, const char* end) {
+      trailing_blank(const Byte*& cur, const Byte* end) {
          bool result = true;
          for (; cur < end and not saw_newline(cur, end); ++cur)
             result = isspace(*cur);
@@ -203,7 +203,7 @@ namespace OpenAxiom {
       // Attempt to advance `cur' past the double left angle brackets
       // starting a chunk name.  Returm true on success.
       static bool
-      chunk_name_began(const char*& cur, const char* end) {
+      chunk_name_began(const Byte*& cur, const Byte* end) {
          if (cur[0] == '<' and cur + 1 < end and cur[1] == '<') {
             cur += 2;
             return true;
@@ -214,7 +214,7 @@ namespace OpenAxiom {
       // Attempt to move `cur' past the double right angle brackets
       // terminating a chunk name.  Returm true on success.
       static bool
-      chunk_name_ended(const char*& cur, const char* end) {
+      chunk_name_ended(const Byte*& cur, const Byte* end) {
          if (cur[0] == '>' and cur + 1 < end and cur[1] == '>') {
             cur += 2;
             return true;
@@ -225,7 +225,7 @@ namespace OpenAxiom {
       // We've just seen the start of a chunk reference; skip
       // characters till we seen of the chunk's name.
       static void
-      skip_to_end_of_chunk_name(const char*& cur, const char* end) {
+      skip_to_end_of_chunk_name(const Byte*& cur, const Byte* end) {
          while (cur < end) {
             if (looking_at_newline(*cur)
                 or (cur + 1 < end and cur[0] == '>' and cur[1] == '>'))
@@ -236,7 +236,7 @@ namespace OpenAxiom {
 
       // Move the cursor until end of line.
       static void
-      skip_to_end_of_line(const char*& cur, const char* end) {
+      skip_to_end_of_line(const Byte*& cur, const Byte* end) {
          while (cur < end) {
             if (saw_newline(cur, end))
                break;
@@ -246,22 +246,22 @@ namespace OpenAxiom {
       
       void
       Document::parse(const Memory::FileMapping& file) {
-         const char* cur = text_start;
-         const char* last = file.end();
+         auto cur = text_start;
+         auto last = file.end();
          // Process one line at a time.
          while (cur < last) {
             // 1. `@' ends previous chunk
             if (*cur == '@') {
-               const char* p = cur;
+               auto p = cur;
                if (trailing_blank(++cur, last))
                   finish_chunk(p);
             }
             // 2. `<<' introduces a chunk reference or a chunk definition.
             else if (chunk_name_began(cur, last)) {
-               const char* label_start = cur;
+               auto label_start = cur;
                skip_to_end_of_chunk_name(cur, last);
                if (chunk_name_ended(cur, last)) {
-                  const char* label_end = cur - 2;
+                  auto label_end = cur - 2;
                   if (cur < last and *cur == '=') {
                      if (trailing_blank(++cur, last)) {
                         // chunk definition or extension
diff --git a/src/utils/sexpr.H b/src/utils/sexpr.H
index 7b62a328..358dd506 100644
--- a/src/utils/sexpr.H
+++ b/src/utils/sexpr.H
@@ -111,8 +111,8 @@ namespace OpenAxiom {
          Lexer(StringPool& pool, std::vector<Token>& toks)
                : strings(pool), tokens(toks) { }
 
-         const char* tokenize(const char*, const char*);
-         BasicString intern(const char* s, size_t n) {
+         const Byte* tokenize(const Byte*, const Byte*);
+         BasicString intern(const Byte* s, size_t n) {
             return strings.intern(s, n);
          }
 
diff --git a/src/utils/sexpr.cc b/src/utils/sexpr.cc
index 9e57765f..77ed3060 100644
--- a/src/utils/sexpr.cc
+++ b/src/utils/sexpr.cc
@@ -122,16 +122,16 @@ namespace OpenAxiom {
 
       // Move `cur' past all consecutive blank characters, and
       // return the new position.
-      static const char* 
-      skip_blank(const char*& cur, const char* end) {
+      static const Byte* 
+      skip_blank(const Byte*& cur, const Byte* end) {
          while (cur < end and is_blank(*cur))
             ++cur;
          return cur;
       }
 
       // Move `cur' to end-of-line marker.
-      static const char*
-      skip_to_eol(const char*& cur, const char* end) {
+      static const Byte*
+      skip_to_eol(const Byte*& cur, const Byte* end) {
          // FIXME: properly handle CR+LF.
          while (cur < end and *cur != '\n')
             ++cur;
@@ -139,8 +139,8 @@ namespace OpenAxiom {
       }
 
       // Move `cur' until a word boundary is reached.
-      static const char*
-      skip_to_word_boundary(const char*& cur, const char* end) {
+      static const Byte*
+      skip_to_word_boundary(const Byte*& cur, const Byte* end) {
          bool saw_escape = false;
          for (; cur < end; ++cur) {
             if (saw_escape)
@@ -156,7 +156,7 @@ namespace OpenAxiom {
       // Move `cur' one-past a non-esacaped character `c'.
       // Return true if the character was seen.
       static bool
-      skip_to_nonescaped_char(const char*& cur, const char* end, char c) {
+      skip_to_nonescaped_char(const Byte*& cur, const Byte* end, char c) {
          bool saw_escape = false;
          for (; cur < end; ++cur)
             if (saw_escape)
@@ -173,7 +173,7 @@ namespace OpenAxiom {
       // Move `cur' past the closing quote of string literal.
       // Return true if the closing fence was effectively seen.
       static inline bool
-      skip_to_quote(const char*& cur, const char* end) {
+      skip_to_quote(const Byte*& cur, const Byte* end) {
          return skip_to_nonescaped_char(cur, end, '"');
       }
 
@@ -204,7 +204,7 @@ namespace OpenAxiom {
       // an integer followrd by the equal sign or the sharp sign.
       // `cur' is moved along the way.
       static bool
-      only_digits_before_equal_or_shap(const char*& cur, const char* end) {
+      only_digits_before_equal_or_shap(const Byte*& cur, const Byte* end) {
          while (cur < end and isdigit(*cur))
             ++cur;
          return cur < end and (*cur == '#' or *cur == '=');
@@ -215,8 +215,8 @@ namespace OpenAxiom {
       // entirely of digits.
       static void
       maybe_reclassify(Token& t) {
-         const char* cur = t.lexeme->begin();
-         const char* end = t.lexeme->end();
+         const Byte* cur = t.lexeme->begin();
+         const Byte* end = t.lexeme->end();
          while (cur < end and isdigit(*cur))
             ++cur;
          if (cur == end)
@@ -226,7 +226,7 @@ namespace OpenAxiom {
       // Returns true if the first characters in the range
       // [cur, last) start an identifier.
       static bool
-      start_symbol(const char* cur, const char* last) {
+      start_symbol(const Byte* cur, const Byte* last) {
          if (cur >= last)
             return false;
          return identifier_part(*cur)
@@ -236,7 +236,7 @@ namespace OpenAxiom {
       // We are processing a symbol token.  Accumulate all
       // legitimate characters till the end of the token.
       static void
-      skip_to_end_of_symbol(const char*& cur, const char* end) {
+      skip_to_end_of_symbol(const Byte*& cur, const Byte* end) {
          const char c = *cur;
          if (*cur == '|') 
             skip_to_nonescaped_char(++cur, end, c);
@@ -247,22 +247,22 @@ namespace OpenAxiom {
       }
 
       static Token
-      match_maybe_symbol(Lexer* lexer, const char*& cur, const char* end) {
+      match_maybe_symbol(Lexer* lexer, const Byte*& cur, const Byte* end) {
          Token t = { Token::identifier, 0 };
-         const char* start = cur;
+         const Byte* start = cur;
          skip_to_end_of_symbol(cur, end);
          t.lexeme = lexer->intern(start, cur - start);
          maybe_reclassify(t);
          return t;
       }
 
-      const char*
-      Lexer::tokenize(const char* cur, const char* end) {
+      const Byte*
+      Lexer::tokenize(const Byte* cur, const Byte* end) {
          while (skip_blank(cur, end) < end) {
             Token t = { Token::unknown, 0 };
             switch (*cur) {
             case ';': {
-               const char* start = cur;
+               const Byte* start = cur;
                t.type = Token::semicolon;
                skip_to_eol(cur, end);
                t.lexeme = intern(start, cur - start);
@@ -276,7 +276,7 @@ namespace OpenAxiom {
                break;
 
             case ',': {
-               const char* start = cur;
+               const Byte* start = cur;
                if (++cur < end and *cur == '@') {
                   t.type = Token::comma_at;
                   ++cur;
@@ -292,7 +292,7 @@ namespace OpenAxiom {
                break;
 
             case '#': {
-               const char* start = cur;
+               const Byte* start = cur;
                if (cur + 1 < end and special_after_sharp(cur[1])) {
                   t.type = Token::Type(OPENAXIOM_SEXPR_TOKEN2(cur[0], cur[1]));
                   t.lexeme = intern(cur, 2);
@@ -322,7 +322,7 @@ namespace OpenAxiom {
             }
 
             case '"': {
-               const char* start = cur;
+               const Byte* start = cur;
                skip_to_quote(++cur, end);
                t.type = Token::string;
                t.lexeme = intern(start, cur - start);
@@ -333,7 +333,7 @@ namespace OpenAxiom {
                if (start_symbol(cur, end))
                   t = match_maybe_symbol(this, cur, end);
                else {
-                  const char* start = cur;
+                  const Byte* start = cur;
                   skip_to_word_boundary(++cur, end);
                   t.lexeme = intern(start, cur - start);
                }
@@ -651,7 +651,7 @@ namespace OpenAxiom {
       // The sequence of characters in [cur, last) consists
       // entirely of digits.  Return the corresponding natural value.
       static size_t
-      natural_value(const char* cur, const char* last) {
+      natural_value(const Byte* cur, const Byte* last) {
          size_t n = 0;
          for (; cur < last; ++cur)
             // FIXME: check for overflow.
@@ -678,7 +678,7 @@ namespace OpenAxiom {
       equal_character_name(BasicString lhs, const char* rhs) {
          if (lhs->size() != strlen(rhs))
             return false;
-         for (const char* cur = lhs->begin(); cur != lhs->end(); ++cur)
+         for (const Byte* cur = lhs->begin(); cur != lhs->end(); ++cur)
             if (tolower(*cur) != *rhs++)
                return false;
          return true;
@@ -839,6 +839,11 @@ namespace OpenAxiom {
       Parser::Parser(Allocator& a, std::vector<const Syntax*>& v)
             : alloc(a), syns(v) { }
 
+      static std::string
+      to_string(BasicString s) {
+         return { s->begin(), s->end() };
+      }
+
       const Syntax*
       Parser::parse_syntax(const Token*& cur, const Token* last) {
          if (not skip_ignorable_tokens(cur, last))
@@ -898,7 +903,7 @@ namespace OpenAxiom {
 
          default:
             parse_error(std::string("parse error before ")
-                        + cur->lexeme->begin());
+                        + to_string(cur->lexeme));
             return 0;           // never executed
          }
       }
@@ -915,7 +920,7 @@ namespace OpenAxiom {
          std::vector<Token> tokens;
          Memory::FileMapping input(s);
          Lexer lexer(raw_strs, tokens);
-         const char* rest = lexer.tokenize(input.begin(), input.end());
+         const Byte* rest = lexer.tokenize(input.begin(), input.end());
          if (rest != input.end())
             syntax_error("syntax error");
          Parser parser(allocator, *this);
diff --git a/src/utils/storage.cc b/src/utils/storage.cc
index 54d14761..183dcd20 100644
--- a/src/utils/storage.cc
+++ b/src/utils/storage.cc
@@ -266,7 +266,8 @@ namespace OpenAxiom {
          HANDLE mapping = CreateFileMapping(file, 0, PAGE_READONLY, 0, 0, 0);
          if (mapping == 0)
             filesystem_error("could not map file " + path);
-         start = MapViewOfFile(mapping, FILE_MAP_READ, 0, 0, 0);
+         start = static_cast<Byte*>
+            (MapViewOfFile(mapping, FILE_MAP_READ, 0, 0, 0));
          extent = GetFileSize(file, 0);
          CloseHandle(mapping);
          CloseHandle(file);
@@ -280,7 +281,8 @@ namespace OpenAxiom {
          int fd = open(path.c_str(), O_RDONLY);
          if (fd < 0)
             filesystem_error("could not open " + path);
-         start = mmap(Pointer(), s.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
+         start = static_cast<Byte*>
+            (mmap(Pointer(), s.st_size, PROT_READ, MAP_PRIVATE, fd, 0));
          close(fd);
          if (start == MAP_FAILED)
             filesystem_error("could not map file " + path);
diff --git a/src/utils/string-pool.H b/src/utils/string-pool.H
index 311db1bf..5ed06fe3 100644
--- a/src/utils/string-pool.H
+++ b/src/utils/string-pool.H
@@ -47,12 +47,12 @@ namespace OpenAxiom {
    // ----------------
    // String data allocated from a stringpool.
    struct StringItem {
-      const char* begin() const { return text; }
-      const char* end() const { return text + length; }
+      const Byte* begin() const { return text; }
+      const Byte* end() const { return text + length; }
       size_t size() const { return length; }
-      bool equal(const char*, size_t) const;
+      bool equal(const Byte*, size_t) const;
    protected:
-      const char* text;         // pointer to the byte sequence
+      const Byte* text;         // pointer to the byte sequence
       size_t length;            // number of bytes in this string
       friend class StringPool;
       StringItem() : text(), length() { }
@@ -72,11 +72,11 @@ namespace OpenAxiom {
       EntryType* intern(const char*);
 
       // Intern a sequence of characters given by its start and length.
-      EntryType* intern(const char*, size_t);
+      EntryType* intern(const Byte*, size_t);
    private:
-      Memory::Arena<char> strings; // character blub
+      Memory::Arena<Byte> strings; // character blub
       // Allocate a string from the internal arena.
-      const char* make_copy(const char*, size_t);
+      const Byte* make_copy(const Byte*, size_t);
    };
 
    typedef const StringPool::EntryType* BasicString;
diff --git a/src/utils/string-pool.cc b/src/utils/string-pool.cc
index db5036a0..39186362 100644
--- a/src/utils/string-pool.cc
+++ b/src/utils/string-pool.cc
@@ -39,7 +39,7 @@ namespace OpenAxiom {
    // -- StringItem --
    // ----------------
    bool
-   StringItem::equal(const char* str, size_t sz) const {
+   StringItem::equal(const Byte* str, size_t sz) const {
       if (length != sz)
          return false;
       for (size_t i = 0; i < sz; ++i)
@@ -60,23 +60,23 @@ namespace OpenAxiom {
    // Return a hash for the string starting from `str'
    // of length `sz'.
    static size_t
-   hash(const char* str, size_t sz) {
+   hash(const Byte* str, size_t sz) {
       size_t h = 0;
       for(size_t i = 0; i < sz; ++i)
          h = str[i] + (h << 6) + (h << 16) - h;
       return h;
    }
 
-   const char*
-   StringPool::make_copy(const char* f, size_t sz) {
-      char* s = strings.allocate(sz + 1);
+   const Byte*
+   StringPool::make_copy(const Byte* f, size_t sz) {
+      Byte* s = strings.allocate(sz + 1);
       memcpy(s, f, sz);
       s[sz] = '\0';
       return s;
    }
    
    StringPool::EntryType*
-   StringPool::intern(const char* src, size_t sz) {
+   StringPool::intern(const Byte* src, size_t sz) {
       const size_t h = hash(src, sz);
       EntryType* e = hash_chain(h);
       if (sz == 0)
@@ -97,6 +97,6 @@ namespace OpenAxiom {
 
    StringPool::EntryType*
    StringPool::intern(const char* s) {
-      return intern(s, strlen(s));
+      return intern(reinterpret_cast<const Byte*>(s), strlen(s));
    }
 }