Support C-Style enum in preproc

Makefile

@@ -311,7 +311,7 @@ ifeq ($(NODEP),1)
 $(C_BUILDDIR)/%.o: $(C_SUBDIR)/%.c
 ifeq (,$(KEEP_TEMPS))
 	@echo "$(CC1) <flags> -o $@ $<"
-	@$(CPP) $(CPPFLAGS) $< | $(PREPROC) $< charmap.txt -i | $(CC1) $(CFLAGS) -o - - | cat - <(echo -e ".text\n\t.align\t2, 0") | $(AS) $(ASFLAGS) -o $@ -
+	@$(CPP) $(CPPFLAGS) $< | $(PREPROC) -i $< charmap.txt | $(CC1) $(CFLAGS) -o - - | cat - <(echo -e ".text\n\t.align\t2, 0") | $(AS) $(ASFLAGS) -o $@ -
 else
 	@$(CPP) $(CPPFLAGS) $< -o $(C_BUILDDIR)/$*.i
 	@$(PREPROC) $(C_BUILDDIR)/$*.i charmap.txt | $(CC1) $(CFLAGS) -o $(C_BUILDDIR)/$*.s
@@ -323,7 +323,7 @@ define C_DEP
 $1: $2 $$(shell $(SCANINC) -I include -I tools/agbcc/include -I gflib $2)
 ifeq (,$$(KEEP_TEMPS))
 	@echo "$$(CC1) <flags> -o $$@ $$<"
-	@$$(CPP) $$(CPPFLAGS) $$< | $$(PREPROC) $$< charmap.txt -i | $$(CC1) $$(CFLAGS) -o - - | cat - <(echo -e ".text\n\t.align\t2, 0") | $$(AS) $$(ASFLAGS) -o $$@ -
+	@$$(CPP) $$(CPPFLAGS) $$< | $$(PREPROC) -i $$< charmap.txt | $$(CC1) $$(CFLAGS) -o - - | cat - <(echo -e ".text\n\t.align\t2, 0") | $$(AS) $$(ASFLAGS) -o $$@ -
 else
 	@$$(CPP) $$(CPPFLAGS) $$< -o $$(C_BUILDDIR)/$3.i
 	@$$(PREPROC) $$(C_BUILDDIR)/$3.i charmap.txt | $$(CC1) $$(CFLAGS) -o $$(C_BUILDDIR)/$3.s
@@ -338,7 +338,7 @@ ifeq ($(NODEP),1)
 $(GFLIB_BUILDDIR)/%.o: $(GFLIB_SUBDIR)/%.c $$(c_dep)
 ifeq (,$(KEEP_TEMPS))
 	@echo "$(CC1) <flags> -o $@ $<"
-	@$(CPP) $(CPPFLAGS) $< | $(PREPROC) $< charmap.txt -i | $(CC1) $(CFLAGS) -o - - | cat - <(echo -e ".text\n\t.align\t2, 0") | $(AS) $(ASFLAGS) -o $@ -
+	@$(CPP) $(CPPFLAGS) $< | $(PREPROC) -i $< charmap.txt | $(CC1) $(CFLAGS) -o - - | cat - <(echo -e ".text\n\t.align\t2, 0") | $(AS) $(ASFLAGS) -o $@ -
 else
 	@$(CPP) $(CPPFLAGS) $< -o $(GFLIB_BUILDDIR)/$*.i
 	@$(PREPROC) $(GFLIB_BUILDDIR)/$*.i charmap.txt | $(CC1) $(CFLAGS) -o $(GFLIB_BUILDDIR)/$*.s
@@ -350,7 +350,7 @@ define GFLIB_DEP
 $1: $2 $$(shell $(SCANINC) -I include -I tools/agbcc/include -I gflib $2)
 ifeq (,$$(KEEP_TEMPS))
 	@echo "$$(CC1) <flags> -o $$@ $$<"
-	@$$(CPP) $$(CPPFLAGS) $$< | $$(PREPROC) $$< charmap.txt -i | $$(CC1) $$(CFLAGS) -o - - | cat - <(echo -e ".text\n\t.align\t2, 0") | $$(AS) $$(ASFLAGS) -o $$@ -
+	@$$(CPP) $$(CPPFLAGS) $$< | $$(PREPROC) -i $$< charmap.txt | $$(CC1) $$(CFLAGS) -o - - | cat - <(echo -e ".text\n\t.align\t2, 0") | $$(AS) $$(ASFLAGS) -o $$@ -
 else
 	@$$(CPP) $$(CPPFLAGS) $$< -o $$(GFLIB_BUILDDIR)/$3.i
 	@$$(PREPROC) $$(GFLIB_BUILDDIR)/$3.i charmap.txt | $$(CC1) $$(CFLAGS) -o $$(GFLIB_BUILDDIR)/$3.s
@@ -363,11 +363,11 @@ endif
 
 ifeq ($(NODEP),1)
 $(C_BUILDDIR)/%.o: $(C_SUBDIR)/%.s
-	$(PREPROC) $< charmap.txt | $(CPP) -I include - | $(AS) $(ASFLAGS) -o $@
+	$(PREPROC) $< charmap.txt | $(CPP) -I include - | $(PREPROC) -i $$< charmap.txt | $(AS) $(ASFLAGS) -o $@
 else
 define SRC_ASM_DATA_DEP
 $1: $2 $$(shell $(SCANINC) -I include -I "" $2)
-	$$(PREPROC) $$< charmap.txt | $$(CPP) -I include - | $$(AS) $$(ASFLAGS) -o $$@
+	$$(PREPROC) $$< charmap.txt | $$(CPP) -I include - | $$(PREPROC) -ie $$< charmap.txt | $$(AS) $$(ASFLAGS) -o $$@
 endef
 $(foreach src, $(C_ASM_SRCS), $(eval $(call SRC_ASM_DATA_DEP,$(patsubst $(C_SUBDIR)/%.s,$(C_BUILDDIR)/%.o, $(src)),$(src))))
 endif
@@ -385,7 +385,7 @@ endif
 
 ifeq ($(NODEP),1)
 $(DATA_ASM_BUILDDIR)/%.o: $(DATA_ASM_SUBDIR)/%.s
-	$(PREPROC) $< charmap.txt | $(CPP) -I include - | $(AS) $(ASFLAGS) -o $@
+	$(PREPROC) $< charmap.txt | $(CPP) -I include - | $(PREPROC) -ie $$< charmap.txt | $(AS) $(ASFLAGS) -o $@
 else
 $(foreach src, $(REGULAR_DATA_ASM_SRCS), $(eval $(call SRC_ASM_DATA_DEP,$(patsubst $(DATA_ASM_SUBDIR)/%.s,$(DATA_ASM_BUILDDIR)/%.o, $(src)),$(src))))
 endif

tools/preproc/Makefile

@@ -3,10 +3,10 @@ CXX ?= g++
 CXXFLAGS := -std=c++11 -O2 -Wall -Wno-switch -Werror
 
 SRCS := asm_file.cpp c_file.cpp charmap.cpp preproc.cpp string_parser.cpp \
-	utf8.cpp
+	utf8.cpp io.cpp
 
 HEADERS := asm_file.h c_file.h char_util.h charmap.h preproc.h string_parser.h \
-	utf8.h
+	utf8.h io.h
 
 ifeq ($(OS),Windows_NT)
 EXE := .exe

tools/preproc/asm_file.cpp

@@ -27,33 +27,12 @@
 #include "utf8.h"
 #include "string_parser.h"
 #include "../../gflib/characters.h"
+#include "io.h"
 
-AsmFile::AsmFile(std::string filename) : m_filename(filename)
+AsmFile::AsmFile(std::string filename, bool isStdin, bool doEnum) : m_filename(filename)
 {
-    FILE *fp = std::fopen(filename.c_str(), "rb");
-
-    if (fp == NULL)
-        FATAL_ERROR("Failed to open \"%s\" for reading.\n", filename.c_str());
-
-    std::fseek(fp, 0, SEEK_END);
-
-    m_size = std::ftell(fp);
-
-    if (m_size < 0)
-        FATAL_ERROR("File size of \"%s\" is less than zero.\n", filename.c_str());
-    else if (m_size == 0)
-        return; // Empty file
-
-    m_buffer = new char[m_size + 1];
-
-    std::rewind(fp);
-
-    if (std::fread(m_buffer, m_size, 1, fp) != 1)
-        FATAL_ERROR("Failed to read \"%s\".\n", filename.c_str());
-
-    m_buffer[m_size] = 0;
-
-    std::fclose(fp);
+    m_buffer = ReadFileToBuffer(filename.c_str(), isStdin, &m_size);
+    m_doEnum = doEnum;
 
     m_pos = 0;
     m_lineNum = 1;
@@ -65,6 +44,7 @@ AsmFile::AsmFile(std::string filename) : m_filename(filename)
 AsmFile::AsmFile(AsmFile&& other) : m_filename(std::move(other.m_filename))
 {
     m_buffer = other.m_buffer;
+    m_doEnum = other.m_doEnum;
     m_pos = other.m_pos;
     m_size = other.m_size;
     m_lineNum = other.m_lineNum;
@@ -174,6 +154,8 @@ Directive AsmFile::GetDirective()
         return Directive::String;
     else if (CheckForDirective(".braille"))
         return Directive::Braille;
+    else if (CheckForDirective("enum"))
+        return Directive::Enum;
     else
         return Directive::Unknown;
 }
@@ -527,6 +509,70 @@ void AsmFile::OutputLine()
     }
 }
 
+// parses an assumed C `enum`. Returns false if `enum { ...` is not matched
+bool AsmFile::ParseEnum()
+{
+    if (!m_doEnum)
+        return false;
+
+    long fallbackPosition = m_pos;
+    std::string headerFilename = "";
+    long currentHeaderLine = SkipWhitespaceAndEol();
+    std::string enumName = ReadIdentifier();
+    currentHeaderLine += SkipWhitespaceAndEol();
+    long enumCounter = 0;
+    long symbolCount = 0;
+
+    if (m_buffer[m_pos] != '{') // assume assembly macro, otherwise assume enum and report errors accordingly
+    {
+        m_pos = fallbackPosition - 4;
+        return false;
+    }
+
+    currentHeaderLine += FindLastLineNumber(headerFilename);
+    m_pos++;
+    for (;;)
+    {
+        currentHeaderLine += SkipWhitespaceAndEol();
+        std::string currentIdentName = ReadIdentifier();
+        if (!currentIdentName.empty())
+        {
+            std::printf("# %ld \"%s\"\n", currentHeaderLine, headerFilename.c_str());
+            currentHeaderLine += SkipWhitespaceAndEol();
+            if (m_buffer[m_pos] == '=')
+            {
+                m_pos++;
+                currentHeaderLine += SkipWhitespaceAndEol();
+                enumCounter = ReadInteger(headerFilename, currentHeaderLine);
+                currentHeaderLine += SkipWhitespaceAndEol();
+            }
+            std::printf(".equiv %s, %ld\n", currentIdentName.c_str(), enumCounter);
+            enumCounter++;
+            symbolCount++;
+        }
+        else if (symbolCount == 0)
+        {
+            RaiseError("%s:%ld: empty enum is invalid", headerFilename.c_str(), currentHeaderLine);
+        }
+
+        if (m_buffer[m_pos] != ',')
+        {
+            currentHeaderLine += SkipWhitespaceAndEol();
+            if (m_buffer[m_pos++] == '}' && m_buffer[m_pos++] == ';')
+            {
+                ExpectEmptyRestOfLine();
+                break;
+            }
+            else
+            {
+                RaiseError("unterminated enum from included file %s:%ld", headerFilename.c_str(), currentHeaderLine);
+            }
+        }
+        m_pos++;
+    }
+    return true;
+}
+
 // Asserts that the rest of the line is empty and moves to the next one.
 void AsmFile::ExpectEmptyRestOfLine()
 {
@@ -599,3 +645,130 @@ void AsmFile::RaiseWarning(const char* format, ...)
 {
     DO_REPORT("warning");
 }
+
+// Skips Whitespace including newlines and returns the amount of newlines skipped
+int AsmFile::SkipWhitespaceAndEol()
+{
+    int newlines = 0;
+    while (m_buffer[m_pos] == '\t' || m_buffer[m_pos] == ' ' || m_buffer[m_pos] == '\n')
+    {
+        if (m_buffer[m_pos] == '\n')
+            newlines++;
+        m_pos++;
+    }
+    return newlines;
+}
+
+// returns the last line indicator and its corresponding file name without modifying the token index
+int AsmFile::FindLastLineNumber(std::string& filename)
+{
+    long pos = m_pos;
+    long linebreaks = 0;
+    while (m_buffer[pos] != '#' && pos >= 0)
+    {
+        if (m_buffer[pos] == '\n')
+            linebreaks++;
+        pos--;
+    }
+
+    if (pos < 0)
+        RaiseError("line indicator for header file not found before `enum`");
+
+    pos++;
+    while (m_buffer[pos] == ' ' || m_buffer[pos] == '\t')
+        pos++;
+
+    if (!IsAsciiDigit(m_buffer[pos]))
+        RaiseError("malformatted line indicator found before `enum`, expected line number");
+
+    unsigned n = 0;
+    int digit = 0;
+    while ((digit = ConvertDigit(m_buffer[pos++], 10)) != -1)
+        n = 10 * n + digit;
+
+    while (m_buffer[pos] == ' ' || m_buffer[pos] == '\t')
+        pos++;
+
+    if (m_buffer[pos++] != '"')
+        RaiseError("malformatted line indicator found before `enum`, expected filename");
+
+    while (m_buffer[pos] != '"')
+    {
+        unsigned char c = m_buffer[pos++];
+
+        if (c == 0)
+        {
+            if (pos >= m_size)
+                RaiseError("unexpected EOF in line indicator");
+            else
+                RaiseError("unexpected null character in line indicator");
+        }
+
+        if (!IsAsciiPrintable(c))
+            RaiseError("unexpected character '\\x%02X' in line indicator", c);
+
+        if (c == '\\')
+        {
+            c = m_buffer[pos];
+            RaiseError("unexpected escape '\\%c' in line indicator", c);
+        }
+
+        filename += c;
+    }
+
+    return n + linebreaks - 1;
+}
+
+std::string AsmFile::ReadIdentifier()
+{
+    long start = m_pos;
+    if (!IsIdentifierStartingChar(m_buffer[m_pos]))
+        return std::string();
+
+    m_pos++;
+
+    while (IsIdentifierChar(m_buffer[m_pos]))
+        m_pos++;
+
+    return std::string(&m_buffer[start], m_pos - start);
+}
+
+long AsmFile::ReadInteger(std::string filename, long line)
+{
+    bool negate = false;
+    int radix = 10;
+    if (!IsAsciiDigit(m_buffer[m_pos]))
+    {
+        if (m_buffer[m_pos++] == '-')
+            negate = true;
+        else
+            RaiseError("expected number in included file %s:%ld", filename.c_str(), line);
+    }
+
+    if (m_buffer[m_pos] == '0' && m_buffer[m_pos + 1] == 'x')
+    {
+        radix = 16;
+        m_pos += 2;
+    }
+    else if (m_buffer[m_pos] == '0' && m_buffer[m_pos + 1] == 'b')
+    {
+        radix = 2;
+        m_pos += 2;
+    }
+    else if (m_buffer[m_pos] == '0' && IsAsciiDigit(m_buffer[m_pos+1]))
+    {
+        radix = 8;
+        m_pos++;
+    }
+
+    long n = 0;
+    int digit;
+
+    while ((digit = ConvertDigit(m_buffer[m_pos], radix)) != -1)
+    {
+        n = n * radix + digit;
+        m_pos++;
+    }
+
+    return negate ? -n : n;
+}

tools/preproc/asm_file.h

@@ -31,13 +31,14 @@ enum class Directive
     Include,
     String,
     Braille,
+    Enum,
     Unknown
 };
 
 class AsmFile
 {
 public:
-    AsmFile(std::string filename);
+    AsmFile(std::string filename, bool isStdin, bool doEnum);
     AsmFile(AsmFile&& other);
     AsmFile(const AsmFile&) = delete;
     ~AsmFile();
@@ -49,9 +50,11 @@ class AsmFile
     bool IsAtEnd();
     void OutputLine();
     void OutputLocation();
+    bool ParseEnum();
 
 private:
     char* m_buffer;
+    bool m_doEnum;
     long m_pos;
     long m_size;
     long m_lineNum;
@@ -68,6 +71,10 @@ class AsmFile
     void RaiseError(const char* format, ...);
     void RaiseWarning(const char* format, ...);
     void VerifyStringLength(int length);
+    int SkipWhitespaceAndEol();
+    int FindLastLineNumber(std::string& filename);
+    std::string ReadIdentifier();
+    long ReadInteger(std::string filename, long line);
 };
 
 #endif // ASM_FILE_H