Merge pull request #74 from Servostar/69-better-parser-error-messages

69 better parser error messages

.gitignore

@@ -15,3 +15,6 @@ parser.tab.c
 parser.tab.h
 build
 /Testing/
+CTestTestfile.cmake
+DartConfiguration.tcl
+*.cmake

CMakeLists.txt

@@ -62,7 +62,7 @@ set(YACC_GENERATED_SOURCE_FILE ${PROJECT_SOURCE_DIR}/src/yacc/parser.tab.c)
 
 add_custom_command(OUTPUT ${YACC_GENERATED_SOURCE_FILE}
         COMMAND yacc
-        ARGS -Wcounterexamples -d -o ${YACC_GENERATED_SOURCE_FILE} ${YACC_SOURCE_FILE}
+        ARGS -Wno-yacc -Wcounterexamples -d -o ${YACC_GENERATED_SOURCE_FILE} ${YACC_SOURCE_FILE}
         COMMENT "generate C source file for parser"
         VERBATIM)
 

src/lex/lexer.l

@@ -2,9 +2,21 @@
 %{
     #include <yacc/parser.tab.h>
     #include <sys/log.h>
+    #include <lex/util.h>
 
     int yyLineNumber = 1;
+  
     int yylex();
+
+    extern int yyerror(const char* s);
+  
+    #define YY_USER_ACTION beginToken(yytext);
+
+    #define YY_INPUT(buf,result,max_size)  {\
+        result = nextChar(buf); \
+        if (  result <= 0  ) \
+          result = YY_NULL; \
+        }
 %}
 
 /* disable the following functions */
@@ -81,5 +93,6 @@
 [a-zA-Z_0-9]+ {DEBUG("\"%s\" tokenized with \'Ident\'", yytext); yylval.string = strdup(yytext); return(Ident); };
 \"([^\"\n])*\" {DEBUG("\"%s\" tokenized with \'ValStr\'", yytext); yylval.string = strdup(yytext); return(ValStr);};
 \"\"\"([^\"\n]|\\\n)*\"\"\" {DEBUG("\"%s\" tokenized with \'ValMultistr\'", yytext); yylval.string = strdup(yytext); return(ValMultistr);};
-.;
+[ \r\t] { /* ignore whitespace */ };
+. { return yytext[0]; /* passthrough unknown token, let parser handle the error */ };
 %%

src/lex/util.c

@@ -0,0 +1,79 @@
+
+#include <lex/util.h>
+#include <string.h>
+#include <stdlib.h>
+
+// implementation based on:
+// https://github.com/sunxfancy/flex-bison-examples/blob/master/error-handling/ccalc.c
+
+char* buffer = NULL;
+
+static int eof = 0;
+static int nRow = 0;
+static int nBuffer = 0;
+static int lBuffer = 0;
+static int nTokenStart = 0;
+static int nTokenLength = 0;
+static int nTokenNextStart = 0;
+
+static void lex_deinit(void) {
+  free(buffer);
+}
+
+void lex_init(void) {
+  buffer = malloc(MAX_READ_BUFFER_SIZE);
+  atexit(lex_deinit);
+}
+
+void beginToken(char *t) {
+  nTokenStart = nTokenNextStart;
+  nTokenLength = (int) strlen(t);
+  nTokenNextStart = nBuffer + 1;
+
+  yylloc.first_line = nRow;
+  yylloc.first_column = nTokenStart;
+  yylloc.last_line = nRow;
+  yylloc.last_column = nTokenStart + nTokenLength - 1;
+}
+
+int nextChar(char *dst) {
+  int frc;
+
+  if (eof)
+    return 0;
+  
+  while (nBuffer >= lBuffer) {
+    frc = getNextLine();
+    if (frc != 0) {
+      return 0;
+    }
+  }
+
+  dst[0] = buffer[nBuffer];
+  nBuffer += 1;
+
+  return dst[0] != 0;
+}
+
+int getNextLine(void) {
+  char *p;
+  
+  nBuffer = 0;
+  nTokenStart = -1;
+  nTokenNextStart = 1;
+  eof = 0;
+
+  p = fgets(buffer, MAX_READ_BUFFER_SIZE, yyin);
+  if (p == NULL) {
+    if (ferror(yyin)) {
+      return -1;
+    }
+    eof = 1;
+    return 1;
+  }
+
+  nRow += 1;
+  lBuffer = (int) strlen(buffer);
+
+  return 0;
+}

src/lex/util.h

@@ -0,0 +1,37 @@
+
+#ifndef LEX_UTIL_H_
+#define LEX_UTIL_H_
+
+#include <yacc/parser.tab.h>
+#include <stdio.h>
+
+#define MAX_READ_BUFFER_SIZE 1000
+
+extern FILE* yyin;
+extern YYLTYPE yylloc;
+extern char* buffer;
+
+/**
+ * @brief Initialize global state needed for the lexer
+ */
+void lex_init(void);
+
+/**
+ * @brief Begin counting a new token. This will fill the global struct yylloc.
+ * @param t the text of the token. Must be null terminated
+ */
+[[gnu::nonnull(1)]]
+void beginToken(char *t);
+
+/**
+ * @brief Stores the next character into the supplied buffer
+ * @param dst the buffer to store character in
+ */
+int nextChar(char *dst);
+
+/**
+ * @brief Reads the next line from yyin into a global buffer
+ */
+int getNextLine(void);
+
+#endif // LEX_UTIL_H_

src/main.c

@@ -2,11 +2,11 @@
 #include <stdlib.h>
 #include <sys/log.h>
 #include <yacc/parser.tab.h>
+#include <sys/col.h>
+#include <lex/util.h>
 
 #define LOG_LEVEL LOG_LEVEL_DEBUG
 
-extern FILE *yyin;
-
 /**
  * @brief Log a debug message to inform about beginning exit procedures
  *
@@ -41,6 +41,10 @@ void setup(void) {
   // actual setup
   AST_init();
 
+  col_init();
+
+  lex_init();
+
   DEBUG("finished starting up gemstone...");
 }
 

src/sys/col.c

@@ -0,0 +1,87 @@
+
+#include <stdlib.h>
+#include <string.h>
+#include <sys/col.h>
+#include <sys/log.h>
+
+#ifdef __unix__
+#include <unistd.h>
+#elif defined(_WIN32) || defined(WIN32)
+#include <Windows.h>
+#endif
+
+char *RED;
+char *YELLOW;
+char *MAGENTA;
+char *CYAN;
+char *GREEN;
+char *RESET;
+char *BOLD;
+char *FAINT;
+
+void col_init(void) {
+  if (stdout_supports_ansi_esc()) {
+    enable_ansi_colors();
+  } else {
+    disable_ansi_colors();
+  }
+}
+
+void disable_ansi_colors() {
+  DEBUG("disabling ANSI escape codes");
+
+  RED = "";
+  YELLOW = "";
+  MAGENTA = "";
+  CYAN = "";
+  GREEN = "";
+  RESET = "";
+  BOLD = "";
+  FAINT = "";
+}
+
+void enable_ansi_colors() {
+  DEBUG("enabling ANSI escape codes");
+
+  RED = "\x1b[31m";
+  YELLOW = "\x1b[33m";
+  MAGENTA = "\x1b[35m";
+  CYAN = "\x1b[36m";
+  GREEN = "\x1b[32m";
+  RESET = "\x1b[0m";
+  BOLD = "\x1b[1m";
+  FAINT = "\x1b[2m";
+}
+
+int stdout_supports_ansi_esc() {
+
+#ifdef __unix__
+  // check if TTY
+  if (isatty(STDOUT_FILENO)) {
+    const char *colors = getenv("COLORTERM");
+    // check if colors are set and allowed
+    if (colors != NULL && (strcmp(colors, "truecolor") == 0 || strcmp(colors, "24bit") == 0)) {
+      return ANSI_ENABLED;
+    }
+  }
+#elif defined(_WIN32) || defined(WIN32)
+  // see:
+  // https://stackoverflow.com/questions/63913005/how-to-test-if-console-supports-ansi-color-codes
+  DWORD mode;
+  HANDLE hConsole = GetStdHandle(STD_OUTPUT_HANDLE);
+
+  if (!GetConsoleMode(hConsole, &mode)) {
+    ERROR("failed to get console mode");
+    return ANSI_ENABLED;
+  }
+
+  if ((mode & ENABLE_VIRTUAL_TERMINAL_INPUT) |
+      (mode & ENABLE_VIRTUAL_TERMINAL_PROCESSING)) {
+    return ANSI_ENABLED;
+  }
+#else
+#warning "unsupported platform, ASNI escape codes disabled by default"
+#endif
+
+  return ASNI_DISABLED;
+}

src/sys/col.h

@@ -0,0 +1,43 @@
+
+#ifndef COLORS_H_
+#define COLORS_H_
+
+#define ANSI_ENABLED 1
+#define ASNI_DISABLED 0
+
+// Common escape codes
+// can be used to print colored text
+extern char *RED;
+extern char *YELLOW;
+extern char *MAGENTA;
+extern char *CYAN;
+extern char *GREEN;
+extern char *RESET;
+extern char *BOLD;
+extern char *FAINT;
+
+/**
+ * @brief Initialize global state
+ */
+void col_init(void);
+
+/**
+ * @brief Enable ANSI escape codes. This will set the correct escape codes to
+ *        the global strings above.
+ */
+void enable_ansi_colors();
+
+/**
+ * @brief Disable ANSI escape codes. This will set all the above global strings to be empty.
+ */
+void disable_ansi_colors();
+
+/**
+ * @brief Check if stdout may support ANSI escape codes.
+ * @attention This function may report escape codes to be unavailable even if they actually are.
+ * @return ANSI_ENABLED if escape sequences are supported ASNI_DISABLED otherwise 
+ */
+[[nodiscard]]
+int stdout_supports_ansi_esc();
+
+#endif // COLORS_H_

src/yacc/parser.y

@@ -1,9 +1,15 @@
+%locations
+%define parse.error verbose
+
 %{
     #include <sys/log.h>
+    #include <sys/col.h>
+
+    int yyerror(const char*);
+
+    extern char* buffer;
     extern int yylineno;
-
-    int yyerror(char*);
-
+    
     extern int yylex();
 %}
 
@@ -55,6 +61,7 @@
 %token FunFunname
 %token FunLineno
 %token FunExtsupport
+%token Invalid
 
 /* Operator associativity */
 %right '='
@@ -222,7 +229,70 @@ opbit: expr OpBitand expr
     | OpBitnot expr %prec OpBitand;
 %%
 
-int yyerror(char *s) {
-    ERROR("%s", s);
-    return 0;
+ 
+const char* ERROR = "error";
+const char* WARNING = "warning";
+const char* NOTE = "note";
+
+int print_message(const char* kind, const char* message) {
+    // number of characters written
+    int char_count = 0;
+    // highlight to use
+    char* HIGHLIGHT = CYAN;
+
+    // convert message kind into color
+    if (kind == ERROR) {
+        HIGHLIGHT = RED;
+    } else if (kind == WARNING) {
+        HIGHLIGHT = YELLOW;
+    }
+
+    // print message
+    char_count += printf("%sfilename:%d:%d%s:%s%s %s: %s%s\n", BOLD, yylloc.first_line, yylloc.first_column, RESET, HIGHLIGHT, BOLD, kind, RESET, message);
+    
+    // print line in which error occurred
+    
+    char_count += printf(" %4d | ", yylloc.first_line);
+
+    for (int i = 0; i < yylloc.first_column - 1; i++) {
+        if (buffer[i] == '\n') {
+            break;
+        }
+        printf("%c", buffer[i]);
+    }
+
+    char_count += printf("%s%s", BOLD, HIGHLIGHT);
+
+    for (int i = yylloc.first_column - 1; i < yylloc.last_column; i++) {
+        if (buffer[i] == '\n') {
+            break;
+        }
+        char_count += printf("%c", buffer[i]);
+    }
+
+    char_count += printf("%s", RESET);
+
+    for (int i = yylloc.last_column; buffer[i] != '\0' && buffer[i] != '\n'; i++) {
+        printf("%c", buffer[i]);
+    }
+
+    char_count += printf("\n      | ");
+
+    for (int i = 0; i < yylloc.first_column - 1; i++) {
+        char_count += printf(" ");
+    }
+
+    char_count += printf("%s^", HIGHLIGHT);
+
+    for (int i = 0; i < yylloc.last_column - yylloc.first_column; i++) {
+        printf("~");
+    }
+    
+    char_count += printf("%s\n\n", RESET);
+
+    return char_count;
+}
+
+int yyerror(const char *s) {
+    return print_message(ERROR, s);
 }