Improve error reporting, by printing the errored line and a caret (^), taking care to display the beginning of the token the error is at.

lslopt/lslparse.py

@@ -60,14 +60,12 @@ def fieldpos(inp, sep, n):
 class EParse(Exception):
 
     def __init__(self, parser, msg):
-        errorpos = parser.errorpos
-        if parser.script[errorpos:errorpos+1] == '\n':
-            errorpos += 1
-        lno = parser.script.count('\n', 0, errorpos)
-        cno = errorpos - fieldpos(parser.script, '\n', lno)
+        self.errorpos = parser.errorpos
+        self.lno = parser.script.count('\n', 0, self.errorpos)
+        self.cno = self.errorpos - fieldpos(parser.script, '\n', self.lno)
         # Note the column number reported is in bytes.
 
-        msg = u"(Line %d char %d): ERROR: %s" % (lno + 1, cno + 1, msg)
+        msg = u"(Line %d char %d): ERROR: %s" % (self.lno + 1, self.cno + 1, msg)
         super(EParse, self).__init__(msg)
 
 class EParseUEOF(EParse):
@@ -309,11 +307,11 @@ class parser(object):
     def GetToken(self):
         """Lexer"""
 
-        # Keep track of the current position. If an error occurs, it will happen at the start of this token.
-        self.errorpos = self.pos
-
         try:
             while self.pos < self.length:
+                # If an error occurs, it will happen at the start of this token.
+                self.errorpos = self.pos
+
                 c = self.script[self.pos]
                 self.pos += 1
 
@@ -2388,14 +2386,14 @@ list lazy_list_set(list L, integer i, list v)
         # incomplete parsing pass, gathering globals with their types and
         # function arguments. And that's what we do.
 
-        self.pos = 0
+        self.pos = self.errorpos = 0
         self.tok = self.GetToken()
 
         self.globals = self.BuildTempGlobalsTable()
 
         # Restart
 
-        self.pos = 0
+        self.pos = self.errorpos = 0
         self.tok = self.GetToken()
 
         # Reserve spots at the beginning for functions we add

main.py

@@ -19,16 +19,22 @@
 
 # This is the main executable program that imports the libraries.
 
-from lslopt.lslparse import parser,EParse
+from lslopt.lslparse import parser,EParse,fieldpos
 from lslopt.lsloutput import outscript
 from lslopt.lsloptimizer import optimizer
 import sys, os, getopt, re
 import lslopt.lslcommon
 
 
-VERSION = '0.1.3alpha'
+VERSION = '0.2.0beta'
 
 
+def ReportError(script, e):
+    sys.stderr.write(script[fieldpos(script, '\n', e.lno):
+                            fieldpos(script, '\n', e.lno+1)-1] + '\n')
+    sys.stderr.write(' ' * e.cno + '^\n')
+    sys.stderr.write(e[0] + '\n')
+
 class UniConvScript(object):
     '''Converts the script to Unicode, setting the properties required by
     EParse to report a meaningful error position.
@@ -467,6 +473,8 @@ def main():
                 # need the result.
                 UniConvScript(script).to_unicode()
             except EParse as e:
+                # We don't call ReportError to prevent problems due to
+                # displaying invalid UTF-8
                 sys.stderr.write(e[0] + '\n')
                 return 1
         script = PreparePreproc(script)
@@ -502,7 +510,7 @@ def main():
         try:
             ts = p.parse(script, options)
         except EParse as e:
-            sys.stderr.write(e[0] + '\n')
+            ReportError(script, e)
             return 1
         del p, script