The reformatter for compiler's error messsages

terror.py

#!/usr/bin/env python3

import fnmatch
import os
import re
import sys

msgtype = "gcc"
rootdir = os.getcwd()
workdir = rootdir

linkdict = {}

def readlink(source):
	global linkdict, rootdir, workdir
	if source in linkdict:
		return linkdict[source]
	else:
		target = source
		while True:
			old = target
			try:
				target = os.readlink(old)
			except OSError:
				break
			if target[0] != '/':
				target = os.path.normpath( \
				           os.path.join( \
				             os.path.dirname(old), \
				             target))
		#print rootdir, workdir, target
		if target[0] != '/' and rootdir != workdir:
			target = os.path.join(workdir, target)
		if target[0] == '/':
			target = os.path.relpath(target, rootdir)
		linkdict[source] = target
		return target

opened_file = None
opened_file_line = 0
opened_file_s = ""

def fixcol(source, line, col):
	global msgtype, opened_file, opened_file_line, opened_file_s
	if msgtype == "ocaml":
		return col + 1
	if opened_file \
	     and (opened_file.name != source or opened_file_line > line):
		opened_file.close()
		opened_file = None
	if opened_file is None:
		opened_file = open(source, "r")
		opened_file_line = 0
	while opened_file_line < line:
		opened_file_s = opened_file.readline()
		opened_file_line += 1
	tabwdith = 8
	col1 = 1
	col8 = 1
	while col8 < col:
		try:
			c = opened_file_s[col1 - 1]
		except IndexError:
			return None
		if c == '\t':
			col8 += tabwdith - (col8 - 1) % tabwdith
		else:
			col8 += 1
		col1 += 1
	if col1 < len(opened_file_s) and opened_file_s[col1 - 1] == '.':
		col1 += 1 # fix "." included into a next identifier
	return col1 # for vscode

fixcol_filter = []

def have_to_fixcol(source):
	global fixcol_filter
	for i in fixcol_filter:
		if fnmatch.fnmatch(source, i):
			return True
	return False

def parse_args():
	global msgtype, rootdir, workdir, fixcol_filter
	i = 1
	while i < len(sys.argv):
		if sys.argv[i] == "--ocaml":
			msgtype = "ocaml"
		elif sys.argv[i] == "--root":
			i += 1
			rootdir = sys.argv[i]
		elif sys.argv[i] == "--cd" or sys.argv[i] == '-C':
			i += 1
			workdir = sys.argv[i]
		elif sys.argv[i] == '--fixcol':
			i += 1
			fixcol_filter.append(sys.argv[i])
		elif sys.argv[i][0] == '-':
			raise Exception
		i = i + 1

parse_args()

if msgtype == "ocaml":
	errhea_re = None
	anyhea_re = re.compile(r"^File \"([^:\"]*)\", lines? (\d+)(?:-\d+)?, " \
	                         + r"characters (\d+)-(?P<endcol>\d+):$")
	locmsg_re = re.compile(r"^(\d+ \|| +\^)")
	errmsg_re = re.compile(r"^(?P<level>Error):")
	anymsg_re = re.compile(r"^(?:(?P<level>Alert \w+:|Warning \d+:)" \
	                         + "|[A-Z]|$|stdlib-shims library:)")
else:
	errhea_re = re.compile(r"^([^:\"]+):(\d+)(?::(\d+)?):? (?P<level>(?:error:)?)")
	anyhea_re = re.compile(r"^([^:\"]+):(\d+)(?::(\d+)?):? " \
	                         + "(?P<level>warning|note):")
	locmsg_re = re.compile(r"^ ")
	errmsg_re = None
	anymsg_re = None

makemsg_re = re.compile(r"^Makefile:\d+: recipe for target .* failed$" \
                          + r"|^make(?:\[\d+\])?: " \
                          + r"|^error: ocamldep.opt ") # ocamlmake

def ein(level):
	if level == 'E':
		return "\x1b[91m"
	else:
		return "\x1b[95m"

has_error = False
in_error = None
for s in sys.stdin:
	makemsg = makemsg_re.match(s)
	result = None
	location_msg = None
	additional_msg = None
	if makemsg is None:
		result = anyhea_re.match(s)
		if result:
			in_error = 'L'
		if result is None and errhea_re:
			result = errhea_re.match(s)
			if result:
				in_error = 'E'
		if result is None and in_error:
			if in_error != 'A':
				location_msg = locmsg_re.match(s)
			if location_msg is None \
			     and errmsg_re \
			     and in_error == 'L':
				additional_msg = errmsg_re.match(s)
				if additional_msg:
					in_error = 'E'
			if location_msg is None \
			     and additional_msg is None \
			     and anymsg_re:
				additional_msg = anymsg_re.match(s)
				if additional_msg and in_error == 'L':
					in_error = 'A'
			if location_msg is None and additional_msg is None:
				in_error = None
	if result:
		source = result.group(1) # filename
		target = readlink(source)
		line = int(result.group(2))
		resetted = False
		# level
		try:
			(s5, e5) = result.span('level')
		except IndexError:
			(s5, e5) = (-1, -1)
		if s5 >= 0:
			s = s[:s5] + ein(in_error) + s[s5:e5] + "\x1b[0m" + s[e5:]
			resetted = True
		# endcol
		try:
			(s4, e4) = result.span('endcol')
		except IndexError:
			(s4, e4) = (-1, -1)
		if s4 >= 0 and have_to_fixcol(target):
			endcol = int(result.group(4))
			endcol = fixcol(target, line, endcol)
			if endcol:
				s = s[:s4] + str(endcol) + s[e4:]
			else:
				s = s[:s4 - 1] + s[e4:]
		# col
		(s3, e3) = result.span(3)
		if s3 >= 0:
			col = int(result.group(3))
			if have_to_fixcol(target):
				col = fixcol(target, line, col)
				if col:
					s = s[:s3] + str(col) + s[e3:]
				else:
					s = s[:s3 - 1] + s[e3:]
		if target != source:
			s = s[:result.start(1)] + target + s[result.end(1):]
		sys.stdout.write("\x1b[97m")
		if msgtype == "ocaml":
			msg = target + ":" + str(line) + ":" + str(col) + ":\n"
			sys.stdout.write(msg)
			sys.stdout.write("\x1b[0m")
			resetted = True
		sys.stdout.write(s)
		if not resetted:
			sys.stdout.write("\x1b[0m")
		sys.stdout.flush()
		has_error = True
	elif location_msg:
		sys.stdout.write(s)
		sys.stdout.flush()
	elif additional_msg:
		try:
			(s1, e1) = additional_msg.span('level')
		except IndexError:
			(s1, e1) = (-1, -1)
		if s1 >= 0:
			sys.stdout.write(ein(in_error))
			sys.stdout.write(s[:e1])
			sys.stdout.write("\x1b[0m")
			sys.stdout.write(s[e1:])
		else:
			sys.stdout.write(s)
		sys.stdout.flush()
	elif not has_error or not makemsg:
		sys.stdout.write("\x1b[37m")
		sys.stdout.write(s)
		sys.stdout.write("\x1b[0m")
		sys.stdout.flush()

if opened_file:
	opened_file.close()