[python] more UTF-8 fun

Summary:
Make sure `inferTraceBugs` works with non-ascii characters. Harden the code a
bit more on the way.

Fixes #592

Reviewed By: mbouaziz

Differential Revision: D4659016

fbshipit-source-id: 79f7a80

infer/lib/python/inferTraceBugs

@@ -123,6 +123,9 @@ class Tracer(object):
     def build_report(self, report):
         self.build_trace(report[issues.JSON_INDEX_TRACE])
 
+    def __unicode__(self):
+        return unicode(self.indenter)
+
     def __str__(self):
         return str(self.indenter)
 
@@ -278,7 +281,7 @@ def html_bug_trace(args, report, bug_id):
     bug_trace += '%s\n' % issues.text_of_report(report)
     tracer = Tracer(args)
     tracer.build_report(report)
-    bug_trace += str(tracer)
+    bug_trace += unicode(tracer)
     return bug_trace
 
 
@@ -389,7 +392,7 @@ def main():
 
     tracer = Tracer(args, max_level)
     tracer.build_report(report)
-    utils.stdout(str(tracer))
+    utils.stdout(unicode(tracer))
 
 
 if __name__ == '__main__':

infer/lib/python/inferlib/source.py

@@ -13,14 +13,14 @@ from __future__ import unicode_literals
 import codecs
 import os
 
-from . import colorize, config
+from . import colorize, config, utils
 
 BASE_INDENT = 2
 # how many lines of context around each report
 SOURCE_CONTEXT = 2
 
 
-class Indenter(str):
+class Indenter(unicode):
     def __init__(self):
         super(Indenter, self).__init__()
         self.text = ''
@@ -46,7 +46,7 @@ class Indenter(str):
 
     def add(self, x):
         if type(x) != unicode:
-            x = x.decode(config.CODESET)
+            x = utils.decode(x)
         lines = x.splitlines()
         indent = self.indent_get()
         lines = [indent + l for l in lines]
@@ -57,7 +57,7 @@ class Indenter(str):
         return self.text
 
     def __str__(self):
-        return unicode(self).encode(config.CODESET)
+        return utils.encode(unicode(self))
 
 
 def build_source_context(source_name, mode, report_line):

infer/tests/build_systems/utf8_in_procname/Makefile

@@ -17,11 +17,16 @@ SOURCES = ../codetoanalyze/make/utf8_in_function_names.c
 
 include $(TESTS_DIR)/clang.make
 
-infer-out/report.json: $(CLANG_DEPS) $(SOURCES) $(HEADERS)
+infer-out/report.json: $(CLANG_DEPS) $(INFERTRACEBUGS_BIN) $(SOURCES) $(HEADERS) $(MAKEFILE_LIST)
 #	set non-utf8-supporting locale
 	LC_ALL=C; \
 	$(call silent_on_success,\
 	  $(INFER_BIN) --check-duplicate-symbols $(INFER_OPTIONS) -a $(ANALYZER) -- clang $(CLANG_OPTIONS) $(SOURCES) 2>duplicates.txt)
+	grep "DUPLICATE_SYMBOLS" duplicates.txt; test $$? -ne 0
+#	make sure inferTraceBugs is immune to UTF-8
+	$(call silent_on_success, $(INFERTRACEBUGS_BIN) --max-level max --select 0)
+	$(call silent_on_success, $(INFERTRACEBUGS_BIN) --html)
+	[ -f infer-out/report.html/index.html ]
 #	run again to check that infer manages to delete the results directory
 	LC_ALL=C; \
 	$(call silent_on_success,\