move gathering of match telementry to match_info

common/python/nomadcore/annotator.py

@@ -41,31 +41,64 @@ class Annotator(object):
         self._formatNameWidth = formatNameWidth
         self._update_annotate_format()
 
-    def annotate(self, match, line, matcher, targetStartEnd):
-        # classify match
-        full = match and match.start() == 0 and match.end() == len(line)
-        local_ignore = False
-        global_ignore = False
-        matcher_does_nothing = False
+    def match_info(self, match, line, matcher, targetStartEnd):
+        result = {
+            # raw line
+            'line': line,
+            # information about SimpleMatcher
+            'name': matcher.name if matcher.name else 'UNNAMED',
+            'defFile': matcher.defFile,
+            'defLine': matcher.defLine,
+            'matcher_does_nothing': matcher.does_nothing, # matcher without effect
+            'which_re': 'end' if targetStartEnd else 'start',
+            # classification of match
+            'match': 0, # 0 - no, 1 - partial, 2 - full
+            'coverageIgnore': 0, # 0 - no, 1 - local, 2 - global
+            # overall span of match
+            'span': [],
+            # capture groups
+            'group': [],
+        }
         if match:
-            if matcher.coverageIgnore:
-                local_ignore = True
-            elif matcher.does_nothing:
-                matcher_does_nothing = True
+            result['span'] = [match.start(), match.end()]
+            if match.start() == 0 and match.end() == len(line):
+                result['match'] = 2 # full match
+            else:
+                result['match'] = 1 # partial match
+            if matcher.coverageIgnore: # matcher is local coverageIgnore
+                result['coverageIgnore'] = 1
+            # capture groups
+            #   todo: move this to SimpleMatcher/CompiledMatcher
+            groupname=['all'] + [ str(num+1) for num in range(match.re.groups)]
+            for name,num in match.re.groupindex.items():
+                groupname[num]=name
+            # extract capture group info
+            for groupi in range(1, match.re.groups+1):
+                # Forward compatibility with 'regex' or 're2', which support
+                # multiple captures per named group:
+                #   span: list of start/end indices
+                #   capture: list of matched strings
+                result['group'].append({'name': groupname[groupi], 'span': [],
+                                        'capture': []})
+                if match.group(groupi) is not None:
+                    result['group'][-1]['capture'].append(match.group(groupi))
+                    result['group'][-1]['span'].append(match.span(groupi))
         else:
-            m2 = self.coverageIgnore.match(line)
-            if m2:
-                global_ignore = True
-                match = m2
+            m_ci = self.coverageIgnore.match(line)
+            if m_ci:
+                result['coverageIgnore'] = 2 # global coverageIgnore matched
+                result['span'] = [m_ci.start(), m_ci.end()]
+        return result
 
+    def annotate(self, match, line, matcher, targetStartEnd):
+        minfo = self.match_info(match, line, matcher, targetStartEnd)
         # update counters
         self.counter['total'] += 1
-        if local_ignore or global_ignore:
+        if minfo['coverageIgnore']:
             self.counter['ignore'] += 1
-        elif match:
-            if full:
+        elif minfo['match'] > 1:
             self.counter['match'] += 1
-            else:
+        elif minfo['match']:
             self.counter['partial'] += 1
         else:
             self.counter['unmatched'] += 1
@@ -76,36 +109,33 @@ class Annotator(object):
 
         # setup match label
         matchlabel = ''
-        if local_ignore:
+        if minfo['coverageIgnore'] == 1:
             matchlabel = 'l_ign'
-        elif global_ignore:
+        elif minfo['coverageIgnore']:
             matchlabel = 'g_ign'
-        elif match:
-            if matcher_does_nothing:
+        elif minfo['match']:
+            if minfo['matcher_does_nothing']:
                 matchlabel='n_'
-            if not full:
+            if minfo['match']<2:
                 matchlabel += 'p_'
-            matchlabel += 'end' if targetStartEnd else 'start'
+            matchlabel += minfo['which_re']
         else:
             matchlabel = 'no'
         # highlight line
-        if local_ignore or global_ignore:
-            highlighted = self.matchHighlighter.highlight(
-                match, line, linecolor=ANSI.FG_BLUE)
-        elif matcher_does_nothing:
-            highlighted = self.matchHighlighter.highlight(
-                match, line, linecolor=ANSI.FG_MAGENTA)
-        elif match:
-            highlighted = self.matchHighlighter.highlight(match, line)
+        if minfo['coverageIgnore']:
+            highlighted = self.matchHighlighter.highlight_minfo(
+                minfo, linecolor=ANSI.FG_BLUE)
+        elif minfo['matcher_does_nothing']:
+            highlighted = self.matchHighlighter.highlight_minfo(
+                minfo, linecolor=ANSI.FG_MAGENTA)
+        elif minfo['match']:
+            highlighted = self.matchHighlighter.highlight_minfo(minfo)
         else:
             highlighted = line
-        # set matcher name
-        if matcher.name:
-            name=matcher.name[-self._formatNameWidth:]
-        else:
-            name='UNNAMED'
-        defFile=matcher.defFile[-self._formatSourceWidth:]
+        # shorted matcher- and source file names
+        name = minfo['name'][-self._formatNameWidth:]
+        defFile = minfo['defFile'][-self._formatSourceWidth:]
         self.annotateFile.write(self._annotate_format % (
-            defFile, matcher.defLine, name,
+            defFile, minfo['defLine'], name,
             matchlabel, highlighted,
         ))