static code analysis for octave

Markus Bergholz 4d9c8f6f85 codeformat		3 years ago
README.md	update readme	3 years ago
about.m	regexp hell	3 years ago
depfun.m	add depfun as a wrapper	3 years ago
mlint.m	codeformat	3 years ago

in progress …

segfault warning

Currently it can happen that the PCRE1 library that is used by regexp() dies in a stack overflow. Octave will segfault then.

One workaround is to compile pcre with option --disable-stack-for-recursion, but that slows performance. Another workaround is to increase the max stacksize of your system.
A long term fix will be an improvement of the used regular expressions (_it’s not perfect yet….identify variables is a huge challenge (for me))
A much longer term fix will be to port octave using PCRE2

mlint example

testfile.m

m = rand(100,100);
[index,auchdas] = find(m < 0.1);
if isempty(index)
disp('not found')
end


>> mlint('testfile.m')
testfile.m:2:2 - variable shadows function: index -> /usr/share/octave/4.0.0/m/strings/index.m
testfile.m:2:8 - variable is assigned but never used: auchdas
testfile.m:3:12 - variable shadows function: index -> /usr/share/octave/4.0.0/m/strings/index.m
testfile.m:4 - code indent is 0, should be at least 2
>> mlint('testfile.m', 'CodeIndent', false)
testfile.m:2:2 - variable shadows function: index -> /usr/share/octave/4.0.0/m/strings/index.m
testfile.m:2:8 - variable is assigned but never used: auchdas
testfile.m:3:12 - variable shadows function: index -> /usr/share/octave/4.0.0/m/strings/index.m

about()

octave:1> about('mlint.m')
ans =

  scalar structure containing the fields:

    functions = 
    {
      [1,1] = about @ /home/markus/git/OctaveCodeAnalyzer/depfun/about.m
      [2,1] = all @ libinterp/corefcn/data.cc
      [3,1] = cell2mat @ /usr/share/octave/4.0.0/m/general/cell2mat.m
      [4,1] = cellfun @ libinterp/corefcn/cellfun.cc
      [5,1] = fclose @ libinterp/corefcn/file-io.cc
      [6,1] = fgetl @ libinterp/corefcn/file-io.cc
      [7,1] = find @ libinterp/corefcn/find.cc
      [8,1] = fopen @ libinterp/corefcn/file-io.cc
      [9,1] = fprintf @ libinterp/corefcn/file-io.cc
      [10,1] = ischar @ libinterp/corefcn/strfns.cc
      [11,1] = isempty @ libinterp/corefcn/data.cc
      [12,1] = length @ libinterp/corefcn/data.cc
      [13,1] = regexp @ libinterp/corefcn/regexp.cc
      [14,1] = strcmp @ libinterp/corefcn/strfns.cc
      [15,1] = strfind @ libinterp/corefcn/strfind.cc
      [16,1] = strtrim @ /usr/share/octave/4.0.0/m/strings/strtrim.m
      [17,1] = struct @ libinterp/octave-value/ov-struct.cc
      [18,1] = uint8 @ libinterp/octave-value/ov-uint8.cc
    }
    shadowed = {}(0x0)
    variables =

      scalar structure containing the fields:

        used = 
        {
          [1,1] = current
          [1,2] = fid
          [1,3] = funcs
          [1,4] = ind
          [1,5] = lineOfCode
          [1,6] = lint
          [1,7] = lint.indent
          [1,8] = lint.indent.is
          [1,9] = lint.indent.next
          [1,10] = lint.indent.size
          [1,11] = lint.line
          [1,12] = next
          [1,13] = pos
          [1,14] = ret
          [1,15] = wasLineBreak
          [1,16] =  currentIndentFactor
        }
        overloaded = {}(0x0)

    linesOfCode =

      scalar structure containing the fields:

        blanklines =                   21
        total =                  114
        comments =                   11
        code =                   82


>> about('../../go-redis/inst/redis.m')
ans =

  scalar structure containing the fields:

    functions =
    {
      [1,1] = NaN @ libinterp/corefcn/data.cc
      [2,1] = OCTAVE_VERSION @ libinterp/corefcn/defaults.cc
      [3,1] = and @ libinterp/corefcn/data.cc
      [4,1] = any @ libinterp/corefcn/data.cc
      [5,1] = assert @ /usr/share/octave/4.0.0/m/testfun/assert.m
      [6,1] = builtin @ libinterp/parse-tree/oct-parse.cc
      [7,1] = cell @ libinterp/octave-value/ov-cell.cc
      [8,1] = cell2mat @ /usr/share/octave/4.0.0/m/general/cell2mat.m
      [9,1] = cellfun @ libinterp/corefcn/cellfun.cc
      [10,1] = char @ libinterp/corefcn/strfns.cc
      [11,1] = delete @ /usr/share/octave/4.0.0/m/miscellaneous/delete.m
      [12,1] = disp @ libinterp/corefcn/pr-output.cc
      [13,1] = error @ libinterp/corefcn/error.cc
      [14,1] = exist @ libinterp/corefcn/variables.cc
      [15,1] = fclose @ libinterp/corefcn/file-io.cc
      [16,1] = file @ variable
      [17,1] = fopen @ libinterp/corefcn/file-io.cc
      [18,1] = fread @ libinterp/corefcn/file-io.cc
      [19,1] = get @ libinterp/corefcn/graphics.cc
      [20,1] = index @ /usr/share/octave/4.0.0/m/strings/index.m
      [21,1] = input @ libinterp/corefcn/input.cc
      [22,1] = inputname @ /usr/share/octave/4.0.0/m/miscellaneous/inputname.m
      [23,1] = iscell @ libinterp/octave-value/ov-cell.cc
      [24,1] = ischar @ libinterp/corefcn/strfns.cc
      [25,1] = isempty @ libinterp/corefcn/data.cc
      [26,1] = isnan @ libinterp/corefcn/mappers.cc
      [27,1] = isnumeric @ libinterp/corefcn/data.cc
      [28,1] = isspace @ libinterp/corefcn/mappers.cc
      [29,1] = load @ libinterp/corefcn/load-save.cc
      [30,1] = max @ libinterp/corefcn/max.cc
      [31,1] = meshgrid @ /usr/share/octave/4.0.0/m/plot/util/meshgrid.m
      [32,1] = min @ libinterp/corefcn/max.cc
      [33,1] = nargin @ libinterp/octave-value/ov-usr-fcn.cc
      [34,1] = num2str @ /usr/share/octave/4.0.0/m/general/num2str.m
      [35,1] = numel @ libinterp/corefcn/data.cc
      [36,1] = ones @ libinterp/corefcn/data.cc
      [37,1] = or @ libinterp/corefcn/data.cc
      [38,1] = prod @ libinterp/corefcn/data.cc
      [39,1] = quit @ libinterp/corefcn/toplev.cc
      [40,1] = rand @ libinterp/corefcn/rand.cc
      [41,1] = regexp @ libinterp/corefcn/regexp.cc
      [42,1] = rename @ libinterp/corefcn/dirfns.cc
      [43,1] = reshape @ libinterp/corefcn/data.cc
      [44,1] = save @ libinterp/corefcn/load-save.cc
      [45,1] = set @ libinterp/corefcn/graphics.cc
      [46,1] = size @ libinterp/corefcn/data.cc
      [47,1] = source @ libinterp/parse-tree/oct-parse.cc
      [48,1] = sprintf @ libinterp/corefcn/file-io.cc
      [49,1] = str2double @ libinterp/corefcn/str2double.cc
      [50,1] = strcmp @ libinterp/corefcn/strfns.cc
      [51,1] = strfind @ libinterp/corefcn/strfind.cc
      [52,1] = sub2ind @ libinterp/corefcn/sub2ind.cc
      [53,1] = type @ /usr/share/octave/4.0.0/m/help/type.m
      [54,1] = varargin @
      [55,1] = zeros @ libinterp/corefcn/data.cc
    }
    shadowed =
    {
      [1,1] = get @ redis.m -> libinterp/corefcn/graphics.cc
      [2,1] = numel @ redis.m -> libinterp/corefcn/data.cc
      [3,1] = ones @ redis.m -> libinterp/corefcn/data.cc
      [4,1] = rand @ redis.m -> libinterp/corefcn/rand.cc
      [5,1] = rename @ redis.m -> libinterp/corefcn/dirfns.cc
      [6,1] = save @ redis.m -> libinterp/corefcn/load-save.cc
      [7,1] = set @ redis.m -> libinterp/corefcn/graphics.cc
      [8,1] = size @ redis.m -> libinterp/corefcn/data.cc
      [9,1] = type @ redis.m -> /usr/share/octave/4.0.0/m/help/type.m
      [10,1] = zeros @ redis.m -> libinterp/corefcn/data.cc
    }
    variables =

      scalar structure containing the fields:

        used =
        {
          [1,1] = dimension
          [1,2] = dimensionVar
          [1,3] = emptyInd
          [1,4] = fid
          [1,5] = linInd
          [1,6] = luastring
          [1,7] = n
          [1,8] = newhost
          [1,9] = newobjectHandle
          [1,10] = newport
          [1,11] = origin_dimension
          [1,12] = origin_linInd
          [1,13] = origin_pairs
          [1,14] = pairs
          [1,15] = ret
          [1,16] = ret1
          [1,17] = ret2
          [1,18] = ret3
          [1,19] = retVar
          [1,20] = self
          [1,21] = self.array_hash
          [1,22] = self.batchsize
          [1,23] = self.count
          [1,24] = self.dbnr
          [1,25] = self.gaussian_hash
          [1,26] = self.hostname
          [1,27] = self.is_cluster
          [1,28] = self.is_octave
          [1,29] = self.objectHandle
          [1,30] = self.passwd
          [1,31] = self.port
          [1,32] = self.precision
          [1,33] = self.swap
          [1,34] = self.verboseCluster
          [1,35] = tmp
          [1,36] = valueVar
          [1,37] = varname
          [1,38] = x
          [1,39] = y
        }
        overloaded = {}(0x0)

    linesOfCode =

      scalar structure containing the fields:

        blanklines =                   95
        total =                  600
        comments =                   52
        code =                  453

about this

caution, bad english follows

about about.m

determine every available function, script and constant when calling about. result is a unique list
open the file and add the file itself to a reserved list which will be sorted out from list which was created before
call _identifyFunctions(str) which tries to list all functions and variables (no comments and no strings). at this point, we can’t determine if a result is a variable or a function.
identify functions which are declared in the same file with _identifyFunktionsInFunction(str)
identify variables with _identifyVars(str). this is everything in front of a = .
in a process of elimination, functions and shadow functions are assigned.
in a 2nd process of elimination variables are assigned to “overloads a function” or not.

about mlint.m

about() is called to get a list of: functions, shadowed functions, variables and overloaded variables.
when mlint should inform about unused variables, the file is read-in at once to determine the count of matches for earch variable.
now mlint start iterating over each line. it checks first for the code indent in the current line and determine the code indent of the next line. this is done by calling CheckforIndent() two times.
the next check is for overloaded variables: CheckVariable()
the last check is for assigned but never used variables on base of what was done in step 2.

step 3, 4 and 5 are repeated for every line of code to determine the line and position where it happen.

README.md