#!/usr/bin/env lua

local arg = {...}

local options = {

word = false,

insensitive = true,

comments = false,

strings = false,

keys = false,

recursive = true,

}

local set_options = {

["-w"] = "word",

["-i"] = "insensitive",

["-c"] = "comments",

["-s"] = "strings",

["-k"] = "keys",

["-r"] = "recursive",

}

repeat

local set = false

for i, a in ipairs(arg) do

if a == "--help" then

options.help = true

set = false

break

end

if a:sub(1,1) == "-" and #a > 2 then

table.remove(arg, i)

for i = 2, #a do

table.insert(arg, "-"..a:sub(i,i))

end

set = true

break

end

if a == "--" then

table.remove(arg, i)

set = false

break

end

local is_set = set_options[a]

if is_set then

table.remove(arg, i)

options[is_set] = true

set = true

break

end

local is_unset = set_options[a:lower()]

if is_unset then

table.remove(arg, i)

options[is_unset] = false

set = true

break

end

end

until not set

if options.help or #arg == 0 then

print([[

codegrep - a grep-like tool tailored for code

codegrep is like grep, but it is able to skip comments and strings in the code.

Usage:

codegrep [-wWiIcCsS] <pattern> [files...]

Pattern is a Lua pattern, which is similar to regex but not the same,

except that '-' is taken literally.

Source files supported are: *.c *.h *.lua *.java *.grace *.js *.html

If files are not given, it recursively searches for source files in the

current directory (-r is accepted for compatibility but ignored).

Options:

-w -W Enable or disable whole-word search (default is off)

-i -I Enable or disable case-insensitive search (default is on)

-c -C Enable or disable search in comments (default is off: skips comments)

-s -S Enable or disable search in strings (default is off: skips strings)

-k -K Enable or disable search in key-like strings, that is, strings

containing only the search term (default is off: skips those strings)

Example:

codegrep -wc read

Looks for the word read in all source files of the tree, including comments.

]])

os.exit(0)

end

local pattern = arg[1]

table.remove(arg, 1)

pattern = pattern:gsub("%-", "%%-")

if options.insensitive then

pattern = pattern:lower()

end

if #arg == 0 then

for name in io.popen("find . -type f -and '(' -name '*.[ch]' -or -name '*.lua' -or -name '*.cpp' -or -name '*.hh' -or -name '*.cc' -or -name '*.java' -or -name '*.grace' -or -name '*.grc' -or -name '*.js' -or -name '*.html' -or -name '*.htm' ')'"):lines() do

table.insert(arg, name)

end

end

local function single_line_cut(line, marker, except)

local sc = line:find(marker, 1, true)

if sc then

if except then

if line:sub(sc, sc+#except-1) == except then

return line

end

end

line = line:sub(1, sc - 1)

end

return line

end

local function check_key(s)

if options.insensitive then

s = s:lower()

end

return s:match("^"..pattern.."$")

end

local function multi_line_cut(line, opener, closer)

local comment = false

local cc = 1

while true do

local mc = line:find(opener, cc, true)

if mc then

cc = line:find(closer, mc+#opener, true)

if cc then

if ((not options.keys) or not check_key(line:sub(mc+#opener, cc-1))) then

line = line:sub(1, mc - 1)..(" "):rep(cc + #closer - mc)..line:sub(cc + #closer, -1)

end

cc = cc + #closer

else

line = line:sub(1, mc - 1)

comment = true

break

end

else

break

end

end

return line, comment

end

local function in_multi_line(line, closer)

local comment = true

local cc = line:find(closer, 1, true)

if cc then

line = (" "):rep(cc + 1)..line:sub(cc + #closer, -1)

comment = false

else

line = ""

end

return line, comment

end

local function single_and_multi_cuts(line, comment, mopen, mclose, sopen, sexcept)

if comment then

line, comment = in_multi_line(line, mclose)

end

if not comment then

line = single_line_cut(line, sopen, sexcept)

line, comment = multi_line_cut(line, mopen, mclose)

end

if comment then

line = single_line_cut(line, sopen, sexcept)

end

return line, comment

end

local found_any = false

local styles = {

c = "c",

cpp = "c",

cc = "c",

hh = "c",

h = "c",

java = "c",

js = "c",

grace = "grace",

grc = "grace",

lua = "lua",

html = "html",

htm = "html",

}

for _, name in ipairs(arg) do

local file = io.open(name)

if file then

local ext = name:match("%.([^%.]*)$")

local style = styles[ext]

local comment = false

local multistring = false

local n = 1

local htmlmode = "html"

for line in file:lines() do

local orig = line

if not options.comments then

if style == "c" then

line, comment = single_and_multi_cuts(line, comment, "/*", "*/", "//")

elseif style == "grace" then

if not comment then

line = single_line_cut(line, "//")

end

elseif style == "lua" then

line, comment = single_and_multi_cuts(line, comment, "--[[", "]]", "--", "--[[")

elseif style == "html" then

if htmlmode == "html" then

if not comment then

line, comment = multi_line_cut(line, "<!--", "-->")

else

line, comment = in_multi_line(line, "-->")

end

if not comment then

local in_js

line, in_js = multi_line_cut(line, "<script", "</script>")

if in_js then

htmlmode = "js"

else

local in_css

line, in_css = multi_line_cut(line, "<style", "</style>")

if in_css then

htmlmode = "css"

end

end

end

elseif htmlmode == "js" then

line, comment = single_and_multi_cuts(line, comment, "/*", "*/", "//")

local in_js

_, in_js = in_multi_line(line, "</script>")

if not in_js then

htmlmode = "html"

end

elseif htmlmode == "css" then

line, comment = single_line_cut(line, "//")

local in_css

_, in_css = in_multi_line(line, "</style>")

if not in_css then

htmlmode = "html"

end

end

end

end

if not options.strings then

if style == "c" then

line = multi_line_cut(line, '"', '"')

line = multi_line_cut(line, "'", "'")

elseif style == "grace" then

line = multi_line_cut(line, '"', '"')

elseif style == "lua" then

if multistring then

line, multistring = in_multi_line(line, ']]')

end

if not multistring then

line = multi_line_cut(line, '"', '"')

line, multistring = multi_line_cut(line, '[[', ']]')

end

elseif style == "html" then

if htmlmode == "js" then

line = multi_line_cut(line, '"', '"')

line = multi_line_cut(line, "'", "'")

end

end

end

if options.insensitive then

line = line:lower()

end

local show = false

local starts, stops = {}, {}

for start, _, stop in line:gmatch("()("..pattern..")()") do

if (not options.word) or ((not line:sub(start-1,start-1):match("[%w_]")) and (not line:sub(stop,stop):match("[%w_]"))) then

table.insert(starts, start)

table.insert(stops, stop)

show = true

end

end

if show then

local out = { "\27[1;35m", name, " \27[1;32m", tostring(n), ":\27[0m" }

local at = 1

for i=1, #starts do

table.insert(out, orig:sub(at, starts[i]-1))

table.insert(out, "\27[1;31m")

table.insert(out, orig:sub(starts[i], stops[i]-1))

table.insert(out, "\27[0m")

at = stops[i]

end

table.insert(out, orig:sub(at, -1))

print(table.concat(out))

found_any = true

end

n = n + 1

end

end

end

os.exit(found_any and 0 or 1)