bin/find-unused-defines.py

   1 #!/usr/bin/python3
   2
   3 # Search for unused constants in header files.
   4 #
   5 # Note that sometimes these constants are calculated, so some careful checking of the output is necessary.
   6 #
   7 # Takes about 4 hours to run this on a fast machine with an SSD
   8 #
   9
  10 import subprocess
  11 import sys
  12 import re
  13 import codecs
  14
  15 exclusionSet = set([
  16     # List of RID constants where we compute a value using a base before calling one of the RESSTR methods
  17     # Found with: git grep -P 'RID_\w+\s*\+' -- :/ ':!*.hrc' ':!*.src' ':!*.java' ':!*.py' ':!*.xba'
  18     "RID_SVXSTR_KEY_",
  19     "RID_UPDATE_BUBBLE_TEXT_",
  20     "RID_UPDATE_BUBBLE_T_TEXT_",
  21     "RID_SVXSTR_TBLAFMT_",
  22     "RID_BMP_CONTENT_",
  23     "RID_DROPMODE_",
  24     "RID_BMP_LEVEL",
  25     "RID_SVXSTR_BULLET_DESCRIPTION",
  26     "RID_SVXSTR_SINGLENUM_DESCRIPTION",
  27     "RID_SVXSTR_OUTLINENUM_DESCRIPTION",
  28     "RID_SVXSTR_RULER_",
  29     "RID_GALLERYSTR_THEME_",
  30     "RID_SVXSTR_BULLET_DESCRIPTION",
  31     "RID_SVXSTR_SINGLENUM_DESCRIPTION",
  32     "RID_SVXSTR_OUTLINENUM_DESCRIPTION",
  33     # doing some weird stuff in svx/source/unodraw/unoprov.cxx involving mapping of UNO api names to translated names and back again
  34     "RID_SVXSTR_GRDT",
  35     "RID_SVXSTR_HATCH",
  36     "RID_SVXSTR_BMP",
  37     "RID_SVXSTR_DASH",
  38     "RID_SVXSTR_LEND",
  39     "RID_SVXSTR_TRASNGR",
  40     # other places doing calculations
  41     "RID_SVXSTR_DEPTH",
  42     "RID_SUBSETSTR_",
  43     "ANALYSIS_",
  44     "FLD_DOCINFO_CHANGE",
  45     "FLD_EU_",
  46     "FLD_INPUT_",
  47     "FLD_PAGEREF_",
  48     "FLD_STAT_",
  49     "FMT_AUTHOR_",
  50     "FMT_CHAPTER_",
  51     "FMT_DBFLD_",
  52     "FMT_FF_",
  53     "FMT_GETVAR_",
  54     "FMT_MARK_",
  55     "FMT_REF_",
  56     "FMT_SETVAR_",
  57     "STR_AUTH_FIELD_ADDRESS_",
  58     "STR_AUTH_TYPE_",
  59     "STR_AUTOFMTREDL_",
  60     "STR_CONTENT_TYPE_",
  61     "STR_UPDATE_ALL",
  62     "STR_UPDATE_INDEX",
  63     "STR_UPDATE_LINK",
  64     "BMP_PLACEHOLDER_",
  65     "STR_RPT_HELP_",
  66     "STR_TEMPLATE_NAME",
  67     "UID_BRWEVT_",
  68     "HID_EVT_",
  69     "HID_PROP_",
  70     "STR_VOBJ_MODE_",
  71     "STR_COND_",
  72     "SCSTR_CONTENT_",
  73     "DATE_FUNCDESC_",
  74     "DATE_FUNCNAME_",
  75     "DATE_DEFFUNCNAME_",
  76     "PRICING_DEFFUNCNAME_",
  77     "PRICING_FUNCDESC_",
  78     "PRICING_FUNCNAME_",
  79     "STR_ItemValCAPTION",
  80     "STR_ItemValCIRC",
  81     "STR_ItemValEDGE",
  82     "STR_ItemValFITTOSIZE",
  83     "STR_ItemValMEASURE_",
  84     "STR_ItemValMEASURETEXT_",
  85     "STR_ItemValTEXTANI_",
  86     "STR_ItemValTEXTHADJ",
  87     "STR_ItemValTEXTVADJ",
  88     "RID_SVXITEMS_VERJUST",
  89     "RID_SVXITEMS_ORI",
  90     "RID_SVXITEMS_JUSTMETHOD",
  91     "RID_SVXITEMS_HORJUST",
  92     "MM_PART",
  93     ])
  94
  95
  96 def in_exclusion_set( a ):
  97     for f in exclusionSet:
  98         if a.startswith(f):
  99             return True
 100     return False
 101
 102
 103 # Hack to turn off unicode decoding errors, which sometimes happens in the output and I really don't
 104 # care
 105 def strict_handler(exception):
 106     return u"", exception.end
 107 codecs.register_error("strict", strict_handler)
 108
 109 # find defines, excluding the externals folder
 110 a = subprocess.Popen("git grep -hP '^#define\\s+\\w\\w\\w\\w+\\s*' -- \"[!e][!x][!t]*\" | sort -u", stdout=subprocess.PIPE, shell=True, encoding='UTF-8')
 111
 112 name_re = re.compile(r"#define\s+(\w+)")
 113 with a.stdout as txt:
 114     for line in txt:
 115         idName = name_re.match(str(line)).group(1)
 116         if idName.startswith("INCLUDED_"):
 117             continue
 118         # the various _START and _END constants are normally unused outside of the .hrc and .src files, and that's fine
 119         if idName.endswith("_START"):
 120             continue
 121         if idName.endswith("_BEGIN"):
 122             continue
 123         if idName.endswith("_END"):
 124             continue
 125         if idName == "RID_SVX_FIRSTFREE":
 126             continue
 127         if idName == "":
 128             continue
 129         if idName.startswith("__com"):
 130             continue # these are the include/header macros for the UNO stuff
 131         if in_exclusion_set(idName):
 132             continue
 133         # search for the constant
 134         b = subprocess.Popen(["git", "grep", "-w", idName], stdout=subprocess.PIPE, encoding='UTF-8')
 135         found_reason_to_exclude = False
 136         with b.stdout as txt2:
 137             cnt = 0
 138             for line2 in txt2:
 139                 line2 = line2.strip() # otherwise the comparisons below will not work
 140                 # ignore if/undef magic, does not indicate an actual use (most of the time)
 141                 if "ifdef" in line2:
 142                     continue
 143                 if "undef" in line2:
 144                     continue
 145                 # ignore commented out code
 146                 if line2.startswith("//"):
 147                     continue
 148                 if line2.startswith("/*"):
 149                     continue
 150                 # check if we found one in actual code
 151                 if idName.startswith("SID_"):
 152                     if ".hrc:" not in line2 and ".src:" not in line2 and ".sdi:" not in line2:
 153                         found_reason_to_exclude = True
 154                 else:
 155                     if ".hrc:" not in line2 and ".src:" not in line2:
 156                         found_reason_to_exclude = True
 157                 if idName.startswith("RID_"):
 158                         # is the constant being used as an identifier by entries in .src files?
 159                         if ".src:" in line2 and "Identifier = " in line2:
 160                             found_reason_to_exclude = True
 161                         # is the constant being used by the property controller extension or reportdesigner inspection,
 162                         # which use macros to declare constants, hiding them from a search
 163                         if "extensions/source/propctrlr" in line2:
 164                             found_reason_to_exclude = True
 165                         if "reportdesign/source/ui/inspection/inspection.src" in line2:
 166                             found_reason_to_exclude = True
 167                 if idName.startswith("HID_"):
 168                         # is the constant being used as an identifier by entries in .src files
 169                         if ".src:" in line2 and "HelpId = " in line2:
 170                             found_reason_to_exclude = True
 171                 # is it being used as a constant in an ItemList  in .src files?
 172                 if ".src:" in line2 and (";> ;" in line2 or "; >;" in line2):
 173                     found_reason_to_exclude = True
 174                 # these are used in calculations in other .hrc files
 175                 if "sw/inc/rcid.hrc:" in line2:
 176                     found_reason_to_exclude = True
 177                 # calculations
 178                 if "sw/source/uibase/inc/ribbar.hrc:" in line2 and "ST_" in idName:
 179                     found_reason_to_exclude = True
 180                 if "sw/source/uibase/inc/ribbar.hrc:" in line2 and "STR_IMGBTN_" in idName:
 181                     found_reason_to_exclude = True
 182                 if "sw/source/core/undo/undo.hrc:" in line2:
 183                     found_reason_to_exclude = True
 184                 if "sw/inc/poolfmt.hrc:" in line2:
 185                     found_reason_to_exclude = True
 186                 # used via a macro that hides them from search
 187                 if "dbaccess/" in line2 and idName.startswith("PROPERTY_ID_"):
 188                     found_reason_to_exclude = True
 189                 if "reportdesign/" in line2 and idName.startswith("HID_RPT_PROP_"):
 190                     found_reason_to_exclude = True
 191                 if "reportdesign/" in line2 and idName.startswith("RID_STR_"):
 192                     found_reason_to_exclude = True
 193                 if "forms/" in line2 and idName.startswith("PROPERTY_"):
 194                     found_reason_to_exclude = True
 195                 if "svx/source/tbxctrls/extrusioncontrols.hrc:" in line2 and idName.startswith("DIRECTION_"):
 196                     found_reason_to_exclude = True
 197                 if "svx/source/tbxctrls/extrusioncontrols.hrc:" in line2 and idName.startswith("FROM_"):
 198                     found_reason_to_exclude = True
 199                 # if we see more than a few lines then it's probably one of the BASE/START/BEGIN things
 200                 cnt = cnt + 1
 201                 if cnt > 2:
 202                     found_reason_to_exclude = True
 203         if not found_reason_to_exclude:
 204             print(idName)
 205             # otherwise the previous line of output will be incorrectly mixed into the below git output, because of buffering
 206             sys.stdout.flush()
 207             # search again, so we log the location and filename of stuff we want to remove
 208             subprocess.call(["git", "grep", "-wn", idName])
 209