bin/find-can-be-private-symbols.py

   1 #!/usr/bin/python3
   2 #
   3 # Find exported symbols that can be made non-exported.
   4 #
   5 # Noting that (a) parsing these commands is a pain, the output is quite irregular and (b) I'm fumbling in the
   6 # dark here, trying to guess what exactly constitutes an "import" vs an "export" of a symbol, linux linking
   7 # is rather complex.
   8 #
   9 # Takes about 5min to run on a decent machine.
  10 #
  11 # The standalone function analysis is reasonable reliable, but the class/method analysis is less so
  12 #   (something to do with destructor thunks not showing up in my results?)
  13 #
  14 # Also, the class/method analysis will not catch problems like
  15 #    'dynamic_cast from 'Foo' with hidden type visibility to 'Bar' with default type visibility'
  16 #    but loplugin:dyncastvisibility will do that for you
  17 #
  18
  19 import subprocess
  20 import sys
  21 import re
  22
  23 exported_symbols1 = set()
  24 imported_symbols1 = set()
  25 exported_symbols2 = set() # decoded
  26 imported_symbols2 = set() # decoded
  27 # all names that exist in the source code
  28 #all_source_names = set()
  29
  30
  31 #subprocess_find_all_source_names = subprocess.Popen("git grep -oh -P '\\b\\w\\w\\w+\\b' -- '*.h*' | sort -u",
  32 #    stdout=subprocess.PIPE, stderr=subprocess.STDOUT, shell=True)
  33 #with subprocess_find_all_source_names.stdout as txt:
  34 #    for line in txt:
  35 #        line = line.strip()
  36 #        all_source_names.add(line)
  37 #subprocess_find_all_source_names.terminate()
  38
  39 # find all our shared libs
  40 subprocess_find = subprocess.Popen("find ./instdir -name *.so && find ./workdir/LinkTarget/CppunitTest -name *.so",
  41     stdout=subprocess.PIPE, shell=True)
  42 with subprocess_find.stdout as txt:
  43     for line in txt:
  44         sharedlib = line.strip()
  45         # look for exported symbols
  46         subprocess_nm = subprocess.Popen(b"nm -D " + sharedlib, stdout=subprocess.PIPE, shell=True)
  47         with subprocess_nm.stdout as txt2:
  48             # We are looking for lines something like:
  49             # 0000000000036ed0 T flash_component_getFactory
  50             line_regex = re.compile(r'^[0-9a-fA-F]+ T ')
  51             for line2_bytes in txt2:
  52                 line2 = line2_bytes.strip().decode("utf-8")
  53                 if line_regex.match(line2):
  54                     sym = line2.split(" ")[2].strip()
  55                     exported_symbols1.add(sym)
  56         subprocess_nm.terminate()
  57         # look for imported symbols
  58         subprocess_objdump = subprocess.Popen(b"objdump -T " + sharedlib, stdout=subprocess.PIPE, shell=True)
  59         with subprocess_objdump.stdout as txt2:
  60             # ignore some header bumpf
  61             txt2.readline()
  62             txt2.readline()
  63             txt2.readline()
  64             txt2.readline()
  65             # We are looking for lines something like:
  66             # 0000000000000000      DF *UND*  0000000000000000     _ZN16FilterConfigItem10WriteInt32ERKN3rtl8OUStringEi
  67             for line2_bytes in txt2:
  68                 line2 = line2_bytes.strip().decode("utf-8")
  69                 if not("*UND*"in line2): continue
  70                 tokens = line2.split(" ")
  71                 sym = tokens[len(tokens)-1].strip()
  72                 imported_symbols1.add(sym)
  73         subprocess_objdump.terminate()
  74 subprocess_find.terminate()
  75
  76 # look for imported symbols in executables
  77 subprocess_find = subprocess.Popen("find ./instdir -name *.bin", stdout=subprocess.PIPE, shell=True)
  78 with subprocess_find.stdout as txt:
  79     for line in txt:
  80         executable = line.strip()
  81         # look for exported symbols
  82         subprocess_nm = subprocess.Popen(b"nm -D " + executable + b" | grep -w U", stdout=subprocess.PIPE, shell=True)
  83         with subprocess_nm.stdout as txt2:
  84             # We are looking for lines something like:
  85             # U sal_detail_deinitialize
  86             for line2_bytes in txt2:
  87                 line2 = line2_bytes.strip().decode("utf-8")
  88                 sym = line2.split(" ")[1]
  89                 imported_symbols1.add(sym)
  90 subprocess_find.terminate()
  91
  92 #progress = 0;
  93 #for sym in sorted(imported_symbols - exported_symbols):
  94 #    progress += 1
  95 #    if (progress % 128 == 0): print( str(int(progress * 100 / len(diff))) + "%")
  96 #    filtered_sym = subprocess.check_output(["c++filt", sym]).strip().decode("utf-8")
  97 #    if filtered_sym.startswith("non-virtual thunk to "): filtered_sym = filtered_sym[21:]
  98 #    elif filtered_sym.startswith("virtual thunk to "): filtered_sym = filtered_sym[17:]
  99 #    print("Symbol imported but not exported? " + filtered_sym)
 100
 101 # Now we have to symbolize before comparing because sometimes (due to thunks) two
 102 # different encoded names symbolize to the same method/func name
 103 #
 104 progress = 0;
 105 progress_max_len = len(imported_symbols1) + len(exported_symbols1)
 106 for sym in imported_symbols1:
 107     progress += 1
 108     if (progress % 128 == 0): print( str(int(progress * 100 / progress_max_len)) + "%")
 109     filtered_sym = subprocess.check_output(["c++filt", sym]).strip().decode("utf-8")
 110     if filtered_sym.startswith("non-virtual thunk to "): filtered_sym = filtered_sym[21:]
 111     elif filtered_sym.startswith("virtual thunk to "): filtered_sym = filtered_sym[17:]
 112     imported_symbols2.add(filtered_sym)
 113 progress = 0;
 114 for sym in exported_symbols1:
 115     progress += 1
 116     if (progress % 128 == 0): print( str(int(progress * 100 / progress_max_len)) + "%")
 117     filtered_sym = subprocess.check_output(["c++filt", sym]).strip().decode("utf-8")
 118     if filtered_sym.startswith("non-virtual thunk to "): filtered_sym = filtered_sym[21:]
 119     elif filtered_sym.startswith("virtual thunk to "): filtered_sym = filtered_sym[17:]
 120     exported_symbols2.add(filtered_sym)
 121
 122 unused_exports = exported_symbols2 - imported_symbols2
 123 print("exported       = " + str(len(exported_symbols2)))
 124 print("imported       = " + str(len(imported_symbols2)))
 125 print("unused_exports = " + str(len(unused_exports)))
 126
 127 #def extractFunctionNameFromSignature(sym):
 128 #    i = sym.find("(")
 129 #    if i == -1: return sym
 130 #    return sym[:i]
 131
 132 with open("bin/find-can-be-private-symbols.functions.results", "wt") as f:
 133     for sym in sorted(unused_exports):
 134         # Filter out most of the noise.
 135         # No idea where these are coming from, but not our code.
 136         if sym.startswith("CERT_"): continue
 137         elif sym.startswith("DER_"): continue
 138         elif sym.startswith("FORM_"): continue
 139         elif sym.startswith("FPDF"): continue
 140         elif sym.startswith("HASH_"): continue
 141         elif sym.startswith("Hunspell_"): continue
 142         elif sym.startswith("LL_"): continue
 143         elif sym.startswith("LP_"): continue
 144         elif sym.startswith("LU"): continue
 145         elif sym.startswith("MIP"): continue
 146         elif sym.startswith("MPS"): continue
 147         elif sym.startswith("NSS"): continue
 148         elif sym.startswith("NSC_"): continue
 149         elif sym.startswith("PK11"): continue
 150         elif sym.startswith("PL_"): continue
 151         elif sym.startswith("PQ"): continue
 152         elif sym.startswith("PBE_"): continue
 153         elif sym.startswith("PORT_"): continue
 154         elif sym.startswith("PRP_"): continue
 155         elif sym.startswith("PR_"): continue
 156         elif sym.startswith("PT_"): continue
 157         elif sym.startswith("QS_"): continue
 158         elif sym.startswith("REPORT_"): continue
 159         elif sym.startswith("RSA_"): continue
 160         elif sym.startswith("SEC"): continue
 161         elif sym.startswith("SGN"): continue
 162         elif sym.startswith("SOS"): continue
 163         elif sym.startswith("SSL_"): continue
 164         elif sym.startswith("VFY_"): continue
 165         elif sym.startswith("_PR_"): continue
 166         elif sym.startswith("ber_"): continue
 167         elif sym.startswith("bfp_"): continue
 168         elif sym.startswith("ldap_"): continue
 169         elif sym.startswith("ne_"): continue
 170         elif sym.startswith("opj_"): continue
 171         elif sym.startswith("pg_"): continue
 172         elif sym.startswith("pq"): continue
 173         elif sym.startswith("presolve_"): continue
 174         elif sym.startswith("sqlite3_"): continue
 175         elif sym.startswith("libepubgen::"): continue
 176         elif sym.startswith("lucene::"): continue
 177         elif sym.startswith("Hunspell::"): continue
 178         elif sym.startswith("sk_"): continue
 179         elif sym.startswith("_Z"): continue
 180         # dynamically loaded
 181         elif sym.endswith("get_implementation"): continue
 182         elif sym.endswith("component_getFactory"): continue
 183         elif sym == "CreateDialogFactory": continue
 184         elif sym == "CreateUnoWrapper": continue
 185         elif sym == "ExportDOC": continue
 186         elif sym == "ExportRTF": continue
 187         elif sym == "GetSaveWarningOfMSVBAStorage_ww8": continue
 188         elif sym == "GetSpecialCharsForEdit": continue
 189         elif sym.startswith("Import"): continue
 190         elif sym.startswith("Java_com_sun_star_"): continue
 191         elif sym.startswith("TestImport"): continue
 192         elif sym.startswith("getAllCalendars_"): continue
 193         elif sym.startswith("getAllCurrencies_"): continue
 194         elif sym.startswith("getAllFormats"): continue
 195         elif sym.startswith("getBreakIteratorRules_"): continue
 196         elif sym.startswith("getCollationOptions_"): continue
 197         elif sym.startswith("getCollatorImplementation_"): continue
 198         elif sym.startswith("getContinuousNumberingLevels_"): continue
 199         elif sym.startswith("getDateAcceptancePatterns_"): continue
 200         elif sym.startswith("getForbiddenCharacters_"): continue
 201         elif sym.startswith("getIndexAlgorithm_"): continue
 202         elif sym.startswith("getLCInfo_"): continue
 203         elif sym.startswith("getLocaleItem_"): continue
 204         elif sym.startswith("getOutlineNumberingLevels_"): continue
 205         elif sym.startswith("getReservedWords_"): continue
 206         elif sym.startswith("getSTC_"): continue
 207         elif sym.startswith("getSearchOptions_"): continue
 208         elif sym.startswith("getTransliterations_"): continue
 209         elif sym.startswith("getUnicodeScripts_"): continue
 210         elif sym.startswith("lok_"): continue
 211         # UDK API
 212         elif sym.startswith("osl_"): continue
 213         elif sym.startswith("rtl_"): continue
 214         elif sym.startswith("typelib_"): continue
 215         elif sym.startswith("typereg_"): continue
 216         elif sym.startswith("uno_"): continue
 217         # remove things we found that do not exist in our source code, they're not ours
 218         #if not(extractFunctionNameFromSignature(sym) in all_source_names): continue
 219         f.write(sym + "\n")
 220