hphp/parser/make-parser.sh

   1 #!/bin/bash
   2
   3 unset CDPATH
   4 DIR="$( cd "$( dirname "$0" )" && pwd )"
   5
   6 # If we're using buck, then we'll be in a sandboxed source directory instead of
   7 # the repo.  We want the path to the repo so we can check in the generated
   8 # parser artifacts.
   9 if [ -n "${FBCODE_DIR}" ]; then
  10   DIR="${FBCODE_DIR}/hphp/parser"
  11 fi
  12
  13 INFILE=hphp.y
  14
  15 if [ -z "${INSTALL_DIR}" ]; then
  16   INSTALL_DIR="${DIR}"
  17 fi
  18
  19 BISON=$(which bison)
  20 if [ ! -x "${BISON}" ]; then
  21   echo "bison not found" 1>&2
  22   exit 1
  23 fi
  24
  25 OUTFILE5=${INSTALL_DIR}/hphp.5.tab.cpp
  26 OUTFILE7=${INSTALL_DIR}/hphp.7.tab.cpp
  27 OUTHEADER5=${INSTALL_DIR}/hphp.5.tab.hpp
  28 OUTHEADER7=${INSTALL_DIR}/hphp.7.tab.hpp
  29 OUTHEADER=${INSTALL_DIR}/hphp.tab.hpp
  30
  31 BISON_OPTS="--verbose --locations -d"
  32
  33 TMP=$(mktemp)
  34
  35 # Use a potentially-scary awk script to split the single hphp.y parser into two
  36 # outputs.  It's actually not that bad.
  37 #  - The idea is that anything between a /* !PHP5_ONLY */ line and a /* !END */
  38 #    line are removed from the PHP7 parser, and vice versa for PHP5.
  39 #  - This is done just by looking for those tokens and setting the "flag".
  40 #  - We make sure to still print things for the removed lines, to keep line
  41 #    numbers consistent to ease debugging.
  42
  43 awk \
  44   '/!END/{flag=0} flag{print "/* REMOVED */"} /!PHP7_ONLY/{print; flag=1} !flag' \
  45   "${INFILE}" > "${TMP}"
  46 # shellcheck disable=SC2086
  47 "${BISON}" ${BISON_OPTS} -pCompiler5 "-o${OUTFILE5}" "${TMP}"
  48 if [ $? -ne 0 ]; then
  49   echo "Bison failed to compile PHP5 parser!"
  50   exit 1
  51 fi
  52
  53 awk \
  54   '/!END/{flag=0} flag{print "/* REMOVED */"} /!PHP5_ONLY/{print; flag=1} !flag' \
  55   "${INFILE}" > "${TMP}"
  56 # shellcheck disable=SC2086
  57 "${BISON}" ${BISON_OPTS} -pCompiler7 "-o${OUTFILE7}" "${TMP}"
  58 if [ $? -ne 0 ] ; then
  59   echo "Bison failed to compile PHP7 parser!"
  60   exit 1
  61 fi
  62
  63 rm "${TMP}"
  64
  65 # Remove alpha variance in "#line" directives.
  66 if [ "${INSTALL_DIR}" = "${DIR}" ]; then
  67   sed -i -e "s#${TMP}#${INFILE}#g" -e "s#${DIR}/##g" "${OUTFILE5}"
  68   sed -i -e "s#${TMP}#${INFILE}#g" -e "s#${DIR}/##g" "${OUTFILE7}"
  69 else
  70   sed -i "s#${TMP}#${INFILE}#g" "${OUTFILE5}"
  71   sed -i "s#${TMP}#${INFILE}#g" "${OUTFILE7}"
  72 fi
  73
  74 sed -i \
  75     -e 's@int Compiler[57]parse.*@@' \
  76     -e 's@.*int Compiler[57]debug.*@@' \
  77     -e "s@#ifndef YY_COMPILER[57]_.*@@g" \
  78     -e "s@# define YY_COMPILER[57]_.*@@g" \
  79     -e "s@#endif /\* !YY_COMPILER[57]_.*@@g" \
  80     "${OUTHEADER5}" "${OUTHEADER7}"
  81
  82 cmp "${OUTHEADER5}" "${OUTHEADER7}"
  83 if [ $? -ne 0 ]; then
  84   echo "PHP5 and PHP7 headers differ, must be the same tokens"
  85   exit 1
  86 fi
  87 cp "${OUTHEADER5}" "${OUTHEADER}"
  88
  89 # Lots of our logic relies on knowing the shape of the token table.  Sadly it is
  90 # an enum without introspection, so instead make it macros so we can control its
  91 # shape on re-requires of the .hpp file.
  92 sed -i -r \
  93     -e 's/(T_\w+)\s+=\s+([0-9]+)\s*,?/YYTOKEN(\2, \1)/g' \
  94     -e "s/\s+enum\s+yytokentype/#ifndef YYTOKEN_MAP\n#define YYTOKEN_MAP enum yytokentype\n#define YYTOKEN(num, name) name = num,\n#endif\n   YYTOKEN_MAP/" \
  95     "${OUTHEADER}"
  96
  97 # Remove the include guard's #endif (-e doesn't work for this).
  98 sed -i '$ d' "${OUTHEADER}"
  99
 100 # We don't want to rely on the grammar to have a fixed start and end token, so
 101 # let's parse the file and make two macros for the min and max.
 102 TOKEN_MIN=$(grep "^\s\+YYTOKEN(" "${OUTHEADER}" -m 1 | \
 103   sed -r -e 's/\s+YYTOKEN.([0-9]+).*/#ifndef YYTOKEN_MIN\n#define YYTOKEN_MIN \1\n#endif/')
 104 TOKEN_MAX=$(grep "^\s\+YYTOKEN(" "${OUTHEADER}" | tail -n 1 | \
 105   sed -r -e 's/\s+YYTOKEN.([0-9]+).*/#ifndef YYTOKEN_MAX\n#define YYTOKEN_MAX \1\n#endif/')
 106
 107 echo -e "${TOKEN_MIN}\n\n${TOKEN_MAX}" >> "${OUTHEADER}"
 108
 109 # We still want the files in our tree since they are checked in.
 110 if [ "${INSTALL_DIR}" != "${DIR}" ]; then
 111   sed -i -e "1i// @""generated" "${OUTHEADER}"
 112   cp "${OUTHEADER}" "${DIR}/hphp.tab.hpp"
 113 fi