From: Xavier Deguillard <xavierd@fb.com>
Date: Tue, 26 Jul 2022 13:51:58 -0700
Subject: pcre: move to pcre2

Summary:
The pcre API is no longer maintained. According to pcre.org:

. The older, but still widely deployed PCRE library, originally released in
. 1997, is at version 8.45. This version of PCRE is now at end of life, and is no
. longer being actively maintained. Version 8.45 is expected to be the final
. release of the older PCRE library, and new projects should use PCRE2 instead.

Let's thus migrate to pcre2.

Reviewed By: chadaustin

Differential Revision: D38134776

fbshipit-source-id: 0d2eaeb84013cd03a7426fffa23afe9f48d57cdf

Comment: ported to older watchman by Jeremy Bicha
---
 TARGETS           |  2 +-
 configure.ac      | 18 +++++++--------
 query/pcre.cpp    | 66 ++++++++++++++++++++++++++++++++-----------------------
 watchman_system.h |  5 +++--
 4 files changed, 52 insertions(+), 39 deletions(-)

diff --git a/TARGETS b/TARGETS
index 5b3ddb3..4ec3d25 100644
--- a/TARGETS
+++ b/TARGETS
@@ -191,7 +191,7 @@ cpp_library(
 cpp_library(
     name = "pcre",
     srcs = ["query/pcre.cpp"],
-    compiler_flags = ["-DHAVE_PCRE_H"] + compiler_flags,
+    compiler_flags = ["-DHAVE_PCRE2_H"] + compiler_flags,
     link_whole = True,
     deps = [":headers"],
     external_deps = ["pcre"],
diff --git a/configure.ac b/configure.ac
index 7e809a5..a57deb2 100644
--- a/configure.ac
+++ b/configure.ac
@@ -152,8 +152,8 @@ AM_CONDITIONAL(HAVE_RUBY, [test "x$want_ruby" = "xyes"])
 
 AC_ARG_WITH(pcre, [
   --without-pcre       Don't enable pcre support.
-  --with-pcre=PATH     Enable pcre support.  PATH is location of pcre-config.
-                       Default is to enable and look for pcre-config in your
+  --with-pcre=PATH     Enable pcre support.  PATH is location of pcre2-config.
+                       Default is to enable and look for pcre2-config in your
                        $PATH
 ],[
   pcre_config="$withval"
@@ -163,10 +163,10 @@ AC_ARG_WITH(pcre, [
 
 if test "$pcre_config" != "no" ; then
   if test "$pcre_config" = "yes" -o "$pcre_config" = "check"; then
-    AC_PATH_PROG(PCRE_CONFIG_PATH, pcre-config, false)
+    AC_PATH_PROG(PCRE_CONFIG_PATH, pcre2-config, false)
     dnl If --with-pcre was specified but pcre-config not found, fail hard now.
     if test "$pcre_config" = "yes" -a "$PCRE_CONFIG_PATH" = "false"; then
-      AC_MSG_FAILURE([--with-pcre was given, but pcre-config not found in PATH])
+      AC_MSG_FAILURE([--with-pcre was given, but pcre2-config not found in PATH])
     fi
   else
     PCRE_CONFIG_PATH="$pcre_config"
@@ -181,18 +181,18 @@ if test "$pcre_config" != "no" ; then
   fi
 
   if test -n "$pcre_version"; then
-    PCRE_LIBS=`$PCRE_CONFIG_PATH --libs`
+    PCRE_LIBS=`$PCRE_CONFIG_PATH --libs8`
     LIBS="$LIBS $PCRE_LIBS"
     PCRE_CFLAGS=`$PCRE_CONFIG_PATH --cflags`
     CFLAGS="$CFLAGS $PCRE_CFLAGS"
     CPPFLAGS="$CPPFLAGS $PCRE_CFLAGS"
-    AC_CHECK_HEADERS(pcre.h, [], [
+    AC_CHECK_HEADERS(pcre2.h, [], [
       if test "$pcre_config" != "check"; then
          AC_MSG_FAILURE([--with-pcre was given, but pcre not found:
-  pcre-config --libs=$PCRE_LIBS
-  pcre-config --cflags=$PCRE_CFLAGS])
+  pcre2-config --libs8=$PCRE_LIBS
+  pcre2-config --cflags=$PCRE_CFLAGS])
       fi
-    ])
+    ], [#define PCRE2_CODE_UNIT_WIDTH 8])
   elif test "$pcre_config" != "check"; then
     AC_MSG_FAILURE([$PCRE_CONFIG_PATH failed to run, could not check for PCRE])
   fi
diff --git a/query/pcre.cpp b/query/pcre.cpp
index 60c201f..1329d3b 100644
--- a/query/pcre.cpp
+++ b/query/pcre.cpp
@@ -5,25 +5,25 @@
 
 #include "make_unique.h"
 
-#ifdef HAVE_PCRE_H
+#ifdef HAVE_PCRE2_H
 
 using watchman::CaseSensitivity;
 
 class PcreExpr : public QueryExpr {
-  pcre *re;
-  pcre_extra *extra;
+  pcre2_code *re;
+  pcre2_match_data* matchData;
   bool wholename;
 
  public:
-  explicit PcreExpr(pcre* re, pcre_extra* extra, bool wholename)
-      : re(re), extra(extra), wholename(wholename) {}
+  explicit PcreExpr(pcre2_code* re, pcre2_match_data* matchData, bool wholename)
+      : re(re), matchData(matchData), wholename(wholename) {}
 
   ~PcreExpr() override {
     if (re) {
-      pcre_free(re);
+      pcre2_code_free(re);
     }
-    if (extra) {
-      pcre_free(extra);
+    if (matchData) {
+      pcre2_match_data_free(matchData);
     }
   }
 
@@ -37,16 +37,17 @@ class PcreExpr : public QueryExpr {
       str = file->baseName();
     }
 
-    rc = pcre_exec(re, extra, str.data(), str.size(), 0, 0, nullptr, 0);
-
-    if (rc == PCRE_ERROR_NOMATCH) {
-      return false;
-    }
-    if (rc >= 0) {
-      return true;
-    }
-    // An error.  It's not actionable here
-    return false;
+    rc = pcre2_match(
+        re,
+        reinterpret_cast<const unsigned char*>(str.data()),
+        str.size(),
+        0,
+        0,
+        matchData,
+        nullptr);
+    // Errors are either PCRE2_ERROR_NOMATCH or non actionable. Thus only match
+    // when we get a positive return value.
+    return rc >= 0;
   }
 
   static std::unique_ptr<QueryExpr>
@@ -54,9 +55,7 @@ class PcreExpr : public QueryExpr {
     const char *ignore, *pattern, *scope = "basename";
     const char *which =
         caseSensitive == CaseSensitivity::CaseInSensitive ? "ipcre" : "pcre";
-    pcre* re;
-    const char* errptr = nullptr;
-    int erroff = 0;
+    size_t erroff = 0;
     int errcode = 0;
 
     if (json_unpack(term, "[s,s,s]", &ignore, &pattern, &scope) != 0 &&
@@ -70,29 +69,42 @@ class PcreExpr : public QueryExpr {
           "Invalid scope '", scope, "' for ", which, " expression"));
     }
 
-    re = pcre_compile2(
-        pattern,
-        caseSensitive == CaseSensitivity::CaseInSensitive ? PCRE_CASELESS : 0,
+    auto re = pcre2_compile(
+        reinterpret_cast<const unsigned char*>(pattern),
+        PCRE2_ZERO_TERMINATED,
+        caseSensitive == CaseSensitivity::CaseInSensitive ? PCRE2_CASELESS : 0,
         &errcode,
-        &errptr,
         &erroff,
         nullptr);
     if (!re) {
+      // From PCRE2 documentation:
+      // https://www.pcre.org/current/doc/html/pcre2api.html#SEC32: "None of the
+      // messages are very long; a buffer size of 120 code units is ample"
+      PCRE2_UCHAR buffer[120];
+      static_assert(
+          sizeof(char) == sizeof(PCRE2_UCHAR),
+          "Watchman uses the 8-bit PCRE2 library");
+      pcre2_get_error_message(errcode, buffer, 120);
       throw QueryParseError(watchman::to<std::string>(
           "invalid ",
           which,
           ": code ",
           errcode,
           " ",
-          errptr,
+          reinterpret_cast<const char*>(&buffer),
           " at offset ",
           erroff,
           " in ",
           pattern));
     }
 
+    auto matchData = pcre2_match_data_create_from_pattern(re, nullptr);
+    if (!matchData) {
+      throw std::bad_alloc();
+    }
+
     return watchman::make_unique<PcreExpr>(
-        re, pcre_study(re, 0, &errptr), !strcmp(scope, "wholename"));
+        re, matchData, !strcmp(scope, "wholename"));
   }
   static std::unique_ptr<QueryExpr> parsePcre(
       w_query* query,
diff --git a/watchman_system.h b/watchman_system.h
index 6235d01..baeb636 100644
--- a/watchman_system.h
+++ b/watchman_system.h
@@ -64,8 +64,9 @@
 #include <poll.h>
 #include <sys/wait.h>
 #endif
-#ifdef HAVE_PCRE_H
-# include <pcre.h>
+#ifdef HAVE_PCRE2_H
+#define PCRE2_CODE_UNIT_WIDTH 8
+# include <pcre2.h>
 #endif
 #ifdef HAVE_EXECINFO_H
 # include <execinfo.h>