diff options
63 files changed, 1439 insertions, 283 deletions
diff --git a/README.chromium b/README.chromium index 66dd3d8..bbb7f8d 100644 --- a/README.chromium +++ b/README.chromium @@ -238,3 +238,16 @@ This directory contains the source code of ICU 52.1 for C/C++ 11. Cherry-pick an upstream patch to fix a bug in bidi. - patches/bidi.patch - upstream bug : http://bugs.icu-project.org/trac/ticket/11054 + +12. Apply the following patches for regex + - patches/regex.patch and patches/regex2.patch + - upstream bugs : http://bugs.icu-project.org/trac/ticket/11369 + http://bugs.icu-project.org/trac/ticket/11370 + http://bugs.icu-project.org/trac/ticket/11371 + +13. Apply the patch for collator + - patches/col.patch + - No upstream bug filed because the upstream code was completely + rewritten during the ICU 53 cycle. + + @@ -247,17 +247,6 @@ }, }, }], - ['OS == "android" and use_system_stlport == 1', { - 'target_conditions': [ - ['_toolset == "target"', { - # ICU requires RTTI, which is not present in the system's - # stlport, so we have to include gabi++. - 'include_dirs': [ - '<(android_src)/abi/cpp/include', - ], - }], - ], - }], ], # conditions }, { @@ -354,17 +343,6 @@ }, }, }], - ['OS == "android" and use_system_stlport == 1', { - 'target_conditions': [ - ['_toolset == "target"', { - # ICU requires RTTI, which is not present in the system's - # stlport, so we have to include gabi++. - 'include_dirs': [ - '<(android_src)/abi/cpp/include', - ], - }], - ], - }], ], # conditions }, ], # targets @@ -375,20 +353,6 @@ 'target_name': 'system_icu', 'type': 'none', 'conditions': [ - ['OS=="android"', { - 'direct_dependent_settings': { - 'include_dirs': [ - '<(android_src)/external/icu/icu4c/source/common', - '<(android_src)/external/icu/icu4c/source/i18n', - ], - }, - 'link_settings': { - 'libraries': [ - '-licui18n', - '-licuuc', - ], - }, - }], ['OS=="qnx"', { 'link_settings': { 'libraries': [ @@ -397,7 +361,7 @@ ], }, }], - ['OS!="android" and OS!="qnx"', { + ['OS!="qnx"', { 'link_settings': { 'ldflags': [ '<!@(icu-config --ldflags)', diff --git a/icudata.host.darwin-arm.mk b/icudata.host.darwin-arm.mk index 4bf0311..3004c4e 100644 --- a/icudata.host.darwin-arm.mk +++ b/icudata.host.darwin-arm.mk @@ -30,7 +30,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -111,7 +110,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icudata.host.darwin-arm64.mk b/icudata.host.darwin-arm64.mk index 3c3882e..490eb00 100644 --- a/icudata.host.darwin-arm64.mk +++ b/icudata.host.darwin-arm64.mk @@ -30,7 +30,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -110,7 +109,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icudata.host.darwin-x86.mk b/icudata.host.darwin-x86.mk index 4bf0311..3004c4e 100644 --- a/icudata.host.darwin-x86.mk +++ b/icudata.host.darwin-x86.mk @@ -30,7 +30,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -111,7 +110,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icudata.host.darwin-x86_64.mk b/icudata.host.darwin-x86_64.mk index f23daaf..4faeff6 100644 --- a/icudata.host.darwin-x86_64.mk +++ b/icudata.host.darwin-x86_64.mk @@ -30,7 +30,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -111,7 +110,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icudata.host.linux-arm.mk b/icudata.host.linux-arm.mk index 4a707e9..f9654d6 100644 --- a/icudata.host.linux-arm.mk +++ b/icudata.host.linux-arm.mk @@ -30,7 +30,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -111,7 +110,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icudata.host.linux-arm64.mk b/icudata.host.linux-arm64.mk index 505658d..8d4dfc1 100644 --- a/icudata.host.linux-arm64.mk +++ b/icudata.host.linux-arm64.mk @@ -30,7 +30,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -110,7 +109,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icudata.host.linux-x86.mk b/icudata.host.linux-x86.mk index 4a707e9..f9654d6 100644 --- a/icudata.host.linux-x86.mk +++ b/icudata.host.linux-x86.mk @@ -30,7 +30,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -111,7 +110,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icudata.host.linux-x86_64.mk b/icudata.host.linux-x86_64.mk index bfdcde9..c9f49b9 100644 --- a/icudata.host.linux-x86_64.mk +++ b/icudata.host.linux-x86_64.mk @@ -30,7 +30,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -111,7 +110,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icudata.target.darwin-arm.mk b/icudata.target.darwin-arm.mk index 66d2a18..05ffa0b 100644 --- a/icudata.target.darwin-arm.mk +++ b/icudata.target.darwin-arm.mk @@ -30,7 +30,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -137,7 +136,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icudata.target.darwin-arm64.mk b/icudata.target.darwin-arm64.mk index 1ff229e..0b335ee 100644 --- a/icudata.target.darwin-arm64.mk +++ b/icudata.target.darwin-arm64.mk @@ -29,7 +29,6 @@ LOCAL_SRC_FILES := \ # Flags passed to both C and C++ files. MY_CFLAGS_Debug := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -124,7 +123,6 @@ LOCAL_CPPFLAGS_Debug := \ # Flags passed to both C and C++ files. MY_CFLAGS_Release := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icudata.target.darwin-x86.mk b/icudata.target.darwin-x86.mk index 0b75684..d0b551a 100644 --- a/icudata.target.darwin-x86.mk +++ b/icudata.target.darwin-x86.mk @@ -29,7 +29,6 @@ LOCAL_SRC_FILES := \ # Flags passed to both C and C++ files. MY_CFLAGS_Debug := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -130,7 +129,6 @@ LOCAL_CPPFLAGS_Debug := \ # Flags passed to both C and C++ files. MY_CFLAGS_Release := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icudata.target.darwin-x86_64.mk b/icudata.target.darwin-x86_64.mk index 905bb9d..0e0acca 100644 --- a/icudata.target.darwin-x86_64.mk +++ b/icudata.target.darwin-x86_64.mk @@ -30,7 +30,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -130,7 +129,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icudata.target.linux-arm.mk b/icudata.target.linux-arm.mk index 66d2a18..05ffa0b 100644 --- a/icudata.target.linux-arm.mk +++ b/icudata.target.linux-arm.mk @@ -30,7 +30,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -137,7 +136,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icudata.target.linux-arm64.mk b/icudata.target.linux-arm64.mk index 1ff229e..0b335ee 100644 --- a/icudata.target.linux-arm64.mk +++ b/icudata.target.linux-arm64.mk @@ -29,7 +29,6 @@ LOCAL_SRC_FILES := \ # Flags passed to both C and C++ files. MY_CFLAGS_Debug := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -124,7 +123,6 @@ LOCAL_CPPFLAGS_Debug := \ # Flags passed to both C and C++ files. MY_CFLAGS_Release := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icudata.target.linux-x86.mk b/icudata.target.linux-x86.mk index 0b75684..d0b551a 100644 --- a/icudata.target.linux-x86.mk +++ b/icudata.target.linux-x86.mk @@ -29,7 +29,6 @@ LOCAL_SRC_FILES := \ # Flags passed to both C and C++ files. MY_CFLAGS_Debug := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -130,7 +129,6 @@ LOCAL_CPPFLAGS_Debug := \ # Flags passed to both C and C++ files. MY_CFLAGS_Release := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icudata.target.linux-x86_64.mk b/icudata.target.linux-x86_64.mk index 905bb9d..0e0acca 100644 --- a/icudata.target.linux-x86_64.mk +++ b/icudata.target.linux-x86_64.mk @@ -30,7 +30,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -130,7 +129,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icui18n.host.darwin-arm.mk b/icui18n.host.darwin-arm.mk index af8904c..485e5d8 100644 --- a/icui18n.host.darwin-arm.mk +++ b/icui18n.host.darwin-arm.mk @@ -191,7 +191,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -280,7 +279,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icui18n.host.darwin-arm64.mk b/icui18n.host.darwin-arm64.mk index 6188bc6..23ae105 100644 --- a/icui18n.host.darwin-arm64.mk +++ b/icui18n.host.darwin-arm64.mk @@ -191,7 +191,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -279,7 +278,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icui18n.host.darwin-x86.mk b/icui18n.host.darwin-x86.mk index af8904c..485e5d8 100644 --- a/icui18n.host.darwin-x86.mk +++ b/icui18n.host.darwin-x86.mk @@ -191,7 +191,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -280,7 +279,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icui18n.host.darwin-x86_64.mk b/icui18n.host.darwin-x86_64.mk index de8608c..7fc23f2 100644 --- a/icui18n.host.darwin-x86_64.mk +++ b/icui18n.host.darwin-x86_64.mk @@ -191,7 +191,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -280,7 +279,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icui18n.host.linux-arm.mk b/icui18n.host.linux-arm.mk index af8904c..485e5d8 100644 --- a/icui18n.host.linux-arm.mk +++ b/icui18n.host.linux-arm.mk @@ -191,7 +191,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -280,7 +279,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icui18n.host.linux-arm64.mk b/icui18n.host.linux-arm64.mk index 6188bc6..23ae105 100644 --- a/icui18n.host.linux-arm64.mk +++ b/icui18n.host.linux-arm64.mk @@ -191,7 +191,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -279,7 +278,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icui18n.host.linux-x86.mk b/icui18n.host.linux-x86.mk index af8904c..485e5d8 100644 --- a/icui18n.host.linux-x86.mk +++ b/icui18n.host.linux-x86.mk @@ -191,7 +191,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -280,7 +279,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icui18n.host.linux-x86_64.mk b/icui18n.host.linux-x86_64.mk index de8608c..7fc23f2 100644 --- a/icui18n.host.linux-x86_64.mk +++ b/icui18n.host.linux-x86_64.mk @@ -191,7 +191,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -280,7 +279,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icui18n.target.darwin-arm.mk b/icui18n.target.darwin-arm.mk index 91b1cdf..55f780c 100644 --- a/icui18n.target.darwin-arm.mk +++ b/icui18n.target.darwin-arm.mk @@ -191,7 +191,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -302,7 +301,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icui18n.target.darwin-arm64.mk b/icui18n.target.darwin-arm64.mk index c0b1bf8..108a751 100644 --- a/icui18n.target.darwin-arm64.mk +++ b/icui18n.target.darwin-arm64.mk @@ -190,7 +190,6 @@ LOCAL_SRC_FILES := \ # Flags passed to both C and C++ files. MY_CFLAGS_Debug := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -289,7 +288,6 @@ LOCAL_CPPFLAGS_Debug := \ # Flags passed to both C and C++ files. MY_CFLAGS_Release := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icui18n.target.darwin-x86.mk b/icui18n.target.darwin-x86.mk index 369f066..17a7c8d 100644 --- a/icui18n.target.darwin-x86.mk +++ b/icui18n.target.darwin-x86.mk @@ -190,7 +190,6 @@ LOCAL_SRC_FILES := \ # Flags passed to both C and C++ files. MY_CFLAGS_Debug := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -295,7 +294,6 @@ LOCAL_CPPFLAGS_Debug := \ # Flags passed to both C and C++ files. MY_CFLAGS_Release := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icui18n.target.darwin-x86_64.mk b/icui18n.target.darwin-x86_64.mk index ad57c31..131d6b9 100644 --- a/icui18n.target.darwin-x86_64.mk +++ b/icui18n.target.darwin-x86_64.mk @@ -191,7 +191,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -295,7 +294,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icui18n.target.linux-arm.mk b/icui18n.target.linux-arm.mk index 91b1cdf..55f780c 100644 --- a/icui18n.target.linux-arm.mk +++ b/icui18n.target.linux-arm.mk @@ -191,7 +191,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -302,7 +301,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icui18n.target.linux-arm64.mk b/icui18n.target.linux-arm64.mk index c0b1bf8..108a751 100644 --- a/icui18n.target.linux-arm64.mk +++ b/icui18n.target.linux-arm64.mk @@ -190,7 +190,6 @@ LOCAL_SRC_FILES := \ # Flags passed to both C and C++ files. MY_CFLAGS_Debug := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -289,7 +288,6 @@ LOCAL_CPPFLAGS_Debug := \ # Flags passed to both C and C++ files. MY_CFLAGS_Release := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icui18n.target.linux-x86.mk b/icui18n.target.linux-x86.mk index 369f066..17a7c8d 100644 --- a/icui18n.target.linux-x86.mk +++ b/icui18n.target.linux-x86.mk @@ -190,7 +190,6 @@ LOCAL_SRC_FILES := \ # Flags passed to both C and C++ files. MY_CFLAGS_Debug := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -295,7 +294,6 @@ LOCAL_CPPFLAGS_Debug := \ # Flags passed to both C and C++ files. MY_CFLAGS_Release := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icui18n.target.linux-x86_64.mk b/icui18n.target.linux-x86_64.mk index ad57c31..131d6b9 100644 --- a/icui18n.target.linux-x86_64.mk +++ b/icui18n.target.linux-x86_64.mk @@ -191,7 +191,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -295,7 +294,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icuuc.host.darwin-arm.mk b/icuuc.host.darwin-arm.mk index 639abaa..3a8b5e1 100644 --- a/icuuc.host.darwin-arm.mk +++ b/icuuc.host.darwin-arm.mk @@ -200,7 +200,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -289,7 +288,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icuuc.host.darwin-arm64.mk b/icuuc.host.darwin-arm64.mk index 925abfb..f0632f0 100644 --- a/icuuc.host.darwin-arm64.mk +++ b/icuuc.host.darwin-arm64.mk @@ -200,7 +200,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -288,7 +287,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icuuc.host.darwin-x86.mk b/icuuc.host.darwin-x86.mk index 639abaa..3a8b5e1 100644 --- a/icuuc.host.darwin-x86.mk +++ b/icuuc.host.darwin-x86.mk @@ -200,7 +200,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -289,7 +288,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icuuc.host.darwin-x86_64.mk b/icuuc.host.darwin-x86_64.mk index a5c83ad..d790e4f 100644 --- a/icuuc.host.darwin-x86_64.mk +++ b/icuuc.host.darwin-x86_64.mk @@ -200,7 +200,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -289,7 +288,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icuuc.host.linux-arm.mk b/icuuc.host.linux-arm.mk index 639abaa..3a8b5e1 100644 --- a/icuuc.host.linux-arm.mk +++ b/icuuc.host.linux-arm.mk @@ -200,7 +200,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -289,7 +288,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icuuc.host.linux-arm64.mk b/icuuc.host.linux-arm64.mk index 925abfb..f0632f0 100644 --- a/icuuc.host.linux-arm64.mk +++ b/icuuc.host.linux-arm64.mk @@ -200,7 +200,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -288,7 +287,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icuuc.host.linux-x86.mk b/icuuc.host.linux-x86.mk index 639abaa..3a8b5e1 100644 --- a/icuuc.host.linux-x86.mk +++ b/icuuc.host.linux-x86.mk @@ -200,7 +200,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -289,7 +288,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icuuc.host.linux-x86_64.mk b/icuuc.host.linux-x86_64.mk index a5c83ad..d790e4f 100644 --- a/icuuc.host.linux-x86_64.mk +++ b/icuuc.host.linux-x86_64.mk @@ -200,7 +200,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ @@ -289,7 +288,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -pthread \ -fno-strict-aliasing \ -Wno-unused-parameter \ diff --git a/icuuc.target.darwin-arm.mk b/icuuc.target.darwin-arm.mk index 00e7234..36496f6 100644 --- a/icuuc.target.darwin-arm.mk +++ b/icuuc.target.darwin-arm.mk @@ -200,7 +200,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -311,7 +310,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icuuc.target.darwin-arm64.mk b/icuuc.target.darwin-arm64.mk index a7dd589..22a7848 100644 --- a/icuuc.target.darwin-arm64.mk +++ b/icuuc.target.darwin-arm64.mk @@ -199,7 +199,6 @@ LOCAL_SRC_FILES := \ # Flags passed to both C and C++ files. MY_CFLAGS_Debug := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -298,7 +297,6 @@ LOCAL_CPPFLAGS_Debug := \ # Flags passed to both C and C++ files. MY_CFLAGS_Release := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icuuc.target.darwin-x86.mk b/icuuc.target.darwin-x86.mk index 71c5441..1374ac3 100644 --- a/icuuc.target.darwin-x86.mk +++ b/icuuc.target.darwin-x86.mk @@ -199,7 +199,6 @@ LOCAL_SRC_FILES := \ # Flags passed to both C and C++ files. MY_CFLAGS_Debug := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -304,7 +303,6 @@ LOCAL_CPPFLAGS_Debug := \ # Flags passed to both C and C++ files. MY_CFLAGS_Release := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icuuc.target.darwin-x86_64.mk b/icuuc.target.darwin-x86_64.mk index 79a2456..680e3e8 100644 --- a/icuuc.target.darwin-x86_64.mk +++ b/icuuc.target.darwin-x86_64.mk @@ -200,7 +200,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -304,7 +303,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icuuc.target.linux-arm.mk b/icuuc.target.linux-arm.mk index 00e7234..36496f6 100644 --- a/icuuc.target.linux-arm.mk +++ b/icuuc.target.linux-arm.mk @@ -200,7 +200,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -311,7 +310,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icuuc.target.linux-arm64.mk b/icuuc.target.linux-arm64.mk index a7dd589..22a7848 100644 --- a/icuuc.target.linux-arm64.mk +++ b/icuuc.target.linux-arm64.mk @@ -199,7 +199,6 @@ LOCAL_SRC_FILES := \ # Flags passed to both C and C++ files. MY_CFLAGS_Debug := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -298,7 +297,6 @@ LOCAL_CPPFLAGS_Debug := \ # Flags passed to both C and C++ files. MY_CFLAGS_Release := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icuuc.target.linux-x86.mk b/icuuc.target.linux-x86.mk index 71c5441..1374ac3 100644 --- a/icuuc.target.linux-x86.mk +++ b/icuuc.target.linux-x86.mk @@ -199,7 +199,6 @@ LOCAL_SRC_FILES := \ # Flags passed to both C and C++ files. MY_CFLAGS_Debug := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -304,7 +303,6 @@ LOCAL_CPPFLAGS_Debug := \ # Flags passed to both C and C++ files. MY_CFLAGS_Release := \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/icuuc.target.linux-x86_64.mk b/icuuc.target.linux-x86_64.mk index 79a2456..680e3e8 100644 --- a/icuuc.target.linux-x86_64.mk +++ b/icuuc.target.linux-x86_64.mk @@ -200,7 +200,6 @@ LOCAL_SRC_FILES := \ MY_CFLAGS_Debug := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ @@ -304,7 +303,6 @@ LOCAL_CPPFLAGS_Debug := \ MY_CFLAGS_Release := \ -fstack-protector \ --param=ssp-buffer-size=4 \ - \ -fno-strict-aliasing \ -Wno-unused-parameter \ -Wno-missing-field-initializers \ diff --git a/patches/col.patch b/patches/col.patch new file mode 100644 index 0000000..aa117fe --- /dev/null +++ b/patches/col.patch @@ -0,0 +1,29 @@ +Index: source/i18n/ucol.cpp +=================================================================== +--- source/i18n/ucol.cpp (revision 292709) ++++ source/i18n/ucol.cpp (working copy) +@@ -2259,6 +2259,9 @@ + if (data->pos + 1 == data->endp) { + return *(data->pos ++); + } ++ if (data->pos >= data->endp) { ++ return (UChar) -1; // return U+FFFF (non-char) to indicate an error ++ } + } + else { + if (innormbuf) { +@@ -2821,7 +2824,13 @@ + } + } + } else if (U16_IS_LEAD(schar)) { +- miss = U16_GET_SUPPLEMENTARY(schar, getNextNormalizedChar(source)); ++ UChar nextChar = getNextNormalizedChar(source); ++ const UChar* prevPos = source->pos; ++ if (U16_IS_TRAIL(nextChar)) { ++ miss = U16_GET_SUPPLEMENTARY(schar, nextChar); ++ } else if (prevPos < source->pos) { ++ goBackOne(source); ++ } + } + + uint8_t sCC; diff --git a/patches/regex.patch b/patches/regex.patch new file mode 100644 index 0000000..4f74ee1 --- /dev/null +++ b/patches/regex.patch @@ -0,0 +1,86 @@ +Index: source/i18n/regexcmp.h +=================================================================== +--- source/i18n/regexcmp.h (revision 292476) ++++ source/i18n/regexcmp.h (working copy) +@@ -182,7 +182,9 @@ + int32_t fMatchOpenParen; // The position in the compiled pattern + // of the slot reserved for a state save + // at the start of the most recently processed +- // parenthesized block. ++ // parenthesized block. Updated when processing ++ // a close to the location for the corresponding open. ++ + int32_t fMatchCloseParen; // The position in the pattern of the first + // location after the most recently processed + // parenthesized block. +Index: source/i18n/regexcmp.cpp +=================================================================== +--- source/i18n/regexcmp.cpp (revision 292476) ++++ source/i18n/regexcmp.cpp (working copy) +@@ -2133,6 +2133,10 @@ + int32_t patEnd = fRXPat->fCompiledPat->size() - 1; + int32_t minML = minMatchLength(fMatchOpenParen, patEnd); + int32_t maxML = maxMatchLength(fMatchOpenParen, patEnd); ++ if (URX_TYPE(maxML) != 0) { ++ error(U_REGEX_LOOK_BEHIND_LIMIT); ++ break; ++ } + if (maxML == INT32_MAX) { + error(U_REGEX_LOOK_BEHIND_LIMIT); + break; +@@ -2166,6 +2170,10 @@ + int32_t patEnd = fRXPat->fCompiledPat->size() - 1; + int32_t minML = minMatchLength(fMatchOpenParen, patEnd); + int32_t maxML = maxMatchLength(fMatchOpenParen, patEnd); ++ if (URX_TYPE(maxML) != 0) { ++ error(U_REGEX_LOOK_BEHIND_LIMIT); ++ break; ++ } + if (maxML == INT32_MAX) { + error(U_REGEX_LOOK_BEHIND_LIMIT); + break; +@@ -2329,7 +2337,15 @@ + int32_t topOfBlock = blockTopLoc(FALSE); + if (fIntervalUpper == 0) { + // Pathological case. Attempt no matches, as if the block doesn't exist. ++ // Discard the generated code for the block. ++ // If the block included parens, discard the info pertaining to them as well. + fRXPat->fCompiledPat->setSize(topOfBlock); ++ if (fMatchOpenParen >= topOfBlock) { ++ fMatchOpenParen = -1; ++ } ++ if (fMatchCloseParen >= topOfBlock) { ++ fMatchCloseParen = -1; ++ } + return TRUE; + } + +Index: source/test/testdata/regextst.txt +=================================================================== +--- source/test/testdata/regextst.txt (revision 292476) ++++ source/test/testdata/regextst.txt (working copy) +@@ -1173,6 +1173,24 @@ + "(?<=(?:){11})bc" "<0>bc</0>" # Empty (?:) expression. + + ++# Bug 11369 ++# Incorrect optimization of patterns with a zero length quantifier {0} ++ ++"(.|b)(|b){0}\$(?#xxx){3}(?>\D*)" "AAAAABBBBBCCCCCDDDDEEEEE" ++"(|b)ab(c)" "<0><1></1>ab<2>c</2></0>" ++"(|b){0}a{3}(D*)" "<0>aaa<2></2></0>" ++"(|b){0,1}a{3}(D*)" "<0><1></1>aaa<2></2></0>" ++"((|b){0})a{3}(D*)" "<0><1></1>aaa<3></3></0>" ++ ++# Bug 11370 ++# Max match length computation of look-behind expression gives result that is too big to fit in the ++# in the 24 bit operand portion of the compiled code. Expressions should fail to compile ++# (Look-behind match length must be bounded. This case is treated as unbounded, an error.) ++ ++"(?<!(0123456789a){10000000})x" E "no match" ++"(?<!\\ubeaf(\\ubeaf{11000}){11000})" E "no match" ++ ++ + # Random debugging, Temporary + # + #"^(?:a?b?)*$" "a--" diff --git a/patches/regex2.patch b/patches/regex2.patch new file mode 100644 index 0000000..808553e --- /dev/null +++ b/patches/regex2.patch @@ -0,0 +1,991 @@ +Index: source/common/unicode/utypes.h +=================================================================== +--- source/common/unicode/utypes.h (revision 292709) ++++ source/common/unicode/utypes.h (working copy) +@@ -647,6 +647,7 @@ + U_REGEX_STACK_OVERFLOW, /**< Regular expression backtrack stack overflow. */ + U_REGEX_TIME_OUT, /**< Maximum allowed match time exceeded */ + U_REGEX_STOPPED_BY_CALLER, /**< Matching operation aborted by user callback fn. */ ++ U_REGEX_PATTERN_TOO_BIG, /**< Pattern exceeds limits on size or complexity. */ + U_REGEX_ERROR_LIMIT, /**< This must always be the last value to indicate the limit for regexp errors */ + + /* +Index: source/common/utypes.c +=================================================================== +--- source/common/utypes.c (revision 292709) ++++ source/common/utypes.c (working copy) +@@ -165,7 +165,8 @@ + "U_REGEX_INVALID_RANGE", + "U_REGEX_STACK_OVERFLOW", + "U_REGEX_TIME_OUT", +- "U_REGEX_STOPPED_BY_CALLER" ++ "U_REGEX_STOPPED_BY_CALLER", ++ "U_REGEX_PATTERN_TOO_BIG" + }; + + static const char * const +Index: source/i18n/regexcmp.cpp +=================================================================== +--- source/i18n/regexcmp.cpp (revision 292943) ++++ source/i18n/regexcmp.cpp (working copy) +@@ -302,7 +302,7 @@ + // present in the saved state: the input string position (int64_t) and + // the position in the compiled pattern. + // +- fRXPat->fFrameSize+=RESTACKFRAME_HDRCOUNT; ++ allocateStackData(RESTACKFRAME_HDRCOUNT); + + // + // Optimization pass 1: NOPs, back-references, and case-folding +@@ -368,9 +368,9 @@ + // the start of an ( grouping. + //4 NOP Resreved, will be replaced by a save if there are + // OR | operators at the top level +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_STATE_SAVE, 2), *fStatus); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_JMP, 3), *fStatus); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_FAIL, 0), *fStatus); ++ appendOp(URX_BUILD(URX_STATE_SAVE, 2)); ++ appendOp(URX_BUILD(URX_JMP, 3)); ++ appendOp(URX_BUILD(URX_FAIL, 0)); + + // Standard open nonCapture paren action emits the two NOPs and + // sets up the paren stack frame. +@@ -393,7 +393,7 @@ + } + + // add the END operation to the compiled pattern. +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_END, 0), *fStatus); ++ appendOp(URX_BUILD(URX_END, 0)); + + // Terminate the pattern compilation state machine. + returnVal = FALSE; +@@ -422,7 +422,7 @@ + // the JMP will eventually be the location following the ')' for the + // group. This will be patched in later, when the ')' is encountered. + op = URX_BUILD(URX_JMP, 0); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + // Push the position of the newly added JMP op onto the parentheses stack. + // This registers if for fixup when this block's close paren is encountered. +@@ -431,7 +431,7 @@ + // Append a NOP to the compiled pattern. This is the slot reserved + // for a SAVE in the event that there is yet another '|' following + // this one. +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); ++ appendOp(URX_BUILD(URX_NOP, 0)); + fParenStack.push(fRXPat->fCompiledPat->size()-1, *fStatus); + } + break; +@@ -457,12 +457,11 @@ + // END_CAPTURE is encountered. + { + fixLiterals(); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); +- int32_t varsLoc = fRXPat->fFrameSize; // Reserve three slots in match stack frame. +- fRXPat->fFrameSize += 3; +- int32_t cop = URX_BUILD(URX_START_CAPTURE, varsLoc); +- fRXPat->fCompiledPat->addElement(cop, *fStatus); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); ++ appendOp(URX_BUILD(URX_NOP, 0)); ++ int32_t varsLoc = allocateStackData(3); // Reserve three slots in match stack frame. ++ int32_t cop = URX_BUILD(URX_START_CAPTURE, varsLoc); ++ appendOp(cop); ++ appendOp(URX_BUILD(URX_NOP, 0)); + + // On the Parentheses stack, start a new frame and add the postions + // of the two NOPs. Depending on what follows in the pattern, the +@@ -487,8 +486,8 @@ + // is an '|' alternation within the parens. + { + fixLiterals(); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); ++ appendOp(URX_BUILD(URX_NOP, 0)); ++ appendOp(URX_BUILD(URX_NOP, 0)); + + // On the Parentheses stack, start a new frame and add the postions + // of the two NOPs. +@@ -510,12 +509,11 @@ + // is an '|' alternation within the parens. + { + fixLiterals(); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); +- int32_t varLoc = fRXPat->fDataSize; // Reserve a data location for saving the +- fRXPat->fDataSize += 1; // state stack ptr. +- int32_t stoOp = URX_BUILD(URX_STO_SP, varLoc); +- fRXPat->fCompiledPat->addElement(stoOp, *fStatus); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); ++ appendOp(URX_BUILD(URX_NOP, 0)); ++ int32_t varLoc = allocateData(1); // Reserve a data location for saving the state stack ptr. ++ int32_t stoOp = URX_BUILD(URX_STO_SP, varLoc); ++ appendOp(stoOp); ++ appendOp(URX_BUILD(URX_NOP, 0)); + + // On the Parentheses stack, start a new frame and add the postions + // of the two NOPs. Depending on what follows in the pattern, the +@@ -558,26 +556,25 @@ + // Two data slots are reserved, for saving the stack ptr and the input position. + { + fixLiterals(); +- int32_t dataLoc = fRXPat->fDataSize; +- fRXPat->fDataSize += 2; ++ int32_t dataLoc = allocateData(2); + int32_t op = URX_BUILD(URX_LA_START, dataLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + op = URX_BUILD(URX_STATE_SAVE, fRXPat->fCompiledPat->size()+ 2); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + op = URX_BUILD(URX_JMP, fRXPat->fCompiledPat->size()+ 3); +- fRXPat->fCompiledPat->addElement(op, *fStatus); +- ++ appendOp(op); ++ + op = URX_BUILD(URX_LA_END, dataLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + op = URX_BUILD(URX_BACKTRACK, 0); +- fRXPat->fCompiledPat->addElement(op, *fStatus); +- ++ appendOp(op); ++ + op = URX_BUILD(URX_NOP, 0); +- fRXPat->fCompiledPat->addElement(op, *fStatus); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); ++ appendOp(op); + + // On the Parentheses stack, start a new frame and add the postions + // of the NOPs. +@@ -602,16 +599,15 @@ + // an alternate (transparent) region. + { + fixLiterals(); +- int32_t dataLoc = fRXPat->fDataSize; +- fRXPat->fDataSize += 2; ++ int32_t dataLoc = allocateData(2); + int32_t op = URX_BUILD(URX_LA_START, dataLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + op = URX_BUILD(URX_STATE_SAVE, 0); // dest address will be patched later. +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + op = URX_BUILD(URX_NOP, 0); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + // On the Parentheses stack, start a new frame and add the postions + // of the StateSave and NOP. +@@ -649,23 +645,22 @@ + fixLiterals(); + + // Allocate data space +- int32_t dataLoc = fRXPat->fDataSize; +- fRXPat->fDataSize += 4; ++ int32_t dataLoc = allocateData(4); + + // Emit URX_LB_START + int32_t op = URX_BUILD(URX_LB_START, dataLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + // Emit URX_LB_CONT + op = URX_BUILD(URX_LB_CONT, dataLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); +- fRXPat->fCompiledPat->addElement(0, *fStatus); // MinMatchLength. To be filled later. +- fRXPat->fCompiledPat->addElement(0, *fStatus); // MaxMatchLength. To be filled later. ++ appendOp(op); ++ appendOp(0); // MinMatchLength. To be filled later. ++ appendOp(0); // MaxMatchLength. To be filled later. + + // Emit the NOP + op = URX_BUILD(URX_NOP, 0); +- fRXPat->fCompiledPat->addElement(op, *fStatus); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); ++ appendOp(op); + + // On the Parentheses stack, start a new frame and add the postions + // of the URX_LB_CONT and the NOP. +@@ -705,24 +700,23 @@ + fixLiterals(); + + // Allocate data space +- int32_t dataLoc = fRXPat->fDataSize; +- fRXPat->fDataSize += 4; ++ int32_t dataLoc = allocateData(4); + + // Emit URX_LB_START + int32_t op = URX_BUILD(URX_LB_START, dataLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + // Emit URX_LBN_CONT + op = URX_BUILD(URX_LBN_CONT, dataLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); +- fRXPat->fCompiledPat->addElement(0, *fStatus); // MinMatchLength. To be filled later. +- fRXPat->fCompiledPat->addElement(0, *fStatus); // MaxMatchLength. To be filled later. +- fRXPat->fCompiledPat->addElement(0, *fStatus); // Continue Loc. To be filled later. ++ appendOp(op); ++ appendOp(0); // MinMatchLength. To be filled later. ++ appendOp(0); // MaxMatchLength. To be filled later. ++ appendOp(0); // Continue Loc. To be filled later. + + // Emit the NOP + op = URX_BUILD(URX_NOP, 0); +- fRXPat->fCompiledPat->addElement(op, *fStatus); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); ++ appendOp(op); + + // On the Parentheses stack, start a new frame and add the postions + // of the URX_LB_CONT and the NOP. +@@ -793,11 +787,10 @@ + if (URX_TYPE(repeatedOp) == URX_SETREF) { + // Emit optimized code for [char set]+ + int32_t loopOpI = URX_BUILD(URX_LOOP_SR_I, URX_VAL(repeatedOp)); +- fRXPat->fCompiledPat->addElement(loopOpI, *fStatus); +- frameLoc = fRXPat->fFrameSize; +- fRXPat->fFrameSize++; ++ appendOp(loopOpI); ++ frameLoc = allocateStackData(1); + int32_t loopOpC = URX_BUILD(URX_LOOP_C, frameLoc); +- fRXPat->fCompiledPat->addElement(loopOpC, *fStatus); ++ appendOp(loopOpC); + break; + } + +@@ -813,11 +806,10 @@ + if (fModeFlags & UREGEX_UNIX_LINES) { + loopOpI |= 2; + } +- fRXPat->fCompiledPat->addElement(loopOpI, *fStatus); +- frameLoc = fRXPat->fFrameSize; +- fRXPat->fFrameSize++; ++ appendOp(loopOpI); ++ frameLoc = allocateStackData(1); + int32_t loopOpC = URX_BUILD(URX_LOOP_C, frameLoc); +- fRXPat->fCompiledPat->addElement(loopOpC, *fStatus); ++ appendOp(loopOpC); + break; + } + +@@ -831,18 +823,17 @@ + // Zero length match is possible. + // Emit the code sequence that can handle it. + insertOp(topLoc); +- frameLoc = fRXPat->fFrameSize; +- fRXPat->fFrameSize++; ++ frameLoc = allocateStackData(1); + + int32_t op = URX_BUILD(URX_STO_INP_LOC, frameLoc); + fRXPat->fCompiledPat->setElementAt(op, topLoc); + + op = URX_BUILD(URX_JMP_SAV_X, topLoc+1); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + } else { + // Simpler code when the repeated body must match something non-empty + int32_t jmpOp = URX_BUILD(URX_JMP_SAV, topLoc); +- fRXPat->fCompiledPat->addElement(jmpOp, *fStatus); ++ appendOp(jmpOp); + } + } + break; +@@ -855,7 +846,7 @@ + { + int32_t topLoc = blockTopLoc(FALSE); + int32_t saveStateOp = URX_BUILD(URX_STATE_SAVE, topLoc); +- fRXPat->fCompiledPat->addElement(saveStateOp, *fStatus); ++ appendOp(saveStateOp); + } + break; + +@@ -892,10 +883,10 @@ + fRXPat->fCompiledPat->setElementAt(jmp1_op, jmp1_loc); + + int32_t jmp2_op = URX_BUILD(URX_JMP, jmp2_loc+2); +- fRXPat->fCompiledPat->addElement(jmp2_op, *fStatus); ++ appendOp(jmp2_op); + + int32_t save_op = URX_BUILD(URX_STATE_SAVE, jmp1_loc+1); +- fRXPat->fCompiledPat->addElement(save_op, *fStatus); ++ appendOp(save_op); + } + break; + +@@ -937,10 +928,9 @@ + // Emit optimized code for a [char set]* + int32_t loopOpI = URX_BUILD(URX_LOOP_SR_I, URX_VAL(repeatedOp)); + fRXPat->fCompiledPat->setElementAt(loopOpI, topLoc); +- dataLoc = fRXPat->fFrameSize; +- fRXPat->fFrameSize++; ++ dataLoc = allocateStackData(1); + int32_t loopOpC = URX_BUILD(URX_LOOP_C, dataLoc); +- fRXPat->fCompiledPat->addElement(loopOpC, *fStatus); ++ appendOp(loopOpC); + break; + } + +@@ -957,10 +947,9 @@ + loopOpI |= 2; + } + fRXPat->fCompiledPat->setElementAt(loopOpI, topLoc); +- dataLoc = fRXPat->fFrameSize; +- fRXPat->fFrameSize++; ++ dataLoc = allocateStackData(1); + int32_t loopOpC = URX_BUILD(URX_LOOP_C, dataLoc); +- fRXPat->fCompiledPat->addElement(loopOpC, *fStatus); ++ appendOp(loopOpC); + break; + } + } +@@ -975,8 +964,7 @@ + // extra loop-breaking code. + if (minMatchLength(saveStateLoc, fRXPat->fCompiledPat->size()-1) == 0) { + insertOp(saveStateLoc); +- dataLoc = fRXPat->fFrameSize; +- fRXPat->fFrameSize++; ++ dataLoc = allocateStackData(1); + + int32_t op = URX_BUILD(URX_STO_INP_LOC, dataLoc); + fRXPat->fCompiledPat->setElementAt(op, saveStateLoc+1); +@@ -992,7 +980,7 @@ + fRXPat->fCompiledPat->setElementAt(saveStateOp, saveStateLoc); + + // Append the URX_JMP_SAV or URX_JMPX operation to the compiled pattern. +- fRXPat->fCompiledPat->addElement(jmpOp, *fStatus); ++ appendOp(jmpOp); + } + break; + +@@ -1009,7 +997,7 @@ + int32_t jmpOp = URX_BUILD(URX_JMP, saveLoc); + int32_t stateSaveOp = URX_BUILD(URX_STATE_SAVE, jmpLoc+1); + fRXPat->fCompiledPat->setElementAt(jmpOp, jmpLoc); +- fRXPat->fCompiledPat->addElement(stateSaveOp, *fStatus); ++ appendOp(stateSaveOp); + } + break; + +@@ -1078,9 +1066,9 @@ + + // First the STO_SP before the start of the loop + insertOp(topLoc); +- int32_t varLoc = fRXPat->fDataSize; // Reserve a data location for saving the +- fRXPat->fDataSize += 1; // state stack ptr. +- int32_t op = URX_BUILD(URX_STO_SP, varLoc); ++ ++ int32_t varLoc = allocateData(1); // Reserve a data location for saving the ++ int32_t op = URX_BUILD(URX_STO_SP, varLoc); + fRXPat->fCompiledPat->setElementAt(op, topLoc); + + int32_t loopOp = (int32_t)fRXPat->fCompiledPat->popi(); +@@ -1090,7 +1078,7 @@ + + // Then the LD_SP after the end of the loop + op = URX_BUILD(URX_LD_SP, varLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + } + + break; +@@ -1134,7 +1122,7 @@ + } else { + op = URX_BUILD(URX_DOTANY, 0); + } +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + } + break; + +@@ -1151,7 +1139,7 @@ + } else if ((fModeFlags & UREGEX_MULTILINE) != 0 && (fModeFlags & UREGEX_UNIX_LINES) != 0) { + op = URX_CARET_M_UNIX; + } +- fRXPat->fCompiledPat->addElement(URX_BUILD(op, 0), *fStatus); ++ appendOp(URX_BUILD(op, 0)); + } + break; + +@@ -1168,13 +1156,13 @@ + } else if ((fModeFlags & UREGEX_MULTILINE) != 0 && (fModeFlags & UREGEX_UNIX_LINES) != 0) { + op = URX_DOLLAR_MD; + } +- fRXPat->fCompiledPat->addElement(URX_BUILD(op, 0), *fStatus); ++ appendOp(URX_BUILD(op, 0)); + } + break; + + case doBackslashA: + fixLiterals(FALSE); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_CARET, 0), *fStatus); ++ appendOp(URX_BUILD(URX_CARET, 0)); + break; + + case doBackslashB: +@@ -1186,7 +1174,7 @@ + #endif + fixLiterals(FALSE); + int32_t op = (fModeFlags & UREGEX_UWORD)? URX_BACKSLASH_BU : URX_BACKSLASH_B; +- fRXPat->fCompiledPat->addElement(URX_BUILD(op, 1), *fStatus); ++ appendOp(URX_BUILD(op, 1)); + } + break; + +@@ -1199,63 +1187,59 @@ + #endif + fixLiterals(FALSE); + int32_t op = (fModeFlags & UREGEX_UWORD)? URX_BACKSLASH_BU : URX_BACKSLASH_B; +- fRXPat->fCompiledPat->addElement(URX_BUILD(op, 0), *fStatus); ++ appendOp(URX_BUILD(op, 0)); + } + break; + + case doBackslashD: + fixLiterals(FALSE); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_D, 1), *fStatus); ++ appendOp(URX_BUILD(URX_BACKSLASH_D, 1)); + break; + + case doBackslashd: + fixLiterals(FALSE); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_D, 0), *fStatus); ++ appendOp(URX_BUILD(URX_BACKSLASH_D, 0)); + break; + + case doBackslashG: + fixLiterals(FALSE); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_G, 0), *fStatus); ++ appendOp(URX_BUILD(URX_BACKSLASH_G, 0)); + break; + + case doBackslashS: + fixLiterals(FALSE); +- fRXPat->fCompiledPat->addElement( +- URX_BUILD(URX_STAT_SETREF_N, URX_ISSPACE_SET), *fStatus); ++ appendOp(URX_BUILD(URX_STAT_SETREF_N, URX_ISSPACE_SET)); + break; + + case doBackslashs: + fixLiterals(FALSE); +- fRXPat->fCompiledPat->addElement( +- URX_BUILD(URX_STATIC_SETREF, URX_ISSPACE_SET), *fStatus); ++ appendOp(URX_BUILD(URX_STATIC_SETREF, URX_ISSPACE_SET)); + break; + + case doBackslashW: + fixLiterals(FALSE); +- fRXPat->fCompiledPat->addElement( +- URX_BUILD(URX_STAT_SETREF_N, URX_ISWORD_SET), *fStatus); ++ appendOp(URX_BUILD(URX_STAT_SETREF_N, URX_ISWORD_SET)); + break; + + case doBackslashw: + fixLiterals(FALSE); +- fRXPat->fCompiledPat->addElement( +- URX_BUILD(URX_STATIC_SETREF, URX_ISWORD_SET), *fStatus); ++ appendOp(URX_BUILD(URX_STATIC_SETREF, URX_ISWORD_SET)); + break; + + case doBackslashX: + fixLiterals(FALSE); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_X, 0), *fStatus); ++ appendOp(URX_BUILD(URX_BACKSLASH_X, 0)); + break; + + + case doBackslashZ: + fixLiterals(FALSE); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_DOLLAR, 0), *fStatus); ++ appendOp(URX_BUILD(URX_DOLLAR, 0)); + break; + + case doBackslashz: + fixLiterals(FALSE); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_Z, 0), *fStatus); ++ appendOp(URX_BUILD(URX_BACKSLASH_Z, 0)); + break; + + case doEscapeError: +@@ -1321,7 +1305,7 @@ + } else { + op = URX_BUILD(URX_BACKREF, groupNum); + } +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + } + break; + +@@ -1342,22 +1326,21 @@ + { + // Emit the STO_SP + int32_t topLoc = blockTopLoc(TRUE); +- int32_t stoLoc = fRXPat->fDataSize; +- fRXPat->fDataSize++; // Reserve the data location for storing save stack ptr. ++ int32_t stoLoc = allocateData(1); // Reserve the data location for storing save stack ptr. + int32_t op = URX_BUILD(URX_STO_SP, stoLoc); + fRXPat->fCompiledPat->setElementAt(op, topLoc); + + // Emit the STATE_SAVE + op = URX_BUILD(URX_STATE_SAVE, fRXPat->fCompiledPat->size()+2); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + // Emit the JMP + op = URX_BUILD(URX_JMP, topLoc+1); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + // Emit the LD_SP + op = URX_BUILD(URX_LD_SP, stoLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + } + break; + +@@ -1377,8 +1360,7 @@ + insertOp(topLoc); + + // emit STO_SP loc +- int32_t stoLoc = fRXPat->fDataSize; +- fRXPat->fDataSize++; // Reserve the data location for storing save stack ptr. ++ int32_t stoLoc = allocateData(1); // Reserve the data location for storing save stack ptr. + int32_t op = URX_BUILD(URX_STO_SP, stoLoc); + fRXPat->fCompiledPat->setElementAt(op, topLoc); + +@@ -1389,11 +1371,11 @@ + + // Append the JMP operation. + op = URX_BUILD(URX_JMP, topLoc+1); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + // Emit the LD_SP loc + op = URX_BUILD(URX_LD_SP, stoLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + } + break; + +@@ -1412,8 +1394,7 @@ + insertOp(topLoc); + + // Emit the STO_SP +- int32_t stoLoc = fRXPat->fDataSize; +- fRXPat->fDataSize++; // Reserve the data location for storing save stack ptr. ++ int32_t stoLoc = allocateData(1); // Reserve the data location for storing save stack ptr. + int32_t op = URX_BUILD(URX_STO_SP, stoLoc); + fRXPat->fCompiledPat->setElementAt(op, topLoc); + +@@ -1424,7 +1405,7 @@ + + // Emit the LD_SP + op = URX_BUILD(URX_LD_SP, stoLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + } + break; + +@@ -1481,8 +1462,8 @@ + // is an '|' alternation within the parens. + { + fixLiterals(FALSE); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); ++ appendOp(URX_BUILD(URX_NOP, 0)); ++ appendOp(URX_BUILD(URX_NOP, 0)); + + // On the Parentheses stack, start a new frame and add the postions + // of the two NOPs (a normal non-capturing () frame, except for the +@@ -1862,7 +1843,7 @@ + } else { + op = URX_BUILD(URX_ONECHAR, lastCodePoint); + } +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + } else { + // Two or more chars, emit a URX_STRING to match them. + if (fModeFlags & UREGEX_CASE_INSENSITIVE) { +@@ -1872,12 +1853,19 @@ + // into two single char ops, for efficiency. + op = URX_BUILD(URX_STRING, fRXPat->fLiteralText.length()); + } +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + op = URX_BUILD(URX_STRING_LEN, fLiteralChars.length()); +- fRXPat->fCompiledPat->addElement(op, *fStatus); +- ++ appendOp(op); ++ + // Add this string into the accumulated strings of the compiled pattern. ++ // The total size of the accumulated strings must be restricted to 24 bits because ++ // string indexes appear as compiled pattern operand values. ++ // This is the only place that the pattern.fLiteralText string is modified. ++ + fRXPat->fLiteralText.append(fLiteralChars); ++ if (U_SUCCESS(*fStatus) && fRXPat->fLiteralText.length() > 0x00ffffff) { ++ *fStatus = U_REGEX_PATTERN_TOO_BIG; ++ } + } + + fLiteralChars.remove(); +@@ -1884,10 +1872,22 @@ + } + + ++//------------------------------------------------------------------------------ ++// ++// appendOp() Append a new instruction onto the compiled pattern ++// Includes error checking, limiting the size of the ++// pattern to lengths that can be represented in the ++// 24 bit operand field of an instruction. ++// ++//------------------------------------------------------------------------------ ++void RegexCompile::appendOp(int32_t op) { ++ fRXPat->fCompiledPat->addElement(op, *fStatus); ++ if ((fRXPat->fCompiledPat->size() > 0x00fffff0) && U_SUCCESS(*fStatus)) { ++ *fStatus = U_REGEX_PATTERN_TOO_BIG; ++ } ++} + + +- +- + //------------------------------------------------------------------------------ + // + // insertOp() Insert a slot for a new opcode into the already +@@ -1947,9 +1947,61 @@ + } + + ++//------------------------------------------------------------------------------ ++// ++// allocateData() Allocate storage in the matcher's static data area. ++// Return the index for the newly allocated data. ++// The storage won't actually exist until we are running a match ++// operation, but the storage indexes are inserted into various ++// opcodes while compiling the pattern. ++// ++//------------------------------------------------------------------------------ ++int32_t RegexCompile::allocateData(int32_t size) { ++ if (U_FAILURE(*fStatus)) { ++ return 0; ++ } ++ if (size <= 0 || size > 0x100 || fRXPat->fDataSize < 0) { ++ *fStatus = U_REGEX_INTERNAL_ERROR; ++ return 0; ++ } ++ int32_t dataIndex = fRXPat->fDataSize; ++ fRXPat->fDataSize += size; ++ if (fRXPat->fDataSize >= 0x00fffff0) { ++ *fStatus = U_REGEX_PATTERN_TOO_BIG; ++ } ++ return dataIndex; ++} + ++ + //------------------------------------------------------------------------------ + // ++// allocateStackData() Allocate space in the back-tracking stack frame. ++// Return the index for the newly allocated data. ++// The frame indexes are inserted into various ++// opcodes while compiling the pattern, meaning that frame ++// size must be restricted to the size that will fit ++// as an operand (24 bits). ++// ++//------------------------------------------------------------------------------ ++int32_t RegexCompile::allocateStackData(int32_t size) { ++ if (U_FAILURE(*fStatus)) { ++ return 0; ++ } ++ if (size <= 0 || size > 0x100 || fRXPat->fFrameSize < 0) { ++ *fStatus = U_REGEX_INTERNAL_ERROR; ++ return 0; ++ } ++ int32_t dataIndex = fRXPat->fFrameSize; ++ fRXPat->fFrameSize += size; ++ if (fRXPat->fFrameSize >= 0x00fffff0) { ++ *fStatus = U_REGEX_PATTERN_TOO_BIG; ++ } ++ return dataIndex; ++} ++ ++ ++//------------------------------------------------------------------------------ ++// + // blockTopLoc() Find or create a location in the compiled pattern + // at the start of the operation or block that has + // just been compiled. Needed when a quantifier (* or +@@ -2065,7 +2117,7 @@ + + int32_t frameVarLocation = URX_VAL(captureOp); + int32_t endCaptureOp = URX_BUILD(URX_END_CAPTURE, frameVarLocation); +- fRXPat->fCompiledPat->addElement(endCaptureOp, *fStatus); ++ appendOp(endCaptureOp); + } + break; + case atomic: +@@ -2077,7 +2129,7 @@ + U_ASSERT(URX_TYPE(stoOp) == URX_STO_SP); + int32_t stoLoc = URX_VAL(stoOp); + int32_t ldOp = URX_BUILD(URX_LD_SP, stoLoc); +- fRXPat->fCompiledPat->addElement(ldOp, *fStatus); ++ appendOp(ldOp); + } + break; + +@@ -2087,7 +2139,7 @@ + U_ASSERT(URX_TYPE(startOp) == URX_LA_START); + int32_t dataLoc = URX_VAL(startOp); + int32_t op = URX_BUILD(URX_LA_END, dataLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + } + break; + +@@ -2098,11 +2150,11 @@ + U_ASSERT(URX_TYPE(startOp) == URX_LA_START); + int32_t dataLoc = URX_VAL(startOp); + int32_t op = URX_BUILD(URX_LA_END, dataLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + op = URX_BUILD(URX_BACKTRACK, 0); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + op = URX_BUILD(URX_LA_END, dataLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + // Patch the URX_SAVE near the top of the block. + // The destination of the SAVE is the final LA_END that was just added. +@@ -2123,9 +2175,9 @@ + U_ASSERT(URX_TYPE(startOp) == URX_LB_START); + int32_t dataLoc = URX_VAL(startOp); + int32_t op = URX_BUILD(URX_LB_END, dataLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); +- op = URX_BUILD(URX_LA_END, dataLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); ++ op = URX_BUILD(URX_LA_END, dataLoc); ++ appendOp(op); + + // Determine the min and max bounds for the length of the + // string that the pattern can match. +@@ -2162,7 +2214,7 @@ + U_ASSERT(URX_TYPE(startOp) == URX_LB_START); + int32_t dataLoc = URX_VAL(startOp); + int32_t op = URX_BUILD(URX_LBN_END, dataLoc); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + // Determine the min and max bounds for the length of the + // string that the pattern can match. +@@ -2228,7 +2280,7 @@ + case 0: + { + // Set of no elements. Always fails to match. +- fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKTRACK, 0), *fStatus); ++ appendOp(URX_BUILD(URX_BACKTRACK, 0)); + delete theSet; + } + break; +@@ -2250,7 +2302,7 @@ + int32_t setNumber = fRXPat->fSets->size(); + fRXPat->fSets->addElement(theSet, *fStatus); + int32_t setOp = URX_BUILD(URX_SETREF, setNumber); +- fRXPat->fCompiledPat->addElement(setOp, *fStatus); ++ appendOp(setOp); + } + } + } +@@ -2289,11 +2341,8 @@ + // counterLoc --> Loop counter + // +1 --> Input index (for breaking non-progressing loops) + // (Only present if unbounded upper limit on loop) +- int32_t counterLoc = fRXPat->fFrameSize; +- fRXPat->fFrameSize++; +- if (fIntervalUpper < 0) { +- fRXPat->fFrameSize++; +- } ++ int32_t dataSize = fIntervalUpper < 0 ? 2 : 1; ++ int32_t counterLoc = allocateStackData(dataSize); + + int32_t op = URX_BUILD(InitOp, counterLoc); + fRXPat->fCompiledPat->setElementAt(op, topOfBlock); +@@ -2313,7 +2362,7 @@ + // Apend the CTR_LOOP op. The operand is the location of the CTR_INIT op. + // Goes at end of the block being looped over, so just append to the code so far. + op = URX_BUILD(LoopOp, topOfBlock); +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + + if ((fIntervalLow & 0xff000000) != 0 || + (fIntervalUpper > 0 && (fIntervalUpper & 0xff000000) != 0)) { +@@ -2380,12 +2429,12 @@ + int32_t i; + for (i=1; i<fIntervalUpper; i++ ) { + if (i == fIntervalLow) { +- fRXPat->fCompiledPat->addElement(saveOp, *fStatus); ++ appendOp(saveOp); + } + if (i > fIntervalLow) { +- fRXPat->fCompiledPat->addElement(saveOp, *fStatus); ++ appendOp(saveOp); + } +- fRXPat->fCompiledPat->addElement(op, *fStatus); ++ appendOp(op); + } + return TRUE; + } +Index: source/i18n/regexcmp.h +=================================================================== +--- source/i18n/regexcmp.h (revision 292943) ++++ source/i18n/regexcmp.h (working copy) +@@ -103,6 +103,11 @@ + void fixLiterals(UBool split=FALSE); // Generate code for pending literal characters. + void insertOp(int32_t where); // Open up a slot for a new op in the + // generated code at the specified location. ++ void appendOp(int32_t op); // Append a new op to the compiled pattern. ++ int32_t allocateData(int32_t size); // Allocate space in the matcher data area. ++ // Return index of the newly allocated data. ++ int32_t allocateStackData(int32_t size); // Allocate space in the match back-track stack frame. ++ // Return offset index in the frame. + int32_t minMatchLength(int32_t start, + int32_t end); + int32_t maxMatchLength(int32_t start, +Index: source/i18n/regeximp.cpp +=================================================================== +--- source/i18n/regeximp.cpp (revision 292709) ++++ source/i18n/regeximp.cpp (working copy) +@@ -12,6 +12,8 @@ + + #if !UCONFIG_NO_REGULAR_EXPRESSIONS + #include "regeximp.h" ++ ++#include "uassert.h" + #include "unicode/utf16.h" + + U_NAMESPACE_BEGIN +@@ -113,7 +115,27 @@ + return fIndex; + } + ++// Assemble a pcode instruction from the opcode and operand values. ++// Out-of-range values should not occur - if they do it is from an internal ++// error in the regex compiler. + ++// TODO: move into regexcmp, where it has access to fStatus. ++// NOP cleanly if U_FAILURE. ++// Set U_REGEX_INTERNAL_ERROR on bad operands. ++ ++int32_t URX_BUILD(int32_t type, int32_t val) { ++ if (type < 0 || type > 255) { ++ U_ASSERT(FALSE); ++ type = URX_RESERVED_OP; ++ } ++ if (val > 0x00ffffff) { ++ U_ASSERT(FALSE); ++ val = 0; ++ } ++ return (type << 24) | val; ++} ++ ++ + U_NAMESPACE_END + + #endif +Index: source/i18n/regeximp.h +=================================================================== +--- source/i18n/regeximp.h (revision 292709) ++++ source/i18n/regeximp.h (working copy) +@@ -254,7 +254,7 @@ + // + // Convenience macros for assembling and disassembling a compiled operation. + // +-#define URX_BUILD(type, val) (int32_t)((type << 24) | (val)) ++int32_t URX_BUILD(int32_t val, int32_t type); + #define URX_TYPE(x) ((uint32_t)(x) >> 24) + #define URX_VAL(x) ((x) & 0xffffff) + +Index: source/test/intltest/regextst.cpp +=================================================================== +--- source/test/intltest/regextst.cpp (revision 292709) ++++ source/test/intltest/regextst.cpp (working copy) +@@ -131,6 +131,9 @@ + case 21: name = "Bug 9283"; + if (exec) Bug9283(); + break; ++ case 22: name = "TestBug11371"; ++ if (exec) TestBug11371(); ++ break; + + default: name = ""; + break; //needed to end loop +@@ -5229,5 +5232,47 @@ + } + } + ++void RegexTest::TestBug11371() { ++ UErrorCode status = U_ZERO_ERROR; ++ UnicodeString patternString; ++ ++ for (int i=0; i<8000000; i++) { ++ patternString.append(UnicodeString("()")); ++ } ++ LocalPointer<RegexPattern> compiledPat(RegexPattern::compile(patternString, 0, status)); ++ if (status != U_REGEX_PATTERN_TOO_BIG) { ++ errln("File %s, line %d expected status=U_REGEX_PATTERN_TOO_BIG; got %s.", ++ __FILE__, __LINE__, u_errorName(status)); ++ } ++ ++ status = U_ZERO_ERROR; ++ patternString = "("; ++ for (int i=0; i<20000000; i++) { ++ patternString.append(UnicodeString("A++")); ++ } ++ patternString.append(UnicodeString("){0}B++")); ++ LocalPointer<RegexPattern> compiledPat2(RegexPattern::compile(patternString, 0, status)); ++ if (status != U_REGEX_PATTERN_TOO_BIG) { ++ errln("File %s, line %d expected status=U_REGEX_PATTERN_TOO_BIG; got %s.", ++ __FILE__, __LINE__, u_errorName(status)); ++ } ++ ++ // Pattern with too much string data, such that string indexes overflow operand data. ++ status = U_ZERO_ERROR; ++ patternString = ""; ++ while (patternString.length() < 0x00ffffff) { ++ patternString.append(UnicodeString("stuff and things dont you know, these are a few of my favorite strings\n")); ++ } ++ patternString.append(UnicodeString("X? trailing string")); ++ LocalPointer<RegexPattern> compiledPat3(RegexPattern::compile(patternString, 0, status)); ++ compiledPat3->dumpPattern(); ++ if (status != U_REGEX_PATTERN_TOO_BIG) { ++ errln("File %s, line %d expected status=U_REGEX_PATTERN_TOO_BIG; got %s.", ++ __FILE__, __LINE__, u_errorName(status)); ++ } ++ ++ ++ ++} ++ + #endif /* !UCONFIG_NO_REGULAR_EXPRESSIONS */ +- +Index: source/test/intltest/regextst.h +=================================================================== +--- source/test/intltest/regextst.h (revision 292709) ++++ source/test/intltest/regextst.h (working copy) +@@ -47,6 +47,7 @@ + virtual void Bug7029(); + virtual void Bug9283(); + virtual void CheckInvBufSize(); ++ virtual void TestBug11371(); + + // The following functions are internal to the regexp tests. + virtual void assertUText(const char *expected, UText *actual, const char *file, int line); diff --git a/source/common/unicode/utypes.h b/source/common/unicode/utypes.h index 0cf78da..a53aafa 100644 --- a/source/common/unicode/utypes.h +++ b/source/common/unicode/utypes.h @@ -647,6 +647,7 @@ typedef enum UErrorCode { U_REGEX_STACK_OVERFLOW, /**< Regular expression backtrack stack overflow. */ U_REGEX_TIME_OUT, /**< Maximum allowed match time exceeded */ U_REGEX_STOPPED_BY_CALLER, /**< Matching operation aborted by user callback fn. */ + U_REGEX_PATTERN_TOO_BIG, /**< Pattern exceeds limits on size or complexity. */ U_REGEX_ERROR_LIMIT, /**< This must always be the last value to indicate the limit for regexp errors */ /* diff --git a/source/common/utypes.c b/source/common/utypes.c index c28e727..a5cadaf 100644 --- a/source/common/utypes.c +++ b/source/common/utypes.c @@ -165,7 +165,8 @@ _uRegexErrorName[U_REGEX_ERROR_LIMIT - U_REGEX_ERROR_START] = { "U_REGEX_INVALID_RANGE", "U_REGEX_STACK_OVERFLOW", "U_REGEX_TIME_OUT", - "U_REGEX_STOPPED_BY_CALLER" + "U_REGEX_STOPPED_BY_CALLER", + "U_REGEX_PATTERN_TOO_BIG" }; static const char * const diff --git a/source/i18n/regexcmp.cpp b/source/i18n/regexcmp.cpp index 0ec6154..afad2bd 100644 --- a/source/i18n/regexcmp.cpp +++ b/source/i18n/regexcmp.cpp @@ -302,7 +302,7 @@ void RegexCompile::compile( // present in the saved state: the input string position (int64_t) and // the position in the compiled pattern. // - fRXPat->fFrameSize+=RESTACKFRAME_HDRCOUNT; + allocateStackData(RESTACKFRAME_HDRCOUNT); // // Optimization pass 1: NOPs, back-references, and case-folding @@ -368,9 +368,9 @@ UBool RegexCompile::doParseActions(int32_t action) // the start of an ( grouping. //4 NOP Resreved, will be replaced by a save if there are // OR | operators at the top level - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_STATE_SAVE, 2), *fStatus); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_JMP, 3), *fStatus); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_FAIL, 0), *fStatus); + appendOp(URX_BUILD(URX_STATE_SAVE, 2)); + appendOp(URX_BUILD(URX_JMP, 3)); + appendOp(URX_BUILD(URX_FAIL, 0)); // Standard open nonCapture paren action emits the two NOPs and // sets up the paren stack frame. @@ -393,7 +393,7 @@ UBool RegexCompile::doParseActions(int32_t action) } // add the END operation to the compiled pattern. - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_END, 0), *fStatus); + appendOp(URX_BUILD(URX_END, 0)); // Terminate the pattern compilation state machine. returnVal = FALSE; @@ -422,7 +422,7 @@ UBool RegexCompile::doParseActions(int32_t action) // the JMP will eventually be the location following the ')' for the // group. This will be patched in later, when the ')' is encountered. op = URX_BUILD(URX_JMP, 0); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); // Push the position of the newly added JMP op onto the parentheses stack. // This registers if for fixup when this block's close paren is encountered. @@ -431,7 +431,7 @@ UBool RegexCompile::doParseActions(int32_t action) // Append a NOP to the compiled pattern. This is the slot reserved // for a SAVE in the event that there is yet another '|' following // this one. - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); + appendOp(URX_BUILD(URX_NOP, 0)); fParenStack.push(fRXPat->fCompiledPat->size()-1, *fStatus); } break; @@ -457,12 +457,11 @@ UBool RegexCompile::doParseActions(int32_t action) // END_CAPTURE is encountered. { fixLiterals(); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); - int32_t varsLoc = fRXPat->fFrameSize; // Reserve three slots in match stack frame. - fRXPat->fFrameSize += 3; - int32_t cop = URX_BUILD(URX_START_CAPTURE, varsLoc); - fRXPat->fCompiledPat->addElement(cop, *fStatus); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); + appendOp(URX_BUILD(URX_NOP, 0)); + int32_t varsLoc = allocateStackData(3); // Reserve three slots in match stack frame. + int32_t cop = URX_BUILD(URX_START_CAPTURE, varsLoc); + appendOp(cop); + appendOp(URX_BUILD(URX_NOP, 0)); // On the Parentheses stack, start a new frame and add the postions // of the two NOPs. Depending on what follows in the pattern, the @@ -487,8 +486,8 @@ UBool RegexCompile::doParseActions(int32_t action) // is an '|' alternation within the parens. { fixLiterals(); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); + appendOp(URX_BUILD(URX_NOP, 0)); + appendOp(URX_BUILD(URX_NOP, 0)); // On the Parentheses stack, start a new frame and add the postions // of the two NOPs. @@ -510,12 +509,11 @@ UBool RegexCompile::doParseActions(int32_t action) // is an '|' alternation within the parens. { fixLiterals(); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); - int32_t varLoc = fRXPat->fDataSize; // Reserve a data location for saving the - fRXPat->fDataSize += 1; // state stack ptr. - int32_t stoOp = URX_BUILD(URX_STO_SP, varLoc); - fRXPat->fCompiledPat->addElement(stoOp, *fStatus); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); + appendOp(URX_BUILD(URX_NOP, 0)); + int32_t varLoc = allocateData(1); // Reserve a data location for saving the state stack ptr. + int32_t stoOp = URX_BUILD(URX_STO_SP, varLoc); + appendOp(stoOp); + appendOp(URX_BUILD(URX_NOP, 0)); // On the Parentheses stack, start a new frame and add the postions // of the two NOPs. Depending on what follows in the pattern, the @@ -558,26 +556,25 @@ UBool RegexCompile::doParseActions(int32_t action) // Two data slots are reserved, for saving the stack ptr and the input position. { fixLiterals(); - int32_t dataLoc = fRXPat->fDataSize; - fRXPat->fDataSize += 2; + int32_t dataLoc = allocateData(2); int32_t op = URX_BUILD(URX_LA_START, dataLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); op = URX_BUILD(URX_STATE_SAVE, fRXPat->fCompiledPat->size()+ 2); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); op = URX_BUILD(URX_JMP, fRXPat->fCompiledPat->size()+ 3); - fRXPat->fCompiledPat->addElement(op, *fStatus); - + appendOp(op); + op = URX_BUILD(URX_LA_END, dataLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); op = URX_BUILD(URX_BACKTRACK, 0); - fRXPat->fCompiledPat->addElement(op, *fStatus); - + appendOp(op); + op = URX_BUILD(URX_NOP, 0); - fRXPat->fCompiledPat->addElement(op, *fStatus); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); + appendOp(op); // On the Parentheses stack, start a new frame and add the postions // of the NOPs. @@ -602,16 +599,15 @@ UBool RegexCompile::doParseActions(int32_t action) // an alternate (transparent) region. { fixLiterals(); - int32_t dataLoc = fRXPat->fDataSize; - fRXPat->fDataSize += 2; + int32_t dataLoc = allocateData(2); int32_t op = URX_BUILD(URX_LA_START, dataLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); op = URX_BUILD(URX_STATE_SAVE, 0); // dest address will be patched later. - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); op = URX_BUILD(URX_NOP, 0); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); // On the Parentheses stack, start a new frame and add the postions // of the StateSave and NOP. @@ -649,23 +645,22 @@ UBool RegexCompile::doParseActions(int32_t action) fixLiterals(); // Allocate data space - int32_t dataLoc = fRXPat->fDataSize; - fRXPat->fDataSize += 4; + int32_t dataLoc = allocateData(4); // Emit URX_LB_START int32_t op = URX_BUILD(URX_LB_START, dataLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); // Emit URX_LB_CONT op = URX_BUILD(URX_LB_CONT, dataLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); - fRXPat->fCompiledPat->addElement(0, *fStatus); // MinMatchLength. To be filled later. - fRXPat->fCompiledPat->addElement(0, *fStatus); // MaxMatchLength. To be filled later. + appendOp(op); + appendOp(0); // MinMatchLength. To be filled later. + appendOp(0); // MaxMatchLength. To be filled later. // Emit the NOP op = URX_BUILD(URX_NOP, 0); - fRXPat->fCompiledPat->addElement(op, *fStatus); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); + appendOp(op); // On the Parentheses stack, start a new frame and add the postions // of the URX_LB_CONT and the NOP. @@ -705,24 +700,23 @@ UBool RegexCompile::doParseActions(int32_t action) fixLiterals(); // Allocate data space - int32_t dataLoc = fRXPat->fDataSize; - fRXPat->fDataSize += 4; + int32_t dataLoc = allocateData(4); // Emit URX_LB_START int32_t op = URX_BUILD(URX_LB_START, dataLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); // Emit URX_LBN_CONT op = URX_BUILD(URX_LBN_CONT, dataLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); - fRXPat->fCompiledPat->addElement(0, *fStatus); // MinMatchLength. To be filled later. - fRXPat->fCompiledPat->addElement(0, *fStatus); // MaxMatchLength. To be filled later. - fRXPat->fCompiledPat->addElement(0, *fStatus); // Continue Loc. To be filled later. + appendOp(op); + appendOp(0); // MinMatchLength. To be filled later. + appendOp(0); // MaxMatchLength. To be filled later. + appendOp(0); // Continue Loc. To be filled later. // Emit the NOP op = URX_BUILD(URX_NOP, 0); - fRXPat->fCompiledPat->addElement(op, *fStatus); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); + appendOp(op); // On the Parentheses stack, start a new frame and add the postions // of the URX_LB_CONT and the NOP. @@ -793,11 +787,10 @@ UBool RegexCompile::doParseActions(int32_t action) if (URX_TYPE(repeatedOp) == URX_SETREF) { // Emit optimized code for [char set]+ int32_t loopOpI = URX_BUILD(URX_LOOP_SR_I, URX_VAL(repeatedOp)); - fRXPat->fCompiledPat->addElement(loopOpI, *fStatus); - frameLoc = fRXPat->fFrameSize; - fRXPat->fFrameSize++; + appendOp(loopOpI); + frameLoc = allocateStackData(1); int32_t loopOpC = URX_BUILD(URX_LOOP_C, frameLoc); - fRXPat->fCompiledPat->addElement(loopOpC, *fStatus); + appendOp(loopOpC); break; } @@ -813,11 +806,10 @@ UBool RegexCompile::doParseActions(int32_t action) if (fModeFlags & UREGEX_UNIX_LINES) { loopOpI |= 2; } - fRXPat->fCompiledPat->addElement(loopOpI, *fStatus); - frameLoc = fRXPat->fFrameSize; - fRXPat->fFrameSize++; + appendOp(loopOpI); + frameLoc = allocateStackData(1); int32_t loopOpC = URX_BUILD(URX_LOOP_C, frameLoc); - fRXPat->fCompiledPat->addElement(loopOpC, *fStatus); + appendOp(loopOpC); break; } @@ -831,18 +823,17 @@ UBool RegexCompile::doParseActions(int32_t action) // Zero length match is possible. // Emit the code sequence that can handle it. insertOp(topLoc); - frameLoc = fRXPat->fFrameSize; - fRXPat->fFrameSize++; + frameLoc = allocateStackData(1); int32_t op = URX_BUILD(URX_STO_INP_LOC, frameLoc); fRXPat->fCompiledPat->setElementAt(op, topLoc); op = URX_BUILD(URX_JMP_SAV_X, topLoc+1); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); } else { // Simpler code when the repeated body must match something non-empty int32_t jmpOp = URX_BUILD(URX_JMP_SAV, topLoc); - fRXPat->fCompiledPat->addElement(jmpOp, *fStatus); + appendOp(jmpOp); } } break; @@ -855,7 +846,7 @@ UBool RegexCompile::doParseActions(int32_t action) { int32_t topLoc = blockTopLoc(FALSE); int32_t saveStateOp = URX_BUILD(URX_STATE_SAVE, topLoc); - fRXPat->fCompiledPat->addElement(saveStateOp, *fStatus); + appendOp(saveStateOp); } break; @@ -892,10 +883,10 @@ UBool RegexCompile::doParseActions(int32_t action) fRXPat->fCompiledPat->setElementAt(jmp1_op, jmp1_loc); int32_t jmp2_op = URX_BUILD(URX_JMP, jmp2_loc+2); - fRXPat->fCompiledPat->addElement(jmp2_op, *fStatus); + appendOp(jmp2_op); int32_t save_op = URX_BUILD(URX_STATE_SAVE, jmp1_loc+1); - fRXPat->fCompiledPat->addElement(save_op, *fStatus); + appendOp(save_op); } break; @@ -937,10 +928,9 @@ UBool RegexCompile::doParseActions(int32_t action) // Emit optimized code for a [char set]* int32_t loopOpI = URX_BUILD(URX_LOOP_SR_I, URX_VAL(repeatedOp)); fRXPat->fCompiledPat->setElementAt(loopOpI, topLoc); - dataLoc = fRXPat->fFrameSize; - fRXPat->fFrameSize++; + dataLoc = allocateStackData(1); int32_t loopOpC = URX_BUILD(URX_LOOP_C, dataLoc); - fRXPat->fCompiledPat->addElement(loopOpC, *fStatus); + appendOp(loopOpC); break; } @@ -957,10 +947,9 @@ UBool RegexCompile::doParseActions(int32_t action) loopOpI |= 2; } fRXPat->fCompiledPat->setElementAt(loopOpI, topLoc); - dataLoc = fRXPat->fFrameSize; - fRXPat->fFrameSize++; + dataLoc = allocateStackData(1); int32_t loopOpC = URX_BUILD(URX_LOOP_C, dataLoc); - fRXPat->fCompiledPat->addElement(loopOpC, *fStatus); + appendOp(loopOpC); break; } } @@ -975,8 +964,7 @@ UBool RegexCompile::doParseActions(int32_t action) // extra loop-breaking code. if (minMatchLength(saveStateLoc, fRXPat->fCompiledPat->size()-1) == 0) { insertOp(saveStateLoc); - dataLoc = fRXPat->fFrameSize; - fRXPat->fFrameSize++; + dataLoc = allocateStackData(1); int32_t op = URX_BUILD(URX_STO_INP_LOC, dataLoc); fRXPat->fCompiledPat->setElementAt(op, saveStateLoc+1); @@ -992,7 +980,7 @@ UBool RegexCompile::doParseActions(int32_t action) fRXPat->fCompiledPat->setElementAt(saveStateOp, saveStateLoc); // Append the URX_JMP_SAV or URX_JMPX operation to the compiled pattern. - fRXPat->fCompiledPat->addElement(jmpOp, *fStatus); + appendOp(jmpOp); } break; @@ -1009,7 +997,7 @@ UBool RegexCompile::doParseActions(int32_t action) int32_t jmpOp = URX_BUILD(URX_JMP, saveLoc); int32_t stateSaveOp = URX_BUILD(URX_STATE_SAVE, jmpLoc+1); fRXPat->fCompiledPat->setElementAt(jmpOp, jmpLoc); - fRXPat->fCompiledPat->addElement(stateSaveOp, *fStatus); + appendOp(stateSaveOp); } break; @@ -1078,9 +1066,9 @@ UBool RegexCompile::doParseActions(int32_t action) // First the STO_SP before the start of the loop insertOp(topLoc); - int32_t varLoc = fRXPat->fDataSize; // Reserve a data location for saving the - fRXPat->fDataSize += 1; // state stack ptr. - int32_t op = URX_BUILD(URX_STO_SP, varLoc); + + int32_t varLoc = allocateData(1); // Reserve a data location for saving the + int32_t op = URX_BUILD(URX_STO_SP, varLoc); fRXPat->fCompiledPat->setElementAt(op, topLoc); int32_t loopOp = (int32_t)fRXPat->fCompiledPat->popi(); @@ -1090,7 +1078,7 @@ UBool RegexCompile::doParseActions(int32_t action) // Then the LD_SP after the end of the loop op = URX_BUILD(URX_LD_SP, varLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); } break; @@ -1134,7 +1122,7 @@ UBool RegexCompile::doParseActions(int32_t action) } else { op = URX_BUILD(URX_DOTANY, 0); } - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); } break; @@ -1151,7 +1139,7 @@ UBool RegexCompile::doParseActions(int32_t action) } else if ((fModeFlags & UREGEX_MULTILINE) != 0 && (fModeFlags & UREGEX_UNIX_LINES) != 0) { op = URX_CARET_M_UNIX; } - fRXPat->fCompiledPat->addElement(URX_BUILD(op, 0), *fStatus); + appendOp(URX_BUILD(op, 0)); } break; @@ -1168,13 +1156,13 @@ UBool RegexCompile::doParseActions(int32_t action) } else if ((fModeFlags & UREGEX_MULTILINE) != 0 && (fModeFlags & UREGEX_UNIX_LINES) != 0) { op = URX_DOLLAR_MD; } - fRXPat->fCompiledPat->addElement(URX_BUILD(op, 0), *fStatus); + appendOp(URX_BUILD(op, 0)); } break; case doBackslashA: fixLiterals(FALSE); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_CARET, 0), *fStatus); + appendOp(URX_BUILD(URX_CARET, 0)); break; case doBackslashB: @@ -1186,7 +1174,7 @@ UBool RegexCompile::doParseActions(int32_t action) #endif fixLiterals(FALSE); int32_t op = (fModeFlags & UREGEX_UWORD)? URX_BACKSLASH_BU : URX_BACKSLASH_B; - fRXPat->fCompiledPat->addElement(URX_BUILD(op, 1), *fStatus); + appendOp(URX_BUILD(op, 1)); } break; @@ -1199,63 +1187,59 @@ UBool RegexCompile::doParseActions(int32_t action) #endif fixLiterals(FALSE); int32_t op = (fModeFlags & UREGEX_UWORD)? URX_BACKSLASH_BU : URX_BACKSLASH_B; - fRXPat->fCompiledPat->addElement(URX_BUILD(op, 0), *fStatus); + appendOp(URX_BUILD(op, 0)); } break; case doBackslashD: fixLiterals(FALSE); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_D, 1), *fStatus); + appendOp(URX_BUILD(URX_BACKSLASH_D, 1)); break; case doBackslashd: fixLiterals(FALSE); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_D, 0), *fStatus); + appendOp(URX_BUILD(URX_BACKSLASH_D, 0)); break; case doBackslashG: fixLiterals(FALSE); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_G, 0), *fStatus); + appendOp(URX_BUILD(URX_BACKSLASH_G, 0)); break; case doBackslashS: fixLiterals(FALSE); - fRXPat->fCompiledPat->addElement( - URX_BUILD(URX_STAT_SETREF_N, URX_ISSPACE_SET), *fStatus); + appendOp(URX_BUILD(URX_STAT_SETREF_N, URX_ISSPACE_SET)); break; case doBackslashs: fixLiterals(FALSE); - fRXPat->fCompiledPat->addElement( - URX_BUILD(URX_STATIC_SETREF, URX_ISSPACE_SET), *fStatus); + appendOp(URX_BUILD(URX_STATIC_SETREF, URX_ISSPACE_SET)); break; case doBackslashW: fixLiterals(FALSE); - fRXPat->fCompiledPat->addElement( - URX_BUILD(URX_STAT_SETREF_N, URX_ISWORD_SET), *fStatus); + appendOp(URX_BUILD(URX_STAT_SETREF_N, URX_ISWORD_SET)); break; case doBackslashw: fixLiterals(FALSE); - fRXPat->fCompiledPat->addElement( - URX_BUILD(URX_STATIC_SETREF, URX_ISWORD_SET), *fStatus); + appendOp(URX_BUILD(URX_STATIC_SETREF, URX_ISWORD_SET)); break; case doBackslashX: fixLiterals(FALSE); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_X, 0), *fStatus); + appendOp(URX_BUILD(URX_BACKSLASH_X, 0)); break; case doBackslashZ: fixLiterals(FALSE); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_DOLLAR, 0), *fStatus); + appendOp(URX_BUILD(URX_DOLLAR, 0)); break; case doBackslashz: fixLiterals(FALSE); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_Z, 0), *fStatus); + appendOp(URX_BUILD(URX_BACKSLASH_Z, 0)); break; case doEscapeError: @@ -1321,7 +1305,7 @@ UBool RegexCompile::doParseActions(int32_t action) } else { op = URX_BUILD(URX_BACKREF, groupNum); } - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); } break; @@ -1342,22 +1326,21 @@ UBool RegexCompile::doParseActions(int32_t action) { // Emit the STO_SP int32_t topLoc = blockTopLoc(TRUE); - int32_t stoLoc = fRXPat->fDataSize; - fRXPat->fDataSize++; // Reserve the data location for storing save stack ptr. + int32_t stoLoc = allocateData(1); // Reserve the data location for storing save stack ptr. int32_t op = URX_BUILD(URX_STO_SP, stoLoc); fRXPat->fCompiledPat->setElementAt(op, topLoc); // Emit the STATE_SAVE op = URX_BUILD(URX_STATE_SAVE, fRXPat->fCompiledPat->size()+2); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); // Emit the JMP op = URX_BUILD(URX_JMP, topLoc+1); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); // Emit the LD_SP op = URX_BUILD(URX_LD_SP, stoLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); } break; @@ -1377,8 +1360,7 @@ UBool RegexCompile::doParseActions(int32_t action) insertOp(topLoc); // emit STO_SP loc - int32_t stoLoc = fRXPat->fDataSize; - fRXPat->fDataSize++; // Reserve the data location for storing save stack ptr. + int32_t stoLoc = allocateData(1); // Reserve the data location for storing save stack ptr. int32_t op = URX_BUILD(URX_STO_SP, stoLoc); fRXPat->fCompiledPat->setElementAt(op, topLoc); @@ -1389,11 +1371,11 @@ UBool RegexCompile::doParseActions(int32_t action) // Append the JMP operation. op = URX_BUILD(URX_JMP, topLoc+1); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); // Emit the LD_SP loc op = URX_BUILD(URX_LD_SP, stoLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); } break; @@ -1412,8 +1394,7 @@ UBool RegexCompile::doParseActions(int32_t action) insertOp(topLoc); // Emit the STO_SP - int32_t stoLoc = fRXPat->fDataSize; - fRXPat->fDataSize++; // Reserve the data location for storing save stack ptr. + int32_t stoLoc = allocateData(1); // Reserve the data location for storing save stack ptr. int32_t op = URX_BUILD(URX_STO_SP, stoLoc); fRXPat->fCompiledPat->setElementAt(op, topLoc); @@ -1424,7 +1405,7 @@ UBool RegexCompile::doParseActions(int32_t action) // Emit the LD_SP op = URX_BUILD(URX_LD_SP, stoLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); } break; @@ -1481,8 +1462,8 @@ UBool RegexCompile::doParseActions(int32_t action) // is an '|' alternation within the parens. { fixLiterals(FALSE); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); + appendOp(URX_BUILD(URX_NOP, 0)); + appendOp(URX_BUILD(URX_NOP, 0)); // On the Parentheses stack, start a new frame and add the postions // of the two NOPs (a normal non-capturing () frame, except for the @@ -1862,7 +1843,7 @@ void RegexCompile::fixLiterals(UBool split) { } else { op = URX_BUILD(URX_ONECHAR, lastCodePoint); } - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); } else { // Two or more chars, emit a URX_STRING to match them. if (fModeFlags & UREGEX_CASE_INSENSITIVE) { @@ -1872,20 +1853,39 @@ void RegexCompile::fixLiterals(UBool split) { // into two single char ops, for efficiency. op = URX_BUILD(URX_STRING, fRXPat->fLiteralText.length()); } - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); op = URX_BUILD(URX_STRING_LEN, fLiteralChars.length()); - fRXPat->fCompiledPat->addElement(op, *fStatus); - + appendOp(op); + // Add this string into the accumulated strings of the compiled pattern. + // The total size of the accumulated strings must be restricted to 24 bits because + // string indexes appear as compiled pattern operand values. + // This is the only place that the pattern.fLiteralText string is modified. + fRXPat->fLiteralText.append(fLiteralChars); + if (U_SUCCESS(*fStatus) && fRXPat->fLiteralText.length() > 0x00ffffff) { + *fStatus = U_REGEX_PATTERN_TOO_BIG; + } } fLiteralChars.remove(); } - - +//------------------------------------------------------------------------------ +// +// appendOp() Append a new instruction onto the compiled pattern +// Includes error checking, limiting the size of the +// pattern to lengths that can be represented in the +// 24 bit operand field of an instruction. +// +//------------------------------------------------------------------------------ +void RegexCompile::appendOp(int32_t op) { + fRXPat->fCompiledPat->addElement(op, *fStatus); + if ((fRXPat->fCompiledPat->size() > 0x00fffff0) && U_SUCCESS(*fStatus)) { + *fStatus = U_REGEX_PATTERN_TOO_BIG; + } +} //------------------------------------------------------------------------------ @@ -1947,6 +1947,58 @@ void RegexCompile::insertOp(int32_t where) { } +//------------------------------------------------------------------------------ +// +// allocateData() Allocate storage in the matcher's static data area. +// Return the index for the newly allocated data. +// The storage won't actually exist until we are running a match +// operation, but the storage indexes are inserted into various +// opcodes while compiling the pattern. +// +//------------------------------------------------------------------------------ +int32_t RegexCompile::allocateData(int32_t size) { + if (U_FAILURE(*fStatus)) { + return 0; + } + if (size <= 0 || size > 0x100 || fRXPat->fDataSize < 0) { + *fStatus = U_REGEX_INTERNAL_ERROR; + return 0; + } + int32_t dataIndex = fRXPat->fDataSize; + fRXPat->fDataSize += size; + if (fRXPat->fDataSize >= 0x00fffff0) { + *fStatus = U_REGEX_PATTERN_TOO_BIG; + } + return dataIndex; +} + + +//------------------------------------------------------------------------------ +// +// allocateStackData() Allocate space in the back-tracking stack frame. +// Return the index for the newly allocated data. +// The frame indexes are inserted into various +// opcodes while compiling the pattern, meaning that frame +// size must be restricted to the size that will fit +// as an operand (24 bits). +// +//------------------------------------------------------------------------------ +int32_t RegexCompile::allocateStackData(int32_t size) { + if (U_FAILURE(*fStatus)) { + return 0; + } + if (size <= 0 || size > 0x100 || fRXPat->fFrameSize < 0) { + *fStatus = U_REGEX_INTERNAL_ERROR; + return 0; + } + int32_t dataIndex = fRXPat->fFrameSize; + fRXPat->fFrameSize += size; + if (fRXPat->fFrameSize >= 0x00fffff0) { + *fStatus = U_REGEX_PATTERN_TOO_BIG; + } + return dataIndex; +} + //------------------------------------------------------------------------------ // @@ -2065,7 +2117,7 @@ void RegexCompile::handleCloseParen() { int32_t frameVarLocation = URX_VAL(captureOp); int32_t endCaptureOp = URX_BUILD(URX_END_CAPTURE, frameVarLocation); - fRXPat->fCompiledPat->addElement(endCaptureOp, *fStatus); + appendOp(endCaptureOp); } break; case atomic: @@ -2077,7 +2129,7 @@ void RegexCompile::handleCloseParen() { U_ASSERT(URX_TYPE(stoOp) == URX_STO_SP); int32_t stoLoc = URX_VAL(stoOp); int32_t ldOp = URX_BUILD(URX_LD_SP, stoLoc); - fRXPat->fCompiledPat->addElement(ldOp, *fStatus); + appendOp(ldOp); } break; @@ -2087,7 +2139,7 @@ void RegexCompile::handleCloseParen() { U_ASSERT(URX_TYPE(startOp) == URX_LA_START); int32_t dataLoc = URX_VAL(startOp); int32_t op = URX_BUILD(URX_LA_END, dataLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); } break; @@ -2098,11 +2150,11 @@ void RegexCompile::handleCloseParen() { U_ASSERT(URX_TYPE(startOp) == URX_LA_START); int32_t dataLoc = URX_VAL(startOp); int32_t op = URX_BUILD(URX_LA_END, dataLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); op = URX_BUILD(URX_BACKTRACK, 0); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); op = URX_BUILD(URX_LA_END, dataLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); // Patch the URX_SAVE near the top of the block. // The destination of the SAVE is the final LA_END that was just added. @@ -2123,9 +2175,9 @@ void RegexCompile::handleCloseParen() { U_ASSERT(URX_TYPE(startOp) == URX_LB_START); int32_t dataLoc = URX_VAL(startOp); int32_t op = URX_BUILD(URX_LB_END, dataLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); - op = URX_BUILD(URX_LA_END, dataLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); + op = URX_BUILD(URX_LA_END, dataLoc); + appendOp(op); // Determine the min and max bounds for the length of the // string that the pattern can match. @@ -2133,6 +2185,10 @@ void RegexCompile::handleCloseParen() { int32_t patEnd = fRXPat->fCompiledPat->size() - 1; int32_t minML = minMatchLength(fMatchOpenParen, patEnd); int32_t maxML = maxMatchLength(fMatchOpenParen, patEnd); + if (URX_TYPE(maxML) != 0) { + error(U_REGEX_LOOK_BEHIND_LIMIT); + break; + } if (maxML == INT32_MAX) { error(U_REGEX_LOOK_BEHIND_LIMIT); break; @@ -2158,7 +2214,7 @@ void RegexCompile::handleCloseParen() { U_ASSERT(URX_TYPE(startOp) == URX_LB_START); int32_t dataLoc = URX_VAL(startOp); int32_t op = URX_BUILD(URX_LBN_END, dataLoc); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); // Determine the min and max bounds for the length of the // string that the pattern can match. @@ -2166,6 +2222,10 @@ void RegexCompile::handleCloseParen() { int32_t patEnd = fRXPat->fCompiledPat->size() - 1; int32_t minML = minMatchLength(fMatchOpenParen, patEnd); int32_t maxML = maxMatchLength(fMatchOpenParen, patEnd); + if (URX_TYPE(maxML) != 0) { + error(U_REGEX_LOOK_BEHIND_LIMIT); + break; + } if (maxML == INT32_MAX) { error(U_REGEX_LOOK_BEHIND_LIMIT); break; @@ -2220,7 +2280,7 @@ void RegexCompile::compileSet(UnicodeSet *theSet) case 0: { // Set of no elements. Always fails to match. - fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKTRACK, 0), *fStatus); + appendOp(URX_BUILD(URX_BACKTRACK, 0)); delete theSet; } break; @@ -2242,7 +2302,7 @@ void RegexCompile::compileSet(UnicodeSet *theSet) int32_t setNumber = fRXPat->fSets->size(); fRXPat->fSets->addElement(theSet, *fStatus); int32_t setOp = URX_BUILD(URX_SETREF, setNumber); - fRXPat->fCompiledPat->addElement(setOp, *fStatus); + appendOp(setOp); } } } @@ -2281,11 +2341,8 @@ void RegexCompile::compileInterval(int32_t InitOp, int32_t LoopOp) // counterLoc --> Loop counter // +1 --> Input index (for breaking non-progressing loops) // (Only present if unbounded upper limit on loop) - int32_t counterLoc = fRXPat->fFrameSize; - fRXPat->fFrameSize++; - if (fIntervalUpper < 0) { - fRXPat->fFrameSize++; - } + int32_t dataSize = fIntervalUpper < 0 ? 2 : 1; + int32_t counterLoc = allocateStackData(dataSize); int32_t op = URX_BUILD(InitOp, counterLoc); fRXPat->fCompiledPat->setElementAt(op, topOfBlock); @@ -2305,7 +2362,7 @@ void RegexCompile::compileInterval(int32_t InitOp, int32_t LoopOp) // Apend the CTR_LOOP op. The operand is the location of the CTR_INIT op. // Goes at end of the block being looped over, so just append to the code so far. op = URX_BUILD(LoopOp, topOfBlock); - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); if ((fIntervalLow & 0xff000000) != 0 || (fIntervalUpper > 0 && (fIntervalUpper & 0xff000000) != 0)) { @@ -2329,7 +2386,15 @@ UBool RegexCompile::compileInlineInterval() { int32_t topOfBlock = blockTopLoc(FALSE); if (fIntervalUpper == 0) { // Pathological case. Attempt no matches, as if the block doesn't exist. + // Discard the generated code for the block. + // If the block included parens, discard the info pertaining to them as well. fRXPat->fCompiledPat->setSize(topOfBlock); + if (fMatchOpenParen >= topOfBlock) { + fMatchOpenParen = -1; + } + if (fMatchCloseParen >= topOfBlock) { + fMatchCloseParen = -1; + } return TRUE; } @@ -2364,12 +2429,12 @@ UBool RegexCompile::compileInlineInterval() { int32_t i; for (i=1; i<fIntervalUpper; i++ ) { if (i == fIntervalLow) { - fRXPat->fCompiledPat->addElement(saveOp, *fStatus); + appendOp(saveOp); } if (i > fIntervalLow) { - fRXPat->fCompiledPat->addElement(saveOp, *fStatus); + appendOp(saveOp); } - fRXPat->fCompiledPat->addElement(op, *fStatus); + appendOp(op); } return TRUE; } diff --git a/source/i18n/regexcmp.h b/source/i18n/regexcmp.h index 0041beb..b17e6bd 100644 --- a/source/i18n/regexcmp.h +++ b/source/i18n/regexcmp.h @@ -103,6 +103,11 @@ private: void fixLiterals(UBool split=FALSE); // Generate code for pending literal characters. void insertOp(int32_t where); // Open up a slot for a new op in the // generated code at the specified location. + void appendOp(int32_t op); // Append a new op to the compiled pattern. + int32_t allocateData(int32_t size); // Allocate space in the matcher data area. + // Return index of the newly allocated data. + int32_t allocateStackData(int32_t size); // Allocate space in the match back-track stack frame. + // Return offset index in the frame. int32_t minMatchLength(int32_t start, int32_t end); int32_t maxMatchLength(int32_t start, @@ -182,7 +187,9 @@ private: int32_t fMatchOpenParen; // The position in the compiled pattern // of the slot reserved for a state save // at the start of the most recently processed - // parenthesized block. + // parenthesized block. Updated when processing + // a close to the location for the corresponding open. + int32_t fMatchCloseParen; // The position in the pattern of the first // location after the most recently processed // parenthesized block. diff --git a/source/i18n/regeximp.cpp b/source/i18n/regeximp.cpp index 32df5fe..395b3fc 100644 --- a/source/i18n/regeximp.cpp +++ b/source/i18n/regeximp.cpp @@ -12,6 +12,8 @@ #if !UCONFIG_NO_REGULAR_EXPRESSIONS #include "regeximp.h" + +#include "uassert.h" #include "unicode/utf16.h" U_NAMESPACE_BEGIN @@ -113,6 +115,26 @@ int64_t CaseFoldingUCharIterator::getIndex() { return fIndex; } +// Assemble a pcode instruction from the opcode and operand values. +// Out-of-range values should not occur - if they do it is from an internal +// error in the regex compiler. + +// TODO: move into regexcmp, where it has access to fStatus. +// NOP cleanly if U_FAILURE. +// Set U_REGEX_INTERNAL_ERROR on bad operands. + +int32_t URX_BUILD(int32_t type, int32_t val) { + if (type < 0 || type > 255) { + U_ASSERT(FALSE); + type = URX_RESERVED_OP; + } + if (val > 0x00ffffff) { + U_ASSERT(FALSE); + val = 0; + } + return (type << 24) | val; +} + U_NAMESPACE_END diff --git a/source/i18n/regeximp.h b/source/i18n/regeximp.h index 31d333c..00da653 100644 --- a/source/i18n/regeximp.h +++ b/source/i18n/regeximp.h @@ -254,7 +254,7 @@ enum { // // Convenience macros for assembling and disassembling a compiled operation. // -#define URX_BUILD(type, val) (int32_t)((type << 24) | (val)) +int32_t URX_BUILD(int32_t val, int32_t type); #define URX_TYPE(x) ((uint32_t)(x) >> 24) #define URX_VAL(x) ((x) & 0xffffff) diff --git a/source/i18n/ucol.cpp b/source/i18n/ucol.cpp index b0d819f..e5ff44e 100644 --- a/source/i18n/ucol.cpp +++ b/source/i18n/ucol.cpp @@ -2259,6 +2259,9 @@ inline UChar getNextNormalizedChar(collIterate *data) if (data->pos + 1 == data->endp) { return *(data->pos ++); } + if (data->pos >= data->endp) { + return (UChar) -1; // return U+FFFF (non-char) to indicate an error + } } else { if (innormbuf) { @@ -2821,7 +2824,13 @@ uint32_t ucol_prv_getSpecialCE(const UCollator *coll, UChar ch, uint32_t CE, col } } } else if (U16_IS_LEAD(schar)) { - miss = U16_GET_SUPPLEMENTARY(schar, getNextNormalizedChar(source)); + UChar nextChar = getNextNormalizedChar(source); + const UChar* prevPos = source->pos; + if (U16_IS_TRAIL(nextChar)) { + miss = U16_GET_SUPPLEMENTARY(schar, nextChar); + } else if (prevPos < source->pos) { + goBackOne(source); + } } uint8_t sCC; diff --git a/source/test/intltest/regextst.cpp b/source/test/intltest/regextst.cpp index 8e750c9..9fd9f43 100644 --- a/source/test/intltest/regextst.cpp +++ b/source/test/intltest/regextst.cpp @@ -131,6 +131,9 @@ void RegexTest::runIndexedTest( int32_t index, UBool exec, const char* &name, ch case 21: name = "Bug 9283"; if (exec) Bug9283(); break; + case 22: name = "TestBug11371"; + if (exec) TestBug11371(); + break; default: name = ""; break; //needed to end loop @@ -5229,5 +5232,47 @@ void RegexTest::CheckInvBufSize() { } } -#endif /* !UCONFIG_NO_REGULAR_EXPRESSIONS */ +void RegexTest::TestBug11371() { + UErrorCode status = U_ZERO_ERROR; + UnicodeString patternString; + + for (int i=0; i<8000000; i++) { + patternString.append(UnicodeString("()")); + } + LocalPointer<RegexPattern> compiledPat(RegexPattern::compile(patternString, 0, status)); + if (status != U_REGEX_PATTERN_TOO_BIG) { + errln("File %s, line %d expected status=U_REGEX_PATTERN_TOO_BIG; got %s.", + __FILE__, __LINE__, u_errorName(status)); + } + + status = U_ZERO_ERROR; + patternString = "("; + for (int i=0; i<20000000; i++) { + patternString.append(UnicodeString("A++")); + } + patternString.append(UnicodeString("){0}B++")); + LocalPointer<RegexPattern> compiledPat2(RegexPattern::compile(patternString, 0, status)); + if (status != U_REGEX_PATTERN_TOO_BIG) { + errln("File %s, line %d expected status=U_REGEX_PATTERN_TOO_BIG; got %s.", + __FILE__, __LINE__, u_errorName(status)); + } + // Pattern with too much string data, such that string indexes overflow operand data. + status = U_ZERO_ERROR; + patternString = ""; + while (patternString.length() < 0x00ffffff) { + patternString.append(UnicodeString("stuff and things dont you know, these are a few of my favorite strings\n")); + } + patternString.append(UnicodeString("X? trailing string")); + LocalPointer<RegexPattern> compiledPat3(RegexPattern::compile(patternString, 0, status)); + compiledPat3->dumpPattern(); + if (status != U_REGEX_PATTERN_TOO_BIG) { + errln("File %s, line %d expected status=U_REGEX_PATTERN_TOO_BIG; got %s.", + __FILE__, __LINE__, u_errorName(status)); + } + + + +} + +#endif /* !UCONFIG_NO_REGULAR_EXPRESSIONS */ diff --git a/source/test/intltest/regextst.h b/source/test/intltest/regextst.h index 22a77ea..de3af29 100644 --- a/source/test/intltest/regextst.h +++ b/source/test/intltest/regextst.h @@ -47,6 +47,7 @@ public: virtual void Bug7029(); virtual void Bug9283(); virtual void CheckInvBufSize(); + virtual void TestBug11371(); // The following functions are internal to the regexp tests. virtual void assertUText(const char *expected, UText *actual, const char *file, int line); diff --git a/source/test/testdata/regextst.txt b/source/test/testdata/regextst.txt index 5716ab5..f0b00ab 100644 --- a/source/test/testdata/regextst.txt +++ b/source/test/testdata/regextst.txt @@ -1173,6 +1173,24 @@ "(?<=(?:){11})bc" "<0>bc</0>" # Empty (?:) expression. +# Bug 11369 +# Incorrect optimization of patterns with a zero length quantifier {0} + +"(.|b)(|b){0}\$(?#xxx){3}(?>\D*)" "AAAAABBBBBCCCCCDDDDEEEEE" +"(|b)ab(c)" "<0><1></1>ab<2>c</2></0>" +"(|b){0}a{3}(D*)" "<0>aaa<2></2></0>" +"(|b){0,1}a{3}(D*)" "<0><1></1>aaa<2></2></0>" +"((|b){0})a{3}(D*)" "<0><1></1>aaa<3></3></0>" + +# Bug 11370 +# Max match length computation of look-behind expression gives result that is too big to fit in the +# in the 24 bit operand portion of the compiled code. Expressions should fail to compile +# (Look-behind match length must be bounded. This case is treated as unbounded, an error.) + +"(?<!(0123456789a){10000000})x" E "no match" +"(?<!\\ubeaf(\\ubeaf{11000}){11000})" E "no match" + + # Random debugging, Temporary # #"^(?:a?b?)*$" "a--" |