Remove all usages of FXSYS_iswASCIIalnum

Instances are either replaced with FXSYS_iswalnum, which calls out to
the ICU library to do the proper Unicode operations, or have been
converted to a isascii && isalnum pair, if ASCII alnum is actually
what was wanted.

BUG=pdfium:1035

Change-Id: I959ec8739a4d020e61562180393ab8113a81577c
Reviewed-on: https://pdfium-review.googlesource.com/28430
Reviewed-by: dsinclair <dsinclair@chromium.org>
Commit-Queue: Ryan Harrison <rharrison@chromium.org>

8 files changed, 14 insertions, 9 deletions

diff --git a/BUILD.gn b/BUILD.gn
index 485afe28bf..d4e2d46a31 100644
--- a/BUILD.gn
+++ b/BUILD.gn
@@ -883,6 +883,7 @@ jumbo_static_library("fxcrt") {
   public_deps = [
     ":freetype_common",
     "third_party:pdfium_base",
+    "//third_party/icu:icuuc",
   ]
 
   if (pdf_enable_xfa) {
diff --git a/core/fpdftext/cpdf_linkextract.cpp b/core/fpdftext/cpdf_linkextract.cpp
index f8144a171b..c3cf4fc1ef 100644
--- a/core/fpdftext/cpdf_linkextract.cpp
+++ b/core/fpdftext/cpdf_linkextract.cpp
@@ -246,7 +246,7 @@ bool CPDF_LinkExtract::CheckMailLink(WideString* str) {
   size_t pPos = aPos.value();  // Used to track the position of '@' or '.'.
   for (size_t i = aPos.value(); i > 0; i--) {
     wchar_t ch = (*str)[i - 1];
-    if (ch == L'_' || ch == L'-' || FXSYS_iswASCIIalnum(ch))
+    if (ch == L'_' || ch == L'-' || FXSYS_iswalnum(ch))
       continue;
 
     if (ch != L'.' || i == pPos || i == 1) {
@@ -282,7 +282,7 @@ bool CPDF_LinkExtract::CheckMailLink(WideString* str) {
   pPos = 0;  // Used to track the position of '.'.
   for (size_t i = aPos.value() + 1; i < nLen; i++) {
     wchar_t wch = (*str)[i];
-    if (wch == L'-' || FXSYS_iswASCIIalnum(wch))
+    if (wch == L'-' || FXSYS_iswalnum(wch))
       continue;
 
     if (wch != L'.' || i == pPos + 1) {
diff --git a/core/fpdftext/cpdf_textpage.cpp b/core/fpdftext/cpdf_textpage.cpp
index 5019eabaf5..7541dae6a7 100644
--- a/core/fpdftext/cpdf_textpage.cpp
+++ b/core/fpdftext/cpdf_textpage.cpp
@@ -1210,7 +1210,7 @@ bool CPDF_TextPage::IsHyphen(wchar_t curChar) const {
 
   if ((iter + 1) != curText.rend()) {
     iter++;
-    if (FXSYS_iswASCIIalpha(*iter) && FXSYS_iswASCIIalnum(curChar))
+    if (FXSYS_iswASCIIalpha(*iter) && FXSYS_iswalnum(curChar))
       return true;
   }
 
diff --git a/core/fxcrt/DEPS b/core/fxcrt/DEPS
new file mode 100644
index 0000000000..2be03524b6
--- /dev/null
+++ b/core/fxcrt/DEPS
@@ -0,0 +1,3 @@
+include_rules = [
+  '+third_party/icu',
+]
diff --git a/core/fxcrt/css/cfx_cssselector.cpp b/core/fxcrt/css/cfx_cssselector.cpp
index 76cb846890..8a22b12287 100644
--- a/core/fxcrt/css/cfx_cssselector.cpp
+++ b/core/fxcrt/css/cfx_cssselector.cpp
@@ -16,9 +16,9 @@ namespace {
 int32_t GetCSSNameLen(const wchar_t* psz, const wchar_t* pEnd) {
   const wchar_t* pStart = psz;
   while (psz < pEnd) {
-    wchar_t wch = *psz;
-    if (!FXSYS_iswASCIIalnum(wch) && wch != '_' && wch != '-')
+    if (!isascii(*psz) || (!isalnum(*psz) && *psz != '_' && *psz != '-')) {
       break;
+    }
     ++psz;
   }
   return psz - pStart;
diff --git a/core/fxcrt/fx_extension.h b/core/fxcrt/fx_extension.h
index 491d4b29b7..bd0ac8b303 100644
--- a/core/fxcrt/fx_extension.h
+++ b/core/fxcrt/fx_extension.h
@@ -12,6 +12,7 @@
 #include <memory>
 
 #include "core/fxcrt/fx_string.h"
+#include "third_party/icu/source/common/unicode/uchar.h"
 
 #define FX_INVALID_OFFSET static_cast<uint32_t>(-1)
 
@@ -45,8 +46,8 @@ inline bool FXSYS_iswASCIIalpha(wchar_t wch) {
   return FXSYS_isASCIIupper(wch) || FXSYS_isASCIIlower(wch);
 }
 
-inline bool FXSYS_iswASCIIalnum(wchar_t wch) {
-  return FXSYS_iswASCIIalpha(wch) || std::iswdigit(wch);
+inline bool FXSYS_iswalnum(wchar_t c) {
+  return u_isalnum(c);
 }
 
 inline bool FXSYS_iswASCIIspace(wchar_t c) {
diff --git a/fxjs/cjs_publicmethods.cpp b/fxjs/cjs_publicmethods.cpp
index 28df0a305b..81a84d286b 100644
--- a/fxjs/cjs_publicmethods.cpp
+++ b/fxjs/cjs_publicmethods.cpp
@@ -285,7 +285,7 @@ bool CJS_PublicMethods::MaskSatisfied(wchar_t c_Change, wchar_t c_Mask) {
     case L'A':
       return FXSYS_iswASCIIalpha(c_Change);
     case L'O':
-      return FXSYS_iswASCIIalnum(c_Change);
+      return isascii(c_Change) && isalnum(c_Change);
     case L'X':
       return true;
     default:
diff --git a/fxjs/cjs_util.cpp b/fxjs/cjs_util.cpp
index d552fcdf34..d73c238488 100644
--- a/fxjs/cjs_util.cpp
+++ b/fxjs/cjs_util.cpp
@@ -311,7 +311,7 @@ WideString CJS_Util::printx(const WideString& wsFormat,
       } break;
       case 'X': {
         if (iSourceIdx < wsSource.GetLength()) {
-          if (FXSYS_iswASCIIalnum(wsSource[iSourceIdx])) {
+          if (isascii(wsSource[iSourceIdx]) && isalnum(wsSource[iSourceIdx])) {
             wsResult += TranslateCase(wsSource[iSourceIdx], eCaseMode);
             ++iFormatIdx;
           }