intermediate changes

ref:cde9a383711a11544ce7e107a78147fb96cc4029

5 files changed, 158 insertions, 0 deletions

diff --git a/library/cpp/html/pcdata/pcdata.cpp b/library/cpp/html/pcdata/pcdata.cpp
new file mode 100644
index 0000000000..740c240fd2
--- /dev/null
+++ b/library/cpp/html/pcdata/pcdata.cpp
@@ -0,0 +1,81 @@
+#include "pcdata.h"
+
+#include <util/string/strspn.h>
+
+static TCompactStrSpn sspn("\"<>&'");
+
+static void EncodeHtmlPcdataAppendInternal(const TStringBuf str, TString& strout, bool qAmp) {
+    const char* s = str.data();
+    const char* e = s + str.length();
+
+    for (;;) {
+        const char* next = sspn.FindFirstOf(s, e);
+
+        strout.AppendNoAlias(s, next - s);
+        s = next;
+
+        if (s == e)
+            break;
+
+        switch (*s) {
+            case '\"':
+                strout += TStringBuf("&quot;");
+                ++s;
+                break;
+
+            case '<':
+                strout += TStringBuf("&lt;");
+                ++s;
+                break;
+
+            case '>':
+                strout += TStringBuf("&gt;");
+                ++s;
+                break;
+
+            case '\'':
+                strout += TStringBuf("&#39;");
+                ++s;
+                break;
+
+            case '&':
+                if (qAmp)
+                    strout += TStringBuf("&amp;");
+                else
+                    strout += TStringBuf("&");
+                ++s;
+                break;
+        }
+    }
+}
+
+void EncodeHtmlPcdataAppend(const TStringBuf str, TString& strout) {
+    EncodeHtmlPcdataAppendInternal(str, strout, true);
+}
+
+TString EncodeHtmlPcdata(const TStringBuf str, bool qAmp) {
+    TString strout;
+    EncodeHtmlPcdataAppendInternal(str, strout, qAmp);
+    return strout;
+}
+
+TString DecodeHtmlPcdata(const TString& sz) {
+    TString res;
+    const char* codes[] = {"&quot;", "&lt;", "&gt;", "&#39;", "&#039;", "&amp;", "&apos;", nullptr};
+    const char chars[] = {'\"', '<', '>', '\'', '\'', '&', '\''};
+    for (size_t i = 0; i < sz.length(); ++i) {
+        char c = sz[i];
+        if (c == '&') {
+            for (const char** p = codes; *p; ++p) {
+                size_t len = strlen(*p);
+                if (strncmp(sz.c_str() + i, *p, len) == 0) {
+                    i += len - 1;
+                    c = chars[p - codes];
+                    break;
+                }
+            }
+        }
+        res += c;
+    }
+    return res;
+}
diff --git a/library/cpp/html/pcdata/pcdata.h b/library/cpp/html/pcdata/pcdata.h
new file mode 100644
index 0000000000..7dd741f53d
--- /dev/null
+++ b/library/cpp/html/pcdata/pcdata.h
@@ -0,0 +1,10 @@
+#pragma once
+
+#include <util/generic/fwd.h>
+
+/// Converts a text into HTML-code. Special characters of HTML («<», «>», ...) replaced with entities.
+TString EncodeHtmlPcdata(const TStringBuf str, bool qAmp = true);
+void EncodeHtmlPcdataAppend(const TStringBuf str, TString& strout);
+
+/// Reverse of EncodeHtmlPcdata()
+TString DecodeHtmlPcdata(const TString& sz);
diff --git a/library/cpp/html/pcdata/pcdata_ut.cpp b/library/cpp/html/pcdata/pcdata_ut.cpp
new file mode 100644
index 0000000000..5833f8bc59
--- /dev/null
+++ b/library/cpp/html/pcdata/pcdata_ut.cpp
@@ -0,0 +1,48 @@
+#include "pcdata.h"
+
+#include <library/cpp/testing/unittest/registar.h>
+
+Y_UNIT_TEST_SUITE(TPcdata) {
+    Y_UNIT_TEST(TestStress) {
+        {
+            ui64 key = 0x000017C0B76C4E87ull;
+            TString res = EncodeHtmlPcdata(TStringBuf((const char*)&key, sizeof(key)));
+        }
+
+        for (size_t i = 0; i < 1000; ++i) {
+            const TString s = NUnitTest::RandomString(i, i);
+
+            UNIT_ASSERT_VALUES_EQUAL(DecodeHtmlPcdata(EncodeHtmlPcdata(s)), s);
+        }
+    }
+
+    Y_UNIT_TEST(Test1) {
+        const TString tests[] = {
+            "qw&qw",
+            "&<",
+            ">&qw",
+            "\'&aaa"};
+
+        for (auto s : tests) {
+            UNIT_ASSERT_VALUES_EQUAL(DecodeHtmlPcdata(EncodeHtmlPcdata(s)), s);
+        }
+    }
+
+    Y_UNIT_TEST(Test2) {
+        UNIT_ASSERT_VALUES_EQUAL(EncodeHtmlPcdata("&qqq"), "&amp;qqq");
+    }
+
+    Y_UNIT_TEST(TestEncodeHtmlPcdataAppend) {
+        TString s;
+        EncodeHtmlPcdataAppend("m&m", s);
+        EncodeHtmlPcdataAppend("'s", s);
+        UNIT_ASSERT_VALUES_EQUAL(EncodeHtmlPcdata("m&m's"), s);
+        UNIT_ASSERT_VALUES_EQUAL("m&amp;m&#39;s", s);
+    }
+
+    Y_UNIT_TEST(TestStrangeAmpParameter) {
+        UNIT_ASSERT_VALUES_EQUAL(EncodeHtmlPcdata("m&m's", true), "m&amp;m&#39;s");
+        UNIT_ASSERT_VALUES_EQUAL(EncodeHtmlPcdata("m&m's"), "m&amp;m&#39;s"); //default
+        UNIT_ASSERT_VALUES_EQUAL(EncodeHtmlPcdata("m&m's", false), "m&m&#39;s");
+    }
+}
diff --git a/library/cpp/html/pcdata/ut/ya.make b/library/cpp/html/pcdata/ut/ya.make
new file mode 100644
index 0000000000..bc947baa89
--- /dev/null
+++ b/library/cpp/html/pcdata/ut/ya.make
@@ -0,0 +1,9 @@
+UNITTEST_FOR(library/cpp/html/pcdata)
+
+OWNER(vladon)
+
+SRCS(
+    pcdata_ut.cpp
+)
+
+END()
diff --git a/library/cpp/html/pcdata/ya.make b/library/cpp/html/pcdata/ya.make
new file mode 100644
index 0000000000..637220e706
--- /dev/null
+++ b/library/cpp/html/pcdata/ya.make
@@ -0,0 +1,10 @@
+LIBRARY()
+
+OWNER(vladon)
+
+SRCS(
+    pcdata.cpp
+    pcdata.h
+)
+
+END()