intermediate changes

ref:cde9a383711a11544ce7e107a78147fb96cc4029

11 files changed, 303 insertions, 0 deletions

diff --git a/library/cpp/html/escape/escape.cpp b/library/cpp/html/escape/escape.cpp
new file mode 100644
index 0000000000..5b8ed60f04
--- /dev/null
+++ b/library/cpp/html/escape/escape.cpp
@@ -0,0 +1,66 @@
+#include "escape.h"
+
+#include <util/generic/array_size.h>
+#include <util/generic/strbuf.h>
+
+namespace NHtml {
+    namespace {
+        struct TReplace {
+            char Char;
+            bool ForText;
+            TStringBuf Entity;
+        };
+
+        TReplace Escapable[] = {
+            {'"', false, TStringBuf("&quot;")},
+            {'&', true, TStringBuf("&amp;")},
+            {'<', true, TStringBuf("&lt;")},
+            {'>', true, TStringBuf("&gt;")},
+        };
+
+        TString EscapeImpl(const TString& value, bool isText) {
+            auto ci = value.begin();
+            // Looking for escapable characters.
+            for (; ci != value.end(); ++ci) {
+                for (size_t i = (isText ? 1 : 0); i < Y_ARRAY_SIZE(Escapable); ++i) {
+                    if (*ci == Escapable[i].Char) {
+                        goto escape;
+                    }
+                }
+            }
+
+            // There is no escapable characters, so return original value.
+            return value;
+
+        escape:
+            TString tmp = TString(value.begin(), ci);
+
+            for (; ci != value.end(); ++ci) {
+                size_t i = (isText ? 1 : 0);
+
+                for (; i < Y_ARRAY_SIZE(Escapable); ++i) {
+                    if (*ci == Escapable[i].Char) {
+                        tmp += Escapable[i].Entity;
+                        break;
+                    }
+                }
+
+                if (i == Y_ARRAY_SIZE(Escapable)) {
+                    tmp += *ci;
+                }
+            }
+
+            return tmp;
+        }
+
+    }
+
+    TString EscapeAttributeValue(const TString& value) {
+        return EscapeImpl(value, false);
+    }
+
+    TString EscapeText(const TString& value) {
+        return EscapeImpl(value, true);
+    }
+
+}
diff --git a/library/cpp/html/escape/escape.h b/library/cpp/html/escape/escape.h
new file mode 100644
index 0000000000..1c45fc5193
--- /dev/null
+++ b/library/cpp/html/escape/escape.h
@@ -0,0 +1,9 @@
+#pragma once
+
+#include <util/generic/string.h>
+
+namespace NHtml {
+    TString EscapeAttributeValue(const TString& value);
+    TString EscapeText(const TString& value);
+
+}
diff --git a/library/cpp/html/escape/ut/escape_ut.cpp b/library/cpp/html/escape/ut/escape_ut.cpp
new file mode 100644
index 0000000000..cd7b955138
--- /dev/null
+++ b/library/cpp/html/escape/ut/escape_ut.cpp
@@ -0,0 +1,16 @@
+#include <library/cpp/html/escape/escape.h>
+#include <library/cpp/testing/unittest/registar.h>
+
+using namespace NHtml;
+
+Y_UNIT_TEST_SUITE(TEscapeHtml) {
+    Y_UNIT_TEST(Escape) {
+        UNIT_ASSERT_EQUAL(EscapeText("in & out"), "in &amp; out");
+        UNIT_ASSERT_EQUAL(EscapeText("&&"), "&amp;&amp;");
+        UNIT_ASSERT_EQUAL(EscapeText("&amp;"), "&amp;amp;");
+
+        UNIT_ASSERT_EQUAL(EscapeText("<script>"), "&lt;script&gt;");
+
+        UNIT_ASSERT_EQUAL(EscapeText("text"), "text");
+    }
+}
diff --git a/library/cpp/html/escape/ut/ya.make b/library/cpp/html/escape/ut/ya.make
new file mode 100644
index 0000000000..27d3a3d65a
--- /dev/null
+++ b/library/cpp/html/escape/ut/ya.make
@@ -0,0 +1,13 @@
+UNITTEST()
+
+OWNER(stanly)
+
+SRCS(
+    escape_ut.cpp
+)
+
+PEERDIR(
+    library/cpp/html/escape
+)
+
+END()
diff --git a/library/cpp/html/escape/ya.make b/library/cpp/html/escape/ya.make
new file mode 100644
index 0000000000..93a31e33d6
--- /dev/null
+++ b/library/cpp/html/escape/ya.make
@@ -0,0 +1,9 @@
+LIBRARY()
+
+OWNER(stanly)
+
+SRCS(
+    escape.cpp
+)
+
+END()
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()
diff --git a/library/cpp/html/ya.make b/library/cpp/html/ya.make
new file mode 100644
index 0000000000..ab904be3b9
--- /dev/null
+++ b/library/cpp/html/ya.make
@@ -0,0 +1,32 @@
+RECURSE(
+    dehtml
+    dehtml/ut
+    detect
+    blob
+    entity
+    entity/ut
+    escape
+    escape/ut
+    face
+    face/blob
+    html5
+    html5/tests
+    html5/ut
+    lexer
+    lexer/ut
+    pcdata
+    pcdata/ut
+    pdoc
+    print
+    relalternate
+    relalternate/ut
+    sanitize
+    spec
+    storage
+    storage/ut
+    strip
+    tree
+    url
+    zoneconf
+    zoneconf/ut
+)