Rework Text stamps. Now we have String stamps that produce Pulp-based string in desired encoding, and StampText<T> template that joins these strings with space separator

nataraj-hates-MS-for-stealing-github · nataraj-hates-MS-for-stealing-github · commit 712cdf62a4e9 · 2025-06-10T19:30:45.000+03:00
diff --git a/blobstamper/stamp_text.cpp b/blobstamper/stamp_text.cpp
@@ -19,52 +19,38 @@
 #include "stamp_text.h"
 
 std::string
-StampTextPulp::ExtractStr(std::shared_ptr<Blob> blob)
+StampStringLatin1::ExtractStr(std::shared_ptr<Blob> blob)
 {
     std::vector<char> data = blob->Chop(minSize(), maxSize())->AsByteVector();
 
-    std::vector<char>::iterator the_iterator;
+    std::vector<char>::iterator iterator;
 
-    the_iterator = data.begin();
+    iterator = data.begin();
     std::string res;
-    while (the_iterator != data.end())
+    while (iterator != data.end())
     {
-      if (*the_iterator == '\0') { *the_iterator = ' '; }
-      res.push_back(*the_iterator++);
+      if (*iterator == '\0') { *iterator = ' '; }
+      res.push_back(*iterator++);
     }
 
     return res;
 }
 
-std::string StampTextPulpWords::ExtractStr(std::shared_ptr<Blob> blob)
+std::string
+StampStringASCII::ExtractStr(std::shared_ptr<Blob> blob)
 {
-  std::vector<std::string> data = ExtractStrVector(blob);
-  std::string res = "";
-
-  for(std::string s : data)
-  {
-    if (!res.empty())
-    {
-      res+=" ";
-    }
-    res+= s;
-  }
-  return res;
-}
+    std::vector<char> data = blob->Chop(minSize(), maxSize())->AsByteVector();
 
-std::string StampTextDictWords::ExtractStr(std::shared_ptr<Blob> blob)
-{
-  std::vector<std::string> data = ExtractStrVector(blob);
-  std::string res = "";
+    std::vector<char>::iterator iterator;
 
-  for(std::string s : data)
-  {
-    if (!res.empty())
+    iterator = data.begin();
+    std::string res;
+    while (iterator != data.end())
     {
-      res+=" ";
+      *iterator &= 0x7F; // Remove first bit. This will made it ASCII (or more precisely 7bit encoding including character 0x01..0x1F)
+      if (*iterator == '\0') { *iterator = ' '; }
+      res.push_back(*iterator++);
     }
-    res+= s;
-  }
-  return res;
+    return res;
 }
 
diff --git a/blobstamper/stamp_text.h b/blobstamper/stamp_text.h
@@ -24,24 +24,48 @@
 
 #include "stamp_dict.h"
 
-class StampTextPulp: public StampBaseStr
+class StampStringLatin1: public StampBaseStr
 {
   public:
     virtual int minSize() override { return  1; }
     virtual int maxSize() override { return -1; }
     std::string ExtractStr(std::shared_ptr<Blob> blob) override;
 };
 
-class StampTextPulpWords:  public GalleyVectorStrStampBase<StampTextPulp>
+class StampStringASCII: public StampBaseStr
 {
   public:
-    virtual std::string ExtractStr(std::shared_ptr<Blob> blob) override;
+    virtual int minSize() override { return  1; }
+    virtual int maxSize() override { return -1; }
+    std::string ExtractStr(std::shared_ptr<Blob> blob) override;
 };
 
-class StampTextDictWords:  public GalleyVectorStrStampBase<StampDictLCAlphaSmall>
+template<class T> class StampText: public GalleyVectorStrStampBase<T>
 {
   public:
     virtual std::string ExtractStr(std::shared_ptr<Blob> blob) override;
 };
 
+
+/* StampText create a string that consist of word that are separated by spaces.
+ * Each word is created by T stamp, (T is template parameter, you can set any
+ * stamp type there */
+
+template<class T>
+std::string StampText<T>::ExtractStr(std::shared_ptr<Blob> blob)
+{
+  std::vector<std::string> data = this->ExtractStrVector(blob);
+  std::string res = "";
+
+  for(std::string s : data)
+  {
+    if (!res.empty())
+    {
+      res+=" ";
+    }
+    res+= s;
+  }
+  return res;
+}
+
 #endif /* STAMP_TEXT_H */
diff --git a/t/150-stamp_text.cpp b/t/150-stamp_text.cpp
@@ -31,30 +31,48 @@ using namespace TAP;
 int
 main()
 {
-  TEST_START(3);
+  TEST_START(5);
   { /* 1..1 */
     char data[] = "папа\0мама\0бабушка\0дедушка\0братик\0сестричка";
     std::shared_ptr<Blob> blob = std::make_shared<Blob>(data, (sizeof data)-1);
-    StampTextPulp stamp;
-    std::string s = stamp.ExtractStr(blob);
-    is(s, "папа мама бабушка дедушка братик сестричка", "StampTextSimple");
+    StampStringLatin1 stamp_str_l1;
+    std::string s = stamp_str_l1.ExtractStr(blob);
+    is(s, "папа мама бабушка дедушка братик сестричка", "StampStringLatin1");
   }
 
   { /* 2..2 */
+    char data[] = "папа\0мама\0бабушка\0дедушка\0братик\0сестричка";
+    std::shared_ptr<Blob> blob = std::make_shared<Blob>(data, (sizeof data)-1);
+    StampStringASCII stamp_str_ascii;
+    std::string s = stamp_str_ascii.ExtractStr(blob);
+    is(s,
+       "P?P0P?P0 P<P0P<P0 P1P0P1Q\x3Q\x8P:P0 P4P5P4Q\x3Q\x8P:P0 P1Q P0Q\x2P8P: Q\x1P5Q\x1Q\x2Q P8Q\x7P:P0",
+        "StampTextASCII");
+  }
+
+  { /* 3..3 */
     char data[] = "dad\0mam\0granddad\0grandmam\0brother\0sister";
     std::shared_ptr<Blob> blob = std::make_shared<Blob>(data, (sizeof data)-1);
-    StampTextPulpWords stamp;
+    StampText<StampStringLatin1> stamp;
     std::string s = stamp.ExtractStr(blob);
-    is(s, "d dad gra n dmam broth er siste", "GalleyTextSimple");
+    is(s, "d dad gra n dmam broth er siste", "StampText<StampStringLatin1>");
   }
 
-  { /* 3..3 */
+  { /* 4..4 */
     char data[] = "abcdef" "abcdef" "ABCDEF" "012345";
     std::shared_ptr<Blob> blob = std::make_shared<Blob>(data, (sizeof data)-1);
-    StampTextDictWords stamp;
+    StampText<StampDictLCAlphaSmall> stamp;
     std::string s = stamp.ExtractStr(blob);
-    is(s, "gleam godfather graffiti greened grouping gunshots gleam godfather graffiti greened grouping gunshots dismally dissented divested doorstep dread drunks convertors corpulent counterparts cranking crippled crusades", "GalleyLCAlphaSmall");
+    is(s, "gleam godfather graffiti greened grouping gunshots gleam godfather graffiti greened grouping gunshots dismally dissented divested doorstep dread drunks convertors corpulent counterparts cranking crippled crusades", "StampText<StampDictLCAlphaSmall>");
+  }
 
+  { /* 5..5 */
+    char data[] = "Некоторый текст написанный русскими буквами в кодировке utf-8";
+    std::shared_ptr<Blob> blob = std::make_shared<Blob>(data, (sizeof data)-1);
+    StampText<StampStringASCII> stamp;
+    std::string s = stamp.ExtractStr(blob);
+    is(s, "P9  Q  Q\x3Q \x1Q \x1P: P8 P<P 8 P 1Q\x3 P:P 2P0 P<P 8 P 2  P: P>P 4P8 Q  P>P 2P :P5  ut f-8", "StampText<StampStringASCII>");
   }
+
   TEST_END;
 }
