[19_3] Move hex related routines to lolly::data

+ Rename as_hexadecimal to to_Hex
+ Add routine: to_hex
+ Fix INT32_MIN bug of to_hex

Kernel/Types/analyze.cpp

@@ -10,6 +10,7 @@
  ******************************************************************************/
 
 #include "analyze.hpp"
+#include "lolly/data/numeral.hpp"
 #include "ntuple.hpp"
 
 /******************************************************************************
@@ -520,53 +521,6 @@ fnsymbol_nr (int nr) {
   return r;
 }
 
-/******************************************************************************
- * Conversions to and from hexadecimal
- ******************************************************************************/
-
-static const char* hex_string= "0123456789ABCDEF";
-
-string
-as_hex (uint8_t i) {
-  uint8_t i_low = i & 15;
-  uint8_t i_high= i >> 4;
-  return locase_all (string (hex_string[i_high]) * string (hex_string[i_low]));
-}
-
-string
-as_hexadecimal (int i) {
-  if (i < 0) return "-" * as_hexadecimal (-i);
-  if (i < 16) return hex_string[i & 15];
-  return as_hexadecimal (i >> 4) * hex_string[i & 15];
-}
-
-string
-as_hexadecimal (pointer ptr) {
-  intptr_t i= (intptr_t) ptr;
-  if (i < 0) return "-" * as_hexadecimal (-i);
-  if (i < 16) return hex_string[i & 15];
-  return as_hexadecimal (i >> 4) * hex_string[i & 15];
-}
-
-string
-as_hexadecimal (int i, int len) {
-  if (len == 1) return hex_string[i & 15];
-  else return as_hexadecimal (i >> 4, len - 1) * hex_string[i & 15];
-}
-
-int
-from_hexadecimal (string s) {
-  int i, n= N (s), res= 0;
-  if ((n > 0) && (s[0] == '-')) return -from_hexadecimal (s (1, n));
-  for (i= 0; i < n; i++) {
-    res= res << 4;
-    if (is_digit (s[i])) res+= (int) (s[i] - '0');
-    if ((s[i] >= 'A') && (s[i] <= 'F')) res+= (int) (s[i] + 10 - 'A');
-    if ((s[i] >= 'a') && (s[i] <= 'f')) res+= (int) (s[i] + 10 - 'a');
-  }
-  return res;
-}
-
 /******************************************************************************
  * Routines for the TeXmacs encoding
  ******************************************************************************/
@@ -955,7 +909,7 @@ unescape_guile (string s) {
       else if (i + 3 < n && s[i + 1] == 'x' && is_hex_digit (s[i + 2]) &&
                is_hex_digit (s[i + 3])) {
         string e= s (i + 2, i + 4);
-        r << (unsigned char) from_hexadecimal (e);
+        r << (unsigned char) lolly::data::from_hex (e);
         i+= 3;
       }
       else r << s[i];

Kernel/Types/analyze.hpp

@@ -301,52 +301,6 @@ string Alpha_nr (int nr);
  */
 string fnsymbol_nr (int nr);
 
-/**
- * @brief Converts an 8-bit unsigned integer to a fixed-length (2) hex string.
- *
- * @param i The integer to be converted to a fixed-length hex string.
- * @return The fixed-length (2) hexadecimal string representation of the input
- * integer.
- */
-string as_hex (uint8_t i);
-
-/**
- * @brief Converts an integer to a hexadecimal string.
- *
- * @param i The integer to be converted to a hexadecimal string.
- * @return The hexadecimal string representation of the input integer.
- */
-string as_hexadecimal (int i);
-
-/**
- * @brief Converts a pointer to a hexadecimal string.
- *
- * @param ptr The pointer to be converted to a hexadecimal string.
- * @return The hexadecimal string representation of the input pointer.
- */
-string as_hexadecimal (pointer ptr);
-
-/**
- * @brief Converts an integer to a hexadecimal string with a fixed length.
- *
- * @param i The integer to be converted to a hexadecimal string.
- * @param len The length of the output hexadecimal string.
- * @return The fixed-length hexadecimal string representation of the input
- * integer.
- */
-string as_hexadecimal (int i, int length);
-
-/**
- * @brief Converts a hexadecimal string to an integer.
- *
- * This function takes a hexadecimal string as input and converts it into its
- * integer representation.
- *
- * @param s The hexadecimal string to be converted to an integer.
- * @return The integer representation of the input hexadecimal string.
- */
-int from_hexadecimal (string s);
-
 /**
  * @brief Encodes a string by converting special characters to TeXmacs encoding.
  *

lolly/data/numeral.cpp

@@ -15,13 +15,20 @@
 
 namespace lolly {
 namespace data {
-static string roman_ones[10]    = {"",  "i",  "ii",  "iii",  "iv",
-                                   "v", "vi", "vii", "viii", "ix"};
-static string roman_tens[10]    = {"",  "x",  "xx",  "xxx",  "xl",
-                                   "l", "lx", "lxx", "lxxx", "xc"};
-static string roman_hundreds[10]= {"",  "c",  "cc",  "ccc",  "cd",
-                                   "d", "dc", "dcc", "dccc", "cm"};
-static string roman_thousands[4]= {"", "m", "mm", "mmm"};
+static const string roman_ones[10]    = {"",  "i",  "ii",  "iii",  "iv",
+                                         "v", "vi", "vii", "viii", "ix"};
+static const string roman_tens[10]    = {"",  "x",  "xx",  "xxx",  "xl",
+                                         "l", "lx", "lxx", "lxxx", "xc"};
+static const string roman_hundreds[10]= {"",  "c",  "cc",  "ccc",  "cd",
+                                         "d", "dc", "dcc", "dccc", "cm"};
+static const string roman_thousands[4]= {"", "m", "mm", "mmm"};
+
+// 0 should not be used as index of this array, or bug occurs. because digit 0
+// is handled specially according to position of digit.
+static const string chars_han[10]= {"?",  "一", "二", "三", "四",
+                                    "五", "六", "七", "八", "九"};
+
+static const char* hex_string= "0123456789ABCDEF";
 
 string
 to_roman (int32_t nr) {
@@ -37,11 +44,6 @@ to_Roman (int32_t nr) {
   return upcase_all (to_roman (nr));
 }
 
-// 0 should not be used as index of this array, or bug occurs. because digit 0
-// is handled specially according to position of digit.
-static string chars_han[10]= {
-    "<unspecified>", "一", "二", "三", "四", "五", "六", "七", "八", "九"};
-
 string
 hanzi_sub (int16_t nr, bool leading_zero) {
   short thousand= (nr % 10000) / 1000, hundred= (nr % 1000) / 100,
@@ -120,7 +122,7 @@ hanzi_sub (int16_t nr, bool leading_zero) {
   case 0x1F:
     return "";
   default:
-    return "<unspecified>" * as_string (cases);
+    return "?" * as_string (cases);
   }
 }
 
@@ -140,5 +142,62 @@ to_hanzi (int32_t nr) {
   return hanzi_sub (nr, false);
 }
 
+string
+to_padded_Hex (uint8_t i) {
+  uint8_t i_low = i & 15;
+  uint8_t i_high= i >> 4;
+  return string (hex_string[i_high]) * string (hex_string[i_low]);
+}
+
+string
+to_padded_hex (uint8_t i) {
+  return locase_all (to_padded_Hex (i));
+}
+
+string
+to_Hex (int32_t i) {
+  if (i == INT32_MIN) return "-80000000";
+  if (i < 0) return "-" * to_Hex (-i);
+  if (i < 16) return hex_string[i & 15];
+  return to_Hex (i >> 4) * hex_string[i & 15];
+}
+
+string
+to_hex (int32_t i) {
+  return locase_all (to_Hex (i));
+}
+
+string
+to_Hex (pointer ptr) {
+  intptr_t i= (intptr_t) ptr;
+  if (i < 0) return "-" * to_Hex (-i);
+  if (i < 16) return hex_string[i & 15];
+  return to_Hex (i >> 4) * hex_string[i & 15];
+}
+
+string
+to_hex (pointer ptr) {
+  return locase_all (to_Hex (ptr));
+}
+
+int
+from_hex (string s) {
+  int i, n= N (s), res= 0;
+  if ((n > 0) && (s[0] == '-')) return -from_hex (s (1, n));
+  for (i= 0; i < n; i++) {
+    res= res << 4;
+    if (is_digit (s[i])) res+= (int) (s[i] - '0');
+    if ((s[i] >= 'A') && (s[i] <= 'F')) res+= (int) (s[i] + 10 - 'A');
+    if ((s[i] >= 'a') && (s[i] <= 'f')) res+= (int) (s[i] + 10 - 'a');
+  }
+  return res;
+}
+
+string
+as_hexadecimal (int i, int len) {
+  if (len == 1) return hex_string[i & 15];
+  else return as_hexadecimal (i >> 4, len - 1) * hex_string[i & 15];
+}
+
 } // namespace data
 } // namespace lolly

lolly/data/numeral.hpp

@@ -39,5 +39,56 @@ string to_Roman (int32_t nr);
  * @return A string representing the Chinese numeral.
  */
 string to_hanzi (int32_t nr);
+
+/**
+ * @brief Converts an 8-bit unsigned integer to a fixed-length (2) hex string.
+ *
+ * @param i The integer to be converted to a fixed-length hex string.
+ * @return The fixed-length (2) hexadecimal string representation of the input
+ * integer.
+ */
+string to_padded_hex (uint8_t i);
+
+/**
+ * @brief Converts an integer to a hexadecimal string.
+ *
+ * @param i The integer to be converted to a hexadecimal string.
+ * @return The hexadecimal string representation of the input integer.
+ */
+string to_hex (int i);
+
+string to_Hex (int i);
+
+/**
+ * @brief Converts a pointer to a hexadecimal string.
+ *
+ * @param ptr The pointer to be converted to a hexadecimal string.
+ * @return The hexadecimal string representation of the input pointer.
+ */
+string to_hex (pointer ptr);
+
+string to_Hex (pointer ptr);
+
+/**
+ * @brief Converts a hexadecimal string to an integer.
+ *
+ * This function takes a hexadecimal string as input and converts it into its
+ * integer representation.
+ *
+ * @param s The hexadecimal string to be converted to an integer.
+ * @return The integer representation of the input hexadecimal string.
+ */
+int from_hex (string s);
+
+/**
+ * @brief Converts an integer to a hexadecimal string with a fixed length.
+ *
+ * @param i The integer to be converted to a hexadecimal string.
+ * @param len The length of the output hexadecimal string.
+ * @return The fixed-length hexadecimal string representation of the input
+ * integer.
+ */
+string as_hexadecimal (int i, int length);
+
 } // namespace data
 } // namespace lolly

lolly/hash/md5.cpp

@@ -12,6 +12,7 @@
 #include "md5.hpp"
 #include "analyze.hpp"
 #include "file.hpp"
+#include "lolly/data/numeral.hpp"
 
 #include <tbox/tbox.h>
 
@@ -49,7 +50,7 @@ md5_hexdigest (url u) {
 
     string md5_hex= string ();
     for (int i= 0; i < 16; ++i) {
-      md5_hex << as_hex (o_buffer[i]);
+      md5_hex << data::to_padded_hex (o_buffer[i]);
     }
     return md5_hex;
   }

lolly/hash/sha.cpp

@@ -14,6 +14,7 @@
 #include "sha.hpp"
 #include "analyze.hpp"
 #include "file.hpp"
+#include "lolly/data/numeral.hpp"
 #include "string.hpp"
 #include "url.hpp"
 
@@ -74,7 +75,7 @@ sha_hexdigest (url u, sha_mode mode) {
 
     string ret= string ();
     for (int i= 0; i < o_size; ++i) {
-      ret << as_hex (o_buffer[i]);
+      ret << data::to_padded_hex (o_buffer[i]);
     }
     return ret;
   }

tests/Kernel/Types/analyze_test.cpp

@@ -63,18 +63,6 @@ TEST_CASE ("is_alphanum") {
   CHECK (!is_alphanum ("!"));
 }
 
-TEST_CASE ("as_hex") {
-  SUBCASE ("0~255") {
-    string_eq (as_hex ((uint8_t) 0), "00");
-    string_eq (as_hex ((uint8_t) 1), "01");
-    string_eq (as_hex ((uint8_t) 255), "ff");
-  }
-  SUBCASE ("overflow") {
-    string_eq (as_hex ((uint8_t) -1), "ff");
-    string_eq (as_hex ((uint8_t) 256), "00");
-  }
-}
-
 TEST_CASE ("test locase all") {
   CHECK_EQ (locase_all (string ("true")) == string ("true"), true);
   CHECK_EQ (locase_all (string ("TRue")) == string ("true"), true);