charset: Refactor vim_str2nr

src/nvim/charset.c

@@ -1611,8 +1611,9 @@ bool vim_isblankline(char_u *lbuf)
 /// If maxlen > 0, check at a maximum maxlen chars.
 ///
 /// @param start
-/// @param prep Returns type of number 0 = decimal, 'x' or 'X' is hex,
+/// @param prep Returns guessed type of number 0 = decimal, 'x' or 'X' is
-///        '0' = octal, 'b' or 'B' is bin
+///             hexadecimal, '0' = octal, 'b' or 'B' is binary. When using
+///             STR2NR_FORCE is always zero.
 /// @param len Returns the detected length of number.
 /// @param what Recognizes what number passed.
 /// @param nptr Returns the signed result.
@@ -1627,55 +1628,84 @@ void vim_str2nr(const char_u *const start, int *const prep, int *const len,
 #define STRING_ENDED(ptr) \
     (!(maxlen == 0 || (int)((ptr) - (const char *)start) < maxlen))
   int pre = 0;  // default is decimal
-  bool negative = false;
+  const bool negative = (ptr[0] == '-');
+  uvarnumber_T un = 0;
 
-  if (ptr[0] == '-') {
+  if (negative) {
-    negative = true;
     ptr++;
   }
 
-  // Recognize hex, octal and bin.
+  if (what & STR2NR_FORCE) {
-  if ((what & (STR2NR_HEX|STR2NR_OCT|STR2NR_BIN))
+    // When forcing main consideration is skipping the prefix. Octal and decimal
-      && !STRING_ENDED(ptr + 1)
+    // numbers have no prefixes to skip. pre is not set.
-      && ptr[0] == '0' && ptr[1] != '8' && ptr[1] != '9') {
+    switch ((unsigned)what & (~(unsigned)STR2NR_FORCE)) {
+      case STR2NR_HEX: {
+        if (!STRING_ENDED(ptr + 2)
+            && ptr[0] == '0'
+            && (ptr[1] == 'x' || ptr[1] == 'X')
+            && ascii_isxdigit(ptr[2])) {
+          ptr += 2;
+        }
+        goto vim_str2nr_hex;
+      }
+      case STR2NR_BIN: {
+        if (!STRING_ENDED(ptr + 2)
+            && ptr[0] == '0'
+            && (ptr[1] == 'b' || ptr[1] == 'B')
+            && ascii_isbdigit(ptr[2])) {
+          ptr += 2;
+        }
+        goto vim_str2nr_bin;
+      }
+      case STR2NR_OCT: {
+        goto vim_str2nr_oct;
+      }
+      case 0: {
+        goto vim_str2nr_dec;
+      }
+      default: {
+        assert(false);
+      }
+    }
+  } else if ((what & (STR2NR_HEX|STR2NR_OCT|STR2NR_BIN))
+             && !STRING_ENDED(ptr + 1)
+             && ptr[0] == '0' && ptr[1] != '8' && ptr[1] != '9') {
     pre = ptr[1];
-
+    // Detect hexadecimal: 0x or 0X follwed by hex digit
     if ((what & STR2NR_HEX)
         && !STRING_ENDED(ptr + 2)
         && (pre == 'X' || pre == 'x')
         && ascii_isxdigit(ptr[2])) {
-      // hexadecimal
       ptr += 2;
-    } else if ((what & STR2NR_BIN)
+      goto vim_str2nr_hex;
-               && !STRING_ENDED(ptr + 2)
+    }
-               && (pre == 'B' || pre == 'b')
+    // Detect binary: 0b or 0B follwed by 0 or 1
-               && ascii_isbdigit(ptr[2])) {
+    if ((what & STR2NR_BIN)
-      // binary
+        && !STRING_ENDED(ptr + 2)
+        && (pre == 'B' || pre == 'b')
+        && ascii_isbdigit(ptr[2])) {
       ptr += 2;
-    } else {
+      goto vim_str2nr_bin;
-      // decimal or octal, default is decimal
+    }
-      pre = 0;
+    // Detect octal number: zero followed by octal digits without '8' or '9'
-
+    pre = 0;
-      if (what & STR2NR_OCT
+    if (!(what & STR2NR_OCT)
-          && !STRING_ENDED(ptr + 1)
+        || !('0' <= ptr[1] && ptr[1] <= '7')) {
-          && ('0' <= ptr[1] && ptr[1] <= '7')) {
+      goto vim_str2nr_dec;
-        // Assume octal now: what we already know is that string starts with
+    }
-        // zero and some octal digit.
+    for (int i = 2; !STRING_ENDED(ptr + i) && ascii_isdigit(ptr[i]); i++) {
-        pre = '0';
+      if (ptr[i] > '7') {
-        // Don’t interpret "0", "008" or "0129" as octal.
+        goto vim_str2nr_dec;
-        for (int i = 2; !STRING_ENDED(ptr + i) && ascii_isdigit(ptr[i]); i++) {
-          if (ptr[i] > '7') {
-            // Can’t be octal.
-            pre = 0;
-            break;
-          }
-        }
       }
     }
+    pre = '0';
+    goto vim_str2nr_oct;
+  } else {
+    goto vim_str2nr_dec;
   }
 
   // Do the string-to-numeric conversion "manually" to avoid sscanf quirks.
-  uvarnumber_T un = 0;
+  assert(false);  // Should’ve used goto earlier.
 #define PARSE_NUMBER(base, cond, conv) \
   do { \
     while (!STRING_ENDED(ptr) && (cond)) { \
@@ -1688,18 +1718,29 @@ void vim_str2nr(const char_u *const start, int *const prep, int *const len,
       ptr++; \
     } \
   } while (0)
-  if (pre == 'B' || pre == 'b' || what == (STR2NR_BIN|STR2NR_FORCE)) {
+  switch (pre) {
-    // Binary number.
+    case 'b':
-    PARSE_NUMBER(2, (*ptr == '0' || *ptr == '1'), (*ptr - '0'));
+    case 'B': {
-  } else if (pre == '0' || what == (STR2NR_OCT|STR2NR_FORCE)) {
+vim_str2nr_bin:
-    // Octal number.
+      PARSE_NUMBER(2, (*ptr == '0' || *ptr == '1'), (*ptr - '0'));
-    PARSE_NUMBER(8, ('0' <= *ptr && *ptr <= '7'), (*ptr - '0'));
+      break;
-  } else if (pre == 'X' || pre == 'x' || what == (STR2NR_HEX|STR2NR_FORCE)) {
+    }
-    // Hexadecimal number.
+    case '0': {
-    PARSE_NUMBER(16, (ascii_isxdigit(*ptr)), (hex2nr(*ptr)));
+vim_str2nr_oct:
-  } else {
+      PARSE_NUMBER(8, ('0' <= *ptr && *ptr <= '7'), (*ptr - '0'));
-    // Decimal number.
+      break;
-    PARSE_NUMBER(10, (ascii_isdigit(*ptr)), (*ptr - '0'));
+    }
+    case 0: {
+vim_str2nr_dec:
+      PARSE_NUMBER(10, (ascii_isdigit(*ptr)), (*ptr - '0'));
+      break;
+    }
+    case 'x':
+    case 'X': {
+vim_str2nr_hex:
+      PARSE_NUMBER(16, (ascii_isxdigit(*ptr)), (hex2nr(*ptr)));
+      break;
+    }
   }
 #undef PARSE_NUMBER
 

test/symbolic/klee/nvim/charset.c

@@ -31,55 +31,83 @@ void vim_str2nr(const char_u *const start, int *const prep, int *const len,
 #define STRING_ENDED(ptr) \
     (!(maxlen == 0 || (int)((ptr) - (const char *)start) < maxlen))
   int pre = 0;  // default is decimal
-  bool negative = false;
+  const bool negative = (ptr[0] == '-');
+  uvarnumber_T un = 0;
 
-  if (ptr[0] == '-') {
+  if (negative) {
-    negative = true;
     ptr++;
   }
 
-  // Recognize hex, octal and bin.
+  if (what & STR2NR_FORCE) {
-  if ((what & (STR2NR_HEX|STR2NR_OCT|STR2NR_BIN))
+    // When forcing main consideration is skipping the prefix. Octal and decimal
-      && !STRING_ENDED(ptr + 1)
+    // numbers have no prefixes to skip. pre is not set.
-      && ptr[0] == '0' && ptr[1] != '8' && ptr[1] != '9') {
+    switch ((unsigned)what & (~(unsigned)STR2NR_FORCE)) {
+      case STR2NR_HEX: {
+        if (!STRING_ENDED(ptr + 2)
+            && ptr[0] == '0'
+            && (ptr[1] == 'x' || ptr[1] == 'X')
+            && ascii_isxdigit(ptr[2])) {
+          ptr += 2;
+        }
+        goto vim_str2nr_hex;
+      }
+      case STR2NR_BIN: {
+        if (!STRING_ENDED(ptr + 2)
+            && ptr[0] == '0'
+            && (ptr[1] == 'b' || ptr[1] == 'B')
+            && ascii_isbdigit(ptr[2])) {
+          ptr += 2;
+        }
+        goto vim_str2nr_bin;
+      }
+      case STR2NR_OCT: {
+        goto vim_str2nr_oct;
+      }
+      case 0: {
+        goto vim_str2nr_dec;
+      }
+      default: {
+        assert(false);
+      }
+    }
+  } else if ((what & (STR2NR_HEX|STR2NR_OCT|STR2NR_BIN))
+             && !STRING_ENDED(ptr + 1)
+             && ptr[0] == '0' && ptr[1] != '8' && ptr[1] != '9') {
     pre = ptr[1];
-
+    // Detect hexadecimal: 0x or 0X follwed by hex digit
     if ((what & STR2NR_HEX)
         && !STRING_ENDED(ptr + 2)
         && (pre == 'X' || pre == 'x')
         && ascii_isxdigit(ptr[2])) {
-      // hexadecimal
       ptr += 2;
-    } else if ((what & STR2NR_BIN)
+      goto vim_str2nr_hex;
-               && !STRING_ENDED(ptr + 2)
+    }
-               && (pre == 'B' || pre == 'b')
+    // Detect binary: 0b or 0B follwed by 0 or 1
-               && ascii_isbdigit(ptr[2])) {
+    if ((what & STR2NR_BIN)
-      // binary
+        && !STRING_ENDED(ptr + 2)
+        && (pre == 'B' || pre == 'b')
+        && ascii_isbdigit(ptr[2])) {
       ptr += 2;
-    } else {
+      goto vim_str2nr_bin;
-      // decimal or octal, default is decimal
+    }
-      pre = 0;
+    // Detect octal number: zero followed by octal digits without '8' or '9'
-
+    pre = 0;
-      if (what & STR2NR_OCT
+    if (!(what & STR2NR_OCT)) {
-          && !STRING_ENDED(ptr + 1)
+      goto vim_str2nr_dec;
-          && ('0' <= ptr[1] && ptr[1] <= '7')) {
+    }
-        // Assume octal now: what we already know is that string starts with
+    for (int i = 2; !STRING_ENDED(ptr + i) && ascii_isdigit(ptr[i]); i++) {
-        // zero and some octal digit.
+      if (ptr[i] > '7') {
-        pre = '0';
+        goto vim_str2nr_dec;
-        // Don’t interpret "0", "008" or "0129" as octal.
-        for (int i = 2; !STRING_ENDED(ptr + i) && ascii_isdigit(ptr[i]); i++) {
-          if (ptr[i] > '7') {
-            // Can’t be octal.
-            pre = 0;
-            break;
-          }
-        }
       }
     }
+    pre = '0';
+    goto vim_str2nr_oct;
+  } else {
+    goto vim_str2nr_dec;
   }
 
   // Do the string-to-numeric conversion "manually" to avoid sscanf quirks.
-  uvarnumber_T un = 0;
+  assert(false);  // Should’ve used goto earlier.
 #define PARSE_NUMBER(base, cond, conv) \
   do { \
     while (!STRING_ENDED(ptr) && (cond)) { \
@@ -92,18 +120,29 @@ void vim_str2nr(const char_u *const start, int *const prep, int *const len,
       ptr++; \
     } \
   } while (0)
-  if (pre == 'B' || pre == 'b' || what == (STR2NR_BIN|STR2NR_FORCE)) {
+  switch (pre) {
-    // Binary number.
+    case 'b':
-    PARSE_NUMBER(2, (*ptr == '0' || *ptr == '1'), (*ptr - '0'));
+    case 'B': {
-  } else if (pre == '0' || what == (STR2NR_OCT|STR2NR_FORCE)) {
+vim_str2nr_bin:
-    // Octal number.
+      PARSE_NUMBER(2, (*ptr == '0' || *ptr == '1'), (*ptr - '0'));
-    PARSE_NUMBER(8, ('0' <= *ptr && *ptr <= '7'), (*ptr - '0'));
+      break;
-  } else if (pre == 'X' || pre == 'x' || what == (STR2NR_HEX|STR2NR_FORCE)) {
+    }
-    // Hexadecimal number.
+    case '0': {
-    PARSE_NUMBER(16, (ascii_isxdigit(*ptr)), (hex2nr(*ptr)));
+vim_str2nr_oct:
-  } else {
+      PARSE_NUMBER(8, ('0' <= *ptr && *ptr <= '7'), (*ptr - '0'));
-    // Decimal number.
+      break;
-    PARSE_NUMBER(10, (ascii_isdigit(*ptr)), (*ptr - '0'));
+    }
+    case 0: {
+vim_str2nr_dec:
+      PARSE_NUMBER(10, (ascii_isdigit(*ptr)), (*ptr - '0'));
+      break;
+    }
+    case 'x':
+    case 'X': {
+vim_str2nr_hex:
+      PARSE_NUMBER(16, (ascii_isxdigit(*ptr)), (hex2nr(*ptr)));
+      break;
+    }
   }
 #undef PARSE_NUMBER
 

test/unit/charset/vim_str2nr_spec.lua

@@ -0,0 +1 @@
+-- FIXME