Complete implementation of strto* and ato* functions

code/errno.c

@@ -0,0 +1,4 @@
+
+#include <errno.h>
+
+int errno;

code/stdlib/conv.c

@@ -9,6 +9,14 @@
 // TODO: i made a function that parses longs and i'm fucken
 // tired. someone make float parsing :kekw:
 
+// Call me weak if you want but I'm actually too lazy to type
+// them out every time, also they take up a lot of horiz space.
+typedef long long int          intll;
+typedef long int               intl;
+typedef unsigned long long int intull;
+typedef unsigned long int      intul;
+typedef unsigned int           intu;
+
 #define inrange(start, c, end) ((start) <= (c) && (c) <= (end))
 
 static bool isbase(int c, int base) {
@@ -43,92 +51,144 @@ static long todigitl(int c) {
     return val;
 }
 
-long
+static intull strto_generic(
-strtol(const char *restrict nptr, char **restrict endptr, int inbase) {
+    const char *restrict nptr,
-    if(!inrange(0, inbase, 36)) {
+    char **restrict endptr,
-        *endptr = NULL;
+    int inbase,
-        return 0;
+    intl *coefptr,
+    intull int_max
+) {
+    const char *restrict str = nptr;
+    intull value = 0;
+    int digits_read = 0;
+    bool is_signed = (coefptr != NULL);
+    // Find max{abs(int)}. Signed integers have negative,
+    // whose absolute value is 1 bigger than int_max.
+    intull int_abs_max = int_max;
+    if(is_signed) {
+        ++int_abs_max;
     }
+    if(!inrange(0, inbase, 36)) {
+        goto finish;
+    }
+    intull base = (intull)inbase;
     // Skip space on the beginning
-    while(isspace(*nptr)) {
+    while(isspace(*str)) {
-        ++nptr;
+        ++str;
     }
     // Parse sign
-    long coef = 1;
+    intl coef = 1;
-    if(*nptr == '-') {
+    if(is_signed) {
-        coef = -1;
+        if(*str == '-') {
-        ++nptr;
+            coef = -1;
+            ++str;
+        }
     }
-    if(*nptr == '+') {
+    if(*str == '+') {
-        ++nptr;
+        ++str;
     }
-    unsigned long base = (unsigned long)inbase;
-    unsigned long value = 0;
     // See if we need to parse base in C-like format
-    if(*nptr == '0' && *(nptr+1) == 'x') {
+    // then set the base accordingly
-        ++nptr;
+    if(*str == '0' && (*(str+1) == 'x' || *(str+1) == 'X')) {
+        ++str;
         if(base == 16 || base == 0) {
-            ++nptr;
+            ++str;
             base = 16;
         }
         else {
-            value = 0;
+            goto finish;
-            goto end;
         }
     }
-    else if(*nptr == '0') {
+    else if(*str == '0') {
-        ++nptr;
+        ++str;
+        ++digits_read;
         if(base == 8 || base == 0) {
             base = 8;
         }
     }
-    while(isbase(*nptr, (int)base)) {
+    // Parse the string of digits in the given base. If the value
-        unsigned long digit = (unsigned long)todigitl(*nptr);
+    // exceeds abs(int_min) we exit with range error.
-        if(value > (ULONG_MAX - digit)/base) {
+    while(isbase(*str, (int)base)) {
-            errno = ERANGE;
+        intull digit = (intull)todigitl(*str);
-            value = 0;
+        if(value > (int_abs_max - digit)/base) {
-            goto end;
+            goto error_out_of_range;
         }
         value = base*value + digit;
-        ++nptr;
+        ++str;
+        ++digits_read;
     }
-    unsigned long max_modulo = (unsigned long)LONG_MAX+1;
+    // We only allow the modulo of value equal to abs(int_min) if it is
-    if(value > max_modulo) {
+    // preceeded by the minus sign.
-        errno = ERANGE;
+    if(is_signed) {
+        if(value == int_abs_max && coef != -1) {
+            goto error_out_of_range;
+        }
+    }
+    goto finish;
+error_out_of_range:
+    // Skip the remainder of the subject string
+    while(isbase(*str, (int)base)) {
+        ++str;
+    }
+    errno = ERANGE;
+    value = int_max;
+    goto finish;
+finish:
+    // If no conversion is performed we return the value of 0 and *endptr
+    // is set to the nptr.
+    bool conv_performed = (digits_read > 0);
+    if(!conv_performed) {
         value = 0;
-        goto end;
     }
-    if(value == max_modulo) {
+    if(endptr != NULL) {
-        if(coef == 1) {
+        if(!conv_performed) {
-            errno = ERANGE;
+            *endptr = (char *)nptr;
-            value = 0;
-            goto end;
         }
         else {
-            value = LONG_MIN;
+            *endptr = (char *)str;
-            coef = 1;
         }
     }
-end:
+    *coefptr = coef;
-    if(endptr != NULL) {
+    return value;
-        *endptr = (char *)nptr;
+}
+
+intl strtol(const char *restrict nptr, char **restrict endptr, int base) {
+    intull int_max = (intull)LONG_MAX;
+    intl coef;
+    intull modulo = strto_generic(nptr, endptr, base, &coef, int_max);
+    intl value;
+    if(modulo == int_max) {
+        value = LONG_MIN;
     }
-    return coef*(long)value;
+    else {
+        value = coef * (intl)modulo;
+    }
+    return value;
 }
 
-long long
+intll strtoll(const char *restrict nptr, char **restrict endptr, int base) {
-strtoll(const char *restrict nptr, char **restrict endptr, int base) {
+    intull int_max = (intull)LLONG_MAX;
-    return 0;
+    intl coef;
+    intull modulo = strto_generic(nptr, endptr, base, &coef, int_max);
+    intll value;
+    if(modulo == int_max) {
+        value = LLONG_MIN;
+    }
+    else {
+        value = (intll)coef * (intll)modulo;
+    }
+    return value;
 }
 
-unsigned long
+intul strtoul(const char *restrict nptr, char **restrict endptr, int base) {
-strtoul(const char *restrict nptr, char **restrict endptr, int base) {
+    intull int_max = (intull)ULONG_MAX;
-    return 0;
+    intull value = strto_generic(nptr, endptr, base, NULL, int_max);
+    return (intul)value;
 }
 
-unsigned long long
+intull strtoull(const char *restrict nptr, char **restrict endptr, int base) {
-strtoull(const char *restrict nptr, char **restrict endptr, int base) {
+    intull int_max = (intull)ULLONG_MAX;
-    return 0;
+    return strto_generic(nptr, endptr, base, NULL, int_max);
 }
 
 double atof(const char *nptr) {

inc/errno.h

@@ -5,4 +5,4 @@
 #define ERANGE 3
 
 // TODO: make it thread-local
-int errno;
+extern int errno;

test/test.c

@@ -1,13 +1,44 @@
 #include <assert.h>
 #include <ctype.h>
+#include <stdlib.h>
+#include <stddef.h>
+#include <stdbool.h>
+#include <errno.h>
 
 int main(int argc, char** argv) {
-    /*for (int i = 0; i < argv; i++) {
+    const char *p = "10 200000000000000000000000000000 30 -40 junk";
-        printf("[%d] = %s\n", argv[i]);
+    // printf("Parsing '%s':\n", p);
-    }*/
+ 
-
+    for (;;)
-    for (char c = 'a'; c != 'z'; ++c) {
+    {
-        assert(isupper(toupper(c)));
+        // errno can be set to any non-zero value by a library function call
+        // regardless of whether there was an error, so it needs to be cleared
+        // in order to check the error set by strtol
+        errno = 0;
+        char *end;
+        const long i = strtol(p, &end, 10);
+        if (p == end)
+            break;
+ 
+        const bool range_error = errno == ERANGE;
+        // printf("Extracted '%.*s', strtol returned %ld.", (int)(end-p), p, i);
+        p = end;
+ 
+        // if (range_error)
+        //     printf(" Range error occurred.");
+ 
+        // putchar('\n');
     }
+ 
+    // printf("Unextracted leftover: '%s'\n\n", p);
+ 
+    // parsing without error handling
+    long v1 = strtol("1010", NULL, 2);
+    long v2 = strtol("12",   NULL, 8);
+    long v3 = strtol("A",    NULL, 16);
+    long v4 = strtol("junk", NULL, 36);
+    long v5 = strtol("012",  NULL, 0);
+    long v6 = strtol("0xA",  NULL, 0);
+    long v7 = strtol("junk", NULL, 0);
     return 0;
 }