[CDRIVER-5983] Refactor String Handling Around URI Parsing

build/sphinx/mongoc_common.py

@@ -8,6 +8,7 @@
 
 from sphinx.application import Sphinx
 from sphinx.application import logger as sphinx_log
+
 try:
     from sphinx.builders.dirhtml import DirectoryHTMLBuilder
 except ImportError:
@@ -16,7 +17,8 @@
 from sphinx.config import Config
 from docutils.parsers.rst import Directive
 
-needs_sphinx = "1.7" # Do not require newer sphinx. EPEL packages build man pages with Sphinx 1.7.6. Refer: CDRIVER-4767
+# Do not require newer sphinx. EPEL packages build man pages with Sphinx 1.7.6. Refer: CDRIVER-4767
+needs_sphinx = "1.7"
 author = "MongoDB, Inc"
 
 # -- Options for HTML output ----------------------------------------------
@@ -38,7 +40,8 @@ def _file_man_page_name(fpath: Path) -> Union[str, None]:
             continue
         return mat[1]
 
-def _collect_man (app: Sphinx):
+
+def _collect_man(app: Sphinx):
     # Note: 'app' is partially-formed, as this is called from the Sphinx.__init__
     docdir = Path(app.srcdir)
     # Find everything:
@@ -61,6 +64,7 @@ def _collect_man (app: Sphinx):
         assert docname, filepath
         man_pages.append((docname, man_name, "", [author], 3))
 
+
 # -- Options for manual page output ---------------------------------------
 
 # NOTE: This starts empty, but we populate it in `setup` in _collect_man() (see above)
@@ -168,6 +172,7 @@ def generate_html_redirs(app: Sphinx, page: str, templatename: str, context: Dic
     builder.css_files[:] = prev_css
     sphinx_log.debug("Wrote redirect: %r -> %r", path, page)
 
+
 def mongoc_common_setup(app: Sphinx):
     _collect_man(app)
     app.connect("html-page-context", generate_html_redirs)

src/common/src/mlib/intencode.h

@@ -21,8 +21,11 @@
 
 #include <mlib/config.h>
 #include <mlib/loop.h>
+#include <mlib/str.h>
 
+#include <errno.h>
 #include <stdint.h>
+#include <stdlib.h>
 #include <string.h>
 
 /**
@@ -165,3 +168,183 @@ mlib_write_f64le (void *out, double d)
    memcpy (&bits, &d, sizeof d);
    return mlib_write_u64le (out, bits);
 }
+
+/**
+ * @brief Decode a 64-bit natural number
+ *
+ * @param in The input string to be decoded. Does not support a sign or base prefix!
+ * @param base The base to be decoded. Must not be zero!
+ * @param out Pointer that receives the decoded value
+ * @return int A result code for the operation.
+ *
+ * See `mlib_i64_parse` for more details.
+ */
+static inline int
+mlib_nat64_parse (mstr_view in, unsigned base, uint64_t *out)
+{
+   if (in.len == 0) {
+      // Empty string is not valid
+      return EINVAL;
+   }
+
+   // Accummulate into this value:
+   uint64_t value = 0;
+   // Whether any operation in the parse overflowed the integer value
+   bool did_overflow = false;
+   // Loop until we have consumed the full string, or encounter an invalid digit
+   while (in.len) {
+      // Shift place value for another digit
+      did_overflow = mlib_mul (&value, base) || did_overflow;
+      // Case-fold for alpha digits
+      int32_t digit = mlib_latin_tolower (in.data[0]);
+      unsigned digit_value = 0;
+      // Only standard digits
+      if (digit >= '0' && digit <= '9') {
+         // Normal digit
+         digit_value = (unsigned) (digit - '0');
+      } else if (digit >= 'a' && digit <= 'z') {
+         // Letter digits
+         digit_value = (unsigned) (digit - 'a') + 10;
+      } else {
+         // Not a valid alnum digit
+         return EINVAL;
+      }
+      if (digit_value >= base) {
+         // The digit value is out-of-range for our chosen base
+         return EINVAL;
+      }
+      // Accumulate the new digit value
+      did_overflow = mlib_add (&value, digit_value) || did_overflow;
+      // Jump to the next digit in the string
+      in = mstr_substr (in, 1);
+   }
+
+   if (did_overflow) {
+      return ERANGE;
+   }
+
+   (void) (out && (*out = value));
+   return 0;
+}
+
+/**
+ * @brief Parse a string as a 64-bit signed integer
+ *
+ * @param in The string of digits to be parsed.
+ * @param base Optional: The base to use for parsing. Use "0" to infer the base.
+ * @param out Optional storage for an int64 value to be updated with the result
+ * @return int Returns an errno value for the parse
+ *
+ * - A value of `0` indicates that the parse was successful.
+ * - A value of `EINVAL` indicates that the input string is not a valid
+ *   representation of an integer.
+ * - A value of `ERANGE` indicates that the input string is a valid integer,
+ *   but the actual encoded value cannot be represented in an `int64_t`
+ * - If the parse fails (returns non-zero), then the value at `*out` will remain
+ *   unmodified.
+ *
+ * This differs from `strtoll` in that it requires that the entire string be
+ * parsed as a valid integer. If parsing stops early, then the result will indicate
+ * an error of EINVAL.
+ */
+static inline int
+mlib_i64_parse (mstr_view in, unsigned base, int64_t *out)
+{
+   if (in.len == 0) {
+      // Empty string is not a valid integer
+      return EINVAL;
+   }
+   // Parse the possible sign prefix
+   int sign = 1;
+   // Check for a "+"
+   if (in.data[0] == '+') {
+      // Just a plus. Drop it and do nothing with it.
+      in = mstr_substr (in, 1);
+   }
+   // Check for a negative prefix
+   else if (in.data[0] == '-') {
+      // Negative sign. We'll negate the value later.
+      in = mstr_substr (in, 1);
+      sign = -1;
+   }
+
+   // Infer the base value, if we have one
+   if (base == 0) {
+      if (in.len && in.data[0] == '0') {
+         if (in.len > 1) {
+            if (mlib_latin_tolower (in.data[1]) == 'x') {
+               // Hexadecimal
+               base = 16;
+               in = mstr_substr (in, 2);
+            } else if (mlib_latin_tolower (in.data[1]) == 'o') {
+               // Octal
+               base = 8;
+               in = mstr_substr (in, 2);
+            } else if (mlib_latin_tolower (in.data[1]) == 'b') {
+               // Binary
+               base = 2;
+               in = mstr_substr (in, 2);
+            }
+         }
+         if (base == 0) {
+            // Other: Octal with a single "0" prefix. Don't trim this, because
+            // it may be a literal "0"
+            base = 8;
+         }
+      } else {
+         // No '0' prefix. Treat it as decimal
+         base = 10;
+      }
+   }
+
+   // Try to parse the natural number now that we have removed all prefixes and
+   // have a non-zero base.
+   uint64_t nat;
+   int rc = mlib_nat64_parse (in, base, &nat);
+   if (rc) {
+      return rc;
+   }
+
+   // Try to narrow from the u64 to i64 and apply the sign. This must be done as
+   // one operation because of the pathological case of parsing INT64_MIN
+   int64_t i64 = 0;
+   if (mlib_mul (&i64, nat, sign)) {
+      return ERANGE;
+   }
+
+   (void) (out && (*out = i64));
+   return 0;
+}
+
+#define mlib_i64_parse(...) MLIB_ARGC_PICK (_mlib_i64_parse, __VA_ARGS__)
+#define _mlib_i64_parse_argc_2(S, Ptr) _mlib_i64_parse_argc_3 ((S), 0, (Ptr))
+#define _mlib_i64_parse_argc_3(S, Base, Ptr) mlib_i64_parse (mstr_view_from ((S)), Base, Ptr)
+
+/**
+ * @brief Parse a 32-bit integer from a string.
+ *
+ * See `mlib_i64_parse` for more details.
+ */
+static inline int
+mlib_i32_parse (mstr_view in, unsigned base, int32_t *out)
+{
+   int64_t tmp;
+   int ec = mlib_i64_parse (in, base, &tmp);
+   if (ec) {
+      // Failed to parse the int64 value.
+      return ec;
+   }
+   // Attempt to narrow to a 32-bit value
+   int32_t i32 = 0;
+   if (mlib_narrow (&i32, tmp)) {
+      // Value is out-of-range
+      return ERANGE;
+   }
+   // Success
+   (void) (out && (*out = i32));
+   return 0;
+}
+
+#define mlib_i32_parse(...) MLIB_ARGC_PICK (_mlib_i32_parse, __VA_ARGS__)
+#define _mlib_i32_parse_argc_2(S, Ptr) _mlib_i32_parse_argc_3 ((S), 0, (Ptr))
+#define _mlib_i32_parse_argc_3(S, Base, Ptr) mlib_i32_parse (mstr_view_from ((S)), Base, Ptr)