Merge pull request #1720 from evoskuil/master

Re-implement URI parser using boost.url (dependency), remove "strict".

Author: evoskuil

include/bitcoin/system/boost.hpp

@@ -26,12 +26,14 @@
 
 // Include boost only from here, so exception disable works.
 #include <boost/asio.hpp>
+#include <boost/beast.hpp>
 #include <boost/format.hpp>
 #include <boost/iostreams/stream.hpp>
 #include <boost/json.hpp>
 #include <boost/locale.hpp>
 #include <boost/multiprecision/cpp_int.hpp>
 #include <boost/program_options.hpp>
+#include <boost/url.hpp>
 
 // ADL free functions for use with boost-json.
 #define DECLARE_JSON_VALUE_CONVERTORS(name) \

include/bitcoin/system/impl/unicode/ascii.ipp

@@ -54,6 +54,11 @@ constexpr bool is_ascii_alpha(char32_t point) NOEXCEPT
     return is_ascii_lower(point) || is_ascii_upper(point);
 }
 
+constexpr bool is_ascii_alphanumeric(char32_t point) NOEXCEPT
+{
+    return is_ascii_alpha(point) || is_ascii_number(point);
+}
+
 // see char32_separators[]
 constexpr bool is_ascii_separator(char32_t point) NOEXCEPT
 {

include/bitcoin/system/unicode/ascii.hpp

@@ -30,6 +30,7 @@ constexpr bool is_ascii_number(char32_t point) NOEXCEPT;
 constexpr bool is_ascii_lower(char32_t point) NOEXCEPT;
 constexpr bool is_ascii_upper(char32_t point) NOEXCEPT;
 constexpr bool is_ascii_alpha(char32_t point) NOEXCEPT;
+constexpr bool is_ascii_alphanumeric(char32_t point) NOEXCEPT;
 constexpr bool is_ascii_separator(char32_t point) NOEXCEPT;
 constexpr bool is_ascii_whitespace(char32_t point) NOEXCEPT;
 

include/bitcoin/system/wallet/addresses/bitcoin_uri.hpp

@@ -39,7 +39,7 @@ class BC_API bitcoin_uri
 
     /// Constructors.
     bitcoin_uri() NOEXCEPT;
-    bitcoin_uri(const std::string& uri, bool strict=true) NOEXCEPT;
+    bitcoin_uri(const std::string& uri) NOEXCEPT;
 
     /// Operators.
     bool operator<(const bitcoin_uri& other) const NOEXCEPT;
@@ -75,7 +75,6 @@ class BC_API bitcoin_uri
     void set_address(const stealth_address& stealth) NOEXCEPT;
 
     /// uri_reader implementation.
-    void set_strict(bool strict) NOEXCEPT;
     bool set_scheme(const std::string& scheme) NOEXCEPT;
     bool set_authority(const std::string& authority) NOEXCEPT;
     bool set_path(const std::string& path) NOEXCEPT;
@@ -87,7 +86,6 @@ class BC_API bitcoin_uri
     bool set_amount(const std::string& satoshis) NOEXCEPT;
 
     /// Member state.
-    bool strict_;
     std::string scheme_;
     std::string address_;
     std::map<std::string, std::string> query_;

include/bitcoin/system/wallet/addresses/uri.hpp

@@ -30,42 +30,42 @@ namespace wallet {
 class BC_API uri
 {
 public:
+    typedef std::map<std::string, std::string> query_map;
+
     DEFAULT_COPY_MOVE_DESTRUCT(uri);
     uri() NOEXCEPT {}
 
     /// Decodes a URI from a string.
-    /// @param strict set to false to tolerate unescaped special characters.
-    bool decode(const std::string& encoded, bool strict=true) NOEXCEPT;
+    bool decode(const std::string& encoded) NOEXCEPT;
     std::string encoded() const NOEXCEPT;
 
-    /// Returns the lowercased URI scheme.
+    /// Returns the lowercased URI scheme (set empty to remove).
     std::string scheme() const NOEXCEPT;
-    void set_scheme(const std::string& scheme) NOEXCEPT;
+    bool has_scheme() const NOEXCEPT;
+    bool set_scheme(const std::string& scheme) NOEXCEPT;
 
     /// Obtains the unescaped authority part, if any (user@server:port).
     std::string authority() const NOEXCEPT;
     bool has_authority() const NOEXCEPT;
-    void set_authority(const std::string& authority) NOEXCEPT;
+    bool set_authority(const std::string& authority) NOEXCEPT;
     void remove_authority() NOEXCEPT;
 
     /// Obtains the unescaped path part.
     std::string path() const NOEXCEPT;
-    void set_path(const std::string& path) NOEXCEPT;
+    bool set_path(const std::string& path) NOEXCEPT;
 
     /// Returns the unescaped query string, if any.
     std::string query() const NOEXCEPT;
     bool has_query() const NOEXCEPT;
-    void set_query(const std::string& query) NOEXCEPT;
+    bool set_query(const std::string& query) NOEXCEPT;
     void remove_query() NOEXCEPT;
 
     /// Returns the unescaped fragment string, if any.
     std::string fragment() const NOEXCEPT;
     bool has_fragment() const NOEXCEPT;
-    void set_fragment(const std::string& fragment) NOEXCEPT;
+    bool set_fragment(const std::string& fragment) NOEXCEPT;
     void remove_fragment() NOEXCEPT;
 
-    typedef std::map<std::string, std::string> query_map;
-
     /// Interprets the query string as a sequence of key-value pairs.
     /// All query strings are valid, so this function cannot fail.
     /// The results are unescaped. Both keys and values can be zero-length,
@@ -74,16 +74,8 @@ class BC_API uri
     void encode_query(const query_map& map) NOEXCEPT;
 
 private:
-    // All parts are stored with their original escaping.
-    std::string scheme_;
-    std::string authority_;
-    std::string path_;
-    std::string query_;
-    std::string fragment_;
-
-    bool has_authority_ = false;
-    bool has_query_ = false;
-    bool has_fragment_ = false;
+    // Mutable URL object to store the parsed URI.
+    boost::urls::url url_;
 };
 
 } // namespace wallet

include/bitcoin/system/wallet/addresses/uri_reader.hpp

@@ -39,15 +39,16 @@ class BC_API uri_reader
     /// @return The parsed URI or a default instance if the `uri` is malformed
     /// according to the  `UriReader`.
     template <class UriReader>
-    static UriReader parse(const std::string& uri, bool strict=true) NOEXCEPT
+    static UriReader parse(const std::string& uri) NOEXCEPT
     {
         wallet::uri parsed;
-        if (!parsed.decode(uri, strict))
+        if (!parsed.decode(uri))
+            return {};
+
+        UriReader out{};
+        if (!out.set_scheme(parsed.scheme()))
             return {};
 
-        UriReader out;
-        out.set_strict(strict);
-        out.set_scheme(parsed.scheme());
         if (parsed.has_authority() && !out.set_authority(parsed.authority()))
             return {};
 
@@ -70,7 +71,6 @@ class BC_API uri_reader
     }
 
     /// uri_reader interface.
-    virtual void set_strict(bool strict) NOEXCEPT = 0;
     virtual bool set_scheme(const std::string& scheme) NOEXCEPT = 0;
     virtual bool set_authority(const std::string& authority) NOEXCEPT = 0;
     virtual bool set_path(const std::string& path) NOEXCEPT = 0;

src/chain/script.cpp

@@ -45,13 +45,13 @@ BC_PUSH_WARNING(NO_ARRAY_INDEXING)
 // that indicates height size. This is inconsistent with an extreme future
 // where the size byte overflows. However satoshi actually requires nominal
 // encoding.
-//*************************************************************************
+//*****************************************************************************
 bool script::is_coinbase_pattern(const operations& ops, size_t height) NOEXCEPT
 {
     using namespace machine::number;
     return !ops.empty()
         && ops[0].is_nominal_push()
-        && ops[0].data() == chunk::from_integer(to_unsigned(height));
+        && ops[0].data() == chunk::from_integer(to_signed(height));
 }
 
 // Constructors.

src/wallet/addresses/bitcoin_uri.cpp

@@ -38,12 +38,11 @@ static const auto parameter_req_ = "req-";
 static constexpr size_t parameter_req_length = 4;
 
 bitcoin_uri::bitcoin_uri() NOEXCEPT
-  : strict_(true)
 {
 }
 
-bitcoin_uri::bitcoin_uri(const std::string& uri, bool strict) NOEXCEPT
-  : bitcoin_uri(uri_reader::parse<bitcoin_uri>(uri, strict))
+bitcoin_uri::bitcoin_uri(const std::string& uri) NOEXCEPT
+  : bitcoin_uri(uri_reader::parse<bitcoin_uri>(uri))
 {
 }
 
@@ -171,7 +170,7 @@ void bitcoin_uri::set_address(const stealth_address& stealth) NOEXCEPT
 bool bitcoin_uri::set_amount(const std::string& satoshis) NOEXCEPT
 {
     uint64_t decoded;
-    if (!decode_base10(decoded, satoshis, btc_decimal_places, strict_))
+    if (!decode_base10(decoded, satoshis, btc_decimal_places, true))
         return false;
 
     // Normalize the encoding for string-based getter (parameter).
@@ -182,11 +181,6 @@ bool bitcoin_uri::set_amount(const std::string& satoshis) NOEXCEPT
 // uri_reader implementation.
 // ----------------------------------------------------------------------------
 
-void bitcoin_uri::set_strict(bool strict) NOEXCEPT
-{
-    strict_ = strict;
-}
-
 bool bitcoin_uri::set_scheme(const std::string& scheme) NOEXCEPT
 {
     if (scheme == bitcoin_scheme)
@@ -201,8 +195,8 @@ bool bitcoin_uri::set_scheme(const std::string& scheme) NOEXCEPT
 bool bitcoin_uri::set_authority(const std::string& authority) NOEXCEPT
 {
     // Using "bitcoin://" instead of "bitcoin:" is a common mistake, so we
-    // allow the authority in place of the path when not strict.
-    return !strict_ && set_path(authority);
+    // allow the authority in place of the path.
+    return set_path(authority);
 }
 
 bool bitcoin_uri::set_path(const std::string& path) NOEXCEPT
@@ -248,8 +242,8 @@ bool bitcoin_uri::operator<(const bitcoin_uri& other) const NOEXCEPT
 
 bool bitcoin_uri::operator==(const bitcoin_uri& other) const NOEXCEPT
 {
-    return strict_ == other.strict_ && scheme_ == other.scheme_ &&
-        address_ == other.address_ && query_ == other.query_;
+    return scheme_ == other.scheme_ && address_ == other.address_ &&
+        query_ == other.query_;
 }
 
 bool bitcoin_uri::operator!=(const bitcoin_uri& other) const NOEXCEPT