Add marks and regex addressing systems

- Hook up parser to buffer edits.
- Other minor fixes.

src/internal/address/address.cc

@@ -29,11 +29,12 @@ Address::Address(std::string &cmd, uint64_t &i) {
     break;
   case '\'': {
     i++;
-    if (i < cmd.size() && 'a' <= cmd[i] && cmd[i] <= 'z')
+    if (i < cmd.size()
+        && (('a' <= cmd[i] && cmd[i] <= 'z') || ('A' <= cmd[i] && cmd[i] <= 'Z')))
       i++;
     else
       throw address_error("Invalid mark.");
-    base = Mark(cmd[i]);
+    base = Mark(cmd[i - 1]);
   } break;
   case '/': {
     i++;
@@ -54,7 +55,9 @@ Address::Address(std::string &cmd, uint64_t &i) {
       else
         i++;
     }
-    base = RegexLine(Direction::Forward, cmd.substr(start, i - start));
+    base = Regex(Direction::Forward, cmd.substr(start, i - start));
+    if (i < cmd.size())
+      i++;
   } break;
   case '?': {
     i++;
@@ -75,7 +78,9 @@ Address::Address(std::string &cmd, uint64_t &i) {
       else
         i++;
     }
-    base = RegexLine(Direction::Backward, cmd.substr(start, i - start));
+    base = Regex(Direction::Backward, cmd.substr(start, i - start));
+    if (i < cmd.size())
+      i++;
   } break;
   case '+': {
     base = Current();

src/internal/address/resolve.cc

@@ -17,6 +17,21 @@ uint64_t Address::resolve(BEd &ctx) {
         line = ctx.active->vase.lines();
       } else if constexpr (std::is_same_v<T, Number>) {
         line = addr.i;
+      } else if constexpr (std::is_same_v<T, Mark>) {
+        line = ctx.active->marks.get(addr.m);
+        if (line == UINT64_MAX)
+          throw address_error("Mark not set.");
+      } else if constexpr (std::is_same_v<T, Regex>) {
+        std::string_view re = addr.re;
+        if (re.size() == 0)
+          re = ctx.last_regex;
+        if (re.size() == 0)
+          throw address_error("No regex given.");
+        if (addr.dir == Direction::Forward)
+          line = ctx.active->vase.find_next(re, ctx.active->line - 1) + 1;
+        else
+          line = ctx.active->vase.find_prev(re, ctx.active->line - 1) + 1;
+        ctx.last_regex = re;
       } else if constexpr (std::is_same_v<T, Block>) {
         uint64_t current_line = ctx.active->line;
         if (ctx.active->vase.lines() > 0 && current_line == 0)

src/internal/buffer/buffer.cc

@@ -64,8 +64,6 @@ void Buffer::jump(uint64_t n_line) {
   if (n_line > vase.lines())
     throw ed_error("Line number too high.");
   line = n_line;
-  prev_range.start = line;
-  prev_range.end = line;
 }
 
 void Buffer::append(std::string text, uint64_t line) {
@@ -82,6 +80,9 @@ void Buffer::append(std::string text, uint64_t line) {
   prev_range.start = p.row + 1;
   vase.insert(&p, text);
   prev_range.end = p.row + 1;
+  marks.insert(prev_range.start, prev_range.end);
+  if (parser)
+    parser->insert(vase, prev_range.start, prev_range.end);
   modified = true;
 }
 
@@ -89,6 +90,9 @@ void Buffer::remove(uint64_t start_line, uint64_t end_line) {
   vase.erase({{start_line - 1, 0}, {end_line, 0}});
   prev_range.start = start_line;
   prev_range.end = start_line;
+  marks.erase(start_line, end_line - start_line + 1);
+  if (parser)
+    parser->erase(vase, start_line, end_line - start_line + 1);
   modified = true;
 }
 
@@ -96,6 +100,9 @@ void Buffer::join(uint64_t start_line, uint64_t end_line) {
   vase.regex_search_replace(R"(\n)", {{start_line - 1, 0}, {end_line, 0}}, "", "g");
   prev_range.start = start_line;
   prev_range.end = start_line;
+  marks.collapse(start_line, end_line - start_line);
+  if (parser)
+    parser->erase(vase, start_line, end_line - start_line);
   modified = true;
 }
 
@@ -132,6 +139,8 @@ inline void reset(std::ostream &out) {
 }
 
 void Buffer::print(BEd &ctx, uint64_t start_line, uint64_t end_line) {
+  prev_range.start = start_line;
+  prev_range.end = end_line;
   if (parser) {
     std::optional<syntax::Parser::Iterator> it_o = parser->get_hl(vase, start_line - 1);
     auto &it = *it_o;
@@ -169,8 +178,52 @@ void Buffer::print(BEd &ctx, uint64_t start_line, uint64_t end_line) {
     while (it.next() && start_line++ <= end_line)
       std::cout << it.line << std::endl;
   }
+}
+
+void Buffer::number_print(BEd &ctx, uint64_t start_line, uint64_t end_line) {
   prev_range.start = start_line;
   prev_range.end = end_line;
+  uint8_t width = 1;
+  for (uint64_t n = end_line; n >= 10; n /= 10)
+    ++width;
+  if (parser) {
+    std::optional<syntax::Parser::Iterator> it_o = parser->get_hl(vase, start_line - 1);
+    auto &it = *it_o;
+    while (start_line <= end_line) {
+      it.next();
+      std::cout << std::setw(width) << it.at << "\t";
+      const std::string &line = it.it->line;
+      const auto &tokens = it.tokens;
+      uint32_t cursor = 0;
+      for (const auto &token : tokens) {
+        const uint32_t start = token.start;
+        const uint32_t end = token.end;
+        if (start > line.size())
+          break;
+        if (end > line.size())
+          break;
+        if (cursor < start) {
+          std::cout.write(
+            line.data() + cursor,
+            start - cursor
+          );
+        }
+        const auto highlight = ctx.theme.get(token);
+        apply(std::cout, highlight);
+        std::cout.write(line.data() + start, end - start);
+        reset(std::cout);
+        cursor = end;
+      }
+      if (cursor < line.size())
+        std::cout.write(line.data() + cursor, line.size() - cursor);
+      std::cout << std::endl;
+      ++start_line;
+    }
+  } else {
+    vase::Iterator it = vase.iterate(start_line - 1, Direction::Forward);
+    while (it.next() && start_line <= end_line)
+      std::cout << std::setw(width) << start_line++ << "\t" << it.line << std::endl;
+  }
 }
 
 std::string Buffer::list_string(std::string_view s) {

src/internal/commands/commands.cc

@@ -9,6 +9,15 @@ void Suffix::register_suffixes(BEd &ctx) {
     .handle = [](BEd &ctx) {
       auto line = ctx.active->line;
       ctx.active->print(ctx, line, line);
+      ctx.active->jump(line);
+    }
+  };
+  ctx.suffixes['n' - 'a'] = Suffix{
+    .desc = "Prints current line with line number.",
+    .handle = [](BEd &ctx) {
+      auto line = ctx.active->line;
+      ctx.active->number_print(ctx, line, line);
+      ctx.active->jump(line);
     }
   };
 }
@@ -55,6 +64,26 @@ void Command::register_posix(BEd &ctx) {
       }
     }
   );
+  ctx.commands.insert(
+    "j",
+    Command{
+      .address_mode = Command::AddressMode::Range,
+      .suffix = Command::SuffixKind::Suffix,
+      .desc = "Join a set of lines (default: .,.+1)",
+      .accept_zero = true,
+      .handle = [](BEd &ctx, std::span<const uint64_t> addresses, std::string_view) {
+        std::cout << addresses.size() << "\n";
+        uint64_t start_line = ctx.active->line;
+        uint64_t end_line = ctx.active->line + 1;
+        if (addresses.size() == 1)
+          return;
+        else if (addresses.size() == 2)
+          start_line = addresses[0], end_line = addresses[1];
+        ctx.active->join(start_line, end_line);
+        ctx.active->jump(start_line);
+      }
+    }
+  );
   ctx.commands.insert(
     "q",
     Command{
@@ -92,27 +121,67 @@ void Command::register_posix(BEd &ctx) {
       .handle = [](BEd &ctx, std::span<const uint64_t> addresses, std::string_view) {
         uint64_t start_line = ctx.active->line;
         uint64_t end_line = ctx.active->line;
-        if (!addresses.size())
-          ctx.active->print(ctx, start_line, end_line);
-        else if (addresses.size() == 1)
-          ctx.active->print(ctx, addresses[0], addresses[0]);
-        else
-          ctx.active->print(ctx, addresses[0], addresses[1]);
+        if (addresses.size() == 1)
+          start_line = addresses[0], end_line = addresses[0];
+        else if (addresses.size() == 2)
+          start_line = addresses[0], end_line = addresses[1];
+        ctx.active->print(ctx, start_line, end_line);
+        ctx.active->jump(end_line);
+      }
+    }
+  );
+  ctx.commands.insert(
+    "n",
+    Command{
+      .address_mode = Command::AddressMode::Range,
+      .suffix = Command::SuffixKind::Suffix,
+      .desc = "Print range with line numbers (default .,.)",
+      .accept_zero = false,
+      .handle = [](BEd &ctx, std::span<const uint64_t> addresses, std::string_view) {
+        uint64_t start_line = ctx.active->line;
+        uint64_t end_line = ctx.active->line;
+        if (addresses.size() == 1)
+          start_line = addresses[0], end_line = addresses[0];
+        else if (addresses.size() == 2)
+          start_line = addresses[0], end_line = addresses[1];
+        ctx.active->number_print(ctx, start_line, end_line);
+        ctx.active->jump(end_line);
       }
     }
   );
   ctx.commands.insert(
     "=",
     Command{
-      .address_mode = Command::AddressMode::Single,
+      .address_mode = Command::AddressMode::Range,
       .suffix = Command::SuffixKind::Suffix,
-      .desc = "Print line number",
+      .desc = "Print line number(s)",
       .accept_zero = true,
       .handle = [](BEd &ctx, std::span<const uint64_t> addresses, std::string_view) {
         if (!addresses.size())
           std::cout << ctx.active->vase.lines() << std::endl;
-        else
+        else if (addresses.size() == 1)
           std::cout << addresses[0] << std::endl;
+        else
+          std::cout << addresses[0] << "," << addresses[1] << std::endl;
+      }
+    }
+  );
+  ctx.commands.insert(
+    "k",
+    Command{
+      .address_mode = Command::AddressMode::Single,
+      .suffix = Command::SuffixKind::Continuation,
+      .desc = "Mark a line.",
+      .accept_zero = true,
+      .handle = [](BEd &ctx, std::span<const uint64_t> addresses, std::string_view args) {
+        if (args.size() < 1 || args.size() > 2)
+          throw ed_error("Malformed mark command");
+        if (addresses.size())
+          ctx.active->marks.set(args[0], addresses[0]);
+        else
+          ctx.active->marks.set(args[0], ctx.active->line);
+        if (args.size() > 1)
+          ctx.suffix_handle(args[1]);
       }
     }
   );

src/internal/syntax/parser.cc

@@ -243,15 +243,12 @@ void Parser::modify(vase::Vase &vase, uint64_t target, uint64_t count) {
     events.clear();
     lang.parse(&state, it.line, at == 0, &tokens, &events);
     for (const auto &ev : events) {
-      const auto t = uint8_t(ev.ev_type);
-      if (t > ParseEvent::Closing)
-        continue;
       if (c.leaf->n == c.leaf->cap) {
         uint32_t cap = c.leaf->cap ? c.leaf->cap * 2 : 8;
         c.leaf->blocks = (uint16_t *)realloc(c.leaf->blocks, cap * sizeof(uint16_t));
         c.leaf->cap = cap;
       }
-      c.leaf->blocks[c.leaf->n++] = t << 15 | (at - chunk_start);
+      c.leaf->blocks[c.leaf->n++] = ev.closing << 15 | (at - chunk_start);
     }
     at++;
   }

src/internal/syntax/ruby/parse.cc

@@ -225,7 +225,7 @@ bool handle_line_markers(RubyParser &p, std::vector<Token> *tokens, std::vector<
   if (p.len() == 6 && p.peek_str(6) == "=begin") {
     p.current().state = RubyState::RubyInternalState::COMMENT;
     p.current().flags &= ~RubyState::RubyInternalState::EXPECTING_EXPRESSION;
-    events->push_back(ParseEvent::Opening);
+    events->push_back(false);
     tokens->push_back({0, p.len(), Token::Comment});
     return true;
   }
@@ -266,7 +266,7 @@ bool handle_syntax(RubyParser &p, std::vector<Token> *tokens, std::vector<ParseE
     case RubyState::RubyInternalState::CLASS_NAME: {
       if (p.peek() == '<' && p.peek(1) == '<') {
         p.advance(2);
-        events->push_back(ParseEvent::Opening);
+        events->push_back(false);
         p.current().flags = (p.current().flags & ~RubyState::RubyInternalState::NAME_MASK);
         p.current().flags |= RubyState::RubyInternalState::EXPECTING_EXPRESSION;
         return false;
@@ -285,7 +285,7 @@ bool handle_syntax(RubyParser &p, std::vector<Token> *tokens, std::vector<ParseE
           return false;
         }
       }
-      events->push_back(ParseEvent::Opening);
+      events->push_back(false);
       p.current().flags = (p.current().flags & ~RubyState::RubyInternalState::NAME_MASK);
       return false;
     }
@@ -311,7 +311,7 @@ bool handle_syntax(RubyParser &p, std::vector<Token> *tokens, std::vector<ParseE
           return false;
         }
       }
-      events->push_back(ParseEvent::Opening);
+      events->push_back(false);
       p.current().flags = (p.current().flags & ~RubyState::RubyInternalState::NAME_MASK);
       return false;
     }
@@ -330,7 +330,7 @@ bool handle_syntax(RubyParser &p, std::vector<Token> *tokens, std::vector<ParseE
           return false;
         }
       }
-      events->push_back(ParseEvent::Opening);
+      events->push_back(false);
       p.current().flags = (p.current().flags & ~RubyState::RubyInternalState::NAME_MASK);
       return false;
     }
@@ -899,21 +899,21 @@ bool handle_syntax(RubyParser &p, std::vector<Token> *tokens, std::vector<ParseE
         p.advance(j);
         return false;
       } else if (j == tries.expecting_end_keywords_trie.longest_match(p.peek_str(j))) {
-        events->push_back(ParseEvent::Opening);
+        events->push_back(false);
         p.current().flags |= RubyState::RubyInternalState::EXPECTING_EXPRESSION;
         tokens->push_back({p.i, p.i + j, Token::Keyword});
         p.advance(j);
         return false;
       } else if (j == tries.conditional_keywords_trie.longest_match(p.peek_str(j))) {
         if (p.current().flags & RubyState::RubyInternalState::NEWLINE || p.op_last)
-          events->push_back(ParseEvent::Opening);
+          events->push_back(false);
         p.current().flags |= RubyState::RubyInternalState::EXPECTING_EXPRESSION;
         tokens->push_back({p.i, p.i + j, Token::Keyword});
         p.advance(j);
         return false;
       } else if (j == tries.end_keywords_trie.longest_match(p.peek_str(j))) {
         p.current().flags |= RubyState::RubyInternalState::EXPECTING_EXPRESSION;
-        events->push_back(ParseEvent::Opening);
+        events->push_back(false);
         tokens->push_back({p.i, p.i + j, Token::Keyword});
         p.advance(j);
         return false;
@@ -950,7 +950,7 @@ bool handle_syntax(RubyParser &p, std::vector<Token> *tokens, std::vector<ParseE
           return false;
         }
         if (j == 3 && p.peek_str(3) == "end") {
-          events->push_back(ParseEvent::Closing);
+          events->push_back(true);
           tokens->push_back({p.i, p.i + j, Token::Keyword});
           p.advance(3);
           return false;
@@ -980,17 +980,18 @@ bool handle_syntax(RubyParser &p, std::vector<Token> *tokens, std::vector<ParseE
           return false;
         }
         uint32_t start = p.i;
-        p.advance(j);
-        if (p.peek() == ':') {
+        if (p.peek(j) == ':') {
+          p.advance(j);
           p.advance();
           tokens->push_back({start, p.i, Token::Label});
           return false;
-        } else if (p.peek(-1) == '!' || p.peek(-1) == '?') {
-          p.advance();
+        } else if (p.peek(j - 1) == '!' || p.peek(j - 1) == '?') {
+          p.advance(j);
           tokens->push_back({start, p.i, Token::Function});
           return false;
         } else {
-          uint32_t j = 0;
+          p.advance(j);
+          j = 0;
           if (p.peek(j) == '(' || p.peek(j) == '{') {
             tokens->push_back({start, p.i, Token::Function});
             return false;
@@ -1066,7 +1067,7 @@ void ruby_parse(
       tokens->push_back({p.i, p.len(), Token::Comment});
       if (p.i == 0 && p.peek_str(4) == "=end") {
         p.current().state = RubyState::RubyInternalState::NONE;
-        events->push_back(ParseEvent::Closing);
+        events->push_back(true);
       }
       return;
     }

src/internal/vase/regex/api.cc

@@ -156,4 +156,122 @@ std::vector<Range> Vase::regex_search(
     result.push_back({point_of(match.start), point_of(match.end)});
   return result;
 }
+
+uint64_t Vase::find_next(std::string_view pattern, uint64_t start) {
+  std::vector<RegexMatch> results;
+  int errornumber;
+  PCRE2_SIZE erroroffset;
+  pcre2_code *re = pcre2_compile(
+    (PCRE2_SPTR)pattern.data(),
+    pattern.size(),
+    PCRE2_UTF | PCRE2_EXTENDED,
+    &errornumber,
+    &erroroffset,
+    NULL
+  );
+  if (re == NULL)
+    throw ed_error("Can't compile regex.");
+  pcre2_match_data *match_data = pcre2_match_data_create_from_pattern(re, nullptr);
+  if (!match_data) {
+    pcre2_code_free(re);
+    throw ed_error("Can't create regex match data.");
+  }
+  uint64_t at = (start + 1) % lines();
+  LineIterator it(root, at, Direction::Forward);
+  std::string line;
+  while (it.next(&line)) {
+    int rc = pcre2_match(re, (PCRE2_SPTR)line.data(), line.size(), 0, 0, match_data, nullptr);
+    if (rc >= 0) {
+      pcre2_match_data_free(match_data);
+      pcre2_code_free(re);
+      return at;
+    }
+    if (rc != PCRE2_ERROR_NOMATCH) {
+      pcre2_match_data_free(match_data);
+      pcre2_code_free(re);
+      throw ed_error("Regex matching failed.");
+    }
+    at++;
+  }
+  at = 0;
+  LineIterator it2(root, at, Direction::Forward);
+  while (it2.next(&line)) {
+    if (at >= start)
+      break;
+    int rc = pcre2_match(re, (PCRE2_SPTR)line.data(), line.size(), 0, 0, match_data, nullptr);
+    if (rc >= 0) {
+      pcre2_match_data_free(match_data);
+      pcre2_code_free(re);
+      return at;
+    }
+    if (rc != PCRE2_ERROR_NOMATCH) {
+      pcre2_match_data_free(match_data);
+      pcre2_code_free(re);
+      throw ed_error("Regex matching failed.");
+    }
+    at++;
+  }
+  pcre2_match_data_free(match_data);
+  pcre2_code_free(re);
+  throw ed_error("No line matched.");
+}
+
+uint64_t Vase::find_prev(std::string_view pattern, uint64_t start) {
+  std::vector<RegexMatch> results;
+  int errornumber;
+  PCRE2_SIZE erroroffset;
+  pcre2_code *re = pcre2_compile(
+    (PCRE2_SPTR)pattern.data(),
+    pattern.size(),
+    PCRE2_UTF | PCRE2_EXTENDED,
+    &errornumber,
+    &erroroffset,
+    NULL
+  );
+  if (re == NULL)
+    throw ed_error("Can't compile regex.");
+  pcre2_match_data *match_data = pcre2_match_data_create_from_pattern(re, nullptr);
+  if (!match_data) {
+    pcre2_code_free(re);
+    throw ed_error("Can't create regex match data.");
+  }
+  uint64_t at = (start == 0 ? lines() : start) - 1;
+  LineIterator it(root, at, Direction::Backward);
+  std::string line;
+  while (it.next(&line)) {
+    int rc = pcre2_match(re, (PCRE2_SPTR)line.data(), line.size(), 0, 0, match_data, nullptr);
+    if (rc >= 0) {
+      pcre2_match_data_free(match_data);
+      pcre2_code_free(re);
+      return at;
+    }
+    if (rc != PCRE2_ERROR_NOMATCH) {
+      pcre2_match_data_free(match_data);
+      pcre2_code_free(re);
+      throw ed_error("Regex matching failed.");
+    }
+    at--;
+  }
+  at = lines();
+  LineIterator it2(root, at, Direction::Backward);
+  while (it2.next(&line)) {
+    if (at <= start)
+      break;
+    int rc = pcre2_match(re, (PCRE2_SPTR)line.data(), line.size(), 0, 0, match_data, nullptr);
+    if (rc >= 0) {
+      pcre2_match_data_free(match_data);
+      pcre2_code_free(re);
+      return at;
+    }
+    if (rc != PCRE2_ERROR_NOMATCH) {
+      pcre2_match_data_free(match_data);
+      pcre2_code_free(re);
+      throw ed_error("Regex matching failed.");
+    }
+    at--;
+  }
+  pcre2_match_data_free(match_data);
+  pcre2_code_free(re);
+  throw ed_error("No line matched.");
+}
 } // namespace bed::internal::vase