Add some more ed commands.

include/commands/ed/ed.h

@@ -17,12 +17,15 @@ struct Command {
     HelpToggle,
     Help,
     Print,
+    List,
     Number,
     Append,
     Insert,
     Change,
     Delete,
     Write,
+    Join,
+    Mark,
     Edit,
     ForceEdit,
     Filename,
@@ -85,6 +88,8 @@ struct Ed {
   Command parse(std::string cmd, bool eof);
   void append(std::string text, uint64_t line);
   void remove(uint64_t start_line, uint64_t end_line);
+  std::string list_string(std::string_view s);
+  void join(uint64_t start_line, uint64_t end_line);
   bool handle(std::string cmd, bool eof);
 };
 

src/commands/ed/edit.cc

@@ -1,21 +0,0 @@
-#include "commands/ed/ed.h"
-
-namespace crib::commands::ed {
-void Ed::append(std::string text, uint64_t line) {
-  using namespace crib::internal::vase;
-  Point p = {line, 0};
-  if (!vase.lines()) {
-    text.pop_back();
-  } else if (line == vase.lines()) {
-    p.row--;
-    p.col = UINT64_MAX;
-    text = "\n" + text;
-    text.pop_back();
-  }
-  vase.insert(&p, text);
-}
-
-void Ed::remove(uint64_t start_line, uint64_t end_line) {
-  vase.erase({{start_line - 1, 0}, {end_line, 0}});
-}
-} // namespace crib::commands::ed

src/commands/ed/handle.cc

@@ -70,6 +70,24 @@ bool Ed::handle(std::string cmd, bool eof) {
         std::cout << it.line << std::endl;
       line = line_end;
     } break;
+    case Command::Type::List: {
+      uint64_t line_start = line;
+      uint64_t line_end = line;
+      if (command.start.type != Command::Address::Type::None) {
+        resolve_address(command.start, &line_start);
+        line_end = line_start;
+        if (command.address_flags & Command::RANGE)
+          resolve_address(command.end, &line_end);
+      }
+      if (line_start == 0)
+        throw ed_error("Line 0 is invalid.");
+      if (line_end < line_start)
+        throw ed_error("Invalid address range.");
+      Iterator it = vase.iterate(line_start - 1, Direction::Forward);
+      while (it.next() && line_start++ <= line_end)
+        std::cout << list_string(it.line) << std::endl;
+      line = line_end;
+    } break;
     case Command::Type::Number: {
       uint64_t line_start = line;
       uint64_t line_end = line;
@@ -89,6 +107,12 @@ bool Ed::handle(std::string cmd, bool eof) {
       line = line_end;
     } break;
     case Command::Type::Append: {
+      if (command.start.type != Command::Address::Type::None) {
+        if (command.address_flags & Command::RANGE)
+          resolve_address(command.end, &line);
+        else
+          resolve_address(command.start, &line);
+      }
       std::string text;
       std::string cline;
       uint64_t line_count = 0;
@@ -99,12 +123,6 @@ bool Ed::handle(std::string cmd, bool eof) {
         text.append(cline);
         text.push_back('\n');
       }
-      if (command.start.type != Command::Address::Type::None) {
-        if (command.address_flags & Command::RANGE)
-          resolve_address(command.end, &line);
-        else
-          resolve_address(command.start, &line);
-      }
       if (text.empty()) {
         if (line == 0)
           line = 1;
@@ -116,6 +134,14 @@ bool Ed::handle(std::string cmd, bool eof) {
       line += line_count;
     } break;
     case Command::Type::Insert: {
+      if (command.start.type != Command::Address::Type::None) {
+        if (command.address_flags & Command::RANGE)
+          resolve_address(command.end, &line);
+        else
+          resolve_address(command.start, &line);
+      }
+      if (line == 0)
+        line = 1;
       std::string text;
       std::string cline;
       uint64_t line_count = 0;
@@ -126,14 +152,6 @@ bool Ed::handle(std::string cmd, bool eof) {
         text.append(cline);
         text.push_back('\n');
       }
-      if (command.start.type != Command::Address::Type::None) {
-        if (command.address_flags & Command::RANGE)
-          resolve_address(command.end, &line);
-        else
-          resolve_address(command.start, &line);
-      }
-      if (line == 0)
-        line = 1;
       if (text.empty())
         break;
       vase.snapshot();
@@ -247,6 +265,39 @@ bool Ed::handle(std::string cmd, bool eof) {
       if (!suppress_mode)
         std::cout << bytes << std::endl;
     } break;
+    case Command::Type::Join: {
+      uint64_t line_start = line;
+      uint64_t line_end = line + 1;
+      if (command.start.type != Command::Address::Type::None) {
+        resolve_address(command.start, &line_start);
+        line_end = line_start;
+        if (command.address_flags & Command::RANGE)
+          resolve_address(command.end, &line_end);
+      }
+      if (line_end < line_start)
+        throw ed_error("Invalid address range.");
+      if (line_start == 0)
+        throw ed_error("Invalid line number given.");
+      if (line_end == line_start)
+        break;
+      vase.snapshot();
+      join(line_start, line_end);
+      line = line_start;
+      modified = true;
+    } break;
+    case Command::Type::Mark: {
+      uint64_t mark_line = line;
+      if (command.start.type != Command::Address::Type::None) {
+        if (command.address_flags & Command::RANGE)
+          resolve_address(command.end, &mark_line);
+        else
+          resolve_address(command.start, &mark_line);
+      }
+      if (mark_line == 0)
+        throw ed_error("Invalid line number given.");
+      char mark = command.argument[0];
+      marks[mark - 'a'] = mark_line;
+    } break;
     case Command::Type::Edit: {
       if (modified && !was_editing) {
         editing = true;
@@ -318,8 +369,7 @@ bool Ed::handle(std::string cmd, bool eof) {
         std::cout << line << "\t" << it.line << std::endl;
         break;
       case 'l':
-        // TODO: escaping.
+        std::cout << list_string(it.line) << std::endl;
-        std::cout << it.line << std::endl;
         break;
       }
     }

src/commands/ed/helpers.cc

@@ -0,0 +1,83 @@
+#include "commands/ed/ed.h"
+
+namespace crib::commands::ed {
+void Ed::append(std::string text, uint64_t line) {
+  using namespace crib::internal::vase;
+  Point p = {line, 0};
+  if (!vase.lines()) {
+    text.pop_back();
+  } else if (line == vase.lines()) {
+    p.row--;
+    p.col = UINT64_MAX;
+    text = "\n" + text;
+    text.pop_back();
+  }
+  vase.insert(&p, text);
+}
+
+void Ed::remove(uint64_t start_line, uint64_t end_line) {
+  vase.erase({{start_line - 1, 0}, {end_line, 0}});
+}
+
+void Ed::join(uint64_t start_line, uint64_t end_line) {
+  vase.regex_search_replace("\\n", {{start_line - 1, 0}, {end_line, 0}}, "", "");
+}
+
+std::string Ed::list_string(std::string_view s) {
+  uint32_t width = 80;
+  winsize ws{};
+  if (ioctl(STDOUT_FILENO, TIOCGWINSZ, &ws) == 0 && ws.ws_col != 0)
+    width = ws.ws_col;
+  std::string out;
+  out.reserve(s.size());
+  const uint32_t max_width = width > 1 ? width - 1 : 1;
+  uint32_t column = 0;
+  auto append = [&](std::string_view text) {
+    if (column + text.size() > max_width) {
+      out += "\\\n";
+      column = 0;
+    }
+    out += text;
+    column += text.size();
+  };
+  for (unsigned char c : s) {
+    switch (c) {
+    case '\\':
+      append("\\\\");
+      break;
+    case '$':
+      append("\\$");
+      break;
+    case '\a':
+      append("\\a");
+      break;
+    case '\b':
+      append("\\b");
+      break;
+    case '\f':
+      append("\\f");
+      break;
+    case '\r':
+      append("\\r");
+      break;
+    case '\t':
+      append("\\t");
+      break;
+    case '\v':
+      append("\\v");
+      break;
+    default:
+      if (!std::isprint(c)) {
+        char buf[5];
+        std::snprintf(buf, sizeof(buf), "\\%03o", c);
+        append(buf);
+      } else {
+        append(std::string_view((const char *)&c, 1));
+      }
+      break;
+    }
+  }
+  out += '$';
+  return out;
+}
+} // namespace crib::commands::ed

src/commands/ed/parsers/command.cc

@@ -59,6 +59,10 @@ Command Ed::parse(std::string cmd, bool eof) {
     command.type = Command::Type::Print;
     i++;
     break;
+  case 'l':
+    command.type = Command::Type::List;
+    i++;
+    break;
   case 'n':
     command.type = Command::Type::Number;
     i++;
@@ -91,6 +95,10 @@ Command Ed::parse(std::string cmd, bool eof) {
     command.type = Command::Type::Delete;
     i++;
     break;
+  case 'j':
+    command.type = Command::Type::Join;
+    i++;
+    break;
   case 'w':
     command.type = Command::Type::Write;
     i++;
@@ -135,6 +143,16 @@ Command Ed::parse(std::string cmd, bool eof) {
       throw ed_error("Invalid command.");
     command.argument = cmd.substr(i);
     return command;
+  case 'k':
+    command.type = Command::Type::Mark;
+    i++;
+    if (i >= cmd.size())
+      throw ed_error("Invalid command.");
+    if ('a' <= cmd[i] && cmd[i] <= 'z')
+      command.argument = cmd[i];
+    else
+      throw ed_error("Invalid mark used.");
+    break;
   case '#':
     command.type = Command::Type::Dump;
     i++;

src/internal/vase/regex/api.cc

@@ -7,6 +7,12 @@ std::vector<Vase::ReplacePart> Vase::parse_replace(std::string_view s) {
   auto flush_constant = [&]() {
     if (!constant.empty()) {
       uint64_t lines = 0;
+      const char *p = constant.data();
+      const char *end = p + constant.size();
+      while ((p = (const char *)memchr(p, '\n', end - p))) {
+        ++lines;
+        ++p;
+      }
       uint64_t pos = append->append(constant.data(), (uint64_t)constant.size());
       parts.push_back(
         ReplacePart{

src/internal/vase/regex/core.cc

@@ -7,8 +7,6 @@ std::vector<Vase::RegexMatch> Vase::_regex_search(
   bool global = false;
   uint64_t flags = PCRE2_MULTILINE | PCRE2_UTF;
 
-  const char *dot = "(?:(?!\\n)\\X)";
-
   for (char c : options) {
     switch (c) {
     case 'g':
@@ -18,7 +16,7 @@ std::vector<Vase::RegexMatch> Vase::_regex_search(
       flags |= PCRE2_CASELESS;
       break;
     case 's':
-      dot = "(?:\\X)";
+      flags |= PCRE2_DOTALL;
       break;
     case 'x':
       flags |= PCRE2_EXTENDED;
@@ -40,57 +38,9 @@ std::vector<Vase::RegexMatch> Vase::_regex_search(
   int errornumber;
   PCRE2_SIZE erroroffset;
 
-  std::string out;
-  out.reserve(pattern.size() * 2);
-  bool escaped = false;
-  bool in_class = false;
-  bool in_quote = false;
-  for (size_t i = 0; i < pattern.size(); i++) {
-    char c = pattern[i];
-    if (escaped) {
-      out += c;
-      escaped = false;
-      continue;
-    }
-    if (c == '\\') {
-      out += c;
-      if (i + 1 < pattern.size()) {
-        char next = pattern[i + 1];
-        if (next == 'Q')
-          in_quote = true;
-        else if (next == 'E')
-          in_quote = false;
-        out += next;
-        i++;
-        continue;
-      }
-      escaped = true;
-      continue;
-    }
-    if (in_quote) {
-      out += c;
-      continue;
-    }
-    if (c == '[') {
-      in_class = true;
-      out += c;
-      continue;
-    }
-    if (c == ']' && in_class) {
-      in_class = false;
-      out += c;
-      continue;
-    }
-    if (c == '.' && !in_class) {
-      out += dot;
-      continue;
-    }
-    out += c;
-  }
-
   pcre2_code *re = pcre2_compile(
-    (PCRE2_SPTR)out.data(),
+    (PCRE2_SPTR)pattern.data(),
-    out.size(),
+    pattern.size(),
     flags,
     &errornumber,
     &erroroffset,