Fix garbage throttle values in the status bar on 32-bit archs.

Closes #1683.

doc/manual/paths.md

@@ -0,0 +1,60 @@
+# Resolving paths
+
+Commands that return a path have `.realpath` variants that resolve it to a
+canonical one, with symlinks followed and any `.` or `..` components removed.
+
+The intent is to make paths safer to hand to an external script.
+A script called through `execute` receives whatever path rtorrent gives it, and
+many scripts do no sanity checking of their own, so resolving the path before it
+leaves rtorrent removes a class of surprises: a download directory that is a
+symlink into somewhere unexpected, or a torrent whose name walks upwards out of
+the directory it is supposed to live in.
+
+    # Instead of this
+    execute = ~/bin/on-finished, (d.base_path)
+
+    # Pass the resolved path
+    execute = ~/bin/on-finished, (d.base_path.realpath.or_throw)
+
+## Available variants
+
+Each command below comes in an `.or_empty` and an `.or_throw` form.
+
+| Command | Resolves |
+| --- | --- |
+| `d.base_path.realpath.*` | `d.base_path` |
+| `d.directory.realpath.*` | `d.directory` |
+| `d.tied_to_file.realpath.*` | `d.tied_to_file` |
+| `d.loaded_file.realpath.*` | `d.loaded_file` |
+| `f.frozen_path.realpath.*` | `f.frozen_path` |
+| `session.path.realpath.*` | `session.path` |
+| `directory.default.realpath.*` | `directory.default` |
+
+A leading `~` is expanded first, exactly as it is for `execute`, so
+`~/downloads` resolves the same way it would on the command line.
+
+## Paths that do not exist
+
+Resolving requires the path to name an existing file or directory, which is
+often not the case for a download whose data has not been written yet.
+The two forms differ only in what they do about it.
+
+`.or_empty` returns an empty string:
+
+    print = (d.base_path.realpath.or_empty)    # ""
+
+This keeps `d.multicall` usable over a view that mixes started and unstarted
+downloads, since a single unresolvable path does not abort the whole call.
+A script receiving one of these paths should still check that it is not empty
+before acting on it.
+
+`.or_throw` raises an error instead:
+
+    print = (d.base_path.realpath.or_throw)    # Could not resolve path: '...'
+
+Prefer this one wherever an unresolvable path means the command should not run
+at all, such as a single `execute` on `event.download.finished`.
+
+Note that a download only has a file list once it has been opened, so
+`d.base_path` and `f.frozen_path` are empty at `event.download.inserted` time
+and their `.realpath` variants resolve nothing.

doc/manual/string_functions.md

@@ -0,0 +1,139 @@
+# String functions
+
+The `string.*` commands manipulate and inspect text from within the
+configuration file.
+They are pure functions with no side effects, so they can be nested freely and
+are safe to call over RPC.
+
+Every argument is converted to its string representation before use, which means
+numbers can be passed where text is expected.
+Commands that count characters, such as `string.length` and `string.substr`, count
+utf-8 characters rather than bytes.
+
+## string.length
+
+    # string.length = «text»
+
+    string.length = "héllo"    # 5
+
+Returns the number of utf-8 characters in the text.
+
+## string.equals
+
+    # string.equals = «text», «other»[, ...]
+
+    string.equals = (d.name), "first.iso", "second.iso"
+
+Returns `1` if the first argument equals any of the following arguments,
+otherwise `0`.
+
+## string.starts_with, string.ends_with
+
+    # string.starts_with = «text», «prefix»[, ...]
+    # string.ends_with = «text», «tail»[, ...]
+
+    string.starts_with = (t.url), "http://", "https://"
+    string.ends_with = (d.name), ".iso"
+
+Returns `1` if the text begins, or ends, with any of the given prefixes or
+tails.
+
+## string.contains, string.contains_i
+
+    # string.contains = «haystack», «needle»[, ...]
+    # string.contains_i = «haystack», «needle»[, ...]
+
+    string.contains = (t.url), "retracker.local"
+    string.contains_i = (t.url), "RETRACKER.local"
+
+Returns `1` if the haystack contains any of the needles.
+The `_i` variant compares case-insensitively, and only handles ascii.
+
+## string.substr
+
+    # string.substr = «text»[, «position»[, «count»[, «default»]]]
+
+    string.substr = "abcdef", 2, 3       # "cde"
+    string.substr = "abcdef", -2         # "ef"
+    string.substr = "abcdef", 10, 1, "?" # "?"
+
+Extracts a part of the text, starting at `position` and spanning `count`
+characters.
+The position defaults to the start of the text, and the count to the rest of it.
+A negative position is relative to the end of the text.
+If the position falls outside the text, the default value is returned instead,
+which is the empty string unless given.
+
+## string.split
+
+    # string.split = «text», «delimiter»
+
+    string.split = "a.b.c", "."    # {"a", "b", "c"}
+    string.split = "abc", ""       # {"a", "b", "c"}
+
+Splits the text into a list, keeping empty fields.
+An empty delimiter splits the text into its utf-8 characters.
+
+## string.join
+
+    # string.join = «delimiter»[, «object»[, ...]]
+
+    string.join = "-", (string.split, "a.b.c", ".")    # "a-b-c"
+
+Concatenates the objects, inserting the delimiter between them.
+Lists are flattened, so the result of `string.split` can be joined back
+together.
+
+## string.lpad, string.rpad
+
+    # string.lpad = «text», «length»[, «padding»]
+    # string.rpad = «text», «length»[, «padding»]
+
+    string.lpad = 7, 3, 0    # "007"
+    string.rpad = "a", 3     # "a  "
+
+Pads the text at the start, or the end, until it is `length` characters long.
+The padding defaults to a single space and is repeated as needed.
+Text that is already long enough is returned unchanged.
+
+## string.strip, string.lstrip, string.rstrip
+
+    # string.strip = «text»[, «strippable»[, ...]]
+    # string.lstrip = «text»[, «head»[, ...]]
+    # string.rstrip = «text»[, «tail»[, ...]]
+
+    string.strip = "  padded  "      # "padded"
+    string.strip = "//path//", "/"   # "path"
+
+Removes characters from both ends of the text, or from only the start or the
+end.
+The arguments after the text form a set of utf-8 characters to remove.
+Without them, whitespace is removed.
+
+## string.map
+
+    # string.map = «text», {«old», «new»}[, ...]
+
+    string.map = (d.state), {0, "stopped"}, {1, "started"}
+
+Returns the replacement of the first pair whose `old` value equals the whole
+text.
+If no pair matches, the text is returned unchanged.
+
+## string.replace
+
+    # string.replace = «text», {«old», «new»}[, ...]
+
+    string.replace = "a-b-c", {"-", "+"}    # "a+b+c"
+
+Replaces every occurrence of `old` with `new`.
+The pairs are applied in order, so a later pair operates on the result of the
+earlier ones.
+
+## Example: dropping unwanted trackers
+
+The following disables every tracker that points at `retracker.local` as soon as
+a download is inserted.
+
+    method.set_key = event.download.inserted, drop_retracker, \
+      ((t.multicall, default, "branch=(string.contains,(t.url),retracker.local),((t.disable))"))

src/Makefile.am

@@ -194,6 +194,7 @@ libsub_root_a_SOURCES = \
 	command_throttle.cc \
 	command_tracker.cc \
 	command_scheduler.cc \
+	command_string.cc \
 	command_ui.cc \
 	control.cc \
 	control.h \

src/command_download.cc

@@ -655,6 +655,8 @@ initialize_command_download() {
   CMD2_DL("d.base_path.as_binary",            [](auto* download, auto) { return retrieve_d_base_path(download).object_as_binary(); });
   CMD2_DL("d.base_path.or_base64",            [](auto* download, auto) { return retrieve_d_base_path(download).object_utf8_or_base64(); });
   CMD2_DL("d.base_path.or_as_binary",         [](auto* download, auto) { return retrieve_d_base_path(download).object_utf8_or_as_binary(); });
+  CMD2_DL("d.base_path.realpath.or_empty",    [](auto* download, auto) { return resolve_path(retrieve_d_base_path(download).str()); });
+  CMD2_DL("d.base_path.realpath.or_throw",    [](auto* download, auto) { return resolve_path_or_throw(retrieve_d_base_path(download).str()); });
   CMD2_DL("d.base_filename",                  [](auto* download, auto) { return retrieve_d_base_filename(download).str(); });
   CMD2_DL("d.base_filename.hex",              [](auto* download, auto) { return retrieve_d_base_filename(download).object_hex(); });
   CMD2_DL("d.base_filename.base64",           [](auto* download, auto) { return retrieve_d_base_filename(download).object_base64(); });
@@ -771,6 +773,11 @@ initialize_command_download() {
   CMD2_DL_VAR_STRING_PUBLIC("d.tied_to_file", "rtorrent", "tied_to_file");
   CMD2_DL_VAR_STRING("d.loaded_file",  "rtorrent", "loaded_file");
 
+  CMD2_DL("d.tied_to_file.realpath.or_empty", [](auto* download, auto) { return resolve_path(rpc::convert_to_string(download_get_variable(download, "rtorrent", "tied_to_file"))); });
+  CMD2_DL("d.tied_to_file.realpath.or_throw", [](auto* download, auto) { return resolve_path_or_throw(rpc::convert_to_string(download_get_variable(download, "rtorrent", "tied_to_file"))); });
+  CMD2_DL("d.loaded_file.realpath.or_empty",  [](auto* download, auto) { return resolve_path(rpc::convert_to_string(download_get_variable(download, "rtorrent", "loaded_file"))); });
+  CMD2_DL("d.loaded_file.realpath.or_throw",  [](auto* download, auto) { return resolve_path_or_throw(rpc::convert_to_string(download_get_variable(download, "rtorrent", "loaded_file"))); });
+
   // The "state_changed" variable is required to be a valid unix time
   // value, it indicates the last time the torrent changed its state,
   // resume/pause.
@@ -878,6 +885,8 @@ initialize_command_download() {
   CMD2_DL_VALUE_V ("d.tracker.send_scrape",           [](auto download, uint64_t arg) { download->tracker_controller().scrape_request(arg); });
 
   CMD2_DL         ("d.directory",          CMD2_ON_FL(root_dir));
+  CMD2_DL         ("d.directory.realpath.or_empty", [](auto* download, auto) { return resolve_path(download->file_list()->root_dir()); });
+  CMD2_DL         ("d.directory.realpath.or_throw", [](auto* download, auto) { return resolve_path_or_throw(download->file_list()->root_dir()); });
   CMD2_DL_STRING_V("d.directory.set",      std::bind(&apply_d_directory, std::placeholders::_1, std::placeholders::_2));
   CMD2_DL         ("d.directory_base",     CMD2_ON_FL(root_dir));
   CMD2_DL_STRING_V("d.directory_base.set", std::bind(&core::Download::set_root_directory, std::placeholders::_1, std::placeholders::_2));
@@ -912,6 +921,14 @@ initialize_command_download() {
   rpc::rpc.mark_safe("d.local_id_html");
   rpc::rpc.mark_safe("d.bitfield");
   rpc::rpc.mark_safe("d.base_path");
+  rpc::rpc.mark_safe("d.base_path.realpath.or_empty");
+  rpc::rpc.mark_safe("d.base_path.realpath.or_throw");
+  rpc::rpc.mark_safe("d.directory.realpath.or_empty");
+  rpc::rpc.mark_safe("d.directory.realpath.or_throw");
+  rpc::rpc.mark_safe("d.tied_to_file.realpath.or_empty");
+  rpc::rpc.mark_safe("d.tied_to_file.realpath.or_throw");
+  rpc::rpc.mark_safe("d.loaded_file.realpath.or_empty");
+  rpc::rpc.mark_safe("d.loaded_file.realpath.or_throw");
   rpc::rpc.mark_safe("d.base_path.hex");
   rpc::rpc.mark_safe("d.base_path.base64");
   rpc::rpc.mark_safe("d.base_path.base64_as_binary");

src/command_file.cc

@@ -116,6 +116,8 @@ initialize_command_file() {
   CMD2_FILE("f.frozen_path.as_binary",            [](auto* file, auto)   { return file->frozen_path().object_as_binary(); });
   CMD2_FILE("f.frozen_path.or_base64",            [](auto* file, auto)   { return file->frozen_path().object_utf8_or_base64(); });
   CMD2_FILE("f.frozen_path.or_as_binary",         [](auto* file, auto)   { return file->frozen_path().object_utf8_or_as_binary(); });
+  CMD2_FILE("f.frozen_path.realpath.or_empty",    [](auto* file, auto)   { return resolve_path(file->frozen_path().str()); });
+  CMD2_FILE("f.frozen_path.realpath.or_throw",    [](auto* file, auto)   { return resolve_path_or_throw(file->frozen_path().str()); });
 
   CMD2_FILE("f.match_depth_prev",       std::bind(&torrent::File::match_depth_prev, std::placeholders::_1));
   CMD2_FILE("f.match_depth_next",       std::bind(&torrent::File::match_depth_next, std::placeholders::_1));
@@ -129,6 +131,8 @@ initialize_command_file() {
   rpc::rpc.mark_safe("f.path_components");
   rpc::rpc.mark_safe("f.path_depth");
   rpc::rpc.mark_safe("f.frozen_path");
+  rpc::rpc.mark_safe("f.frozen_path.realpath.or_empty");
+  rpc::rpc.mark_safe("f.frozen_path.realpath.or_throw");
   rpc::rpc.mark_safe("f.frozen_path.hex");
   rpc::rpc.mark_safe("f.frozen_path.base64");
   rpc::rpc.mark_safe("f.frozen_path.base64_as_binary");

src/command_helpers.cc

@@ -19,6 +19,7 @@ void initialize_command_groups();
 void initialize_command_throttle();
 void initialize_command_tracker();
 void initialize_command_scheduler();
+void initialize_command_string();
 void initialize_command_ui();
 
 void
@@ -37,4 +38,5 @@ initialize_commands() {
   initialize_command_throttle();
   initialize_command_tracker();
   initialize_command_scheduler();
+  initialize_command_string();
 }

src/command_local.cc

@@ -296,9 +296,13 @@ initialize_command_local() {
   CMD_VAR_BOOL    ("pieces.hash.on_completion",       true);
 
   CMD_VAR_STRING  ("directory.default",               "./");
+  CMD_ANY         ("directory.default.realpath.or_empty", [](auto, auto)    { return resolve_path(rpc::call_command_string("directory.default")); });
+  CMD_ANY         ("directory.default.realpath.or_throw", [](auto, auto)    { return resolve_path_or_throw(rpc::call_command_string("directory.default")); });
 
   CMD_VAR_STRING  ("session.name",                    "");
   CMD_ANY         ("session.path",                    [](auto, auto)        { return session_thread::manager()->path(); });
+  CMD_ANY         ("session.path.realpath.or_empty",  [](auto, auto)        { return resolve_path(session_thread::manager()->path()); });
+  CMD_ANY         ("session.path.realpath.or_throw",  [](auto, auto)        { return resolve_path_or_throw(session_thread::manager()->path()); });
   CMD_ANY_STRING_V("session.path.set",                [](auto, auto& str)   { return session_thread::manager()->set_path(str); });
   CMD_ANY         ("session.use_lock",                [](auto, auto)        { return session_thread::manager()->use_lock(); });
   CMD_ANY_VALUE_V ("session.use_lock.set",            [](auto, auto& value) { return session_thread::manager()->set_use_lock(value); });
@@ -372,6 +376,10 @@ initialize_command_local() {
 
   rpc::rpc.mark_safe("directory.default");
   rpc::rpc.mark_safe("session.path");
+  rpc::rpc.mark_safe("session.path.realpath.or_empty");
+  rpc::rpc.mark_safe("session.path.realpath.or_throw");
+  rpc::rpc.mark_safe("directory.default.realpath.or_empty");
+  rpc::rpc.mark_safe("directory.default.realpath.or_throw");
   rpc::rpc.mark_safe("session.use_lock");
   rpc::rpc.mark_safe("session.on_completion");
 

src/command_string.cc

@@ -0,0 +1,359 @@
+#include "config.h"
+
+#include <algorithm>
+#include <cctype>
+#include <iterator>
+#include <set>
+#include <string>
+#include <utility>
+#include <vector>
+#include <torrent/exceptions.h>
+#include <torrent/object.h>
+
+#include "rpc/parse.h"
+#include "rpc/rpc_manager.h"
+
+#include "globals.h"
+#include "control.h"
+#include "command_helpers.h"
+
+namespace {
+
+const std::string whitespace_characters = " \t\n\r\f\v";
+
+// The byte offset of every utf-8 character in 'text', terminated by the offset
+// past the last character. Bytes that are not valid utf-8 lead bytes are
+// treated as single characters.
+std::vector<size_t>
+utf8_offsets(const std::string& text) {
+  std::vector<size_t> offsets;
+
+  for (size_t i = 0; i < text.size(); i++)
+    if (i == 0 || (static_cast<unsigned char>(text[i]) & 0xC0) != 0x80)
+      offsets.push_back(i);
+
+  offsets.push_back(text.size());
+  return offsets;
+}
+
+int64_t
+utf8_length(const std::string& text) {
+  int64_t result = 0;
+
+  for (size_t i = 0; i < text.size(); i++)
+    if (i == 0 || (static_cast<unsigned char>(text[i]) & 0xC0) != 0x80)
+      result++;
+
+  return result;
+}
+
+std::string
+utf8_substr(const std::string& text, const std::vector<size_t>& offsets, size_t first, size_t last) {
+  return text.substr(offsets[first], offsets[last] - offsets[first]);
+}
+
+std::set<std::string>
+utf8_character_set(const std::string& text) {
+  auto offsets = utf8_offsets(text);
+  std::set<std::string> result;
+
+  for (size_t i = 0; i + 1 < offsets.size(); i++)
+    result.insert(utf8_substr(text, offsets, i, i + 1));
+
+  return result;
+}
+
+std::string
+ascii_lowercase(std::string text) {
+  std::transform(text.begin(), text.end(), text.begin(), [](unsigned char c) { return std::tolower(c); });
+  return text;
+}
+
+// A 'max_count' of zero means the command accepts any number of arguments.
+void
+verify_argument_count(const char* name, const torrent::Object::list_type& args, size_t min_count, size_t max_count) {
+  if (args.size() < min_count || (max_count != 0 && args.size() > max_count))
+    throw torrent::input_error(std::string(name) + ": invalid number of arguments.");
+}
+
+const torrent::Object&
+argument_at(const torrent::Object::list_type& args, size_t index) {
+  return *std::next(args.begin(), index);
+}
+
+void
+flatten_argument(const torrent::Object& arg, std::vector<std::string>* dest) {
+  if (!arg.is_list()) {
+    dest->push_back(rpc::convert_to_string(arg));
+    return;
+  }
+
+  for (const auto& child : arg.as_list())
+    flatten_argument(child, dest);
+}
+
+// The {old, new} pairs shared by 'string.map' and 'string.replace'.
+std::pair<std::string, std::string>
+argument_to_pair(const char* name, const torrent::Object& arg) {
+  if (!arg.is_list() || arg.as_list().size() != 2)
+    throw torrent::input_error(std::string(name) + ": arguments after the text must be {old, new} pairs.");
+
+  return {rpc::convert_to_string(arg.as_list().front()), rpc::convert_to_string(arg.as_list().back())};
+}
+
+// Compares the first argument against every remaining argument, returning true
+// as soon as one of them matches.
+torrent::Object
+apply_string_predicate(const char* name, const torrent::Object::list_type& args, bool (*predicate)(const std::string&, const std::string&)) {
+  verify_argument_count(name, args, 2, 0);
+
+  auto text = rpc::convert_to_string(args.front());
+
+  for (auto itr = std::next(args.begin()); itr != args.end(); itr++)
+    if (predicate(text, rpc::convert_to_string(*itr)))
+      return int64_t(1);
+
+  return int64_t(0);
+}
+
+bool
+text_equals(const std::string& text, const std::string& other) {
+  return text == other;
+}
+
+bool
+text_starts_with(const std::string& text, const std::string& prefix) {
+  return text.size() >= prefix.size() && text.compare(0, prefix.size(), prefix) == 0;
+}
+
+bool
+text_ends_with(const std::string& text, const std::string& tail) {
+  return text.size() >= tail.size() && text.compare(text.size() - tail.size(), tail.size(), tail) == 0;
+}
+
+bool
+text_contains(const std::string& text, const std::string& needle) {
+  return text.find(needle) != std::string::npos;
+}
+
+bool
+text_contains_i(const std::string& text, const std::string& needle) {
+  return ascii_lowercase(text).find(ascii_lowercase(needle)) != std::string::npos;
+}
+
+torrent::Object
+apply_string_length(const torrent::Object::list_type& args) {
+  verify_argument_count("string.length", args, 1, 1);
+
+  return utf8_length(rpc::convert_to_string(args.front()));
+}
+
+torrent::Object
+apply_string_substr(const torrent::Object::list_type& args) {
+  verify_argument_count("string.substr", args, 1, 4);
+
+  auto text        = rpc::convert_to_string(args.front());
+  auto offsets     = utf8_offsets(text);
+  auto text_length = static_cast<int64_t>(offsets.size() - 1);
+
+  auto position = args.size() > 1 ? rpc::convert_to_value(argument_at(args, 1)) : 0;
+  auto fallback = args.size() > 3 ? rpc::convert_to_string(argument_at(args, 3)) : std::string();
+
+  // Negative positions are relative to the end of the string.
+  if (position < 0)
+    position += text_length;
+
+  if (position < 0 || position > text_length)
+    return fallback;
+
+  auto last = text_length;
+
+  if (args.size() > 2) {
+    auto count = rpc::convert_to_value(argument_at(args, 2));
+
+    if (count < 0)
+      throw torrent::input_error("string.substr: the character count cannot be negative.");
+
+    last = std::min(text_length, position + std::min(count, text_length));
+  }
+
+  return utf8_substr(text, offsets, position, last);
+}
+
+torrent::Object
+apply_string_split(const torrent::Object::list_type& args) {
+  verify_argument_count("string.split", args, 2, 2);
+
+  auto text   = rpc::convert_to_string(args.front());
+  auto delim  = rpc::convert_to_string(args.back());
+  auto result = torrent::Object::create_list();
+
+  // An empty delimiter splits the text into its utf-8 characters.
+  if (delim.empty()) {
+    auto offsets = utf8_offsets(text);
+
+    for (size_t i = 0; i + 1 < offsets.size(); i++)
+      result.as_list().push_back(utf8_substr(text, offsets, i, i + 1));
+
+    return result;
+  }
+
+  size_t first = 0;
+
+  while (true) {
+    auto pos = text.find(delim, first);
+
+    if (pos == std::string::npos) {
+      result.as_list().push_back(text.substr(first));
+      return result;
+    }
+
+    result.as_list().push_back(text.substr(first, pos - first));
+    first = pos + delim.size();
+  }
+}
+
+torrent::Object
+apply_string_join(const torrent::Object::list_type& args) {
+  verify_argument_count("string.join", args, 1, 0);
+
+  auto delim = rpc::convert_to_string(args.front());
+
+  std::vector<std::string> parts;
+
+  for (auto itr = std::next(args.begin()); itr != args.end(); itr++)
+    flatten_argument(*itr, &parts);
+
+  std::string result;
+
+  for (auto itr = parts.begin(); itr != parts.end(); itr++) {
+    if (itr != parts.begin())
+      result += delim;
+
+    result += *itr;
+  }
+
+  return result;
+}
+
+torrent::Object
+apply_string_pad(const char* name, const torrent::Object::list_type& args, bool pad_start) {
+  verify_argument_count(name, args, 2, 3);
+
+  auto text     = rpc::convert_to_string(args.front());
+  auto pad_size = rpc::convert_to_value(argument_at(args, 1));
+  auto padding  = args.size() > 2 ? rpc::convert_to_string(argument_at(args, 2)) : std::string(" ");
+
+  auto text_length = utf8_length(text);
+
+  if (pad_size <= text_length || padding.empty())
+    return text;
+
+  auto padding_offsets = utf8_offsets(padding);
+  auto padding_length  = static_cast<int64_t>(padding_offsets.size() - 1);
+
+  std::string result;
+
+  for (int64_t i = 0; i < pad_size - text_length; i++) {
+    auto index = static_cast<size_t>(i % padding_length);
+    result += utf8_substr(padding, padding_offsets, index, index + 1);
+  }
+
+  return pad_start ? result + text : text + result;
+}
+
+torrent::Object
+apply_string_strip(const char* name, const torrent::Object::list_type& args, bool strip_start, bool strip_end) {
+  verify_argument_count(name, args, 1, 0);
+
+  auto text = rpc::convert_to_string(args.front());
+
+  std::string strippable;
+
+  for (auto itr = std::next(args.begin()); itr != args.end(); itr++)
+    strippable += rpc::convert_to_string(*itr);
+
+  if (args.size() == 1)
+    strippable = whitespace_characters;
+
+  auto characters = utf8_character_set(strippable);
+  auto offsets    = utf8_offsets(text);
+
+  size_t first = 0;
+  size_t last  = offsets.size() - 1;
+
+  while (strip_start && first < last && characters.count(utf8_substr(text, offsets, first, first + 1)) != 0)
+    first++;
+
+  while (strip_end && last > first && characters.count(utf8_substr(text, offsets, last - 1, last)) != 0)
+    last--;
+
+  return utf8_substr(text, offsets, first, last);
+}
+
+torrent::Object
+apply_string_map(const torrent::Object::list_type& args) {
+  verify_argument_count("string.map", args, 2, 0);
+
+  auto text = rpc::convert_to_string(args.front());
+
+  for (auto itr = std::next(args.begin()); itr != args.end(); itr++) {
+    auto pair = argument_to_pair("string.map", *itr);
+
+    if (text == pair.first)
+      return pair.second;
+  }
+
+  return text;
+}
+
+torrent::Object
+apply_string_replace(const torrent::Object::list_type& args) {
+  verify_argument_count("string.replace", args, 2, 0);
+
+  auto text = rpc::convert_to_string(args.front());
+
+  for (auto itr = std::next(args.begin()); itr != args.end(); itr++) {
+    auto pair = argument_to_pair("string.replace", *itr);
+
+    if (pair.first.empty())
+      throw torrent::input_error("string.replace: the replaced text cannot be empty.");
+
+    for (auto pos = text.find(pair.first); pos != std::string::npos; pos = text.find(pair.first, pos + pair.second.size()))
+      text.replace(pos, pair.first.size(), pair.second);
+  }
+
+  return text;
+}
+
+}
+
+void
+initialize_command_string() {
+  // clang-format off
+  CMD2_ANY_LIST("string.length",        [](auto, const auto& args) { return apply_string_length(args); });
+  CMD2_ANY_LIST("string.substr",     [](auto, const auto& args) { return apply_string_substr(args); });
+  CMD2_ANY_LIST("string.split",      [](auto, const auto& args) { return apply_string_split(args); });
+  CMD2_ANY_LIST("string.join",       [](auto, const auto& args) { return apply_string_join(args); });
+  CMD2_ANY_LIST("string.map",        [](auto, const auto& args) { return apply_string_map(args); });
+  CMD2_ANY_LIST("string.replace",    [](auto, const auto& args) { return apply_string_replace(args); });
+
+  CMD2_ANY_LIST("string.equals",     [](auto, const auto& args) { return apply_string_predicate("string.equals", args, &text_equals); });
+  CMD2_ANY_LIST("string.starts_with", [](auto, const auto& args) { return apply_string_predicate("string.starts_with", args, &text_starts_with); });
+  CMD2_ANY_LIST("string.ends_with",   [](auto, const auto& args) { return apply_string_predicate("string.ends_with", args, &text_ends_with); });
+  CMD2_ANY_LIST("string.contains",   [](auto, const auto& args) { return apply_string_predicate("string.contains", args, &text_contains); });
+  CMD2_ANY_LIST("string.contains_i", [](auto, const auto& args) { return apply_string_predicate("string.contains_i", args, &text_contains_i); });
+
+  CMD2_ANY_LIST("string.lpad",       [](auto, const auto& args) { return apply_string_pad("string.lpad", args, true); });
+  CMD2_ANY_LIST("string.rpad",       [](auto, const auto& args) { return apply_string_pad("string.rpad", args, false); });
+
+  CMD2_ANY_LIST("string.strip",      [](auto, const auto& args) { return apply_string_strip("string.strip", args, true, true); });
+  CMD2_ANY_LIST("string.lstrip",     [](auto, const auto& args) { return apply_string_strip("string.lstrip", args, true, false); });
+  CMD2_ANY_LIST("string.rstrip",     [](auto, const auto& args) { return apply_string_strip("string.rstrip", args, false, true); });
+  // clang-format on
+
+  for (const auto name : {"string.length", "string.substr", "string.split", "string.join", "string.map", "string.replace",
+                          "string.equals", "string.starts_with", "string.ends_with", "string.contains", "string.contains_i",
+                          "string.lpad", "string.rpad", "string.strip", "string.lstrip", "string.rstrip"})
+    rpc::rpc.mark_safe(name);
+}

src/display/utils.cc

@@ -345,7 +345,7 @@ print_status_info(char* first, char* last) {
   if (!torrent::up_throttle_global()->is_throttled()) {
     first = print_buffer(first, last, "[Throttle off");
   } else {
-    first = print_buffer(first, last, "[Throttle %3i", torrent::up_throttle_global()->max_rate() / 1024);
+    first = print_buffer(first, last, "[Throttle %3i", (int)(torrent::up_throttle_global()->max_rate() / 1024));
 
     if (!throttle_up_names.empty())
       first = print_status_throttle_limit(first, last, true, throttle_up_names);
@@ -354,7 +354,7 @@ print_status_info(char* first, char* last) {
   if (!torrent::down_throttle_global()->is_throttled()) {
     first = print_buffer(first, last, " / off KB]");
   } else {
-    first = print_buffer(first, last, " / %3i", torrent::down_throttle_global()->max_rate() / 1024);
+    first = print_buffer(first, last, " / %3i", (int)(torrent::down_throttle_global()->max_rate() / 1024));
 
     if (!throttle_down_names.empty())
       first = print_status_throttle_limit(first, last, false, throttle_down_names);

src/globals.cc

@@ -3,6 +3,7 @@
 #include "globals.h"
 
 #include <cstdlib>
+#include <stdlib.h>
 #include <torrent/exceptions.h>
 
 rpc::ip_table_list ip_tables;
@@ -28,3 +29,32 @@ expand_path(const std::string& path) {
 
   return path;
 }
+
+// Resolves a path to a canonical one with no symlinks or relative components,
+// so it can safely be handed to an external script. Returns an empty string if
+// the path does not name an existing file or directory.
+std::string
+resolve_path(const std::string& path) {
+  if (path.empty())
+    return std::string();
+
+  char* resolved = ::realpath(expand_path(path).c_str(), nullptr);
+
+  if (resolved == nullptr)
+    return std::string();
+
+  std::string result(resolved);
+  std::free(resolved);
+
+  return result;
+}
+
+std::string
+resolve_path_or_throw(const std::string& path) {
+  auto result = resolve_path(path);
+
+  if (result.empty())
+    throw torrent::input_error("Could not resolve path: '" + path + "'.");
+
+  return result;
+}

src/globals.h

@@ -11,6 +11,8 @@ extern rpc::ip_table_list ip_tables;
 extern Control*           control;
 
 std::string expand_path(const std::string& path);
+std::string resolve_path(const std::string& path);
+std::string resolve_path_or_throw(const std::string& path);
 
 namespace rpc {
 class SCgi;

src/rpc/xmlrpc_c.cc

@@ -451,6 +451,11 @@ XmlRpc::process(const char* inBuffer, uint32_t length, slot_write slotWrite) {
   if (local_env.fault_occurred && local_env.fault_code == XMLRPC_INTERNAL_ERROR)
     throw torrent::internal_error("Internal error in XMLRPC.");
 
+  if (memblock == nullptr) {
+    xmlrpc_env_clean(&local_env);
+    return false;
+  }
+
   bool result = slotWrite((const char*)xmlrpc_mem_block_contents(memblock),
                           xmlrpc_mem_block_size(memblock));
 

test/Makefile.am

@@ -51,6 +51,10 @@ rtorrent_Test_Src_SOURCES = $(rtorrent_Test_Common) \
 	src/test_command_dynamic.h \
 	src/test_command_local.cc \
 	src/test_command_local.h \
+	src/test_command_path.cc \
+	src/test_command_path.h \
+	src/test_command_string.cc \
+	src/test_command_string.h \
 	src/test_watch_ready_queue.cc \
 	src/test_watch_ready_queue.h
 

test/src/test_command_path.cc

@@ -0,0 +1,116 @@
+#include "config.h"
+
+#include "test/src/test_command_path.h"
+
+#include <cstdlib>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <torrent/exceptions.h>
+#include <torrent/torrent.h>
+
+#include "control.h"
+#include "globals.h"
+#include "rpc/parse_commands.h"
+
+CPPUNIT_TEST_SUITE_REGISTRATION(TestCommandPath);
+
+void initialize_command_local();
+
+void
+TestCommandPath::setUp() {
+  char temp_dir[] = "/tmp/rtorrent_test_path_XXXXXX";
+
+  CPPUNIT_ASSERT(mkdtemp(temp_dir) != nullptr);
+
+  // The temporary directory itself may sit behind a symlink, as /tmp does on
+  // macOS, so resolve it up front to keep the expected values exact.
+  m_temp_dir = resolve_path(temp_dir);
+
+  CPPUNIT_ASSERT(!m_temp_dir.empty());
+  CPPUNIT_ASSERT_EQUAL(0, mkdir((m_temp_dir + "/data").c_str(), 0755));
+  CPPUNIT_ASSERT_EQUAL(0, symlink((m_temp_dir + "/data").c_str(), (m_temp_dir + "/link").c_str()));
+}
+
+void
+TestCommandPath::tearDown() {
+  unlink((m_temp_dir + "/link").c_str());
+  rmdir((m_temp_dir + "/data").c_str());
+  rmdir(m_temp_dir.c_str());
+}
+
+void
+TestCommandPath::test_resolves_symlink() {
+  CPPUNIT_ASSERT_EQUAL(m_temp_dir + "/data", resolve_path(m_temp_dir + "/link"));
+
+  // A path that lies below a symlinked directory is resolved as well.
+  CPPUNIT_ASSERT_EQUAL(0, mkdir((m_temp_dir + "/data/below").c_str(), 0755));
+  CPPUNIT_ASSERT_EQUAL(m_temp_dir + "/data/below", resolve_path(m_temp_dir + "/link/below"));
+  rmdir((m_temp_dir + "/data/below").c_str());
+}
+
+void
+TestCommandPath::test_removes_relative_components() {
+  CPPUNIT_ASSERT_EQUAL(m_temp_dir, resolve_path(m_temp_dir + "/data/.."));
+  CPPUNIT_ASSERT_EQUAL(m_temp_dir + "/data", resolve_path(m_temp_dir + "/./data"));
+  CPPUNIT_ASSERT_EQUAL(m_temp_dir + "/data", resolve_path(m_temp_dir + "/data/"));
+
+  // Trailing slashes and duplicated separators collapse.
+  CPPUNIT_ASSERT_EQUAL(m_temp_dir + "/data", resolve_path(m_temp_dir + "//data//"));
+}
+
+void
+TestCommandPath::test_expands_tilde() {
+  const char* home = std::getenv("HOME");
+
+  if (home == nullptr || *home == '\0')
+    return;
+
+  CPPUNIT_ASSERT_EQUAL(resolve_path(home), resolve_path("~"));
+  CPPUNIT_ASSERT_THROW(resolve_path("~root/somewhere"), torrent::input_error);
+}
+
+void
+TestCommandPath::test_missing_path_throws() {
+  CPPUNIT_ASSERT_THROW(resolve_path_or_throw(""), torrent::input_error);
+  CPPUNIT_ASSERT_THROW(resolve_path_or_throw(m_temp_dir + "/does_not_exist"), torrent::input_error);
+  CPPUNIT_ASSERT_EQUAL(m_temp_dir + "/data", resolve_path_or_throw(m_temp_dir + "/link"));
+}
+
+void
+TestCommandPath::test_missing_path_is_empty() {
+  CPPUNIT_ASSERT_EQUAL(std::string(), resolve_path(""));
+  CPPUNIT_ASSERT_EQUAL(std::string(), resolve_path(m_temp_dir + "/does_not_exist"));
+  CPPUNIT_ASSERT_EQUAL(std::string(), resolve_path(m_temp_dir + "/does_not_exist/below"));
+
+  // A dangling symlink does not name an existing path either.
+  CPPUNIT_ASSERT_EQUAL(0, symlink((m_temp_dir + "/gone").c_str(), (m_temp_dir + "/dangling").c_str()));
+  CPPUNIT_ASSERT_EQUAL(std::string(), resolve_path(m_temp_dir + "/dangling"));
+  unlink((m_temp_dir + "/dangling").c_str());
+}
+
+void
+TestCommandPath::test_commands() {
+  torrent::initialize_main_thread();
+  torrent::initialize();
+
+  if (control == nullptr)
+    control = new Control;
+
+  if (!rpc::commands.has("directory.default.realpath.or_empty"))
+    initialize_command_local();
+
+  rpc::commands.call_command("directory.default.set", m_temp_dir + "/link");
+
+  CPPUNIT_ASSERT_EQUAL(m_temp_dir + "/link", rpc::commands.call_command("directory.default", torrent::Object()).as_string());
+  CPPUNIT_ASSERT_EQUAL(m_temp_dir + "/data", rpc::commands.call_command("directory.default.realpath.or_empty", torrent::Object()).as_string());
+  CPPUNIT_ASSERT_EQUAL(m_temp_dir + "/data", rpc::commands.call_command("directory.default.realpath.or_throw", torrent::Object()).as_string());
+
+  // A directory that has not been created yet resolves to nothing, and the
+  // or_throw variant reports it instead.
+  rpc::commands.call_command("directory.default.set", m_temp_dir + "/missing");
+
+  CPPUNIT_ASSERT_EQUAL(std::string(), rpc::commands.call_command("directory.default.realpath.or_empty", torrent::Object()).as_string());
+  CPPUNIT_ASSERT_THROW(rpc::commands.call_command("directory.default.realpath.or_throw", torrent::Object()), torrent::input_error);
+
+  torrent::cleanup();
+}

test/src/test_command_path.h

@@ -0,0 +1,30 @@
+#include "test/helpers/test_fixture.h"
+
+#include <string>
+
+class TestCommandPath : public test_fixture {
+  CPPUNIT_TEST_SUITE(TestCommandPath);
+
+  CPPUNIT_TEST(test_resolves_symlink);
+  CPPUNIT_TEST(test_removes_relative_components);
+  CPPUNIT_TEST(test_expands_tilde);
+  CPPUNIT_TEST(test_missing_path_is_empty);
+  CPPUNIT_TEST(test_missing_path_throws);
+  CPPUNIT_TEST(test_commands);
+
+  CPPUNIT_TEST_SUITE_END();
+
+public:
+  void setUp();
+  void tearDown();
+
+  void test_resolves_symlink();
+  void test_removes_relative_components();
+  void test_expands_tilde();
+  void test_missing_path_is_empty();
+  void test_missing_path_throws();
+  void test_commands();
+
+private:
+  std::string m_temp_dir;
+};

test/src/test_command_string.cc

@@ -0,0 +1,261 @@
+#include "config.h"
+
+#include "test/src/test_command_string.h"
+
+#include "rpc/parse_commands.h"
+
+CPPUNIT_TEST_SUITE_REGISTRATION(TestCommandString);
+
+void initialize_command_string();
+
+namespace {
+
+torrent::Object
+args(std::initializer_list<torrent::Object> objects) {
+  auto result = torrent::Object::create_list();
+
+  for (const auto& object : objects)
+    result.as_list().push_back(object);
+
+  return result;
+}
+
+std::string
+call_string(const char* key, std::initializer_list<torrent::Object> objects) {
+  return rpc::commands.call_command(key, args(objects)).as_string();
+}
+
+int64_t
+call_value(const char* key, std::initializer_list<torrent::Object> objects) {
+  return rpc::commands.call_command(key, args(objects)).as_value();
+}
+
+torrent::Object::list_type
+call_list(const char* key, std::initializer_list<torrent::Object> objects) {
+  return rpc::commands.call_command(key, args(objects)).as_list();
+}
+
+// Runs a command the way a line in the configuration file would.
+torrent::Object
+parse(const char* command) {
+  return rpc::parse_command_single(rpc::make_target(), command);
+}
+
+}
+
+void
+TestCommandString::setUp() {
+  if (!rpc::commands.has("string.length"))
+    initialize_command_string();
+}
+
+void
+TestCommandString::tearDown() {
+}
+
+void
+TestCommandString::test_length() {
+  CPPUNIT_ASSERT_EQUAL(int64_t(0), call_value("string.length", {""}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(3), call_value("string.length", {"abc"}));
+
+  // The length is counted in utf-8 characters, not bytes.
+  CPPUNIT_ASSERT_EQUAL(int64_t(5), call_value("string.length", {"héllo"}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(3), call_value("string.length", {"日本語"}));
+
+  // Values are converted to their string representation.
+  CPPUNIT_ASSERT_EQUAL(int64_t(4), call_value("string.length", {int64_t(1234)}));
+}
+
+void
+TestCommandString::test_equals() {
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), call_value("string.equals", {"abc", "abc"}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(0), call_value("string.equals", {"abc", "abd"}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(0), call_value("string.equals", {"abc", "ab"}));
+
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), call_value("string.equals", {"abc", "x", "abc"}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(0), call_value("string.equals", {"abc", "x", "y"}));
+
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), call_value("string.equals", {int64_t(42), "42"}));
+}
+
+void
+TestCommandString::test_starts_with() {
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), call_value("string.starts_with", {"abcdef", "abc"}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), call_value("string.starts_with", {"abcdef", ""}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(0), call_value("string.starts_with", {"abcdef", "bcd"}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(0), call_value("string.starts_with", {"ab", "abc"}));
+
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), call_value("string.starts_with", {"abcdef", "x", "ab"}));
+}
+
+void
+TestCommandString::test_ends_with() {
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), call_value("string.ends_with", {"abcdef", "def"}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), call_value("string.ends_with", {"abcdef", ""}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(0), call_value("string.ends_with", {"abcdef", "cde"}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(0), call_value("string.ends_with", {"ef", "def"}));
+
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), call_value("string.ends_with", {"a.torrent", ".rar", ".torrent"}));
+}
+
+void
+TestCommandString::test_contains() {
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), call_value("string.contains", {"abcdef", "cde"}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(0), call_value("string.contains", {"abcdef", "ace"}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), call_value("string.contains", {"abcdef", "x", "bcd"}));
+
+  CPPUNIT_ASSERT_EQUAL(int64_t(0), call_value("string.contains", {"retracker.local", "RETRACKER"}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), call_value("string.contains_i", {"retracker.local", "RETRACKER"}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), call_value("string.contains_i", {"RETRACKER.LOCAL", "retracker"}));
+  CPPUNIT_ASSERT_EQUAL(int64_t(0), call_value("string.contains_i", {"abcdef", "xyz"}));
+}
+
+void
+TestCommandString::test_substr() {
+  CPPUNIT_ASSERT_EQUAL(std::string("abcdef"), call_string("string.substr", {"abcdef"}));
+  CPPUNIT_ASSERT_EQUAL(std::string("cdef"), call_string("string.substr", {"abcdef", int64_t(2)}));
+  CPPUNIT_ASSERT_EQUAL(std::string("cde"), call_string("string.substr", {"abcdef", int64_t(2), int64_t(3)}));
+  CPPUNIT_ASSERT_EQUAL(std::string("cdef"), call_string("string.substr", {"abcdef", int64_t(2), int64_t(100)}));
+  CPPUNIT_ASSERT_EQUAL(std::string(""), call_string("string.substr", {"abcdef", int64_t(2), int64_t(0)}));
+
+  // Negative positions are relative to the end of the string.
+  CPPUNIT_ASSERT_EQUAL(std::string("ef"), call_string("string.substr", {"abcdef", int64_t(-2)}));
+  CPPUNIT_ASSERT_EQUAL(std::string("e"), call_string("string.substr", {"abcdef", int64_t(-2), int64_t(1)}));
+
+  // Out-of-bounds positions return the default value.
+  CPPUNIT_ASSERT_EQUAL(std::string(""), call_string("string.substr", {"abcdef", int64_t(10)}));
+  CPPUNIT_ASSERT_EQUAL(std::string("n/a"), call_string("string.substr", {"abcdef", int64_t(10), int64_t(1), "n/a"}));
+  CPPUNIT_ASSERT_EQUAL(std::string("n/a"), call_string("string.substr", {"abcdef", int64_t(-10), int64_t(1), "n/a"}));
+
+  // Positions and counts are in utf-8 characters, not bytes.
+  CPPUNIT_ASSERT_EQUAL(std::string("本"), call_string("string.substr", {"日本語", int64_t(1), int64_t(1)}));
+  CPPUNIT_ASSERT_EQUAL(std::string("本語"), call_string("string.substr", {"日本語", int64_t(-2)}));
+}
+
+void
+TestCommandString::test_split() {
+  auto parts = call_list("string.split", {"a,b,c", ","});
+
+  CPPUNIT_ASSERT_EQUAL(size_t(3), parts.size());
+  CPPUNIT_ASSERT_EQUAL(std::string("a"), parts.front().as_string());
+  CPPUNIT_ASSERT_EQUAL(std::string("c"), parts.back().as_string());
+
+  // Empty fields are preserved.
+  CPPUNIT_ASSERT_EQUAL(size_t(3), call_list("string.split", {"a,,b", ","}).size());
+  CPPUNIT_ASSERT_EQUAL(size_t(1), call_list("string.split", {"abc", ","}).size());
+
+  // A multi-character delimiter is matched as a whole.
+  CPPUNIT_ASSERT_EQUAL(size_t(2), call_list("string.split", {"a::b", "::"}).size());
+
+  // An empty delimiter splits the text into its utf-8 characters.
+  auto characters = call_list("string.split", {"日本語", ""});
+
+  CPPUNIT_ASSERT_EQUAL(size_t(3), characters.size());
+  CPPUNIT_ASSERT_EQUAL(std::string("日"), characters.front().as_string());
+}
+
+void
+TestCommandString::test_join() {
+  CPPUNIT_ASSERT_EQUAL(std::string("a, b"), call_string("string.join", {", ", "a", "b"}));
+  CPPUNIT_ASSERT_EQUAL(std::string("ab"), call_string("string.join", {"", "a", "b"}));
+  CPPUNIT_ASSERT_EQUAL(std::string(""), call_string("string.join", {", "}));
+  CPPUNIT_ASSERT_EQUAL(std::string("a"), call_string("string.join", {", ", "a"}));
+  CPPUNIT_ASSERT_EQUAL(std::string("1-2"), call_string("string.join", {"-", int64_t(1), int64_t(2)}));
+
+  // Lists are flattened, so the output of string.split can be joined again.
+  CPPUNIT_ASSERT_EQUAL(std::string("a-b-c"), call_string("string.join", {"-", args({"a", "b"}), "c"}));
+}
+
+void
+TestCommandString::test_pad() {
+  CPPUNIT_ASSERT_EQUAL(std::string("  7"), call_string("string.lpad", {"7", int64_t(3)}));
+  CPPUNIT_ASSERT_EQUAL(std::string("7  "), call_string("string.rpad", {"7", int64_t(3)}));
+  CPPUNIT_ASSERT_EQUAL(std::string("007"), call_string("string.lpad", {"7", int64_t(3), "0"}));
+  CPPUNIT_ASSERT_EQUAL(std::string("700"), call_string("string.rpad", {"7", int64_t(3), "0"}));
+
+  // Text that is already long enough is returned unchanged.
+  CPPUNIT_ASSERT_EQUAL(std::string("abcd"), call_string("string.lpad", {"abcd", int64_t(2)}));
+  CPPUNIT_ASSERT_EQUAL(std::string("abcd"), call_string("string.rpad", {"abcd", int64_t(4)}));
+
+  // Multi-character padding is repeated, and an empty padding is a no-op.
+  CPPUNIT_ASSERT_EQUAL(std::string("axyx"), call_string("string.rpad", {"a", int64_t(4), "xy"}));
+  CPPUNIT_ASSERT_EQUAL(std::string("a"), call_string("string.rpad", {"a", int64_t(4), ""}));
+
+  // Padding is counted in utf-8 characters, not bytes.
+  CPPUNIT_ASSERT_EQUAL(std::string("00日"), call_string("string.lpad", {"日", int64_t(3), "0"}));
+}
+
+void
+TestCommandString::test_strip() {
+  CPPUNIT_ASSERT_EQUAL(std::string("a b"), call_string("string.strip", {"  a b  "}));
+  CPPUNIT_ASSERT_EQUAL(std::string("a"), call_string("string.strip", {"\t\n a \r\n"}));
+  CPPUNIT_ASSERT_EQUAL(std::string("a  "), call_string("string.lstrip", {"  a  "}));
+  CPPUNIT_ASSERT_EQUAL(std::string("  a"), call_string("string.rstrip", {"  a  "}));
+
+  CPPUNIT_ASSERT_EQUAL(std::string("a"), call_string("string.strip", {"xxaxx", "x"}));
+  CPPUNIT_ASSERT_EQUAL(std::string("a"), call_string("string.strip", {"/x/a/x/", "/", "x"}));
+  CPPUNIT_ASSERT_EQUAL(std::string(""), call_string("string.strip", {"aaa", "a"}));
+  CPPUNIT_ASSERT_EQUAL(std::string("  a  "), call_string("string.strip", {"  a  ", ""}));
+
+  // The strippable argument is a set of utf-8 characters.
+  CPPUNIT_ASSERT_EQUAL(std::string("a"), call_string("string.strip", {"日a日", "日"}));
+}
+
+void
+TestCommandString::test_map() {
+  CPPUNIT_ASSERT_EQUAL(std::string("b"), call_string("string.map", {"a", args({"a", "b"})}));
+  CPPUNIT_ASSERT_EQUAL(std::string("c"), call_string("string.map", {"c", args({"a", "b"})}));
+
+  // Only whole-string matches are replaced.
+  CPPUNIT_ASSERT_EQUAL(std::string("ab"), call_string("string.map", {"ab", args({"a", "b"})}));
+
+  // The first matching pair wins.
+  CPPUNIT_ASSERT_EQUAL(std::string("y"), call_string("string.map", {"x", args({"a", "b"}), args({"x", "y"}), args({"x", "z"})}));
+}
+
+void
+TestCommandString::test_replace() {
+  CPPUNIT_ASSERT_EQUAL(std::string("a+b+c"), call_string("string.replace", {"a-b-c", args({"-", "+"})}));
+  CPPUNIT_ASSERT_EQUAL(std::string("abc"), call_string("string.replace", {"a-b-c", args({"-", ""})}));
+  CPPUNIT_ASSERT_EQUAL(std::string("a-b-c"), call_string("string.replace", {"a-b-c", args({"x", "y"})}));
+
+  // Pairs are applied in order, left to right.
+  CPPUNIT_ASSERT_EQUAL(std::string("xby"), call_string("string.replace", {"abc", args({"a", "x"}), args({"c", "y"})}));
+
+  // A replacement that contains the replaced text terminates.
+  CPPUNIT_ASSERT_EQUAL(std::string("aaaa"), call_string("string.replace", {"aa", args({"a", "aa"})}));
+}
+
+void
+TestCommandString::test_invalid_arguments() {
+  CPPUNIT_ASSERT_THROW(rpc::commands.call_command("string.length", torrent::Object()), torrent::input_error);
+  CPPUNIT_ASSERT_THROW(call_string("string.length", {"a", "b"}), torrent::input_error);
+  CPPUNIT_ASSERT_THROW(call_value("string.equals", {"a"}), torrent::input_error);
+  CPPUNIT_ASSERT_THROW(call_value("string.contains", {"a"}), torrent::input_error);
+  CPPUNIT_ASSERT_THROW(call_list("string.split", {"a"}), torrent::input_error);
+  CPPUNIT_ASSERT_THROW(call_string("string.lpad", {"a"}), torrent::input_error);
+
+  // The character count of string.substr cannot be negative.
+  CPPUNIT_ASSERT_THROW(call_string("string.substr", {"abc", int64_t(0), int64_t(-1)}), torrent::input_error);
+
+  // Both string.map and string.replace require {old, new} pairs.
+  CPPUNIT_ASSERT_THROW(call_string("string.map", {"a", "b"}), torrent::input_error);
+  CPPUNIT_ASSERT_THROW(call_string("string.replace", {"a", args({"a", "b", "c"})}), torrent::input_error);
+  CPPUNIT_ASSERT_THROW(call_string("string.replace", {"a", args({"", "b"})}), torrent::input_error);
+}
+
+void
+TestCommandString::test_config_syntax() {
+  CPPUNIT_ASSERT_EQUAL(int64_t(3), parse("string.length=abc").as_value());
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), parse("string.contains=retracker.local,retracker").as_value());
+  CPPUNIT_ASSERT_EQUAL(int64_t(1), parse("string.starts_with=udp://tracker.example.com,http://,udp://").as_value());
+  CPPUNIT_ASSERT_EQUAL(std::string("cde"), parse("string.substr=abcdef,2,3").as_string());
+  CPPUNIT_ASSERT_EQUAL(std::string("padded"), parse("string.strip=\"  padded  \"").as_string());
+
+  // The {old, new} pairs are written as a block in the configuration file.
+  CPPUNIT_ASSERT_EQUAL(std::string("a+b+c"), parse("string.replace=a-b-c,{-,+}").as_string());
+  CPPUNIT_ASSERT_EQUAL(std::string("y"), parse("string.map=x,{a,b},{x,y}").as_string());
+
+  // Commands nest, so a split can be joined back together.
+  CPPUNIT_ASSERT_EQUAL(std::string("a-b-c"), parse("string.join=-,(string.split,a.b.c,.)").as_string());
+}

test/src/test_command_string.h

@@ -0,0 +1,41 @@
+#include "test/helpers/test_fixture.h"
+
+class TestCommandString : public test_fixture {
+  CPPUNIT_TEST_SUITE(TestCommandString);
+
+  CPPUNIT_TEST(test_length);
+  CPPUNIT_TEST(test_equals);
+  CPPUNIT_TEST(test_starts_with);
+  CPPUNIT_TEST(test_ends_with);
+  CPPUNIT_TEST(test_contains);
+  CPPUNIT_TEST(test_substr);
+  CPPUNIT_TEST(test_split);
+  CPPUNIT_TEST(test_join);
+  CPPUNIT_TEST(test_pad);
+  CPPUNIT_TEST(test_strip);
+  CPPUNIT_TEST(test_map);
+  CPPUNIT_TEST(test_replace);
+  CPPUNIT_TEST(test_invalid_arguments);
+  CPPUNIT_TEST(test_config_syntax);
+
+  CPPUNIT_TEST_SUITE_END();
+
+public:
+  void setUp();
+  void tearDown();
+
+  void test_length();
+  void test_equals();
+  void test_starts_with();
+  void test_ends_with();
+  void test_contains();
+  void test_substr();
+  void test_split();
+  void test_join();
+  void test_pad();
+  void test_strip();
+  void test_map();
+  void test_replace();
+  void test_invalid_arguments();
+  void test_config_syntax();
+};