totally reworked process argument tokenization, allow to use both single and double quotes and removing some other limitation

docs/source/features/scraping.md

@@ -19,8 +19,24 @@ However, if you choose `Script` option, then you cannot provide URL of your feed
 
 Any errors in your script must be written to [**error output** (stderr)](https://en.wikipedia.org/wiki/Standard_streams#Standard_error_(stderr)).
 
-```{warning}
-As of RSS Guard 4.2.0, you cannot separate your arguments with `#`. If your argument contains spaces, then enclose it with DOUBLE quotes, for example `"my argument"`. DO NOT use SINGLE quotes to do that.
+:::{warning}
+If your path to executable contains backslashes as directory separators, make sure to escape them with another backslash. Quote each individual argument with double quotes `"arg"` or single quotes `'arg'` and separate all arguments with spaces. You have to escape some characters inside double-quoted argument, for example double quote itself like this `"arg with \"quoted\" part"`.
+
+Examples (one per line):
+
+```
+C:\\MyFolder\\My.exe "arg1" "arg2" "my \"quoted\" arg3" 'my "quoted" arg4'
+
+bash "%data%/scripts/download-feed.sh"
+
+%data%\jq.exe '{ version: "1.1", title: "Stars", items: map( . | .title=.full_name | .content_text=.description | .date_published=.pushed_at)}'
+```
+:::
+
+RSS Guard offers [placeholder](userdata.md#data-placeholder) `%data%` which is automatically replaced with full path to RSS Guard user data folder and you can use this placeholder anywhere in your script call line.
+
+```{attention}
+Working directory of process executing the script is set to point to RSS Guard [user data](userdata) folder.
 ```
 
 Format of post-process script execution line can be seen on picture below.
@@ -31,25 +47,7 @@ If everything goes well, script must return `0` as the process exit code, or a n
 
 Executable file must be always be specified, while arguments do not. Be very careful when quoting arguments. Tested examples of valid execution lines are:
 
-| Command       | Explanation   |
-| :---          | ---           |
-| `bash -c "curl 'https://github.com/martinrotter.atom'"`   | Download ATOM feed file using Bash and Curl. |
-| `Powershell Invoke-WebRequest "https://github.com/martinrotter.atom" \| Select-Object -ExpandProperty Content` | Download ATOM feed file with Powershell. |
-| `php tweeper.php -v 0 "https://twitter.com/NSACareers"`     | Scrape Twitter RSS feed file with [Tweeper](https://git.ao2.it/tweeper.git). Tweeper is the utility that produces RSS feed from Twitter and other similar social platforms. |
-
-```{note}
-The above examples are cross-platform. You can use exactly the same command on Windows, Linux or macOS, if your operating system is properly configured.
-```
-
-RSS Guard offers [placeholder](userdata.md#data-placeholder) `%data%` which is automatically replaced with full path to RSS Guard user data folder, allowing you to make your configuration fully portable. You can, therefore, use something like this as a source script line: `bash %data%/scripts/download-feed.sh`.
-
-```{attention}
-Working directory of process executing the script is set to point to RSS Guard [user data](userdata) folder.
-```
-
-There are [examples of website scrapers](https://github.com/martinrotter/rssguard/tree/master/resources/scripts/scrapers). Most of them are written in Python 3, so their execution line is similar to `python script.py`. Make sure to examine each script for more information on how to use it.
-
-----
+## Dataflow
 After your source feed data is downloaded either via URL or custom script, you can optionally post-process it with one more custom script, which will take **raw source data as input**. It must produce valid feed data to standard output while printing all error messages to error output.
 
 Here is little flowchart explaining where and when scripts are used:
@@ -76,6 +74,10 @@ Typical post-processing filter might do things like CSS formatting, localization
 
 It's completely up to you if you decide to only use script as `Source` of the script or separate your custom functionality between `Source` script and `Post-process` script. Sometimes you might need different `Source` scripts for different online sources and the same `Post-process` script and vice versa.
 
+## Example Scrapers
+There are [examples of website scrapers](https://github.com/martinrotter/rssguard/tree/master/resources/scripts/scrapers). Most of them are written in Python 3, so their execution line is similar to `python "script.py"`. Make sure to examine each script for more information on how to use it.
+
+## 3rd-party Tools
 Third-party tools for scraping made to work with RSS Guard:
 * [CSS2RSS](https://github.com/Owyn/CSS2RSS) - can be used to scrape websites with CSS selectors.
 * [RSSGuardHelper](https://github.com/pipiscrew/RSSGuardHelper) - another CSS selectors helper.

src/librssguard/miscellaneous/textfactory.cpp

@@ -175,45 +175,166 @@ QString TextFactory::capitalizeFirstLetter(const QString& sts) {
   }
 }
 
-QStringList TextFactory::tokenizeProcessArguments(QStringView command) {
+enum class TokenState {
+  // We are not inside argument, we are between arguments.
+  Normal,
+
+  // We have detected escape "\" character coming from double-quoted argument.
+  EscapedFromDoubleQuotes,
+
+  // We have detected escape "\" character coming from spaced argument.
+  EscapedFromSpaced,
+
+  // We are inside argument which was separated by spaces.
+  InsideArgSpaced,
+
+  // We are inside argument.
+  InsideArgDoubleQuotes,
+
+  // We are inside argument, do not evaluate anything, just take it all
+  // as arw text.
+  InsideArgSingleQuotes
+};
+
+QStringList TextFactory::tokenizeProcessArguments(const QString& command) {
+  // Each argument containing spaces must be enclosed with single '' or double "" quotes.
+  // Some characters must be escaped with \ to keep their textual values as
+  // long as double-quoted argument is used.
+
+  if (command.isEmpty()) {
+    return {};
+  }
+
+  // We append space to end of command to make sure that
+  // ending space-separated argument is processed.
+  QString my_command = command + u' ';
+
+  TokenState state = TokenState::Normal;
   QStringList args;
-  QString tmp;
-  int quote_count = 0;
-  bool in_quote = false;
+  QString arg;
 
-  for (int i = 0; i < command.size(); ++i) {
-    if (command.at(i) == QL1C('"')) {
-      ++quote_count;
+  for (QChar chr : my_command) {
+    switch (state) {
+      case TokenState::Normal: {
+        switch (chr.unicode()) {
+          case u'"':
+            // We start double-quoted argument.
+            state = TokenState::InsideArgDoubleQuotes;
+            continue;
 
-      if (quote_count == 3) {
-        quote_count = 0;
-        tmp += command.at(i);
+          case u'\'':
+            // We start single-quoted argument.
+            state = TokenState::InsideArgSingleQuotes;
+            continue;
+
+          case u' ':
+            // Whitespace, just go on.
+            continue;
+
+          default:
+            // We found some actual text which marks
+            // beginning of argument, we assume spaced argument.
+            arg.append(chr);
+            state = TokenState::InsideArgSpaced;
+            continue;
+        }
+
+        break;
       }
 
-      continue;
-    }
-
-    if (quote_count) {
-      if (quote_count == 1) {
-        in_quote = !in_quote;
+      case TokenState::EscapedFromDoubleQuotes: {
+        // Previous character was "\".
+        arg.append(chr);
+        state = TokenState::InsideArgDoubleQuotes;
+        break;
       }
 
-      quote_count = 0;
-    }
-
-    if (!in_quote && command.at(i).isSpace()) {
-      if (!tmp.isEmpty()) {
-        args += tmp;
-        tmp.clear();
+      case TokenState::EscapedFromSpaced: {
+        // Previous character was "\".
+        arg.append(chr);
+        state = TokenState::InsideArgSpaced;
+        break;
+      }
+
+      case TokenState::InsideArgSpaced: {
+        switch (chr.unicode()) {
+          case u'\\':
+            // We found escaped!
+            state = TokenState::EscapedFromSpaced;
+            continue;
+
+          case u' ':
+            // We need to end this argument.
+            args.append(arg);
+            arg.clear();
+            state = TokenState::Normal;
+            continue;
+
+          default:
+            arg.append(chr);
+            break;
+        }
+
+        break;
+      }
+
+      case TokenState::InsideArgDoubleQuotes: {
+        switch (chr.unicode()) {
+          case u'\\':
+            // We found escaped!
+            state = TokenState::EscapedFromDoubleQuotes;
+            continue;
+
+          case u'"':
+            // We need to end this argument.
+            args.append(arg);
+            arg.clear();
+            state = TokenState::Normal;
+            continue;
+
+          default:
+            arg.append(chr);
+            break;
+        }
+
+        break;
+      }
+
+      case TokenState::InsideArgSingleQuotes: {
+        switch (chr.unicode()) {
+          case u'\'':
+            // We need to end this argument.
+            args.append(arg);
+            arg.clear();
+            state = TokenState::Normal;
+            continue;
+
+          default:
+            arg.append(chr);
+            break;
+        }
+
+        break;
       }
-    }
-    else {
-      tmp += command.at(i);
     }
   }
 
-  if (!tmp.isEmpty()) {
-    args += tmp;
+  switch (state) {
+    case TokenState::EscapedFromSpaced:
+    case TokenState::EscapedFromDoubleQuotes:
+      throw ApplicationException(QObject::tr("escape sequence not completed"));
+      break;
+
+    case TokenState::InsideArgDoubleQuotes:
+      throw ApplicationException(QObject::tr("closing \" is missing"));
+      break;
+
+    case TokenState::InsideArgSingleQuotes:
+      throw ApplicationException(QObject::tr("closing ' is missing"));
+      break;
+
+    default:
+      break;
   }
 
   return args;

src/librssguard/miscellaneous/textfactory.h

@@ -36,7 +36,7 @@ class TextFactory {
     static QString decrypt(const QString& text, quint64 key = 0);
     static QString newline();
     static QString capitalizeFirstLetter(const QString& sts);
-    static QStringList tokenizeProcessArguments(QStringView args);
+    static QStringList tokenizeProcessArguments(const QString& command);
 
     // Shortens input string according to given length limit.
     static QString shorten(const QString& input, int text_length_limit = TEXT_TITLE_LIMIT);

src/librssguard/services/standard/gui/standardfeeddetails.cpp

@@ -7,6 +7,7 @@
 #include "exceptions/networkexception.h"
 #include "exceptions/scriptexception.h"
 #include "miscellaneous/iconfactory.h"
+#include "miscellaneous/textfactory.h"
 #include "network-web/networkfactory.h"
 #include "services/abstract/category.h"
 #include "services/standard/definitions.h"
@@ -260,11 +261,12 @@ void StandardFeedDetails::onUrlChanged(const QString& new_url) {
     }
   }
   else if (sourceType() == StandardFeed::SourceType::Script) {
-    if (new_url.simplified().isEmpty()) {
-      m_ui.m_txtSource->setStatus(LineEditWithStatus::StatusType::Error, tr("The source is empty."));
+    try {
+      TextFactory::tokenizeProcessArguments(new_url);
+      m_ui.m_txtSource->setStatus(LineEditWithStatus::StatusType::Ok, tr("Source is ok."));
     }
-    else {
-      m_ui.m_txtSource->setStatus(LineEditWithStatus::StatusType::Ok, tr("The source is ok."));
+    catch (const ApplicationException& ex) {
+      m_ui.m_txtSource->setStatus(LineEditWithStatus::StatusType::Error, tr("Error: %1").arg(ex.message()));
     }
   }
   else {
@@ -273,11 +275,12 @@ void StandardFeedDetails::onUrlChanged(const QString& new_url) {
 }
 
 void StandardFeedDetails::onPostProcessScriptChanged(const QString& new_pp) {
-  if (QRegularExpression(QSL(SCRIPT_SOURCE_TYPE_REGEXP)).match(new_pp).hasMatch() || !new_pp.simplified().isEmpty()) {
+  try {
+    TextFactory::tokenizeProcessArguments(new_pp);
     m_ui.m_txtPostProcessScript->setStatus(LineEditWithStatus::StatusType::Ok, tr("Command is ok."));
   }
-  else {
-    m_ui.m_txtPostProcessScript->setStatus(LineEditWithStatus::StatusType::Ok, tr("Command is empty."));
+  catch (const ApplicationException& ex) {
+    m_ui.m_txtPostProcessScript->setStatus(LineEditWithStatus::StatusType::Error, tr("Error: %1").arg(ex.message()));
   }
 }