Merge pull request #86 from JesseMckinzie/bracket_fix

Fix bug when using escaped charactersx in filepattern

Author: sameeul

src/filepattern/cpp/internal/filepattern.cpp

@@ -1,4 +1,5 @@
 #include "filepattern.hpp"
+#include "../util/util.hpp"
 #include <chrono>
 
 using namespace std;
@@ -25,7 +26,7 @@ FilePatternObject::FilePatternObject(const string& path, const string& file_patt
         this->recursive_ = recursive; // Iterate over subdirectories
 
         // check if filepattern contains directory capturing 
-        if (file_pattern.find('/') != std::string::npos || file_pattern.find('\\') != std::string::npos) {
+        if (s::isPath(file_pattern)) {
             this->setCaptureDirectoryNames(true);
             this->recursive_ = true; // need to be recursive to capture directory names
 
@@ -76,6 +77,7 @@ void FilePatternObject::matchFilesOneDir(){
         auto length = end - start;
         throw invalid_argument("Invalid pattern found in bracket expressions in filepattern: \"" + this->getRegexFilePattern().substr(start, length+1) + "\"");
     }
+
     regex pattern_regex = regex(this->getRegexFilePattern());
     smatch sm;
 
@@ -115,7 +117,7 @@ void FilePatternObject::matchFilesMultDir(){
         }
 
         if(regex_match(file, sm, pattern_regex)){
-
+            
             if(this->getJustPath() || this->captureDirectoryNames()) {
                 tup = getVariableMap(file_path, sm);
             } else  {

src/filepattern/cpp/util/util.hpp

@@ -57,12 +57,35 @@ static const std::string SLASH = "/";
 #endif
 
 
+
+
 /**
  * @brief Functions for std::strings and Types.
  *
  */
 namespace s {
 
+    inline bool isPath(const std::string& str) {
+
+        if (str.find('/') != std::string::npos) {
+            return true;
+        }
+
+        // regex characters to escape
+        const std::unordered_set<char> escape_chars = {'*', '?', '^', '$', '(', ')', '[', ']', '|'};
+
+        for (auto i = 0; i < str.size()-1; ++i) {
+            if (str[i] == '\\' &&
+                std::find(std::begin(escape_chars), std::end(escape_chars), str[i+1]) == std::end(escape_chars)
+            ) {
+                return true; // Contains a slash that is not proceeded by a character being escaped
+            }
+        }
+
+        // no slashes were found that are not used for escaping characters
+        return false;
+    }
+
     inline std::string escapeForwardSlashes(const std::string& input) {
         std::string result;
 

tests/test_filepattern.py

@@ -290,6 +290,7 @@ class TestFilePattern():
     test_generate_filepattern_data.generate_data()
     test_generate_filepattern_data.generate_channel_data()
     test_generate_filepattern_data.generate_sorted_data()
+    test_generate_filepattern_data.generate_bracket_data()
 
     def test_file_pattern(self):
 
@@ -636,6 +637,25 @@ def test_file_pattern_sorting(self):
 
         assert sorted(indices) == indices
 
+    def test_file_pattern_brackets(self):
+
+        bracket_path = self.root_directory + '/test_data/bracket_data/'
+
+        pattern = "x\\(0-31\\)_y\\(01-48\\)_c{c:d}.ome.tif"
+
+        files = fp.FilePattern(bracket_path, pattern)
+
+        result = []
+
+        for file in files: # test iterator without call
+            result.append(file)
+
+        print(result)
+
+        for i in range(len(result)):
+            result[i][0]['c'] == i
+            os.path.basename(result[i][1][0]) == f'x(0-31)_y(01-48)_c{i}.ome.tif'
+
 
 # Todo: These tests need new data to be added after replacing the old version of filepattern.
 """

tests/test_generate_filepattern_data.py

@@ -138,12 +138,39 @@ def generate_text_data():
                 filename = f"img_r{r:03}_c{c:03}.tif"
                 file.write(filename + "\n")
 
+def generate_bracket_data():
+    directory = 'test_data'
+    root_directory = os.path.dirname(os.path.realpath(__file__))
+    path = os.path.join(root_directory, directory)
+    data_path = path + '/bracket_data'
+
+    try:
+        os.mkdir(path)
+        print('Data directory created at ' + path)
+    except FileExistsError:
+        print("Data directory already exists")
+
+    try:
+        os.mkdir(data_path)
+        print('Data directory created at ' + data_path)
+    except FileExistsError:
+        print("Data directory already exists")
+
+    for i in range(0, MAX):
+        data1 = 'x(0-31)_y(01-48)_c0.ome.tif'
+        data2 = 'x(0-31)_y(01-48)_c1.ome.tif'
+        f1 = open(data_path + '/' + data1, 'w+')
+        f1.close()
+        f2 = open(data_path + '/' + data2, 'w+')
+        f2.close()
+
 
 if __name__ == '__main__':
     generate_data()
     generate_channel_data()
     generate_sorted_data()
     generate_text_data()
+    generate_bracket_data()
 
 
 MAX = 3