Merge pull request #87 from JesseMckinzie/bug_fixes_1_10_24

bugfix: escaped parenthesis, space in directory name, and capturing directories with special characters

src/filepattern/cpp/internal/filepattern.cpp

@@ -13,7 +13,7 @@ FilePatternObject::FilePatternObject(const string& path, const string& file_patt
         this->getPathFromPattern(path); // set path and file_pattern
 
         try {
-            this->recursive_iterator_ = fs::recursive_directory_iterator(this->getPath());
+            this->recursive_iterator_ = fs::recursive_directory_iterator(fs::path(this->getPath()));
             this->recursive_ = true;
             this->setJustPath(true);
         } catch (const std::runtime_error& e) {
@@ -24,7 +24,8 @@ FilePatternObject::FilePatternObject(const string& path, const string& file_patt
     } else {
 
         this->recursive_ = recursive; // Iterate over subdirectories
-
+        this->setCaptureDirectoryNames(false);
+
         // check if filepattern contains directory capturing 
         if (s::isPath(file_pattern)) {
             this->setCaptureDirectoryNames(true);
@@ -37,12 +38,12 @@ FilePatternObject::FilePatternObject(const string& path, const string& file_patt
 
         this->setJustPath(false);
         this->setPath(path); // store path to target directory
-        
+
         try {
-            if(recursive){
-                this->recursive_iterator_ = fs::recursive_directory_iterator(this->getPath());
+            if(this->recursive_){
+                this->recursive_iterator_ = fs::recursive_directory_iterator(fs::path(this->getPath()));
             } else{
-                this->iterator_ = fs::directory_iterator(this->getPath()); // store iterator for target directory
+                this->iterator_ = fs::directory_iterator(fs::path(this->getPath())); // store iterator for target directory
             }
         } catch (const std::runtime_error& e) {
             string error = "No directory found. Invalid path \"" + path + "\".";
@@ -115,9 +116,9 @@ void FilePatternObject::matchFilesMultDir(){
         } else {
             file = s::getBaseName(file_path);
         }
-
+        
         if(regex_match(file, sm, pattern_regex)){
-            
+
             if(this->getJustPath() || this->captureDirectoryNames()) {
                 tup = getVariableMap(file_path, sm);
             } else  {

src/filepattern/cpp/util/util.hpp

@@ -75,8 +75,9 @@ namespace s {
         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)
+            if ((str[i] == '\\' &&
+                std::find(std::begin(escape_chars), std::end(escape_chars), str[i+1]) == std::end(escape_chars)) ||
+                str[i] == '/'
             ) {
                 return true; // Contains a slash that is not proceeded by a character being escaped
             }

tests/test_filepattern.py

@@ -291,6 +291,8 @@ class TestFilePattern():
     test_generate_filepattern_data.generate_channel_data()
     test_generate_filepattern_data.generate_sorted_data()
     test_generate_filepattern_data.generate_bracket_data()
+    test_generate_filepattern_data.generate_channel_data_sc()
+    test_generate_filepattern_data.generate_channel_data_spaces()
 
     def test_file_pattern(self):
 
@@ -579,7 +581,54 @@ def test_recursive_directory_regex_fp(self):
             assert fp_data.test_recursive_directory_fp[i][0]["directory"] == result[i][0]["directory"]
             assert str(os.path.basename(fp_data.test_recursive_directory_fp[i][1][0])) == os.path.basename(result[i][1][0])
 
+    def test_recursive_directory_regex_special_character_fp(self):
+
+        path = self.root_directory + '/test_data/recursive_data_sc'
+
+        filepattern = '/(?P<directory>.*)/img_r{r:ddd}_c{c:ddd}.tif'
+
+        files = fp.FilePattern(path, filepattern, recursive=True)
+
+        result = []
+
+        for file in files():
+            result.append(file)
 
+        # test that same number of files are returned
+        assert len(result) == len(fp_data.test_recursive_directory_fp) 
+
+        # test that each variable and path are equal for each file in list
+        for i in range(len(fp_data.test_recursive_directory_fp)): 
+            print(result[i])
+            assert fp_data.test_recursive_directory_fp[i][0]["r"] == result[i][0]["r"]
+            assert fp_data.test_recursive_directory_fp[i][0]["c"] == result[i][0]["c"]
+            assert fp_data.test_recursive_directory_fp[i][0]["directory"] + '_TEST' == result[i][0]["directory"]
+            assert str(os.path.basename(fp_data.test_recursive_directory_fp[i][1][0])) == os.path.basename(result[i][1][0])
+
+    def test_recursive_directory_spaces_fp(self):
+
+        path = self.root_directory + '/test_data/recursive_data_spaces/'
+
+        filepattern = 'img_r{r:ddd}_c{c:ddd}.tif'
+
+        files = fp.FilePattern(path, filepattern, recursive=True)
+
+        result = []
+
+        for file in files():
+            result.append(file)
+
+        print(result)
+
+        # test that same number of files are returned
+        assert len(result) == len(fp_data.test_recursive_space) 
+
+        # test that each variable and path are equal for each file in list
+        for i in range(len(fp_data.test_recursive_space)): 
+            print(result[i])
+            assert fp_data.test_recursive_space[i][0]["r"] == result[i][0]["r"]
+            assert fp_data.test_recursive_space[i][0]["c"] == result[i][0]["c"]
+            assert str(os.path.basename(fp_data.test_recursive_space[i][1][0])) == os.path.basename(result[i][1][0])
 
     def test_recursive_multi_directory_regex_fp(self):
 

tests/test_filepattern_data.py

@@ -647,3 +647,42 @@
 ({'c': 1, 'directory': 'TXREAD', 'r': 2}, ['/Users/jmckinzie/Documents/GitHub/filepattern-1/tests/test_data/recursive_data/TXREAD/img_r002_c001.tif']),
 ({'c': 2, 'directory': 'TXREAD', 'r': 2}, ['/Users/jmckinzie/Documents/GitHub/filepattern-1/tests/test_data/recursive_data/TXREAD/img_r002_c002.tif']),
 ]
+
+test_recursive_space = [
+    (
+        {'c': 0, 'r': 0},
+        ['img_r000_c000.tif', 'img_r000_c000.tif', 'img_r000_c000.tif']
+    ),
+    (
+        {'c': 1, 'r': 0},
+        ['img_r000_c001.tif', 'img_r000_c001.tif', 'img_r000_c001.tif']
+    ),
+    (
+        {'c': 2, 'r': 0},
+        ['img_r000_c002.tif', 'img_r000_c002.tif', 'img_r000_c002.tif']
+    ),
+    (
+        {'c': 0, 'r': 1},
+        ['img_r001_c000.tif', 'img_r001_c000.tif', 'img_r001_c000.tif']
+    ),
+    (
+        {'c': 1, 'r': 1},
+        ['img_r001_c001.tif', 'img_r001_c001.tif', 'img_r001_c001.tif']
+    ),
+    (
+        {'c': 2, 'r': 1},
+        ['img_r001_c002.tif', 'img_r001_c002.tif', 'img_r001_c002.tif']
+    ),
+    (
+        {'c': 0, 'r': 2},
+        ['img_r002_c000.tif', 'img_r002_c000.tif', 'img_r002_c000.tif']
+    ),
+    (
+        {'c': 1, 'r': 2},
+        ['img_r002_c001.tif', 'img_r002_c001.tif', 'img_r002_c001.tif']
+    ),
+    (
+        {'c': 2, 'r': 2},
+        ['img_r002_c002.tif', 'img_r002_c002.tif', 'img_r002_c002.tif']
+    )
+]

tests/test_generate_filepattern_data.py

@@ -100,6 +100,112 @@ def generate_channel_data():
 
     print("Files generated.")
 
+def generate_channel_data_sc():
+
+    MAX = 3
+
+    directory = 'test_data'
+    root_directory = os.path.dirname(os.path.realpath(__file__))
+    path = os.path.join(root_directory, directory)
+    data_path = path + '/data'
+    recursive_path = path + '/recursive_data_sc'
+
+    try:
+        os.mkdir(path)
+        os.mkdir(data_path)
+        os.mkdir(recursive_path)
+        print('Data directory created at ' + path)
+    except FileExistsError:
+        print("Data directory already exists")
+
+    try:
+        os.mkdir(data_path)
+        print('Data directory created at ' + path)
+    except FileExistsError:
+        print("Data directory already exists")
+
+    try:
+        os.mkdir(recursive_path)
+        print('Data directory created at ' + path)
+    except FileExistsError:
+        print("Data directory already exists")
+
+    channels = ['DAPI_TEST', 'TXREAD_TEST', 'GFP_TEST']
+
+    for channel in channels:
+        current_path = recursive_path + '/' + channel
+        try:
+            os.mkdir(current_path)
+        except FileExistsError:
+            print("Channel directory already exists")
+
+        for i in range(0, MAX):
+
+            str_i = str(i).zfill(3)
+
+            for j in range(0, MAX):
+
+                str_j = str(j).zfill(3)
+
+                recursive_name = 'img_r{}_c{}.tif'.format(str_i, str_j)
+                f = open(current_path + '/' + recursive_name, 'w+')
+                f.close()
+
+    print("Files generated.")
+
+def generate_channel_data_spaces():
+
+    MAX = 3
+
+    directory = 'test_data'
+    root_directory = os.path.dirname(os.path.realpath(__file__))
+    path = os.path.join(root_directory, directory)
+    data_path = path + '/data'
+    recursive_path = path + '/recursive_data_spaces'
+
+    try:
+        os.mkdir(path)
+        os.mkdir(data_path)
+        os.mkdir(recursive_path)
+        print('Data directory created at ' + path)
+    except FileExistsError:
+        print("Data directory already exists")
+
+    try:
+        os.mkdir(data_path)
+        print('Data directory created at ' + path)
+    except FileExistsError:
+        print("Data directory already exists")
+
+    try:
+        os.mkdir(recursive_path)
+        print('Data directory created at ' + path)
+    except FileExistsError:
+        print("Data directory already exists")
+
+    channels = ['DAPI TEST', 'TXREAD TEST', 'GFP TEST']
+
+    for channel in channels:
+        current_path = recursive_path + '/' + channel
+        try:
+            os.mkdir(current_path)
+        except FileExistsError:
+            print("Channel directory already exists")
+
+        for i in range(0, MAX):
+
+            str_i = str(i).zfill(3)
+
+            for j in range(0, MAX):
+
+                str_j = str(j).zfill(3)
+
+                recursive_name = 'img_r{}_c{}.tif'.format(str_i, str_j)
+                f = open(current_path + '/' + recursive_name, 'w+')
+                f.close()
+
+    print("Files generated.")
+
 def generate_sorted_data():
     MAX = 30
     length = 0
@@ -171,6 +277,8 @@ def generate_bracket_data():
     generate_sorted_data()
     generate_text_data()
     generate_bracket_data()
+    generate_channel_data_sc
+    generate_channel_data_spaces
 
 
 MAX = 3