From c505eaa4c53d6e12115c6edaeb9dbb4f03ba5f03 Mon Sep 17 00:00:00 2001
From: Nasty <bogatenkova.anastasiya@mail.ru>
Date: Fri, 16 Aug 2024 15:20:06 +0300
Subject: [PATCH] TLDR-748 finishing fixes

---
 dedoc/api/api_args.py                         |  2 +-
 dedoc/api/web/index.html                      | 26 ++++++++-
 .../default_structure_extractor.py            | 40 +++++++------
 .../structure_extractors/patterns/__init__.py |  3 +-
 .../patterns/abstract_pattern.py              | 45 ++------------
 .../patterns/bracket_list_pattern.py          |  6 +-
 .../patterns/bracket_roman_list_pattern.py    |  6 +-
 .../patterns/bullet_list_pattern.py           |  6 +-
 .../patterns/dotted_list_pattern.py           | 14 ++---
 .../patterns/letter_list_pattern.py           |  6 +-
 .../patterns/regexp_pattern.py                |  8 ++-
 .../patterns/roman_list_pattern.py            |  6 +-
 .../patterns/start_word_pattern.py            |  8 ++-
 .../patterns/tag_header_pattern.py            | 25 +++++++-
 .../patterns/tag_list_pattern.py              | 22 +++++--
 .../patterns/tag_pattern.py                   | 58 +++++++++++++++++++
 .../patterns/tag_type_pattern.py              | 18 ------
 dedoc/structure_extractors/patterns/utils.py  | 15 +++--
 dedoc/utils/parameter_utils.py                |  7 +++
 tests/api_tests/test_api_doctype_default.py   | 29 ++++++++--
 20 files changed, 215 insertions(+), 135 deletions(-)
 delete mode 100644 dedoc/structure_extractors/patterns/tag_type_pattern.py

diff --git a/dedoc/api/api_args.py b/dedoc/api/api_args.py
index cd75b6e5..8ffdc7b9 100644
--- a/dedoc/api/api_args.py
+++ b/dedoc/api/api_args.py
@@ -8,7 +8,7 @@
 class QueryParameters:
     # type of document structure parsing
     document_type: str = Form("", enum=["", "law", "tz", "diploma", "article", "fintoc"], description="Document domain")
-    patterns: str = Form(None, description='Patterns for default document type (when document_type="")')
+    patterns: str = Form("", description='Patterns for default document type (when document_type="")')
     structure_type: str = Form("tree", enum=["linear", "tree"], description="Output structure type")
     return_format: str = Form("json", enum=["json", "html", "plain_text", "tree", "collapsed_tree", "ujson", "pretty_json"],
                               description="Response representation, most types (except json) are used for debug purposes only")
diff --git a/dedoc/api/web/index.html b/dedoc/api/web/index.html
index 423dbcfe..d0c8b984 100644
--- a/dedoc/api/web/index.html
+++ b/dedoc/api/web/index.html
@@ -28,7 +28,7 @@ <h3>Parameters configuration</h3>
 
         <div class="parameters">
             <h4>Type of document structure parsing</h4>
-            <details><summary>document_type, structure_type, return_format</summary>
+            <details><summary>document_type, patterns, structure_type, return_format</summary>
                 <br>
                 <p>
                     <label>
@@ -43,6 +43,14 @@ <h4>Type of document structure parsing</h4>
                     </label>
                 </p>
 
+                <p>
+                    <div>
+                        Patterns for default structure extractor (document_type="other")<br>
+                        <label><textarea id="patterns" name="patterns" style="width:450px;height:75px;"></textarea></label><br>
+                        <button type="button" onclick="Format()">Format</button>
+                    </div>
+                </p>
+
                 <p>
                     <label>
                         <select name="structure_type">
@@ -114,7 +122,7 @@ <h4>Tables handling </h4>
 
         <div class="parameters">
             <h4>PDF handling</h4>
-            <details><summary>pdf_with_text_layer, language, pages, is_one_column_document, document_orientation, need_header_footer_analysis, need_binarization</summary>
+            <details><summary>pdf_with_text_layer, fast_textual_layer_detection, language, pages, is_one_column_document, document_orientation, need_header_footer_analysis, need_binarization</summary>
                 <br>
                 <p>
                     <label>
@@ -213,4 +221,18 @@ <h3>Useful links</h3>
     </ul>
 
 </body>
+
+<script>
+    function Format() {
+        try {
+            let input = document.getElementById("patterns")
+            let data = JSON.parse(input.value.replaceAll("\\", "\\\\"))
+            input.value = JSON.stringify(data, null, 2).replaceAll("\\\\", "\\")
+        }
+        catch (error) {
+            alert("Incorrect JSON syntax")
+        }
+    }
+</script>
+
 </html>
\ No newline at end of file
diff --git a/dedoc/structure_extractors/concrete_structure_extractors/default_structure_extractor.py b/dedoc/structure_extractors/concrete_structure_extractors/default_structure_extractor.py
index 8b4ac6cb..911780fb 100644
--- a/dedoc/structure_extractors/concrete_structure_extractors/default_structure_extractor.py
+++ b/dedoc/structure_extractors/concrete_structure_extractors/default_structure_extractor.py
@@ -46,32 +46,36 @@ def __get_patterns(self, parameters: dict) -> List[AbstractPattern]:
             from dedoc.structure_extractors.patterns.tag_header_pattern import TagHeaderPattern
             from dedoc.structure_extractors.patterns.tag_list_pattern import TagListPattern
             from dedoc.structure_extractors.patterns.tag_pattern import TagPattern
-            from dedoc.structure_extractors.patterns.tag_type_pattern import TagTypePattern
 
             return [
                 TagHeaderPattern(line_type=HierarchyLevel.header, level_1=1, can_be_multiline=False),
-                TagListPattern(line_type=HierarchyLevel.list_item, can_be_multiline=False),
-                DottedListPattern(line_type=HierarchyLevel.list_item, level_1=2, level_2=1, can_be_multiline=False),
+                TagListPattern(line_type=HierarchyLevel.list_item, default_level_1=2, default_level_2=1, can_be_multiline=False),
+                DottedListPattern(line_type=HierarchyLevel.list_item, level_1=2, can_be_multiline=False),
                 BracketListPattern(line_type=HierarchyLevel.list_item, level_1=3, level_2=1, can_be_multiline=False),
                 LetterListPattern(line_type=HierarchyLevel.list_item, level_1=4, level_2=1, can_be_multiline=False),
                 BulletListPattern(line_type=HierarchyLevel.list_item, level_1=5, level_2=1, can_be_multiline=False),
-                TagTypePattern(),
-                TagPattern(line_type=HierarchyLevel.raw_text)
+                TagPattern(default_line_type=HierarchyLevel.raw_text)
             ]
 
-        try:
-            import ast
-            from dedoc.structure_extractors.patterns.utils import get_pattern
+        import ast
+        from dedoc.structure_extractors.patterns.utils import get_pattern
 
-            patterns = parameters["patterns"]
-            if isinstance(patterns, str):
+        if isinstance(patterns, str):
+            try:
                 patterns = ast.literal_eval(patterns)
-            assert isinstance(patterns, list), "Patterns parameter should contain a list of patterns"
-            assert len(patterns) > 0, "Patterns parameter should contain a non-empty list of patterns"
-            if isinstance(patterns[0], dict):
-                patterns = [get_pattern(pattern) for pattern in patterns]
+            except ValueError as e:
+                raise StructureExtractorError(msg=f"Bad syntax for patterns: {str(e)}")
 
-            assert isinstance(patterns[0], AbstractPattern), "Patterns should be initialized properly"
-        except AssertionError as e:
-            raise StructureExtractorError(msg=str(e))
-        return patterns
+        if not isinstance(patterns, list):
+            raise StructureExtractorError(msg="Patterns parameter should contain a list of patterns")
+
+        pattern_classes = []
+        for pattern in patterns:
+            if isinstance(pattern, dict):
+                pattern_classes.append(get_pattern(pattern))
+            elif isinstance(pattern, AbstractPattern):
+                pattern_classes.append(pattern)
+            else:
+                raise StructureExtractorError(msg="Pattern should be dict or `AbstractPattern`")
+
+        return pattern_classes
diff --git a/dedoc/structure_extractors/patterns/__init__.py b/dedoc/structure_extractors/patterns/__init__.py
index 7f905e46..dad35f6c 100644
--- a/dedoc/structure_extractors/patterns/__init__.py
+++ b/dedoc/structure_extractors/patterns/__init__.py
@@ -9,7 +9,6 @@
 from dedoc.structure_extractors.patterns.tag_header_pattern import TagHeaderPattern
 from dedoc.structure_extractors.patterns.tag_list_pattern import TagListPattern
 from dedoc.structure_extractors.patterns.tag_pattern import TagPattern
-from dedoc.structure_extractors.patterns.tag_type_pattern import TagTypePattern
 
 __all__ = [BracketListPattern, BracketRomanListPattern, BulletListPattern, DottedListPattern, LetterListPattern, RegexpPattern, RomanListPattern,
-           StartWordPattern, TagHeaderPattern, TagListPattern, TagPattern, TagTypePattern]
+           StartWordPattern, TagHeaderPattern, TagListPattern, TagPattern]
diff --git a/dedoc/structure_extractors/patterns/abstract_pattern.py b/dedoc/structure_extractors/patterns/abstract_pattern.py
index c3d6f8b9..199eec22 100644
--- a/dedoc/structure_extractors/patterns/abstract_pattern.py
+++ b/dedoc/structure_extractors/patterns/abstract_pattern.py
@@ -8,15 +8,13 @@
 class AbstractPattern(ABC):
     _name = ""
 
-    def __init__(self,
-                 line_type: Optional[str] = None,
-                 level_1: Optional[int] = None,
-                 level_2: Optional[int] = None,
-                 can_be_multiline: Optional[bool] = None) -> None:
+    def __init__(self, line_type: Optional[str], level_1: Optional[int], level_2: Optional[int], can_be_multiline: Optional[bool or str]) -> None:
+        from dedoc.utils.parameter_utils import get_bool_value
+
         self._line_type = line_type
         self._level_1 = level_1
         self._level_2 = level_2
-        self._can_be_multiline = can_be_multiline
+        self._can_be_multiline = get_bool_value(can_be_multiline, default_value=True)
 
     @classmethod
     def name(cls: "AbstractPattern") -> str:
@@ -26,37 +24,6 @@ def name(cls: "AbstractPattern") -> str:
     def match(self, line: LineWithMeta) -> bool:
         pass
 
+    @abstractmethod
     def get_hierarchy_level(self, line: LineWithMeta) -> HierarchyLevel:
-        return HierarchyLevel(
-            line_type=self._get_line_type(line),
-            level_1=self._get_level_1(line),
-            level_2=self._get_level_2(line),
-            can_be_multiline=self._get_can_be_multiline(line)
-        )
-
-    def _get_line_type(self, line: LineWithMeta) -> str:
-        if self._line_type is not None:
-            return self._line_type
-
-        if line.metadata.tag_hierarchy_level is None:
-            raise ValueError(f"Cannot resolve line type: tag_hierarchy_level is missing and {self._name} line_type isn't configured")
-
-        return line.metadata.tag_hierarchy_level.line_type
-
-    def _get_level_1(self, line: LineWithMeta) -> Optional[int]:
-        if self._level_1 is not None:
-            return self._level_1
-
-        return line.metadata.tag_hierarchy_level.level_1 if line.metadata.tag_hierarchy_level else None
-
-    def _get_level_2(self, line: LineWithMeta) -> Optional[int]:
-        if self._level_2 is not None:
-            return self._level_2
-
-        return line.metadata.tag_hierarchy_level.level_2 if line.metadata.tag_hierarchy_level else None
-
-    def _get_can_be_multiline(self, line: LineWithMeta) -> bool:
-        if self._can_be_multiline is not None:
-            return self._can_be_multiline
-
-        return line.metadata.tag_hierarchy_level.can_be_multiline if line.metadata.tag_hierarchy_level else True
+        pass
diff --git a/dedoc/structure_extractors/patterns/bracket_list_pattern.py b/dedoc/structure_extractors/patterns/bracket_list_pattern.py
index 1c9e5a03..524d2fe1 100644
--- a/dedoc/structure_extractors/patterns/bracket_list_pattern.py
+++ b/dedoc/structure_extractors/patterns/bracket_list_pattern.py
@@ -7,9 +7,5 @@
 class BracketListPattern(RegexpPattern):
     _name = "bracket_list"
 
-    def __init__(self,
-                 line_type: Optional[str] = None,
-                 level_1: Optional[int] = None,
-                 level_2: Optional[int] = None,
-                 can_be_multiline: Optional[bool] = None) -> None:
+    def __init__(self, line_type: str, level_1: int, level_2: int, can_be_multiline: Optional[bool or str] = None) -> None:
         super().__init__(regexp=BracketPrefix.regexp, line_type=line_type, level_1=level_1, level_2=level_2, can_be_multiline=can_be_multiline)
diff --git a/dedoc/structure_extractors/patterns/bracket_roman_list_pattern.py b/dedoc/structure_extractors/patterns/bracket_roman_list_pattern.py
index 69f49b8b..35965cd6 100644
--- a/dedoc/structure_extractors/patterns/bracket_roman_list_pattern.py
+++ b/dedoc/structure_extractors/patterns/bracket_roman_list_pattern.py
@@ -7,9 +7,5 @@
 class BracketRomanListPattern(RegexpPattern):
     _name = "bracket_roman_list"
 
-    def __init__(self,
-                 line_type: Optional[str] = None,
-                 level_1: Optional[int] = None,
-                 level_2: Optional[int] = None,
-                 can_be_multiline: Optional[bool] = None) -> None:
+    def __init__(self, line_type: str, level_1: int, level_2: int, can_be_multiline: Optional[bool or str] = None) -> None:
         super().__init__(regexp=BracketRomanPrefix.regexp, line_type=line_type, level_1=level_1, level_2=level_2, can_be_multiline=can_be_multiline)
diff --git a/dedoc/structure_extractors/patterns/bullet_list_pattern.py b/dedoc/structure_extractors/patterns/bullet_list_pattern.py
index c8cabe0f..e0dacd4b 100644
--- a/dedoc/structure_extractors/patterns/bullet_list_pattern.py
+++ b/dedoc/structure_extractors/patterns/bullet_list_pattern.py
@@ -7,9 +7,5 @@
 class BulletListPattern(RegexpPattern):
     _name = "bullet_list"
 
-    def __init__(self,
-                 line_type: Optional[str] = None,
-                 level_1: Optional[int] = None,
-                 level_2: Optional[int] = None,
-                 can_be_multiline: Optional[bool] = None) -> None:
+    def __init__(self, line_type: str, level_1: int, level_2: int, can_be_multiline: Optional[bool or str] = None) -> None:
         super().__init__(regexp=BulletPrefix.regexp, line_type=line_type, level_1=level_1, level_2=level_2, can_be_multiline=can_be_multiline)
diff --git a/dedoc/structure_extractors/patterns/dotted_list_pattern.py b/dedoc/structure_extractors/patterns/dotted_list_pattern.py
index 1bb1aef8..ef4a071b 100644
--- a/dedoc/structure_extractors/patterns/dotted_list_pattern.py
+++ b/dedoc/structure_extractors/patterns/dotted_list_pattern.py
@@ -9,19 +9,15 @@
 class DottedListPattern(RegexpPattern):
     _name = "dotted_list"
 
-    def __init__(self,
-                 line_type: Optional[str] = None,
-                 level_1: Optional[int] = None,
-                 level_2: Optional[int] = None,
-                 can_be_multiline: Optional[bool] = None) -> None:
-        super().__init__(regexp=DottedPrefix.regexp, line_type=line_type, level_1=level_1, level_2=level_2, can_be_multiline=can_be_multiline)
+    def __init__(self, line_type: str, level_1: int, can_be_multiline: Optional[bool or str] = None) -> None:
+        super().__init__(regexp=DottedPrefix.regexp, line_type=line_type, level_1=level_1, level_2=None, can_be_multiline=can_be_multiline)
 
     def get_hierarchy_level(self, line: LineWithMeta) -> HierarchyLevel:
         return HierarchyLevel(
-            line_type=self._get_line_type(line),
-            level_1=self._get_level_1(line),
+            line_type=self._line_type,
+            level_1=self._level_1,
             level_2=self.__get_list_depth(line=line),
-            can_be_multiline=self._get_can_be_multiline(line)
+            can_be_multiline=self._can_be_multiline
         )
 
     def __get_list_depth(self, line: LineWithMeta) -> int:
diff --git a/dedoc/structure_extractors/patterns/letter_list_pattern.py b/dedoc/structure_extractors/patterns/letter_list_pattern.py
index eb1613a2..ae24ba73 100644
--- a/dedoc/structure_extractors/patterns/letter_list_pattern.py
+++ b/dedoc/structure_extractors/patterns/letter_list_pattern.py
@@ -7,9 +7,5 @@
 class LetterListPattern(RegexpPattern):
     _name = "letter_list"
 
-    def __init__(self,
-                 line_type: Optional[str] = None,
-                 level_1: Optional[int] = None,
-                 level_2: Optional[int] = None,
-                 can_be_multiline: Optional[bool] = None) -> None:
+    def __init__(self, line_type: str, level_1: int, level_2: int, can_be_multiline: Optional[bool or str] = None) -> None:
         super().__init__(regexp=AnyLetterPrefix.regexp, line_type=line_type, level_1=level_1, level_2=level_2, can_be_multiline=can_be_multiline)
diff --git a/dedoc/structure_extractors/patterns/regexp_pattern.py b/dedoc/structure_extractors/patterns/regexp_pattern.py
index 93cc41a3..b8fc2276 100644
--- a/dedoc/structure_extractors/patterns/regexp_pattern.py
+++ b/dedoc/structure_extractors/patterns/regexp_pattern.py
@@ -1,6 +1,7 @@
 import re
 from typing import Optional
 
+from dedoc.data_structures.hierarchy_level import HierarchyLevel
 from dedoc.data_structures.line_with_meta import LineWithMeta
 from dedoc.structure_extractors.patterns.abstract_pattern import AbstractPattern
 
@@ -10,10 +11,10 @@ class RegexpPattern(AbstractPattern):
 
     def __init__(self,
                  regexp: str or re.Pattern,
-                 line_type: Optional[str] = None,
+                 line_type: str,
                  level_1: Optional[int] = None,
                  level_2: Optional[int] = None,
-                 can_be_multiline: Optional[bool] = None) -> None:
+                 can_be_multiline: Optional[bool or str] = None) -> None:
         super().__init__(line_type=line_type, level_1=level_1, level_2=level_2, can_be_multiline=can_be_multiline)
         self._regexp = re.compile(regexp) if isinstance(regexp, str) else regexp
 
@@ -21,3 +22,6 @@ def match(self, line: LineWithMeta) -> bool:
         text = line.line.strip().lower()
         match = self._regexp.match(text)
         return match is not None
+
+    def get_hierarchy_level(self, line: LineWithMeta) -> HierarchyLevel:
+        return HierarchyLevel(line_type=self._line_type, level_1=self._level_1, level_2=self._level_2, can_be_multiline=self._can_be_multiline)
diff --git a/dedoc/structure_extractors/patterns/roman_list_pattern.py b/dedoc/structure_extractors/patterns/roman_list_pattern.py
index 425b990c..619ba420 100644
--- a/dedoc/structure_extractors/patterns/roman_list_pattern.py
+++ b/dedoc/structure_extractors/patterns/roman_list_pattern.py
@@ -7,9 +7,5 @@
 class RomanListPattern(RegexpPattern):
     _name = "roman_list"
 
-    def __init__(self,
-                 line_type: Optional[str] = None,
-                 level_1: Optional[int] = None,
-                 level_2: Optional[int] = None,
-                 can_be_multiline: Optional[bool] = None) -> None:
+    def __init__(self, line_type: str, level_1: int, level_2: int, can_be_multiline: Optional[bool or str] = None) -> None:
         super().__init__(regexp=RomanPrefix.regexp, line_type=line_type, level_1=level_1, level_2=level_2, can_be_multiline=can_be_multiline)
diff --git a/dedoc/structure_extractors/patterns/start_word_pattern.py b/dedoc/structure_extractors/patterns/start_word_pattern.py
index 3194bd70..a7b9b46f 100644
--- a/dedoc/structure_extractors/patterns/start_word_pattern.py
+++ b/dedoc/structure_extractors/patterns/start_word_pattern.py
@@ -1,5 +1,6 @@
 from typing import Optional
 
+from dedoc.data_structures.hierarchy_level import HierarchyLevel
 from dedoc.data_structures.line_with_meta import LineWithMeta
 from dedoc.structure_extractors.patterns.abstract_pattern import AbstractPattern
 
@@ -9,13 +10,16 @@ class StartWordPattern(AbstractPattern):
 
     def __init__(self,
                  start_word: str,
-                 line_type: Optional[str] = None,
+                 line_type: str,
                  level_1: Optional[int] = None,
                  level_2: Optional[int] = None,
-                 can_be_multiline: Optional[bool] = None) -> None:
+                 can_be_multiline: Optional[bool or str] = None) -> None:
         super().__init__(line_type=line_type, level_1=level_1, level_2=level_2, can_be_multiline=can_be_multiline)
         self.__start_word = start_word.strip().lower()
 
     def match(self, line: LineWithMeta) -> bool:
         text = line.line.strip().lower()
         return text.startswith(self.__start_word)
+
+    def get_hierarchy_level(self, line: LineWithMeta) -> HierarchyLevel:
+        return HierarchyLevel(line_type=self._line_type, level_1=self._level_1, level_2=self._level_2, can_be_multiline=self._can_be_multiline)
diff --git a/dedoc/structure_extractors/patterns/tag_header_pattern.py b/dedoc/structure_extractors/patterns/tag_header_pattern.py
index e390db58..927e148d 100644
--- a/dedoc/structure_extractors/patterns/tag_header_pattern.py
+++ b/dedoc/structure_extractors/patterns/tag_header_pattern.py
@@ -1,10 +1,31 @@
+from typing import Optional
+
 from dedoc.data_structures.hierarchy_level import HierarchyLevel
 from dedoc.data_structures.line_with_meta import LineWithMeta
-from dedoc.structure_extractors.patterns.abstract_pattern import AbstractPattern
+from dedoc.structure_extractors.patterns.tag_pattern import TagPattern
 
 
-class TagHeaderPattern(AbstractPattern):
+class TagHeaderPattern(TagPattern):
     _name = "tag_header"
 
+    def __init__(self,
+                 line_type: Optional[str] = None,
+                 level_1: Optional[int] = None,
+                 level_2: Optional[int] = None,
+                 can_be_multiline: Optional[bool or str] = None,
+                 default_line_type: str = HierarchyLevel.header,
+                 default_level_1: int = 1,
+                 default_level_2: Optional[int] = None) -> None:
+        super().__init__(line_type=line_type, level_1=level_1, level_2=level_2, can_be_multiline=can_be_multiline, default_line_type=default_line_type,
+                         default_level_1=default_level_1, default_level_2=default_level_2)
+
     def match(self, line: LineWithMeta) -> bool:
         return line.metadata.tag_hierarchy_level is not None and line.metadata.tag_hierarchy_level.line_type == HierarchyLevel.header
+
+    def get_hierarchy_level(self, line: LineWithMeta) -> HierarchyLevel:
+        return HierarchyLevel(
+            line_type=self._get_line_type(line),
+            level_1=self._get_level_1(line),
+            level_2=self._get_regexp_level_2(line),
+            can_be_multiline=self._get_can_be_multiline(line)
+        )
diff --git a/dedoc/structure_extractors/patterns/tag_list_pattern.py b/dedoc/structure_extractors/patterns/tag_list_pattern.py
index cd802ed2..44da867c 100644
--- a/dedoc/structure_extractors/patterns/tag_list_pattern.py
+++ b/dedoc/structure_extractors/patterns/tag_list_pattern.py
@@ -1,19 +1,31 @@
+from typing import Optional
+
 from dedoc.data_structures.hierarchy_level import HierarchyLevel
 from dedoc.data_structures.line_with_meta import LineWithMeta
-from dedoc.structure_extractors.patterns.abstract_pattern import AbstractPattern
+from dedoc.structure_extractors.patterns.tag_pattern import TagPattern
 
 
-class TagListPattern(AbstractPattern):
+class TagListPattern(TagPattern):
     _name = "tag_list"
 
+    def __init__(self,
+                 line_type: Optional[str] = None,
+                 level_1: Optional[int] = None,
+                 level_2: Optional[int] = None,
+                 can_be_multiline: Optional[bool or str] = None,
+                 default_line_type: str = HierarchyLevel.list_item,
+                 default_level_1: int = 2,
+                 default_level_2: Optional[int] = None) -> None:
+        super().__init__(line_type=line_type, level_1=level_1, level_2=level_2, can_be_multiline=can_be_multiline, default_line_type=default_line_type,
+                         default_level_1=default_level_1, default_level_2=default_level_2)
+
     def match(self, line: LineWithMeta) -> bool:
         return line.metadata.tag_hierarchy_level is not None and line.metadata.tag_hierarchy_level.line_type == HierarchyLevel.list_item
 
     def get_hierarchy_level(self, line: LineWithMeta) -> HierarchyLevel:
-        level_1, level_2 = self._get_level_1(line), self._get_level_2(line)
         return HierarchyLevel(
             line_type=self._get_line_type(line),
-            level_1=level_1 if level_1 is not None else 2,
-            level_2=level_2 if level_2 is not None else 1,
+            level_1=self._get_level_1(line),
+            level_2=self._get_regexp_level_2(line),
             can_be_multiline=self._get_can_be_multiline(line)
         )
diff --git a/dedoc/structure_extractors/patterns/tag_pattern.py b/dedoc/structure_extractors/patterns/tag_pattern.py
index 20ad469b..88adf384 100644
--- a/dedoc/structure_extractors/patterns/tag_pattern.py
+++ b/dedoc/structure_extractors/patterns/tag_pattern.py
@@ -1,3 +1,6 @@
+from typing import Optional
+
+from dedoc.data_structures.hierarchy_level import HierarchyLevel
 from dedoc.data_structures.line_with_meta import LineWithMeta
 from dedoc.structure_extractors.patterns.abstract_pattern import AbstractPattern
 
@@ -5,5 +8,60 @@
 class TagPattern(AbstractPattern):
     _name = "tag"
 
+    def __init__(self,
+                 line_type: Optional[str] = None,
+                 level_1: Optional[int] = None,
+                 level_2: Optional[int] = None,
+                 can_be_multiline: Optional[bool or str] = None,
+                 default_line_type: str = HierarchyLevel.raw_text,
+                 default_level_1: Optional[int] = None,
+                 default_level_2: Optional[int] = None) -> None:
+        super().__init__(line_type=line_type, level_1=level_1, level_2=level_2, can_be_multiline=can_be_multiline)
+        self._can_be_multiline_none = can_be_multiline is None
+        self._default_line_type = default_line_type
+        self._default_level_1 = default_level_1
+        self._default_level_2 = default_level_2
+
     def match(self, line: LineWithMeta) -> bool:
         return line.metadata.tag_hierarchy_level is not None
+
+    def get_hierarchy_level(self, line: LineWithMeta) -> HierarchyLevel:
+        return HierarchyLevel(
+            line_type=self._get_line_type(line),
+            level_1=self._get_level_1(line),
+            level_2=self._get_level_2(line),
+            can_be_multiline=self._get_can_be_multiline(line)
+        )
+
+    def _get_line_type(self, line: LineWithMeta) -> str:
+        if self._line_type is not None:
+            return self._line_type
+
+        return self._default_line_type if line.metadata.tag_hierarchy_level.is_unknown() else line.metadata.tag_hierarchy_level.line_type
+
+    def _get_level_1(self, line: LineWithMeta) -> Optional[int]:
+        if self._level_1 is not None:
+            return self._level_1
+
+        return self._default_level_1 if line.metadata.tag_hierarchy_level.level_1 is None else line.metadata.tag_hierarchy_level.level_1
+
+    def _get_level_2(self, line: LineWithMeta) -> Optional[int]:
+        if self._level_2 is not None:
+            return self._level_2
+
+        return self._default_level_2 if line.metadata.tag_hierarchy_level.level_2 is None else line.metadata.tag_hierarchy_level.level_2
+
+    def _get_regexp_level_2(self, line: LineWithMeta) -> int:
+        if self._level_2 is not None:
+            return self._level_2
+        elif line.metadata.tag_hierarchy_level.level_2 is not None:
+            return line.metadata.tag_hierarchy_level.level_2
+        elif self._default_level_2 is not None:
+            return self._default_level_2
+
+        from dedoc.structure_extractors.feature_extractors.list_features.list_utils import get_dotted_item_depth
+        depth = get_dotted_item_depth(line.line.strip())
+        return depth if depth > 0 else 1
+
+    def _get_can_be_multiline(self, line: LineWithMeta) -> bool:
+        return line.metadata.tag_hierarchy_level.can_be_multiline if self._can_be_multiline_none else self._can_be_multiline
diff --git a/dedoc/structure_extractors/patterns/tag_type_pattern.py b/dedoc/structure_extractors/patterns/tag_type_pattern.py
deleted file mode 100644
index a99f4974..00000000
--- a/dedoc/structure_extractors/patterns/tag_type_pattern.py
+++ /dev/null
@@ -1,18 +0,0 @@
-from dedoc.data_structures.hierarchy_level import HierarchyLevel
-from dedoc.data_structures.line_with_meta import LineWithMeta
-from dedoc.structure_extractors.patterns.abstract_pattern import AbstractPattern
-
-
-class TagTypePattern(AbstractPattern):
-    _name = "tag_type"
-
-    def match(self, line: LineWithMeta) -> bool:
-        return line.metadata.tag_hierarchy_level is not None and not line.metadata.tag_hierarchy_level.is_unknown()
-
-    def get_hierarchy_level(self, line: LineWithMeta) -> HierarchyLevel:
-        return HierarchyLevel(
-            line_type=line.metadata.tag_hierarchy_level.line_type,
-            level_1=self._get_level_1(line),
-            level_2=self._get_level_2(line),
-            can_be_multiline=self._get_can_be_multiline(line)
-        )
diff --git a/dedoc/structure_extractors/patterns/utils.py b/dedoc/structure_extractors/patterns/utils.py
index 6f79ac5b..e720a8a8 100644
--- a/dedoc/structure_extractors/patterns/utils.py
+++ b/dedoc/structure_extractors/patterns/utils.py
@@ -1,3 +1,5 @@
+from copy import deepcopy
+
 from dedoc.common.exceptions.structure_extractor_error import StructureExtractorError
 from dedoc.structure_extractors.patterns.abstract_pattern import AbstractPattern
 
@@ -5,16 +7,19 @@
 def get_pattern(pattern_parameters: dict) -> AbstractPattern:
     import dedoc.structure_extractors.patterns as patterns_module
 
-    assert isinstance(pattern_parameters, dict), "Pattern configuration must be a dict"
-    assert "name" in pattern_parameters, "Pattern parameter missing 'name'"
+    if "name" not in pattern_parameters:
+        raise StructureExtractorError(msg="Pattern parameter missing 'name'")
 
     supported_patterns = {pattern.name(): pattern for pattern in patterns_module.__all__}
     pattern_class = supported_patterns.get(pattern_parameters["name"])
-    assert pattern_class is not None, f"Pattern {pattern_parameters['name']} is not found in supported patterns: {supported_patterns.keys()}"
 
-    pattern_parameters.pop("name")
+    if pattern_class is None:
+        raise StructureExtractorError(msg=f"Pattern {pattern_parameters['name']} is not found in supported patterns: {supported_patterns.keys()}")
+
+    pattern_parameters_copy = deepcopy(pattern_parameters)
+    pattern_parameters_copy.pop("name")
     try:
-        pattern = pattern_class(**pattern_parameters)
+        pattern = pattern_class(**pattern_parameters_copy)
     except TypeError as e:
         raise StructureExtractorError(msg=str(e))
     return pattern
diff --git a/dedoc/utils/parameter_utils.py b/dedoc/utils/parameter_utils.py
index 461f3654..1b877c9f 100644
--- a/dedoc/utils/parameter_utils.py
+++ b/dedoc/utils/parameter_utils.py
@@ -6,6 +6,13 @@
 from dedoc.config import get_config
 
 
+def get_bool_value(parameter: Optional[bool or str], default_value: bool = False) -> bool:
+    if parameter is None:
+        return default_value
+
+    return parameter if isinstance(parameter, bool) else str(parameter).lower() == "true"
+
+
 def get_param_language(parameters: Optional[dict]) -> str:
     if parameters is None:
         return "rus+eng"
diff --git a/tests/api_tests/test_api_doctype_default.py b/tests/api_tests/test_api_doctype_default.py
index 40345629..fcf4a2c6 100644
--- a/tests/api_tests/test_api_doctype_default.py
+++ b/tests/api_tests/test_api_doctype_default.py
@@ -8,21 +8,40 @@ def test_patterns(self) -> None:
         patterns = [
             {"name": "regexp", "regexp": "^глава\s\d+\.", "line_type": "глава", "level_1": 1},  # noqa
             {"name": "start_word", "start_word": "статья", "level_1": 2, "line_type": "статья"},
-            {"name": "dotted_list", "level_1": 3, "line_type": "list_item"},
-            {"name": "bracket_list", "level_1": 4, "level_2": 1, "line_type": "list_item"}
+            {"name": "dotted_list", "level_1": 3, "line_type": "list_item", "can_be_multiline": False},
+            {"name": "bracket_list", "level_1": 4, "level_2": 1, "line_type": "bracket_list_item", "can_be_multiline": "false"}
         ]
         result = self._send_request(file_name, {"patterns": str(patterns)})
         structure = result["content"]["structure"]
 
         node = self._get_by_tree_path(structure, "0.1")
-        self.assertEqual(node["text"], "Глава 1. Общие положения")
+        self.assertEqual(node["text"].strip(), "Глава 1. Общие положения")
         self.assertEqual(node["metadata"]["paragraph_type"], "глава")
+        node = self._get_by_tree_path(structure, "0.1.1")
+        self.assertIn("Статья 1.1.", node["text"])
+        self.assertEqual(node["metadata"]["paragraph_type"], "статья")
+        node = self._get_by_tree_path(structure, "0.1.1.0")
+        self.assertEqual(node["metadata"]["paragraph_type"], "list")
+        node = self._get_by_tree_path(structure, "0.1.1.0.0")
+        self.assertIn("1. Законодательство", node["text"])
+        self.assertEqual(node["metadata"]["paragraph_type"], "list_item")
+        node = self._get_by_tree_path(structure, "0.1.2.0.0.0")
+        self.assertEqual(node["text"].strip(), "1) предупреждение;")
+        self.assertEqual(node["metadata"]["paragraph_type"], "bracket_list_item")
         node = self._get_by_tree_path(structure, "0.2")
-        self.assertEqual(node["text"], "Глава 2. Административные правонарушения, посягающие на права граждан и здоровье населения")
+        self.assertEqual(node["text"].strip(), "Глава 2. Административные правонарушения, посягающие на права граждан и здоровье населения")
         self.assertEqual(node["metadata"]["paragraph_type"], "глава")
 
+    def test_empty_patterns(self) -> None:
+        file_name = "docx/example.docx"
+        self._send_request(file_name, {"patterns": ""})
+        self._send_request(file_name, {"patterns": "[]"})
+
     def test_wrong_patterns(self) -> None:
         file_name = "docx/example.docx"
         self._send_request(file_name, {"patterns": str([{"regexp": "^глава\s\d+\.", "line_type": "глава", "level_1": 1}])}, expected_code=400)  # noqa
         self._send_request(file_name, {"patterns": str([{"name": "start_word", "line_type": "глава", "level_1": 1}])}, expected_code=400)
-        self._send_request(file_name, {"patterns": str([1])}, expected_code=400)
+        self._send_request(file_name, {"patterns": str([{"name": "unknown", "line_type": "глава", "level_1": 1}])}, expected_code=400)
+        self._send_request(file_name, {"patterns": "{1: blabla}"}, expected_code=400)
+        self._send_request(file_name, {"patterns": "{1: 2}"}, expected_code=400)
+        self._send_request(file_name, {"patterns": "[1]"}, expected_code=400)