toolslm 0.3.0__py3-none-any.whl → 0.3.2__py3-none-any.whl

toolslm/__init__.py

@@ -1 +1 @@
-__version__ = "0.3.0"
+__version__ = "0.3.2"

toolslm/md_hier.py

@@ -1,65 +1,63 @@
 import re
 from fastcore.utils import *
-__all__ = ['markdown_to_dict', 'create_heading_dict']
+__all__ = ['create_heading_dict', 'HeadingDict']
 
-def markdown_to_dict(markdown_content):
-    def clean_heading(text): return re.sub(r'[^A-Za-z0-9 ]+', '', text).strip()
+class HeadingDict(dict):
+    """A dictionary-like object that also stores the markdown text content."""
+    def __init__(self, text="", *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.text = text
 
-    lines = markdown_content.splitlines()
+
+def create_heading_dict(text, rm_fenced=True):
+    "Create a nested dictionary structure from markdown headings."
+    original_text = text
+    original_lines = text.splitlines()
+    
+    # Use fenced-removed text only for finding headings
+    text_for_headings = text
+    if rm_fenced: text_for_headings = re.sub(r'```[\s\S]*?```', '', text)
+
+    lines_for_headings = text_for_headings.splitlines()
     headings = []
-    in_code_block = False
 
     # Parse headings with their levels and line numbers
-    for idx, line in enumerate(lines):
-        # Toggle code block state when encountering fence
-        if line.strip().startswith('```'): in_code_block = not in_code_block
-        
-        # Only detect headings when not in a code block
-        if in_code_block: continue
-        match = re.match(r'^(#{1,6})\s*(.*)', line)
+    for idx, line in enumerate(lines_for_headings):
+        match = re.match(r'^(#{1,6})\s+\S.*', line)
         if match:
             level = len(match.group(1))
-            text = match.group(2).strip()
-            headings.append({'level': level, 'text': text, 'line': idx})
+            title = line.strip('#').strip()
+            headings.append({'level': level, 'title': title, 'line': idx})
 
-    # Assign content to each heading, including subheadings
+    # Assign text content to each heading using original lines
     for i, h in enumerate(headings):
-        start = h['line']  # Include the heading line itself
+        start = h['line']
         # Find the end index: next heading of same or higher level
         for j in range(i + 1, len(headings)):
             if headings[j]['level'] <= h['level']:
                 end = headings[j]['line']
                 break
-        else: end = len(lines)
-        h['content'] = '\n'.join(lines[start:end]).strip()
+        else:
+            end = len(original_lines)
+        h['content'] = '\n'.join(original_lines[start:end]).strip()
 
-    # Build the dictionary with hierarchical keys
-    result,stack = {},[]
-    first_level = headings[0]['level']
-    for h in headings:
-        stack = stack[:h['level'] - first_level] + [clean_heading(h['text'])]
-        key = '.'.join(stack)
-        result[key] = h['content']
-    return dict2obj(result)
-
-def create_heading_dict(text):
-    text = re.sub(r'```[\s\S]*?```', '', text)
-    headings = re.findall(r'^#+.*', text, flags=re.MULTILINE)
-    result = {}
+    # Build the nested structure
+    result = HeadingDict(original_text)
     stack = [result]
-    prev_level = 0
+    stack_levels = [0]
 
-    for heading in headings:
-        level = heading.count('#')
-        title = heading.strip('#').strip()
-        while level <= prev_level:
+    for h in headings:
+        # Pop stack until we find the right parent level
+        while len(stack) > 1 and stack_levels[-1] >= h['level']:
             stack.pop()
-            prev_level -= 1
-        new_dict = {}
-        stack[-1][title] = new_dict
+            stack_levels.pop()
+
+        new_dict = HeadingDict(h['content'])
+        stack[-1][h['title']] = new_dict
         stack.append(new_dict)
-        prev_level = level
-    return dict2obj(result)
+        stack_levels.append(h['level'])
+
+    return result
 
 
 if __name__=='__main__':
@@ -91,63 +89,71 @@ Admin section.
 Admin users management.
 """
 
-    result = markdown_to_dict(md_content)
+    result = create_heading_dict(md_content)
     #for key, value in result.items(): print(f'Key: {key}\nValue:\n{value}\n{"-"*40}')
 
     def test_empty_content():
         md_content = "# Empty Heading"
-        result = markdown_to_dict(md_content)
-        assert result['Empty Heading'] == '# Empty Heading'
+        result = create_heading_dict(md_content)
+        assert 'Empty Heading' in result
+        assert result['Empty Heading'].text == '# Empty Heading'
+        assert result.text == md_content
 
     def test_special_characters():
         md_content = "# Heading *With* Special _Characters_!\nContent under heading."
-        result = markdown_to_dict(md_content)
-        assert 'Heading With Special Characters' in result
-        assert result['Heading With Special Characters'] == '# Heading *With* Special _Characters_!\nContent under heading.'
+        result = create_heading_dict(md_content)
+        assert 'Heading *With* Special _Characters_!' in result
+        assert result['Heading *With* Special _Characters_!'].text == '# Heading *With* Special _Characters_!\nContent under heading.'
+        assert result.text == md_content
 
     def test_duplicate_headings():
         md_content = "# Duplicate\n## Duplicate\n### Duplicate\nContent under duplicate headings."
-        result = markdown_to_dict(md_content)
+        result = create_heading_dict(md_content)
         assert 'Duplicate' in result
-        assert 'Duplicate.Duplicate' in result
-        assert 'Duplicate.Duplicate.Duplicate' in result
-        assert result['Duplicate.Duplicate.Duplicate'] == '### Duplicate\nContent under duplicate headings.'
+        assert 'Duplicate' in result['Duplicate']
+        assert 'Duplicate' in result['Duplicate']['Duplicate']
+        assert result['Duplicate']['Duplicate']['Duplicate'].text == '### Duplicate\nContent under duplicate headings.'
+        assert result.text == md_content
 
     def test_no_content():
         md_content = "# No Content Heading\n## Subheading"
-        result = markdown_to_dict(md_content)
-        assert result['No Content Heading'] == '# No Content Heading\n## Subheading'
-        assert result['No Content Heading.Subheading'] == '## Subheading'
+        result = create_heading_dict(md_content)
+        assert result['No Content Heading'].text == '# No Content Heading\n## Subheading'
+        assert result['No Content Heading']['Subheading'].text == '## Subheading'
+        assert result.text == md_content
 
     def test_different_levels():
         md_content = "### Level 3 Heading\nContent at level 3.\n# Level 1 Heading\nContent at level 1."
-        result = markdown_to_dict(md_content)
+        result = create_heading_dict(md_content)
         assert 'Level 3 Heading' in result
         assert 'Level 1 Heading' in result
-        assert result['Level 3 Heading'] == '### Level 3 Heading\nContent at level 3.'
-        assert result['Level 1 Heading'] == '# Level 1 Heading\nContent at level 1.'
+        assert result['Level 3 Heading'].text == '### Level 3 Heading\nContent at level 3.'
+        assert result['Level 1 Heading'].text == '# Level 1 Heading\nContent at level 1.'
+        assert result.text == md_content
 
     def test_parent_includes_subheadings():
         md_content = "# Parent\nParent content.\n## Child\nChild content.\n### Grandchild\nGrandchild content."
-        result = markdown_to_dict(md_content)
-        assert result['Parent'] == '# Parent\nParent content.\n## Child\nChild content.\n### Grandchild\nGrandchild content.'
-        assert result['Parent.Child'] == '## Child\nChild content.\n### Grandchild\nGrandchild content.'
-        assert result['Parent.Child.Grandchild'] == '### Grandchild\nGrandchild content.'
+        result = create_heading_dict(md_content)
+        assert result['Parent'].text == '# Parent\nParent content.\n## Child\nChild content.\n### Grandchild\nGrandchild content.'
+        assert result['Parent']['Child'].text == '## Child\nChild content.\n### Grandchild\nGrandchild content.'
+        assert result['Parent']['Child']['Grandchild'].text == '### Grandchild\nGrandchild content.'
+        assert result.text == md_content
 
     def test_multiple_level2_siblings():
-        md_content = "##Sib 1\n##Sib 2\n##Sib 3\n##Sib 4\n##Sib 5'"
-        result = markdown_to_dict(md_content)
+        md_content = "## Sib 1\n## Sib 2\n## Sib 3\n## Sib 4\n## Sib 5'"
+        result = create_heading_dict(md_content)
         assert 'Sib 1' in result
         assert 'Sib 2' in result
         assert 'Sib 3' in result
         assert 'Sib 4' in result
-        assert 'Sib 5' in result
-        
+        assert "Sib 5'" in result
+        assert result.text == md_content
+
     def test_code_chunks_escaped():
         md_content = "# Parent\nParent content.\n## Child\nChild content.\n```python\n# Code comment\nprint('Hello, world!')\n```"
-        result = markdown_to_dict(md_content)
-        assert 'Code comment' not in result
-        assert "# Code comment" in result['Parent.Child']
+        result = create_heading_dict(md_content)
+        assert 'Code comment' not in str(result)
+        assert result.text == md_content
 
     test_empty_content()
     test_special_characters()
@@ -159,7 +165,7 @@ Admin users management.
     test_code_chunks_escaped()
     print('tests passed')
 
-    def test_nested_headings():    
+    def test_nested_headings():
         md_content = "# Parent\nParent content.\n## Child\nChild content.\n### Grandchild\nGrandchild content."
         result = create_heading_dict(md_content)
         assert 'Child' in result['Parent']
@@ -169,7 +175,104 @@ Admin users management.
         md_content = "# Parent\nParent content.\n## Child\nChild content.\n```python\n# Code comment\nprint('Hello, world!')\n```"
         result = create_heading_dict(md_content)
         assert 'Code comment' not in result
-    
+
+    def test_fenced_blocks_preserved_in_text():
+        md_content = """# Section
+Content before code.
+
+```python
+# This heading should be ignored for structure
+def hello():
+    print("Hello, world!")
+```
+
+More content after code."""
+        result = create_heading_dict(md_content)
+        # Fenced code should be preserved in text content
+        assert '```python' in result['Section'].text
+        assert 'def hello():' in result['Section'].text
+        assert '```' in result['Section'].text
+        # But headings inside fenced blocks should not create structure
+        assert 'This heading should be ignored for structure' not in result['Section']
+
     test_nested_headings()
     test_code_chunks_escaped()
-    print('tests passed')
+    test_fenced_blocks_preserved_in_text()
+
+    def test_multiple_h1s():
+        md_content = "# First H1\n# Second H1\n# Third H1"
+        result = create_heading_dict(md_content)
+        assert 'First H1' in result
+        assert 'Second H1' in result
+        assert 'Third H1' in result
+        assert result['First H1'] == {}
+        assert result['Second H1'] == {}
+        assert result['Third H1'] == {}
+
+    def test_skip_levels_down():
+        md_content = "# Root\n## Level2\n#### Level4"
+        result = create_heading_dict(md_content)
+        assert 'Root' in result
+        assert 'Level2' in result['Root']
+        assert 'Level4' in result['Root']['Level2']
+
+    def test_skip_levels_up():
+        md_content = "# Root\n#### Deep\n## Back to 2"
+        result = create_heading_dict(md_content)
+        assert 'Root' in result
+        assert 'Deep' in result['Root']
+        assert 'Back to 2' in result['Root']
+        assert result['Root']['Deep'] == {}
+        assert result['Root']['Back to 2'] == {}
+
+    def test_non_h1_start():
+        md_content = "### Starting at 3\n## Going to 2\n# Finally 1"
+        result = create_heading_dict(md_content)
+        assert 'Starting at 3' in result
+        assert 'Going to 2' in result
+        assert 'Finally 1' in result
+
+    test_multiple_h1s()
+    test_skip_levels_down()
+    test_skip_levels_up()
+    test_non_h1_start()
+
+    # Edge case tests
+    def test_empty_input():
+        result = create_heading_dict("")
+        assert result == {}
+        assert result.text == ""
+
+    def test_whitespace_only():
+        result = create_heading_dict("   \n\t  \n   ")
+        assert result == {}
+        assert result.text == "   \n\t  \n   "
+
+    def test_malformed_headings():
+        # Too many #s (matches max 6)
+        md_content = "####### Too Many\nContent"
+        result = create_heading_dict(md_content)
+        assert 'Too Many' not in result
+        assert result.text == md_content
+
+    def test_unicode_and_emojis():
+        # Unicode characters
+        md_content = "# Café & Naïve\nContent with unicode\n## 中文标题\nChinese content"
+        result = create_heading_dict(md_content)
+        assert 'Café & Naïve' in result
+        assert '中文标题' in result['Café & Naïve']
+        assert result.text == md_content
+
+        # Emojis
+        md_content = "# 🚀 Rocket Heading\nRocket content\n## 💻 Computer\nComputer content"
+        result = create_heading_dict(md_content)
+        assert '🚀 Rocket Heading' in result
+        assert '💻 Computer' in result['🚀 Rocket Heading']
+        assert result.text == md_content
+
+    test_empty_input()
+    test_whitespace_only()
+    test_malformed_headings()
+    test_unicode_and_emojis()
+    print('tests passed')
+

{toolslm-0.3.0.dist-info → toolslm-0.3.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: toolslm
-Version: 0.3.0
+Version: 0.3.2
 Summary: Tools to make language models a bit easier to use
 Home-page: https://github.com/AnswerDotAI/toolslm
 Author: Jeremy Howard

toolslm-0.3.2.dist-info/RECORD

@@ -0,0 +1,13 @@
+toolslm/__init__.py,sha256=vNiWJ14r_cw5t_7UDqDQIVZvladKFGyHH2avsLpN7Vg,22
+toolslm/_modidx.py,sha256=-D-B5o30VGs11gBKf96lpADVXnZhdiVEshJpLzmUnDs,4378
+toolslm/download.py,sha256=g3BxUSxylC_575M7RFSJ1GI3Co3EwPDdEeWzxaf2Czk,4451
+toolslm/funccall.py,sha256=7nPfbcvDRMWiVKBKMLlCOMInoUJgDs5e38ef2T7QBHY,8485
+toolslm/md_hier.py,sha256=Havk9Hf0t2Xt67n_r7ZxCsS0pciR85iLcE5quShvkTg,10032
+toolslm/shell.py,sha256=dGInuRKvexu21VmtZkw_0S3BGiTsbAongUG-yG4YHpc,1566
+toolslm/xml.py,sha256=D665Nk7NzyZlXyXrpnIRqfK2xQ-6Gf0bCSgocjF7zik,4061
+toolslm-0.3.2.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+toolslm-0.3.2.dist-info/METADATA,sha256=5lWEv7BWTwdd5cvXgGsQXqr0j6tk8UIcGpRTlcjV3V4,2404
+toolslm-0.3.2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+toolslm-0.3.2.dist-info/entry_points.txt,sha256=xFz0Eymlo5X7BGpaO6DI9gMxvN5A7faebzrlr8ctp5I,95
+toolslm-0.3.2.dist-info/top_level.txt,sha256=4hRTrFWayz_Kz5221XjvlpCwVFrW3WPi1P0fllkTq9s,8
+toolslm-0.3.2.dist-info/RECORD,,

toolslm-0.3.0.dist-info/RECORD

@@ -1,13 +0,0 @@
-toolslm/__init__.py,sha256=VrXpHDu3erkzwl_WXrqINBm9xWkcyUy53IQOj042dOs,22
-toolslm/_modidx.py,sha256=-D-B5o30VGs11gBKf96lpADVXnZhdiVEshJpLzmUnDs,4378
-toolslm/download.py,sha256=g3BxUSxylC_575M7RFSJ1GI3Co3EwPDdEeWzxaf2Czk,4451
-toolslm/funccall.py,sha256=7nPfbcvDRMWiVKBKMLlCOMInoUJgDs5e38ef2T7QBHY,8485
-toolslm/md_hier.py,sha256=4uC12443tPBduYJgIZZIcEat2VG0x7JYC8-SwDdS2JY,6360
-toolslm/shell.py,sha256=dGInuRKvexu21VmtZkw_0S3BGiTsbAongUG-yG4YHpc,1566
-toolslm/xml.py,sha256=D665Nk7NzyZlXyXrpnIRqfK2xQ-6Gf0bCSgocjF7zik,4061
-toolslm-0.3.0.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-toolslm-0.3.0.dist-info/METADATA,sha256=JmN3o1_BAvUgUWi7q8j8uxpweyQzG6qTMt2u_NgASdU,2404
-toolslm-0.3.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-toolslm-0.3.0.dist-info/entry_points.txt,sha256=xFz0Eymlo5X7BGpaO6DI9gMxvN5A7faebzrlr8ctp5I,95
-toolslm-0.3.0.dist-info/top_level.txt,sha256=4hRTrFWayz_Kz5221XjvlpCwVFrW3WPi1P0fllkTq9s,8
-toolslm-0.3.0.dist-info/RECORD,,