markitect-main/tests/fixtures/markdown_samples.py

"""
Markdown document builders and sample generators for testing.
"""

from typing import Dict, List, Optional
import random
import string


class MarkdownDocumentBuilder:
    """Builder pattern for creating test markdown documents."""

    def __init__(self):
        self.content_parts: List[str] = []
        self.metadata: Dict[str, str] = {}

    def with_heading(self, text: str, level: int = 1) -> "MarkdownDocumentBuilder":
        """Add a heading to the document."""
        if level < 1 or level > 6:
            raise ValueError("Heading level must be between 1 and 6")

        heading_marker = "#" * level
        self.content_parts.append(f"{heading_marker} {text}")
        return self

    def with_paragraph(self, text: str) -> "MarkdownDocumentBuilder":
        """Add a paragraph to the document."""
        self.content_parts.append(text)
        return self

    def with_list(self, items: List[str], ordered: bool = False) -> "MarkdownDocumentBuilder":
        """Add a list to the document."""
        if ordered:
            list_items = [f"{i+1}. {item}" for i, item in enumerate(items)]
        else:
            list_items = [f"- {item}" for item in items]

        self.content_parts.append("\n".join(list_items))
        return self

    def with_code_block(self, code: str, language: str = "python") -> "MarkdownDocumentBuilder":
        """Add a code block to the document."""
        self.content_parts.append(f"```{language}\n{code}\n```")
        return self

    def with_link(self, text: str, url: str) -> "MarkdownDocumentBuilder":
        """Add a link to the document."""
        self.content_parts.append(f"[{text}]({url})")
        return self

    def with_metadata(self, key: str, value: str) -> "MarkdownDocumentBuilder":
        """Add metadata (front matter) to the document."""
        self.metadata[key] = value
        return self

    def with_table(self, headers: List[str], rows: List[List[str]]) -> "MarkdownDocumentBuilder":
        """Add a table to the document."""
        table_lines = []

        # Header row
        table_lines.append("| " + " | ".join(headers) + " |")

        # Separator row
        table_lines.append("| " + " | ".join(["-" * len(header) for header in headers]) + " |")

        # Data rows
        for row in rows:
            table_lines.append("| " + " | ".join(row) + " |")

        self.content_parts.append("\n".join(table_lines))
        return self

    def with_blockquote(self, text: str) -> "MarkdownDocumentBuilder":
        """Add a blockquote to the document."""
        quote_lines = [f"> {line}" for line in text.split("\n")]
        self.content_parts.append("\n".join(quote_lines))
        return self

    def build(self) -> str:
        """Build the final markdown document."""
        content = "\n\n".join(self.content_parts)

        if self.metadata:
            metadata_lines = [f"{k}: {v}" for k, v in self.metadata.items()]
            content = "---\n" + "\n".join(metadata_lines) + "\n---\n\n" + content

        return content


class LargeMarkdownGenerator:
    """Generator for creating large markdown documents for performance testing."""

    def __init__(self, seed: Optional[int] = None):
        self.random = random.Random(seed)

    def generate_document(self, size: str = "1mb") -> str:
        """Generate a large markdown document of specified size."""
        size_bytes = self._parse_size(size)
        builder = MarkdownDocumentBuilder()

        # Add metadata
        builder.with_metadata("title", "Large Test Document")
        builder.with_metadata("author", "Test Generator")
        builder.with_metadata("size", size)

        # Add content until we reach target size
        current_size = 0
        section_count = 0

        while current_size < size_bytes:
            section_count += 1
            section_title = f"Section {section_count}"
            builder.with_heading(section_title, level=2)

            # Add paragraphs
            for _ in range(self.random.randint(3, 8)):
                paragraph = self._generate_paragraph()
                builder.with_paragraph(paragraph)
                current_size += len(paragraph) + 2  # +2 for newlines

                if current_size >= size_bytes:
                    break

            # Add a list occasionally
            if self.random.random() < 0.3:
                items = [self._generate_sentence() for _ in range(self.random.randint(3, 7))]
                builder.with_list(items)
                current_size += sum(len(item) for item in items) + len(items) * 3  # Approximate

            # Add a code block occasionally
            if self.random.random() < 0.2:
                code = self._generate_code_block()
                builder.with_code_block(code)
                current_size += len(code) + 10  # +10 for code block markers

        return builder.build()

    def _parse_size(self, size: str) -> int:
        """Parse size string (e.g., '1mb', '500kb') to bytes."""
        size = size.lower()
        if size.endswith("kb"):
            return int(size[:-2]) * 1024
        elif size.endswith("mb"):
            return int(size[:-2]) * 1024 * 1024
        elif size.endswith("gb"):
            return int(size[:-2]) * 1024 * 1024 * 1024
        else:
            return int(size)

    def _generate_paragraph(self) -> str:
        """Generate a paragraph of random text."""
        sentences = []
        for _ in range(self.random.randint(3, 8)):
            sentences.append(self._generate_sentence())
        return " ".join(sentences)

    def _generate_sentence(self) -> str:
        """Generate a random sentence."""
        words = []
        for _ in range(self.random.randint(5, 15)):
            words.append(self._generate_word())

        sentence = " ".join(words).capitalize()
        return sentence + "."

    def _generate_word(self) -> str:
        """Generate a random word."""
        length = self.random.randint(3, 12)
        return "".join(self.random.choices(string.ascii_lowercase, k=length))

    def _generate_code_block(self) -> str:
        """Generate a random code block."""
        lines = []
        for _ in range(self.random.randint(5, 15)):
            line = self._generate_code_line()
            lines.append(line)
        return "\n".join(lines)

    def _generate_code_line(self) -> str:
        """Generate a line of code-like text."""
        templates = [
            "def {func_name}({params}):",
            "    return {expression}",
            "if {condition}:",
            "    {statement}",
            "# {comment}",
            "class {class_name}:",
            "    self.{attr} = {value}",
            "import {module}",
            "from {module} import {name}",
        ]

        template = self.random.choice(templates)
        variables = {
            "func_name": self._generate_word(),
            "params": ", ".join([self._generate_word() for _ in range(self.random.randint(0, 3))]),
            "expression": f"{self._generate_word()}({self._generate_word()})",
            "condition": f"{self._generate_word()} == {self.random.randint(1, 100)}",
            "statement": f"{self._generate_word()} = {self.random.randint(1, 100)}",
            "comment": " ".join([self._generate_word() for _ in range(self.random.randint(2, 6))]),
            "class_name": self._generate_word().capitalize(),
            "attr": self._generate_word(),
            "value": str(self.random.randint(1, 100)),
            "module": self._generate_word(),
            "name": self._generate_word(),
        }

        return template.format(**variables)


# Pre-built sample documents
SAMPLE_SIMPLE_DOCUMENT = """# Simple Document

This is a simple test document.

## Features

- Feature 1
- Feature 2
- Feature 3
"""

SAMPLE_COMPLEX_DOCUMENT = (
    MarkdownDocumentBuilder()
    .with_metadata("title", "Complex Test Document")
    .with_metadata("author", "Test Suite")
    .with_metadata("tags", "test, complex, sample")
    .with_heading("Complex Test Document")
    .with_paragraph("This is a complex test document with various markdown features.")
    .with_heading("Table of Contents", level=2)
    .with_list([
        "Introduction",
        "Features",
        "Examples",
        "Conclusion"
    ], ordered=True)
    .with_heading("Introduction", level=2)
    .with_paragraph("This document demonstrates various markdown features.")
    .with_blockquote("This is an important note about the document.")
    .with_heading("Features", level=2)
    .with_list([
        "**Bold text**",
        "*Italic text*",
        "`Code inline`",
        "[Links](https://example.com)"
    ])
    .with_heading("Code Example", level=3)
    .with_code_block('''def hello_world():
    """Print hello world message."""
    print("Hello, World!")
    return "success"''')
    .with_heading("Data Table", level=3)
    .with_table(
        ["Name", "Type", "Description"],
        [
            ["title", "string", "Document title"],
            ["author", "string", "Document author"],
            ["tags", "array", "Document tags"]
        ]
    )
    .with_heading("Conclusion", level=2)
    .with_paragraph("This document shows the power of markdown for documentation.")
    .build()
)

SAMPLE_TECHNICAL_DOCUMENT = (
    MarkdownDocumentBuilder()
    .with_metadata("title", "API Documentation")
    .with_metadata("version", "1.0.0")
    .with_metadata("category", "technical")
    .with_heading("API Documentation")
    .with_paragraph("This document describes the REST API endpoints.")
    .with_heading("Authentication", level=2)
    .with_paragraph("All API requests require authentication via API key.")
    .with_code_block('''curl -H "Authorization: Bearer YOUR_API_KEY" \\
     https://api.example.com/v1/endpoint''', "bash")
    .with_heading("Endpoints", level=2)
    .with_heading("GET /users", level=3)
    .with_paragraph("Retrieve a list of users.")
    .with_table(
        ["Parameter", "Type", "Required", "Description"],
        [
            ["limit", "integer", "No", "Maximum number of results"],
            ["offset", "integer", "No", "Number of results to skip"],
            ["filter", "string", "No", "Filter criteria"]
        ]
    )
    .with_heading("Response", level=4)
    .with_code_block('''{
  "users": [
    {
      "id": 1,
      "name": "John Doe",
      "email": "john@example.com"
    }
  ],
  "total": 1,
  "offset": 0,
  "limit": 10
}''', "json")
    .build()
)