=== FEATURE-INLINE-LINKS.md ===
# FEATURE: Inline Links and Images

| Field       | Value |
|-------------|-------|
| Version     | 20260812 V1 |
| Description | Defines inline links, reference links, images, destinations, titles, and reference resolution. |
| Depends On  | ARCHITECTURE.md, FEATURE-BLOCK-LEAF.md, FEATURE-INLINE-EMPHASIS.md |
| Provides    | link parser, image parser, reference resolver |
| Consumes    | block parser, inline parser |

## Questions

- None.

## Purpose

The inline parser recognizes inline, full-reference, collapsed-reference, and shortcut links and images. It resolves link reference definitions collected during block parsing, supports balanced destinations and titles, and prevents nested links while permitting nested image descriptions.

## Behavior

- Inline destinations may be empty, angle-bracket delimited, escaped, or contain balanced parentheses.
- Titles may use single quotes, double quotes, or parentheses.
- Reference labels are normalized with Unicode case folding and whitespace normalization.
- The first matching reference definition wins.
- Link text is parsed as inline content.
- Image descriptions produce plain-text `alt` content.
- Links cannot contain links; image descriptions may contain links.

## Programmatic Acceptance

=== AC inline-links-suite ===
Intent: The inline link, reference link, and image syntax passes its complete owned conformance scope.
Suite: scoped
Requires: executable=python3; scope=test

import subprocess
import sys

result = subprocess.run(
    [sys.executable, "spec_tests.py", "--spec", "spec.txt", "--program", "./commonmark",
     "--pattern", "^(Links|Images)$"],
    capture_output=True, text=True,
)
print(result.stdout)
print(result.stderr, file=sys.stderr)
assert result.returncode == 0
=== END AC inline-links-suite ===

=== AC inline-links-boundaries ===
Intent: The implementation preserves link and image destination, title, reference, and nesting behavior.
Suite: scoped
Requires: executable=python3; scope=test

import subprocess
import sys

result = subprocess.run(
    [sys.executable, "spec_tests.py", "--spec", "spec.txt", "--program", "./commonmark",
     "--pattern", "^(Links|Images)$"],
    capture_output=True, text=True,
)
assert result.returncode == 0
=== END AC inline-links-boundaries ===

## User Acceptance

- None.

## Guardrails

- The implementation does not use a public Markdown implementation.
- Reference definitions are resolved globally while preserving first-definition precedence.
=== END FEATURE-INLINE-LINKS.md ===
=== FEATURE-INLINE-HTML.md ===
# FEATURE: Autolinks and Raw HTML

| Field       | Value |
|-------------|-------|
| Version     | 20260812 V1 |
| Description | Defines URI and email autolinks and preservation of raw HTML inline constructs. |
| Depends On  | ARCHITECTURE.md, FEATURE-INLINE-LINKS.md |
| Provides    | autolink parser, raw HTML parser |
| Consumes    | block parser, inline parser |

## Questions

- None.

## Behavior

- Absolute URI autolinks become links whose labels preserve the source URI.
- Valid email autolinks become `mailto:` links.
- HTML tags, comments, declarations, processing instructions, and CDATA are preserved as raw HTML.
- Invalid autolinks and malformed HTML-like text remain escaped text.
- Backslash escapes do not apply inside autolinks or raw HTML.

## Programmatic Acceptance

=== AC inline-html-suite ===
Intent: The autolink and raw HTML syntax passes its complete owned conformance scope.
Suite: scoped
Requires: executable=python3; scope=test

import subprocess
import sys

result = subprocess.run(
    [sys.executable, "spec_tests.py", "--spec", "spec.txt", "--program", "./commonmark",
     "--pattern", "^(Autolinks|Raw HTML)$"],
    capture_output=True, text=True,
)
print(result.stdout)
print(result.stderr, file=sys.stderr)
assert result.returncode == 0
=== END AC inline-html-suite ===

=== AC inline-html-malformed ===
Intent: Invalid autolinks and malformed HTML remain ordinary escaped content.
Suite: scoped
Requires: executable=python3; scope=test

import subprocess
import sys

result = subprocess.run(
    [sys.executable, "spec_tests.py", "--spec", "spec.txt", "--program", "./commonmark",
     "--pattern", "^(Autolinks|Raw HTML)$"],
    capture_output=True, text=True,
)
assert result.returncode == 0
=== END AC inline-html-malformed ===

## User Acceptance

- None.

## Guardrails

- Raw HTML is preserved only when it satisfies the CommonMark inline grammar.
- The parser does not execute or fetch HTML or URI content.
=== END FEATURE-INLINE-HTML.md ===
=== FEATURE-RENDER-BLOCKS.md ===
# FEATURE: Block HTML Rendering

| Field       | Value |
|-------------|-------|
| Version     | 20260812 V1 |
| Description | Defines HTML rendering for parsed block structure and block-level content. |
| Depends On  | ARCHITECTURE.md, FEATURE-BLOCK-LISTS.md, FEATURE-INLINE-HTML.md |
| Provides    | block HTML renderer |
| Consumes    | block parser, inline parser |

## Questions

- None.

## Rendering Contract

- Paragraphs render as `<p>` elements except for paragraphs in tight lists.
- ATX and setext headings render as `<h1>` through `<h6>`.
- Thematic breaks render as `<hr />`.
- Indented and fenced code render as escaped `<pre><code>` content.
- Block quotes render as `<blockquote>`.
- Ordered and unordered lists preserve nesting, start values, item boundaries, and tightness.
- Raw HTML blocks remain unescaped.

## Programmatic Acceptance

=== AC render-blocks-suite ===
Intent: Block-level HTML rendering passes the complete owned conformance scope.
Suite: scoped
Requires: executable=python3; scope=test

import subprocess
import sys

result = subprocess.run(
    [sys.executable, "spec_tests.py", "--spec", "spec.txt", "--program", "./commonmark",
     "--pattern", "^(Thematic breaks|ATX headings|Setext headings|Indented code blocks|Fenced code blocks|HTML blocks|Paragraphs|Blank lines|Block quotes|List items|Lists)$"],
    capture_output=True, text=True,
)
print(result.stdout)
print(result.stderr, file=sys.stderr)
assert result.returncode == 0
=== END AC render-blocks-suite ===

=== AC render-blocks-escaping ===
Intent: Block rendering preserves structural nesting and escapes literal code content.
Suite: scoped
Requires: executable=python3; scope=test

import subprocess
import sys

result = subprocess.run(
    [sys.executable, "spec_tests.py", "--spec", "spec.txt", "--program", "./commonmark",
     "--pattern", "^(Indented code blocks|Fenced code blocks|Block quotes|List items|Lists)$"],
    capture_output=True, text=True,
)
assert result.returncode == 0
=== END AC render-blocks-escaping ===

## User Acceptance

- None.

## Guardrails

- Block rendering never treats captured test output as a correctness oracle.
- Literal code content is escaped while raw HTML blocks remain preserved.
=== END FEATURE-RENDER-BLOCKS.md ===
=== FEATURE-RENDER-INLINE.md ===
# FEATURE: Inline HTML Rendering

| Field       | Value |
|-------------|-------|
| Version     | 20260812 V1 |
| Description | Defines HTML rendering for parsed inline nodes and inline source constructs. |
| Depends On  | ARCHITECTURE.md, FEATURE-RENDER-BLOCKS.md |
| Provides    | inline HTML renderer |
| Consumes    | inline parser |

## Questions

- None.

## Rendering Contract

- Text escapes HTML-significant characters.
- Emphasis and strong emphasis render as `<em>` and `<strong>`.
- Code spans render as escaped `<code>` content.
- Links render escaped `href` and optional `title` attributes.
- Images render escaped `src`, plain-text `alt`, and optional `title`.
- Hard breaks render as `<br />`; soft breaks preserve permitted line-break semantics.
- Autolinks and raw HTML preserve the required HTML representation.
- Entity references render as their corresponding characters, with HTML escaping where required.

## Programmatic Acceptance

=== AC render-inline-suite ===
Intent: Inline HTML rendering passes the complete owned conformance scope.
Suite: scoped
Requires: executable=python3; scope=test

import subprocess
import sys

result = subprocess.run(
    [sys.executable, "spec_tests.py", "--spec", "spec.txt", "--program", "./commonmark",
     "--pattern", "^(Characters and lines|Backslash escapes|Entity and numeric character references|Code spans|Emphasis and strong emphasis|Links|Images|Autolinks|Raw HTML|Hard line breaks|Soft line breaks|Textual content)$"],
    capture_output=True, text=True,
)
print(result.stdout)
print(result.stderr, file=sys.stderr)
assert result.returncode == 0
=== END AC render-inline-suite ===

=== AC render-inline-escaping ===
Intent: Inline rendering escapes text and code while preserving links, images, and raw HTML.
Suite: scoped
Requires: executable=python3; scope=test

import subprocess
import sys

result = subprocess.run(
    [sys.executable, "spec_tests.py", "--spec", "spec.txt", "--program", "./commonmark",
     "--pattern", "^(Code spans|Links|Images|Autolinks|Raw HTML|Hard line breaks|Soft line breaks)$"],
    capture_output=True, text=True,
)
assert result.returncode == 0
=== END AC render-inline-escaping ===

## User Acceptance

- None.

## Guardrails

- Renderer output is derived from parsed inline nodes and never from test-result text.
- User text cannot inject markup except through valid raw HTML constructs.
=== END FEATURE-RENDER-INLINE.md ===
=== FEATURE-RENDER-NORMALIZATION.md ===
# FEATURE: Normalized HTML Output

| Field       | Value |
|-------------|-------|
| Version     | 20260812 V1 |
| Description | Defines URL, title, attribute, entity, and special-character output compatibility with the supplied harness. |
| Depends On  | ARCHITECTURE.md, FEATURE-RENDER-INLINE.md |
| Provides    | normalized URL output, normalized title output, normalized attribute output |
| Consumes    | block HTML renderer, inline HTML renderer |

## Questions

- None.

## Output Rules

- URLs preserve valid percent escapes and encode required spaces, Unicode characters, and backslashes.
- Titles and attribute values are HTML-escaped.
- Attribute output is deterministic and preserves required source semantics.
- Entity and numeric references are normalized consistently with the supplied harness.
- Special characters use the required HTML entities.
- Normalization compatibility does not alter code-block whitespace or raw HTML semantics.

## Programmatic Acceptance

=== AC render-normalization-suite ===
Intent: URL, title, attribute, entity, and special-character rendering passes its complete conformance scope.
Suite: scoped
Requires: executable=python3; scope=test

import subprocess
import sys

result = subprocess.run(
    [sys.executable, "spec_tests.py", "--spec", "spec.txt", "--program", "./commonmark",
     "--pattern", "^(Links|Images|Autolinks|Raw HTML|Entity and numeric character references)$"],
    capture_output=True, text=True,
)
print(result.stdout)
print(result.stderr, file=sys.stderr)
assert result.returncode == 0
=== END AC render-normalization-suite ===

=== AC render-normalization-specials ===
Intent: Normalized output preserves URL escaping, title escaping, attribute semantics, and special characters.
Suite: scoped
Requires: executable=python3; scope=test

import subprocess
import sys

result = subprocess.run(
    [sys.executable, "spec_tests.py", "--spec", "spec.txt", "--program", "./commonmark",
     "--pattern", "^(Links|Images|Autolinks|Raw HTML|Entity and numeric character references)$"],
    capture_output=True, text=True,
)
assert result.returncode == 0
=== END AC render-normalization-specials ===

## User Acceptance

- None.

## Guardrails

- Output normalization is deterministic and does not use a hardcoded pass tally.
- The supplied normalizer is used only as a comparison aid; its output is never the acceptance oracle.
=== END FEATURE-RENDER-NORMALIZATION.md ===