diff --git a/src/pack_liascript_course/pack_liascript_course.py b/src/pack_liascript_course/pack_liascript_course.py index 4c1dcdf..279607f 100644 --- a/src/pack_liascript_course/pack_liascript_course.py +++ b/src/pack_liascript_course/pack_liascript_course.py @@ -133,33 +133,84 @@ def fetch_bytes(source: str) -> bytes: re.compile(r']+\bhref=["\']([^"\']+)["\']', re.IGNORECASE), # HTML script src re.compile(r']+\bsrc=["\']([^"\']+)["\']', re.IGNORECASE), - # LiaScript @import - re.compile(r'@import\s+["\']([^"\']+)["\']', re.IGNORECASE), # HTML audio/video src re.compile(r'<(?:audio|video|source)\b[^>]+\bsrc=["\']([^"\']+)["\']', re.IGNORECASE), ] _ABSOLUTE_PREFIXES = ("http://", "https://", "//", "#", "mailto:", "data:") +# Regex that matches a YAML frontmatter block at the very start of a document. +_YAML_FRONTMATTER_RE = re.compile(r'^---[ \t]*\n(.*?\n)(?:---|\.\.\.)[ \t]*\n', re.DOTALL) + +# LiaScript YAML header fields whose values may be relative file paths. +_YAML_LINK_FIELDS = frozenset({"import", "link", "script", "logo"}) + + +def _extract_yaml_frontmatter_links(content: str) -> list[str]: + """Return file references found in LiaScript YAML header fields. + + Handles both scalar values (``field: path``) and YAML list items + (``- path``) for the fields ``import``, ``link``, ``script``, and + ``logo``. + """ + match = _YAML_FRONTMATTER_RE.match(content) + if not match: + return [] + + frontmatter = match.group(1) + result: list[str] = [] + current_field: str | None = None + + for line in frontmatter.splitlines(): + # A top-level field definition: "key: value" or "key:" + field_match = re.match(r'^([A-Za-z_]\w*)\s*:\s*(.*)', line) + if field_match: + field_name = field_match.group(1).lower() + field_value = field_match.group(2).strip().strip("\"'") + + if field_name in _YAML_LINK_FIELDS: + current_field = field_name + if field_value: # Inline scalar: "field: value" + result.append(field_value) + else: + current_field = None + elif current_field is not None: + # Indented list item under the current field: " - value" + list_match = re.match(r'^\s+-\s+(.*)', line) + if list_match: + value = list_match.group(1).strip().strip("\"'") + if value: + result.append(value) + elif line and not line[0].isspace(): + # Non-indented non-field line: we left the field's block. + current_field = None + + return result + def extract_relative_links(content: str) -> list[str]: """Return a deduplicated list of relative asset paths referenced in *content*.""" seen: set[str] = set() result: list[str] = [] + def _add(raw: str) -> None: + path = raw.strip().split("?")[0].split("#")[0].strip() + if not path: + return + if any(path.startswith(prefix) for prefix in _ABSOLUTE_PREFIXES): + return + if path not in seen: + seen.add(path) + result.append(path) + + # Links declared in the YAML frontmatter take priority. + for link in _extract_yaml_frontmatter_links(content): + _add(link) + + # Links referenced in the document body via Markdown/HTML syntax. for pattern in _LINK_PATTERNS: for match in pattern.finditer(content): - raw = match.group(1).strip() - # Remove query string and fragment - path = raw.split("?")[0].split("#")[0].strip() - if not path: - continue - # Skip absolute links - if any(path.startswith(prefix) for prefix in _ABSOLUTE_PREFIXES): - continue - if path not in seen: - seen.add(path) - result.append(path) + _add(match.group(1)) return result diff --git a/tests/test_pack_liascript_course.py b/tests/test_pack_liascript_course.py index 860d4d8..72c14d6 100644 --- a/tests/test_pack_liascript_course.py +++ b/tests/test_pack_liascript_course.py @@ -122,16 +122,74 @@ def test_html_script(self): links = packer.extract_relative_links(content) assert "js/quiz.js" in links - def test_liascript_import(self): - content = "@import 'macros.md'" + # ------------------------------------------------------------------ + # YAML frontmatter tests + # ------------------------------------------------------------------ + + def test_yaml_import_scalar(self): + content = "---\nimport: macros.md\n---\n# Body" + links = packer.extract_relative_links(content) + assert "macros.md" in links + + def test_yaml_import_list(self): + content = "---\nimport:\n - macros.md\n - helpers.md\n---\n" links = packer.extract_relative_links(content) assert "macros.md" in links + assert "helpers.md" in links + + def test_yaml_link_scalar(self): + content = "---\nlink: styles/custom.css\n---\n" + links = packer.extract_relative_links(content) + assert "styles/custom.css" in links + + def test_yaml_link_list(self): + content = "---\nlink:\n - styles/a.css\n - styles/b.css\n---\n" + links = packer.extract_relative_links(content) + assert "styles/a.css" in links + assert "styles/b.css" in links + + def test_yaml_script_scalar(self): + content = "---\nscript: js/init.js\n---\n" + links = packer.extract_relative_links(content) + assert "js/init.js" in links + + def test_yaml_logo(self): + content = "---\nlogo: images/banner.jpg\n---\n" + links = packer.extract_relative_links(content) + assert "images/banner.jpg" in links - def test_liascript_import_double_quotes(self): - content = '@import "https://example.com/remote.md"' + def test_yaml_skips_absolute_import(self): + content = "---\nimport: https://example.com/remote.md\n---\n" links = packer.extract_relative_links(content) + assert not links + + def test_yaml_skips_absolute_import_in_list(self): + content = "---\nimport:\n - local.md\n - https://example.com/remote.md\n---\n" + links = packer.extract_relative_links(content) + assert "local.md" in links assert "https://example.com/remote.md" not in links + def test_yaml_ignores_other_fields(self): + content = "---\ntitle: My Course\nauthor: Someone\n---\n" + links = packer.extract_relative_links(content) + assert not links + + def test_yaml_and_body_deduplication(self): + content = "---\nlink: styles/custom.css\n---\n" + links = packer.extract_relative_links(content) + assert "styles/custom.css" in links + assert len([l for l in links if l == "styles/custom.css"]) == 1 + + def test_no_yaml_frontmatter(self): + content = "# Just a plain markdown file\n![img](image.png)" + links = packer.extract_relative_links(content) + assert "image.png" in links + + def test_yaml_quoted_values(self): + content = '---\nlogo: "images/logo.png"\n---\n' + links = packer.extract_relative_links(content) + assert "images/logo.png" in links + def test_skips_absolute_urls(self): content = "![logo](https://example.com/logo.png)" links = packer.extract_relative_links(content)