cjm-markdown-decompose-core

A Markdown decomposition core for context graphs: parses Markdown content (frontmatter, structure, and [[wiki-links]]) into provenance-carrying graph nodes and edges. First source = the ecosystem's own memory/decision files; generalizes to any Markdown corpus.

Modules

• cjm_markdown_decompose_core.__init__
• cjm_markdown_decompose_core.blocks — Embedded DERIVED blocks: spans of a body that are not the document's own content.
• cjm_markdown_decompose_core.extract — Map parsed Markdown onto dev-graph-schema nodes (coarse tier).
• cjm_markdown_decompose_core.ingest — Flatten decomposed notes into graph elements for idempotent extension.
• cjm_markdown_decompose_core.parse — Schema-free Markdown parsing (stdlib + PyYAML).
• cjm_markdown_decompose_core.project — Projections from the decomposed content back out to files.
• cjm_markdown_decompose_core.relations — Per-source-type relationship harvesting from parsed Markdown.
• cjm_markdown_decompose_core.sections — Decompose a Markdown body into ordered Section nodes (the navigable unit).

API

cjm_markdown_decompose_core.blocks

• DerivedBlock class — One derived block: its role and its [start, end) span of the body.
• FencedBlock class — One content div whose headings open no Section (ruling d5442ba8 on finding 67a54d21):
• block_link_targets function — A derived block's fingerprint for the render filter: its link targets as Pandoc hands
• derived_blocks function — Classify a body's derived blocks under its source profile.
• fenced_block_parts function — A fenced block's typed parts (ruling d5442ba8 (1)/(2)), derived from the text at read,
• fenced_blocks function — Classify a body's fenced blocks under its source profile.
• find_chrome_includes function — Site-chrome include lines: a line holding only {{< include PATH >}} whose PATH is one
• find_hand_toc function — The hand table of contents: the FIRST run of two or more anchor-link list lines at
• find_series_callouts function — The series callout(s): a top-level ::: {.callout-tip} whose first line is a surveyed
• find_series_nav_lines function — Hand series-navigation lines: a heading holding only Previous: / Next: and links.
• include_target function — A chrome include's fingerprint for the render filter: the partial it includes (the
• mask_blocks function — The body a relation harvest reads: derived blocks contribute no edges (253ac996 (4)).
• quarto_derived_blocks function — The Quarto archive profile: series callouts, the hand TOC, hand series-nav lines, the
• quarto_fenced_blocks function — The Quarto profile's fenced blocks: the tabset and callout divs whose headings open no

cjm_markdown_decompose_core.extract

• corpus_index_files function — Every <dir>/index.md or <dir>/index.qmd under the root, one per post directory.
• note_from_file function — Read a Markdown file and map it to a coarse NoteNode.
• note_from_parsed function — Build a coarse NoteNode from already-parsed Markdown.
• note_from_text function — Parse + map in one step from in-memory text (hashes the UTF-8 bytes).
• note_type_from function — Read the memory category from metadata.type (None when absent).
• slug_from function — Derive the stable slug: frontmatter name if present, else the path.
• title_from function — Derive a display title: explicit frontmatter title, else the slug titleized.

cjm_markdown_decompose_core.ingest

• corpus_graph_elements function — Collect notes into the node + edge wire-dict lists extend_graph expects.

cjm_markdown_decompose_core.parse

• ParsedMarkdown class — The structural decomposition of one Markdown document.
• absorbs_next_line function — Whether line leaves a block open that takes the next line as its own, so an ATX line
• extract_headings function — Extract ATX headings (#..######) as (level, text) pairs.
• extract_wiki_links function — Extract [[wiki-link]] targets from the body, de-duplicated, order-preserved.
• fenced_code_spans function — Locate fenced code blocks (``` / ~~~) as character spans of the body.
• find_headings function — The body's ATX heading matches, skipping every line inside a fenced code block.
• parse_frontmatter function — Parse frontmatter YAML into a dict (empty dict when absent).
• parse_markdown function — Parse a Markdown document into frontmatter + body + wiki-links + headings.
• split_frontmatter function — Split a leading ----delimited YAML frontmatter block from the body.
• strip_code function — Blank out fenced + inline code spans (replaced with a space, length-agnostic).

cjm_markdown_decompose_core.project

• compose_note_text function — Compose a note's text from its Sections' own state walked by the outline relations.
• element_state function — An element's own STATE as a graph-sourced note journals it (design amendment 56b24fd5
• first_sentence function — A deterministic terse hook: the description's first sentence, capped.
• note_index_line function — Render one index line from a note's frontmatter-derived fields.
• note_text_from_graph_nodes function — Reconstruct a note's file text FROM the graph (frontmatter_raw + ordered raw spans).
• note_view_from_graph_node function — Reconstruct an index-relevant NoteNode from a queried graph node.
• render_memory_index function — Render a MEMORY.md index: notes grouped by category, one line each.
• render_memory_index_from_graph_nodes function — Render the memory index FROM queried graph nodes (the self-hosting path).
• render_note_text function — Reassemble a note's exact file text from its verbatim parts (lossless mode).
• render_onboarding_surface function — Render the onboarding surface: orientation + how-to-query + resident PUSH core + landmark map + how-to-pull.
• section_outline function — The outline relations of decomposed Sections, keyed by Section id -- the same

cjm_markdown_decompose_core.relations

• NoteRelations class — The harvested relationship signals for one note (beyond [[wiki-links]]).
• detect_profile function — Detect the source-type profile from the frontmatter shape.
• harvest_aliases function — Harvest aliases (old URLs) -> bare permalinks (alternate identities).
• harvest_categories function — Harvest categories -> normalized Topic keys (the thematic-clustering facet).
• harvest_relations function — Harvest a note's relationships using its (detected or given) source profile.
• harvest_site_links function — Harvest every in-body link to a site PAGE -> its VERBATIM target (ruling d31e9ba7).
• is_page_target function — A link to a page, not to a file: an image, video or download beside a post is never a
• is_site_link function — A link into the site: relative or rooted, or an absolute URL on the site's own host.
• normalize_permalink function — Reduce a post link to its bare permalink — the path AFTER posts/.
• quarto_site_url function — The site URL of the Quarto project a file belongs to (ruling 260119bf).
• slugify function — Normalize a category/tag to a stable key (lowercased, separators collapsed).

cjm_markdown_decompose_core.sections

• decompose_sections function — Decompose a body into ordered SectionNodes (heading-delimited).
• heading_anchor function — Slugify a heading to its anchor — the Pandoc/Quarto auto-identifier shape.

Dependencies

Depends on: cjm-context-graph-layer, cjm-context-graph-primitives, cjm-dev-graph-schema, pyyaml Used by: cjm-context-graph-projection, cjm-notebook-decompose-core, cjm-substrate