#!/usr/bin/env bash # build-graph: Generate Cytoscape.js graph JSON from room/ directory # Reads room artifacts, parses frontmatter, runs analyze-room, # and outputs dashboard/graph.json for the De Stijl knowledge graph dashboard # Must handle empty room gracefully (outputs 8 section group nodes) # macOS bash 3.2 compatibility -- Python3 fallback if [ "${BASH_VERSINFO[0]:-0}" -lt 4 ]; then exec python3 - "$@" << 'PYEOF' import sys, os, re, json, subprocess from pathlib import Path from datetime import datetime, timezone ROOM_DIR = sys.argv[1] if len(sys.argv) > 1 else "./room" OUTPUT_PATH = sys.argv[2] if len(sys.argv) > 2 else "./dashboard/graph.json" # ---- 8 DD-aligned core sections with display names and De Stijl colors ---- SECTION_COLORS = { "problem-definition": "#A63D2F", "market-analysis": "#C8A43C", "solution-design": "#5C5A56", "business-model": "#2D6B4A", "competitive-analysis": "#B5602A", "team-execution": "#1E3A6E", "legal-ip": "#6B4E8B", "financial-model": "#2A6B5E", } SECTION_LABELS = { "problem-definition": "PROBLEM DEFINITION", "market-analysis": "MARKET ANALYSIS", "solution-design": "SOLUTION DESIGN", "business-model": "BUSINESS MODEL", "competitive-analysis": "COMPETITIVE ANALYSIS", "team-execution": "TEAM & EXECUTION", "legal-ip": "LEGAL & IP", "financial-model": "FINANCIAL MODEL", } CORE_SECTIONS = [ "problem-definition", "market-analysis", "solution-design", "business-model", "competitive-analysis", "team-execution", "legal-ip", "financial-model", ] EXTENDED_COLORS = { "opportunity-bank": "#C87137", "funding": "#3A7B5E", "personas": "#7B4A8B", "product": "#1E3A6E", "ip": "#6B4E8B", "decisions": "#A63D2F", "beta-testing": "#2D6B4A", "product-evolution": "#C8A43C", "tech-stack": "#5C5A56", } STRUCTURAL_DIRS = {"meetings", "team"} # ---- Dynamic section discovery ---- is_core = set(CORE_SECTIONS) EXTENDED_SECTIONS = [] room_path = Path(ROOM_DIR) if room_path.is_dir(): for d in sorted(room_path.iterdir()): if not d.is_dir(): continue dn = d.name if dn.startswith("."): continue if dn in STRUCTURAL_DIRS or dn in is_core: continue has_content = False if (d / "STATE.md").is_file(): has_content = True else: for _ in d.rglob("*.md"): has_content = True break if has_content: EXTENDED_SECTIONS.append(dn) SECTION_COLORS[dn] = EXTENDED_COLORS.get(dn, "#5C5A56") SECTION_LABELS[dn] = dn.replace("-", " ").upper() ALL_SECTIONS = CORE_SECTIONS + EXTENDED_SECTIONS # ---- Helpers ---- def extract_frontmatter(filepath): lines = [] try: with open(filepath, "r", encoding="utf-8", errors="replace") as f: raw = f.readlines() except Exception: return [] if not raw or raw[0].strip() != "---": return [] for line in raw[1:]: if line.strip() == "---": break lines.append(line.rstrip("\n")) return lines def get_field(filepath, field): for line in extract_frontmatter(filepath): if line.startswith(field + ":"): val = line[len(field) + 1:].strip().strip('"') return val return "" def get_title(filepath): try: with open(filepath, "r", encoding="utf-8", errors="replace") as f: for line in f: if line.startswith("# "): return line[2:].strip() except Exception: pass return Path(filepath).stem def json_escape(s): s = s.replace("\\", "\\\\") s = s.replace('"', '\\"') s = s.replace("\n", "\\n") s = s.replace("\t", "\\t") return s def find_md_files(directory, recursive=False): """Find .md files, sorted, like find ... | sort""" p = Path(directory) if not p.is_dir(): return [] if recursive: files = sorted(p.rglob("*.md")) else: files = sorted(p.glob("*.md")) return [str(f) for f in files if f.is_file()] # ---- Ensure output directory exists ---- os.makedirs(os.path.dirname(OUTPUT_PATH) or ".", exist_ok=True) # =========================================================== # Phase 1: Build nodes # =========================================================== nodes = [] node_count = 0 artifact_count = 0 # Section group nodes (always present) for section in ALL_SECTIONS: color = SECTION_COLORS[section] label = SECTION_LABELS[section] nodes.append( '{ "data": { "id": "%s", "label": "%s", "color": "%s", "layer": "structure" }, "classes": "section-group" }' % (section, label, color) ) node_count += 1 # Artifact nodes artifact_pipeline = {} artifact_stage = {} artifact_section = {} pipeline_stages = {} if room_path.is_dir(): for section in ALL_SECTIONS: section_dir = room_path / section if not section_dir.is_dir(): continue for f in sorted(section_dir.rglob("*.md")): if not f.is_file(): continue fname = f.name if fname in ("ROOM.md", "STATE.md", "TEAM-STATE.md"): continue rel_path = str(f.relative_to(section_dir)) rel_clean = rel_path.replace(".md", "").replace("/", "-") artifact_id = "%s/%s" % (section, rel_clean) title = json_escape(get_title(str(f))) methodology = json_escape(get_field(str(f), "methodology")) created = get_field(str(f), "created") pipeline = get_field(str(f), "pipeline") pipeline_stage_val = get_field(str(f), "pipeline_stage") color = SECTION_COLORS[section] nodes.append( '{ "data": { "id": "%s", "label": "%s", "section": "%s", "color": "%s", "methodology": "%s", "created": "%s", "pipeline": "%s", "pipeline_stage": "%s", "layer": "content", "parent": "%s" }, "classes": "artifact" }' % (artifact_id, title, section, color, methodology, created, pipeline, pipeline_stage_val, section) ) node_count += 1 artifact_count += 1 artifact_section[artifact_id] = section if pipeline and pipeline_stage_val and pipeline_stage_val != "null": artifact_pipeline[artifact_id] = pipeline artifact_stage[artifact_id] = pipeline_stage_val pipeline_stages["%s:%s" % (pipeline, pipeline_stage_val)] = artifact_id # Meeting nodes meeting_count = 0 meeting_speakers = {} meetings_dir = room_path / "meetings" if meetings_dir.is_dir(): for mdir in sorted(meetings_dir.iterdir()): if not mdir.is_dir(): continue dir_name = mdir.name meeting_date = dir_name[:10] meeting_name = dir_name[11:] if len(dir_name) > 11 else "" meeting_id = "meeting/%s" % dir_name speakers_csv = "" decisions_count = "0" action_items_count = "0" meta_file = mdir / "metadata.yaml" if meta_file.is_file(): in_speakers = False with open(str(meta_file), "r", encoding="utf-8", errors="replace") as mf: for yline in mf: yline = yline.rstrip("\n") if yline.startswith("speakers:"): in_speakers = True continue if in_speakers: m = re.match(r"^\s*-\s+(.*)", yline) if m: spk = m.group(1).strip().strip('"') if speakers_csv: speakers_csv += "," + spk else: speakers_csv = spk else: in_speakers = False m = re.match(r"^decisions_count:\s*(.*)", yline) if m: decisions_count = m.group(1).strip() m = re.match(r"^action_items_count:\s*(.*)", yline) if m: action_items_count = m.group(1).strip() meeting_speakers[meeting_id] = speakers_csv label = json_escape(meeting_name or dir_name) nodes.append( '{ "data": { "id": "%s", "label": "%s", "meeting_date": "%s", "speakers": "%s", "decisions_count": "%s", "action_items_count": "%s", "color": "#D4A843", "layer": "content" }, "classes": "meeting" }' % (meeting_id, label, meeting_date, json_escape(speakers_csv), decisions_count, action_items_count) ) node_count += 1 meeting_count += 1 # Speaker nodes speaker_count = 0 speaker_seen = {} team_dir = room_path / "team" if team_dir.is_dir(): for role_type in ("members", "mentors", "advisors"): role_dir = team_dir / role_type if not role_dir.is_dir(): continue for pdir in sorted(role_dir.iterdir()): if not pdir.is_dir(): continue person_name = pdir.name if person_name in speaker_seen: continue speaker_seen[person_name] = 1 speaker_id = "speaker/%s" % person_name role = "" profile = pdir / "PROFILE.md" if profile.is_file(): role = get_field(str(profile), "role") if not role: role = get_field(str(profile), "primary_role") if not role: role = role_type label = json_escape(person_name) nodes.append( '{ "data": { "id": "%s", "label": "%s", "role": "%s", "role_type": "%s", "color": "#1E3A6E", "layer": "content" }, "classes": "speaker" }' % (speaker_id, label, json_escape(role), role_type) ) node_count += 1 speaker_count += 1 # =========================================================== # Phase 2: Build edges # =========================================================== edges = [] edge_count = 0 def add_edge(source, target, etype, label, css_class, source_type="room"): global edge_count edges.append( '{ "data": { "id": "e%d", "source": "%s", "target": "%s", "type": "%s", "label": "%s", "source_type": "%s" }, "classes": "%s" }' % (edge_count, json_escape(source), json_escape(target), etype, json_escape(label), source_type, css_class) ) edge_count += 1 # FEEDS_INTO edges for key, aid in pipeline_stages.items(): pipeline_name, stage_str = key.split(":", 1) try: next_stage = int(stage_str) + 1 except ValueError: continue next_key = "%s:%d" % (pipeline_name, next_stage) if next_key in pipeline_stages: add_edge(aid, pipeline_stages[next_key], "FEEDS_INTO", "feeds into", "feeds-into") # Run analyze-room analyze_output = "" if room_path.is_dir(): # Find analyze-room script relative to this script's location # When running via heredoc, try multiple candidate paths candidates = [] if "BASH_SOURCE_DIR" in os.environ: candidates.append(Path(os.environ["BASH_SOURCE_DIR"]) / "analyze-room") # Try relative to ROOM_DIR parent candidates.append(Path(ROOM_DIR).resolve().parent / "scripts" / "analyze-room") # Try current working directory candidates.append(Path(".") / "scripts" / "analyze-room") for candidate in candidates: if candidate.is_file() and os.access(str(candidate), os.X_OK): try: analyze_output = subprocess.check_output( ["bash", str(candidate), ROOM_DIR], stderr=subprocess.DEVNULL, text=True ) except Exception: analyze_output = "" break # Parse CONTRADICT/CONVERGE section_latest = {} for aid, sec in artifact_section.items(): section_latest[sec] = aid for line in analyze_output.splitlines(): if line.startswith("CONTRADICT:"): parts = line.split(":") if len(parts) >= 5: sec_a, sec_b, confidence, message = parts[1], parts[2], parts[3], parts[4] source = section_latest.get(sec_a, sec_a) target = section_latest.get(sec_b, sec_b) add_edge(source, target, "CONTRADICTS", message, "contradicts") for line in analyze_output.splitlines(): if line.startswith("CONVERGE:"): parts = line.split(":") if len(parts) >= 5: term, count_str, confidence, message = parts[1], parts[2], parts[3], parts[4] matching = [] for section in ALL_SECTIONS: sdir = room_path / section if sdir.is_dir(): try: result = subprocess.run( ["grep", "-rql", term, str(sdir)], capture_output=True, text=True ) if result.returncode == 0: matching.append(section) except Exception: pass if len(matching) >= 2: add_edge(matching[0], matching[1], "CONVERGES", term, "converges") # INFORMS edges if room_path.is_dir(): for aid, section in artifact_section.items(): fname = aid.split("/", 1)[1] if "/" in aid else aid filepath = room_path / section / (fname + ".md") if not filepath.is_file(): continue try: content = filepath.read_text(encoding="utf-8", errors="replace") except Exception: continue for target_section in ALL_SECTIONS: if target_section == section: continue if "[[%s]]" % target_section in content: add_edge(aid, target_section, "INFORMS", "informs", "informs") # Meeting edges for mid, spk_csv in meeting_speakers.items(): if not spk_csv: continue for spk in spk_csv.split(","): spk = spk.strip() if not spk: continue speaker_id = "speaker/%s" % spk if spk not in speaker_seen: speaker_seen[spk] = 1 label = json_escape(spk) nodes.append( '{ "data": { "id": "%s", "label": "%s", "role": "", "role_type": "", "color": "#1E3A6E", "layer": "content" }, "classes": "speaker" }' % (speaker_id, label) ) node_count += 1 speaker_count += 1 add_edge(speaker_id, mid, "SPOKE_IN", "spoke in", "spoke-in", "meeting") add_edge(speaker_id, mid, "ATTENDED", "attended", "attended", "meeting") # FILED_TO edges dir_name = mid.replace("meeting/", "", 1) filed_dir = room_path / "meetings" / dir_name / "filed-to" if filed_dir.is_dir(): for link in sorted(filed_dir.iterdir()): if not link.exists(): continue if link.is_symlink(): try: target_path = str(link.resolve()) except Exception: target_path = str(link) else: target_path = "" try: with open(str(link), "r", encoding="utf-8", errors="replace") as lf: head_lines = [] for i, ln in enumerate(lf): if i >= 5: break head_lines.append(ln.rstrip("\n")) head_text = "\n".join(head_lines) # Equivalent of: grep -oP '(?<=\().*?(?=\))|(?<=: ).*|^/.*|^room/.*' m = re.search(r'(?<=\().*?(?=\))|(?<=: ).*|^/.*|^room/.*', head_text, re.MULTILINE) if m: target_path = m.group(0) except Exception: pass if not target_path: target_path = link.stem for sec in ALL_SECTIONS: if sec in target_path: add_edge(mid, sec, "FILED_TO", "filed to", "filed-to", "meeting") break # =========================================================== # Phase 2b: Wikilink concept nodes and intelligence edges # =========================================================== concept_count = 0 concept_seen = {} concept_ref_count = {} concept_sources = {} if room_path.is_dir(): for mdfile in sorted(room_path.rglob("*.md")): if not mdfile.is_file(): continue if mdfile.name == "transcript.md": continue rel_path = str(mdfile.relative_to(room_path)) source_id = rel_path.replace(".md", "") try: content = mdfile.read_text(encoding="utf-8", errors="replace") except Exception: continue wlinks = sorted(set(re.findall(r'\[\[([^\]]+)\]\]', content))) if not wlinks: continue for concept in wlinks: if not concept: continue concept_key = concept.lower().replace(" ", "-") prev_sources = concept_sources.get(concept_key, "") if source_id not in prev_sources: concept_ref_count[concept_key] = concept_ref_count.get(concept_key, 0) + 1 if prev_sources: concept_sources[concept_key] = prev_sources + "," + source_id else: concept_sources[concept_key] = source_id # Create concept nodes for concepts referenced in 2+ files for concept_key, ref_count in concept_ref_count.items(): if ref_count < 2: continue concept_id = "concept/%s" % concept_key concept_seen[concept_key] = 1 css_class = "concept" resolved = False if room_path.is_dir(): for _ in room_path.rglob(concept_key + ".md"): resolved = True break if not resolved: css_class = "concept unresolved" label = json_escape(concept_key) nodes.append( '{ "data": { "id": "%s", "label": "%s", "ref_count": %d, "color": "#C8A43C", "layer": "intelligence" }, "classes": "%s" }' % (concept_id, label, ref_count, css_class) ) node_count += 1 concept_count += 1 # REFERENCES edges for concept_key, src_csv in concept_sources.items(): concept_id = "concept/%s" % concept_key for src in src_csv.split(","): if not src: continue add_edge(src, concept_id, "REFERENCES", "references", "references", "wikilink") # Cross-meeting intelligence edges from MEETINGS-INTELLIGENCE.md intel_file = room_path / "MEETINGS-INTELLIGENCE.md" if intel_file.is_file(): try: intel_content = intel_file.read_text(encoding="utf-8", errors="replace") except Exception: intel_content = "" meeting_date_re = re.compile(r'[0-9]{4}-[0-9]{2}-[0-9]{2}-[a-z0-9-]+') def find_meeting_refs(line_text): refs = meeting_date_re.findall(line_text) matched = [] for mref in refs: if not mref: continue for mid2 in meeting_speakers: if mref in mid2: matched.append(mid2) break return matched # Parse Convergence Signals in_convergence = False for iline in intel_content.splitlines(): if iline.startswith("## Convergence Signals"): in_convergence = True continue if iline.startswith("## ") and in_convergence: in_convergence = False continue if in_convergence and "-" in iline: conv_meetings = find_meeting_refs(iline) if len(conv_meetings) >= 2: for i in range(len(conv_meetings) - 1): for j in range(i + 1, len(conv_meetings)): add_edge(conv_meetings[i], conv_meetings[j], "REINFORCES", "convergence signal", "reinforces", "intelligence") # Parse Contradictions in_contradictions = False for iline in intel_content.splitlines(): if iline.startswith("## Contradictions"): in_contradictions = True continue if iline.startswith("## ") and in_contradictions: in_contradictions = False continue if in_contradictions and "-" in iline: contra_meetings = find_meeting_refs(iline) if len(contra_meetings) >= 2: for i in range(len(contra_meetings) - 1): for j in range(i + 1, len(contra_meetings)): add_edge(contra_meetings[i], contra_meetings[j], "CONTRADICTS", "contradiction", "contradicts", "intelligence") # =========================================================== # Phase 3: Build intelligence object # =========================================================== gaps_json_parts = [] convergence_json_parts = [] contradictions_json_parts = [] for line in analyze_output.splitlines(): if line.startswith("GAP:"): parts = line.split(":") if len(parts) >= 5: gaps_json_parts.append( '{ "type": "%s", "section": "%s", "confidence": "%s", "message": "%s" }' % (json_escape(parts[1]), json_escape(parts[2]), json_escape(parts[3]), json_escape(parts[4])) ) elif line.startswith("CONVERGE:"): parts = line.split(":") if len(parts) >= 5: convergence_json_parts.append( '{ "term": "%s", "count": %s, "confidence": "%s", "message": "%s" }' % (json_escape(parts[1]), parts[2], json_escape(parts[3]), json_escape(parts[4])) ) elif line.startswith("CONTRADICT:"): parts = line.split(":") if len(parts) >= 5: contradictions_json_parts.append( '{ "section_a": "%s", "section_b": "%s", "confidence": "%s", "message": "%s" }' % (json_escape(parts[1]), json_escape(parts[2]), json_escape(parts[3]), json_escape(parts[4])) ) # Build summary venture_stage = "Pre-Opportunity" room_name = "Data Room" state_file = room_path / "STATE.md" if state_file.is_file(): try: state_content = state_file.read_text(encoding="utf-8", errors="replace") except Exception: state_content = "" for sl in state_content.splitlines(): if re.match(r'(?i)venture_stage:', sl): venture_stage = re.sub(r'.*venture_stage:\s*', '', sl, flags=re.IGNORECASE).strip().strip('"') break for sl in state_content.splitlines(): if re.match(r'(?i)(venture_name|room_name|project_name|name):', sl): room_name = re.sub(r'^[^:]*:\s*', '', sl).strip().strip('"') break if room_name == "Data Room": for sl in state_content.splitlines(): if sl.startswith("# "): h1 = sl[2:].strip() if h1 and h1 != "Data Room State": room_name = h1 break if room_name == "Data Room": dn = room_path.name if dn != "room": room_name = " ".join(w.capitalize() for w in dn.split("-")) room_name_json = room_name.replace("\\", "\\\\").replace('"', '\\"') sections_active = 0 sections_empty = 0 for section in ALL_SECTIONS: sdir = room_path / section if sdir.is_dir(): count = sum( 1 for f in sdir.iterdir() if f.is_file() and f.suffix == ".md" and f.name not in ("ROOM.md", "STATE.md") ) if count > 0: sections_active += 1 else: sections_empty += 1 else: sections_empty += 1 # =========================================================== # Phase 4: Output JSON # =========================================================== generated_at = datetime.now(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ") nodes_str = ",\n ".join(nodes) if nodes else "" edges_str = ",\n ".join(edges) if edges else "" output = """{ "meta": { "roomName": "%s", "ventureStage": "%s", "generatedAt": "%s", "roomDir": "%s", "generator": "MindrianOS build-graph" }, "elements": { "nodes": [ %s ], "edges": [ %s ] }, "intelligence": { "gaps": [%s], "convergence": [%s], "contradictions": [%s], "summary": { "total_artifacts": %d, "meeting_count": %d, "speaker_count": %d, "concept_count": %d, "venture_stage": "%s", "sections_active": %d, "sections_empty": %d } } } """ % ( room_name_json, venture_stage, generated_at, ROOM_DIR, nodes_str, edges_str, ", ".join(gaps_json_parts), ", ".join(convergence_json_parts), ", ".join(contradictions_json_parts), artifact_count, meeting_count, speaker_count, concept_count, venture_stage, sections_active, sections_empty, ) with open(OUTPUT_PATH, "w", encoding="utf-8") as out: out.write(output) print("Generated graph.json: %d nodes, %d edges" % (node_count, edge_count)) PYEOF fi set -euo pipefail # Portable readlink -f (macOS lacks GNU readlink) portable_realpath() { if command -v realpath >/dev/null 2>&1; then realpath "$1" elif command -v python3 >/dev/null 2>&1; then python3 -c "import os; print(os.path.realpath('$1'))" else echo "$1" fi } ROOM_DIR="${1:-./room}" OUTPUT_PATH="${2:-./dashboard/graph.json}" # ── 8 DD-aligned core sections with display names and De Stijl colors ── declare -A SECTION_COLORS SECTION_COLORS=( [problem-definition]="#A63D2F" [market-analysis]="#C8A43C" [solution-design]="#5C5A56" [business-model]="#2D6B4A" [competitive-analysis]="#B5602A" [team-execution]="#1E3A6E" [legal-ip]="#6B4E8B" [financial-model]="#2A6B5E" ) declare -A SECTION_LABELS SECTION_LABELS=( [problem-definition]="PROBLEM DEFINITION" [market-analysis]="MARKET ANALYSIS" [solution-design]="SOLUTION DESIGN" [business-model]="BUSINESS MODEL" [competitive-analysis]="COMPETITIVE ANALYSIS" [team-execution]="TEAM & EXECUTION" [legal-ip]="LEGAL & IP" [financial-model]="FINANCIAL MODEL" ) CORE_SECTIONS=( problem-definition market-analysis solution-design business-model competitive-analysis team-execution legal-ip financial-model ) # ── Pre-assigned colors for known extension sections ── declare -A EXTENDED_COLORS EXTENDED_COLORS=( [opportunity-bank]="#C87137" [funding]="#3A7B5E" [personas]="#7B4A8B" [product]="#1E3A6E" [ip]="#6B4E8B" [decisions]="#A63D2F" [beta-testing]="#2D6B4A" [product-evolution]="#C8A43C" [tech-stack]="#5C5A56" ) # ── Structural directories (not sections) ── STRUCTURAL_DIRS=("meetings" "team") # ── Dynamic section discovery ── EXTENDED_SECTIONS=() declare -A is_core_section declare -A is_structural_dir for s in "${CORE_SECTIONS[@]}"; do is_core_section["$s"]=1; done for s in "${STRUCTURAL_DIRS[@]}"; do is_structural_dir["$s"]=1; done if [ -d "$ROOM_DIR" ]; then for dir in "$ROOM_DIR"/*/; do [ -d "$dir" ] || continue dir_name=$(basename "$dir") [[ "$dir_name" == .* ]] && continue [ -n "${is_structural_dir[$dir_name]:-}" ] && continue [ -n "${is_core_section[$dir_name]:-}" ] && continue # Qualify: must contain at least one .md file (search recursively) has_content=false if [ -f "${dir}STATE.md" ]; then has_content=true else md_file=$(find "$dir" -name "*.md" 2>/dev/null | head -1) [ -n "$md_file" ] && has_content=true fi if $has_content; then EXTENDED_SECTIONS+=("$dir_name") # Assign color and label for extended sections if [ -n "${EXTENDED_COLORS[$dir_name]:-}" ]; then SECTION_COLORS["$dir_name"]="${EXTENDED_COLORS[$dir_name]}" else SECTION_COLORS["$dir_name"]="#5C5A56" fi SECTION_LABELS["$dir_name"]=$(echo "$dir_name" | tr '-' ' ' | tr '[:lower:]' '[:upper:]') fi done fi ALL_SECTIONS=("${CORE_SECTIONS[@]}" "${EXTENDED_SECTIONS[@]}") # ── Helper: extract frontmatter from a file ── extract_frontmatter() { local file="$1" local in_frontmatter=false local line_num=0 while IFS= read -r line; do line_num=$((line_num + 1)) if [ "$line_num" -eq 1 ] && [ "$line" = "---" ]; then in_frontmatter=true continue fi if $in_frontmatter && [ "$line" = "---" ]; then break fi if $in_frontmatter; then echo "$line" fi done < "$file" } # ── Helper: get a specific frontmatter field ── get_field() { local file="$1" local field="$2" extract_frontmatter "$file" | grep "^${field}:" | sed "s/${field}:[[:space:]]*//" | tr -d '"' | head -1 || true } # ── Helper: get first heading from markdown file ── get_title() { local file="$1" local title title=$(grep -m1 '^# ' "$file" 2>/dev/null | sed 's/^# //' || true) if [ -z "$title" ]; then title=$(basename "$file" .md) fi echo "$title" } # ── Helper: JSON-escape a string ── json_escape() { local s="$1" s="${s//\\/\\\\}" s="${s//\"/\\\"}" s="${s//$'\n'/\\n}" s="${s//$'\t'/\\t}" echo "$s" } # ── Ensure output directory exists ── mkdir -p "$(dirname "$OUTPUT_PATH")" # ═══════════════════════════════════════════════ # Phase 1: Build nodes # ═══════════════════════════════════════════════ nodes="" node_count=0 artifact_count=0 # Section group nodes (always present) for section in "${ALL_SECTIONS[@]}"; do color="${SECTION_COLORS[$section]}" label="${SECTION_LABELS[$section]}" node="{ \"data\": { \"id\": \"${section}\", \"label\": \"${label}\", \"color\": \"${color}\", \"layer\": \"structure\" }, \"classes\": \"section-group\" }" if [ -n "$nodes" ]; then nodes="${nodes}, ${node}" else nodes=" ${node}" fi node_count=$((node_count + 1)) done # Artifact nodes (from room/ directory) # Track pipeline info for edge building declare -A artifact_pipeline # id -> pipeline name declare -A artifact_stage # id -> pipeline stage number declare -A artifact_section # id -> section name declare -A pipeline_stages # "pipeline:stage" -> artifact id if [ -d "$ROOM_DIR" ]; then for section in "${ALL_SECTIONS[@]}"; do section_dir="${ROOM_DIR}/${section}" [ -d "$section_dir" ] || continue while IFS= read -r f; do [ -f "$f" ] || continue fname=$(basename "$f") [ "$fname" = "ROOM.md" ] || [ "$fname" = "STATE.md" ] || [ "$fname" = "TEAM-STATE.md" ] && continue # Build artifact ID from relative path within section # Use -- as separator instead of / to avoid breaking Cytoscape CSS selectors rel_path="${f#${section_dir}/}" rel_clean="${rel_path%.md}" rel_clean="${rel_clean////-}" artifact_id="${section}/${rel_clean}" title=$(json_escape "$(get_title "$f")") methodology=$(json_escape "$(get_field "$f" "methodology")") created=$(get_field "$f" "created") pipeline=$(get_field "$f" "pipeline") pipeline_stage=$(get_field "$f" "pipeline_stage") color="${SECTION_COLORS[$section]}" node="{ \"data\": { \"id\": \"${artifact_id}\", \"label\": \"${title}\", \"section\": \"${section}\", \"color\": \"${color}\", \"methodology\": \"${methodology}\", \"created\": \"${created}\", \"pipeline\": \"${pipeline}\", \"pipeline_stage\": \"${pipeline_stage}\", \"layer\": \"content\", \"parent\": \"${section}\" }, \"classes\": \"artifact\" }" nodes="${nodes}, ${node}" node_count=$((node_count + 1)) artifact_count=$((artifact_count + 1)) # Track for pipeline edge building artifact_section["$artifact_id"]="$section" if [ -n "$pipeline" ] && [ -n "$pipeline_stage" ] && [ "$pipeline_stage" != "null" ]; then artifact_pipeline["$artifact_id"]="$pipeline" artifact_stage["$artifact_id"]="$pipeline_stage" pipeline_stages["${pipeline}:${pipeline_stage}"]="$artifact_id" fi done < <(find "$section_dir" -name "*.md" -type f 2>/dev/null | sort) done fi # Meeting nodes (from room/meetings/ directories) meeting_count=0 declare -A meeting_speakers # meeting_id -> comma-separated speakers if [ -d "$ROOM_DIR/meetings" ]; then for mdir in "$ROOM_DIR"/meetings/*/; do [ -d "$mdir" ] || continue dir_name=$(basename "$mdir") meeting_date="${dir_name:0:10}" meeting_name="${dir_name:11}" meeting_id="meeting/${dir_name}" # Read metadata.yaml for speakers list and counts speakers_csv="" decisions_count="0" action_items_count="0" if [ -f "${mdir}metadata.yaml" ]; then in_speakers=false while IFS= read -r yline; do if [[ "$yline" =~ ^speakers: ]]; then in_speakers=true continue fi if $in_speakers; then if [[ "$yline" =~ ^[[:space:]]*-[[:space:]]+(.*) ]]; then spk="${BASH_REMATCH[1]}" spk=$(echo "$spk" | tr -d '"' | xargs) if [ -n "$speakers_csv" ]; then speakers_csv="${speakers_csv},${spk}" else speakers_csv="${spk}" fi else in_speakers=false fi fi if [[ "$yline" =~ ^decisions_count:[[:space:]]*(.*) ]]; then decisions_count="${BASH_REMATCH[1]}" fi if [[ "$yline" =~ ^action_items_count:[[:space:]]*(.*) ]]; then action_items_count="${BASH_REMATCH[1]}" fi done < "${mdir}metadata.yaml" fi meeting_speakers["$meeting_id"]="$speakers_csv" label=$(json_escape "${meeting_name:-$dir_name}") node="{ \"data\": { \"id\": \"${meeting_id}\", \"label\": \"${label}\", \"meeting_date\": \"${meeting_date}\", \"speakers\": \"$(json_escape "$speakers_csv")\", \"decisions_count\": \"${decisions_count}\", \"action_items_count\": \"${action_items_count}\", \"color\": \"#D4A843\", \"layer\": \"content\" }, \"classes\": \"meeting\" }" nodes="${nodes}, ${node}" node_count=$((node_count + 1)) meeting_count=$((meeting_count + 1)) done fi # Speaker nodes (from room/team/ profiles) speaker_count=0 declare -A speaker_seen # speaker name -> 1 if [ -d "$ROOM_DIR/team" ]; then for role_type in members mentors advisors; do role_dir="$ROOM_DIR/team/${role_type}" [ -d "$role_dir" ] || continue for pdir in "$role_dir"/*/; do [ -d "$pdir" ] || continue person_name=$(basename "$pdir") [ -n "${speaker_seen[$person_name]:-}" ] && continue speaker_seen["$person_name"]=1 speaker_id="speaker/${person_name}" role="" if [ -f "${pdir}PROFILE.md" ]; then role=$(get_field "${pdir}PROFILE.md" "role") [ -z "$role" ] && role=$(get_field "${pdir}PROFILE.md" "primary_role") fi role="${role:-$role_type}" label=$(json_escape "${person_name}") node="{ \"data\": { \"id\": \"${speaker_id}\", \"label\": \"${label}\", \"role\": \"$(json_escape "$role")\", \"role_type\": \"${role_type}\", \"color\": \"#1E3A6E\", \"layer\": \"content\" }, \"classes\": \"speaker\" }" nodes="${nodes}, ${node}" node_count=$((node_count + 1)) speaker_count=$((speaker_count + 1)) done done fi # ═══════════════════════════════════════════════ # Phase 2: Build edges # ═══════════════════════════════════════════════ edges="" edge_count=0 add_edge() { local source="$1" local target="$2" local etype="$3" local label="$4" local css_class="$5" local source_type="${6:-room}" local edge="{ \"data\": { \"id\": \"e${edge_count}\", \"source\": \"$(json_escape "$source")\", \"target\": \"$(json_escape "$target")\", \"type\": \"${etype}\", \"label\": \"$(json_escape "$label")\", \"source_type\": \"${source_type}\" }, \"classes\": \"${css_class}\" }" if [ -n "$edges" ]; then edges="${edges}, ${edge}" else edges=" ${edge}" fi edge_count=$((edge_count + 1)) } # FEEDS_INTO edges: consecutive pipeline stages for key in "${!pipeline_stages[@]}"; do pipeline_name="${key%%:*}" stage="${key##*:}" next_stage=$((stage + 1)) next_key="${pipeline_name}:${next_stage}" if [ -n "${pipeline_stages[$next_key]:-}" ]; then add_edge "${pipeline_stages[$key]}" "${pipeline_stages[$next_key]}" "FEEDS_INTO" "feeds into" "feeds-into" fi done # Run analyze-room for CONTRADICT/CONVERGE intelligence analyze_output="" if [ -d "$ROOM_DIR" ]; then script_dir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" if [ -x "${script_dir}/analyze-room" ]; then analyze_output=$(bash "${script_dir}/analyze-room" "$ROOM_DIR" 2>/dev/null || true) fi fi # Parse CONTRADICT lines -> edges between sections' most recent artifacts declare -A section_latest # section -> most recent artifact id (by created date or last found) for aid in "${!artifact_section[@]}"; do sec="${artifact_section[$aid]}" section_latest["$sec"]="$aid" done while IFS= read -r line; do if [[ "$line" == CONTRADICT:* ]]; then IFS=':' read -r _ sec_a sec_b confidence message <<< "$line" source="${section_latest[$sec_a]:-$sec_a}" target="${section_latest[$sec_b]:-$sec_b}" add_edge "$source" "$target" "CONTRADICTS" "$message" "contradicts" fi done <<< "$analyze_output" # Parse CONVERGE lines -> edges between sections sharing a convergence term while IFS= read -r line; do if [[ "$line" == CONVERGE:* ]]; then IFS=':' read -r _ term count confidence message <<< "$line" # Find sections whose artifacts mention this term matching_sections=() for section in "${ALL_SECTIONS[@]}"; do section_dir="${ROOM_DIR}/${section}" if [ -d "$section_dir" ] && grep -rql "$term" "$section_dir" 2>/dev/null; then matching_sections+=("$section") fi done # Connect the first two matching sections if [ ${#matching_sections[@]} -ge 2 ]; then add_edge "${matching_sections[0]}" "${matching_sections[1]}" "CONVERGES" "$term" "converges" fi fi done <<< "$analyze_output" # INFORMS edges: cross-references [[section-name]] in artifact content if [ -d "$ROOM_DIR" ]; then for aid in "${!artifact_section[@]}"; do # Reconstruct file path from artifact id section="${artifact_section[$aid]}" fname="${aid##*/}" filepath="${ROOM_DIR}/${section}/${fname}.md" [ -f "$filepath" ] || continue for target_section in "${ALL_SECTIONS[@]}"; do [ "$target_section" = "$section" ] && continue if grep -q "\[\[${target_section}\]\]" "$filepath" 2>/dev/null; then add_edge "$aid" "$target_section" "INFORMS" "informs" "informs" fi done done fi # Meeting edges: SPOKE_IN, ATTENDED, FILED_TO for mid in "${!meeting_speakers[@]}"; do IFS=',' read -ra spk_arr <<< "${meeting_speakers[$mid]}" for spk in "${spk_arr[@]}"; do spk=$(echo "$spk" | xargs) [ -z "$spk" ] && continue speaker_id="speaker/${spk}" # Create speaker node if not already seen (from metadata but no PROFILE.md) if [ -z "${speaker_seen[$spk]:-}" ]; then speaker_seen["$spk"]=1 label=$(json_escape "${spk}") node="{ \"data\": { \"id\": \"${speaker_id}\", \"label\": \"${label}\", \"role\": \"\", \"role_type\": \"\", \"color\": \"#1E3A6E\", \"layer\": \"content\" }, \"classes\": \"speaker\" }" nodes="${nodes}, ${node}" node_count=$((node_count + 1)) speaker_count=$((speaker_count + 1)) fi add_edge "$speaker_id" "$mid" "SPOKE_IN" "spoke in" "spoke-in" "meeting" add_edge "$speaker_id" "$mid" "ATTENDED" "attended" "attended" "meeting" done # FILED_TO edges from meeting to sections via filed-to/ pointer files # Pointer files are markdown with the target path as text content (not symlinks) dir_name="${mid#meeting/}" filed_dir="$ROOM_DIR/meetings/${dir_name}/filed-to" if [ -d "$filed_dir" ]; then for link in "$filed_dir"/*; do [ -e "$link" ] || continue # Read target path: try symlink first, then parse file content if [ -L "$link" ]; then target_path=$(portable_realpath "$link" 2>/dev/null || echo "$link") else target_path=$(head -5 "$link" 2>/dev/null | sed -n 's/.*(\([^)]*\)).*/\1/p; s/^.*: //p; /^\/\|^room\//p' | head -1 || echo "$link") # If no path found in content, use the filename as hint [ -z "$target_path" ] && target_path=$(basename "$link" .md) fi # Try to match to a section name for sec in "${ALL_SECTIONS[@]}"; do if [[ "$target_path" == *"${sec}"* ]]; then add_edge "$mid" "$sec" "FILED_TO" "filed to" "filed-to" "meeting" break fi done done fi done # ═══════════════════════════════════════════════ # Phase 2b: Wikilink concept nodes and intelligence edges # ═══════════════════════════════════════════════ concept_count=0 declare -A concept_seen # concept name -> 1 (node created) declare -A concept_ref_count # concept name -> number of distinct source files declare -A concept_sources # concept name -> comma-separated source artifact ids # Scan all .md files for [[wikilinks]] (exclude transcript.md to avoid noise) if [ -d "$ROOM_DIR" ]; then while IFS= read -r mdfile; do [ -f "$mdfile" ] || continue fname=$(basename "$mdfile") [ "$fname" = "transcript.md" ] && continue # Determine source artifact id (section/filename or meeting/dirname/filename) rel_path="${mdfile#$ROOM_DIR/}" source_id="${rel_path%.md}" # Extract [[wikilinks]] wlinks=$(grep -oE '\[\[[^\]]+\]\]' "$mdfile" 2>/dev/null | sed 's/\[\[//;s/\]\]//' | sort -u || true) [ -z "$wlinks" ] && continue while IFS= read -r concept; do [ -z "$concept" ] && continue # Normalize: lowercase, spaces to hyphens concept_key=$(echo "$concept" | tr '[:upper:]' '[:lower:]' | tr ' ' '-') # Track reference count (unique source files) prev_count="${concept_ref_count[$concept_key]:-0}" # Check if this source already counted prev_sources="${concept_sources[$concept_key]:-}" if [[ "$prev_sources" != *"$source_id"* ]]; then concept_ref_count["$concept_key"]=$((prev_count + 1)) if [ -n "$prev_sources" ]; then concept_sources["$concept_key"]="${prev_sources},${source_id}" else concept_sources["$concept_key"]="$source_id" fi fi done <<< "$wlinks" done < <(find "$ROOM_DIR" -name "*.md" -type f 2>/dev/null) fi # Create concept nodes for concepts referenced in 2+ files for concept_key in "${!concept_ref_count[@]}"; do ref_count="${concept_ref_count[$concept_key]}" [ "$ref_count" -lt 2 ] && continue concept_id="concept/${concept_key}" concept_seen["$concept_key"]=1 # Check if concept resolves to an existing .md file in room css_class="concept" resolved=false if [ -d "$ROOM_DIR" ]; then while IFS= read -r match; do resolved=true break done < <(find "$ROOM_DIR" -name "${concept_key}.md" -type f 2>/dev/null) fi $resolved || css_class="concept unresolved" label=$(json_escape "${concept_key}") node="{ \"data\": { \"id\": \"${concept_id}\", \"label\": \"${label}\", \"ref_count\": ${ref_count}, \"color\": \"#C8A43C\", \"layer\": \"intelligence\" }, \"classes\": \"${css_class}\" }" nodes="${nodes}, ${node}" node_count=$((node_count + 1)) concept_count=$((concept_count + 1)) done # Add REFERENCES edges from source artifacts to concepts for concept_key in "${!concept_sources[@]}"; do concept_id="concept/${concept_key}" # Only add edges to concepts that have nodes (2+ refs) or all concepts IFS=',' read -ra src_arr <<< "${concept_sources[$concept_key]}" for src in "${src_arr[@]}"; do [ -z "$src" ] && continue add_edge "$src" "$concept_id" "REFERENCES" "references" "references" "wikilink" done done # Cross-meeting intelligence edges from MEETINGS-INTELLIGENCE.md if [ -f "$ROOM_DIR/MEETINGS-INTELLIGENCE.md" ]; then intel_content=$(cat "$ROOM_DIR/MEETINGS-INTELLIGENCE.md") # Parse Convergence Signals section for REINFORCES edges in_convergence=false while IFS= read -r iline; do if [[ "$iline" == "## Convergence Signals"* ]]; then in_convergence=true continue fi if [[ "$iline" == "## "* ]] && $in_convergence; then in_convergence=false continue fi if $in_convergence && [[ "$iline" == *"-"* ]]; then # Extract meeting date patterns (YYYY-MM-DD-name) conv_meetings=() while IFS= read -r mref; do [ -z "$mref" ] && continue # Find matching meeting node for mid in "${!meeting_speakers[@]}"; do if [[ "$mid" == *"$mref"* ]]; then conv_meetings+=("$mid") break fi done done < <(echo "$iline" | grep -oE '[0-9]{4}-[0-9]{2}-[0-9]{2}-[a-z0-9-]+' || true) # Create REINFORCES edge between pairs if [ ${#conv_meetings[@]} -ge 2 ]; then for ((i=0; i<${#conv_meetings[@]}-1; i++)); do for ((j=i+1; j<${#conv_meetings[@]}; j++)); do add_edge "${conv_meetings[$i]}" "${conv_meetings[$j]}" "REINFORCES" "convergence signal" "reinforces" "intelligence" done done fi fi done <<< "$intel_content" # Parse Contradictions section for CONTRADICTS edges in_contradictions=false while IFS= read -r iline; do if [[ "$iline" == "## Contradictions"* ]]; then in_contradictions=true continue fi if [[ "$iline" == "## "* ]] && $in_contradictions; then in_contradictions=false continue fi if $in_contradictions && [[ "$iline" == *"-"* ]]; then contra_meetings=() while IFS= read -r mref; do [ -z "$mref" ] && continue for mid in "${!meeting_speakers[@]}"; do if [[ "$mid" == *"$mref"* ]]; then contra_meetings+=("$mid") break fi done done < <(echo "$iline" | grep -oE '[0-9]{4}-[0-9]{2}-[0-9]{2}-[a-z0-9-]+' || true) if [ ${#contra_meetings[@]} -ge 2 ]; then for ((i=0; i<${#contra_meetings[@]}-1; i++)); do for ((j=i+1; j<${#contra_meetings[@]}; j++)); do add_edge "${contra_meetings[$i]}" "${contra_meetings[$j]}" "CONTRADICTS" "contradiction" "contradicts" "intelligence" done done fi fi done <<< "$intel_content" fi # ═══════════════════════════════════════════════ # Phase 3: Build intelligence object # ═══════════════════════════════════════════════ # Parse gaps gaps_json="" while IFS= read -r line; do if [[ "$line" == GAP:* ]]; then IFS=':' read -r _ gap_type gap_section confidence message _ <<< "$line" gap="{ \"type\": \"$(json_escape "$gap_type")\", \"section\": \"$(json_escape "$gap_section")\", \"confidence\": \"$(json_escape "$confidence")\", \"message\": \"$(json_escape "$message")\" }" if [ -n "$gaps_json" ]; then gaps_json="${gaps_json}, ${gap}" else gaps_json="${gap}" fi fi done <<< "$analyze_output" # Parse convergence convergence_json="" while IFS= read -r line; do if [[ "$line" == CONVERGE:* ]]; then IFS=':' read -r _ term count confidence message <<< "$line" conv="{ \"term\": \"$(json_escape "$term")\", \"count\": ${count}, \"confidence\": \"$(json_escape "$confidence")\", \"message\": \"$(json_escape "$message")\" }" if [ -n "$convergence_json" ]; then convergence_json="${convergence_json}, ${conv}" else convergence_json="${conv}" fi fi done <<< "$analyze_output" # Parse contradictions contradictions_json="" while IFS= read -r line; do if [[ "$line" == CONTRADICT:* ]]; then IFS=':' read -r _ sec_a sec_b confidence message _ <<< "$line" contra="{ \"section_a\": \"$(json_escape "$sec_a")\", \"section_b\": \"$(json_escape "$sec_b")\", \"confidence\": \"$(json_escape "$confidence")\", \"message\": \"$(json_escape "$message")\" }" if [ -n "$contradictions_json" ]; then contradictions_json="${contradictions_json}, ${contra}" else contradictions_json="${contra}" fi fi done <<< "$analyze_output" # Build summary venture_stage="Pre-Opportunity" room_name="Data Room" if [ -f "${ROOM_DIR}/STATE.md" ]; then stage_line=$(grep -i 'venture_stage:' "${ROOM_DIR}/STATE.md" 2>/dev/null | head -1 || true) if [ -n "$stage_line" ]; then venture_stage=$(echo "$stage_line" | sed 's/.*venture_stage:[[:space:]]*//' | tr -d '"' | xargs) fi # Extract room/venture name (try multiple frontmatter keys) name_line=$(grep -iE 'venture_name:|room_name:|project_name:|name:' "${ROOM_DIR}/STATE.md" 2>/dev/null | head -1 || true) if [ -n "$name_line" ]; then room_name=$(echo "$name_line" | sed 's/^[^:]*:[[:space:]]*//' | tr -d '"' | xargs) fi # Fallback: extract from H1 heading (e.g., "# Polygon Eye Data Room") if [ "$room_name" = "Data Room" ]; then h1_line=$(grep -m1 '^# ' "${ROOM_DIR}/STATE.md" 2>/dev/null | sed 's/^# //' || true) if [ -n "$h1_line" ] && [ "$h1_line" != "Data Room State" ]; then room_name="$h1_line" fi fi # Last fallback: use parent directory name (titlecased) if [ "$room_name" = "Data Room" ]; then dir_name=$(basename "$ROOM_DIR") if [ "$dir_name" != "room" ]; then room_name=$(echo "$dir_name" | sed 's/-/ /g; s/\b\(.\)/\u\1/g') fi fi fi # Escape room name for JSON room_name_json=$(echo "$room_name" | sed 's/\\/\\\\/g; s/"/\\"/g') sections_active=0 sections_empty=0 for section in "${ALL_SECTIONS[@]}"; do section_dir="${ROOM_DIR}/${section}" if [ -d "$section_dir" ]; then count=$(find "$section_dir" -maxdepth 1 -name "*.md" ! -name "ROOM.md" ! -name "STATE.md" 2>/dev/null | wc -l | tr -d ' ') if [ "$count" -gt 0 ]; then sections_active=$((sections_active + 1)) else sections_empty=$((sections_empty + 1)) fi else sections_empty=$((sections_empty + 1)) fi done # ═══════════════════════════════════════════════ # Phase 4: Output JSON # ═══════════════════════════════════════════════ cat > "$OUTPUT_PATH" << ENDJSON { "meta": { "roomName": "${room_name_json}", "ventureStage": "${venture_stage}", "generatedAt": "$(date -u +%Y-%m-%dT%H:%M:%SZ)", "roomDir": "${ROOM_DIR}", "generator": "MindrianOS build-graph" }, "elements": { "nodes": [ ${nodes} ], "edges": [ ${edges} ] }, "intelligence": { "gaps": [${gaps_json}], "convergence": [${convergence_json}], "contradictions": [${contradictions_json}], "summary": { "total_artifacts": ${artifact_count}, "meeting_count": ${meeting_count}, "speaker_count": ${speaker_count}, "concept_count": ${concept_count}, "venture_stage": "${venture_stage}", "sections_active": ${sections_active}, "sections_empty": ${sections_empty} } } } ENDJSON echo "Generated graph.json: ${node_count} nodes, ${edge_count} edges"