app.py

# app.py - Intrinsic Intelligence Foundations: Search & Viewer (Gradio)
import os, re, json
import gradio as gr

HF_REPO = os.environ.get("HF_DATASET_REPO", "kadubon/intrinsic-intelligence-foundations")

def _load_dataset():
    try:
        from datasets import load_dataset
        ds = load_dataset(HF_REPO, split="train")
        return ds, "remote"
    except Exception as e:
        local_path = os.environ.get("LOCAL_JSONL", "huggingface_dataset_takahashi.jsonl")
        rows = []
        if os.path.exists(local_path):
            with open(local_path, "r", encoding="utf-8") as f:
                for line in f:
                    if line.strip():
                        rows.append(json.loads(line))
            return rows, "local"
        raise RuntimeError(f"Failed to load dataset: {e}")

DS, MODE = _load_dataset()

def _iter_records():
    if MODE == "remote":
        for r in DS:
            yield r
    else:
        for r in DS:
            yield r

EQ_PATTERN = re.compile(r"\[\[EQ:([^\]]+)\]\]")

def expand_placeholders(text, equations, to="tex"):
    tmap = {}
    for e in equations or []:
        tmap[e.get("id")] = (e.get("tex",""), e.get("mathml",""))
    if to == "tex":
        return EQ_PATTERN.sub(lambda m: f"$${tmap.get(m.group(1), ('',''))[0]}$$", text or "")
    else:
        return EQ_PATTERN.sub(lambda m: tmap.get(m.group(1), ('',''))[1] or "", text or "")

def record_to_md(rec, show="tex", preview_chars=1200):
    title = rec.get("title","(no title)")
    doi = rec.get("doi")
    url = rec.get("urls",{}).get("landing") or (f"https://doi.org/{doi}" if doi else None)
    authors = rec.get("authors") or []
    if authors and isinstance(authors, list):
        auth = ", ".join([f"{a.get('given','').strip()} {a.get('family','').strip()}".strip() if isinstance(a, dict) else str(a) for a in authors])
    else:
        auth = "K. Takahashi"
    kws = rec.get("keywords") or []
    eqs = rec.get("equations") or []
    text = rec.get("fulltext",{}).get("plain","")

    if show == "tex":
        body = expand_placeholders(text, eqs, to="tex")
        md_body = body[:preview_chars] + ("…" if len(body) > preview_chars else "")
    else:
        body = expand_placeholders(text, eqs, to="mathml")
        snippet = body[:preview_chars] + ("…" if len(body) > preview_chars else "")
        md_body = f"<div>{snippet}</div>"

    meta = []
    if url:
        meta.append(f"[DOI]({url})")
    if doi and not url:
        meta.append(f"`{doi}`")
    if kws:
        meta.append("keywords: " + ", ".join(kws[:10]))

    header = f"### 📄 {title}\n\n**Authors:** {auth}  \n" + ("  \n".join(meta) if meta else "")
    return header + "\n\n" + md_body

def search_dataset(query, show="tex", top_k=5):
    q = (query or "").strip().lower()
    if not q:
        return "Type keywords to search titles/keywords/text."
    hits = []
    for rec in _iter_records():
        title = rec.get("title","")
        kw = " ".join(rec.get("keywords") or [])
        text = rec.get("fulltext",{}).get("plain","")
        hay = " ".join([title, kw, text]).lower()
        if q in hay:
            hits.append(rec)
        if len(hits) >= top_k*3:
            break
    if not hits:
        return "No results found."
    md = []
    for rec in hits[:top_k]:
        md.append(record_to_md(rec, show=show))
    return "\n\n---\n\n".join(md)

def view_by_index(idx, show="tex"):
    try:
        idx = int(idx)
    except:
        return "Index must be an integer (0-based)."
    rec = None
    if MODE == "remote":
        if 0 <= idx < len(DS):
            rec = DS[int(idx)]
    else:
        if 0 <= idx < len(DS):
            rec = DS[int(idx)]
    if rec is None:
        return f"Out of range. Available: 0..{len(DS)-1}"
    return record_to_md(rec, show=show, preview_chars=10000)

with gr.Blocks(title="Intrinsic Intelligence Foundations") as demo:
    gr.Markdown(
        """
# Intrinsic Intelligence Foundations — Search & Viewer
Explore the math-aware dataset (TeX/MathML) for autonomous, self-organizing intelligence research.
**Source:** [Hugging Face dataset](https://huggingface.co/datasets/kadubon/intrinsic-intelligence-foundations)
"""
    )
    with gr.Row():
        query = gr.Textbox(label="Search keywords", placeholder="e.g., teleogenesis, fractal category theory, UGV")
        show = gr.Radio(choices=["tex","mathml"], value="tex", label="Render equations as")
        topk = gr.Slider(minimum=1, maximum=10, value=5, step=1, label="Top-K")
        btn = gr.Button("Search")
    out = gr.Markdown()
    btn.click(fn=search_dataset, inputs=[query, show, topk], outputs=out)

    gr.Markdown("### Or view by 0-based index")
    with gr.Row():
        idx = gr.Number(value=0, precision=0, label="Index")
        show2 = gr.Radio(choices=["tex","mathml"], value="tex", label="Render equations as", interactive=True)
        btn2 = gr.Button("View record")
    out2 = gr.Markdown()
    btn2.click(fn=view_by_index, inputs=[idx, show2], outputs=out2)

if __name__ == "__main__":
    demo.launch()