gismap 0.1.0__py3-none-any.whl

gismap/__init__.py

@@ -0,0 +1,20 @@
+"""Top-level package for Analytical Lab Cartography In Computer Science."""
+
+from importlib.metadata import metadata
+
+from gismap.utils.common import get_classes as get_classes
+from gismap.gismo import make_gismo as make_gismo
+from gismap.search import (
+    Search as Search,
+    SearchDocuments as SearchDocuments,
+    SearchLandmarks as SearchLandmarks,
+    SearchFeatures as SearchFeatures,
+    search_to_html as search_to_html,
+    search_to_text as search_to_text,
+)
+
+
+infos = metadata(__name__)
+__version__ = infos["Version"]
+__author__ = """Fabien Mathieu"""
+__email__ = "loufab@gmail.com"

gismap/gismap.py

@@ -0,0 +1 @@
+"""Main module."""

gismap/gismo.py

@@ -0,0 +1,379 @@
+from sklearn.feature_extraction.text import CountVectorizer
+
+from gismo.corpus import Corpus
+from gismo.embedding import Embedding
+from gismo.gismo import Gismo
+
+
+def make_post_publi(lab):
+    """
+    Hook to turn publication key stored in a corpus into actual publication.
+
+    Parameters
+    ----------
+    lab: :class:`~gismap.lab.lab.Lab`
+        Lab that contains the corpus publications.
+
+    Returns
+    -------
+    callable
+    """
+
+    def to_bib(g, i):
+        item = g.corpus[i]
+        return lab.publications[item]
+
+    return to_bib
+
+
+def make_gismo(lab, vectorizer_parameters=None):
+    """
+    Makes a gismo out of a lab.
+
+    Parameters
+    ----------
+    lab: :class:`~gismap.lab.lab.Lab`
+        Lab that contains publications.
+    vectorizer_parameters: :class:`dict`
+        Overriding parameters for the Countvectorizer of the gismo.
+
+    Returns
+    -------
+    gismo: :class:`~gismo.gismo.Gismo`
+        Gismo of the lab.
+    """
+    parameters = {"ngram_range": (1, 3), "dtype": float, "stop_words": sw, "min_df": 3}
+    if vectorizer_parameters is not None:
+        parameters.update(vectorizer_parameters)
+    corpus = Corpus([p for p in lab.publications], to_text=lab.publi_to_text)
+    vectorizer = CountVectorizer(**parameters)
+    embedding = Embedding(vectorizer=vectorizer)
+    embedding.fit_transform(corpus)
+    gismo = Gismo(corpus, embedding)
+    gismo.post_documents_item = make_post_publi(lab)
+    return gismo
+
+
+stop_words = [
+    "01",
+    "20plus",
+    "academia",
+    "academic",
+    "academy",
+    "académie",
+    "acm",
+    "activities",
+    "actualités",
+    "adresse",
+    "advances",
+    "affichertoutesdepuis",
+    "ajouter",
+    "an",
+    "ancitations",
+    "and",
+    "annual",
+    "annéetrier",
+    "antipolis",
+    "are",
+    "article",
+    "articledisponiblesnon",
+    "articles",
+    "articles0",
+    "arxiv",
+    "as",
+    "astérisque",
+    "at",
+    "attended",
+    "au",
+    "auteuradresse",
+    "auteurnouveaux",
+    "auteurnouvelles",
+    "aux",
+    "award",
+    "awarded",
+    "awards",
+    "base",
+    "based",
+    "be",
+    "been",
+    "bibliography",
+    "bibliothèquemétriquesalertesparamètresconnexionconnexionobtenir",
+    "board",
+    "book",
+    "born",
+    "by",
+    "california",
+    "called",
+    "can",
+    "celles",
+    "centrale",
+    "centre",
+    "ces",
+    "cet",
+    "cette",
+    "chair",
+    "chaussées",
+    "ciarlet",
+    "citations",
+    "citationstrier",
+    "cited",
+    "citée",
+    "cnrs",
+    "cnrsadresse",
+    "coauteurscoauteurssuivre",
+    "collaboration",
+    "colloquium",
+    "collège",
+    "columbia",
+    "committee",
+    "comptabilisées",
+    "conference",
+    "contact",
+    "contributions",
+    "council",
+    "cours",
+    "course",
+    "courses",
+    "dans",
+    "de",
+    "des",
+    "diegoadresse",
+    "différentes",
+    "director",
+    "disponiblessur",
+    "doi",
+    "données",
+    "doubleles",
+    "du",
+    "décompte",
+    "early",
+    "earned",
+    "ecole",
+    "ed",
+    "edinburgh",
+    "elected",
+    "en",
+    "english",
+    "ens",
+    "envoi",
+    "et",
+    "europaea",
+    "european",
+    "events",
+    "exigences",
+    "fellow",
+    "financementcoauteurstout",
+    "for",
+    "formerly",
+    "forum",
+    "fr",
+    "france",
+    "franceadresse",
+    "français",
+    "françois",
+    "french",
+    "from",
+    "fusionnéesle",
+    "fusionnés",
+    "google",
+    "grenoble",
+    "gérard",
+    "habilitation",
+    "had",
+    "he",
+    "her",
+    "here",
+    "highly",
+    "his",
+    "home",
+    "hong",
+    "honorary",
+    "ici",
+    "ieee",
+    "imag",
+    "in",
+    "inclut",
+    "informatique",
+    "innovation",
+    "inria",
+    "insa",
+    "institut",
+    "international",
+    "invited",
+    "is",
+    "isbn",
+    "it",
+    "je",
+    "jean",
+    "jour",
+    "journal",
+    "july",
+    "known",
+    "kong",
+    "la",
+    "lab",
+    "le",
+    "lecture",
+    "les",
+    "lille",
+    "liées",
+    "liés",
+    "lncs",
+    "lyon",
+    "mail",
+    "maintenant",
+    "medal",
+    "media",
+    "member",
+    "mises",
+    "mon",
+    "monde",
+    "my",
+    "mécanique",
+    "national",
+    "ne",
+    "nombre",
+    "normale",
+    "notes",
+    "notificationsokmon",
+    "novel",
+    "of",
+    "olivier",
+    "on",
+    "opération",
+    "ordre",
+    "page",
+    "pages",
+    "paper",
+    "par",
+    "paraccès",
+    "parcitée",
+    "paris",
+    "paristech",
+    "partout",
+    "pas",
+    "pdf",
+    "peut",
+    "peuvent",
+    "peux",
+    "ph",
+    "phd",
+    "pierre",
+    "plus",
+    "polytechnique",
+    "ponts",
+    "pour",
+    "pp",
+    "premier",
+    "preprint",
+    "president",
+    "prix",
+    "prize",
+    "proceedings",
+    "professor",
+    "profil",
+    "profilcitée",
+    "profilma",
+    "programme",
+    "programmes",
+    "propre",
+    "présentation",
+    "publications",
+    "publiccoauteurstitretriertrier",
+    "published",
+    "que",
+    "qui",
+    "received",
+    "recherche",
+    "record",
+    "report",
+    "research",
+    "researcher",
+    "réaliser",
+    "réessayer",
+    "résultatsaideconfidentialitéconditions",
+    "saclayadresse",
+    "san",
+    "scholar",
+    "scholarchargement",
+    "school",
+    "sciences",
+    "scientifique",
+    "scientist",
+    "selected",
+    "senior",
+    "she",
+    "sigmod",
+    "silver",
+    "site",
+    "slides",
+    "sont",
+    "sophia",
+    "sorbonne",
+    "southern",
+    "speaker",
+    "springer",
+    "stanford",
+    "sud",
+    "suivants",
+    "suivies",
+    "summer",
+    "supervision",
+    "supérieure",
+    "supérieureadresse",
+    "sur",
+    "symposium",
+    "système",
+    "tard",
+    "temps",
+    "texas",
+    "that",
+    "the",
+    "then",
+    "theses",
+    "titrecitée",
+    "to",
+    "transactions",
+    "travaux",
+    "télécom",
+    "un",
+    "under",
+    "une",
+    "univ",
+    "university",
+    "universityadresse",
+    "université",
+    "upmc",
+    "usa",
+    "using",
+    "validée",
+    "verimag",
+    "verlag",
+    "veuillez",
+    "via",
+    "vol",
+    "was",
+    "with",
+    "won",
+    "worked",
+    "www",
+    "year",
+    "école",
+    "être",
+    "towards",
+    "paradigm",
+    "we",
+    "this",
+    "which",
+    "our",
+    "proposed",
+    "their",
+    "approach",
+    "each",
+    "such",
+    "show",
+    "what",
+    "nous",
+]
+
+sw = stop_words + [str(i) for i in range(2100)]

gismap/lab/__init__.py

@@ -0,0 +1,10 @@
+from gismap.lab.lab import (
+    Lab as Lab,
+    ListLab as ListLab,
+    LabAuthor as LabAuthor,
+    AuthorMetadata as AuthorMetadata,
+)
+from gismap.lab.lip6 import Lip6 as Lip6, Lip6Lab as Lip6Lab
+from gismap.lab.toulouse import Solace as Solace, LaasLab as LaasLab
+from gismap.lab.graph import lab2graph as lab2graph
+from gismap.lab.vis import generate_html as generate_html

gismap/lab/graph.py

@@ -0,0 +1,234 @@
+import numpy as np
+from collections import defaultdict
+from itertools import combinations
+from gismap.lab.vis import generate_html
+
+
+def initials(name):
+    """
+    Parameters
+    ----------
+    name: :class:`str`
+        Person's name.
+
+    Returns
+    -------
+    :class:`str`
+        Person's initials (2 letters only).
+    """
+    first_letters = [w[0] for w in name.split()]
+    return first_letters[0] + first_letters[-1]
+
+
+def author_to_html(author):
+    """
+    Parameters
+    ----------
+    author: :class:`~gismap.sources.models.Author`
+        Searcher.
+
+    Returns
+    -------
+    HTML string with URL if applicable.
+    """
+    name = getattr(author, "name", "Unknown Author")
+    # Try direct URL property (optional)
+    url = getattr(author, "url", None)
+    # For LabAuthor, check metadata.url
+    if hasattr(author, "metadata"):
+        meta_url = getattr(author.metadata, "url", None)
+        if meta_url:
+            url = meta_url
+        elif hasattr(author.sources[0], "url"):
+            url = author.sources[0].url
+    if url:
+        return f'<a href="{url.strip()}" target="_blank">{name}</a>'
+    else:
+        return name
+
+
+def publication_to_html(pub):
+    """
+    Parameters
+    ----------
+    pub: :class:`~gismap.sources.models.Publication`
+        Publication.
+
+    Returns
+    -------
+    HTML string with hyperlinks where applicable.
+    """
+    # Title as link if available
+    if getattr(pub, "url", None):
+        title_html = f'<a href="{pub.url}" target="_blank">{pub.title}</a>'
+    else:
+        title_html = pub.title
+
+    # Authors: render in order, separated by comma
+    author_html_list = [
+        author_to_html(author) for author in getattr(pub, "authors", [])
+    ]
+    authors_html = ", ".join(author_html_list)
+
+    # Venue, Year
+    venue = getattr(pub, "venue", "")
+    year = getattr(pub, "year", "")
+
+    # Basic HTML layout
+    html = f"{title_html}, by <i>{authors_html}</i>. {venue}, {year}."
+    return html.strip()
+
+
+def publications_list_html(publications, n=10):
+    """
+
+    Parameters
+    ----------
+    publications: :class:`list` of :class:`~gismap.sources.models.Publication`
+        Publications to display.
+    n: :class:`int`, default=10
+        Number of publications to display. If there are more publications, a *Show more* option is available to unravel them.
+
+    Returns
+    -------
+    :class:`str`
+    """
+    list_items = []
+    for i, pub in enumerate(publications):
+        item = publication_to_html(pub)
+        if i < n:
+            li = f"<li>{item}</li>"
+        else:
+            li = f'<li class="extra-publication" style="display:none;">{item}</li>'
+        list_items.append(li)
+    ul_content = "\n".join(list_items)
+
+    if len(publications) <= n:
+        show_more_part = ""
+    else:
+        # Add a "Show more" link and JavaScript for toggling
+        show_more_part = """
+<li>
+  <a href="#" onclick="
+    var elts = this.parentElement.parentElement.querySelectorAll('.extra-publication');
+    for (var i = 0; i < elts.length; ++i) {elts[i].style.display = 'list-item';}
+    this.parentElement.style.display = 'none';
+    return false;">Show more…</a>
+</li>
+        """
+
+    html = f"""<ul>
+{ul_content}
+{show_more_part}
+</ul>
+"""
+    return html
+
+
+def to_node(s, node_pubs):
+    """
+    Parameters
+    ----------
+    s: :class:`~gismap.lab.lab.LabAuthor`
+        Searcher.
+    node_pubs: :class:`dict`
+        Lab publications.
+
+    Returns
+    -------
+    :class:`dict`
+        A display-ready representation of the searcher.
+    """
+    res = {
+        "id": s.key,
+        "hover": f"Click for details on {s.name}.",
+        "overlay": f"<div> Publications of {author_to_html(s)}:</div><div>{publications_list_html(node_pubs[s.key])}</div>",
+        "group": s.metadata.group,
+    }
+    if s.metadata.img:
+        res.update({"image": s.metadata.img, "shape": "circularImage"})
+    else:
+        res["label"] = initials(s.name)
+    if s.metadata.position:
+        x, y = s.metadata.position
+        res.update({"x": x, "y": y, "fixed": True})
+    return res
+
+
+def to_edge(k, v, searchers):
+    """
+    Parameters
+    ----------
+    k: :class:`tuple`
+        Keys of the searchers involved.
+    v: :class:`list`
+        List of joint publications.
+    searchers: :class:`dict`
+        Searchers.
+
+    Returns
+    -------
+    :class:`dict`
+        A display-ready representation of the collaboration edge.
+    """
+    strength = 1 + np.log2(len(v))
+    return {
+        "from": k[0],
+        "to": k[1],
+        "hover": f"Show joint publications from {searchers[k[0]].name} and {searchers[k[1]].name}",
+        "overlay": f"<div> Joint publications from {author_to_html(searchers[k[0]])} and {author_to_html(searchers[k[1]])}:</div><div>{publications_list_html(v)}</div>",
+        "width": int(strength),
+        "length": int(200 / strength),
+    }
+
+
+def lab2graph(lab):
+    """
+    Parameters
+    ----------
+    lab: :class:`~gismap.lab.lab.Lab`
+        A lab populated with searchers and publications.
+
+    Returns
+    -------
+    :class:`str`
+        Collaboration graph.
+
+    Examples
+    --------
+
+    >>> from gismap.lab import ListLab
+    >>> lab = ListLab(author_list=['Tixeuil Sébastien', 'Mathieu Fabien'], name='mini')
+    >>> lab.update_authors()
+    >>> lab.update_publis()
+    >>> len(lab.authors)
+    2
+    >>> len(lab.publications)
+    453
+    >>> html = lab2graph(lab)
+    >>> html[:80]  # doctest: +ELLIPSIS
+    '\\n<div id="mynetwork_..."></div>\\n<div id="modal_..." class="modal">\\n  <'
+    """
+    node_pubs = {k: [] for k in lab.authors}
+    edges_dict = defaultdict(list)
+    for p in lab.publications.values():
+        # Some authors are twice in the author list for some reason. This should take care of it.
+        lauths = {a.key: a for a in p.authors if a.__class__.__name__ == "LabAuthor"}
+        lauths = sorted([a for a in lauths.values()], key=lambda a: str(a.key))
+        for a in lauths:
+            node_pubs[a.key].append(p)
+        for a1, a2 in combinations(lauths, 2):
+            edges_dict[a1.key, a2.key].append(p)
+    connected = {k for kl in edges_dict for k in kl}
+
+    for k, v in node_pubs.items():
+        node_pubs[k] = sorted(v, key=lambda p: -p.year)
+    for k, v in edges_dict.items():
+        edges_dict[k] = sorted(v, key=lambda p: -p.year)
+
+    return generate_html(
+        nodes=[
+            to_node(s, node_pubs) for s in lab.authors.values() if s.key in connected
+        ],
+        edges=[to_edge(k, v, lab.authors) for k, v in edges_dict.items()],
+    )