diff --git "a/app.py" "b/app.py" --- "a/app.py" +++ "b/app.py" @@ -1,29 +1,41 @@ +# -*- coding: utf-8 -*- """ -African Books Recovery Project — an observatory for books by African authors -published by African presses, c. 1950–1979. - -The point of the dashboard is not to celebrate a catalogue. It is to make visible -how far each book is from being usable again: located, preserved, rights -identified, cleared, republished, curriculum-ready. Most of this corpus is at -stage one, and the dashboard says so plainly. - -Run locally: streamlit run app.py +African Books Recovery Project — observatoire des livres d'auteurs africains +publies par des presses africaines, v. 1950-1979. + +MEME GRAMMAIRE QUE L'ATLAS CLIMATIQUE AFRIQUE + - streamlit seul en dependance. pandas, numpy et altair sont des dependances + obligatoires de streamlit : rien d'autre a installer, donc rien qui puisse + manquer sur le Space. Pas de pyarrow, pas de plotly, pas de geopandas. + - aucune image : tout est retrace en Altair, interrogeable au survol. + - fichier autonome : il n'importe aucun module du projet, il lit des CSV + qu'il cherche a cote de lui. Vrai sur le Space comme en local, les deux + dossiers etant plats. + - cartes tracees a la main. Vega-Lite sait dessiner du GeoJSON, mais + Streamlit reserialise les donnees d'un graphique en table Arrow avant de + les transmettre, ce qui detruit les geometries imbriquees et laisse la + carte vide. Les contours sont donc deplies en une table plate de sommets. + - bascule clair/sombre et EN/FR, repeinte CSS pour le theme. + +CE QUE LE TABLEAU DE BORD DIT, ET NE DIT PAS + Il ne celebre pas un catalogue : il mesure la distance entre une notice + bibliographique et un texte qu'un enseignant peut utiliser. Six etats, de + « localise » a « pret pour l'enseignement ». Presque tout le corpus est au + premier, et la page d'accueil l'affiche au lieu de l'enfouir. + + Rien n'est fabrique. Une valeur absente reste absente : un pays sans notice + est gris sur les cartes et jamais interpole, une annee inconnue reste « non + datee », et « epuise » n'est jamais rendu par « libre de droits ». + + python build_data.py && python selftest.py && streamlit run app.py """ - -from __future__ import annotations - -import sys -import pathlib +from pathlib import Path import altair as alt +import numpy as np import pandas as pd import streamlit as st -sys.path.insert(0, str(pathlib.Path(__file__).parent)) - -import catalogue as D -import theme as T - st.set_page_config( page_title="African Books Recovery Project", page_icon="📚", @@ -31,1122 +43,1879 @@ st.set_page_config( initial_sidebar_state="expanded", ) -P = T.palette() -T.register_altair_theme(P) -st.markdown(T.css(P), unsafe_allow_html=True) +ICI = Path(__file__).resolve().parent + +# ------------------------------------------------------- langue et apparence +LANGUES = {"English": "en", "Français": "fr"} +THEMES_CLES = ["☀︎ Light", "☾ Dark"] + +# Deux ambiances, aucune n'est un blanc pur ni un noir pur : sur un ecran le +# blanc absolu eblouit et le noir absolu fait vibrer le texte clair. Le clair +# est un papier legerement chaud, le sombre une encre brune tres foncee. La +# teinte de marque est un brun cuir — la toile des reliures de l'epoque — et +# non le bleu institutionnel : ce tableau de bord parle d'imprimerie, et il ne +# doit pas etre pris pour le produit officiel d'une bibliotheque nationale. +PALETTES = { + "clair": dict(page="#3a2b20", fond="#fffdf8", panneau="#f7f3ea", + encre="#2f2418", texte="#4a4239", doux="#867c6f", + grille="#ece6db", axe="#d3cabb", trait="#fffdf8", + cadre="#e3dbce", absent="#c9c7c2", pastille="#fffdf8", + accent="#8c4a2f", actif="#6b3520", sur_accent="#ffffff"), + "sombre": dict(page="#141210", fond="#1e1b17", panneau="#262118", + encre="#f5f1e8", texte="#cec6b9", doux="#978e82", + grille="#332d26", axe="#4a4238", trait="#1e1b17", + cadre="#3a332b", absent="#43423d", pastille="#2c2620", + accent="#c2703f", actif="#8c4a2f", sur_accent="#ffffff"), +} -TITLES = D.titles() -PRESSES = D.presses() -DIGI = D.digitization() -RIGHTS = D.rights() -HOLDINGS = D.holdings() -CONF_ORDER = ["high", "medium", "low"] +def _etat(cle, defaut): + if cle not in st.session_state: + st.session_state[cle] = defaut + return st.session_state[cle] + + +# L'anglais est la langue par defaut : le catalogue est en anglais et la +# majorite des notices viennent de catalogues anglophones. Le francais est a un +# clic, et il compte : un tiers du corpus est francophone. +LANG = _etat("lang", "en") +THEME = _etat("theme", "clair") +PAL = PALETTES[THEME] + +# L'anglais est ici la langue SOURCE. Une chaine sans traduction declaree +# s'affiche en anglais : l'oubli est visible, jamais fatal. +TRAD = { + "African Books Recovery Project": "Projet de récupération du livre africain", + "African authors · African presses · c. 1950–1979": + "Auteurs africains · presses africaines · v. 1950-1979", + "Overview": "Vue d'ensemble", "Atlas": "Atlas", "Presses": "Maisons", + "Catalogue": "Catalogue", "Recovery": "Récupération", "Rights": "Droits", + "Copies": "Exemplaires", "Methods": "Méthodes", + "titles catalogued": "notices de titres", + "presses in the register": "maisons au registre", + "countries of publication": "pays de publication", + "languages of publication": "langues de publication", + "with a known scan": "avec une numérisation connue", + "countries with no record yet": "pays sans aucune notice", + "Titles": "Titres", "Houses": "Maisons", "Country": "Pays", + "Region": "Région", "City": "Ville", "Language": "Langue", + "Decade": "Décennie", "Genre": "Genre", "Year": "Année", + "Author": "Auteur", "Title": "Titre", "Press": "Maison", + "Confidence": "Fiabilité", "Scan": "Numérisation", "In print": "Disponible", + "Stage": "Étape", "Cluster": "Groupe", "Sources": "Sources", + "high": "élevée", "medium": "moyenne", "low": "faible", + "W": "O", "E": "E", "N": "N", "S": "S", + "n.d.": "n.d.", "undated": "non daté", "Undated": "Non daté", + "Titles catalogued": "Titres au catalogue", + "Presses in the register": "Maisons au registre", + "Titles with a scan": "Titres numérisés", + "Download the full catalogue (CSV)": "Télécharger le catalogue complet (CSV)", + "Download the press register (CSV)": "Télécharger le registre des maisons (CSV)", + "Filters": "Filtres", "All": "Tout", "Search": "Rechercher", +} -# --------------------------------------------------------------------------- -# shared bits -# --------------------------------------------------------------------------- +def t(x): + return x if LANG == "en" else TRAD.get(x, x) + + +def bil(en_txt, fr_txt): + """Bloc long : ecrit dans les deux langues plutot que traduit chaine par + chaine. Au-dela de deux phrases, la table de traduction devient illisible et + les oublis passent inapercus.""" + return en_txt if LANG == "en" else fr_txt + + +def nf(v, d=0): + """Ecriture du nombre selon la langue.""" + if v is None or (isinstance(v, float) and not np.isfinite(v)): + return t("n.d.") + brut = f"{v:,.{d}f}" + return brut if LANG == "en" else brut.replace(",", " ").replace(".", ",") + + +# ------------------------------------------------------------------ palettes +# Une teinte porte un sens et ne change jamais de sens d'un graphique a +# l'autre : sepia pour les titres, bleu pour les maisons, vert pour ce qui est +# preserve, violet pour les etapes de la recuperation. Chaque rampe est d'une +# seule teinte, du clair au fonce. +# Rampes sequentielles : une seule teinte, et le pas le plus proche du fond +# reste nettement distinct de lui — une valeur basse doit se lire comme une +# valeur, jamais comme du papier nu. +# +# Le mode sombre n'est PAS un retournement automatique du mode clair : ce sont +# les memes teintes, re-echelonnees pour la surface sombre. Sur papier, la +# valeur forte est la plus foncee ; sur une surface sombre, la valeur foncee +# recule au lieu de ressortir, donc la valeur forte y est la plus claire. Sans +# cela, le Nigeria disparaissait dans le fond alors qu'il est en tete. +_RAMPES = { + "clair": { + "sepia": ["#f0dcc2", "#d9a86e", "#b87c40", "#7d4c1c", "#4a2c0f"], + "bleu": ["#cde2fb", "#86b6ef", "#3987e5", "#256abf", "#0d366b"], + "vert": ["#c9ead9", "#7fcaa8", "#3aa87a", "#1f8a5f", "#0d5c3d"], + }, + "sombre": { + "sepia": ["#6b4520", "#8a5a28", "#b87c40", "#d9a86e", "#f2e0c8"], + "bleu": ["#123a63", "#1c5cab", "#3987e5", "#86b6ef", "#cde2fb"], + "vert": ["#0f4f35", "#1f8a5f", "#3aa87a", "#7fcaa8", "#cdeee0"], + }, +} +SEPIA = _RAMPES[THEME]["sepia"] +BLEUS = _RAMPES[THEME]["bleu"] +VERTS = _RAMPES[THEME]["vert"] + +# Slots categoriels, ordre fixe, jamais recycles ni etendus : deux teintes +# voisines restent distinguables en deuteranopie comme en vision normale. +CAT = ["#2a78d6", "#eb6834", "#1baf7a", "#eda100", "#e87ba4", "#008300", + "#4a3aa7", "#e34948"] +# Teintes d'etat, reservees : elles ne servent jamais de « serie 4 ». +STATUT = {"bon": "#1f8a5f", "attention": "#c78a12", "critique": "#c0392b"} +# Les villes de l'atlas : un bleu franc, etranger a toutes les rampes de fond. +VILLE = "#2a78d6" if THEME == "clair" else "#6da7ec" + +REGIONS = ["West Africa", "Central Africa", "East Africa", "Horn of Africa", + "North Africa", "Southern Africa", "Indian Ocean", "Diaspora (Europe)"] +REGION_FR = {"West Africa": "Afrique de l'Ouest", "Central Africa": "Afrique centrale", + "East Africa": "Afrique de l'Est", "Horn of Africa": "Corne de l'Afrique", + "North Africa": "Afrique du Nord", "Southern Africa": "Afrique australe", + "Indian Ocean": "Océan Indien", "Diaspora (Europe)": "Diaspora (Europe)"} +REGION_COUL = dict(zip(REGIONS, CAT)) + +alt.data_transformers.disable_max_rows() + +# Streamlit a remplace use_container_width par width="stretch" en 1.49 : on +# choisit selon la version installee, pour que l'application tourne aussi bien +# sur un Space recent que sur un environnement plus ancien. +_V = tuple(int(x) for x in st.__version__.split(".")[:2] if x.isdigit()) +LARGE = {"width": "stretch"} if _V >= (1, 49) else {"use_container_width": True} +ETROIT = {"width": "content"} if _V >= (1, 49) else {"use_container_width": False} + + +def theme(ch): + """Meme grammaire visuelle pour tous les graphiques, dans les deux themes. + + Le fond reste transparent : Vega peint un fond blanc par defaut, qui ferait + une dalle claire au milieu d'une page sombre.""" + return (ch.configure(background="transparent") + .configure_view(strokeWidth=0) + .configure_axis(grid=True, gridColor=PAL["grille"], gridDash=[2, 3], + domainColor=PAL["axe"], tickColor=PAL["axe"], + labelColor=PAL["texte"], titleColor=PAL["doux"], + labelFontSize=11.5, titleFontSize=11.5, labelLimit=280, + labelOverlap=False, titleFontWeight="normal", + labelPadding=5) + .configure_legend(labelFontSize=11.5, titleFontSize=11.5, + labelColor=PAL["texte"], titleColor=PAL["doux"], + titleFontWeight="normal", symbolType="square", + symbolSize=110) + .configure_title(fontSize=15, fontWeight=600, color=PAL["encre"], + anchor="start", offset=12, subtitleColor=PAL["doux"], + subtitleFontSize=11.5, subtitlePadding=6, + subtitleFontWeight="normal") + .configure_bar(cornerRadiusEnd=3)) + + +def tracer(ch, hauteur=360): + st.altair_chart(theme(ch.properties(height=hauteur)), **LARGE) + + +def tracer_fixe(ch): + """Graphique aux dimensions imposees — indispensable pour les cartes. + + Streamlit ajoute d'office `autosize: fit` a toute specification qui n'en + porte pas : la hauteur declaree devient alors la hauteur TOTALE, titre, axes + et legende compris, et la carte est comprimee dans ce qui reste. En + declarant `autosize: pad`, largeur et hauteur redeviennent celles de la zone + de trace et les decorations s'ajoutent autour.""" + ch = ch.properties(autosize=alt.AutoSizeParams(type="pad", contains="padding")) + try: + st.altair_chart(theme(ch), **ETROIT) + except TypeError: + st.altair_chart(theme(ch)) + + +# -------------------------------------------------------------- chargements +def trouver(nom): + p = ICI / nom + return p if p.exists() else None + + +@st.cache_data(show_spinner=False) +def table(nom): + p = trouver(nom) + return pd.DataFrame() if p is None else pd.read_csv(p) + + +T = table("titles.csv") +P = table("presses.csv") +RIGHTS = table("rights.csv") +DIGI = table("digitization.csv") +HOLD = table("holdings.csv") +CITIES = table("cities.csv") +COUNTRIES = table("countries.csv") + +if T.empty or P.empty or COUNTRIES.empty: + st.error("titles.csv / presses.csv / countries.csv absent — " + "lancer `python build_data.py`.") + st.stop() + +STAGES = [ + ("stage1_located", "1 · Located", "1 · Localisé", + "A catalogue record or a copy has been identified.", + "Une notice de catalogue ou un exemplaire a été identifié."), + ("stage2_preserved", "2 · Preserved", "2 · Préservé", + "A conservation-quality scan exists.", + "Une numérisation de qualité conservation existe."), + ("stage3_rights_identified", "3 · Rights identified", "3 · Ayants droit identifiés", + "The author, estate or successor publisher has been traced.", + "L'auteur, ses ayants droit ou l'éditeur successeur ont été retrouvés."), + ("stage4_cleared", "4 · Cleared", "4 · Autorisé", + "Written permission to publish has been obtained.", + "L'autorisation écrite de publier a été obtenue."), + ("stage5_republished", "5 · Republished", "5 · Réédité", + "The book is legally available again.", + "Le livre est de nouveau disponible légalement."), + ("stage6_curriculum_ready", "6 · Curriculum-ready", "6 · Prêt pour l'enseignement", + "Teaching notes, vocabulary, activities and context exist.", + "Notes pédagogiques, vocabulaire, activités et contexte existent."), +] + +PILOT_CIBLES = { + "A · Onitsha market pamphlets": 25, + "B · Ghanaian children's & folklore": 20, + "C · East African popular fiction": 20, + "D · Francophone (CLE, NEA & related)": 20, + "E · African-language works": 15, +} +RAISON = { + "A · Onitsha market pamphlets": ( + "Popular writing produced and sold outside every colonial and missionary " + "institution — the first large-scale independent book trade in British West " + "Africa. It teaches literacy, commerce and voice at once, and almost none of " + "it is reliably dated, which makes it a good exercise in reading evidence.", + "Une écriture populaire produite et vendue en dehors de toute institution " + "coloniale ou missionnaire — le premier commerce du livre indépendant à " + "grande échelle en Afrique de l'Ouest britannique. Elle enseigne à la fois " + "l'alphabétisation, le commerce et la voix, et presque rien n'y est daté de " + "façon sûre : bon exercice de lecture des sources."), + "B · Ghanaian children's & folklore": ( + "The most immediately usable material in the catalogue and the most poorly " + "recorded. The Ghana Publishing Corporation's Children's Library series was " + "explicitly commissioned as books by Africans for African children — the " + "exact brief this project inherits.", + "Le matériel le plus immédiatement utilisable du catalogue, et le plus mal " + "enregistré. La collection Children's Library de la Ghana Publishing " + "Corporation fut explicitement commandée comme des livres d'Africains pour " + "des enfants africains — exactement le programme dont ce projet hérite."), + "C · East African popular fiction": ( + "Novels and plays from Nairobi and Dar es Salaam at the height of the East " + "African Publishing House and Tanzania Publishing House, including work by " + "West and Southern African writers who published there. Kenya and Tanzania's " + "life + 50 term also makes this the cluster most likely to be clearable.", + "Romans et pièces de Nairobi et Dar es Salaam à l'apogée de l'East African " + "Publishing House et de la Tanzania Publishing House, y compris des auteurs " + "d'Afrique de l'Ouest et australe qui y publiaient. La durée vie + 50 du " + "Kenya et de la Tanzanie en fait aussi le groupe le plus facile à libérer."), + "D · Francophone (CLE, NEA & related)": ( + "Éditions CLE in Yaoundé and the Nouvelles Éditions Africaines in Dakar are " + "the institutionally anchored African francophone houses, and CLE still " + "exists — a living counterparty rather than an orphan-works problem. " + "Présence Africaine belongs alongside them but was Paris-based, and the " + "catalogue says so.", + "Les Éditions CLE à Yaoundé et les Nouvelles Éditions Africaines à Dakar " + "sont les maisons francophones africaines réellement ancrées, et CLE existe " + "toujours : un interlocuteur vivant plutôt qu'un problème d'œuvres " + "orphelines. Présence Africaine leur est associée mais était parisienne, et " + "le catalogue le dit."), + "E · African-language works": ( + "Swahili, Hausa, Yoruba, Igbo, Shona, isiZulu, Akan, Sesotho, Malagasy and " + "more. The least digitised and least catalogued part of the corpus, and the " + "part whose loss is least recoverable from anywhere else.", + "Swahili, haoussa, yoruba, igbo, shona, isiZulu, akan, sesotho, malgache et " + "d'autres. La part la moins numérisée et la moins cataloguée du corpus, et " + "celle dont la perte est la moins réparable ailleurs."), +} +CONF = ["high", "medium", "low"] -def hero(eyebrow: str, title: str, blurb: str): - st.markdown( - f'
{eyebrow}
' - f"

{title}

{blurb}

", - unsafe_allow_html=True, - ) +def lreg(r): + return r if LANG == "en" else REGION_FR.get(r, r) -def note(text: str, kind: str = "note"): - cls = "abrp-warn" if kind == "warn" else "abrp-note" - st.markdown(f'
{text}
', unsafe_allow_html=True) +# ------------------------------------------------------------------- cartes +# Emprise : le bord ouest va jusqu'a -25,5° pour que le Cabo Verde (-23,8°) +# figure sur les cartes. Un pays present dans un classement et absent de toute +# carte se remarque a la relecture. +LON0, LON1 = -25.5, 52.5 +LAT0, LAT1 = -36.0, 38.0 +RAPPORT = (LON1 - LON0) * np.cos(np.deg2rad(5.0)) / (LAT1 - LAT0) -def bar_h(df, x, y, tooltip, title="", height=None, sort=None, fmt="d"): - """Horizontal bar. One series → no legend; the title names it. - Every category keeps its own label: labelOverlap is off and the plot is - given enough height per band that Vega never drops a tick.""" - h = height or max(130, 26 * len(df) + 44) - return ( - alt.Chart(df, title=title) - .mark_bar(cornerRadiusEnd=4, color=P.series[0], height=14) - .encode( - x=alt.X(f"{x}:Q", title=None, axis=alt.Axis(format=fmt, tickCount=4)), - y=alt.Y(f"{y}:N", title=None, sort=sort or "-x", - axis=alt.Axis(labelOverlap=False, labelLimit=190)), - tooltip=tooltip, - ) - .properties(height=h) - ) +@st.cache_data(show_spinner=False) +def sommets(): + p = trouver("contours_afrique.csv") + if p is None: + return pd.DataFrame(columns=["pays", "piece", "ordre", "lon", "lat"]) + return pd.read_csv(p) -def footer(): - st.markdown( - '
African Books Recovery Project · every record in this ' - "dashboard carries its sources and a researcher confidence flag. Undated means " - "undated: no year has been estimated. Rights information is a factual description " - "of published legal materials, not legal advice.
", - unsafe_allow_html=True, - ) - - -# --------------------------------------------------------------------------- -# 1. Overview -# --------------------------------------------------------------------------- - -def page_overview(): - hero( - "Observatory", - "African Books Recovery Project", - "Books by African authors, published by presses based in Africa, c. 1950–1979 — the " - "pamphlet houses of Onitsha, the state and mission presses, the independence-era " - "literary imprints. Most of this writing left circulation without ever entering a " - "durable catalogue. This observatory tracks what has been located, what survives in " - "scannable form, and what stands between a located book and a book a teacher can use.", - ) - - dated = TITLES[TITLES["year"].notna()] - africa_presses = PRESSES[PRESSES["based_in_africa"]] +def _axe(vals, neg, pos, horizontal): + return alt.Axis(values=vals, title=None, + labelExpr=(f"datum.value == 0 ? '0°'" + f" : datum.value < 0 ? abs(datum.value) + '° {neg}'" + f" : datum.value + '° {pos}'"), + labelFontSize=10, labelColor=PAL["doux"], grid=True, + gridColor=PAL["grille"], gridDash=[1, 3], + tickColor=PAL["axe"], domainColor=PAL["axe"], + labelFlush=False if horizontal else alt.Undefined) + + +def carte(valeurs, titre, couleurs, unite, largeur=520, sous_titre="", villes=None): + """Choroplèthe des 54 pays d'Afrique. + + Aucune valeur n'est ecrite dans les polygones : a cette densite elles se + chevauchent. Elles se lisent au survol et sur le classement qui accompagne + chaque carte. Un pays sans notice est gris — jamais interpole, jamais rendu + comme un pays a faible valeur : le sous-titre le compte et le dit.""" + ct = sommets() + if ct.empty: + st.info(bil("`contours_afrique.csv` missing: maps unavailable.", + "`contours_afrique.csv` absent : cartes indisponibles.")) + return + vals = {k: float(v) for k, v in valeurs.items() + if v is not None and np.isfinite(v) and v > 0} + df = ct.copy() + df[t("Country")] = df["pays"] + df["valeur"] = df["pays"].map(lambda p: vals.get(p, np.nan)) + # Colonne booleenne explicite plutot qu'un test isValid() sur la valeur : un + # NaN pandas ne traverse pas toujours la serialisation en null, et le pays + # sans notice reprenait alors le bas de la rampe — exactement la confusion + # que la regle « le gris n'est jamais zero » interdit. + df["a_notice"] = df["valeur"].notna() + df["etiquette"] = df["valeur"].map( + lambda v: bil("no record yet", "aucune notice à ce jour") + if not np.isfinite(v) else f"{int(v)}") + n_sans = int((~df.groupby("pays")["a_notice"].first()).sum()) + + ex = alt.Scale(domain=[LON0, LON1], nice=False, zero=False) + ey = alt.Scale(domain=[LAT0, LAT1], nice=False, zero=False) + lg = alt.Legend(orient="bottom", direction="horizontal", + gradientLength=min(largeur, 300), gradientThickness=10, + titleFontSize=11, labelFontSize=10.5, titleLimit=420, format="d") + # DEUX COUCHES PLUTOT QU'UNE CONDITION + # Une condition Vega sur la valeur nulle ne survit pas au passage par + # Streamlit : le pays sans notice reprenait le bas de la rampe, c'est-a-dire + # exactement la lecture que ce tableau de bord interdit — « pas cherche » ne + # doit jamais ressembler a « peu ». On peint donc d'abord les 54 pays dans la + # teinte d'absence, puis on repeint par-dessus les seuls pays qui ont une + # notice. Ce qui reste gris est gris parce que rien n'a ete peint dessus. + geo = dict(x=alt.X("lon:Q", scale=ex, + axis=_axe(list(range(-10, 51, 10)), t("W"), t("E"), True)), + y=alt.Y("lat:Q", scale=ey, + axis=_axe(list(range(-30, 31, 10)), t("S"), t("N"), False)), + detail=alt.Detail("piece:N"), order=alt.Order("ordre:Q")) + infobulle = [alt.Tooltip(f"{t('Country')}:N"), + alt.Tooltip("etiquette:N", title=unite)] + absent = (alt.Chart(df).mark_line(strokeWidth=0.7, stroke=PAL["trait"], + fill=PAL["absent"]) + .encode(tooltip=infobulle, **geo)) + porte = (alt.Chart(df[df["a_notice"]]).mark_line(strokeWidth=0.7, + stroke=PAL["trait"]) + .encode(fill=alt.Fill("valeur:Q", title=unite, legend=lg, + scale=alt.Scale(range=couleurs, type="sqrt")), + tooltip=infobulle, **geo)) + couches = [absent, porte] + if villes is not None and len(villes): + v = villes.copy() + v[t("City")] = v["city"] + couches.append( + # Le point de ville ne doit pas emprunter une teinte de la rampe du + # fond, sinon on ne sait plus si l'on lit un pays ou une ville. + alt.Chart(v).mark_point(filled=True, opacity=0.92, stroke=PAL["fond"], + strokeWidth=1.0, color=VILLE) + .encode(x=alt.X("lon:Q", scale=ex), y=alt.Y("lat:Q", scale=ey), + size=alt.Size("titles:Q", title=t("Titles"), + scale=alt.Scale(type="sqrt", range=[24, 520]), + legend=alt.Legend(orient="bottom", format="d", + symbolType="circle", + symbolFillColor=VILLE, + symbolStrokeColor=PAL["fond"])), + tooltip=[alt.Tooltip(f"{t('City')}:N"), + alt.Tooltip("titles:Q", title=t("Titles"), format="d"), + alt.Tooltip("presses:Q", title=t("Houses"), format="d")])) + st_ = sous_titre + if n_sans: + gris = bil(f"{n_sans} countries in grey — no record yet, which is not the " + "same as nothing to find", + f"{n_sans} pays en gris — aucune notice à ce jour, ce qui n'est " + "pas la même chose que rien à trouver") + st_ = f"{sous_titre} · {gris}" if sous_titre else gris + tracer_fixe(alt.layer(*couches).properties(width=largeur, height=int(round(largeur / RAPPORT)), + title=alt.TitleParams(titre, subtitle=st_))) + + +def barres(d, champ, libelle, titre, sous_titre="", couleur=None, hauteur=None, + ordre="-x", cle="label", fmt="d"): + """Barres horizontales. Une seule serie : pas de legende, le titre nomme.""" + coul = couleur if couleur is not None else alt.value(PAL["accent"]) + ch = (alt.Chart(d).mark_bar(height=13) + .encode(y=alt.Y(f"{cle}:N", sort=ordre, title=None), + x=alt.X(f"{champ}:Q", title=libelle, + axis=alt.Axis(tickCount=4, format=fmt)), + color=coul, + tooltip=[alt.Tooltip(f"{cle}:N", title=""), + alt.Tooltip(f"{champ}:Q", title=libelle, format=fmt)]) + .properties(title=alt.TitleParams(titre, subtitle=sous_titre))) + tracer(ch, hauteur or (22 * len(d) + 96)) + + +def classement(d, champ, libelle, titre, n=16, cle="label"): + """Le panneau qui accompagne chaque carte : la carte porte la forme, le + classement porte les chiffres. C'est la contrepartie de la regle « aucune + valeur ecrite dans les polygones ».""" + dd = d.nlargest(n, champ) + barres(dd, champ, libelle, titre, cle=cle, hauteur=22 * len(dd) + 86) + + +# ------------------------------------------------------------------- tuiles +def tuiles(items): + """items : (valeur, cle, sous-texte).""" + cells = "".join( + f'
{v}
' + f'
{k}
{s}
' + for v, k, s in items) + st.markdown(f'
{cells}
', unsafe_allow_html=True) + + +def note(txt, kind="note"): + st.markdown(f'
{txt}
', unsafe_allow_html=True) + + +def fiche(titre, meta, corps, pied=""): st.markdown( - T.tiles( - [ - (f"{len(TITLES)}", "titles catalogued", f"{len(dated)} dated, {len(TITLES) - len(dated)} undated"), - (f"{len(PRESSES)}", "presses in the register", f"{len(africa_presses)} with an African imprint"), - (f"{TITLES['country'].nunique()}", "countries of publication", f"{TITLES['region'].nunique()} regions"), - (f"{TITLES['language'].nunique()}", "languages of publication", - f"{int(TITLES['african_language'].sum())} titles in African languages"), - (f"{int(TITLES['stage2_preserved'].sum())}", "with a known scan", - f"{int(TITLES['stage2_preserved'].sum())/len(TITLES):.1%} of the catalogue"), - (f"{int(TITLES['stage6_curriculum_ready'].sum())}", "curriculum-ready", - "no teaching apparatus written yet"), - ] - ), - unsafe_allow_html=True, - ) - - note( - "Read the last two tiles together. The catalogue is the easy part. Of " - f"{len(TITLES)} titles, {int(TITLES['stage2_preserved'].sum())} have a scan anyone can point to and none has " - "teaching material. That distance is the project." - ) + f'

{titre}

' + + (f'
{meta}
' if meta else "") + + (f"

{corps}

" if corps else "") + + (f'
{pied}
' if pied else "") + + "
", unsafe_allow_html=True) + + +def puce(conf): + cls = {"high": "hi", "medium": "me", "low": "lo"}.get(conf, "") + return f'{t(conf)}' + + +def liens(row, limite=4): + urls = [u for u in str(row.get("source_urls", "")).split(" | ") if u.strip()] + noms = str(row.get("source_titles", "")).split(" | ") + out = [] + for i, u in enumerate(urls[:limite]): + lab = noms[i].strip() if i < len(noms) and noms[i].strip() else u.split("/")[2] + out.append(f'{lab}') + reste = len(urls) - limite + if reste > 0: + out.append(f"+{reste}") + return " · ".join(out) if out else "—" + + +def txt(v, defaut=""): + return v if isinstance(v, str) and v.strip() else defaut + + +# --------------------------------------------------- barre de reglages et CSS +def choix(options, cle, defaut=None, format_func=None, ou=st, etiquette=None): + """Liste cliquable — jamais de menu deroulant. + + Un menu deroulant cache ses options : il faut cliquer pour savoir ce qu'il + contient. Ici tout est visible et se prend en un clic. On essaie `pills`, + puis `segmented_control`, puis des boutons radio, de sorte que l'application + tourne aussi sur une version anterieure de Streamlit.""" + ff = format_func or (lambda x: x) + # Un libelle reel, meme masque : Streamlit journalise un avertissement pour + # chaque widget a libelle vide, et un journal de Space couvert + # d'avertissements finit par cacher les vraies erreurs. + etiquette = etiquette or cle.replace("_", " ") + d = options[0] if defaut is None else defaut + for nom in ("pills", "segmented_control"): + f = getattr(ou, nom, None) + if f is None: + continue + try: + v = f(etiquette, options, default=d, format_func=ff, + key=f"{nom}_{cle}", label_visibility="collapsed") + if v is None: + # Recliquer l'option deja selectionnee la deselectionne : on + # remet le defaut plutot que de laisser une rangee vide en face + # d'une carte qui, elle, affiche toujours quelque chose. + st.session_state[f"{nom}_{cle}"] = d + st.rerun() + return v + except Exception: # noqa: BLE001 + continue + return ou.radio(etiquette, options, index=options.index(d), format_func=ff, + horizontal=True, key=f"radio_{cle}", + label_visibility="collapsed") + + +def _bascule(libelles, cle): + inverse = {v: k for k, v in libelles.items()} + courant = inverse[st.session_state[cle]] + try: + c = st.segmented_control(cle, list(libelles), default=courant, + key=f"seg_{cle}", label_visibility="collapsed") + except Exception: # noqa: BLE001 + c = st.radio(cle, list(libelles), index=list(libelles).index(courant), + horizontal=True, key=f"rad_{cle}", + label_visibility="collapsed") + if c and libelles[c] != st.session_state[cle]: + st.session_state[cle] = libelles[c] + st.rerun() + + +_r1, _r2, _r3 = st.columns([1.0, 1.0, 3.4]) +with _r1: + _bascule(dict(zip(THEMES_CLES, ["clair", "sombre"])), "theme") +with _r2: + _bascule(LANGUES, "lang") + +# Le theme de base de Streamlit ne se change pas a l'execution — cela se regle +# dans un fichier de configuration, or le Space doit rester a plat, sans +# sous-dossier .streamlit. On repeint donc l'habillage en CSS ; la palette des +# graphiques, elle, est entierement sous notre controle. +st.markdown(f"""""", unsafe_allow_html=True) + + +# --------------------------------------------------------------- barre laterale +with st.sidebar: + st.markdown(f"### 📚 {t('African Books Recovery Project')}") + st.caption(t("African authors · African presses · c. 1950–1979")) + st.divider() + st.markdown(f"**{t('Filters')}**") + st.caption(bil("They apply to the Atlas, Presses and Catalogue tabs.", + "Ils s'appliquent aux onglets Atlas, Maisons et Catalogue.")) + f_reg = st.multiselect(t("Region"), REGIONS, format_func=lreg, key="f_reg") + f_dec = st.multiselect( + t("Decade"), + [d for d in ["1940s", "1950s", "1960s", "1970s", "1980s", "Undated"] + if d in set(T["decade"])], + format_func=lambda d: t(d) if d == "Undated" else d, key="f_dec") + f_lang = st.multiselect(t("Language"), sorted(T["language"].unique()), key="f_lang") + f_conf = st.multiselect(t("Confidence"), CONF, format_func=t, key="f_conf") + q = st.text_input(t("Search"), key="f_q", + placeholder=bil("title, author or press", + "titre, auteur ou maison")) + st.divider() + st.caption(bil( + "Every record carries its sources and a confidence flag. Nothing is " + "estimated: an unknown year stays undated. Rights information describes " + "published statutes and is not legal advice.", + "Chaque notice porte ses sources et un indice de fiabilité. Rien n'est " + "estimé : une année inconnue reste non datée. L'information juridique " + "décrit des textes publiés et ne vaut pas conseil juridique.")) + + +def filtrer(d): + if f_reg: + d = d[d["region"].isin(f_reg)] + if f_lang and "language" in d.columns: + d = d[d["language"].isin(f_lang)] + if f_dec and "decade" in d.columns: + d = d[d["decade"].isin(f_dec)] + if f_conf: + d = d[d["confidence"].isin(f_conf)] + if q: + ql = q.lower() + cols = [c for c in ["title", "author", "press_name", "name", "description"] + if c in d.columns] + m = pd.Series(False, index=d.index) + for c in cols: + m = m | d[c].astype(str).str.lower().str.contains(ql, na=False) + d = d[m] + return d + + +TF = filtrer(T) +PF = filtrer(P) +FILTRE = bool(f_reg or f_lang or f_dec or f_conf or q) + + +def bandeau_filtre(): + if FILTRE: + note(bil(f"Filtered view — {len(TF)} of {len(T)} titles, " + f"{len(PF)} of {len(P)} presses. Clear the filters in the sidebar " + "to see the whole catalogue.", + f"Vue filtrée — {len(TF)} titres sur {len(T)}, {len(PF)} " + f"maisons sur {len(P)}. Videz les filtres dans la barre latérale " + "pour revoir l'ensemble.")) + + +# ------------------------------------------------------------------- entete +st.markdown(f"# {t('African Books Recovery Project')}") +st.markdown(bil( + "Books written by African authors and published by presses **based in " + "Africa**, roughly 1950 to 1979 — the pamphlet houses of Onitsha, the state " + "and mission presses, the university imprints, the independence-era literary " + "houses of Ibadan, Accra, Nairobi, Dar es Salaam, Yaoundé, Dakar, Lusaka and " + "Johannesburg. Most of this writing left circulation without ever entering a " + "durable catalogue.", + "Des livres écrits par des auteurs africains et publiés par des maisons " + "**établies en Afrique**, de 1950 à 1979 environ — les imprimeurs de pamphlets " + "d'Onitsha, les presses d'État et de mission, les presses universitaires, les " + "maisons littéraires des indépendances à Ibadan, Accra, Nairobi, Dar es Salaam, " + "Yaoundé, Dakar, Lusaka et Johannesburg. L'essentiel de cette écriture a quitté " + "la circulation sans jamais entrer dans un catalogue durable.")) + +onglets = st.tabs([t("Overview"), t("Atlas"), t("Presses"), t("Catalogue"), + t("Recovery"), t("Rights"), t("Copies"), t("Methods")]) +N_ZERO = int((~COUNTRIES["searched"]).sum()) + +# ============================================================ 1. VUE D'ENSEMBLE +with onglets[0]: + dated = T[T["year"].notna()] + tuiles([ + (nf(len(T)), t("titles catalogued"), + bil(f"{len(dated)} dated · {len(T) - len(dated)} undated", + f"{len(dated)} datés · {len(T) - len(dated)} non datés")), + (nf(len(P)), t("presses in the register"), + bil(f"{int(P['based_in_africa'].sum())} with an African imprint", + f"{int(P['based_in_africa'].sum())} à adresse africaine")), + (nf(T["country"].nunique()), t("countries of publication"), + bil(f"{T['region'].nunique()} regions", f"{T['region'].nunique()} régions")), + (nf(T["language"].nunique()), t("languages of publication"), + bil(f"{int(T['african_language'].sum())} titles in African languages", + f"{int(T['african_language'].sum())} titres en langues africaines")), + (nf(int(T["stage2_preserved"].sum())), t("with a known scan"), + f"{int(T['stage2_preserved'].sum()) / len(T):.1%} " + + bil("of the catalogue", "du catalogue")), + (nf(N_ZERO), t("countries with no record yet"), + bil("half the continent, unsearched", "la moitié du continent, non explorée")), + ]) + + note(bil( + "Read the last two tiles together. The catalogue is the easy part. " + f"Of {len(T)} titles, {int(T['stage2_preserved'].sum())} have a scan anyone " + f"can point to, none has teaching material, and {N_ZERO} African countries " + "carry no record here at all. That distance — between a book that exists " + "and a book a teacher can use — is the project.", + "Lisez les deux dernières tuiles ensemble. Le catalogue est la partie " + f"facile. Sur {len(T)} titres, {int(T['stage2_preserved'].sum())} ont une " + "numérisation que l'on peut montrer, aucun n'a de matériel pédagogique, et " + f"{N_ZERO} pays africains ne portent ici aucune notice. Cette distance — " + "entre un livre qui existe et un livre qu'un enseignant peut utiliser — " + "c'est le projet.")) c1, c2 = st.columns([1.15, 1]) - with c1: - dec = ( - TITLES.groupby("decade").size().reset_index(name="titles") - .sort_values("decade") - ) - order = [d for d in ["1940s", "1950s", "1960s", "1970s", "1980s", "Undated"] if d in set(dec["decade"])] - ch = ( - alt.Chart(dec, title="Titles by decade of publication") - .mark_bar(cornerRadiusEnd=4, color=P.series[0], size=44) - .encode( - x=alt.X("decade:N", title=None, sort=order), - y=alt.Y("titles:Q", title="titles", axis=alt.Axis(tickCount=4)), - tooltip=[alt.Tooltip("decade:N", title="Decade"), alt.Tooltip("titles:Q", title="Titles")], - ) - .properties(height=250) - ) - st.altair_chart(ch, width="stretch") - st.caption( - "The 1960s peak is the independence decade: state publishing corporations, " - "university presses and the Onitsha trade at full volume. Undated records are " - "almost all Onitsha pamphlets, where library dates are conjectural." - ) - + dec = T.groupby("decade").size().reset_index(name="n") + ordre = [d for d in ["1940s", "1950s", "1960s", "1970s", "1980s", "Undated"] + if d in set(dec["decade"])] + dec["label"] = dec["decade"].map(lambda d: t(d) if d == "Undated" else d) + ch = (alt.Chart(dec).mark_bar(size=46, color=PAL["accent"]) + .encode(x=alt.X("label:N", title=None, + sort=[t(d) if d == "Undated" else d for d in ordre], + axis=alt.Axis(labelAngle=0)), + y=alt.Y("n:Q", title=t("Titles"), axis=alt.Axis(tickCount=4)), + tooltip=[alt.Tooltip("label:N", title=t("Decade")), + alt.Tooltip("n:Q", title=t("Titles"))]) + .properties(title=alt.TitleParams( + bil("Titles by decade of publication", + "Titres par décennie de publication"), + subtitle=bil("The 1960s peak is the independence decade.", + "Le pic des années 1960 est celui des indépendances.")))) + tracer(ch, 300) + st.caption(bil( + "State publishing corporations, university presses and the Onitsha trade " + "all at full volume. Undated records are almost all Onitsha pamphlets, " + "where library dates are conjectural — shown as undated rather than " + "given a plausible decade.", + "Sociétés d'édition d'État, presses universitaires et commerce d'Onitsha " + "à plein régime. Les notices non datées sont presque toutes des " + "pamphlets d'Onitsha, dont les dates de bibliothèque sont conjecturales : " + "elles restent non datées plutôt que rattachées à une décennie plausible.")) with c2: - reg = TITLES.groupby("region").size().reset_index(name="titles").sort_values("titles", ascending=False) - st.altair_chart( - bar_h(reg, "titles", "region", - [alt.Tooltip("region:N", title="Region"), alt.Tooltip("titles:Q", title="Titles")], - title="Titles by region of publication"), - width="stretch", - ) - st.caption( + reg = T.groupby("region").size().reset_index(name="n") + reg["label"] = reg["region"].map(lreg) + barres(reg, "n", t("Titles"), + bil("Titles by region of publication", + "Titres par région de publication"), hauteur=300, + couleur=alt.Color("region:N", sort=REGIONS, legend=None, + scale=alt.Scale(domain=REGIONS, + range=[REGION_COUL[r] for r in REGIONS]))) + st.caption(bil( "Coverage reflects where the research reached, not where the books are. " - "North Africa, the Horn and Lusophone Africa are thin and should be read as gaps." - ) + "North Africa, the Horn and Lusophone Africa are thin and should be read " + "as gaps in the search, never as gaps in the record.", + "La couverture reflète l'avancée de la recherche, pas la géographie " + "réelle des livres. L'Afrique du Nord, la Corne et l'Afrique lusophone " + "sont minces : ce sont des lacunes de la recherche, jamais du fonds.")) st.divider() c3, c4 = st.columns(2) - with c3: - own = ( - PRESSES.groupby("ownership").size().reset_index(name="presses") - .sort_values("presses", ascending=False) - ) - st.altair_chart( - bar_h(own, "presses", "ownership", - [alt.Tooltip("ownership:N", title="Ownership"), alt.Tooltip("presses:Q", title="Presses")], - title="Who owned the presses"), - width="stretch", - ) - st.caption( - "Ownership decides who holds the rights today. African-owned houses that closed " - "leave orphan works; branches of foreign firms leave rights with a company that " - "still exists — and often still says no." - ) - + own = P.groupby("ownership").size().reset_index(name="n") + own["label"] = own["ownership"] + barres(own, "n", t("Houses"), + bil("Who owned the presses", "À qui appartenaient les maisons"), + hauteur=250) + st.caption(bil( + "Ownership decides who holds the rights today. An African-owned house " + "that closed leaves orphan works; a branch of a foreign firm leaves the " + "rights with a company that still exists — and often still says no.", + "La propriété décide de qui détient les droits aujourd'hui. Une maison " + "africaine disparue laisse des œuvres orphelines ; une filiale de groupe " + "étranger laisse les droits à une société qui existe toujours — et qui " + "refuse souvent.")) with c4: - lang = ( - TITLES.groupby("language").size().reset_index(name="titles") - .sort_values("titles", ascending=False).head(12) - ) - st.altair_chart( - bar_h(lang, "titles", "language", - [alt.Tooltip("language:N", title="Language"), alt.Tooltip("titles:Q", title="Titles")], - title="Languages of publication (top 12)"), - width="stretch", - ) - st.caption( - "English and French dominate the surviving record. That is partly what was " - "published and partly what got catalogued — African-language imprints were the " - "least likely to reach a library that indexed them." - ) + lg = T.groupby("language").size().reset_index(name="n").nlargest(12, "n") + lg["label"] = lg["language"] + barres(lg, "n", t("Titles"), + bil("Languages of publication (top 12)", + "Langues de publication (12 premières)"), hauteur=340) + st.caption(bil( + "English and French dominate the surviving record. That is partly what " + "was published and partly what got catalogued — African-language " + "imprints were the least likely to reach a library that indexed them.", + "L'anglais et le français dominent le fonds conservé. C'est en partie ce " + "qui fut publié, en partie ce qui fut catalogué : les impressions en " + "langues africaines avaient le moins de chances d'atteindre une " + "bibliothèque qui les indexait.")) st.divider() - st.markdown("##### What the evidence rests on") - conf = ( - TITLES.groupby("confidence").size().reindex(CONF_ORDER).fillna(0) - .reset_index(name="titles").rename(columns={"index": "confidence"}) - ) - conf["confidence"] = pd.Categorical(conf["confidence"], CONF_ORDER, ordered=True) - cc = ( - alt.Chart(conf, title="Researcher confidence in each title record") - .mark_bar(cornerRadiusEnd=4, height=18) - .encode( - x=alt.X("titles:Q", title=None), - y=alt.Y("confidence:N", title=None, sort=CONF_ORDER, - axis=alt.Axis(labelOverlap=False)), - color=alt.Color("confidence:N", sort=CONF_ORDER, legend=None, - scale=alt.Scale(domain=CONF_ORDER, - range=[P.status["good"], P.status["warning"], P.status["critical"]])), - tooltip=[alt.Tooltip("confidence:N", title="Confidence"), alt.Tooltip("titles:Q", title="Titles")], - ) - .properties(height=145) - ) - st.altair_chart(cc, width="stretch") - st.caption( - "High = a library or publisher catalogue record. Medium = a scholarly source or a " - "dealer listing transcribed from a title page. Low = a single unconfirmed mention. " - "Nothing here is a guess; low-confidence rows are leads, not facts." - ) - footer() - - -# --------------------------------------------------------------------------- -# 2. Recovery pipeline -# --------------------------------------------------------------------------- - -def page_pipeline(): - hero( - "Pipeline", - "From located to curriculum-ready", - "A recovered book passes through six states. Only the first is bibliographic; the " - "rest are physical, legal and editorial work. Tracking them separately keeps the " - "project honest about what a catalogue entry does and does not mean.", - ) - - st.write("") - sc = D.stage_counts(TITLES) - base = alt.Chart(sc).encode( - y=alt.Y("stage:N", title=None, sort=list(sc["stage"]), - axis=alt.Axis(labelOverlap=False, labelLimit=220)), - ) - bars = base.mark_bar(cornerRadiusEnd=4, height=20).encode( - x=alt.X("count:Q", title="titles", - scale=alt.Scale(domain=[0, len(TITLES) * 1.06]), - axis=alt.Axis(tickCount=5, format="d")), - color=alt.Color("stage:N", sort=list(sc["stage"]), legend=None, - scale=alt.Scale(domain=list(sc["stage"]), range=P.ordinal)), - tooltip=[alt.Tooltip("stage:N", title="Stage"), - alt.Tooltip("count:Q", title="Titles"), - alt.Tooltip("definition:N", title="Means")], - ) - labels = base.mark_text(align="left", dx=7, fontSize=11, color=P.text2).encode( - x=alt.X("count:Q"), text=alt.Text("count:Q", format="d") - ) - st.altair_chart( - (bars + labels).properties(height=290, title="Titles at each stage of recovery"), - width="stretch", - ) - - note( - "These are independent flags, not a funnel. Six titles are commercially in " - "print (stage 5) because a publisher reissued them, not because this project traced " - "their rights. Stages 3, 4 and 6 stand at zero: no estate has been traced, no " - "permission obtained, no teaching apparatus written. That is the true starting line." - ) - - st.markdown("##### What each stage requires") - for col, label, defn in D.STAGES: - n = int(TITLES[col].sum()) - st.markdown( - f'

{label} — {n} titles

' - f'
{defn}
', - unsafe_allow_html=True, - ) + st.markdown("##### " + bil("What the evidence rests on", "Sur quoi repose la preuve")) + cf = T.groupby("confidence").size().reindex(CONF).fillna(0).reset_index() + cf.columns = ["confidence", "n"] + cf["label"] = cf["confidence"].map(t) + ch = (alt.Chart(cf).mark_bar(height=17) + .encode(y=alt.Y("label:N", sort=[t(c) for c in CONF], title=None), + x=alt.X("n:Q", title=t("Titles")), + color=alt.Color("confidence:N", legend=None, + scale=alt.Scale(domain=CONF, + range=[STATUT["bon"], + STATUT["attention"], + STATUT["critique"]])), + tooltip=[alt.Tooltip("label:N", title=t("Confidence")), + alt.Tooltip("n:Q", title=t("Titles"))])) + tracer(ch, 150) + st.caption(bil( + "High = a library or publisher catalogue record. Medium = a scholarly " + "source, or a dealer listing transcribed from a title page. Low = a single " + "unconfirmed mention. Nothing here is a guess; low-confidence rows are " + "leads, not facts, and the app never hides which is which.", + "Élevée = une notice de catalogue de bibliothèque ou d'éditeur. Moyenne = " + "une source savante, ou une annonce de libraire transcrite d'une page de " + "titre. Faible = une seule mention non confirmée. Rien ici n'est deviné : " + "les lignes peu fiables sont des pistes, pas des faits, et l'application ne " + "cache jamais lesquelles.")) + +# ==================================================================== 2. ATLAS +with onglets[1]: + bandeau_filtre() + st.markdown("### " + bil("The geography of a vanished publishing industry", + "La géographie d'une édition disparue")) + st.caption(bil( + "Country outlines: Natural Earth 1:50m, public domain. City dots are placed " + "from a public gazetteer and are a cartographic aid only — they are not part " + "of any bibliographic record.", + "Contours des pays : Natural Earth 1:50m, domaine public. Les points de " + "villes sont placés d'après un répertoire géographique public et ne sont " + "qu'une aide à la lecture : ils ne font partie d'aucune notice.")) + + MESURES = { + bil("Titles catalogued", "Titres au catalogue"): ("titles", SEPIA), + bil("Presses in the register", "Maisons au registre"): ("presses", BLEUS), + bil("Titles with a scan", "Titres numérisés"): ("preserved", VERTS), + } + mes = choix(list(MESURES), "atlas_mesure") + champ, rampe = MESURES[mes] + + cc = COUNTRIES.copy() + if FILTRE: + # Le filtre doit se voir sur la carte : sans cela, choisir une region ne + # changerait rien de visible ici et le clic passerait pour sans effet. + tt = TF.groupby("map_country").size().rename("titles") + pp = PF.groupby("map_country").size().rename("presses") + pv = TF[TF["stage2_preserved"]].groupby("map_country").size().rename("preserved") + cc = (cc.drop(columns=["titles", "presses", "preserved"]) + .merge(pd.concat([tt, pp, pv], axis=1), left_on="map_country", + right_index=True, how="left")) + for c_ in ("titles", "presses", "preserved"): + cc[c_] = cc[c_].fillna(0).astype(int) + + m1, m2 = st.columns([1.34, 1]) + with m1: + villes = None + if champ == "titles": + villes = CITIES[CITIES["on_map"]] + if FILTRE: + vt = TF.groupby("city").size().rename("n") + villes = (villes.drop(columns=["titles"]) + .merge(vt, left_on="city", right_index=True, how="inner") + .rename(columns={"n": "titles"})) + carte(dict(zip(cc["map_country"], cc[champ])), mes, rampe, mes, + largeur=520, villes=villes) + with m2: + cl = cc[cc[champ] > 0].copy() + cl["label"] = cl["country"] + if len(cl): + classement(cl, champ, mes, bil("Ranking", "Classement")) + else: + st.info(bil("Nothing to rank under these filters.", + "Rien à classer avec ces filtres.")) + + if champ == "titles": + st.caption(bil( + "The dots are the cities that actually printed: Onitsha, Ibadan, Accra, " + "Nairobi, Dar es Salaam, Johannesburg, Dakar, Yaoundé. Their size is the " + "number of titles catalogued there. A whole country can be coloured by a " + "single city — Nigeria's shade is Onitsha and Ibadan, not Nigeria.", + "Les points sont les villes qui imprimaient réellement : Onitsha, " + "Ibadan, Accra, Nairobi, Dar es Salaam, Johannesburg, Dakar, Yaoundé. " + "Leur taille est le nombre de titres qui y sont catalogués. Un pays " + "entier peut être coloré par une seule ville — la teinte du Nigéria, " + "c'est Onitsha et Ibadan, pas le Nigéria.")) + + zero = sorted(COUNTRIES[~COUNTRIES["searched"]]["country"].tolist()) + note(bil( + f"The grey is the finding. {len(zero)} of the 54 countries on this " + "map carry no record at all — " + ", ".join(zero[:12]) + + f" and {len(zero) - 12} others. Not one of them lacked a printing press " + "between 1950 and 1979. They are grey because this catalogue has not " + "searched them, and the map is drawn so that the unsearched can never be " + "mistaken for the empty.", + f"Le gris est le résultat. {len(zero)} des 54 pays de cette carte ne " + "portent aucune notice — " + ", ".join(zero[:12]) + + f" et {len(zero) - 12} autres. Aucun d'eux n'était dépourvu d'imprimerie " + "entre 1950 et 1979. Ils sont gris parce que ce catalogue ne les a pas " + "explorés, et la carte est faite pour qu'on ne confonde jamais le " + "non-exploré avec le vide."), "warn") st.divider() - st.markdown("### The first phase: a pilot of 100 titles") - st.caption( - "Five clusters chosen so that each tests a different recovery problem — an " - "uncatalogued pamphlet trade, a children's list, a popular-fiction backlist, a " - "francophone imprint with a living publisher, and African-language texts." - ) - - pilot = TITLES[TITLES["pilot_cluster"] != ""] - rows = [] - for cluster, target in D.PILOT_TARGETS.items(): - got = int((TITLES["pilot_cluster"] == cluster).sum()) - rows.append({"cluster": cluster, "assigned": got, "target": target, "shortfall": target - got}) - pdf = pd.DataFrame(rows) - - base = alt.Chart(pdf).encode( - y=alt.Y("cluster:N", title=None, sort=list(pdf["cluster"]), - axis=alt.Axis(labelOverlap=False, labelLimit=260)) - ) - target_bar = base.mark_bar(cornerRadiusEnd=4, height=16, color=P.grid).encode( - x=alt.X("target:Q", title="titles"), - tooltip=[alt.Tooltip("cluster:N", title="Cluster"), alt.Tooltip("target:Q", title="Target")], - ) - got_bar = base.mark_bar(cornerRadiusEnd=4, height=16, color=P.series[0]).encode( - x=alt.X("assigned:Q"), - tooltip=[alt.Tooltip("cluster:N", title="Cluster"), - alt.Tooltip("assigned:Q", title="Assigned"), - alt.Tooltip("target:Q", title="Target")], - ) - st.altair_chart( - (target_bar + got_bar).properties(height=215, title="Pilot clusters: assigned against target"), - width="stretch", - ) - st.caption("Grey is the target; blue is what the catalogue can currently fill.") - - short = pdf[pdf["shortfall"] > 0] - if len(short): - items = "; ".join(f"{r.cluster} is {r.shortfall} short" for r in short.itertuples()) - note( - f"{items}. This is a real finding, not a data problem: Ghanaian children's and " - "folklore publishing of the period is severely under-catalogued, and the Bureau " - "of Ghana Languages output — the largest body of it — is barely indexed anywhere. " - "Filling this cluster needs on-site work in Accra and Tamale, not more searching.", - kind="warn", - ) - - st.markdown("##### Pilot titles") - cl = st.selectbox("Cluster", ["All clusters"] + list(D.PILOT_TARGETS), key="pilot_cl") - show = pilot if cl == "All clusters" else pilot[pilot["pilot_cluster"] == cl] - st.dataframe( - show[["title", "author", "press_name", "city", "year", "language", "genre", - "digitization_status", "confidence", "pilot_cluster"]] - .sort_values(["pilot_cluster", "year"], na_position="last"), - width="stretch", hide_index=True, - column_config={ - "title": st.column_config.TextColumn("Title", width="medium"), - "press_name": st.column_config.TextColumn("Press", width="medium"), - "year": st.column_config.NumberColumn("Year", format="%d"), - "digitization_status": "Scan status", - "pilot_cluster": st.column_config.TextColumn("Cluster", width="medium"), - }, - ) - footer() - - -# --------------------------------------------------------------------------- -# 3. Press register -# --------------------------------------------------------------------------- - -def page_presses(): - hero( - "Register", - "The presses", - "The publishing houses themselves — who owned them, what they printed, when they " - "stopped. A press record is the fastest route to a rights holder: a house that still " - "trades can license its backlist, a house that closed leaves orphan works.", - ) - - c1, c2, c3 = st.columns(3) - with c1: - regions = st.multiselect("Region", sorted(PRESSES["region"].unique()), key="pr_reg") - with c2: - owner = st.multiselect("Ownership", sorted(PRESSES["ownership"].unique()), key="pr_own") - with c3: - kinds = st.multiselect("Type", sorted(PRESSES["type"].unique()), key="pr_type") - - df = PRESSES.copy() - if regions: - df = df[df["region"].isin(regions)] - if owner: - df = df[df["ownership"].isin(owner)] - if kinds: - df = df[df["type"].isin(kinds)] - - counts = TITLES.groupby("press_id").size().rename("titles_in_catalogue") - df = df.merge(counts, left_on="press_id", right_index=True, how="left") - df["titles_in_catalogue"] = df["titles_in_catalogue"].fillna(0).astype(int) - - st.markdown( - T.tiles([ - (f"{len(df)}", "presses shown", f"of {len(PRESSES)} in the register"), - (f"{int(df['based_in_africa'].sum())}", "African imprint", "printed on the continent"), - (f"{df['country'].nunique()}", "countries", ""), - (f"{int(df['titles_in_catalogue'].sum())}", "titles attached", "in this catalogue"), - ]), - unsafe_allow_html=True, - ) - - founded = df[df["founded"].notna()] - if len(founded): - fc = ( - alt.Chart(founded, title="Presses by decade of founding") - .mark_bar(cornerRadiusEnd=4, color=P.series[0], size=40) - .encode( - x=alt.X("founded:Q", bin=alt.Bin(step=10), title="decade founded", - axis=alt.Axis(format="d")), - y=alt.Y("count():Q", title="presses", axis=alt.Axis(tickCount=4)), - tooltip=[alt.Tooltip("count():Q", title="Presses")], - ) - .properties(height=200) - ) - st.altair_chart(fc, width="stretch") - st.caption( - f"{len(PRESSES) - len(founded)} presses in the register have no verified founding " - "year and are excluded from this chart rather than dated by inference — most are " - "Onitsha pamphlet houses that never registered as companies." - ) + v1, v2 = st.columns(2) + with v1: + cv = CITIES.nlargest(16, "titles").copy() + cv["label"] = cv["city"] + barres(cv, "titles", t("Titles"), + bil("The publishing cities", "Les villes d'édition"), + sous_titre=bil("Titles catalogued, by city of imprint", + "Titres catalogués, par ville d'impression"), + couleur=alt.value(VILLE)) + with v2: + rr = T[T["decade"] != "Undated"].groupby(["region", "decade"]).size() + rr = rr.reset_index(name="n") + rr["label"] = rr["region"].map(lreg) + ordre_r = [r for r in REGIONS if r in set(rr["region"])] + ch = (alt.Chart(rr).mark_bar(stroke=PAL["fond"], strokeWidth=1.5) + .encode(x=alt.X("decade:N", title=None), + y=alt.Y("n:Q", title=t("Titles"), stack=True), + color=alt.Color("label:N", title=None, + sort=[lreg(r) for r in ordre_r], + scale=alt.Scale(domain=[lreg(r) for r in ordre_r], + range=[REGION_COUL[r] for r in ordre_r]), + legend=alt.Legend(orient="bottom", columns=2)), + tooltip=[alt.Tooltip("label:N", title=t("Region")), + alt.Tooltip("decade:N", title=t("Decade")), + alt.Tooltip("n:Q", title=t("Titles"))]) + .properties(title=alt.TitleParams( + bil("Where publishing happened, decade by decade", + "Où l'on publiait, décennie par décennie"), + subtitle=bil("Dated titles only.", "Titres datés seulement.")))) + tracer(ch, 430) + +# ================================================================= 3. MAISONS +with onglets[2]: + bandeau_filtre() + n_titres = TF.groupby("press_id").size().rename("n") + PP = PF.merge(n_titres, left_on="press_id", right_index=True, how="left") + PP["n"] = PP["n"].fillna(0).astype(int) + + tuiles([ + (nf(len(PP)), t("Houses"), bil(f"of {len(P)} in the register", + f"sur {len(P)} au registre")), + (nf(int(PP["based_in_africa"].sum())), + bil("African imprint", "adresse africaine"), + bil("printed on the continent", "imprimé sur le continent")), + (nf(PP["country"].nunique()), t("Country"), + bil("of publication", "de publication")), + (nf(int(PP["n"].sum())), t("Titles"), bil("attached here", "rattachés ici")), + ]) + + if len(PP): + g1, g2 = st.columns(2) + with g1: + ty = PP.groupby("type").size().reset_index(name="n") + ty["label"] = ty["type"] + barres(ty, "n", t("Houses"), + bil("What kind of house", "Quel type de maison"), hauteur=250) + with g2: + fo = PP[PP["founded"].notna()] + if len(fo): + ch = (alt.Chart(fo).mark_bar(size=34, color=PAL["accent"]) + .encode(x=alt.X("founded:Q", bin=alt.Bin(step=10), + title=bil("decade founded", + "décennie de fondation"), + axis=alt.Axis(format="d")), + y=alt.Y("count():Q", title=t("Houses"), + axis=alt.Axis(tickCount=4)), + tooltip=[alt.Tooltip("count():Q", title=t("Houses"))]) + .properties(title=alt.TitleParams( + bil("Presses by decade of founding", + "Maisons par décennie de fondation"), + subtitle=bil( + f"{len(PP) - len(fo)} presses have no verified founding " + "year and are excluded rather than dated by inference.", + f"{len(PP) - len(fo)} maisons n'ont pas d'année de " + "fondation vérifiée : exclues plutôt que datées par " + "déduction.")))) + tracer(ch, 250) st.divider() - sort_by = st.radio("Order by", ["Titles in catalogue", "Founding year", "Name"], - horizontal=True, key="pr_sort") - if sort_by == "Titles in catalogue": - df = df.sort_values("titles_in_catalogue", ascending=False) - elif sort_by == "Founding year": - df = df.sort_values("founded", na_position="last") + OPT_TRI = [bil("Most titles here", "Le plus de titres ici"), + bil("Oldest first", "Les plus anciennes"), bil("By name", "Par nom")] + tri = choix(OPT_TRI, "tri_presses") + if tri == OPT_TRI[0]: + PP = PP.sort_values("n", ascending=False) + elif tri == OPT_TRI[1]: + PP = PP.sort_values("founded", na_position="last") else: - df = df.sort_values("name") - - for r in df.head(40).itertuples(): - series = f" · series: {r.notable_series}" if r.notable_series else "" - figures = f" · key figures: {r.key_figures}" if r.key_figures else "" - flag = "" if r.based_in_africa else ' imprint outside Africa' - st.markdown( - f'

{r.name}{flag}

' - f'
{r.city}, {r.country} · {r.active_span} · {r.ownership} · ' - f"{r.type} · {r.titles_in_catalogue} titles here{figures}{series}
" - f"

{r.description}

" - f'

{r.significance}

' - f'
{T.chip(r.confidence)}
', - unsafe_allow_html=True, - ) - st.markdown( - f'
' - f"Sources: {D.source_links(r._asdict())}
", - unsafe_allow_html=True, - ) - if len(df) > 40: - st.caption(f"Showing the first 40 of {len(df)} presses. Narrow the filters to see the rest.") - footer() - - -# --------------------------------------------------------------------------- -# 4. Title catalogue -# --------------------------------------------------------------------------- - -def page_catalogue(): - hero( - "Catalogue", - "The title record", - "Every row is a book someone can go and find. The record is deliberately provisional: " - "where a year, a scan or a rights holder is unknown, it says unknown rather than " - "filling the gap.", - ) - - c1, c2, c3, c4 = st.columns(4) - with c1: - f_region = st.multiselect("Region", sorted(TITLES["region"].unique()), key="t_reg") - with c2: - f_lang = st.multiselect("Language", sorted(TITLES["language"].unique()), key="t_lang") - with c3: - f_genre = st.multiselect("Genre group", sorted(TITLES["genre_group"].unique()), key="t_gen") - with c4: - f_conf = st.multiselect("Confidence", CONF_ORDER, key="t_conf") - - c5, c6, c7 = st.columns([2, 1, 1]) - with c5: - q = st.text_input("Search title, author or press", key="t_q", placeholder="e.g. Ogali, Onitsha, Rosa Mistika") - with c6: - f_dig = st.multiselect("Scan status", sorted(TITLES["digitization_status"].unique()), key="t_dig") - with c7: - african_only = st.checkbox("African-language titles only", key="t_afr") - - df = TITLES.copy() - if f_region: - df = df[df["region"].isin(f_region)] - if f_lang: - df = df[df["language"].isin(f_lang)] - if f_genre: - df = df[df["genre_group"].isin(f_genre)] - if f_conf: - df = df[df["confidence"].isin(f_conf)] - if f_dig: - df = df[df["digitization_status"].isin(f_dig)] - if african_only: - df = df[df["african_language"]] - if q: - ql = q.lower() - df = df[ - df["title"].str.lower().str.contains(ql, na=False) - | df["author"].str.lower().str.contains(ql, na=False) - | df["press_name"].str.lower().str.contains(ql, na=False) - ] + PP = PP.sort_values("name") + + if not len(PP): + st.info(bil("No press matches these filters.", + "Aucune maison ne correspond à ces filtres.")) + for r in PP.head(30).itertuples(): + drapeau = ("" if r.based_in_africa else ' ' + + bil("imprint outside Africa", "adresse hors d'Afrique") + "") + figs = txt(r.key_figures) + figs = f" · {bil('key figures', 'figures')} : {figs}" if figs else "" + ser = txt(r.notable_series) + ser = f" · {bil('series', 'collections')} : {ser}" if ser else "" + fiche(f"{r.name}{drapeau}", + f"{r.city}, {r.country} · {r.active_span} · {r.ownership} · {r.type} · " + f"{r.n} {t('Titles').lower()}{figs}{ser}", + txt(r.description), + f"{puce(r.confidence)} {t('Sources')} : {liens(r._asdict())}") + if len(PP) > 30: + st.caption(bil(f"Showing the first 30 of {len(PP)} presses. Narrow the " + "filters to see the rest.", + f"30 premières maisons sur {len(PP)}. Affinez les filtres " + "pour voir les autres.")) + +# =============================================================== 4. CATALOGUE +with onglets[3]: + bandeau_filtre() + tuiles([ + (nf(len(TF)), t("Titles"), bil(f"of {len(T)}", f"sur {len(T)}")), + (nf(TF["press_id"].nunique()), t("Houses"), + bil("with a title in this view", "avec un titre dans cette vue")), + (nf(int(TF["stage2_preserved"].sum())), t("Scan"), + bil("open or restricted", "ouverte ou sous condition")), + (nf(int((TF["in_print"] == "no").sum())), + bil("confirmed out of print", "épuisés confirmés"), + bil(f"{int((TF['in_print'] == 'unknown').sum())} unverified", + f"{int((TF['in_print'] == 'unknown').sum())} non vérifiés")), + ]) + + if not len(TF): + st.info(bil("No title matches these filters.", + "Aucun titre ne correspond à ces filtres.")) + else: + k1, k2 = st.columns([1.2, 1]) + with k1: + gg = TF.groupby("genre_group").size().reset_index(name="n") + gg["label"] = gg["genre_group"] + barres(gg, "n", t("Titles"), + bil("What was actually published", "Ce qui fut réellement publié")) + with k2: + dom = ["full_text_open", "scanned_restricted", "catalog_only", "unknown"] + dg = TF.groupby("digitization_status").size() + dg = dg.reindex(dom).fillna(0).reset_index() + dg.columns = ["digitization_status", "n"] + dg["label"] = dg["digitization_status"].str.replace("_", " ") + ch = (alt.Chart(dg).mark_bar(height=17) + .encode(y=alt.Y("label:N", title=None, + sort=[d.replace("_", " ") for d in dom]), + x=alt.X("n:Q", title=t("Titles")), + color=alt.Color("digitization_status:N", legend=None, + scale=alt.Scale(domain=dom, + range=[STATUT["bon"], CAT[0], + STATUT["attention"], + STATUT["critique"]])), + tooltip=[alt.Tooltip("label:N", title=t("Scan")), + alt.Tooltip("n:Q", title=t("Titles"))]) + .properties(title=alt.TitleParams( + bil("Does a scan exist?", "Une numérisation existe-t-elle ?"), + subtitle=bil("'catalog only' means a record exists but no " + "scan. 'unknown' means nobody has checked.", + "« catalog only » : une notice existe, pas de " + "numérisation. « unknown » : personne n'a " + "vérifié.")))) + tracer(ch, 240) + + st.markdown("##### " + bil("Records", "Notices")) + cols = ["title", "author", "press_name", "city", "country", "year", "language", + "genre", "in_print", "digitization_status", "confidence", + "primary_source_url"] + st.dataframe( + TF[cols].sort_values(["year", "title"], na_position="last"), + hide_index=True, height=430, **LARGE, + column_config={ + "title": st.column_config.TextColumn(t("Title"), width="medium"), + "author": st.column_config.TextColumn(t("Author"), width="medium"), + "press_name": st.column_config.TextColumn(t("Press"), width="medium"), + "city": t("City"), "country": t("Country"), + "year": st.column_config.NumberColumn(t("Year"), format="%d"), + "language": t("Language"), "genre": t("Genre"), + "in_print": t("In print"), "digitization_status": t("Scan"), + "confidence": t("Confidence"), + "primary_source_url": st.column_config.LinkColumn(t("Sources"), + display_text="↗")}) + st.download_button( + bil("Download this selection (CSV)", "Télécharger cette sélection (CSV)"), + TF.to_csv(index=False).encode("utf-8"), + file_name="abrp_selection.csv", mime="text/csv") - st.markdown( - T.tiles([ - (f"{len(df)}", "titles shown", f"of {len(TITLES)}"), - (f"{df['press_id'].nunique()}", "presses", ""), - (f"{int(df['stage2_preserved'].sum())}", "with a scan", "open or restricted"), - (f"{int((df['in_print'] == 'no').sum())}", "confirmed out of print", - f"{int((df['in_print'] == 'unknown').sum())} unverified"), - ]), - unsafe_allow_html=True, - ) - - if len(df) == 0: - st.info("No titles match those filters.") - footer() - return + st.divider() + st.markdown("### " + bil("Title dossier", "Dossier de titre")) + st.caption(bil( + "The full record for one book: bibliography, provenance, scan status, " + "the press behind it, the rights context of its country — and, " + "deliberately, what is still missing.", + "La notice complète d'un livre : bibliographie, provenance, état de " + "numérisation, la maison qui l'a publié, le contexte juridique de son " + "pays — et, délibérément, ce qui manque encore.")) + + def _lab(tid): + r = TF[TF["title_id"] == tid].iloc[0] + y = int(r["year"]) if pd.notna(r["year"]) else t("n.d.") + return f"{r['title']} — {r['author']} ({y})" + + tid = st.selectbox(bil("Choose a title", "Choisir un titre"), + TF["title_id"].tolist(), format_func=_lab, key="dossier") + r = TF[TF["title_id"] == tid].iloc[0] + an = int(r["year"]) if pd.notna(r["year"]) else t("undated") + banni = ('' + bil("banned", "interdit") + "" + if str(r.get("banned")) == "True" else "") + fiche(r["title"], + f"{r['author']} · {r['press_name']}, {r['city']} · {an} · " + f"{r['language']} · {r['genre']}", "", + puce(r["confidence"]) + + f'' + f'{str(r["digitization_status"]).replace("_", " ")}' + + f'{t("In print")} : {r["in_print"]}' + + banni) + + d1, d2 = st.columns(2) + with d1: + st.markdown("**" + bil("Bibliographic record", + "Notice bibliographique") + "**") + champs = { + t("Author"): r["author"], t("Title"): r["title"], + t("Language"): r["language"], t("Country"): r["country"], + t("City"): r["city"], t("Press"): r["press_name"], t("Year"): an, + "Pages": (int(r["pages"]) if pd.notna(r["pages"]) + else bil("unrecorded", "non renseigné")), + t("Genre"): r["genre"]} + st.markdown("\n".join(f"- **{k}** · {v}" for k, v in champs.items())) + st.markdown("**" + bil("Still to record", "Reste à renseigner") + "**") + st.markdown("\n".join("- " + x for x in bil( + "Author's dates (nothing can be calculated without them)|" + "Alternative title, edition and printing|" + "ISBN, if one was ever assigned|" + "Illustrator, cover artist and translator|" + "Physical condition of a located copy|" + "Rights holder and rights status|Digitization permission|" + "Subjects, age group and curriculum relevance|" + "Links to original-language, translated and revised editions", + "Dates de l'auteur (rien ne se calcule sans elles)|" + "Titre alternatif, édition et tirage|ISBN, s'il en existe un|" + "Illustrateur, maquettiste et traducteur|" + "État matériel d'un exemplaire localisé|" + "Ayant droit et statut des droits|Autorisation de numériser|" + "Sujets, tranche d'âge et usage pédagogique|" + "Liens entre édition originale, traduite et révisée").split("|"))) + with d2: + st.markdown("**" + bil("Provenance and notes", "Provenance et notes") + "**") + st.markdown(txt(r["notes"], + bil("_No researcher note on this record._", + "_Aucune note de recherche sur cette notice._"))) + st.markdown("**" + t("Sources") + "**") + st.markdown(liens(r.to_dict(), 6), unsafe_allow_html=True) + if txt(r["digitization_url"]): + st.markdown(f"**{t('Scan')}** · " + f"[{txt(r['digitization_holder'], '↗')}]" + f"({r['digitization_url']})") + pr = P[P["press_id"] == r["press_id"]] + if len(pr): + pr = pr.iloc[0] + st.markdown("**" + bil("The press", "La maison") + "**") + st.markdown(f"{pr['name']} — {pr['city']}, {pr['country']} · " + f"{pr['active_span']} · {pr['ownership']}.") + if txt(pr["archive_holdings"]): + st.markdown("**" + bil("Known holdings", "Fonds connus") + "** · " + + pr["archive_holdings"]) + st.markdown("**" + bil("Rights context", "Contexte juridique") + "**") + rr = RIGHTS[RIGHTS["country"] == r["country"]] + if len(rr): + rr = rr.iloc[0] + st.markdown(f"{rr['country']} — {rr['copyright_term']} " + f"({rr['statute']}). {rr['recovery_implication']}") + else: + st.markdown(bil("Not yet researched for this country.", + "Pas encore étudié pour ce pays.")) + +# ============================================================== 5. RECUPERATION +with onglets[4]: + st.markdown("### " + bil("From located to curriculum-ready", + "De « localisé » à « prêt pour l'enseignement »")) + st.caption(bil( + "A recovered book passes through six states. Only the first is " + "bibliographic; the rest are physical, legal and editorial work.", + "Un livre récupéré traverse six états. Seul le premier est " + "bibliographique ; les autres sont matériels, juridiques et éditoriaux.")) + + sc = pd.DataFrame([{"label": (en if LANG == "en" else fr_), "n": int(T[k].sum()), + "def": (den if LANG == "en" else dfr)} + for k, en, fr_, den, dfr in STAGES]) + base = alt.Chart(sc).encode(y=alt.Y("label:N", sort=list(sc["label"]), title=None)) + bars = base.mark_bar(height=19, color=PAL["accent"]).encode( + x=alt.X("n:Q", title=t("Titles"), scale=alt.Scale(domain=[0, len(T) * 1.08]), + axis=alt.Axis(tickCount=5, format="d")), + tooltip=[alt.Tooltip("label:N", title=t("Stage")), + alt.Tooltip("n:Q", title=t("Titles")), + alt.Tooltip("def:N", title="")]) + etiq = base.mark_text(align="left", dx=7, fontSize=11.5, + color=PAL["texte"]).encode(x="n:Q", text=alt.Text("n:Q")) + tracer_fixe((bars + etiq).properties( + width=640, height=250, + title=alt.TitleParams(bil("Titles at each stage of recovery", + "Titres à chaque ��tape de la récupération"), + subtitle=bil("Independent flags, not a funnel.", + "Indicateurs indépendants, pas un entonnoir.")))) + note(bil( + "These are independent flags, not a funnel. Six titles are " + "commercially in print (stage 5) because a publisher reissued them, not " + "because this project traced their rights. Stages 3, 4 and 6 stand at zero: " + "no estate traced, no permission obtained, no teaching apparatus written. " + "That is the true starting line, and the dashboard refuses to flatter it.", + "Ce sont des indicateurs indépendants, pas un entonnoir. Six titres " + "sont commercialement disponibles (étape 5) parce qu'un éditeur les a " + "réédités, non parce que ce projet a retrouvé leurs ayants droit. Les étapes " + "3, 4 et 6 sont à zéro : aucun ayant droit retrouvé, aucune autorisation " + "obtenue, aucun appareil pédagogique écrit. C'est la vraie ligne de départ, " + "et le tableau de bord refuse de l'embellir.")) + with st.expander(bil("What each stage requires", "Ce qu'exige chaque étape")): + for k, en, fr_, den, dfr in STAGES: + st.markdown(f"**{en if LANG == 'en' else fr_}** — {int(T[k].sum())} " + f"{t('Titles').lower()} \n{den if LANG == 'en' else dfr}") - left, right = st.columns([1.25, 1]) - with left: - gg = df.groupby("genre_group").size().reset_index(name="titles").sort_values("titles", ascending=False) - st.altair_chart( - bar_h(gg, "titles", "genre_group", - [alt.Tooltip("genre_group:N", title="Genre"), alt.Tooltip("titles:Q", title="Titles")], - title="What was actually published"), - width="stretch", - ) - with right: - dg = df.groupby("digitization_status").size().reset_index(name="titles") - dom = ["full_text_open", "scanned_restricted", "catalog_only", "unknown"] - dg["digitization_status"] = pd.Categorical(dg["digitization_status"], dom, ordered=True) - dg = dg.sort_values("digitization_status") - dch = ( - alt.Chart(dg, title="Does a scan exist?") - .mark_bar(cornerRadiusEnd=4, height=18) - .encode( - x=alt.X("titles:Q", title=None), - y=alt.Y("digitization_status:N", title=None, sort=dom, - axis=alt.Axis(labelOverlap=False, labelLimit=180)), - color=alt.Color("digitization_status:N", legend=None, - scale=alt.Scale(domain=dom, - range=[P.status["good"], P.series[0], - P.status["warning"], P.status["critical"]])), - tooltip=[alt.Tooltip("digitization_status:N", title="Status"), - alt.Tooltip("titles:Q", title="Titles")], - ) - .properties(height=175) - ) - st.altair_chart(dch, width="stretch") - st.caption("'catalog_only' means a record exists but no scan. 'unknown' means nobody has checked.") - - st.markdown("##### Records") - st.dataframe( - df[["title", "author", "press_name", "city", "country", "year", "language", "genre", - "in_print", "digitization_status", "confidence", "primary_source_url"]] - .sort_values(["year", "title"], na_position="last"), - width="stretch", hide_index=True, height=430, - column_config={ - "title": st.column_config.TextColumn("Title", width="medium"), - "press_name": st.column_config.TextColumn("Press", width="medium"), - "year": st.column_config.NumberColumn("Year", format="%d"), - "in_print": "In print", - "digitization_status": "Scan", - "primary_source_url": st.column_config.LinkColumn("Source", display_text="open"), - }, - ) - st.download_button( - "Download this selection as CSV", - df.to_csv(index=False).encode("utf-8"), - file_name="abrp_titles_selection.csv", - mime="text/csv", - ) + st.divider() + st.markdown("### " + bil("The first phase: a pilot of 100 titles", + "Première phase : un pilote de 100 titres")) + st.caption(bil( + "Five clusters, each testing a different recovery problem — an uncatalogued " + "pamphlet trade, a children's list, a popular-fiction backlist, a " + "francophone imprint with a living publisher, and African-language texts.", + "Cinq groupes, chacun éprouvant un problème différent — un commerce de " + "pamphlets non catalogué, un fonds pour la jeunesse, un catalogue de fiction " + "populaire, une maison francophone encore vivante, et des textes en langues " + "africaines.")) + + pil = pd.DataFrame([{"cluster": c, "assigned": int((T["pilot_cluster"] == c).sum()), + "target": n} for c, n in PILOT_CIBLES.items()]) + b = alt.Chart(pil).encode(y=alt.Y("cluster:N", sort=list(pil["cluster"]), title=None)) + cible = b.mark_bar(height=15, color=PAL["absent"]).encode( + x=alt.X("target:Q", title=t("Titles"), axis=alt.Axis(tickCount=5, format="d")), + tooltip=[alt.Tooltip("cluster:N", title=t("Cluster")), + alt.Tooltip("target:Q", title=bil("Target", "Cible"))]) + fait = b.mark_bar(height=15, color=PAL["accent"]).encode( + x="assigned:Q", + tooltip=[alt.Tooltip("cluster:N", title=t("Cluster")), + alt.Tooltip("assigned:Q", title=bil("Assigned", "Affectés")), + alt.Tooltip("target:Q", title=bil("Target", "Cible"))]) + tracer_fixe((cible + fait).properties( + width=640, height=215, + title=alt.TitleParams(bil("Pilot clusters: assigned against target", + "Groupes du pilote : affectés / cible"), + subtitle=bil("Grey is the target; colour is what the " + "catalogue can currently fill.", + "Le gris est la cible ; la couleur est ce " + "que le catalogue peut remplir.")))) + court = pil[pil["assigned"] < pil["target"]] + if len(court): + n_ = int((court["target"] - court["assigned"]).sum()) + note(bil( + f"Cluster B is {n_} titles short, and that is a finding rather than a " + "data problem. Ghanaian children's and folklore publishing of the " + "period is severely under-catalogued: the Bureau of Ghana Languages " + "published in eleven languages from 1951 and almost none of it is indexed " + "anywhere online. Filling this cluster needs on-site work in Accra and " + "Tamale, not more searching from a desk.", + f"Le groupe B est incomplet de {n_} titres, et c'est un résultat, non " + "un défaut de données. L'édition ghanéenne pour la jeunesse et le " + "folklore de cette période est très mal cataloguée : le Bureau of Ghana " + "Languages a publié en onze langues à partir de 1951 et presque rien n'en " + "est indexé en ligne. Compléter ce groupe demande un travail sur place à " + "Accra et Tamale, pas davantage de recherche à distance."), "warn") + + for cl, cible_n in PILOT_CIBLES.items(): + sub = T[T["pilot_cluster"] == cl] + if not len(sub): + continue + sp = sub["year"].dropna() + etendue = f"{int(sp.min())}–{int(sp.max())}" if len(sp) else t("undated") + with st.expander(f"{cl} — {len(sub)}/{cible_n} · {etendue} · " + + ", ".join(sorted(sub["language"].unique())[:5])): + st.markdown(f"**{RAISON[cl][0 if LANG == 'en' else 1]}**") + st.dataframe( + sub[["title", "author", "press_name", "year", "language", "genre", + "confidence"]].sort_values("year", na_position="last"), + hide_index=True, **LARGE, + column_config={"title": t("Title"), "author": t("Author"), + "press_name": t("Press"), + "year": st.column_config.NumberColumn(t("Year"), + format="%d"), + "language": t("Language"), "genre": t("Genre"), + "confidence": t("Confidence")}) st.divider() - st.markdown("### Title dossier") - st.caption("The full record for one book: bibliography, provenance, scan status and what is still missing.") - pick = st.selectbox( - "Choose a title", - df["title_id"].tolist(), - format_func=lambda tid: _label(df, tid), - key="dossier", - ) - _dossier(df[df["title_id"] == pick].iloc[0]) - footer() - - -def _label(df, tid): - r = df[df["title_id"] == tid].iloc[0] - yr = int(r["year"]) if pd.notna(r["year"]) else "n.d." - return f"{r['title']} — {r['author']} ({yr})" - - -def _dossier(r): - yr = int(r["year"]) if pd.notna(r["year"]) else "undated" - press = PRESSES[PRESSES["press_id"] == r["press_id"]] - press = press.iloc[0] if len(press) else None - - banned_chip = 'banned' if r["banned"] == True else "" # noqa: E712 - scan_label = str(r["digitization_status"]).replace("_", " ") - st.markdown( - f'

{r["title"]}

' - f'
{r["author"]} · {r["press_name"]}, {r["city"]} · {yr} · ' - f'{r["language"]} · {r["genre"]}
' - f'
{T.chip(r["confidence"])}' - f'{scan_label}' - f'in print: {r["in_print"]}' - f"{banned_chip}
", - unsafe_allow_html=True, - ) - - a, b = st.columns(2) - with a: - st.markdown("**Bibliographic record**") - fields = { - "Author": r["author"], - "Title": r["title"], - "Language": r["language"], - "Country of publication": r["country"] or "unrecorded", - "City of publication": r["city"] or "unrecorded", - "Publisher": r["press_name"], - "Year": yr, - "Pages": int(r["pages"]) if pd.notna(r["pages"]) else "unrecorded", - "Genre": r["genre"], - } - st.markdown("\n".join(f"- **{k}** · {v}" for k, v in fields.items())) - - st.markdown("**Still to record**") - missing = [ - "Author's dates (needed before any copyright calculation)", - "Alternative title, edition and printing", - "ISBN, if one was ever assigned", - "Illustrator, cover artist and translator", - "Physical condition of a located copy", - "Rights holder and rights status", - "Digitization permission", - "Subjects, age group and curriculum relevance", - "Links to original-language, translated and revised editions", - ] - st.markdown("\n".join(f"- {m}" for m in missing)) - - with b: - st.markdown("**Provenance and notes**") - st.markdown(r["notes"] or "_No researcher note on this record._") - st.markdown("**Sources**") - st.markdown(D.source_links(r.to_dict(), limit=6)) - if r["digitization_url"]: - st.markdown(f"**Scan** · [{r['digitization_holder'] or 'view'}]({r['digitization_url']})") - if press is not None: - st.markdown("**The press**") - st.markdown( - f"{press['name']} — {press['city']}, {press['country']} · {press['active_span']} · " - f"{press['ownership']}." - ) - if press["archive_holdings"]: - st.markdown(f"**Known holdings** · {press['archive_holdings']}") - - rr = RIGHTS[RIGHTS["country"] == r["country"]] - if len(rr): - rr = rr.iloc[0] - st.markdown("**Rights context for this country**") - st.markdown( - f"{rr['country']} — {rr['copyright_term']} ({rr['statute']}). " - f"{rr['recovery_implication']}" - ) - else: - st.markdown("**Rights context** · not yet researched for this country.") - - -# --------------------------------------------------------------------------- -# 5. Rights & copyright -# --------------------------------------------------------------------------- - -def page_rights(): - hero( - "Rights", - "Out of print is not out of copyright", - "Almost every book in this catalogue is still protected. Literary copyright commonly " - "runs for the author's life plus fifty years, and in several of these countries " - "seventy. Acquiring, cataloguing and preserving a physical copy is lawful in far more " - "situations than publishing its full text online.", - ) - note( - "This page is a factual description of published legal materials — statutes and " - "treaty membership — and not legal advice. Clearance for any individual title should " - "be confirmed with counsel in the relevant jurisdiction.", - kind="warn", - ) - - df = RIGHTS.sort_values(["term_years", "country"]) - base = alt.Chart(df).encode( - y=alt.Y("country:N", title=None, sort=list(df["country"]), - axis=alt.Axis(labelOverlap=False), - scale=alt.Scale(paddingInner=0.35, paddingOuter=0.25)), - ) - ch = base.mark_bar(cornerRadiusEnd=4, color=P.series[0], - stroke=P.surface, strokeWidth=2).encode( - x=alt.X("term_years:Q", title="years after the author's death", - scale=alt.Scale(domain=[0, 80]), axis=alt.Axis(tickCount=5, format="d")), - tooltip=[alt.Tooltip("country:N", title="Country"), - alt.Tooltip("copyright_term:N", title="Term"), - alt.Tooltip("statute:N", title="Statute")], - ) - lbl = base.mark_text(align="left", dx=7, fontSize=11, color=P.text2).encode( - x=alt.X("term_years:Q"), text=alt.Text("copyright_term:N") - ) - st.altair_chart( - (ch + lbl).properties(height=330, title="Copyright term after the author's death"), - width="stretch", - ) - st.caption( - "Kenya, Tanzania and South Africa run life + 50. An author in those three who died in " - "1975 or earlier is out of copyright now — which makes them the obvious place to start " - "a first tranche that needs no clearance at all." - ) + st.markdown("### " + bil("From a recovered book to a taught text", + "Du livre récupéré au texte enseigné")) + note(bil( + "Three layers, three different rights positions. A freely searchable " + "bibliography is publishable now — facts about books are not the books. Full " + "texts require cleared rights, title by title. Curriculum modules sit in " + "between: original teaching material the project owns outright, quoting only " + "short excerpts. That middle layer is how a still-copyrighted book becomes " + "teachable before clearance is finished.", + "Trois couches, trois situations juridiques. Une bibliographie " + "librement consultable est publiable dès maintenant — les faits sur les " + "livres ne sont pas les livres. Les textes intégraux exigent des droits " + "libérés, titre par titre. Les modules pédagogiques sont entre les deux : un " + "matériel original dont le projet est propriétaire, ne citant que de courts " + "extraits. C'est cette couche intermédiaire qui rend enseignable un livre " + "encore protégé, avant que les autorisations soient acquises.")) + banni = T[T["banned"].astype(str) == "True"] + tuiles([ + (nf(int((T["genre_group"] == "Children's & folklore").sum())), + bil("children's & folklore", "jeunesse et folklore"), + bil("the readiest classroom material", + "le matériel le plus immédiatement utilisable")), + (nf(int(T["african_language"].sum())), + bil("African-language titles", "titres en langues africaines"), + f"{T[T['african_language']]['language'].nunique()} " + + bil("languages", "langues")), + (nf(len(banni)), bil("recorded as banned", "signalés comme interdits"), + bil("censorship as a teachable subject", + "la censure comme sujet d'enseignement")), + (nf(T["press_id"].nunique()), t("Houses"), + bil("publishing history itself", "l'histoire de l'édition elle-même")), + ]) + if len(banni): + st.markdown("##### " + bil("Titles recorded as banned", + "Titres signalés comme interdits")) + st.caption(bil( + "Apartheid-era banning is a direct cause of these books leaving " + "circulation, and the ban record is itself curriculum material.", + "L'interdiction sous l'apartheid est une cause directe de la disparition " + "de ces livres, et le registre des interdictions est lui-même un " + "matériau pédagogique.")) + st.dataframe( + banni[["title", "author", "press_name", "city", "year", "notes"]] + .sort_values("year", na_position="last"), hide_index=True, **LARGE, + column_config={"title": t("Title"), "author": t("Author"), + "press_name": t("Press"), "city": t("City"), + "year": st.column_config.NumberColumn(t("Year"), format="%d"), + "notes": st.column_config.TextColumn("Note", width="large")}) + +# ==================================================================== 6. DROITS +with onglets[5]: + st.markdown("### " + bil("Out of print is not out of copyright", + "Épuisé ne veut pas dire libre de droits")) + st.markdown(bil( + "Almost every book in this catalogue is still protected. Literary copyright " + "commonly runs for the author's life plus fifty years, and in several of " + "these countries seventy. Acquiring, cataloguing and preserving a physical " + "copy is lawful in far more situations than publishing its full text online.", + "Presque tous les livres de ce catalogue sont encore protégés. Le droit " + "d'auteur littéraire court communément pendant la vie de l'auteur plus " + "cinquante ans, et dans plusieurs de ces pays soixante-dix. Acquérir, " + "cataloguer et conserver un exemplaire matériel est licite dans beaucoup plus " + "de cas que d'en publier le texte intégral en ligne.")) + note(bil( + "This page describes published legal materials — statutes and treaty " + "membership. It is not legal advice, and clearance for any individual title " + "should be confirmed with counsel in the relevant jurisdiction.", + "Cette page décrit des textes juridiques publiés — lois et adhésions aux " + "traités. Elle ne vaut pas conseil juridique : l'autorisation pour un titre " + "donné doit être confirmée par un juriste de la juridiction concernée."), + "warn") + + R = RIGHTS.sort_values(["term_years", "country"]) + b = alt.Chart(R).encode( + y=alt.Y("country:N", sort=list(R["country"]), title=None, + scale=alt.Scale(paddingInner=0.35, paddingOuter=0.25))) + ch = b.mark_bar(color=PAL["accent"], stroke=PAL["fond"], strokeWidth=2).encode( + x=alt.X("term_years:Q", scale=alt.Scale(domain=[0, 82]), + title=bil("years after the author's death", + "années après la mort de l'auteur"), + axis=alt.Axis(tickCount=5, format="d")), + tooltip=[alt.Tooltip("country:N", title=t("Country")), + alt.Tooltip("copyright_term:N", title=bil("Term", "Durée")), + alt.Tooltip("statute:N", title=bil("Statute", "Texte"))]) + lb = b.mark_text(align="left", dx=7, fontSize=11, color=PAL["texte"]).encode( + x="term_years:Q", text=alt.Text("copyright_term:N")) + tracer_fixe((ch + lb).properties( + width=640, height=330, + title=alt.TitleParams(bil("Copyright term after the author's death", + "Durée de protection après la mort de l'auteur"), + subtitle=bil("Eight jurisdictions researched so far.", + "Huit juridictions étudiées à ce jour.")))) + st.caption(bil( + "Kenya, Tanzania and South Africa run life + 50. An author in those three who " + "died in 1975 or earlier is out of copyright now — which makes them the " + "obvious place to begin a first tranche that needs no clearance at all.", + "Le Kenya, la Tanzanie et l'Afrique du Sud appliquent vie + 50. Un auteur de " + "ces trois pays mort en 1975 ou avant est aujourd'hui dans le domaine " + "public : c'est par là qu'il faut commencer, avec une première tranche qui ne " + "demande aucune autorisation.")) st.divider() - st.markdown("### Public domain check") - st.caption("Arithmetic on the statutory term only. It does not account for anonymous, " - "corporate, posthumous or folklore rules, and it is not a clearance.") - c1, c2, c3 = st.columns([1.2, 1, 2]) - with c1: - country = st.selectbox("Country of publication", sorted(RIGHTS["country"]), key="pd_c") - with c2: - death = st.number_input("Author's year of death", min_value=1850, max_value=2026, - value=1975, step=1, key="pd_y") - row = RIGHTS[RIGHTS["country"] == country].iloc[0] - verdict, expiry, why = D.public_domain_status(death, row["term_years"]) - with c3: - colour = P.status["good"] if "Likely public domain" in verdict else P.status["warning"] - st.markdown( - f'
' - f'

{verdict}

{why}
' - f'

{row["public_domain_notes"]}

', - unsafe_allow_html=True, - ) + st.markdown("### " + bil("Public domain check", "Vérification du domaine public")) + st.caption(bil( + "Arithmetic on the statutory term only. It does not account for anonymous, " + "corporate, posthumous or folklore rules, and it is not a clearance.", + "Arithmétique sur la seule durée légale. Elle ne tient compte ni des règles " + "sur l'anonymat, les personnes morales, les œuvres posthumes ou le folklore, " + "et ne vaut pas autorisation.")) + q1, q2, q3 = st.columns([1.15, 1, 2]) + with q1: + pays = st.selectbox(t("Country"), sorted(RIGHTS["country"]), key="pd_c") + with q2: + mort = st.number_input(bil("Author's year of death", + "Année de mort de l'auteur"), + 1850, 2026, 1975, 1, key="pd_y") + row = RIGHTS[RIGHTS["country"] == pays].iloc[0] + exp = int(mort) + int(row["term_years"]) + libre = 2026 > exp + with q3: + coul = STATUT["bon"] if libre else STATUT["attention"] + verdict = (bil("Likely public domain in this country", + "Probablement dans le domaine public dans ce pays") if libre + else bil("Still in copyright in this country", + "Encore protégé dans ce pays")) + pourquoi = bil( + f"Protection {'ran' if libre else 'runs'} to the end of {exp} under a " + f"life + {int(row['term_years'])} year term.", + f"La protection {'a couru' if libre else 'court'} jusqu'à la fin de " + f"{exp}, sous une durée de vie + {int(row['term_years'])} ans.") + st.markdown(f'
' + f'

{verdict}

{pourquoi}
' + f'

{row["public_domain_notes"]}

', + unsafe_allow_html=True) st.divider() - st.markdown("### Jurisdiction by jurisdiction") + st.markdown("### " + bil("Jurisdiction by jurisdiction", + "Juridiction par juridiction")) for r in RIGHTS.sort_values("country").itertuples(): - orphan = r.orphan_works_mechanism or "No orphan-works mechanism recorded." - st.markdown( - f'

{r.country}

' - f'
{r.copyright_term} · {r.statute} · ' - f'Berne member: {"yes" if r.berne_member else "no"}
' - f"

Corporate / anonymous works · {r.anonymous_or_corporate_term or 'not recorded'}

" - f"

Orphan works · {orphan}

" - f"

What this means for a 1960s book · {r.recovery_implication}

" - f'
{T.chip(r.confidence)}
', - unsafe_allow_html=True, - ) - st.markdown( - f'
' - f"Sources: {D.source_links(r._asdict())}
", - unsafe_allow_html=True, - ) + orph = txt(r.orphan_works_mechanism, + bil("No orphan-works mechanism recorded.", + "Aucun mécanisme pour les œuvres orphelines.")) + fiche(r.country, + f"{r.copyright_term} · {r.statute} · " + + bil("Berne member", "membre de Berne") + + f" : {'✓' if r.berne_member else '—'}", + f"{bil('Corporate / anonymous works', 'Personnes morales / anonymes')}" + f" · {txt(r.anonymous_or_corporate_term, '—')}
" + f"{bil('Orphan works', 'Œuvres orphelines')} · {orph}
" + f"{bil('What this means for a 1960s book', 'Ce que cela implique pour un livre des années 1960')}" + f" · {r.recovery_implication}", + f"{puce(r.confidence)} {t('Sources')} : {liens(r._asdict())}") st.divider() - st.markdown("### The diligent-search file") - st.caption( - "Where the rights holder cannot be found, the defensible position is a documented " - "search, kept per title, showing what was tried and what it returned." - ) - steps = [ - ("Author", "Trace the author, then the estate: national writers' union, university " - "department, obituary, family. Record dates of birth and death — nothing " - "can be calculated without them."), - ("Original publisher", "Establish whether the house still trades, was absorbed, or " - "closed. Company registries settle this faster than literature does."), - ("Successor in title", "If absorbed, follow the chain of acquisition. If closed, look " - "for whether contracts reverted rights to authors — the press " - "archive is where that document lives."), - ("Contributors", "Translator, illustrator and cover artist hold separate rights. A " - "cleared text with an uncleared cover is not cleared."), - ("Territory and format", "Confirm the grant covers digital use and the territories you " - "intend to serve, not just the original print run."), - ("Record the negative", "Log every dead end with date and method. The value of the file " - "is the evidence that the search was real."), + st.markdown("### " + bil("The diligent-search file", + "Le dossier de recherche diligente")) + st.caption(bil( + "Where the rights holder cannot be found, the defensible position is a " + "documented search, kept per title, showing what was tried and what it " + "returned. Its value is the evidence that the search was real.", + "Lorsque l'ayant droit reste introuvable, la position défendable est une " + "recherche documentée, tenue titre par titre, montrant ce qui a été tenté et " + "ce que cela a donné. Sa valeur, c'est la preuve que la recherche a bien eu " + "lieu.")) + ETAPES = [ + ("Author", "L'auteur", + "Trace the author, then the estate: national writers' union, university " + "department, obituary, family. Record dates of birth and death — nothing can " + "be calculated without them.", + "Retrouver l'auteur, puis les ayants droit : union nationale des écrivains, " + "département universitaire, nécrologie, famille. Noter les dates de " + "naissance et de mort — rien ne se calcule sans elles."), + ("Original publisher", "L'éditeur d'origine", + "Establish whether the house still trades, was absorbed, or closed. Company " + "registries settle this faster than literature does.", + "Établir si la maison existe encore, a été absorbée ou a fermé. Les " + "registres du commerce tranchent plus vite que la littérature."), + ("Successor in title", "Le successeur", + "If absorbed, follow the chain of acquisition. If closed, look for whether " + "contracts reverted rights to authors — the press archive is where that " + "document lives.", + "Si absorbée, suivre la chaîne des rachats. Si fermée, chercher si les " + "contrats rendaient les droits aux auteurs — c'est dans les archives de la " + "maison que vit ce document."), + ("Contributors", "Les contributeurs", + "Translator, illustrator and cover artist hold separate rights. A cleared " + "text with an uncleared cover is not cleared.", + "Traducteur, illustrateur et maquettiste détiennent des droits distincts. Un " + "texte autorisé avec une couverture non autorisée n'est pas autorisé."), + ("Territory and format", "Territoire et support", + "Confirm the grant covers digital use and the territories you intend to " + "serve, not just the original print run.", + "Vérifier que la cession couvre l'usage numérique et les territoires visés, " + "et pas seulement le tirage d'origine."), + ("Record the negative", "Consigner les échecs", + "Log every dead end with date and method. The value of the file is the " + "evidence that the search was real.", + "Consigner chaque impasse avec sa date et sa méthode. La valeur du dossier, " + "c'est la preuve que la recherche a été réelle."), ] - for i, (k, v) in enumerate(steps, 1): - st.markdown(f'

{i}. {k}

{v}

', - unsafe_allow_html=True) - - note( - "A fair offer to authors and families. A non-exclusive agreement, proper " - "attribution, a preservation copy given back to the family, transparent usage " - "statistics, and either royalties or an explicitly agreed cultural-access licence. " - "Non-exclusive matters: it leaves the family free to license elsewhere, which is what " - "makes the offer easy to accept." - ) - footer() - - -# --------------------------------------------------------------------------- -# 6. Where the copies are -# --------------------------------------------------------------------------- - -def page_holdings(): - hero( - "Search channels", - "Where the surviving copies are", - "Recovery is a physical problem before it is a digital one. Institutional catalogues " - "find the books that were already privileged enough to be catalogued; the rest are in " - "school storerooms, mission libraries, market stalls and family houses.", - ) - - inafr = HOLDINGS[HOLDINGS["in_africa"]] - out = HOLDINGS[~HOLDINGS["in_africa"]] - documented = out[out["approx_items"].notna()]["approx_items"].sum() - st.markdown( - T.tiles([ - (f"{len(HOLDINGS)}", "search channels", "institutions and routes"), - (f"{len(inafr)}", "in Africa", "libraries, missions, markets, estates"), - (f"{len(out)}", "outside Africa", "mostly northern repositories"), - (f"{int(documented)}+", "items documented abroad", "Onitsha pamphlets, AWS papers, CRL"), - ]), - unsafe_allow_html=True, - ) - - note( - "The asymmetry is the finding. The best-documented, best-preserved holdings of " - "this literature sit in Kansas, Bloomington, London and Chicago, while the in-Africa " - "channels — where most surviving copies actually are — carry the lowest confidence " - "because nobody has surveyed them. Any recovery plan that only works the catalogued " - "repositories will reproduce that asymmetry. Every recovered digital master should be " - "deposited with an African institution as a condition of the work." - ) - - kind_counts = HOLDINGS.groupby(["kind", "in_africa"]).size().reset_index(name="n") - kind_counts["where"] = kind_counts["in_africa"].map({True: "In Africa", False: "Outside Africa"}) - ch = ( - alt.Chart(kind_counts, title="Search channels by kind") - .mark_bar(cornerRadiusEnd=4, height=16, stroke=P.surface, strokeWidth=2) - .encode( - x=alt.X("n:Q", title="channels", axis=alt.Axis(tickCount=4)), - y=alt.Y("kind:N", title=None, sort="-x", - axis=alt.Axis(labelOverlap=False, labelLimit=220)), - color=alt.Color("where:N", title=None, - scale=alt.Scale(domain=["In Africa", "Outside Africa"], - range=[P.series[0], P.series[1]])), - tooltip=[alt.Tooltip("kind:N", title="Kind"), alt.Tooltip("where:N", title="Where"), - alt.Tooltip("n:Q", title="Channels")], - ) - .properties(height=max(200, 34 * kind_counts["kind"].nunique())) - ) - st.altair_chart(ch, width="stretch") + for i, (en, fr_, den, dfr) in enumerate(ETAPES, 1): + fiche(f"{i}. {en if LANG == 'en' else fr_}", "", den if LANG == "en" else dfr) + note(bil( + "A fair offer to authors and families. A non-exclusive agreement, " + "proper attribution, a preservation copy given back to the family, " + "transparent usage statistics, and either royalties or an explicitly agreed " + "cultural-access licence. Non-exclusive matters: it leaves the family free to " + "license elsewhere, which is what makes the offer easy to accept.", + "Une offre honnête aux auteurs et aux familles. Un accord non " + "exclusif, une attribution correcte, une copie de conservation rendue à la " + "famille, des statistiques d'usage transparentes, et soit des redevances soit " + "une licence d'accès culturel explicitement convenue. Le caractère non " + "exclusif compte : il laisse la famille libre de céder ailleurs, et c'est ce " + "qui rend l'offre acceptable.")) + +# ================================================================ 7. EXEMPLAIRES +with onglets[6]: + st.markdown("### " + bil("Where the surviving copies are", + "Où sont les exemplaires survivants")) + st.markdown(bil( + "Recovery is a physical problem before it is a digital one. Institutional " + "catalogues find the books that were already privileged enough to be " + "catalogued; the rest are in school storerooms, mission libraries, market " + "stalls and family houses.", + "La récupération est un problème matériel avant d'être un problème numérique. " + "Les catalogues institutionnels retrouvent les livres déjà assez privilégiés " + "pour avoir été catalogués ; les autres sont dans des réserves d'école, des " + "bibliothèques de mission, des étals de marché et des maisons de famille.")) + + dedans = HOLD[HOLD["in_africa"]] + dehors = HOLD[~HOLD["in_africa"]] + docu = int(dehors["approx_items"].dropna().sum()) + tuiles([ + (nf(len(HOLD)), bil("search channels", "pistes de recherche"), + bil("institutions and routes", "institutions et circuits")), + (nf(len(dedans)), bil("in Africa", "en Afrique"), + bil("libraries, missions, markets, estates", + "bibliothèques, missions, marchés, familles")), + (nf(len(dehors)), bil("outside Africa", "hors d'Afrique"), + bil("mostly northern repositories", "surtout des dépôts du Nord")), + (nf(docu) + "+", bil("items documented abroad", + "pièces documentées à l'étranger"), + bil("Onitsha pamphlets, AWS papers, CRL", + "pamphlets d'Onitsha, archives AWS, CRL")), + ]) + note(bil( + "The asymmetry is the finding. The best-documented, best-preserved " + "holdings of this literature sit in Kansas, Bloomington, London and Chicago, " + "while the in-Africa channels — where most surviving copies actually are — " + "carry the lowest confidence because nobody has surveyed them. Any recovery " + "plan that only works the catalogued repositories will reproduce that " + "asymmetry. Every recovered digital master should be deposited with an " + "African institution as a condition of the work.", + "L'asymétrie est le résultat. Les fonds les mieux documentés et les " + "mieux conservés de cette littérature sont au Kansas, à Bloomington, à " + "Londres et à Chicago, tandis que les pistes africaines — où se trouvent la " + "plupart des exemplaires survivants — sont les moins fiables parce que " + "personne ne les a recensées. Tout plan qui n'exploite que les dépôts " + "catalogués reproduira cette asymétrie. Chaque master numérique produit " + "devrait être déposé dans une institution africaine, comme condition du " + "travail.")) + + kk = HOLD.groupby(["kind", "in_africa"]).size().reset_index(name="n") + kk["label"] = kk["kind"] + dom = [bil("In Africa", "En Afrique"), bil("Outside Africa", "Hors d'Afrique")] + kk["ou"] = kk["in_africa"].map({True: dom[0], False: dom[1]}) + ch = (alt.Chart(kk).mark_bar(height=15, stroke=PAL["fond"], strokeWidth=2) + .encode(y=alt.Y("label:N", sort="-x", title=None), + x=alt.X("n:Q", title=bil("channels", "pistes"), + axis=alt.Axis(tickCount=4, format="d")), + color=alt.Color("ou:N", title=None, + scale=alt.Scale(domain=dom, range=[CAT[0], CAT[1]]), + legend=alt.Legend(orient="top")), + tooltip=[alt.Tooltip("label:N", title=""), + alt.Tooltip("ou:N", title=""), + alt.Tooltip("n:Q", title=bil("channels", "pistes"))]) + .properties(title=alt.TitleParams( + bil("Search channels by kind", "Pistes de recherche par nature")))) + tracer(ch, 34 * kk["label"].nunique() + 130) st.divider() - where = st.radio("Show", ["In Africa", "Outside Africa", "All"], horizontal=True, key="h_where") - hf = HOLDINGS - if where == "In Africa": - hf = HOLDINGS[HOLDINGS["in_africa"]] - elif where == "Outside Africa": - hf = HOLDINGS[~HOLDINGS["in_africa"]] - + ou = choix(dom + [t("All")], "hold_ou") + hf = HOLD if ou == t("All") else (HOLD[HOLD["in_africa"]] if ou == dom[0] + else HOLD[~HOLD["in_africa"]]) for r in hf.sort_values(["kind", "name"]).itertuples(): - items = f" · ~{int(r.approx_items)} items documented" if pd.notna(r.approx_items) else "" - link = f' · catalogue' if r.url else "" - prov = f'{r.provenance}' - st.markdown( - f'

{r.name}

' - f'
{r.kind} · {r.city + ", " if r.city else ""}{r.country}{items}{link}
' - f"

{r.what_it_holds}

" - + (f'

{r.notes}

' if r.notes else "") - + f'
{prov}{T.chip(r.confidence)}
', - unsafe_allow_html=True, - ) + items = (f" · ~{int(r.approx_items)} " + bil("items", "pièces") + if pd.notna(r.approx_items) else "") + lien = (f' · ' + + bil("catalogue", "catalogue") + "") if txt(r.url) else "" + ville = f"{r.city}, " if txt(r.city) else "" + note_ = f"
{r.notes}" if txt(r.notes) else "" + fiche(r.name, f"{r.kind} · {ville}{r.country}{items}{lien}", + r.what_it_holds + note_, + f'{r.provenance}{puce(r.confidence)}') st.divider() - st.markdown("### Digitization initiatives already in the field") - st.caption("Existing projects to partner with rather than duplicate. Access and format " - "columns decide whether their material can be reused at all.") - c1, c2 = st.columns([1, 1]) - with c1: - acc = st.multiselect("Access", sorted(DIGI["access"].unique()), key="d_acc") - with c2: - reg = st.multiselect("Region", sorted(DIGI["region"].unique()), key="d_reg") - dd = DIGI - if acc: - dd = dd[dd["access"].isin(acc)] - if reg: - dd = dd[dd["region"].isin(reg)] - + st.markdown("### " + bil("Digitization initiatives already in the field", + "Les initiatives de numérisation déjà à l'œuvre")) + st.caption(bil( + "Existing projects to partner with rather than duplicate. Access and format " + "decide whether their material can be reused at all.", + "Des projets existants avec qui s'associer plutôt que de les refaire. " + "L'accès et le format décident si leur matériel est réutilisable.")) st.dataframe( - dd[["name", "host_institution", "region", "access", "formats", "api", "scope", "relevance", - "confidence", "url"]].sort_values(["region", "name"]), - width="stretch", hide_index=True, height=380, + DIGI[["name", "host_institution", "region", "access", "formats", "api", + "scope", "relevance", "confidence", "url"]] + .sort_values(["region", "name"]), hide_index=True, height=360, **LARGE, column_config={ "name": st.column_config.TextColumn("Initiative", width="medium"), - "host_institution": st.column_config.TextColumn("Host", width="medium"), - "scope": st.column_config.TextColumn("Scope", width="large"), - "relevance": st.column_config.TextColumn("Why it matters", width="large"), - "url": st.column_config.LinkColumn("Link", display_text="open"), - }, - ) - note( - "Aluka is the cautionary precedent. It began as a freely accessible pan-African " - "resource and is now a licensed JSTOR collection, so African institutions may need a " - "subscription to reach material their own archives supplied. Design the licence " - "before the scanning, not after." - ) - footer() - - -# --------------------------------------------------------------------------- -# 7. Curriculum builder -# --------------------------------------------------------------------------- - -def page_curriculum(): - hero( - "Teaching", - "From a recovered book to a taught text", - "A scan is not a syllabus. The platform's third layer is the one that changes what " - "happens in a classroom: an introduction, a timeline, discussion questions, vocabulary " - "and legally permissible excerpts — publishable even where the full text cannot be.", - ) - - note( - "The three layers, and why they have different rights positions. " - "A freely searchable bibliography is publishable now — facts about books are not the " - "books. Full texts require cleared rights, title by title. Curriculum modules sit in " - "between: original teaching material the project owns outright, quoting only short " - "excerpts. That middle layer is how a still-copyrighted book becomes teachable before " - "clearance is finished." - ) - - pilot = TITLES[TITLES["pilot_cluster"] != ""] - for cluster in D.PILOT_TARGETS: - sub = pilot[pilot["pilot_cluster"] == cluster] - if not len(sub): - continue - langs = ", ".join(sorted(sub["language"].unique())[:6]) - span = sub["year"].dropna() - span_s = f"{int(span.min())}–{int(span.max())}" if len(span) else "undated" - with st.expander(f"{cluster} — {len(sub)} titles · {span_s} · {langs}"): - st.markdown(f"**{_cluster_rationale(cluster)}**") - st.dataframe( - sub[["title", "author", "press_name", "year", "language", "genre", "confidence"]] - .sort_values("year", na_position="last"), - width="stretch", hide_index=True, - column_config={"year": st.column_config.NumberColumn("Year", format="%d")}, - ) - st.markdown("**A module for this cluster would contain**") - st.markdown( - "- an introduction placing the press and the moment it published into\n" - "- a timeline linking the books to the political history around them\n" - "- discussion questions written for the age group the cluster suits\n" - "- vocabulary and language notes, including for the African-language titles\n" - "- short quoted excerpts within fair dealing, with full text where cleared\n" - "- a note on what is missing from the record and how a class might help fill it" - ) - - st.divider() - st.markdown("### What the catalogue can already teach") - ages = TITLES[TITLES["genre_group"] == "Children's & folklore"] - banned = TITLES[TITLES["banned"] == True] # noqa: E712 - st.markdown( - T.tiles([ - (f"{len(ages)}", "children's & folklore titles", "the readiest classroom material"), - (f"{int(TITLES['african_language'].sum())}", "African-language titles", - f"{TITLES[TITLES['african_language']]['language'].nunique()} languages"), - (f"{len(banned)}", "titles recorded as banned", "censorship as a teachable subject"), - (f"{TITLES['press_id'].nunique()}", "presses represented", "publishing history itself"), - ]), - unsafe_allow_html=True, - ) - - if len(banned): - st.markdown("##### Titles recorded as banned") - st.caption( - "Apartheid-era banning is a direct cause of these books leaving circulation, and " - "the ban record is itself curriculum material." - ) - st.dataframe( - banned[["title", "author", "press_name", "city", "year", "notes"]] - .sort_values("year", na_position="last"), - width="stretch", hide_index=True, - column_config={ - "year": st.column_config.NumberColumn("Year", format="%d"), - "notes": st.column_config.TextColumn("Note", width="large"), - }, - ) - footer() - - -def _cluster_rationale(cluster: str) -> str: - return { - "A · Onitsha market pamphlets": - "Popular writing produced and sold outside every colonial and missionary " - "institution — the first large-scale independent book trade in British West " - "Africa. It teaches literacy, commerce and voice at once, and almost none of it " - "is dated reliably, which makes it a good exercise in reading evidence.", - "B · Ghanaian children's & folklore": - "The most immediately usable material in the catalogue and the most poorly " - "recorded. The Ghana Publishing Corporation's Children's Library series was " - "explicitly commissioned as books by Africans for African children — the exact " - "brief this project inherits.", - "C · East African popular fiction": - "Novels and plays from Nairobi and Dar es Salaam at the height of the East African " - "Publishing House and Tanzania Publishing House, including work by West and " - "Southern African writers who published there. Kenya and Tanzania's life + 50 term " - "also makes this the cluster most likely to be clearable.", - "D · Francophone (CLE, NEA & related)": - "Éditions CLE in Yaoundé and the Nouvelles Éditions Africaines in Dakar are the " - "institutionally anchored African francophone houses, and CLE still exists — a " - "living counterparty rather than an orphan-works problem. Présence Africaine " - "belongs alongside them but was Paris-based, and the catalogue says so.", - "E · African-language works": - "Swahili, Hausa, Yoruba, Igbo, Shona, isiZulu, Akan, Sesotho, Malagasy and more. " - "The least digitised and least catalogued part of the corpus, and the part whose " - "loss is least recoverable from elsewhere.", - }.get(cluster, "") - - -# --------------------------------------------------------------------------- -# 8. Methods & sources -# --------------------------------------------------------------------------- - -def page_methods(): - hero( - "Methods", - "How this catalogue was built, and what it is not", - "A recovery project is only as good as its willingness to say what it does not know. " - "These are the rules the data follows and the gaps that remain.", - ) - - st.markdown("### Data governance rules") - rules = [ - ("Every record carries its sources", - "A record with no citable source is dropped at build time, not published with a " - "shrug. Source URLs travel with the row into the CSV export."), - ("Nothing is estimated", - f"{int(TITLES['year'].isna().sum())} titles have no year. They are shown as undated " - "rather than given a plausible decade. Library conjecture dates — the bracketed " - "[1964?] of the Onitsha records — are recorded at medium confidence and flagged."), - ("Confidence is always visible", - "High means a library or publisher catalogue record. Medium means a scholarly source " - "or a dealer listing transcribed from a title page. Low means a single unconfirmed " - "mention and should be treated as a lead."), + "host_institution": st.column_config.TextColumn( + bil("Host", "Institution"), width="medium"), + "region": t("Region"), "access": bil("Access", "Accès"), + "formats": "Format", "api": "API", + "scope": st.column_config.TextColumn(bil("Scope", "Périmètre"), + width="large"), + "relevance": st.column_config.TextColumn( + bil("Why it matters", "Pourquoi c'est utile"), width="large"), + "confidence": t("Confidence"), + "url": st.column_config.LinkColumn("↗", display_text="↗")}) + note(bil( + "Aluka is the cautionary precedent. It began as a freely accessible " + "pan-African resource and is now a licensed JSTOR collection, so African " + "institutions may need a subscription to reach material their own archives " + "supplied. Design the licence before the scanning, not after.", + "Aluka est le précédent qui doit servir d'avertissement. Née comme " + "ressource panafricaine librement accessible, c'est aujourd'hui une " + "collection JSTOR sous licence : des institutions africaines peuvent devoir " + "s'abonner pour atteindre un matériel que leurs propres archives ont fourni. " + "La licence se conçoit avant la numérisation, pas après.")) + +# ================================================================== 8. METHODES +with onglets[7]: + st.markdown("### " + bil("How this catalogue was built, and what it is not", + "Comment ce catalogue fut construit, et ce qu'il n'est pas")) + st.markdown(bil( + "A recovery project is only as good as its willingness to say what it does " + "not know. These are the rules the data follows — all of them enforced in " + "`build_data.py` and checked in `selftest.py`, which fails the build rather " + "than shipping a dashboard that overstates its evidence.", + "Un projet de récupération ne vaut que par sa capacité à dire ce qu'il " + "ignore. Voici les règles que suivent les données — toutes appliquées dans " + "`build_data.py` et vérifiées par `selftest.py`, qui fait échouer la " + "construction plutôt que de livrer un tableau de bord qui exagérerait ses " + "preuves.")) + REGLES = [ + ("Every record carries its sources", "Chaque notice porte ses sources", + "A record with no citable source is dropped at build time, not published " + "with a shrug. Source URLs travel with the row into the CSV export.", + "Une notice sans source citable est écartée à la construction, non publiée " + "avec un haussement d'épaules. Les URL des sources voyagent avec la ligne " + "jusqu'à l'export CSV."), + ("Nothing is estimated", "Rien n'est estimé", + f"{int(T['year'].isna().sum())} titles have no year. They are shown as " + "undated rather than given a plausible decade. Library conjecture dates — " + "the bracketed [1964?] of the Onitsha records — are recorded at medium " + "confidence and flagged in the note.", + f"{int(T['year'].isna().sum())} titres n'ont pas d'année. Ils sont affichés " + "comme non datés plutôt que rattachés à une décennie plausible. Les dates " + "conjecturales des bibliothèques — les [1964?] des notices d'Onitsha — sont " + "enregistrées en fiabilité moyenne et signalées dans la note."), + ("Confidence is always visible", "La fiabilité est toujours affichée", + "High means a library or publisher catalogue record. Medium means a " + "scholarly source or a dealer listing transcribed from a title page. Low " + "means a single unconfirmed mention and should be treated as a lead.", + "Élevée : une notice de catalogue de bibliothèque ou d'éditeur. Moyenne : " + "une source savante ou une annonce de libraire transcrite d'une page de " + "titre. Faible : une seule mention non confirmée, à traiter comme une piste."), ("Out of print is never rendered as out of copyright", - "They are separate fields, computed separately, and the rights page carries the " - "warning on every view."), + "« Épuisé » n'est jamais rendu par « libre de droits »", + "They are separate fields, computed separately, and the rights tab carries " + "the warning on every view.", + "Ce sont deux champs distincts, calculés séparément, et l'onglet Droits " + "porte l'avertissement à chaque affichage."), ("Presses outside Africa are flagged, not hidden", - "Présence Africaine, Éditions Nubia and the Casa dos Estudantes do Império were " - "African-founded and African-directed but printed in Paris and Lisbon. They are in " - "the register with the imprint flagged, because excluding them would falsify the " - "intellectual history and including them silently would falsify the geography."), + "Les maisons hors d'Afrique sont signalées, pas cachées", + "Présence Africaine, Éditions Nubia and the Casa dos Estudantes do Império " + "were African-founded and African-directed but printed in Paris and Lisbon. " + "Excluding them would falsify the intellectual history; including them " + "silently would falsify the geography.", + "Présence Africaine, les Éditions Nubia et la Casa dos Estudantes do Império " + "furent fondées et dirigées par des Africains, mais imprimaient à Paris et à " + "Lisbonne. Les exclure fausserait l'histoire intellectuelle ; les inclure en " + "silence fausserait la géographie."), + ("Grey is never zero", "Le gris n'est jamais zéro", + "On every map, a country with no record is grey and the subtitle counts " + "them. Nothing is interpolated from a neighbour, and an unsearched country " + "is never allowed to read as an empty one.", + "Sur chaque carte, un pays sans notice est gris et le sous-titre les " + "dénombre. Rien n'est interpolé depuis un voisin, et un pays non exploré ne " + "doit jamais se lire comme un pays vide."), + ("City coordinates are a cartographic aid, not a record", + "Les coordonnées de villes sont une aide cartographique, pas une notice", + "The dots on the atlas come from a public gazetteer and place a city on a " + "map; they are declared in villes.json, they carry no bibliographic weight, " + "and every city spelling variant found in the sources is normalised there in " + "the open rather than by a hidden regex.", + "Les points de l'atlas viennent d'un répertoire géographique public et " + "placent une ville sur une carte ; ils sont déclarés dans villes.json, ne " + "portent aucune valeur bibliographique, et chaque variante d'orthographe " + "trouvée dans les sources y est normalisée à découvert, non par une " + "expression régulière cachée."), ("Derived fields are computed, never hand-entered", - "Region, decade, genre grouping, recovery stage and pilot cluster are all produced " - "deterministically by build_data.py and can be recomputed from the source JSON."), + "Les champs dérivés sont calculés, jamais saisis", + "Region, decade, genre grouping, recovery stage, pilot cluster and city " + "normalisation all come out of build_data.py and can be recomputed from the " + "source JSON.", + "Région, décennie, regroupement de genres, étape de récupération, groupe " + "pilote et normalisation des villes sortent tous de build_data.py et sont " + "recalculables depuis le JSON source."), ] - for k, v in rules: - st.markdown(f'

{k}

{v}

', unsafe_allow_html=True) + for en, fr_, den, dfr in REGLES: + fiche(en if LANG == "en" else fr_, "", den if LANG == "en" else dfr) st.divider() - st.markdown("### Known gaps") - gaps = [ - ("Coverage is uneven by design of the search, not of the field", - "Nigeria, Ghana, East Africa, Francophone Africa and Southern Africa were researched " - "in depth. North Africa, the Horn, Lusophone Africa, the Indian Ocean and Anglophone " - "Cameroon are thin. Cape Verde and Swaziland are empty. Read a small number as " - "'not yet searched', never as 'nothing there'."), + st.markdown("### " + bil("Known gaps", "Lacunes connues")) + LACUNES = [ + ("Half the continent has not been searched", + "La moitié du continent n'a pas été explorée", + f"{N_ZERO} of the 54 African countries carry no record in this catalogue. " + "Nigeria, Ghana, East Africa, Francophone Africa and Southern Africa were " + "researched in depth; North Africa, the Horn, Lusophone Africa and the " + "Indian Ocean were not. Read a small number as 'not yet searched', never as " + "'nothing there'.", + f"{N_ZERO} des 54 pays africains ne portent aucune notice dans ce catalogue. " + "Le Nigéria, le Ghana, l'Afrique de l'Est, l'Afrique francophone et " + "l'Afrique australe ont été étudiés en profondeur ; l'Afrique du Nord, la " + "Corne, l'Afrique lusophone et l'océan Indien ne l'ont pas été. Un petit " + "nombre se lit « pas encore cherché », jamais « rien à trouver »."), ("Northern Nigerian and Hausa publishing is the thinnest major gap", - "Five Hausa titles stand in for the output of Gaskiya Corporation and the Northern " - "Nigerian Publishing Company. The IFRA-Nigeria volume on Northern Nigerian publishing " - "is the highest-value unexploited source."), + "L'édition du Nord-Nigéria et en haoussa est la plus grande lacune", + "Five Hausa titles stand in for the output of the Gaskiya Corporation and " + "the Northern Nigerian Publishing Company. The IFRA-Nigeria volume on " + "Northern Nigerian publishing is the highest-value unexploited source.", + "Cinq titres en haoussa tiennent lieu de toute la production de la Gaskiya " + "Corporation et de la Northern Nigerian Publishing Company. Le volume " + "IFRA-Nigeria sur l'édition du Nord est la source inexploitée la plus " + "prometteuse."), ("Ghanaian-language publishing is barely represented", - "The Bureau of Ghana Languages published in eleven languages from 1951 and only one " - "in-period imprint could be verified online. This is the largest recoverable-but-" - "invisible body in West Africa and needs on-site work."), - ("Scan status is mostly unverified", - f"{int((TITLES['digitization_status'] == 'unknown').sum())} titles have no checked " - "digitization status. That is a research task not yet done, not a finding that no " - "scan exists."), - ("Author dates are absent", - "No copyright calculation in this dashboard is applied to a specific author, because " - "the catalogue does not yet hold dates of death. That field is the single highest-" - "value addition to the next pass."), + "L'édition en langues ghanéennes est à peine représentée", + "The Bureau of Ghana Languages published in eleven languages from 1951 and " + "only one in-period imprint could be verified online. This is the largest " + "recoverable-but-invisible body in West Africa and it needs on-site work.", + "Le Bureau of Ghana Languages a publié en onze langues à partir de 1951 et " + "une seule impression de la période a pu être vérifiée en ligne. C'est le " + "plus grand fonds récupérable mais invisible d'Afrique de l'Ouest, et il " + "exige un travail sur place."), + ("Scan status is mostly unverified", "L'état de numérisation est peu vérifié", + f"{int((T['digitization_status'] == 'unknown').sum())} titles have no checked " + "digitization status. That is a research task not yet done, not a finding " + "that no scan exists.", + f"{int((T['digitization_status'] == 'unknown').sum())} titres n'ont pas " + "d'état de numérisation vérifié. C'est une tâche de recherche non faite, pas " + "un constat d'absence de numérisation."), + ("Author dates are absent", "Les dates des auteurs manquent", + "No copyright calculation in this dashboard is applied to a specific author, " + "because the catalogue does not yet hold dates of death. That field is the " + "single highest-value addition to the next pass — it is what unlocks stage 3.", + "Aucun calcul de droits n'est ici appliqué à un auteur précis, car le " + "catalogue ne contient pas encore les dates de décès. C'est l'ajout le plus " + "précieux de la prochaine passe : c'est lui qui débloque l'étape 3."), ("Dealer listings underpin a meaningful minority of records", - "Booksellers transcribe title pages, so publisher and place are usually sound, but " - "years get mis-keyed. These rows are capped at medium confidence and should be " - "re-verified against WorldCat or a national bibliography before publication."), + "Les annonces de libraires fondent une minorité notable de notices", + "Booksellers transcribe title pages, so publisher and place are usually " + "sound, but years get mis-keyed. These rows are capped at medium confidence " + "and should be re-verified against WorldCat or a national bibliography " + "before publication.", + "Les libraires transcrivent les pages de titre : éditeur et lieu sont donc " + "généralement fiables, mais les années sont souvent mal saisies. Ces lignes " + "sont plafonnées en fiabilité moyenne et doivent être revérifiées sur " + "WorldCat ou une bibliographie nationale avant publication."), ] - for k, v in gaps: - st.markdown(f'

{k}

{v}

', unsafe_allow_html=True) + for en, fr_, den, dfr in LACUNES: + fiche(en if LANG == "en" else fr_, "", den if LANG == "en" else dfr) st.divider() - st.markdown("### Reference tools this project should be built on") - st.markdown( - "- **The African Book Publishing Record** — the historical volumes remain the single " - "most complete contemporaneous record of African imprints and are not fully online.\n" - "- **National bibliographies** — Nigeria, Ghana, Kenya and Tanzania each produced one; " - "they capture what legal deposit caught and nothing else.\n" - "- **Publishers' own catalogues** — often the only list of a defunct house's backlist.\n" - "- **Hans Zell's bibliographic research on African publishing** — the standard " - "secondary apparatus for the whole field.\n" - "- **The BnF Catalogue général SRU API** — the francophone records here were built from " - "it and it remains the most complete francophone African imprint source." - ) + st.markdown("### " + bil("Sources and reference tools", + "Sources et instruments de référence")) + st.markdown(bil( + "The catalogue was assembled from library digital collections (the University " + "of Kansas Onitsha Market Literature collection, Indiana, Yale, the Center " + "for Research Libraries, SOAS), the BnF *Catalogue général* SRU API for " + "francophone imprints, publisher histories, scholarly work on African " + "publishing, national bibliographies and antiquarian listings. Rights " + "information comes from national statutes via WIPO Lex, cited per " + "jurisdiction. Country outlines are Natural Earth 1:50m (public domain).\n\n" + "The instruments this project should be built on next: the historical volumes " + "of the **African Book Publishing Record**, which remain the most complete " + "contemporaneous record of African imprints and are not fully online; the " + "**national bibliographies** of Nigeria, Ghana, Kenya and Tanzania; " + "**publishers' own catalogues**, often the only list of a defunct house's " + "backlist; and **Hans Zell's bibliographic research**, the standard secondary " + "apparatus for the whole field.", + "Le catalogue a été constitué à partir de collections numériques de " + "bibliothèques (fonds Onitsha Market Literature de l'université du Kansas, " + "Indiana, Yale, Center for Research Libraries, SOAS), de l'API SRU du " + "*Catalogue général* de la BnF pour les impressions francophones, d'histoires " + "d'éditeurs, de travaux savants sur l'édition africaine, de bibliographies " + "nationales et d'annonces de libraires. L'information juridique provient des " + "lois nationales via WIPO Lex, citée par juridiction. Les contours sont ceux " + "de Natural Earth 1:50m (domaine public).\n\n" + "Les instruments sur lesquels bâtir la suite : les volumes historiques de " + "l'**African Book Publishing Record**, qui restent le relevé contemporain le " + "plus complet des impressions africaines et ne sont pas entièrement en " + "ligne ; les **bibliographies nationales** du Nigéria, du Ghana, du Kenya et " + "de la Tanzanie ; les **catalogues des éditeurs eux-mêmes**, souvent seule " + "liste du fonds d'une maison disparue ; et les **travaux bibliographiques de " + "Hans Zell**, appareil de référence de tout le domaine.")) st.divider() - st.markdown("### The data") - c1, c2 = st.columns(2) - with c1: - st.download_button("Download the full title catalogue (CSV)", - TITLES.to_csv(index=False).encode("utf-8"), - file_name="abrp_titles.csv", mime="text/csv") - with c2: - st.download_button("Download the press register (CSV)", - PRESSES.to_csv(index=False).encode("utf-8"), - file_name="abrp_presses.csv", mime="text/csv") - st.caption( - "The source JSON, the ETL that produces these tables and this application are all in " - "the Space repository. Rebuild with `python build_data.py`." - ) - footer() - - -# --------------------------------------------------------------------------- - -PAGES = { - "Overview": page_overview, - "Recovery pipeline": page_pipeline, - "Press register": page_presses, - "Title catalogue": page_catalogue, - "Rights & copyright": page_rights, - "Where the copies are": page_holdings, - "Curriculum builder": page_curriculum, - "Methods & sources": page_methods, -} - - -def main(): - with st.sidebar: - st.markdown("### 📚 African Books\n#### Recovery Project") - st.caption("African authors · African presses · c. 1950–1979") - choice = st.radio("Section", list(PAGES), label_visibility="collapsed") - st.divider() - st.caption( - f"**{len(TITLES)}** titles · **{len(PRESSES)}** presses\n\n" - f"**{int(TITLES['stage2_preserved'].sum())}** with a known scan · " - f"**0** cleared for republication" - ) - st.caption( - "Every record carries its sources and a confidence flag. Rights information is " - "descriptive, not legal advice." - ) - PAGES[choice]() - - -if __name__ == "__main__": - main() + st.markdown("### " + bil("The data", "Les données")) + e1, e2, e3 = st.columns(3) + with e1: + st.download_button(t("Download the full catalogue (CSV)"), + T.to_csv(index=False).encode("utf-8"), + file_name="abrp_titles.csv", mime="text/csv", **LARGE) + with e2: + st.download_button(t("Download the press register (CSV)"), + P.to_csv(index=False).encode("utf-8"), + file_name="abrp_presses.csv", mime="text/csv", **LARGE) + with e3: + st.download_button(bil("Download country coverage (CSV)", + "Télécharger la couverture par pays (CSV)"), + COUNTRIES.to_csv(index=False).encode("utf-8"), + file_name="abrp_countries.csv", mime="text/csv", **LARGE) + st.caption(bil( + "The sourced research JSON, the ETL that produces these tables and this " + "application are all in the Space repository, in one flat folder. Rebuild " + "with `python build_data.py`, then `python selftest.py`.", + "Le JSON de recherche sourcé, l'ETL qui produit ces tables et cette " + "application sont tous dans le dépôt du Space, dans un seul dossier plat. " + "Reconstruire avec `python build_data.py`, puis `python selftest.py`.")) + +st.markdown("---") +st.caption(bil( + "African Books Recovery Project · every record carries its sources and a " + "confidence flag · undated means undated: no year has been estimated · rights " + "information describes published statutes and is not legal advice · country " + "outlines Natural Earth 1:50m, public domain.", + "Projet de récupération du livre africain · chaque notice porte ses sources et un " + "indice de fiabilité · non daté veut dire non daté : aucune année n'a été " + "estimée · l'information juridique décrit des textes publiés et ne vaut pas " + "conseil juridique · contours Natural Earth 1:50m, domaine public."))